Summary of the invention
For the problems of the prior art, the embodiment of the present invention provides a kind of smart machine controlling party based on machine learningMethod and device.
In order to solve the above technical problems, the embodiment of the present invention the following technical schemes are provided:
In a first aspect, the embodiment of the invention provides a kind of smart machine control method based on machine learning, comprising:
If detecting the application management program starting on mobile terminal, obtains the location information of the mobile terminal and work asPreceding operating time information；
By the location information and the current operating temporal information input into usual smart machine identification model, outputThe identification information of usual smart machine；
The application management program is controlled according to the identification information of the usual smart machine, so that the applicationManagement program shows the control information for remotely operating the usual smart machine；
Wherein, the usual smart machine identification model is according to historical operating data, based on machine learning algorithm trainingIt obtains；The historical operating data includes that the current operating temporal foregoing description mobile terminal utilizes the application management journeyAny smart machine of ordered pair completes location information where the mobile terminal locates when long-range control, operating time information and describedThe identification information of one smart machine.
Further, it is identified by the location information and the current operating temporal information input to usual smart machineIn model, before the identification information for exporting usual smart machine, the smart machine control method based on machine learning is also wrappedIt includes: establishing the usual smart machine identification model；
It is wherein, described to establish the usual smart machine identification model, comprising:
Detecting the mobile terminal using the application management program to the long-range control of any smart machine completion every timeWhen processed, the mark letter of location information where the mobile terminal locates, operating time information and any smart machine is obtainedBreath, the location information where the mobile terminal locates and operating time information that will acquire as sample input data, and, by instituteThe identification information of any smart machine is stated as sample output data, machine learning algorithm is based on, carries out model training, obtain instituteState usual smart machine identification model.
Further, the smart machine control method based on machine learning, further includes:
During establishing the usual smart machine identification model, detecting the mobile terminal using institute every timeWhen stating application management program to the long-range control of any smart machine completion, the operation content to any smart machine is also obtainedInformation；Correspondingly, the location information where the mobile terminal locates that will acquire and operating time information as sample input data,And using the identification information of any smart machine and the operation content information as sample output data, it is based on machineLearning algorithm carries out model training, obtains the usual smart machine identification model；
Correspondingly, described to identify the location information and the current operating temporal information input to usual smart machineIn model, when exporting the identification information of usual smart machine, corresponding operation content information is also exported；
Correspondingly, the identification information according to the usual smart machine controls the application management program,When so that the application management program showing the control information for remotely operating the usual smart machine, also make the controlIt include the corresponding operation content information in information processed.
Further, the location information for obtaining the mobile terminal, specifically includes:
Obtain the wireless signal strength indicating RSSI value conduct for the default specified smart machine that the mobile terminal listens toThe location information of the mobile terminal；Wherein, the default specified smart machine is fixed smart machine and quantity is greater than or waitsIn 2；
Correspondingly, described mobile whole detecting every time during establishing the usual smart machine identification modelWhen long-range control is completed to any smart machine using the application management program in end, the institute that the mobile terminal listens to is obtainedThe RSSI value for stating default specified smart machine is monitored as location information where the mobile terminal locates, and by the mobile terminalThe RSSI value and operating time information of the default specified smart machine arrived as sample input data, and, will be describedThe identification information of one smart machine is based on machine learning algorithm as sample output data, carries out model training, obtains described usedWith smart machine identification model.
Further, the location information for obtaining the mobile terminal, specifically includes:
The location information that the mobile terminal is presently in is obtained by the positioning software installed on the mobile terminal；
Or, obtaining the location information that the mobile terminal is presently in by global position system GPS locator；
Or, obtaining the position that the mobile terminal is presently in by the multiple pyroelectric infrared sensor nodes arranged in advanceConfidence breath；
Or, obtaining the position that the mobile terminal is presently in by the image processing algorithm based on computer machine visionInformation.
Second aspect, the embodiment of the invention also provides a kind of smart machine control device based on machine learning, comprising:
Module is obtained, if obtaining the mobile terminal for detecting that the application management program on mobile terminal startsLocation information and current operating temporal information；
Processing module, for knowing the location information and the current operating temporal information input to usual smart machineIn other model, the identification information of usual smart machine is exported；
Control module, for being controlled according to the identification information of the usual smart machine to the application management programSystem, so that the application management program shows the control information for remotely operating the usual smart machine；
Wherein, the usual smart machine identification model is according to historical operating data, based on machine learning algorithm trainingIt obtains；The historical operating data includes that the current operating temporal foregoing description mobile terminal utilizes the application management journeyAny smart machine of ordered pair completes location information where the mobile terminal locates when long-range control, operating time information and describedThe identification information of one smart machine.
Further, the smart machine control device based on machine learning, further includes: model construction module；
Wherein, the model construction module, is specifically used for:
Detecting the mobile terminal using the application management program to the long-range control of any smart machine completion every timeWhen processed, the mark letter of location information where the mobile terminal locates, operating time information and any smart machine is obtainedBreath, the location information where the mobile terminal locates and operating time information that will acquire as sample input data, and, by instituteThe identification information of any smart machine is stated as sample output data, machine learning algorithm is based on, carries out model training, obtain instituteState usual smart machine identification model.
Further, the model construction module is during establishing the usual smart machine identification model, every timeWhen detecting that the mobile terminal completes long-range control to any smart machine using the application management program, also acquisition pairThe operation content information of any smart machine；Correspondingly, the mobile terminal institute that the model construction module will acquireLocation information and operating time information as sample input data, and, by the identification information of any smart machineWith the operation content information as sample output data, it is based on machine learning algorithm, carries out model training, is obtained described usualSmart machine identification model；
Correspondingly, the processing module is by the location information and the current operating temporal information input to usual intelligenceIn energy equipment identification model, when exporting the identification information of usual smart machine, corresponding operation content information is also exported；
Correspondingly, the control module according to the identification information of the usual smart machine to the application management programIt is controlled, when so that the application management program showing the control information for remotely operating the usual smart machine, alsoSo that including the corresponding operation content information in the control information.
Further, the acquisition module, is specifically used for:
Obtain the wireless signal strength indicating RSSI value conduct for the default specified smart machine that the mobile terminal listens toThe location information of the mobile terminal；Wherein, the default specified smart machine is fixed smart machine and quantity is greater than or waitsIn 2；
Correspondingly, the model construction module exists every time during establishing the usual smart machine identification modelWhen detecting that the mobile terminal completes long-range control to any smart machine using the application management program, the shifting is obtainedDescribed preset that dynamic terminal monitoring arrives specifies the RSSI value of smart machine as location information where the mobile terminal locates, and willThe RSSI value and operating time information for the default specified smart machine that the mobile terminal listens to input number as sampleAccording to, and, using the identification information of any smart machine as sample output data, it is based on machine learning algorithm, carries out mouldType training obtains the usual smart machine identification model.
Further, the acquisition module, is specifically used for:
The location information for obtaining the mobile terminal, specifically includes:
The location information that the mobile terminal is presently in is obtained by the positioning software installed on the mobile terminal；
Or, obtaining the location information that the mobile terminal is presently in by global position system GPS locator；
Or, obtaining the position that the mobile terminal is presently in by the multiple pyroelectric infrared sensor nodes arranged in advanceConfidence breath；
Or, obtaining the position that the mobile terminal is presently in by the image processing algorithm based on computer machine visionInformation.
The third aspect the embodiment of the invention also provides a kind of electronic equipment, including memory, processor and is stored inOn reservoir and the computer program that can run on a processor, the processor are realized when executing described program such as first aspect instituteThe step of stating the smart machine control method based on machine learning.
Fourth aspect, the embodiment of the invention also provides a kind of non-transient computer readable storage mediums, are stored thereon withComputer program is realized when the computer program is executed by processor as described in relation to the first aspect based on the smart machine of machine learningThe step of control method.
As shown from the above technical solution, smart machine control method provided in an embodiment of the present invention based on machine learning andDevice is based on machine learning mode, according to user's habit (where when region is in being accustomed to operating that intelligence is setIt is standby) start the control interface of corresponding smart machine automatically, so as to save user stored from application management program it is multipleThe trouble for the smart machine to be controlled is found in smart machine, so that the operation of user becomes easy, it can directly quicklyGround controls the smart machine to be controlled, to improve user experience.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present inventionIn attached drawing, the technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment isA part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the artEvery other embodiment obtained without creative efforts, shall fall within the protection scope of the present invention.
The present inventor after research alone and based on the big data analysis to user operation habits data statistics, use by discoveryFamily, which controls the location of specific intelligent appliance and user and time using mobile terminal remote, has certain correlation.ExampleSuch as, certain user group is controlled in the sofa position in parlor between workaday 21 points to 22 points or before smart television with mobile phoneWhen intelligent appliance, there is biggish probability operation for controlling smart television, and at 23 points to 1 point of next day near the bed in bedroomWhen, then there is biggish probability operation for controlling intelligent air condition.
Based on the above cognition, this application provides the smart machine control methods based on machine learning, by believing positionBreath and current operating temporal information, using machine learning algorithm well known to neural network etc., can train to obtain as input dataThe high-precision user that user's specific intelligence equipment to be controlled (or specific operation function) can be exported is accustomed to prediction model(namely the subsequent described usual smart machine identification model), to be accustomed to automatic obtain to corresponding smart machine according to userControl.
Based on the above analysis, specific embodiment will be passed through below to the smart machine provided by the present application based on machine learningExplanation is explained in detail in the working principle and the course of work of control method.
Fig. 1 shows the flow chart of the smart machine control method based on machine learning of one embodiment of the invention offer,Referring to Fig. 1, the smart machine control method provided in an embodiment of the present invention based on machine learning, comprising:
Step 101: if detecting the application management program starting on mobile terminal, obtaining the position of the mobile terminalInformation and current operating temporal information.
In the present embodiment, the location information can be absolute location information (such as GPS geographical position coordinates), can also be with(movement end such as can be determined according to the range information of mobile terminal and indoor several default fixed points for relative position informationThe relative position information of end indoors).
In the present embodiment, the current operating temporal information, which refers to, is detecting the application management program on mobile terminalCorresponding temporal information when starting.The temporal information can refer to work Time of Day and time at weekend, can also refer to the noonIt with, can also refer to any time period in one day time in the evening, can also refer to other while include date information and clock informationTemporal information, can also be the temporal information of other various forms, the present embodiment is not construed as limiting this.
In the present embodiment, the application management program refers to the APP for being managed control to smart machine, theseAPP is typically mounted on mobile terminal, is remotely operated corresponding smart machine for user and is used.
In the present embodiment, the step of location information for obtaining the mobile terminal can detect mobile terminalOn application management program starting when execute automatically, can also detect on mobile terminal application management program starting when byUser triggers execution manually.
In the present embodiment, the mobile terminal can refer to smart phone, intelligent remote controller etc..The smart machine can be withRefer to various household appliances or office equipment, such as refrigerator, TV, air-conditioning, washing machine.
Step 102: the location information and the current operating temporal information input to usual smart machine are identified into mouldIn type, the identification information of usual smart machine is exported.
In this step, the usual smart machine identification model is according to current operating temporal foregoing description mobile terminalLocation information where the mobile terminal locates, behaviour when completing long-range control to any smart machine using the application management programMake temporal information and the identification information of any smart machine these historical operating datas, trained based on machine learning algorithmIt arrives.Therefore, after step 101 gets location information and current operating temporal information, by the location information and described work asPreceding operating time information is input in usual smart machine identification model as input data, is set so as to export usual intelligenceStandby identification information.
Step 103: the application management program is controlled according to the identification information of the usual smart machine, so thatThe application management program shows the control information for remotely operating the usual smart machine.
In the present embodiment, identification information (the also referred to as identification of smart machine of the smart machine is obtained in step 102Code) after, the identification information of the smart machine obtained according to step 102 controls the application management program, so that describedApplication management program shows the control information for remotely operating the smart machine, and controlling information here can be for for long-rangeIt operates the control button of the smart machine, may be the control instruction for remotely operating the smart machine, it can be withFor the control interface etc. comprising control button or control instruction.
For example, referring to fig. 2, it is assumed that find that user's habit controls sky in the noon in position A1 according to historical data1 is adjusted, and is accustomed to controlling TV in position A1 the time in the evening, then according to these historical operating datas, is based on machine learning algorithmIt can train to obtain usual smart machine identification model.If being obtained when detecting the application management program starting on mobile terminalThe location information of the mobile terminal taken is position A1 and the current operating time (time of starting application management program) is noonIt is time, then the location of mobile terminal information (position A1) and current operating temporal information (noon such as 12:30) is defeatedEnter into the usual smart machine identification model, corresponding usual intelligence is then exported by the usual smart machine identification modelThe identification information of energy equipment (air-conditioning 1), the identification information for being then based on air-conditioning 1 control the application management program, makeThe control information that shows of the application management program as shown in figure 3, its Fig. 3 and subsequent Fig. 4 and Fig. 5 are with control interfaceFor be illustrated.Referring to Fig. 3 it is found that shown on the application management program in the position with current operating temporal userIt is accustomed to the control interface of the smart machine (air-conditioning 1) of operation, consequently facilitating user controls air-conditioning 1 accordingly.Here, inIt can specifically be set between the period of the day from 11 a.m. to 1 p.m with, the general noon refers to 12. -14 points, and refer to 18. -22 the time in the evening the time in the eveningPoint.
For another example, it is assumed that find that user's habit carries out air-conditioning 2 in position A3 in summer the time in the evening according to historical dataControl, therefore, if the location information of the mobile terminal of acquisition is when detecting the application management program starting on mobile terminalPosition A3 and current operating time are in summer time in the evening, then by the location of mobile terminal information (position A3) and work asPreceding operating time information 20 points of June 22 (in summer such as) is input to the usual smart machine identification model time in the eveningIn, the identification information of corresponding usual smart machine (air-conditioning 2) is then exported by the usual smart machine identification model, thenThe application management program is controlled, so that the control interface that the application management program is shown is as shown in Figure 4.Referring to figure4 it is found that show the smart machine (sky for being accustomed to operation with current operating temporal user in the position on the application management programAdjust 2) control interface (it is shown in the control interface various for controlling the virtual keys of air-conditioning 2, such as can choose mode,Set temperature, control power supply opening and closing etc.), consequently facilitating user controls air-conditioning 2 accordingly.
In addition, it should be noted that, being not limited to Fig. 3 or Fig. 4 in the control interface that the application management program is shownShown in only comprising a kind of or a smart machine control interface, can also for it is as shown in Figure 5 include multiple smart machinesControl interface.For example, it is assumed that user is accustomed in summer in position A1 while operating air conditioner 1 and TV the time in the evening, thenAccording to these historical operating datas, can train to obtain usual smart machine identification model based on machine learning algorithm.If examiningWhen measuring the application management program starting on mobile terminal, the location information of the mobile terminal of acquisition is position A1 and current behaviourMake the time is summer time in the evening, then can believe the location of mobile terminal information (position A1) and current operating temporalBreath () is input in the usual smart machine identification model time in the evening, then defeated by the usual smart machine identification modelThe identification information of corresponding usual smart machine (air-conditioning 1 and TV) out, is then based on the identification information of air-conditioning 1 and TV to instituteApplication management program is stated to be controlled so that the control interface that the application management program is shown as shown in figure 5, consequently facilitating withFamily controls air-conditioning 1 and TV accordingly.It should be noted that the quantity of smart machine is not due in general familyLess than 10, therefore, using this control method provided in this embodiment, user can be saved and stored from application management programMultiple smart machines in find the smart machine to be controlled trouble so that user can directly be seen that the intelligence to be controlled is setStandby control interface, thus improve user experience.
In the present embodiment, it should be noted that usual smart machine identification model recited above is according to current behaviour(user's history remotely controls smart machine by the application management program historical operating data before making the timeData) generated based on machine learning algorithm training.Wherein, it is set based on the machine learning algorithm training generation usual intelligenceWhen standby identification model, generally by mobile terminal using application management program to smart machine progress history control when where positionInformation and corresponding operating time information make the identification information of the corresponding smart machine controlled as sample input dataFor sample output data, initial machine learning model is trained after meeting the model condition of convergence, is generated described usualSmart machine identification model.In the present embodiment, model training can be carried out using CNN or RNN machine learning model.In addition,It should be noted that generating the usual smart machine identification mould being based on machine learning algorithm training according to historical operating dataWhen type, the historical operating data of use is all newest historical operating data, for example, it may be nearest one week, nearest one month,Nearest three months or the nearest half a year closer historical operating data of equidistant current operating temporal, in order to improve described usedWith the recognition accuracy of smart machine identification model, it can relatively accurately identify user in corresponding position and operationTime most thinks the smart machine of control.
It should be noted that the present embodiment according to user habit start the control interface of corresponding smart machine automatically when,Not only consider user where region habit operate what smart machine, and also contemplate user where areaWhen domain is in being accustomed to that smart machine operated, to automatically and accurately provide user's intelligence currently to be controlled for userThe control interface of energy equipment, thus eliminate user and search from the more smart machine stored in application management program in the positionIt sets and in the trouble of the common smart machine of current time, to improve user experience.
In addition, it should be noted that, the smart machine control method based on machine learning provided by the present embodiment may be used alsoWith for user where region habit operate what smart machine control, namely relative to being recited above inHold, when to the usual smart machine identification model input data, need to only input the location information where mobile terminal,Without inputting current operating temporal information.Correspondingly, when being trained to the usual smart machine identification model, sample numberIt also no longer needs in comprising this content of temporal information.For example, referring to fig. 2, it is assumed that user is within the past timeGet used to always controlling air-conditioning 1 and TV in position A1, then it, can be with based on machine learning algorithm according to these historical operating datasTraining obtains usual smart machine identification model, then when detecting that mobile terminal is currently located at position A1, then can will moveThe location of dynamic terminal information A1 is input in the usual smart machine identification model, then by the usual smart machineIdentification model exports the identification information of corresponding usual smart machine (air-conditioning 1 and TV), is then based on the mark of air-conditioning 1 and TVKnow information to control the application management program, so that comprising useful in the control interface that the application management program is shownThe smart machine air-conditioning 1 of operation and the control interface of TV are accustomed in family in the position, consequently facilitating user to air-conditioning 1 and TV intoThe corresponding control of row.
In the present embodiment, it should be noted that the training data used when being trained based on machine learning algorithmThe source of collection can there are many modes, the present embodiment to be not construed as limiting to this, such as may is that A: coming solely from the use gathered in advanceThe historical operating data at family；B: coming solely from default training dataset, which is inventor according to adopting in advanceCollection multiple other users use habit data and generate, for example, can according to the user purchase when register age,Gender, the user correspond to the register informations such as the household electrical appliances type that family is bought to determine matched training dataset for the user；C: on the basis of default training dataset, the operating habit data set that further training obtains of the user is utilized.
As shown from the above technical solution, smart machine control method provided in an embodiment of the present invention, based on machine learningMode starts corresponding intelligence according to user's habit (where when region is in being accustomed to that smart machine operated) automaticallyCan equipment control interface, found from the multiple smart machines stored in application management program so as to saving user and be intended to controlThe trouble of the smart machine of system can directly and quickly set the intelligence to be controlled so that the operation of user becomes easyIt is standby to be controlled, to improve user experience.
It should be noted that smart machine control method provided in this embodiment can be by mobile terminal execution, it can also be withIt is executed, can also be executed jointly in a manner of information exchange by mobile terminal and server by server.
For example, in one implementation, smart machine control method provided in this embodiment can be held by mobile terminalRow.In this implementation, the training of the usual smart machine identification model is completed by mobile terminal.Meanwhile it is mobileTerminal obtains the position letter of mobile terminal itself when detecting thereon for controlling the application management program starting of smart machineBreath and current operating time information, then by the location information and the current operating temporal information input to described usualIn smart machine identification model, and obtain from the output end of the usual smart machine identification model mark of corresponding smart machineInformation is known, to control according to the identification information of the smart machine the application management program, so that the applicationManagement program shows the control interface or control command for remotely operating the smart machine, consequently facilitating using the movementThe user of terminal sees the control interface (or control command) for being accustomed to the smart machine of operation in the position at the first time, thus justIt is controlled in corresponding smart machine, thus eliminates and searched from the more smart machine stored in application management programThe trouble of smart machine is commonly used in the position.Furthermore, it is necessary to specified otherwise, smart machine controlling party provided in this embodimentMethod is different from some show on mobile terminal application management program according to location information of mobile terminal and is located near mobile terminalOr the control method of the smart machine information list of surrounding, because the central inventive thought of the present embodiment is to be accustomed to according to userStart the control interface of corresponding smart machine for user, rather than is to be provided near the position according to the position of user for userSmart machine control interface.Finally, it is emphasized that the present embodiment is intended to embody according to user's habit (in what positionRegion is set in when being accustomed to that smart machine operated) start this core of the control interface of corresponding smart machine hair automaticallyBright thought.
For another example, in one implementation, smart machine control method provided in this embodiment can be executed by server.In this implementation, the training of the usual smart machine identification model is completed by server.Correspondingly, server is logicalCross certain way detect on mobile terminal application management program starting when, obtain the mobile terminal location information andThen current operating temporal information sets the location information and the current operating temporal information input to the usual intelligenceIn standby identification model, and believe from the mark that the output end of the usual smart machine identification model obtains corresponding smart machineThen breath passes through the application management program of certain way on mobile terminals to installation according to the identification information of the smart machineIt is controlled, so that the application management program shows control interface or control life for remotely operating the smart machineIt enables.
It should be noted that in this implementation, server can be detected in several ways and is used on mobile terminalWhether the application management program of control smart machine starts, for example, such processing logic can be preset: in advance describedMonitoring software is installed, which can be communicated with the server, when the monitoring software monitors on mobile terminalWhen on to mobile terminal for controlling the application management program starting of smart machine, prompt information just is sent to the server.It wherein, further include the current location information and current operating temporal of the mobile terminal for thering is monitoring software to obtain in the prompt informationInformation.For another example, it can also realize by other means, logic is handled as can preset: when mobile terminal detectsWhen on to mobile terminal for controlling the application management program starting of smart machine, Xiang Suoshu server sends prompt information.ItsIn, it further include the current location information and current operating temporal information by acquisition for mobile terminal in the prompt information.
Similarly, server obtains the intelligence in the output end of the usual smart machine identification model from server localAfter the identification information of equipment, the application management program of installation on mobile terminals can also be controlled by the monitoring softwareSystem, so that the application management program shows the control interface or control command for remotely operating the smart machine.Alternatively,The identification information of the smart machine can be sent to mobile whole by server after the identification information for obtaining the smart machineEnd, controls the application management program by mobile terminal, so that the application management program is shown for remotely operatingThe control interface or control command of the smart machine.
In this implementation, similarly, server generates the usual intelligence in the mode training based on machine learningWhen equipment identification model, the sample data from acquisition for mobile terminal training is needed, when the sample data includes: historyBetween location information of middle user when remotely being controlled by the application management program smart machine, operating time information withAnd the identification information of the corresponding smart machine operated.Server when from these sample datas of the acquisition for mobile terminal,Monitoring software the relevant technologies recited above can also be used.For example, monitoring answering on the mobile terminal using monitoring softwareWhether remote control operation is completed to smart machine with management program, and is monitoring the application management journey on the mobile terminalOrdered pair smart machine complete remote control operation when, obtain this time remotely control corresponding location information, operating time information withAnd the identification information of the smart machine operated, and these information that will acquire are sent to server, so that server canTo complete the training process of the usual smart machine identification model in server local.
For another example, in one implementation, smart machine control method provided in this embodiment can by mobile terminal andServer is executed jointly in a manner of information exchange.In this implementation, the training of the usual smart machine identification modelWork is completed by server.Correspondingly, it is detected by mobile terminal for controlling whether the application management program of smart machine starts,And when detecting starting, the location information and current operating temporal information of mobile terminal itself are obtained, then by the positionInformation and the current operating temporal information are sent to server, so that server is by the location information and the current operationTemporal information is input in the usual smart machine identification model, then by the usual smart machine identification model output pairThe identification information for the smart machine answered, server is after the identification information for obtaining the smart machine, by the smart machineIdentification information is sent to mobile terminal, by mobile terminal according to the identification information of the smart machine to the application management programControlled so that the application management program show for remotely operating the smart machine control interface (or control lifeIt enables).
Further, content based on the above embodiment, in the present embodiment, the step 101 or step 102 itBefore, the smart machine control method based on machine learning, further includes:
Step 100: establishing the usual smart machine identification model.
It is in this step, described to establish the usual smart machine identification model, comprising:
Detecting the mobile terminal using the application management program to the long-range control of any smart machine completion every timeWhen processed, the mark letter of location information where the mobile terminal locates, operating time information and any smart machine is obtainedBreath, the location information where the mobile terminal locates and operating time information that will acquire as sample input data, and, by instituteThe identification information of any smart machine is stated as sample output data, machine learning algorithm is based on, carries out model training, obtain instituteState usual smart machine identification model.
In the present embodiment, it should be noted that the usual smart machine identification model is the mistake of a dynamic trainingJourney and the usual smart machine identification model are not unalterable, but according to constantly occurring to smart machineThe long-range controlling behavior location information, operating time information and the corresponding smart machine that are included identification information, be based onMachine learning algorithm obtains after constantly training to model progress.It is also the usual smart machine identification model according to recentlyUser's operation behavior can constantly update so that improving the recognition accuracy of the usual smart machine identification model.In the present embodiment, to keep the usual smart machine identification model that can be continuously updated, when detecting the shifting every timeWhen dynamic terminal completes long-range control to any smart machine using the application management program, it is required to obtain the mobile terminalThe identification information of the location information at place, operating time information and any smart machine, and the movement that will acquireLocation information and operating time information where terminal as sample input data, and, by the mark of any smart machineInformation is known as sample output data, is then based on machine learning algorithm, is carried out model training, obtains the usual smart machineIdentification model.
For example, with reference to Fig. 2, it is assumed that in 1-3 month, user gets used to mobile terminal and carries out remotely in the position A1 to TVControl, but found since April, user's mobile terminal accustomed to using remotely controls TV in the position A2 (mayIt is that the owner of family goes on business, that carry out family lodging is other friends, and friend habit remotely controls TV in position A2System), therefore, according to the operation behavior several times of the friend, so that it may be based on machine learning algorithm, carry out constantly training more to modelNewly, so that the recognition accuracy of usual smart machine identification model improves, gradually meet the habit of active user.
In the present embodiment, when carrying out model training by way of machine learning, CNN or RNN model can be used.It is illustrated by taking CNN model as an example below with reference to Fig. 6, it should be noted that Fig. 6 is a schematic model, wherein only simpleTwo convolutional layers and two pond layers are illustrated, in practical applications, the number of convolutional layer and pond layer is generally greater than 2It is a.Specifically, the structure of CNN model specifically includes that an input layer, n convolutional layer, n pond layer, m full articulamentums, oneA output layer；Wherein, the input of the input layer is defeated for the sample of the location information comprising mobile terminal and operating time informationEnter data, input layer is connected with convolutional layer C1；The convolutional layer C1 contains the convolution kernel that k1 size is a1 × a1, described defeatedThe sample input data for entering layer obtains k1 characteristic pattern by convolutional layer C1, and then obtained characteristic pattern is sent to pond layerP1；The pond layer P1 carries out pond to the characteristic pattern that the convolutional layer C1 is generated with the sample size of b1 × b1, obtains correspondingK1 sampling after characteristic pattern, then obtained characteristic pattern is sent to next convolutional layer C2；The n convolutional layer and pondLayer constantly extracts the Sampling characters of sample input data profound level to being sequentially connected with, the last one pond layer Pn with entirelyArticulamentum F1 is connected, wherein convolutional layer Ci contains the convolution kernel that ki size is ai × ai, and the sample size of pond layer Pj isBj × bj, Ci indicate that i-th of convolutional layer, Pj indicate j-th of pond layer；The full articulamentum F1 is the last one described pond layerOne-dimensional layer made of the pixel mapping of the resulting all kn characteristic patterns of Pn, each pixel represent the one of the full articulamentum F1A neuron node, F1 layers of all neuron nodes are connect entirely with the neuron node of next full articulamentum F2；Through mA full articulamentum is sequentially connected with, the last one full articulamentum Fm is connect entirely with the output layer；The output layer output packetThe sample output data of identification information containing usual smart machine.In the present embodiment, believed using the position comprising mobile terminalThe sample input data of breath and operating time information, and the sample output data of the identification information comprising usual smart machine, baseIn machine learning algorithm, above-mentioned CNN model is trained until above-mentioned CNN model is restrained, and then is obtained described usualSmart machine identification model.
For example, when carrying out model training based on machine learning algorithm, it is assumed that by the note to historical operating dataRecord, the sample data for training pattern got are as shown in table 1 below.
Table 1
For upper table 1, it is assumed that position Ax is the sofa in family, first time period 12:00-14:00, described theTwo periods were 19:00-22:00, and smart machine a is air-conditioning, smart machine b is washing machine, and smart machine c is TV, becauseWhatsoever the time on the sofa, opens air-conditioning when being all accustomed to opening TV, and there was only hotter at noon to user,And it can just be taken using laundry washer when only at night.As it can be seen that using above-mentioned sample data to model training after, canAllow to the obtained usual smart machine identification model of training and the recognition result for more matching user demand is provided, and then can be withThe control interface of more matching user habit is provided for user, so as to improve user experience.In addition, it should be noted that,Upper table 1 is intended merely to facilitate one of citing to illustrate, and data volume will be far longer than institute in above-mentioned table 1 when reality carries out sample trainingThe content shown.
Further, content based on the above embodiment, in the present embodiment, the smart machine based on machine learningControl method further include:
During establishing the usual smart machine identification model, detecting the mobile terminal using institute every timeWhen stating application management program to the long-range control of any smart machine completion, the operation content to any smart machine is also obtainedInformation；Correspondingly, the location information where the mobile terminal locates that will acquire and operating time information as sample input data,And using the identification information of any smart machine and the operation content information as sample output data, it is based on machineLearning algorithm carries out model training, obtains the usual smart machine identification model；
Correspondingly, described to identify the location information and the current operating temporal information input to usual smart machineIn model, when exporting the identification information of usual smart machine, corresponding operation content information is also exported；
Correspondingly, the identification information according to the usual smart machine controls the application management program,When so that the application management program showing the control information for remotely operating the usual smart machine, also make the controlIt include the corresponding operation content information in information processed.
It should be noted that the present embodiment is on the basis of above-described embodiment, to increase in the corresponding operation of displayHold this content of information, so that the content that control interface is shown more matches the demand of user, to further saveOperating time of user, and then improve user experience.
As an example it is assumed that user's habit sees the program in TV in noon (first time period) habit in position Ax1 (such as midday news), and be accustomed to seeing program 2 (such as football match) in (second time period) habit in position Ax the time in the evening.CauseThis, when user is currently located at position Ax, and current time be at night, when starting application management program, the application management programNot only show the control interface for remotely operating TV, be also used to show the triggering to program 2 or control button, in order toFamily fast implements the starting to program 2.
In the present embodiment, the operation content information can be control model, the operating mode of smart machine, work frequencyThe information such as road, operational detail, the present embodiment are not construed as limiting this.For example, can be series channel, electricity for TVThe channel informations such as shadow channel, music channel, or specific programme information.In addition, for refrigerator, in the operationHolding to be the operating mode of refrigerator, as battery saving mode (being relatively specific for working day) and normal mode (are relatively specific for weekEnd).In addition, the operation content can be refrigeration mode, heating mode, the dehumidification mode etc. of air-conditioning for air-conditioning, intoOne step can also be the specific set temperature etc. under refrigeration mode.For the concrete meaning of the operation content, the present embodimentNo longer illustrate one by one.
In the present embodiment, when carrying out model training based on machine learning algorithm, it is assumed that by historical operating dataRecord, the sample data for training pattern got is as shown in table 2 below.
Table 2
For upper table 2, it is assumed that position Ax is the sofa in family, first time period 12:00-14:00, described theTwo periods were 19:00-22:00, and smart machine a is TV, because time (first time period) habit sees section to user at noonMesh 1 (such as midday news), and time (second time period) habit sees program 2 (such as football match) at night, it is seen then that using above-mentionedAfter sample data is to model training, enables to the usual smart machine identification model that can be trained to provide more matching and useThe recognition result of family demand, and then the control interface of more matching user habit can be provided for user, so as to improve useFamily experience.In addition, it should be noted that, upper table 2 is intended merely to facilitate one of citing to illustrate, it is practical to carry out number when sample trainingTo be far longer than content shown in above-mentioned table 2 according to amount and data content.
Further, content based on the above embodiment, it is described to obtain the movement eventually in a kind of optional embodimentThe location information at end, specifically includes:
The location information that the mobile terminal is presently in is obtained by the positioning software installed on the mobile terminal.
In the present embodiment, the location information that the mobile terminal is presently in can be by pacifying on the mobile terminalThe positioning software of dress directly acquires.It should be noted that be generally in room by the location information obtained in this present embodiment orThe more specific location information of Office Area, therefore it is required that the positioning accuracy of positioning software wants sufficiently high, can distinguish in room or doDifferent zones position is minimum requirements in public area.For example, being subject to the rooms of 90 square meters, the positioning accuracy of positioning software needs fullFoot can distinguish parlor position and bedroom position is minimum requirements.
In addition, the location information that the mobile terminal is presently in can also be by global position system GPS (GlobalPositioning System) locator acquisition.In addition, the location information that the mobile terminal is presently in can also pass through peopleBody infrared sensor (pyroelectric infrared sensor) obtains.Further, it is also possible to pass through computer machine visual correlation technology (such as baseIn the image processing algorithm of computer machine vision) it is positioned, since the contents of the section can be using current comparative maturityLocation technology, therefore I will not elaborate.It should be noted that the positioning result obtained in this way is typically all one straightConnect positioning result, namely what the location information obtained referred to is exactly absolute location information that the mobile terminal is presently in.
Further, content based on the above embodiment, it is described to obtain the movement in another optional embodimentThe location information of terminal, specifically includes:
Obtain the wireless signal strength indicating RSSI value conduct for the default specified smart machine that the mobile terminal listens toThe location information of the mobile terminal；Wherein, the default specified smart machine is fixed smart machine and quantity is greater than or waitsIn 2；
Correspondingly, described mobile whole detecting every time during establishing the usual smart machine identification modelWhen long-range control is completed to any smart machine using the application management program in end, the institute that the mobile terminal listens to is obtainedThe RSSI value for stating default specified smart machine is monitored as location information where the mobile terminal locates, and by the mobile terminalThe RSSI value and operating time information of the default specified smart machine arrived as sample input data, and, will be describedThe identification information of one smart machine is based on machine learning algorithm as sample output data, carries out model training, obtains described usedWith smart machine identification model.
In the present embodiment, when obtaining the location information of the mobile terminal, there is no obtained using positioning softwareThe absolute position of mobile terminal, but specify smart machine as location reference point using presetting in room, then according to movementTerminal monitoring to the wireless signal strength indicating RSSI value of the default specified smart machine determine the position of the mobile terminalIt sets.
In the present embodiment, the default specified smart machine is fixed smart machine and quantity is greater than or equal to 2, itIt requires that quantity is greater than or equal to 2 is to guarantee positioning accuracy.For example, 3 default specified smart machines can be set, byIt is more accurate in three-point fix, therefore pass through the setting of 3 default specified smart machines, it is ensured that positioning accuracy.
It is set it should be noted that being provided with multiple intelligence with wireless communication function in general user family or in officeStandby (smart home device, intelligent appliance), the smart machine including fixed position and non-fixed position smart machine.ItsIn, after the smart machine of fixed position refers to that air-conditioning, refrigerator etc. one is set, usually will not shift position easily intelligent familyElectricity is suitable as the reference point of positioning mobile terminal.Correspondingly, the small intelligents equipment such as intelligent sound box is due to when in useSetting position may be often replaced, therefore is not suitable for as a reference point.For example, the default specified smart machine can be withFor refrigerator, television set, air-conditioning 1, air-conditioning 2, air-conditioning 3 and washing machine in Fig. 2, these are fixed into the smart machine of position as described inThe wireless signal strength of default specified smart machine, the default specified smart machine that can be listened to according to mobile terminal refers toShow that RSSI value accurately determines the position of the mobile terminal.Here, the position of the mobile terminal can be understood as the position of userIt sets.
It should be noted that the wireless signal strength indicating RSSI value mentioned in the present embodiment can be WiFi signal intensityIndicating RSSI value may be bluetooth signal intensity indicating RSSI value.
In addition, in the present embodiment, the monitoring of RSSI value can also be used in combination with infrared sensor of the human body, for example,Can be after infrared sensor of the human body detect human body, then RSSI value is obtained, to reduce power consumption and operand.
It should be noted that mobile terminal can be in promiscuous mode when preset monitored specifies the RSSI value of smart machineUnder, read the data packet of each smart machine.Alternatively, all smart machines under the user account is ordered to be broadcasted for measuring RSSIThe beacon frame of value, to obtain the RSSI value of default specified smart machine.Wherein, the default specified smart machine eitherBy user's manual setting, it is also possible to what mobile terminal was identified by the identification code of smart machine.
For example, user utilizes smart phone (or intelligent remote controller) to start intelligent device management APP in the roomWhen, after APP logins user account, smart phone monitors (such as promiscuous mode) fixed intelligent appliance-ice by WiFi moduleBeacon frame that case, television set, air-conditioning 1, air-conditioning 2, air-conditioning 3, washing machine are issued (standard beacon frame or with beacon functionCustomized normal data frame), acquire the RSSI value of smart phone position.It should be noted that if smart phone is logicalThe RSSI value that bluetooth module monitors each intelligent appliance is crossed, then can not influence the normal use of mobile phone WiFi network.
It should be noted that although present embodiment has references to RSSI Indoor Position Techniques Based on Location Fingerprint, but the present embodiment withoutOffline sample phase and real-time positioning stage need to be divided into, each operation of user is all recorded in foundation of the database as positioning(fingerprint location method must be sampled offline in advance, and only record RSSI value in offline sample phase, in real-time positioning stageRSSI value be only used for positioning, which can not be recorded in database), significantly reduce the learning cost of user.ThisOutside, present embodiment is without location coordinate and access point layout is established, without the positioning accurate of fingerprint location technology requirementDegree, therefore reduce operand and required precision.
In addition, in a preferred embodiment, it, can also basis before the RSSI value of acquisition mobile terminal positionThe SSID or network segment of the current GPS location of mobile terminal or the Wi-Fi hotspot of connection carry out the group of smart machineFiltering (for example, the intelligent appliance for being placed on office is group, office, the intelligent appliance for being placed on family is group, family),So that leaving the smart machine group being adapted to current scene in APP.For example, filtering out and doing when mobile terminal is located at houseGong Shi group filters out group, family when mobile terminal is located at office.The advantages of this processing, is: facilitating user manualThe default specified smart machine is set.In addition, the default specified smart machine can be the intelligence bound with user accountEquipment, or the smart machine that do not bound with user account.
Fig. 7 shows the structural representation of the smart machine control device based on machine learning of one embodiment of the invention offerFigure, referring to Fig. 7, the smart machine control device provided in an embodiment of the present invention based on machine learning, comprising: acquisition module 21,Processing module 22 and control module 23, in which:
Module 21 is obtained, if obtaining described mobile whole for detecting that the application management program on mobile terminal startsThe location information and current operating temporal information at end；
Processing module 22, for by the location information and the current operating temporal information input to usual smart machineIn identification model, the identification information of usual smart machine is exported；
Control module 23, for being controlled according to the identification information of the usual smart machine to the application management programSystem, so that the application management program shows the control information for remotely operating the usual smart machine；
Wherein, the usual smart machine identification model is according to historical operating data, based on machine learning algorithm trainingIt obtains；The historical operating data includes that the current operating temporal foregoing description mobile terminal utilizes the application management journeyAny smart machine of ordered pair completes location information where the mobile terminal locates when long-range control, operating time information and describedThe identification information of one smart machine.
Further, content based on the above embodiment, in the present embodiment, the smart machine based on machine learningControl device, further includes: model construction module；
Wherein, the model construction module, is specifically used for:
Detecting the mobile terminal using the application management program to the long-range control of any smart machine completion every timeWhen processed, the mark letter of location information where the mobile terminal locates, operating time information and any smart machine is obtainedBreath, the location information where the mobile terminal locates and operating time information that will acquire as sample input data, and, by instituteThe identification information of any smart machine is stated as sample output data, machine learning algorithm is based on, carries out model training, obtain instituteState usual smart machine identification model.
Further, content based on the above embodiment, in the present embodiment, the model construction module is described in the foundationDuring usual smart machine identification model, every time detect the mobile terminal using the application management program to appointWhen one smart machine completes long-range control, the operation content information to any smart machine is also obtained；Correspondingly, the mouldThe type location information where the mobile terminal locates that will acquire of building module and operating time information as sample input data, withAnd using the identification information of any smart machine and the operation content information as sample output data, it is based on engineeringAlgorithm is practised, model training is carried out, obtains the usual smart machine identification model；
Correspondingly, the processing module is by the location information and the current operating temporal information input to usual intelligenceIn energy equipment identification model, when exporting the identification information of usual smart machine, corresponding operation content information is also exported；
Correspondingly, the control module according to the identification information of the usual smart machine to the application management programIt is controlled, when so that the application management program showing the control information for remotely operating the usual smart machine, alsoSo that including the corresponding operation content information in the control information.
Further, content based on the above embodiment, in a kind of optional embodiment, the acquisition module, specificallyFor:
The location information that the mobile terminal is presently in is obtained by the positioning software installed on the mobile terminal.
Further, content based on the above embodiment, in another optional embodiment, the acquisition module, toolBody is used for:
Obtain the wireless signal strength indicating RSSI value conduct for the default specified smart machine that the mobile terminal listens toThe location information of the mobile terminal；Wherein, the default specified smart machine is fixed smart machine and quantity is greater than or waitsIn 2；
Correspondingly, the model construction module exists every time during establishing the usual smart machine identification modelWhen detecting that the mobile terminal completes long-range control to any smart machine using the application management program, the shifting is obtainedDescribed preset that dynamic terminal monitoring arrives specifies the RSSI value of smart machine as location information where the mobile terminal locates, and willThe RSSI value and operating time information for the default specified smart machine that the mobile terminal listens to input number as sampleAccording to, and, using the identification information of any smart machine as sample output data, it is based on machine learning algorithm, carries out mouldType training obtains the usual smart machine identification model.
Since the smart machine control device provided in this embodiment based on machine learning can be used for executing above-mentioned secondSmart machine control method described in a embodiment based on machine learning, working principle is similar with beneficial effect, therefore hereinIt is no longer described in detail, particular content can be found in the introduction of above-described embodiment.
Based on identical inventive concept, further embodiment of this invention provides a kind of electronic equipment, referring to Fig. 8, the electricitySub- equipment specifically includes following content: processor 601, memory 602, communication interface 603 and communication bus 604；
Wherein, the processor 601, memory 602, communication interface 603 are completed each other by the communication bus 604Communication；The communication interface 603 is for realizing between the relevant devices such as each modeling software and intelligent manufacturing equipment module libraryInformation transmission；
The processor 601 is used to call the computer program in the memory 602, and the processor executes the meterThe Overall Steps that the above-mentioned smart machine control method based on machine learning is realized when calculation machine program, for example, the processor is heldFollowing step is realized when the row computer program: if detecting the application management program starting on mobile terminal, obtaining instituteState the location information and current operating temporal information of mobile terminal；The location information and the current operating temporal information is defeatedEnter into usual smart machine identification model, exports the identification information of usual smart machine；According to the usual smart machineIdentification information controls the application management program, so that the application management program is shown for remotely operating described be used toWith the control information of smart machine；Wherein, the usual smart machine identification model is to be based on machine according to historical operating dataLearning algorithm training obtains；The historical operating data includes that the current operating temporal foregoing description mobile terminal utilizes instituteState location information where the mobile terminal locates when application management program completes long-range control to any smart machine, operating timeThe identification information of information and any smart machine.
It should be noted that the electronic equipment mentioned in the present embodiment can be mobile terminal, or cloud serviceDevice.
Based on identical inventive concept, further embodiment of this invention provides a kind of non-transient computer readable storage mediumMatter is stored with computer program on the computer readable storage medium, which realizes above-mentioned when being executed by processorThe Overall Steps of smart machine control method based on machine learning, for example, when the processor executes the computer programIt realizes following step: if detecting the application management program starting on mobile terminal, obtaining the position letter of the mobile terminalBreath and current operating temporal information；The location information and the current operating temporal information input to usual smart machine are knownIn other model, the identification information of usual smart machine is exported；According to the identification information of the usual smart machine to the applicationManagement program is controlled, so that the application management program shows that the control for remotely operating the usual smart machine is believedBreath；Wherein, the usual smart machine identification model is to be obtained according to historical operating data based on machine learning algorithm training's；The historical operating data includes that the current operating temporal foregoing description mobile terminal utilizes the application management program pairAny smart machine completes location information, operating time information and any intelligence where the mobile terminal locates when long-range controlThe identification information of energy equipment.
In addition, the logical order in above-mentioned memory can be realized and as independence by way of SFU software functional unitProduct when selling or using, can store in a computer readable storage medium.Based on this understanding, of the inventionTechnical solution substantially the part of the part that contributes to existing technology or the technical solution can be with software in other wordsThe form of product embodies, which is stored in a storage medium, including some instructions use so thatOne computer equipment (can be personal computer, server or the network equipment etc.) executes each embodiment institute of the present inventionState all or part of the steps of method.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic or disk etc. are various can be withStore the medium of program code.
The apparatus embodiments described above are merely exemplary, wherein described, unit can as illustrated by the separation memberIt is physically separated with being or may not be, component shown as a unit may or may not be physics listMember, it can it is in one place, or may be distributed over multiple network units.It can be selected according to the actual needsIn some or all of the modules realize the purpose of the embodiment of the present invention.Those of ordinary skill in the art are not paying woundIn the case where the labour for the property made, it can understand and implement.
Through the above description of the embodiments, those skilled in the art can be understood that each embodiment canIt realizes by means of software and necessary general hardware platform, naturally it is also possible to pass through hardware.Based on this understanding, onStating technical solution, substantially the part that contributes to existing technology can be embodied in the form of software products in other words, shouldComputer software product may be stored in a computer readable storage medium, such as ROM/RAM, magnetic disk, CD, including several fingersIt enables and using so that a computer equipment (can be personal computer, server or the network equipment etc.) executes each implementationSmart machine control method based on machine learning described in certain parts of example or embodiment.
In the description of the present invention, it should be noted that the orientation or positional relationship of the instructions such as term " on ", "lower" is baseIn orientation or positional relationship shown in the drawings, it is merely for convenience of description of the present invention and simplification of the description, rather than indication or suggestionSignified device or element must have a particular orientation, be constructed and operated in a specific orientation, therefore should not be understood as to thisThe limitation of invention.Unless otherwise clearly defined and limited, term " installation ", " connected ", " connection " shall be understood in a broad sense, exampleSuch as, it may be fixed connection or may be dismantle connection, or integral connection；It can be mechanical connection, be also possible to be electrically connectedIt connects；It can be directly connected, the connection inside two elements can also be can be indirectly connected through an intermediary.For thisFor the those of ordinary skill in field, the specific meanings of the above terms in the present invention can be understood according to specific conditions.
It should also be noted that, herein, relational terms such as first and second and the like are used merely to oneEntity or operation are distinguished with another entity or operation, without necessarily requiring or implying between these entities or operationThere are any actual relationship or orders.Moreover, the terms "include", "comprise" or its any other variant are intended to containLid non-exclusive inclusion, so that the process, method, article or equipment including a series of elements is not only wanted including thoseElement, but also including other elements that are not explicitly listed, or further include for this process, method, article or equipmentIntrinsic element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded thatThere is also other identical elements in process, method, article or equipment including the element.
Finally, it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；AlthoughPresent invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: it still may be usedTo modify the technical solutions described in the foregoing embodiments or equivalent replacement of some of the technical features；And these are modified or replaceed, technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution spirit andRange.