Data prediction model training, photovoltaic inverter fault prediction method and device

Abstract

The application discloses a data prediction model training method, a photovoltaic inverter fault prediction method and device, and the data prediction model training method comprises the following steps: acquiring a data set of historical state quantities of a photovoltaic inverter, wherein each historical state quantity comprises phase voltage, phase current, line voltage, line current and IGBT temperature of each phase; taking the phase voltage, the phase current, the line voltage and the line current in each historical state quantity as features, and taking the IGBT temperature of each phase in each historical state quantity as a label value; training a preset data prediction model by using the data set, obtaining target weights of the data prediction model, and obtaining a trained data prediction model. The data prediction model training method, the photovoltaic inverter fault prediction method and the device can effectively improve the generalization ability of the model, thereby effectively improving the accuracy of the predicted data, and making the photovoltaic inverter fault prediction more accurate.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of photovoltaic power generation, in particular to a data prediction model training method and photovoltaic inverter fault prediction method and device. BACKGROUND

[0002] Photovoltaic energy is a clean energy, and with more and more photovoltaic power generation systems being integrated into the power grid around the world, effective maintenance of the equipment is an important task at present. The photovoltaic inverter is the core equipment in the photovoltaic power generation system, and effective analysis of its state is related to the safe and stable operation of the entire photovoltaic power generation system.

[0003] At present, most of the photovoltaic inverter fault prediction methods are based on real-time monitoring of state quantities to predict data, and use the prediction data to predict the fault of the photovoltaic inverter. However, the training method of the existing data prediction model for the photovoltaic inverter is relatively simple, which leads to poor generalization ability of the model, and further leads to inaccurate prediction data, so that the fault prediction of the photovoltaic inverter is also inaccurate. SUMMARY

[0004] The purpose of the present application is to provide a data prediction model training method, photovoltaic inverter fault prediction method, device, computer equipment and storage medium, which can effectively improve the generalization ability of the model, and further improve the accuracy of the prediction data, so that the fault prediction of the photovoltaic inverter is more accurate.

[0005] In order to achieve the above purpose, in a first aspect, the present application provides a data prediction model training method, comprising:

[0006] Obtaining a data set of historical state quantities of a photovoltaic inverter, each of the historical state quantities comprising phase voltage, phase current, line voltage, line current and IGBT temperature of each phase;

[0007] Taking the phase voltage, phase current, line voltage and line current in each of the historical state quantities as features, and taking the IGBT temperature of each phase in each of the historical state quantities as a label value;

[0008] Training a preset data prediction model using the data set to obtain a target weight of the data prediction model, and obtaining a trained data prediction model.

[0009] In a preferred embodiment of the present application, the training of the preset data prediction model using the data set to obtain the target weight of the data prediction model comprises:

[0010] Determining two target sub-data sets conforming to a preset dispersion from the data set;

[0011] The data prediction model is trained by using the two target sub-data sets, and target weights of the data prediction model are obtained.

[0012] In the preferred embodiment of the present application, the two target sub-data sets meeting the preset dispersion are determined from the data set, comprising:

[0013] The data set is repeatedly sampled by using a preset sampling method and a preset proportion, and a plurality of groups of sampling data sets are obtained, each group of the sampling data sets comprising two sub-data sets;

[0014] The dispersion of each group of the sampling data sets is calculated respectively, and a target sampling data set meeting the preset dispersion is determined, and the two sub-data sets of the target sampling data set are determined as the two target sub-data sets.

[0015] In the preferred embodiment of the present application, the data prediction model is trained by using the two target sub-data sets, and target weights of the data prediction model are obtained, comprising:

[0016] The two target sub-data sets are respectively taken as source domain data and target domain data;

[0017] The source domain data and the target domain data are input into a preset data prediction model for training, and target weights of the data prediction model are obtained.

[0018] In the preferred embodiment of the present application, the expression of the loss function used by the data prediction model is as follows:

[0019]

[0020] wherein, LOSS is a loss value, D s is source domain feature data, y s is source domain label data, D t is target domain feature data, λ∈[0, +∞) is a hyperparameter, and n is the number of the source domain feature data D s .

[0021] In a second aspect, the present application provides a photovoltaic inverter fault prediction method, comprising:

[0022] Obtaining real-time state quantities of a photovoltaic inverter, wherein the real-time state quantities include phase voltage, phase current, line voltage, line current and IGBT temperature of each phase;

[0023] Taking the phase voltage, phase current, line voltage and line current in the real-time state quantities as features, and taking the IGBT temperature of each phase in the real-time state quantities as a label value;

[0024] obtain the IGBT temperature at the next moment by using the real-time state quantity and the data prediction model obtained by the data prediction model training method;

[0025] perform fault prediction on the photovoltaic inverter according to the IGBT temperature at the next moment to obtain a fault prediction result.

[0026] In a third aspect, the present application provides a data prediction model training device, comprising:

[0027] a data acquisition module configured to acquire a data set of historical state quantities of a photovoltaic inverter, each of the historical state quantities comprising phase voltage, phase current, line voltage, line current and IGBT temperature of each phase;

[0028] a data processing module configured to take the phase voltage, phase current, line voltage and line current in each of the historical state quantities as features and take the IGBT temperature of each phase in each of the historical state quantities as a label value;

[0029] a training module configured to train a preset data prediction model by using the data set to obtain target weights of the data prediction model and obtain the trained data prediction model.

[0030] In a fourth aspect, the present application provides a photovoltaic inverter fault prediction device, comprising:

[0031] a data acquisition module configured to acquire real-time state quantities of a photovoltaic inverter, the real-time state quantities comprising phase voltage, phase current, line voltage, line current and IGBT temperature of each phase;

[0032] a data processing module configured to take the phase voltage, phase current, line voltage and line current in the real-time state quantities as features and take the IGBT temperature of each phase in the real-time state quantities as a label value;

[0033] a data prediction module configured to obtain the IGBT temperature at the next moment by using the real-time state quantity and the data prediction model obtained by the data prediction model training device;

[0034] a fault prediction module configured to perform fault prediction on the photovoltaic inverter according to the IGBT temperature at the next moment to obtain a fault prediction result.

[0035] In a fifth aspect, the present application provides a computer device, comprising a memory and a processor, the memory being configured to store a computer program, and the processor being configured to run the computer program to enable the computer device to perform the data prediction model training method or the photovoltaic inverter fault prediction method.

[0036] In a sixth aspect, the present application provides a computer readable storage medium storing a computer program, which, when executed by a processor, implements the data prediction model training method or the photovoltaic inverter fault prediction method described above.

[0037] The data prediction model training method, the photovoltaic inverter fault prediction method, the device, the computer device and the storage medium provided by the embodiments of the present application have the beneficial effects that, compared with the prior art:

[0038] The present application more clearly divides the historical state quantity of the photovoltaic inverter by taking the phase voltage, the phase current, the line voltage and the line current in each historical state quantity as features and taking the temperature of each phase IGBT in each historical state quantity as a label value, can more effectively utilize the historical state quantity of the photovoltaic inverter, and then reuse the data set to train the preset data prediction model to obtain the target weight of the data prediction model and obtain the trained data prediction model, which can effectively improve the generalization ability of the model, thereby effectively improving the accuracy of the predicted data and making the fault prediction of the photovoltaic inverter more accurate. BRIEF DESCRIPTION OF DRAWINGS

[0039] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiments of the present application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope, and for those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0040] Figure 1 is a flowchart of the data prediction model training method provided by the first embodiment of the present application;

[0041] Figure 2 is a structure block diagram of the data prediction model training device provided by the second embodiment of the present application;

[0042] Figure 3 is a flowchart of the photovoltaic inverter fault prediction method provided by the third embodiment of the present application;

[0043] Figure 4 is a structure block diagram of the photovoltaic inverter fault prediction device provided by the fourth embodiment of the present application;

[0044] Figure 5 is a schematic diagram of the internal structure of the computer device provided by the fifth embodiment of the present application. DETAILED DESCRIPTION

[0045] The specific embodiments of the present application will be further described in detail below with reference to the accompanying drawings and examples. The following examples are used to illustrate the present application, but are not used to limit the scope of the present application.

[0046] At present, most of the fault prediction methods of photovoltaic inverters are based on real-time monitoring of state quantities to predict data, and the predicted data is used for fault prediction of photovoltaic inverters. However, the training methods of the existing data prediction models for photovoltaic inverters are relatively simple, which leads to poor generalization ability of the model, and further leads to inaccurate prediction data, so that the fault prediction of the photovoltaic inverter is not accurate.

[0047] In view of the problems in the prior art, the embodiments of the present application provide a data prediction model training method, a photovoltaic inverter fault prediction method, a device, a computer device and a storage medium, which can effectively improve the generalization ability of the model, and further effectively improve the accuracy of the prediction data, so that the fault prediction of the photovoltaic inverter is more accurate.

[0048] Embodiment one

[0049] Referring to Figure 1 , Figure 1 is a flowchart of the data prediction model training method provided by the embodiments of the present application.

[0050] The data prediction model training method described in the embodiments of the present application can be applied to a computer device such as a server.

[0051] In one embodiment, the present application provides a data prediction model training method, comprising the following steps:

[0052] Step S110, obtaining a data set of historical state quantities of a photovoltaic inverter, each historical state quantity comprising phase voltage, phase current, line voltage, line current and temperature of each phase IGBT.

[0053] In one embodiment, the data prediction model trained by the data prediction model training method of the present application is used for data prediction of a photovoltaic inverter.

[0054] In one embodiment, the computer device can obtain the data set of historical state quantities of the photovoltaic inverter through historical data collected by the sensor; the temperature of each phase IGBT (Insulated Gate Bipolar Transistor, Insulated Gate Bipolar Transistor) is the temperature of each phase Insulated Gate Bipolar Transistor; and optionally, the data set of historical state quantities of the photovoltaic inverter is a time series data set.

[0055] Step S120, taking the phase voltage, phase current, line voltage and line current in each historical state quantity as a feature, and taking the temperature of each phase IGBT in each historical state quantity as a label value.

[0056] In one embodiment, the computer device can use a sliding window method, taking the phase voltage, phase current, line voltage, line current in each historical state quantity as a feature, and taking each phase IGBT temperature in each historical state quantity as a label value.

[0057] In step S130, the preset data prediction model is trained using the data set to obtain the target weight of the data prediction model and obtain the trained data prediction model.

[0058] In one embodiment, the preset data prediction model uses a neural network model.

[0059] In one embodiment, when the computer device trains the preset data prediction model using the data set to obtain the target weight of the data prediction model, the computer device can:

[0060] Two target sub-data sets that meet the preset divergence are determined from the data set.

[0061] The preset data prediction model is trained using the two target sub-data sets to obtain the target weight of the data prediction model.

[0062] In this embodiment, the preset divergence can be the maximum KL divergence.

[0063] In one embodiment, the above method can greatly reduce the data used for training the data prediction model, greatly reduce the training time of the data prediction model, and the two target sub-data sets that meet the preset divergence are more meaningful and effective for training. The data prediction model can be trained better while reducing the data used for training the data prediction model, and the model training efficiency is improved.

[0064] In this embodiment, when the computer device determines two target sub-data sets that meet the preset divergence from the data set, the computer device can:

[0065] The data set is repeatedly sampled using a preset sampling method and a preset proportion to obtain multiple groups of sampled data sets, each group of sampled data sets including two sub-data sets.

[0066] The divergence of each group of sampled data sets is calculated respectively to determine a target sampled data set that meets the preset divergence, and the two sub-data sets of the target sampled data set are determined as the two target sub-data sets.

[0067] In this embodiment, the computer device can use a Monte Carlo sampling method to repeatedly sample the data set according to a preset proportion to obtain multiple groups of sampled data sets, each group of sampled data sets including two sub-data sets.

[0068] In this embodiment, it is understandable that the dispersion of each set of sampling data set is calculated, that is, the dispersion of two sub-data sets in each set of sampling data set is calculated.

[0069] In this embodiment, the data set can be divided more simply, quickly and reasonably by the above-mentioned manner, and the sub-data set and the target sub-data set can be obtained more reasonably.

[0070] In this embodiment, when the computer device trains the preset data prediction model by using the two target sub-data sets to obtain the target weight of the data prediction model, the computer device can:

[0071] The two target sub-data sets are respectively taken as source domain data and target domain data.

[0072] The source domain data and the target domain data are input to the preset data prediction model for training to obtain the target weight of the data prediction model.

[0073] In this embodiment, the two target sub-data sets are respectively taken as the source domain data and the target domain data and input to the preset data prediction model for training, so that the training effect of the data prediction model is better.

[0074] In this embodiment, the data prediction model can include a feedforward network and a loss function, where optionally, the feedforward network can be a GRU-Attention network; optionally, the loss function adopted by the data prediction model has the following expression:

[0075]

[0076] where LOSS is a loss value, D s is source domain feature data, y s is source domain label data, D t is target domain feature data, λ∈[0,+∞) is a hyperparameter, and n is the number of source domain feature data D s .

[0077] The above data prediction model training method takes the phase voltage, the phase current, the line voltage and the line current in each historical state quantity as the feature and takes each phase IGBT temperature in each historical state quantity as the label value by using the obtained data set of the historical state quantity of the photovoltaic inverter, so that the data division of the historical state quantity of the photovoltaic inverter is clearer, the historical state quantity of the photovoltaic inverter can be more effectively utilized, the data set is then used to train the preset data prediction model to obtain the target weight of the data prediction model, and the trained data prediction model is obtained, the generalization ability of the model can be effectively improved, the accuracy of the predicted data can be effectively improved, the fault prediction of the photovoltaic inverter is more accurate, and the generalization ability of the model can be effectively improved.

[0078] Embodiment two

[0079] In order to implement the method corresponding to the above-mentioned embodiments, and achieve the corresponding functions and technical effects, a data prediction model training device is provided below.

[0080] Referring to Figure 2 , Figure 2 is a structural block diagram of the data prediction model training device provided by the embodiments of the present application.

[0081] In one embodiment, the data prediction model training device of the present application comprises:

[0082] The acquisition module 210 is configured to acquire a data set of historical state quantities of the photovoltaic inverter, each historical state quantity including phase voltage, phase current, line voltage, line current, and IGBT temperature of each phase.

[0083] The data processing module 220 is configured to take the phase voltage, phase current, line voltage, and line current in each historical state quantity as features, and take the IGBT temperature of each phase in each historical state quantity as a label value.

[0084] The training module 230 is configured to train a preset data prediction model using the data set, obtain a target weight of the data prediction model, and obtain a trained data prediction model.

[0085] The data prediction model training device described above, by acquiring the data set of the historical state quantities of the photovoltaic inverter, taking the phase voltage, phase current, line voltage, and line current in each historical state quantity as features, and taking the IGBT temperature of each phase in each historical state quantity as a label value, more clearly divides the data of the historical state quantities of the photovoltaic inverter, which can more effectively utilize the historical state quantities of the photovoltaic inverter, and then train the preset data prediction model using the data set, obtain the target weight of the data prediction model, and obtain the trained data prediction model, which can effectively improve the generalization ability of the model, thereby effectively improving the accuracy of the predicted data and making the fault prediction of the photovoltaic inverter more accurate.

[0086] In one embodiment, the training module 230 can be specifically configured to:

[0087] determine two target sub-data sets that meet a preset dispersion from the data set;

[0088] train the preset data prediction model using the two target sub-data sets to obtain a target weight of the data prediction model.

[0089] In this embodiment, when the training module 230 determines the two target sub-data sets that meet the preset dispersion from the data set, the training module 230 can:

[0090] The preset sampling method and the preset proportion are repeatedly used to sample the data set, and a plurality of groups of sampling data sets are obtained, each group of sampling data sets comprising two sub-data sets;

[0091] The dispersion of each group of sampling data sets is calculated respectively, a target sampling data set meeting a preset dispersion is determined, and the two sub-data sets of the target sampling data set are determined as two target sub-data sets.

[0092] In the embodiment, when the training module 230 trains the preset data prediction model by using the two target sub-data sets to obtain a target weight of the data prediction model, the training module 230 can:

[0093] The two target sub-data sets are respectively taken as source domain data and target domain data.

[0094] The source domain data and the target domain data are input to the preset data prediction model for training to obtain the target weight of the data prediction model.

[0095] The data prediction model training device described above can implement the data prediction model training method described above. The specific limitations of the data prediction model training device described above and the remaining content can be referred to the content of the data prediction model training method described above, and will not be described in detail in the embodiments.

[0096] Embodiment three

[0097] Referring to Figure 3 , Figure 3 is a flowchart of a photovoltaic inverter fault prediction method provided by the embodiment of the application.

[0098] The photovoltaic inverter fault prediction method described above can be applied to a computer device such as a server.

[0099] In one embodiment, the application provides a photovoltaic inverter fault prediction method, comprising the following steps:

[0100] In step S310, real-time state quantities of the photovoltaic inverter are obtained, the real-time state quantities comprising phase voltage, phase current, line voltage, line current and temperature of each phase IGBT.

[0101] In one embodiment, the computer device can obtain the real-time state quantities of the photovoltaic inverter through a sensor; the temperature of each phase IGBT (Insulated Gate Bipolar Transistor, IGBT) is the temperature of each phase IGBT.

[0102] In step S320, the phase voltage, the phase current, the line voltage and the line current in the real-time state quantities are taken as features, and the temperature of each phase IGBT in the real-time state quantities is taken as a label value.

[0103] In one embodiment, the computer device can adopt a sliding window method, taking the phase voltage, the phase current, the line voltage, and the line current in the real-time state quantity as features, and taking each phase IGBT temperature in the real-time state quantity as a label value.

[0104] In step S330, the IGBT temperature at the next moment is obtained by using the real-time state quantity and the data prediction model.

[0105] It can be understood that the data prediction model is the data prediction model obtained by training in the above-mentioned embodiment one.

[0106] It can be understood that the next moment mentioned above is the next moment of the moment corresponding to the real-time state quantity.

[0107] For the data prediction model, please refer to the content in the above-mentioned embodiment one, which will not be described in detail in the embodiment.

[0108] In step S340, the photovoltaic inverter is fault-predicted according to the IGBT temperature at the next moment, and a fault prediction result is obtained.

[0109] In one embodiment, the computer device can perform fault prediction on the photovoltaic inverter according to whether the IGBT temperature at the next moment is greater than a preset temperature, and obtain a fault prediction result; it can be understood that when the IGBT temperature at the next moment is greater than the preset temperature, a fault prediction result of a photovoltaic inverter fault is obtained; and when the IGBT temperature at the next moment is not greater than the preset temperature, a fault prediction result of a normal photovoltaic inverter is obtained.

[0110] In one embodiment, the photovoltaic inverter fault prediction method of the present application can further include the following steps:

[0111] According to the IGBT temperature at the next moment and the IGBT temperature in the real-time state quantity at the next moment, the target weight of the data prediction model is updated online.

[0112] The photovoltaic inverter fault prediction method uses the obtained real-time state quantity of the photovoltaic inverter and the data prediction model in the above-mentioned embodiment one, which has stronger generalization ability and higher accuracy of predicted data, to more accurately predict the IGBT temperature at the next moment, and thus the fault prediction of the photovoltaic inverter can be more accurate.

[0113] Embodiment four

[0114] In order to perform the method corresponding to the above-mentioned embodiment to realize the corresponding functions and technical effects, a photovoltaic inverter fault prediction device is provided below.

[0115] Reference Figure 4 , Figure 4is a structural block diagram of a photovoltaic inverter fault prediction device provided by an embodiment of the present application.

[0116] In one embodiment, the photovoltaic inverter fault prediction device of the present application comprises:

[0117] The acquisition module 410 is configured to acquire real-time state quantities of the photovoltaic inverter, wherein the real-time state quantities include phase voltage, phase current, line voltage, line current and IGBT temperature of each phase.

[0118] The data processing module 420 is configured to take the phase voltage, the phase current, the line voltage and the line current in the real-time state quantities as features, and take the IGBT temperature of each phase in the real-time state quantities as a label value.

[0119] The data prediction module 430 is configured to obtain the IGBT temperature at the next moment by using the real-time state quantities and the data prediction model.

[0120] The fault prediction module 440 is configured to perform fault prediction on the photovoltaic inverter according to the IGBT temperature at the next moment, and obtain a fault prediction result.

[0121] The photovoltaic inverter fault prediction device can more accurately predict the IGBT temperature at the next moment by using the acquired real-time state quantities of the photovoltaic inverter and the data prediction model with stronger generalization ability and higher prediction data accuracy in the above-mentioned embodiment one, and thus can make the fault prediction of the photovoltaic inverter more accurate.

[0122] In one embodiment, the photovoltaic inverter fault prediction device of the present application can further comprise:

[0123] The online updating module is configured to perform online updating on the target weight of the data prediction model according to the IGBT temperature at the next moment and the IGBT temperature in the real-time state quantities at the next moment.

[0124] The photovoltaic inverter fault prediction device can implement the photovoltaic inverter fault prediction method described above. The specific limitations of the photovoltaic inverter fault prediction device and the remaining content can be referred to the content of the photovoltaic inverter fault prediction method described above, and will not be described herein.

[0125] Embodiment five

[0126] In one embodiment, the present application provides a computer device comprising a memory and a processor, wherein the memory is configured to store a computer program, and the processor is configured to run the computer program to enable the computer device to execute the data prediction model training method described above or the photovoltaic inverter fault prediction method described above.

[0127] Optionally, the computer device described above can be a server.

[0128] In one embodiment, the internal structure of the computer device of the present application can be as shown in Figure 5

[0129] In one embodiment, the present application provides a computer readable storage medium storing a computer program, which, when executed by a processor, implements the above-mentioned data prediction model training method or the above-mentioned photovoltaic inverter fault prediction method.

[0130] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can also be implemented in other manners. The above described device embodiments are merely illustrative, for example, the flowcharts and block diagrams in the accompanying drawings show the possible implementation architectures, functions and operation of the devices, methods and computer program products according to the embodiments of the present application. In this regard, each block in the flowcharts or block diagrams can represent a module, a program segment or a part of code, which contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur in a different order than that noted in the accompanying drawings. For example, two consecutive blocks can actually be executed substantially in parallel, and sometimes they can be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and the combination of blocks in the block diagrams and / or flowcharts, can be implemented by a dedicated hardware-based system that performs the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0131] In addition, each functional module in the various embodiments of the present application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0132] ​If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the parts that contribute to the prior art or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0133] The above merely describes the embodiments of the present application and is not used to limit the protection scope of the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application. It should be noted that similar reference numbers and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0134] The above merely describes the embodiments of the present application and is not used to limit the protection scope of the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application. It should be noted that similar reference numbers and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0135] It should be noted that, in this document, relational terms such as first and second, and the like can be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without limitation, an element preceded by "comprises... a" does not, without more constraints, foreclose the existence of additional identical elements in the process, method, article, or apparatus that comprises the element.

Claims

1. A data prediction model training method, characterized in that, include: Obtain a dataset of historical state parameters of the photovoltaic inverter, each of which includes phase voltage, phase current, line voltage, line current and temperature of each phase IGBT; The phase voltage, phase current, line voltage, and line current in each of the historical state quantities are used as features, and the temperature of each phase IGBT in each of the historical state quantities is used as a tag value. The dataset is used to train a preset data prediction model to obtain the target weights of the data prediction model and to obtain the trained data prediction model. The step of training a preset data prediction model using the dataset to obtain the target weights of the data prediction model includes: From the dataset, determine two target sub-datasets that meet a preset dispersion. The preset data prediction model is trained using the two target subsets to obtain the target weights of the data prediction model; The step of training a preset data prediction model using two target subsets to obtain the target weights of the data prediction model includes: The two target subsets are used as source domain data and target domain data, respectively. Input the source domain data and the target domain data into a preset data prediction model for training, and obtain the target weights of the data prediction model; The loss function used in the data prediction model is expressed as follows: in, The loss value. For source domain prediction data, For source domain label data, Predict data for the target domain. For hyperparameters, Predicting data for the source domain The quantity.

2. The data prediction model training method according to claim 1, characterized in that, The step of determining two target subsets of the dataset that meet a preset dispersion includes: The dataset is repeatedly sampled using a preset sampling method and a preset ratio to obtain multiple sets of sampled datasets, each set of sampled datasets including two sub-datasets; Calculate the dispersion of each of the sampled datasets, determine the target sampled dataset that meets the preset dispersion, and determine the two sub-datasets of the target sampled dataset as two target sub-datasets.

3. A method for predicting faults in a photovoltaic inverter, characterized in that, include: The real-time status parameters of the photovoltaic inverter are obtained, including phase voltage, phase current, line voltage, line current and the temperature of each phase IGBT. The phase voltage, phase current, line voltage, and line current in the real-time state quantities are used as features; The IGBT temperature at the current moment is obtained by using the real-time state variables and the data prediction model training method according to any one of claims 1 to 2. Based on the current IGBT temperature, the photovoltaic inverter is used to predict faults and obtain the fault prediction results.

4. A data prediction model training device, characterized in that, include: The acquisition module is used to acquire a dataset of historical state quantities of the photovoltaic inverter, each of which includes phase voltage, phase current, line voltage, line current and temperature of each phase IGBT. The data processing module is used to take the phase voltage, phase current, line voltage, and line current in each of the historical state quantities as features, and take the temperature of each phase IGBT in each of the historical state quantities as a tag value. The training module is used to train a preset data prediction model using the dataset, obtain the target weights of the data prediction model, and obtain the trained data prediction model. The step of training a preset data prediction model using the dataset to obtain the target weights of the data prediction model includes: From the dataset, determine two target sub-datasets that meet a preset dispersion. The preset data prediction model is trained using the two target subsets to obtain the target weights of the data prediction model; The step of training a preset data prediction model using two target subsets to obtain the target weights of the data prediction model includes: The two target subsets are used as source domain data and target domain data, respectively. Input the source domain data and the target domain data into a preset data prediction model for training, and obtain the target weights of the data prediction model; The loss function used in the data prediction model is expressed as follows: in, The loss value. For source domain prediction data, For source domain label data, Predict data for the target domain. For hyperparameters, Predicting data for the source domain The quantity.

5. A photovoltaic inverter fault prediction device, characterized in that, include: The acquisition module is used to acquire the real-time status parameters of the photovoltaic inverter, including phase voltage, phase current, line voltage, line current and the temperature of each phase IGBT. The data processing module is used to take the phase voltage, phase current, line voltage, and line current in the real-time state quantities as features; The data prediction module is used to obtain the IGBT temperature at the current moment using the real-time state quantity and the data prediction model obtained by the data prediction model training device according to claim 4. The fault prediction module is used to predict faults in the photovoltaic inverter based on the current IGBT temperature and obtain the fault prediction result.

6. A computer device, characterized in that, The device includes a memory and a processor, wherein the memory stores a computer program and the processor runs the computer program to enable the computer device to perform the data prediction model training method according to any one of claims 1 to 2, or the photovoltaic inverter fault prediction method according to claim 3.

7. A computer-readable storage medium, characterized in that, It stores a computer program, which, when executed by a processor, implements the data prediction model training method as described in any one of claims 1 to 2, or the photovoltaic inverter fault prediction method as described in claim 3.

Images