CN114151293A - Fault early warning method, system, equipment and storage medium of fan variable pitch system - Google Patents

Abstract

The invention discloses a fault early warning method and device for a fan variable pitch system, terminal equipment and a computer storage medium. The method comprises the following steps: acquiring real-time operation parameters of a current frequency converter to be pre-warned, and determining the working environment of a fan variable pitch system to which the frequency converter belongs; calling a target fault early warning model corresponding to the working environment from preset fault early warning models according to the working environment, wherein the fault early warning models are obtained by performing federal learning training respectively based on characteristic parameters of fan variable pitch systems in the corresponding working environment; the real-time operation parameters are arranged as model input of a target fault early warning model, and a fault prediction result of a frequency converter in a working environment, which is output after the target fault early warning model is calculated based on the real-time operation parameters, is obtained; and carrying out fault early warning on the frequency converter according to the fault prediction result. The invention can effectively improve the accuracy of fault early warning on the frequency converter of the variable pitch system.

Description

Fault early warning method, system, equipment and storage medium of fan variable pitch system

Technical Field

The invention relates to the technical field of wind motors, in particular to a fault early warning method and system for a fan pitch system, terminal equipment and a computer storage medium.

Background

In a wind turbine generator, a pitch system is an important and key component, but because the wind turbine generator is in a complex working environment for a long time, the fault of the pitch system becomes one of the most common faults in the fault of the wind turbine generator, the integral shutdown of a fan can be caused once the pitch system fails, the generated energy is seriously influenced, meanwhile, other equipment of the fan can be damaged in different degrees, and the fault which is most likely to occur in the pitch system is the fault of a frequency converter of the pitch system.

Although there are many schemes for performing fault early warning on the variable-pitch system frequency converter in the prior art, for example, fault prediction schemes based on Boosting algorithms such as GBDT (Gradient Boosting Decision Tree, an iterative Decision Tree algorithm, also called MART (Multiple Additive Regression Tree) and XGBoost (eXtreme Gradient Boosting), the various existing schemes not only need to collect a large amount of sample data to perform a Boosting training process of the algorithm, but also are difficult to adapt to variable-pitch fault early warning of a fan in a complex working environment due to the limitation of the finally trained algorithm model based on the sample data.

In summary, in the existing scheme for performing fault early warning on the variable pitch system frequency converter, the process for algorithm training is complex and the limitation of complex working environment of the fan is difficult to break through, so that fault early warning on the variable pitch system frequency converter of the fan in various environments is difficult to achieve.

Disclosure of Invention

The invention mainly aims to provide a fault early warning method, a fault early warning system, terminal equipment and a computer storage medium for a fan pitch system, and aims to solve the technical problems that in the existing fault early warning scheme, the process for algorithm training is complex, the limit of the complex working environment of a fan is difficult to break through, and therefore fault early warning is difficult to accurately carry out on a frequency converter of the fan pitch system in various environments.

In order to achieve the above object, the present invention provides a fault early warning method for a fan pitch system, which is applied to early warning a fault of a frequency converter of the fan pitch system, and comprises the following steps:

acquiring real-time operation parameters of a current frequency converter to be pre-warned, and determining the working environment of a fan variable pitch system to which the frequency converter belongs;

calling a target fault early warning model corresponding to the working environment from preset fault early warning models according to the working environment, wherein each fault early warning model is obtained by performing federal learning training based on characteristic parameters of each fan variable pitch system in the corresponding working environment;

the real-time operation parameters are arranged as model input of the target fault early warning model, and a fault prediction result of the frequency converter in the working environment, which is output after the target fault early warning model is calculated based on the real-time operation parameters, is obtained;

and carrying out fault early warning on the frequency converter according to the fault prediction result.

Preferably, the fault pre-warning method further includes:

and carrying out federal learning training according to the characteristic parameters of each fan variable pitch system in the corresponding working environment to obtain each fault early warning model, and carrying out associated storage on each fault early warning model and each fan variable pitch system in the corresponding working environment.

Preferably, the step of performing federal learning training according to the characteristic parameters of each fan pitch system in the corresponding working environment to obtain each fault early warning model includes:

s1: extracting characteristic parameters of each fan variable pitch system in a corresponding working environment from a preset distributed file system, wherein the characteristic parameters are associated with the frequency converter of the fan variable pitch system;

s2: classifying the characteristic parameters according to the working environment, and respectively training each initial machine learning model by using the classified characteristic parameters to obtain model parameters;

s3: integrating the model parameters to form a fault early warning model to be confirmed;

s4: verifying whether the fault early warning model to be confirmed meets a fault prediction condition or not through preset test sample data, wherein the fault prediction condition is as follows: the fault early warning model to be confirmed has a fault prediction accuracy reaching a preset threshold value based on the test sample data;

s5: and if so, carrying out tuning treatment on the to-be-confirmed fault early warning model according to the working environment to obtain the fault early warning model corresponding to each working environment.

Preferably, after the step of verifying whether the to-be-confirmed fault early warning model meets the fault prediction condition through preset test sample data, the method further includes:

s6: if not, taking the to-be-confirmed fault early warning model as a new initial machine learning model, and sequentially executing the steps S2, S3 and S4 until the to-be-confirmed fault early warning model is verified to meet the fault prediction condition.

Preferably, the fault pre-warning method further includes:

collecting characteristic parameters of each fan variable pitch system in a corresponding working environment;

performing data integration, data cleaning and data conversion on the characteristic parameters, and acquiring process parameters for performing data integration, data cleaning and data conversion on the characteristic parameters;

and coding according to the process parameters to obtain coded data, and dispersedly storing the characteristic parameters, the coded data and metadata obtained by performing data integration, data cleaning and data conversion on the characteristic parameters in a preset distributed file system.

Preferably, the step of extracting the characteristic parameters from the distributed file system includes:

acquiring the coded data from the distributed file system;

parsing the encoded data to extract the process parameters encapsulated in the encoded data;

and determining the characteristic parameters of the metadata mapping based on the process parameters, and acquiring the characteristic parameters.

In addition, in order to achieve the above object, the present invention further provides a fault early warning device for a fan pitch system, where the fault early warning device for the fan pitch system is applied to perform fault early warning for a frequency converter of the fan pitch system, and the fault early warning device includes:

the system comprises an acquisition module, a pre-warning module and a warning module, wherein the acquisition module is used for acquiring real-time operation parameters of a current frequency converter to be pre-warned and determining the working environment of a fan variable pitch system to which the frequency converter belongs;

the model calling module is used for calling a target fault early warning model corresponding to the working environment from preset fault early warning models according to the working environment, wherein each fault early warning model is obtained by performing federal learning training based on characteristic parameters of each fan variable pitch system in the corresponding working environment;

the model prediction module is used for sorting the real-time operation parameters into model input of the target fault early warning model and acquiring a fault prediction result of the frequency converter in the working environment, which is output after the target fault early warning model is calculated based on the real-time operation parameters;

and the early warning module is used for carrying out fault early warning on the frequency converter according to the fault prediction result.

Preferably, the fault early warning device further includes:

and the federal model training module is used for performing federal learning training according to the characteristic parameters of each fan variable pitch system in the corresponding working environment to obtain each fault early warning model, and performing associated storage on each fault early warning model and each fan variable pitch system in the corresponding working environment.

The fault early warning method for the fan pitch system is realized when each functional module of the fault early warning device for the fan pitch system runs.

In addition, to achieve the above object, the present invention also provides a terminal device, including: the fault early warning method comprises a memory, a processor and a fault early warning program of the fan variable pitch system, wherein the fault early warning program of the fan variable pitch system is stored in the memory and can be operated on the processor, and when the fault early warning program of the fan variable pitch system is executed by the processor, the steps of the fault early warning method of the fan variable pitch system are realized.

In addition, in order to achieve the above object, the present invention further provides a computer storage medium, where a fault early warning program of a fan pitch system is stored on the computer storage medium, and when the fault early warning program of the fan pitch system is executed by a processor, the steps of the fault early warning method of the fan pitch system are implemented.

In addition, to achieve the above object, the present invention further provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the computer program implements the steps of the fault early warning method for a wind turbine pitch system as described above.

The invention provides a fault early warning method, a fault early warning device, terminal equipment, a computer storage medium and a computer program product for a fan pitch system, wherein the fault early warning method is applied to fault early warning for a frequency converter of the fan pitch system, and comprises the steps of acquiring real-time operation parameters of the frequency converter to be early warned currently and determining the working environment of the fan pitch system to which the frequency converter belongs; calling a target fault early warning model corresponding to the working environment from preset fault early warning models according to the working environment, wherein each fault early warning model is obtained by performing federal learning training based on characteristic parameters of each fan variable pitch system in the corresponding working environment; the real-time operation parameters are arranged as model input of the target fault early warning model, and a fault prediction result of the frequency converter in the working environment, which is output after the target fault early warning model is calculated based on the real-time operation parameters, is obtained; and carrying out fault early warning on the frequency converter according to the fault prediction result.

Compared with the traditional fault prediction scheme based on the lifting algorithms such as GBDT and XGboost, the fault early warning method based on the frequency converter and the like breaks through the limitation of the complex working environment to algorithm model training in the prior art by carrying out federal learning training on the characteristic parameters of each fan pitch system in the working environment of the fan pitch system to which the frequency converter belongs in advance, so that when fault early warning is carried out on the current frequency converter to be early warned, the real-time operation parameters of the frequency converter and the target fault early warning model corresponding to the working environment to which the frequency converter belongs are used for carrying out model calculation, and therefore, the accurate fault prediction result of the frequency converter in the working environment can be obtained for carrying out fault early warning, and the accuracy of fault early warning on the frequency converter of the pitch system is effectively improved.

Drawings

Fig. 1 is a schematic device structure diagram of a hardware operating environment of a terminal device according to an embodiment of the present invention;

FIG. 2 is a schematic flow chart of an embodiment of a fault early warning method of a fan pitch system according to the present invention;

FIG. 3 is a schematic view of an application scenario of a fan pitch system according to an embodiment of the fault early warning method for the fan pitch system of the present invention;

fig. 4 is a functional module schematic diagram of an embodiment of the fault early warning device of the fan pitch system of the present invention.

The implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.

Detailed Description

It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

Referring to fig. 1, fig. 1 is a schematic device structure diagram of a terminal device hardware operating environment according to an embodiment of the present invention.

The terminal device of the embodiment of the invention can be a terminal device for performing fault early warning on a frequency converter of a fan pitch system, and the terminal device can be a server, a smart phone, a Personal Computer (PC), a tablet Computer, a portable Computer and the like.

As shown in fig. 1, the terminal device may include: aprocessor 1001, such as a CPU, acommunication bus 1002, auser interface 1003, anetwork interface 1004, and amemory 1005. Wherein acommunication bus 1002 is used to enable connective communication between these components. Theuser interface 1003 may include a Display screen (Display), an input unit such as a Keyboard (Keyboard), and theoptional user interface 1003 may also include a standard wired interface, a wireless interface. Thenetwork interface 1004 may optionally include a standard wired interface, a wireless interface (e.g., a Wi-Fi interface). Thememory 1005 may be a high-speed RAM memory or a non-volatile memory (e.g., a magnetic disk memory). Thememory 1005 may alternatively be a storage device separate from theprocessor 1001.

Those skilled in the art will appreciate that the terminal device configuration shown in fig. 1 is not intended to be limiting of the terminal device and may include more or fewer components than those shown, or some components may be combined, or a different arrangement of components.

As shown in fig. 1, amemory 1005, which is a kind of computer storage medium, may include therein an operating system, a network communication module, a user interface module, and a fault warning program of a wind turbine pitch system.

In the terminal shown in fig. 1, thenetwork interface 1004 is mainly used for connecting to a backend server and performing data communication with the backend server; theuser interface 1003 is mainly used for connecting a client and performing data communication with the client; and theprocessor 1001 may be configured to invoke the fault early warning program of the wind turbine pitch system stored in thememory 1005, and execute the following embodiments of the fault early warning method of the wind turbine pitch system.

Based on the hardware structure, the invention provides various embodiments of the fault early warning method for the fan variable pitch system.

It should be noted that, in this embodiment, because the wind turbine generator is in a complex working environment for a long time, developing the wind turbine generator fault early warning can effectively reduce the maintenance cost of the wind turbine generator, improve the availability of the wind turbine generator, and then improve the economic benefits of the wind power generation project. As shown in fig. 3, a pitch system of a wind turbine is a key device of a wind turbine, and can control the balance between the power and the rotation speed of the wind turbine under different working conditions, and a fault of the pitch system is one of the most common faults in the faults of the wind turbine.

At present, researches on fault early warning of a variable pitch system of a wind turbine generator mainly focus on analyzing SCAdA system data and effectively applying a machine learning algorithm to the aspects of fault early warning of the variable pitch system of the wind turbine generator.

The frequency converter of the variable pitch system is a component with the highest fault occurrence rate in the variable pitch system, and in order to reduce the faults of the variable pitch system caused by the faults of the frequency converter of the variable pitch system, the fault early warning of the frequency converter of the variable pitch system is very necessary to research.

Although there are many schemes for performing fault early warning on the variable-pitch system frequency converter in the prior art, for example, fault prediction schemes based on lifting algorithms such as GBDT and XGBoost, the existing various schemes not only need to collect a large amount of sample data to perform a lifting training process of the algorithm, but also are difficult to adapt to variable-pitch fault early warning of a fan in a complex working environment due to the limitation of the finally trained algorithm model based on the sample data.

In summary, in the existing scheme for performing fault early warning on the variable pitch system frequency converter, the process for algorithm training is complex and the limitation of complex working environment of the fan is difficult to break through, so that fault early warning on the variable pitch system frequency converter of the fan in various environments is difficult to achieve.

Aiming at the phenomenon, the invention provides a fault early warning method for a fan variable pitch system. Referring to fig. 2, fig. 2 is a schematic flow chart of a first embodiment of a fault early warning method for a fan pitch system of the present invention, and in this embodiment, the fault early warning method for the fan pitch system is applied to the terminal device. It should be noted that, although a logical order is shown in the flow chart, in some cases, the steps shown or described may be performed in an order different than that shown or described herein.

In this embodiment, the fault early warning method for the fan pitch system of the present invention includes:

step S10, acquiring real-time operation parameters of a current frequency converter to be pre-warned, and determining the working environment of a fan variable pitch system to which the frequency converter belongs;

in this embodiment, in the process of performing real-time early warning on a frequency converter in a pitch system, a terminal device first obtains real-time operation parameters of the current frequency converter to be early warned, and simultaneously determines a working environment of a fan pitch system to which the frequency converter belongs.

It should be noted that, in this embodiment, the terminal device may specifically detect a working environment in which the entire fan of the fan pitch system to which the frequency converter is disposed is located, and determine the working environment as the working environment of the frequency converter.

Step S20, calling a target fault early warning model corresponding to the working environment from preset fault early warning models according to the working environment, wherein each fault early warning model is obtained by performing federal learning training based on characteristic parameters of each fan variable pitch system in the corresponding working environment;

in this embodiment, after obtaining the real-time operating parameters of the frequency converter to be warned and determining the working environment of the frequency converter, the terminal device further extracts and calls a target fault warning model corresponding to the working environment from a database storing fault warning models obtained by performing federal learning training in advance based on the characteristic parameters of each fan pitch system in different working environments.

Step S30, the real-time operation parameters are arranged as model input of the target fault early warning model, and a fault prediction result of the frequency converter in the working environment, which is output after the target fault early warning model is calculated based on the real-time operation parameters, is obtained;

in this embodiment, after the terminal device further extracts a target fault early warning model corresponding to a working environment where the frequency converter to be early warned is located, the real-time operation parameters of the frequency converter are arranged to be model input of the target fault early warning model, so that the real-time operation parameters are input into the target fault early warning model for model training calculation, and then the terminal device further obtains a fault prediction result of the frequency converter in the working environment where the frequency converter is located, where the frequency converter is output after the target fault early warning model performs the model training calculation according to the input real-time operation parameters;

and step S40, carrying out fault early warning on the frequency converter according to the fault prediction result.

In this embodiment, after the terminal device obtains the target fault early warning model, performs model training calculation according to the input real-time operation parameters, and outputs a fault prediction result of the frequency converter in the working environment where the frequency converter is located, if the fault prediction result indicates that the probability that the frequency converter may have a fault in the working environment where the frequency converter is located is greater than a preset threshold, the terminal device immediately performs fault early warning on the frequency converter, so that corresponding staff can maintain the frequency converter.

In this embodiment, in the fault early warning method for the fan pitch system, in the process of performing real-time early warning on the frequency converter in the pitch system through the terminal device, the real-time operation parameters of the current frequency converter to be early warned are firstly acquired, and meanwhile, the working environment of the fan pitch system to which the frequency converter belongs is determined. After acquiring real-time operation parameters of a frequency converter to be subjected to early warning and determining the working environment of the frequency converter, the terminal equipment further extracts and calls a target fault early warning model corresponding to the working environment from a database storing a fault early warning model obtained by carrying out federal learning training based on characteristic parameters of each fan pitch system in different working environments in advance, arranges the real-time operation parameters of the frequency converter into model input of the target fault early warning model, inputs the real-time operation parameters into the target fault early warning model to carry out model training calculation, and then further acquires a fault prediction result of the frequency converter in the working environment, wherein the frequency converter is output after the target fault early warning model carries out model training calculation according to the input real-time operation parameters; and if the fault prediction result shows that the probability of possible fault occurrence of the frequency converter in the working environment is greater than a preset threshold value, immediately performing fault early warning on the frequency converter by the terminal equipment so as to maintain the frequency converter by corresponding staff.

Compared with the traditional fault prediction scheme based on the lifting algorithms such as GBDT and XGboost, the fault early warning method based on the frequency converter and the like breaks through the limitation of the complex working environment to algorithm model training in the prior art by carrying out federal learning training on the characteristic parameters of each fan pitch system in the working environment of the fan pitch system to which the frequency converter belongs in advance, so that when fault early warning is carried out on the current frequency converter to be early warned, the real-time operation parameters of the frequency converter and the target fault early warning model corresponding to the working environment to which the frequency converter belongs are used for carrying out model calculation, and therefore, the accurate fault prediction result of the frequency converter in the working environment can be obtained for carrying out fault early warning, and the accuracy of fault early warning on the frequency converter of the pitch system is effectively improved.

Further, in a feasible embodiment, the fault early warning method for the fan pitch system of the present invention may further include:

and carrying out federal learning training according to the characteristic parameters of each fan variable pitch system in the corresponding working environment to obtain each fault early warning model, and carrying out associated storage on each fault early warning model and each fan variable pitch system in the corresponding working environment.

In the embodiment, the terminal equipment performs federal learning training according to characteristic parameters of each fan variable pitch system in a corresponding working environment to obtain each fault early warning model according to the following steps. Namely:

s1: extracting characteristic parameters of each fan variable pitch system in a corresponding working environment from a preset distributed file system, wherein the characteristic parameters are associated with the frequency converter of the fan variable pitch system;

s2: classifying the characteristic parameters according to the working environment, and respectively training each initial machine learning model by using the classified characteristic parameters to obtain model parameters;

s3: integrating the model parameters to form a fault early warning model to be confirmed;

s4: verifying whether the fault early warning model to be confirmed meets a fault prediction condition or not through preset test sample data, wherein the fault prediction condition is as follows: the fault early warning model to be confirmed has a fault prediction accuracy reaching a preset threshold value based on the test sample data;

s5: and if so, carrying out tuning treatment on the to-be-confirmed fault early warning model according to the working environment to obtain the fault early warning model corresponding to each working environment.

In the embodiment, in the process of obtaining each fault early warning model by the terminal device performing federal learning training according to the characteristic parameters of each fan pitch system in the respective working environment, the terminal device extracts the characteristic parameters of each fan pitch system associated with the frequency converter in the system from a distributed file system which is preset and stores the characteristic parameters of each fan pitch system in the respective working environment, then the terminal device classifies all the characteristic parameters by taking each working environment as one type according to the respective working environment of each graded pitch system, so as to obtain the characteristic parameters associated with the frequency converter in the respective system of one or more fan pitch systems with the same working environment, and further takes the characteristic parameters corresponding to each working environment as model training samples respectively, and training the initial machine learning models corresponding to the working environments, and thus obtaining model parameters of each converged initial machine learning model until the initial machine learning models corresponding to each working environment are trained and converged.

And then, the terminal equipment is distinguished as different types according to different working environments, and model parameters of all initial machine learning models corresponding to each working environment are integrated to be used as a fault early warning model corresponding to the working environment. Specifically, the terminal device can integrate a plurality of model parameters of the same type into one parameter in a weighted average manner among all the model parameters, so that after only one parameter of different types is obtained by integrating all the model parameters, the initial machine learning model is filled with the parameters based on the model parameters of different types, and a fault early warning model can be obtained.

Then, after obtaining a fault early warning model corresponding to each working environment through model parameter integration, the terminal device further uses characteristic parameters of a fan pitch system with a known frequency converter fault result in each working environment as test samples respectively, so as to verify whether the fault early warning model corresponding to the working environment meets a fault prediction condition or not by using the test samples respectively, and only when the fault early warning model is based on a prediction result output by the test samples and compared with the accuracy between the known frequency converter fault results reaches a preset threshold (such as 90%), the fault early warning model is determined to meet the fault prediction condition.

Finally, the terminal device conducts model parameter adaptive adjustment and other optimization processing on the fault early warning model which meets the fault prediction condition based on the working environment, so that the fault early warning model can predict a frequency converter of a fan pitch system in the working environment more accurately, and then the terminal device stores the fault early warning model and the specific identification of the working environment in an associated mode, so that the fault early warning model which conducts fan pitch system frequency converter fault early warning and corresponds to each working environment is obtained.

Further, in an optional embodiment, in the process that the terminal device performs federal learning training to obtain each fault early warning model by using the above steps, in step S4: after the step of verifying whether the fault early warning model to be confirmed meets the fault prediction condition through preset test sample data, the method further comprises the following steps:

s6: if not, taking the to-be-confirmed fault early warning model as a new initial machine learning model, and sequentially executing the steps S2, S3 and S4 until the to-be-confirmed fault early warning model is verified to meet the fault prediction condition.

In this embodiment, if the terminal device verifies and discovers that the terminal device aims at the fault early warning model of the corresponding working environment by using the test sample, when the frequency converter fault prediction result in the fan pitch system output by the model is compared with the known frequency converter fault result, the accuracy does not reach the preset threshold, the terminal device determines that the model does not meet the fault prediction condition at present, and needs to train again.

Therefore, the terminal equipment continues to use the model as a new initial machine learning model corresponding to the working environment, and trains, integrates and verifies the new initial machine learning model again according to the process of using the characteristic parameters corresponding to the working environment as model training samples to train the initial machine learning model until the fault early warning model meets the fault early warning condition.

Further, in a feasible embodiment, the fault early warning method for the fan pitch system of the present invention may further include: and collecting characteristic parameters of each fan variable pitch system in a corresponding working environment, preprocessing the characteristic parameters, and storing the preprocessed characteristic parameters in a preset distributed file system, so that the terminal equipment can extract the characteristic parameters of each fan variable pitch system in the corresponding working environment from the preset distributed file system in the process of carrying out federal learning training by adopting the steps to obtain each fault early warning model.

It should be noted that, in this embodiment, the preprocessing operation includes: data integration, data cleaning and data conversion; when the terminal device collects characteristic parameters of each fan pitch system in a corresponding working environment and stores the characteristic parameters in a preset distributed file system after preprocessing operation is performed on the characteristic parameters, the method specifically includes:

collecting characteristic parameters of each fan variable pitch system in a corresponding working environment;

performing data integration, data cleaning and data conversion on the characteristic parameters, and acquiring process parameters for performing data integration, data cleaning and data conversion on the characteristic parameters;

and coding according to the process parameters to obtain coded data, and dispersedly storing the characteristic parameters, the coded data and metadata obtained by performing data integration, data cleaning and data conversion on the characteristic parameters in a preset distributed file system.

In this embodiment, before extracting the characteristic parameters from the distributed file system and performing federal learning training, the terminal device collects the characteristic parameters of each fan pitch system associated with the frequency converter in advance in the respective working environment of each fan pitch system, and then, the terminal device further performs data integration, data cleaning, data conversion and other operations on the characteristic parameters respectively to obtain metadata meeting the requirements of the distributed file system, and in the process, obtains process parameters for performing data integration, data clarity, data conversion and other operations on the characteristic parameters. Finally, the terminal device respectively encodes the characteristic parameters in each working environment according to the obtained process parameters and corresponding encoding rules to obtain encoded data corresponding to the characteristic parameters, and further, the terminal device dispersedly stores the characteristic data, the encoded data and the metadata in a plurality of places or a plurality of devices of a distributed file system so that subsequent terminal devices can extract the characteristic parameters to perform federal learning training and provide file services.

It should be noted that, in this embodiment, before performing operations such as data integration, data cleaning, and data conversion on parameters, the terminal device further configures an operation restriction rule for screening and classifying feature data, so that before performing each operation on the feature parameters corresponding to each working environment, it is first detected whether the feature parameters meet the rule, and only when the feature parameters meet the rule, the terminal device performs each operation and obtains process parameters of each operation.

In this embodiment, the operation restriction rule configured by the terminal device is used to filter and classify the characteristic parameters, so as to obtain the target parameters suitable for the federate learning training as the model training sample. Specifically, the configuration parameters of the operation restriction rule include: data size of characteristic parameters, file format, and the like. It should be understood that, in this embodiment, the configuration parameters of the operation restriction rule may be set according to the user's requirements, and the configuration parameters of the operation restriction rule are not specifically limited in this embodiment.

Further, in a possible embodiment, the extracting, performed by the terminal device, the characteristic parameter from the distributed file system may include:

acquiring the coded data from the distributed file system;

parsing the encoded data to extract the process parameters encapsulated in the encoded data;

and determining the characteristic parameters of the metadata mapping based on the process parameters, and acquiring the characteristic parameters.

In this embodiment, in the process of extracting the characteristic parameters from the distributed file system for federal learning training, after receiving a file service request initiated by a user through a client and needing to access the distributed file system for characteristic parameter extraction, the terminal device extracts encoded data corresponding to a tenant to which the client belongs from the distributed file system based on parsing of the file service request.

Then, the terminal device analyzes the encoded data to obtain process parameters, included in the encoded data, for performing operations such as data integration, data cleaning, data conversion and the like on the feature parameters, and then, the terminal device performs operations such as reverse data integration, data cleaning, data conversion and the like according to the process parameters, thereby determining target feature source data corresponding to the metadata. Specifically, for example, a problem of data redundancy is generated because strong correlation exists between a plurality of fields in a certain characteristic parameter or several fields can be derived from each other, and unnecessary data can be eliminated after data integration, data cleaning, data conversion and other operations, so that the data quality is improved. Therefore, when the terminal analyzes the coded data, the terminal can acquire the process parameters for reducing the field redundancy and operate based on the process parameter direction to finally return the metadata to the target characteristic parameters which are not subjected to the redundancy reduction operation.

Finally, after the terminal device returns the metadata to the target characteristic parameter by analyzing the encoded data, the terminal device can directly execute downloading operation on the target characteristic parameter on the node or the device of the distributed file system where the terminal device is located, and return the downloaded target characteristic parameter to the client initiating the file service request, so as to achieve the operation of extracting the characteristic parameter for carrying out the federal learning training by executing the file service request initiated by the user from the client.

Further, the present invention also provides a fault early warning device for a fan pitch system, which is applied to early warning a fault of a frequency converter of the fan pitch system, as shown in fig. 4, and includes:

the system comprises anacquisition module 10, a pre-warning module and a warning module, wherein the acquisition module is used for acquiring real-time operation parameters of a current frequency converter to be pre-warned and determining the working environment of a fan variable pitch system to which the frequency converter belongs;

themodel calling module 20 is configured to call a target fault early warning model corresponding to the working environment from among preset fault early warning models according to the working environment, wherein each fault early warning model is obtained by performing federal learning training based on characteristic parameters of each fan pitch system in the corresponding working environment;

themodel prediction module 30 is configured to sort the real-time operation parameters into model inputs of the target fault early warning model, and obtain a fault prediction result of the frequency converter in the working environment, where the fault prediction result is output after the target fault early warning model is calculated based on the real-time operation parameters;

and theearly warning module 40 is used for carrying out fault early warning on the frequency converter according to the fault prediction result.

Further, the fault early warning device of the fan variable pitch system of the invention further comprises:

and the federal learning module is used for performing federal learning training according to the characteristic parameters of each fan variable pitch system in the corresponding working environment to obtain each fault early warning model, and performing associated storage on each fault early warning model and each fan variable pitch system in the corresponding working environment.

Further, the federal learning module is further configured to perform the following steps:

s1: extracting characteristic parameters of each fan variable pitch system in a corresponding working environment from a preset distributed file system, wherein the characteristic parameters are associated with the frequency converter of the fan variable pitch system;

s2: classifying the characteristic parameters according to the working environment, and respectively training each initial machine learning model by using the classified characteristic parameters to obtain model parameters;

s3: integrating the model parameters to form a fault early warning model to be confirmed;

s4: verifying whether the fault early warning model to be confirmed meets a fault prediction condition or not through preset test sample data, wherein the fault prediction condition is as follows: the fault early warning model to be confirmed has a fault prediction accuracy reaching a preset threshold value based on the test sample data;

s5: and if so, carrying out tuning treatment on the to-be-confirmed fault early warning model according to the working environment to obtain the fault early warning model corresponding to each working environment.

Further, after the step of verifying whether the to-be-confirmed fault early warning model meets the fault prediction condition through preset test sample data is executed, the federal learning module is further configured to execute the following steps:

s6: if not, taking the to-be-confirmed fault early warning model as a new initial machine learning model, and sequentially executing the steps S2, S3 and S4 until the to-be-confirmed fault early warning model is verified to meet the fault prediction condition.

Further, the fault early warning device of the fan variable pitch system of the invention further comprises:

the data acquisition module is used for acquiring characteristic parameters of each fan variable pitch system in a corresponding working environment;

the data processing module is used for performing data integration, data cleaning and data conversion on the characteristic parameters and acquiring process parameters for performing data integration, data cleaning and data conversion on the characteristic parameters;

and the data storage module is used for coding according to the process parameters to obtain coded data, and dispersedly storing the characteristic parameters, the coded data and metadata obtained by performing data integration, data cleaning and data conversion on the characteristic parameters in a preset distributed file system.

Further, the fault early warning device of the fan pitch system of the invention further comprises a parameter extraction module for extracting characteristic parameters from the distributed file system, and the parameter extraction module is used for:

acquiring the coded data from the distributed file system;

parsing the encoded data to extract the process parameters encapsulated in the encoded data;

and determining the characteristic parameters of the metadata mapping based on the process parameters, and acquiring the characteristic parameters.

The function implementation of each module in the fault early warning device for the fan pitch system of the invention corresponds to each step in the embodiment of the fault early warning method for the fan pitch system of the invention, and the function and implementation process are not repeated here.

The invention also provides a computer storage medium, wherein the computer storage medium stores a fault early warning program of the fan pitch system, and the fault early warning program of the fan pitch system realizes the steps of the fault early warning method of the fan pitch system according to any one of the embodiments when being executed by a processor.

The specific embodiment of the computer storage medium of the present invention is basically the same as the embodiments of the fault early warning method for the fan pitch system, and is not described herein again.

The invention also provides a computer program product comprising a computer program which, when executed by a processor, implements the steps of the fault warning method for a wind turbine pitch system according to any of the above embodiments.

The specific embodiment of the computer program product of the present invention is basically the same as the embodiments of the fault early warning method for the fan pitch system, and is not described herein again.

The above description is only a preferred embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by using the contents of the present specification and the accompanying drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.

Claims (10)

1. A fault early warning method for a fan variable pitch system is characterized by being applied to fault early warning of a frequency converter of the fan variable pitch system, and comprising the following steps of:

acquiring real-time operation parameters of a current frequency converter to be pre-warned, and determining the working environment of a fan variable pitch system to which the frequency converter belongs;

calling a target fault early warning model corresponding to the working environment from preset fault early warning models according to the working environment, wherein each fault early warning model is obtained by performing federal learning training based on characteristic parameters of each fan variable pitch system in the corresponding working environment;

the real-time operation parameters are arranged as model input of the target fault early warning model, and a fault prediction result of the frequency converter in the working environment, which is output after the target fault early warning model is calculated based on the real-time operation parameters, is obtained;

and carrying out fault early warning on the frequency converter according to the fault prediction result.

2. The fault early warning method of claim 1, further comprising:

and carrying out federal learning training according to the characteristic parameters of each fan variable pitch system in the corresponding working environment to obtain each fault early warning model, and carrying out associated storage on each fault early warning model and each fan variable pitch system in the corresponding working environment.

3. The fault early warning method according to claim 2, wherein the step of performing federal learning training according to the characteristic parameters of each fan pitch system in the corresponding working environment to obtain each fault early warning model comprises:

s1: extracting characteristic parameters of each fan variable pitch system in a corresponding working environment from a preset distributed file system, wherein the characteristic parameters are associated with the frequency converter of the fan variable pitch system;

s2: classifying the characteristic parameters according to the working environment, and respectively training each initial machine learning model by using the classified characteristic parameters to obtain model parameters;

s3: integrating the model parameters to form a fault early warning model to be confirmed;

s4: verifying whether the fault early warning model to be confirmed meets a fault prediction condition or not through preset test sample data, wherein the fault prediction condition is as follows: the fault early warning model to be confirmed has a fault prediction accuracy reaching a preset threshold value based on the test sample data;

s5: and if so, carrying out tuning treatment on the to-be-confirmed fault early warning model according to the working environment to obtain the fault early warning model corresponding to each working environment.

4. The method of claim 3, wherein after the step of verifying whether the fault pre-warning model to be confirmed meets the fault prediction condition through preset test sample data, the method further comprises:

s6: if not, taking the to-be-confirmed fault early warning model as a new initial machine learning model, and sequentially executing the steps S2, S3 and S4 until the to-be-confirmed fault early warning model is verified to meet the fault prediction condition.

5. The fault early warning method according to any one of claims 1 to 4, further comprising:

collecting characteristic parameters of each fan variable pitch system in a corresponding working environment;

performing data integration, data cleaning and data conversion on the characteristic parameters, and acquiring process parameters for performing data integration, data cleaning and data conversion on the characteristic parameters;

and coding according to the process parameters to obtain coded data, and dispersedly storing the characteristic parameters, the coded data and metadata obtained by performing data integration, data cleaning and data conversion on the characteristic parameters in a preset distributed file system.

6. The fault early warning method for the fan pitch system according to claim 5, wherein the extracting characteristic parameters from the distributed file system comprises:

acquiring the coded data from the distributed file system;

parsing the encoded data to extract the process parameters encapsulated in the encoded data;

and determining the characteristic parameters of the metadata mapping based on the process parameters, and acquiring the characteristic parameters.

7. The utility model provides a fault early warning device of fan becomes oar system which characterized in that, fan becomes the trouble early warning device of oar system and is applied to and carries out the trouble early warning to fan becomes the converter of oar system, the fault early warning device includes:

the system comprises an acquisition module, a pre-warning module and a warning module, wherein the acquisition module is used for acquiring real-time operation parameters of a current frequency converter to be pre-warned and determining the working environment of a fan variable pitch system to which the frequency converter belongs;

the model calling module is used for calling a target fault early warning model corresponding to the working environment from preset fault early warning models according to the working environment, wherein each fault early warning model is obtained by performing federal learning training based on characteristic parameters of each fan variable pitch system in the corresponding working environment;

the model prediction module is used for sorting the real-time operation parameters into model input of the target fault early warning model and acquiring a fault prediction result of the frequency converter in the working environment, which is output after the target fault early warning model is calculated based on the real-time operation parameters;

and the early warning module is used for carrying out fault early warning on the frequency converter according to the fault prediction result.

8. The wind turbine pitch system fault early warning device of claim 7, further comprising:

and the federal learning module is used for performing federal learning training according to the characteristic parameters of each fan variable pitch system in the corresponding working environment to obtain each fault early warning model, and performing associated storage on each fault early warning model and each fan variable pitch system in the corresponding working environment.

9. A terminal device, characterized in that the terminal device comprises: the fault early warning method comprises the steps of a fault early warning method of a fan pitch system according to any one of claims 1 to 6, and a fault early warning program of the fan pitch system, wherein the fault early warning program of the fan pitch system is stored in the memory and can run on the processor.

10. A computer storage medium, characterized in that the computer storage medium stores thereon a fault warning program of a wind turbine pitch system, and the fault warning program of the wind turbine pitch system, when executed by a processor, implements the steps of the fault warning method of the wind turbine pitch system according to any one of claims 1 to 6.

Images