Calculating a value of loss to be treated, wherein 1[ k ] is not equal to i]The value range is {0,1}, when [ · is]Is a true functional expression of 1 when [. Cndot.)]Assuming that the function expression is 0, x is a sample characteristic characterization distance, tau is a constant, inputting the sample characteristic characterization distance into a loss function value for calculation to obtain a calculation result of the loss function, and taking the calculation result of the loss function as a loss value to be processed.

After the loss value to be processed is determined according to the sample characteristic characterization information, the target loss value can be determined according to the loss value to be processed and the loss cost values, the loss value to be processed and the loss cost values can be input into the loss value analysis processing model, the loss value to be processed and the loss cost values are analyzed by the loss value analysis processing model, the loss value output by the loss value analysis processing model is obtained, and the loss value obtained through output is used as the target loss value.

S504: and if the target loss value meets the loss condition, taking the Federal machine learning model obtained by training as a cross-domain space-time attribute identification model.

The loss condition refers to a judgment condition which is set in advance for a target loss value, and if the target loss value meets the loss condition, the Federal machine learning model obtained through training can be used as a cross-domain spatiotemporal attribute identification model.

In the embodiment of the disclosure, after the loss value to be processed is determined according to the sample characteristic characterization information and the target loss value is determined according to the loss value to be processed and the loss cost values, whether the target loss value meets the loss condition or not can be judged, and if the target loss value meets the loss condition, the federal machine learning model obtained by training is used as a cross-domain space-time attribute identification model.

In the embodiment, a plurality of sample space-time attribute features are respectively input into an initial federal machine learning model to obtain sample feature characterization information output by the federal machine learning model, a loss value to be processed is determined according to the sample feature characterization information, a target loss value is determined according to the loss value to be processed and a plurality of loss cost values, if the target loss value meets a loss condition, the federal machine learning model obtained by training is used as a cross-domain space-time attribute recognition model, so that the target loss value can be calculated, the training convergence of the federal machine learning model is judged according to the target loss value, the feature extraction effect of the cross-domain space-time attribute recognition model obtained by training is improved, and whether the region to be recognized is a dangerous chemical risk region or not is determined according to the cross-domain space-time attribute features, so that the dangerous chemical risk region recognition effect can be effectively improved.

S505: area information of an area to be identified is determined.

S506: and inputting the region information into the cross-domain space-time attribute identification model to obtain cross-domain space-time attribute characteristics which are output by the cross-domain space-time attribute identification model and correspond to the region information.

S507: and determining whether the area to be identified is a dangerous chemical risk area or not according to the cross-domain space-time attribute characteristics.

For the description of S505 to S507, reference may be made to the above embodiments, which are not described herein again.

In the embodiment, the cross-domain space-time attribute characteristics corresponding to the region information are obtained by determining the region information of the region to be identified, wherein, the cross-domain space-time attribute characteristic is an attribute correlation characteristic between first space-time data and second space-time data which characterize the region to be identified, the first space-time data corresponds to a first data domain, the second space-time data corresponds to a second data domain, and the first data domain is different from the second data domain, and determining whether the region to be identified is a dangerous chemical risk region according to the cross-domain space-time attribute characteristics, analyzing the cross-domain space-time attribute characteristics corresponding to the region to be identified, and realizing effective utilization of different data domain space-time data, and realizes the effective judgment of whether the area to be identified is the dangerous chemical risk area or not by utilizing the cross-domain space-time attribute characteristics, thereby effectively improving the identification efficiency of the risk area of the hazardous chemical substances, effectively improving the identification accuracy of the risk area of the hazardous chemical substances, respectively inputting a plurality of sample space-time attribute characteristics into an initial federated machine learning model to obtain sample characteristic characterization information output by the federated machine learning model, determining a loss value to be processed according to the sample characteristic characterization information, determining a target loss value according to the loss value to be processed and a plurality of loss cost values, if the target loss value meets the loss condition, the Federal machine learning model obtained by training is used as a cross-domain space-time attribute recognition model, thereby calculating a target loss value and judging the training convergence of the Federal machine learning model according to the target loss value so as to improve the characteristic extraction effect of the cross-domain space-time attribute recognition model obtained by training, and determining whether the region to be identified is a dangerous chemical risk region or not according to the cross-domain space-time attribute characteristics, so that the identification effect of the dangerous chemical risk region can be effectively improved.

Fig. 6 is a schematic flow chart of a method for identifying a risk region of a hazardous chemical substance according to another embodiment of the present disclosure.

As shown in fig. 6, the method for identifying a dangerous chemical risk region includes:

s601: an initial federated machine learning model is constructed.

For description of S601, reference may be made to the above embodiments for example, which are not described herein again.

S602: and obtaining a plurality of sample residing area information respectively output by a plurality of federal contrast submodels.

The sample resident area information refers to corresponding resident area information obtained after processing sample spatio-temporal data.

In the embodiment of the disclosure, when obtaining a plurality of sample residing area information respectively output by a plurality of federal contrast submodels, sample time-space data can be input into the federal contrast submodels, the sample time-space data is analyzed and processed by using the federal contrast submodels, vehicle trajectory data in the sample time-space data is calculated and processed to extract sample residing points in the sample time-space data, and then the sample residing area information can be identified according to the sample residing points to obtain a plurality of sample residing area information respectively output by the plurality of federal contrast submodels.

S603: and aligning the space-time attribute characteristics of the plurality of samples according to the resident area information of the samples.

The embodiment of the disclosure can acquire a plurality of sample residing area information respectively output by a plurality of federal contrast submodels before processing a plurality of sample time-space attribute characteristics to obtain a positive sample time-space attribute characteristic pair, and perform alignment processing on the plurality of sample time-space attribute characteristics according to the sample residing area information.

In the embodiment of the present disclosure, when aligning multiple sample time-space attribute features according to sample residence area information, the multiple sample time-space attribute features may be aligned according to area identity information in the sample residence area information based on a federal privacy Protection Set Intersection (PSI), so as to obtain aligned sample time-space attribute features.

S604: processing the plurality of sample space-time attribute features to obtain a positive sample space-time attribute feature pair, wherein the positive sample space-time attribute feature pair comprises: sample spatio-temporal attribute features corresponding to trajectory vectors belonging to the same dwell region.

S605: processing the multiple sample space-time attribute features to obtain a negative sample space-time attribute feature pair, wherein the negative sample space-time attribute feature pair comprises: sample spatio-temporal attribute features corresponding to trajectory vectors belonging to different dwell regions.

S606: and inputting the positive sample space-time attribute feature pairs and the negative sample space-time attribute feature pairs into an initial federated machine learning model to obtain positive sample feature characterization distances between the positive sample space-time attribute feature pairs and negative sample feature characterization distances between the negative sample space-time attribute feature pairs output by the federated machine learning model.

S607: and taking the positive sample characteristic distance and the negative sample characteristic distance as sample characteristic information.

S608: and determining a loss value to be processed according to the sample characteristic characterization information, and determining a target loss value according to the loss value to be processed and the loss cost values.

S609: and if the target loss value meets the loss condition, taking the Federal machine learning model obtained by training as a cross-domain space-time attribute identification model.

For an example, the descriptions of S604 to S609 refer to the above embodiments, which are not described herein again.

Optionally, in some embodiments, after a federal machine learning method is adopted, and an initial federal machine learning model is trained in combination with a plurality of loss cost values respectively output by a plurality of federal contrast submodels and a plurality of loss cost values respectively corresponding to the plurality of federal contrast submodels, a plurality of risk region labels respectively corresponding to a plurality of sample resident region information are determined, a plurality of sample spatio-temporal attribute features aligned according to the sample resident region information are determined, an initial artificial intelligence model is trained according to the aligned plurality of sample spatio-temporal attribute features and the plurality of risk region labels until the artificial intelligence model converges, the trained artificial intelligence model is used as a risk region identification model, so that the artificial intelligence model can be trained according to the sample spatio-temporal attribute features and the risk regions to obtain the risk region identification model, the identification processing effect of the risk region identification model is effectively ensured, and whether the risk region identification model trained to converge is a risk region of a hazardous article is identified according to the cross-domain spatio-temporal attribute features by using the risk region identification model, so that the accuracy of hazardous article risk region identification can be effectively improved.

The risk area marking refers to a risk area mark obtained after marking a risk area corresponding to the sample space-time data.

The initial artificial intelligence model refers to an untrained artificial intelligence model with a feature analysis processing function.

In the embodiment of the disclosure, when determining multiple risk area labels respectively corresponding to multiple pieces of sample residing area information, whether the multiple areas corresponding to the sample residing area information are risk areas or not can be labeled according to a spot check result of a service worker, the corresponding multiple residing areas are divided into risk areas and non-risk areas, and the risk areas and the non-risk areas are correspondingly labeled, so as to determine the multiple risk area labels respectively corresponding to the multiple pieces of sample residing area information.

After determining the multiple risk area labels respectively corresponding to the multiple sample resident area information, the disclosed embodiment can determine multiple sample space-time attribute characteristics aligned according to the sample resident area information, then can construct an initial artificial intelligence model, then can train the initial artificial intelligence model according to the aligned multiple sample space-time attribute characteristics and the multiple risk area labels, can input the aligned multiple sample space-time attribute characteristics and the multiple risk area labels into the constructed initial artificial intelligence model, train the artificial intelligence model, iteratively update model parameters until the artificial intelligence model converges, and use the artificial intelligence model trained to converge as a risk area identification model.

Optionally, in some embodiments, the initial artificial intelligence model comprises: the cross-domain space-time attribute recognition model, the full-connection layer connected with the cross-domain space-time attribute recognition model and the classification layer connected with the full-connection layer, so that an initial artificial intelligence model can be constructed based on the cross-domain space-time attribute recognition model, and the cross-domain space-time attribute recognition model is trained to be convergent, so that the training cost of the artificial intelligence model can be reduced, the training efficiency of training the artificial intelligence model to be convergent to obtain a risk area recognition model is improved, and the risk area recognition processing efficiency of hazardous chemicals is improved in an auxiliary mode.

In the embodiment of the disclosure, when an initial artificial intelligence model is constructed, a cross-domain space-time attribute identification model, a full connection layer connected with the cross-domain space-time attribute identification model, and a classification layer connected with the full connection layer may be obtained respectively, the cross-domain space-time attribute identification model, the full connection layer connected with the cross-domain space-time attribute identification model, and the classification layer connected with the full connection layer are constructed as the initial artificial intelligence model, and then the initial artificial intelligence model may be trained according to aligned multiple sample space-time attribute features and multiple risk region labels until the artificial intelligence model converges, and the artificial intelligence model obtained by training is used as a risk region identification model.

S610: area information of the area to be identified is determined.

S611: and inputting the region information into the cross-domain space-time attribute identification model to obtain cross-domain space-time attribute characteristics which are output by the cross-domain space-time attribute identification model and correspond to the region information.

For an example, the description of S610-S611 may refer to the above embodiments, which are not described herein again.

S612: and inputting the cross-domain space-time attribute characteristics into a risk region identification model trained in advance, and determining whether the region to be identified is a dangerous chemical risk region or not based on the output of the risk region identification model.

According to the method and the device for identifying the dangerous chemical substance risk area, after an artificial intelligence model trained to be convergent is used as a risk area identification model, cross-domain space-time attribute features can be input into the risk area identification model trained in advance, the cross-domain space-time attribute features of the area to be identified are identified and processed based on the risk area identification model, so that an output result of the risk area identification model is obtained, and then whether the area to be identified is a dangerous chemical substance risk area or not can be determined based on the output result of the risk area identification model.

For example, as shown in fig. 7, fig. 7 is a schematic diagram of a hazardous chemical substance risk region identification process in the embodiment of the present disclosure, an initial artificial intelligence model including a cross-domain spatiotemporal attribute identification model, a full connection layer connected to the cross-domain spatiotemporal attribute identification model, and a classification layer connected to the full connection layer may be constructed, the initial artificial intelligence model is trained to converge to obtain a risk region identification model, region information is encoded to obtain cross-domain spatiotemporal attribute features, and then the cross-domain spatiotemporal attribute features are input to a risk region identification model trained in advance to determine whether a region to be identified is a hazardous chemical substance risk region based on output of the risk region identification model.

In the embodiment, by determining the area information of the area to be identified, cross-domain space-time attribute features corresponding to the area information are obtained, wherein the cross-domain space-time attribute features represent attribute association features between first space-time data and second space-time data of the area to be identified, the first space-time data correspond to a first data domain, the second space-time data correspond to a second data domain, the first data domain is different from the second data domain, and according to the cross-domain space-time attribute features, whether the area to be identified is a hazardous article risk area is determined, the cross-domain space-time attribute features corresponding to the area to be identified can be analyzed, effective utilization of different data domain space-time data is achieved, whether the area to be identified is a hazardous article risk area is effectively judged by using the cross-domain space-time attribute features, so that the identification efficiency of the hazardous article risk area identification can be effectively improved, the identification accuracy of the hazardous article risk area is effectively improved, an artificial intelligent model is labeled according to the sample space-time attribute features and the risk area, so that the artificial intelligent model is constructed based on the cross-domain space-time-space attribute, the artificial model identification efficiency is improved, and the artificial risk model is effectively ensured, the identification efficiency of the risk area identification model, the risk area identification is improved, and the artificial model identification efficiency is improved.

Fig. 8 is a schematic structural diagram of a device for identifying a hazardous chemical risk area according to an embodiment of the disclosure.

As shown in fig. 8, the hazardous chemical substance risk area identification device 80 includes:

a first determining module 801, configured to determine area information of an area to be identified;

an obtaining module 802, configured to obtain cross-domain spatio-temporal attribute features corresponding to the region information, where the cross-domain spatio-temporal attribute features represent attribute association features between first spatio-temporal data and second spatio-temporal data of a region to be identified, where the first spatio-temporal data corresponds to a first data domain, the second spatio-temporal data corresponds to a second data domain, and the first data domain is different from the second data domain; and

and a second determining module 803, configured to determine whether the region to be identified is a dangerous chemical risk region according to the cross-domain spatiotemporal attribute feature.

In some embodiments of the present disclosure, the obtaining module 802 is specifically configured to:

inputting the region information into a cross-domain space-time attribute recognition model to obtain cross-domain space-time attribute characteristics which are output by the cross-domain space-time attribute recognition model and correspond to the region information, wherein the cross-domain space-time attribute recognition model is obtained by adopting a federal machine learning method for training.

In some embodiments of the present disclosure, as shown in fig. 9, fig. 9 is a schematic structural diagram of a hazardous chemical substance risk area identification device according to another embodiment of the present disclosure, and further includes:

a third determining module 804, configured to determine a plurality of residence points in the region to be identified before inputting the region information into the cross-domain spatio-temporal attribute identification model to obtain cross-domain spatio-temporal attribute features corresponding to the region information output by the cross-domain spatio-temporal attribute identification model;

an identifying module 805, configured to identify, according to the multiple residence points, residence area information from the area to be identified;

a processing module 806, configured to use the parking area information as the area information.

In some embodiments of the present disclosure, further comprising:

a building module 807 for building an initial federal machine learning model before determining the regional information of the region to be identified;

the training module 808 is used for training an initial federal machine learning model by adopting a federal machine learning method and combining a plurality of sample space-time attribute characteristics respectively output by a plurality of federal comparison submodels and a plurality of loss cost values respectively corresponding to the plurality of federal comparison submodels until the federal machine learning model converges, and taking the obtained federal machine learning model as a cross-domain space-time attribute recognition model;

In some embodiments of the present disclosure, the training module 808 is specifically configured to:

respectively inputting the plurality of sample space-time attribute characteristics into an initial federated machine learning model to obtain sample characteristic representation information output by the federated machine learning model;

determining a loss value to be processed according to the sample characteristic characterization information, and determining a target loss value according to the loss value to be processed and a plurality of loss cost values;

and if the target loss value meets the loss condition, taking the Federal machine learning model obtained by training as a cross-domain space-time attribute identification model.

In some embodiments of the present disclosure, among other things, the training module 808 is further configured to:

processing the plurality of sample space-time attribute features to obtain a positive sample space-time attribute feature pair, wherein the positive sample space-time attribute feature pair comprises: sample space-time attribute features corresponding to trajectory vectors belonging to the same dwell region;

processing the plurality of sample space-time attribute features to obtain a negative sample space-time attribute feature pair, wherein the negative sample space-time attribute feature pair comprises: sample spatio-temporal attribute features corresponding to trajectory vectors belonging to different dwell regions;

inputting the positive sample space-time attribute feature pairs and the negative sample space-time attribute feature pairs into an initial federated machine learning model to obtain positive sample feature characterization distances between the positive sample space-time attribute feature pairs and negative sample feature characterization distances between the negative sample space-time attribute feature pairs, which are output by the federated machine learning model;

and taking the positive sample characteristic characterization distance and the negative sample characteristic characterization distance as sample characteristic characterization information.

before processing a plurality of sample time-space attribute characteristics to obtain a positive sample time-space attribute characteristic pair, obtaining a plurality of sample resident area information respectively output by a plurality of federal contrast submodels;

and aligning the space-time attribute characteristics of the plurality of samples according to the resident area information of the samples.

In some embodiments of the present disclosure, the second determining module 803 is specifically configured to:

and inputting the cross-domain space-time attribute characteristics into a risk region identification model trained in advance, and determining whether the region to be identified is a dangerous chemical risk region or not based on the output of the risk region identification model.

after an initial federal machine learning model is trained by combining a plurality of time-space attribute characteristics respectively output by a plurality of federal contrast submodels and a plurality of loss cost values respectively corresponding to the plurality of federal contrast submodels by adopting a federal machine learning method, a plurality of risk region labels respectively corresponding to a plurality of sample resident region information are determined;

determining a plurality of sample space-time attribute characteristics aligned according to the sample residence area information;

and training an initial artificial intelligence model according to the aligned space-time attribute characteristics of the multiple samples and the multiple risk areas until the artificial intelligence model is converged, and taking the artificial intelligence model obtained by training as a risk area identification model.

In some embodiments of the present disclosure, wherein the initial artificial intelligence model comprises: the system comprises a cross-domain space-time attribute identification model, a full connection layer connected with the cross-domain space-time attribute identification model and a classification layer connected with the full connection layer.

Corresponding to the hazardous chemical substance risk area identification method provided in the embodiments of fig. 1 to 7, the present disclosure also provides a hazardous chemical substance risk area identification device, and since the hazardous chemical substance risk area identification device provided in the embodiments of the present disclosure corresponds to the hazardous chemical substance risk area identification method provided in the embodiments of fig. 1 to 7, the implementation manner of the hazardous chemical substance risk area identification method is also applicable to the hazardous chemical substance risk area identification device provided in the embodiments of the present disclosure, and will not be described in detail in the embodiments of the present disclosure.

In order to implement the foregoing embodiment, the present disclosure further provides a computer device, including: the identification method comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein when the processor executes the program, the identification method for the dangerous chemical risk area is realized.

In order to achieve the above embodiments, the present disclosure also proposes a non-transitory computer readable storage medium, on which a computer program is stored, which when executed by a processor, implements the hazardous chemical risk area identification method as proposed by the foregoing embodiments of the present disclosure.

In order to implement the foregoing embodiments, the present disclosure further provides a computer program product, which when executed by an instruction processor in the computer program product, performs the method for identifying a risk region of a hazardous chemical substance according to the foregoing embodiments of the present disclosure.

FIG. 10 illustrates a block diagram of an exemplary computer device suitable for use in implementing embodiments of the present disclosure. Thecomputer device 12 shown in fig. 10 is only one example and should not impose any limitations on the functionality or scope of use of embodiments of the present disclosure.

As shown in FIG. 10,computer device 12 is in the form of a general purpose computing device. The components ofcomputer device 12 may include, but are not limited to: one or more processors orprocessing units 16, asystem memory 28, and abus 18 that couples various system components including thesystem memory 28 and theprocessing unit 16.

Bus 18 represents one or more of any of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, and a processor or local bus using any of a variety of bus architectures. These architectures include, but are not limited to, industry Standard Architecture (ISA) bus, micro Channel Architecture (MAC) bus, enhanced ISA bus, video Electronics Standards Association (VESA) local bus, and Peripheral Component Interconnect (PCI) bus, to name a few.

Computer device 12 typically includes a variety of computer system readable media. Such media may be any available media that is accessible bycomputer device 12 and includes both volatile and nonvolatile media, removable and non-removable media.

Memory 28 may include computer system readable media in the form of volatile Memory, such as Random Access Memory (RAM) 30 and/orcache Memory 32. Thecomputer device 12 may further include other removable/non-removable, volatile/nonvolatile computer system storage media. By way of example only,storage system 34 may be used to read from and write to non-removable, nonvolatile magnetic media (not shown in FIG. 10, and commonly referred to as a "hard drive").

Although not shown in FIG. 10, a disk drive for reading from and writing to a removable, nonvolatile magnetic disk (e.g., a "floppy disk") and an optical disk drive for reading from or writing to a removable, nonvolatile optical disk (e.g., a Compact disk Read Only Memory (CD-ROM), a Digital versatile disk Read Only Memory (DVD-ROM), or other optical media) may be provided. In these cases, each drive may be connected tobus 18 by one or more data media interfaces.Memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to carry out the functions of embodiments of the disclosure.

A program/utility 40 having a set (at least one) ofprogram modules 42 may be stored, for example, inmemory 28,such program modules 42 including but not limited to an operating system, one or more application programs, other program modules, and program data, each of which or some combination of which may comprise an implementation of a network environment.Program modules 42 generally perform the functions and/or methodologies of the embodiments described in this disclosure.

Computer device 12 may also communicate with one or more external devices 14 (e.g., keyboard, pointing device,display 24, etc.), with one or more devices that enable a user to interact withcomputer device 12, and/or with any devices (e.g., network card, modem, etc.) that enablecomputer device 12 to communicate with one or more other computing devices. Such communication may be through an input/output (I/O)interface 22. Moreover,computer device 12 may also communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public Network such as the Internet) viaNetwork adapter 20. As shown,network adapter 20 communicates with the other modules ofcomputer device 12 viabus 18. It should be understood that although not shown in the figures, other hardware and/or software modules may be used in conjunction withcomputer device 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, to name a few.

Theprocessing unit 16 executes various functional applications and data processing by executing programs stored in thesystem memory 28, for example, implementing the hazardous chemical risk area identification method mentioned in the foregoing embodiments.

Other embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed herein. This disclosure is intended to cover any variations, uses, or adaptations of the disclosure following, in general, the principles of the disclosure and including such departures from the present disclosure as come within known or customary practice within the art to which the disclosure pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.

It will be understood that the present disclosure is not limited to the precise arrangements that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

It should be noted that, in the description of the present disclosure, the terms "first", "second", etc. are used for descriptive purposes only and are not to be construed as indicating or implying relative importance. Further, in the description of the present disclosure, "a plurality" means two or more unless otherwise specified.

Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing specific logical functions or steps of the process, and the scope of the preferred embodiments of the present disclosure includes other implementations in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved, as would be understood by those reasonably skilled in the art of the embodiments of the present disclosure.

It should be understood that portions of the present disclosure may be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, various steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any one or combination of the following technologies, which are well known in the art, may be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application specific integrated circuit having an appropriate combinational logic gate circuit, a Programmable Gate Array (PGA), a Field Programmable Gate Array (FPGA), or the like.

It will be understood by those skilled in the art that all or part of the steps carried by the method for implementing the above embodiments may be implemented by hardware related to instructions of a program, which may be stored in a computer readable storage medium, and when the program is executed, the program includes one or a combination of the steps of the method embodiments.

In addition, functional units in the embodiments of the present disclosure may be integrated into one processing module, or each unit may exist alone physically, or two or more units are integrated into one module. The integrated module can be realized in a hardware mode, and can also be realized in a software functional module mode. The integrated module, if implemented in the form of a software functional module and sold or used as a separate product, may also be stored in a computer-readable storage medium.

The storage medium mentioned above may be a read-only memory, a magnetic or optical disk, etc.

In the description herein, references to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present disclosure. In this specification, the schematic representations of the terms used above do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

Although embodiments of the present disclosure have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present disclosure, and that changes, modifications, substitutions and alterations may be made to the above embodiments by those of ordinary skill in the art within the scope of the present disclosure.

Claims

1. A method for identifying a dangerous chemical risk area is characterized by comprising the following steps:

determining area information of an area to be identified;

acquiring cross-domain space-time attribute characteristics corresponding to the area information, wherein the cross-domain space-time attribute characteristics represent attribute association characteristics between first space-time data and second space-time data of the area to be identified, the first space-time data corresponds to a first data domain, the second space-time data corresponds to a second data domain, and the first data domain is different from the second data domain; and

and determining whether the region to be identified is a dangerous chemical risk region or not according to the cross-domain space-time attribute characteristics.

2. The method as claimed in claim 1, wherein said obtaining cross-domain spatiotemporal attribute features corresponding to said region information comprises:

and inputting the region information into a cross-domain space-time attribute recognition model to obtain cross-domain space-time attribute characteristics which are output by the cross-domain space-time attribute recognition model and correspond to the region information, wherein the cross-domain space-time attribute recognition model is obtained by adopting a federal machine learning method for training.

3. The method as claimed in claim 2, wherein before said inputting said region information into a cross-domain spatio-temporal attribute identification model to obtain a cross-domain spatio-temporal attribute feature corresponding to said region information output by said cross-domain spatio-temporal attribute identification model, further comprising:

determining a plurality of residence points in the area to be identified;

identifying resident area information from the area to be identified according to the plurality of resident points;

and taking the resident area information as the area information.

4. The method of claim 2, further comprising, prior to the determining the region information for the region to be identified:

constructing an initial federated machine learning model;

training the initial federal machine learning model by adopting a federal machine learning method and combining a plurality of sample space-time attribute characteristics respectively output by a plurality of federal comparison submodels and a plurality of loss cost values respectively corresponding to the plurality of federal comparison submodels until the federal machine learning model converges, and taking the obtained federal machine learning model as the cross-domain space-time attribute recognition model;

the federated comparison submodel is obtained by local training in a corresponding data domain and is used for processing sample spatio-temporal data provided by the corresponding data domain to obtain sample spatio-temporal attribute characteristics corresponding to the sample spatio-temporal data.

5. The method of claim 4, wherein the training of the initial federated machine learning model using a federated machine learning method in combination with a plurality of spatio-temporal attribute features output by a plurality of federated comparison submodels, respectively, a plurality of loss cost values corresponding to the plurality of federated comparison submodels, respectively, comprises:

respectively inputting the plurality of sample space-time attribute characteristics into the initial federated machine learning model to obtain sample characteristic characterization information output by the federated machine learning model;

determining a loss value to be processed according to the sample characteristic characterization information, and determining a target loss value according to the loss value to be processed and the loss cost values;

and if the target loss value meets the loss condition, taking the federal machine learning model obtained by training as the cross-domain space-time attribute recognition model.

6. The method as claimed in claim 5, wherein said inputting said plurality of sample spatiotemporal attribute features into said initial federated machine learning model to obtain sample feature characterization information output by said federated machine learning model comprises:

processing the plurality of sample spatiotemporal attribute features to obtain a negative sample spatiotemporal attribute feature pair, wherein the negative sample spatiotemporal attribute feature pair comprises: sample space-time attribute features corresponding to trajectory vectors belonging to different dwell regions;

inputting the positive sample space-time attribute feature pairs and the negative sample space-time attribute feature pairs into the initial federated machine learning model to obtain positive sample feature characterization distances between the positive sample space-time attribute feature pairs and negative sample feature characterization distances between the negative sample space-time attribute feature pairs, which are output by the federated machine learning model;

and taking the positive sample characteristic feature characterization distance and the negative sample characteristic feature characterization distance as the sample characteristic characterization information.

7. The method of claim 5, wherein prior to said processing said plurality of sample spatiotemporal attribute features to obtain a positive sample spatiotemporal attribute feature pair, further comprising:

obtaining a plurality of sample residing area information respectively output by the plurality of federal contrast submodels;

and aligning the plurality of sample space-time attribute features according to the sample residence area information.

8. The method of claim 7, wherein the determining whether the region to be identified is a hazardous chemical risk region according to the cross-domain spatiotemporal attribute features comprises:

inputting the cross-domain space-time attribute features into a risk region identification model trained in advance, and determining whether the region to be identified is a dangerous chemical risk region or not based on the output of the risk region identification model.

9. The method of claim 8, wherein after the training the initial federated machine learning model using the federated machine learning method in combination with the plurality of spatio-temporal attribute features output by the plurality of federated comparison submodels, respectively, the plurality of loss cost values corresponding to the plurality of federated comparison submodels, respectively, further comprises:

determining a plurality of risk region labels respectively corresponding to the plurality of sample resident region information;

determining a plurality of sample spatio-temporal attribute characteristics aligned according to the sample residence area information;

and training an initial artificial intelligence model according to the aligned space-time attribute characteristics of the multiple samples and the multiple risk region labels until the artificial intelligence model is converged, and taking the artificial intelligence model obtained by training as the risk region identification model.

10. The method of claim 9, wherein the initial artificial intelligence model comprises: the cross-domain space-time attribute identification model, a full connection layer connected with the cross-domain space-time attribute identification model and a classification layer connected with the full connection layer.

11. An apparatus for identifying a hazardous chemical risk area, the apparatus comprising:

the first determining module is used for determining the area information of the area to be identified;

the acquisition module is used for acquiring cross-domain space-time attribute characteristics corresponding to the area information, wherein the cross-domain space-time attribute characteristics represent attribute association characteristics between first space-time data and second space-time data of the area to be identified, the first space-time data corresponds to a first data domain, the second space-time data corresponds to a second data domain, and the first data domain is different from the second data domain; and

and the second determining module is used for determining whether the region to be identified is a dangerous chemical risk region or not according to the cross-domain space-time attribute characteristics.

12. The apparatus of claim 11, wherein the obtaining module is specifically configured to:

13. The apparatus of claim 12, further comprising:

a third determining module, configured to determine a plurality of residence points in the region to be identified before the region information is input into a cross-domain spatio-temporal attribute identification model to obtain a cross-domain spatio-temporal attribute feature corresponding to the region information output by the cross-domain spatio-temporal attribute identification model;

the identification module is used for identifying resident area information from the area to be identified according to the plurality of resident points;

and the processing module is used for taking the resident area information as the area information.

14. The apparatus of claim 12, further comprising:

the building module is used for building an initial federal machine learning model before determining the regional information of the region to be identified;

the training module is used for training the initial federal machine learning model by adopting a federal machine learning method and combining a plurality of sample space-time attribute characteristics output by a plurality of federal comparison submodels respectively and a plurality of loss cost values corresponding to the plurality of federal comparison submodels respectively until the federal machine learning model converges, and taking the federal machine learning model obtained by training as the cross-domain space-time attribute recognition model;

15. The apparatus of claim 14, wherein the training module is specifically configured to:

and if the target loss value meets the loss condition, taking the trained federated machine learning model as the cross-domain spatiotemporal attribute identification model.

16. The apparatus of claim 15, wherein the training module is further configured to:

processing the plurality of sample spatiotemporal attribute features to obtain a negative sample spatiotemporal attribute feature pair, wherein the negative sample spatiotemporal attribute feature pair comprises: sample spatio-temporal attribute features corresponding to trajectory vectors belonging to different dwell regions;

inputting the positive sample spatiotemporal attribute feature pairs and the negative sample spatiotemporal attribute feature pairs into the initial federated machine learning model to obtain positive sample feature characterization distances between the positive sample spatiotemporal attribute feature pairs and negative sample feature characterization distances between the negative sample spatiotemporal attribute feature pairs, which are output by the federated machine learning model;

and taking the positive sample characteristic feature characterization distance and the negative sample characteristic feature characterization distance as the sample characteristic feature characterization information.

17. The apparatus of claim 15, wherein the training module is further configured to:

before the plurality of sample time-space attribute features are processed to obtain positive sample time-space attribute feature pairs, obtaining a plurality of sample resident area information respectively output by the plurality of federal contrast submodels;

18. The apparatus of claim 17, wherein the second determining module is specifically configured to:

19. The apparatus of claim 18, wherein the training module is further configured to:

after the initial federal machine learning model is trained by combining a plurality of time-space attribute characteristics respectively output by a plurality of federal comparison submodels and a plurality of loss cost values respectively corresponding to the plurality of federal comparison submodels by adopting a federal machine learning method, a plurality of risk region labels respectively corresponding to the information of the plurality of sample residence regions are determined;

and marking and training an initial artificial intelligence model according to the aligned space-time attribute characteristics of the plurality of samples and the plurality of risk areas until the artificial intelligence model is converged, and taking the artificial intelligence model obtained by training as the risk area identification model.

20. The apparatus of claim 19, wherein the initial artificial intelligence model comprises: the cross-domain space-time attribute identification model, a full connection layer connected with the cross-domain space-time attribute identification model and a classification layer connected with the full connection layer.

21. A computer device, comprising:

at least one processor; and

a memory communicatively coupled to the at least one processor; wherein,

the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-10.

22. A non-transitory computer readable storage medium having stored thereon computer instructions for causing a computer to perform the method of any one of claims 1-10.