Data prediction model training, photovoltaic power generation system anomaly early warning method and device

Abstract

The application discloses a data prediction model training method, a photovoltaic power generation system abnormal early warning method and device, and relates to the technical field of data prediction model training. The method comprises the following steps: obtaining a target historical use state quantity of a photovoltaic inverter; training a preset data prediction model by using the target historical use state quantity to obtain an initial weight value of the data prediction model; obtaining a target real-time use state quantity of the photovoltaic inverter; predicting a next use state quantity by using the target real-time use state quantity and the data prediction model; and updating the initial weight value of the data prediction model according to the next use state quantity and the target real-time use state quantity. The data prediction model training method, the photovoltaic power generation system abnormal early warning method and device can effectively improve the generalization ability of the model, and the model can be adaptively adjusted in weight during training, thereby effectively improving the accuracy of the predicted data and making the photovoltaic power generation system abnormal early warning more accurate.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power generation, and in particular to a method and device for data prediction model training and abnormality early warning of a photovoltaic power generation system. Background Art

[0002] Solar photovoltaic power generation is gaining increasing attention due to its clean, environmentally friendly, and sustainable characteristics. With the increasing number of photovoltaic power plants, effective warning of abnormalities in photovoltaic equipment is crucial. Every link in the photovoltaic power generation system is essential, and the health of the equipment affects the overall safety and stability of the system. Timely and effective warning of abnormalities in photovoltaic power generation equipment and the establishment of pre-emptive response plans can effectively ensure the safe and stable operation of the power grid.

[0003] At present, most abnormality warning methods for photovoltaic power generation systems are based on data prediction of real-time monitored state quantities, and use the predicted data for abnormality warning. However, the existing training methods for data prediction models for photovoltaic power generation systems are relatively simple, resulting in poor generalization ability of the model and difficulty in adaptively adjusting weights, which in turn leads to inaccurate prediction data and inaccurate abnormality warnings for photovoltaic power generation systems. Summary of the Invention

[0004] The purpose of the present invention is to provide a data prediction model training, photovoltaic power generation system abnormality warning method, device, computer equipment and storage medium, which can effectively improve the generalization ability of the model, and can adaptively adjust the weights during model training, thereby effectively improving the accuracy of the predicted data, making the abnormality warning of the photovoltaic power generation system more accurate.

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

[0006] Obtain the target historical usage status of the photovoltaic inverter;

[0007] Using the target historical usage state quantity to train a preset data prediction model to obtain initial weights of the data prediction model;

[0008] Obtaining a target real-time usage state quantity of the photovoltaic inverter;

[0009] Using the target real-time usage state quantity and the data prediction model, predicting the next usage state quantity;

[0010] The initial weight of the data prediction model is updated according to the next usage state quantity and the target real-time usage state quantity.

[0011] In a preferred embodiment of the present invention, the step of obtaining the target historical usage state of the photovoltaic inverter includes:

[0012] Obtain the initial historical usage status of the photovoltaic inverter;

[0013] Using a preset data update algorithm to update the non-scalar quantity in the initial historical usage state quantity;

[0014] The updated initial historical usage state quantity is converted into a target historical usage state quantity using a preset data conversion algorithm, where the target historical usage state quantity is supervised data.

[0015] In a preferred embodiment of the present invention, the step of obtaining the target real-time usage state of the photovoltaic inverter includes:

[0016] Obtaining an initial real-time usage state quantity of the photovoltaic inverter;

[0017] When it is detected that the initial real-time usage state quantity is a standard quantity, the initial real-time usage state quantity is converted into a target real-time usage state quantity using a preset data conversion algorithm, and the target real-time usage state quantity is supervised data.

[0018] In a preferred embodiment of the present invention, the step of obtaining the target real-time usage state of the photovoltaic inverter further includes:

[0019] When it is detected that the initial real-time usage state quantity is a non-scalar quantity, using the data prediction model to predict a predicted usage state quantity corresponding to the time of the initial real-time usage state quantity;

[0020] Obtaining a previous usage state quantity of the initial real-time usage state quantity, and obtaining an updated initial real-time usage state quantity based on the previous usage state quantity and the predicted usage state quantity;

[0021] The updated initial real-time usage state quantity is converted into a target real-time usage state quantity using a preset data conversion algorithm, wherein the target real-time usage state quantity is supervised data.

[0022] In a preferred embodiment of the present invention, the error function of the data prediction model is as follows:

[0023]

[0024] in, Using state quantities for said target in real time, It is the prediction result of the usage state quantity at time t using time t-1 and the data prediction model.

[0025] In a second aspect, the present invention provides an abnormality early warning method for a photovoltaic power generation system, comprising:

[0026] Obtain the target real-time usage status of the photovoltaic inverter;

[0027] Using the target real-time usage state quantity and the above-mentioned data prediction model, predict the next usage state quantity;

[0028] When the next usage state quantity is greater than a preset abnormal warning threshold, an abnormal warning prompt is issued.

[0029] In a third aspect, the present invention provides a data prediction model training device, comprising:

[0030] A first acquisition module is used to acquire a target historical usage state quantity of the photovoltaic inverter;

[0031] A training module, configured to train a preset data prediction model using the target historical usage state quantity to obtain an initial weight of the data prediction model;

[0032] A second acquisition module is used to obtain a target real-time usage state quantity of the photovoltaic inverter;

[0033] A data prediction module, configured to predict a next usage state quantity using the target real-time usage state quantity and the data prediction model;

[0034] An online updating module is used to update the initial weight of the data prediction model according to the next usage state quantity and the target real-time usage state quantity.

[0035] In a fourth aspect, the present invention provides an abnormality warning device for a photovoltaic power generation system, comprising:

[0036] An acquisition module is used to obtain the target real-time usage status of the photovoltaic inverter;

[0037] A data prediction module is used to predict the next usage state quantity using the target real-time usage state quantity and the above-mentioned data prediction model;

[0038] The abnormality warning module is used to issue an abnormality warning prompt when the next usage state quantity is greater than a preset abnormality warning threshold.

[0039] In a fifth aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the computer device to execute the above-mentioned data prediction model training method, or the above-mentioned abnormal warning method for the photovoltaic power generation system.

[0040] In a sixth aspect, the present invention 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 abnormality warning method for the photovoltaic power generation system.

[0041] The embodiments of the present invention provide a data prediction model training and photovoltaic power generation system abnormality warning method, device, computer equipment, and storage medium. Compared with the prior art, the advantages thereof are:

[0042] The present invention first performs preliminary training on a preset data prediction model by obtaining the target historical usage state quantity of the photovoltaic inverter to obtain the initial weight of the data prediction model; then, the next usage state quantity is predicted by using the target real-time usage state quantity of the photovoltaic inverter and the data prediction model each time subsequently obtained, and the initial weight of the data prediction model is continuously updated online according to the next usage state quantity and the target real-time usage state quantity, thereby effectively improving the generalization ability of the model, and the weight can be adaptively adjusted during model training, thereby effectively improving the accuracy of the predicted data, making the abnormal warning of the photovoltaic power generation system more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments of the present invention. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0044] Figure 1 1 is a flow chart of a data prediction model training method provided in the first embodiment of the present invention;

[0045] Figure 2 This is a structural block diagram of a data prediction model training device provided in the second embodiment of the present invention;

[0046] Figure 3 1 is a flow chart of an abnormality warning method for a photovoltaic power generation system provided in a third embodiment of the present invention;

[0047] Figure 4 This is a structural block diagram of an abnormality warning device for a photovoltaic power generation system provided by a fourth embodiment of the present invention;

[0048] Figure 5 It is a schematic diagram of the internal structure of the computer device provided in Example 5 of the present invention. DETAILED DESCRIPTION

[0049] The following embodiments of the present invention are described in further detail with reference to the accompanying drawings and examples. The following examples are used to illustrate the present invention but are not intended to limit the scope of the present invention.

[0050] At present, most abnormality warning methods for photovoltaic power generation systems are based on data prediction of real-time monitored state quantities, and use the predicted data for abnormality warning. However, the existing training methods for data prediction models for photovoltaic power generation systems are relatively simple, resulting in poor generalization ability of the model and difficulty in adaptively adjusting weights, which in turn leads to inaccurate prediction data and inaccurate abnormality warnings for photovoltaic power generation systems.

[0051] In response to the above-mentioned problems in the prior art, the embodiments of the present invention provide a method, device, computer equipment and storage medium for data prediction model training and abnormal warning of photovoltaic power generation system, which can effectively improve the generalization ability of the model and adaptively adjust the weights during model training, thereby effectively improving the accuracy of the predicted data and making the abnormal warning of the photovoltaic power generation system more accurate.

[0052] Example 1

[0053] See also Figure 1 , Figure 1 It is a flowchart of the data prediction model training method provided by an embodiment of the present invention.

[0054] The data prediction model training method described below in the embodiment of the present invention can be applied to computer devices such as servers.

[0055] In one embodiment, the present invention provides a data prediction model training method, comprising the following steps:

[0056] Step S110: obtaining a target historical usage state quantity of the photovoltaic inverter.

[0057] In one embodiment, the data prediction model trained by the data prediction model training method of the present invention is used for data prediction of a photovoltaic power generation system, and the photovoltaic power generation system includes a photovoltaic inverter.

[0058] In one embodiment, when the computer device obtains the target historical usage status of the photovoltaic inverter, it may:

[0059] Obtain the initial historical usage status of the photovoltaic inverter;

[0060] Using a preset data update algorithm to update the non-scalar data in the initial historical usage state;

[0061] The updated initial historical usage state quantity is converted into the target historical usage state quantity using a preset data conversion algorithm, and the target historical usage state quantity is supervised data.

[0062] In this embodiment, the initial historical usage state quantity of the photovoltaic inverter may include data on the phase voltage, phase current, line voltage, line current and temperature of each phase IGBT (Insulated Gate Bipolar Transistor) of the photovoltaic inverter at a certain historical moment; in this embodiment, the initial historical usage state quantity of the photovoltaic inverter includes data on the phase voltage, phase current, line voltage, line current and temperature of each phase IGBT (Insulated Gate Bipolar Transistor) of the photovoltaic inverter at multiple historical moments; in this embodiment, the computer device can obtain multiple initial historical usage state quantities of the photovoltaic inverter through historical data collected by sensors.

[0063] In this embodiment, it can be understood that non-scalar quantities are non-standard quantities, and the non-scalar quantities in the initial historical usage state quantities include historical usage state quantities in which erroneous data and / or missing values ​​exist in the initial historical usage state quantities; when the computer device uses a preset data update algorithm to update the non-scalar quantities in the initial historical usage state quantities, the k-nearest neighbor algorithm can be used to update the non-scalar quantities in the initial historical usage state quantities.

[0064] In this embodiment, when the computer device uses a preset data conversion algorithm to convert the updated initial historical usage state quantity into the target historical usage state quantity, a sliding window method can be used to convert the updated initial historical usage state quantity into the target historical usage state quantity.

[0065] In one embodiment, the initial historical usage state quantity of the photovoltaic inverter obtained is converted into the target historical usage state quantity in the above manner, which can fully utilize each historical usage state quantity, make the training effect of the preset data prediction model better, and better obtain the initial weight of the data prediction model.

[0066] Step S120 , training a preset data prediction model using the target historical usage state quantity to obtain an initial weight of the data prediction model.

[0067] In one embodiment, the preset data prediction model adopts a PSO-BP neural network model, which is a BP neural network model optimized by a particle swarm algorithm (PSO).

[0068] Step S130: obtaining the target real-time usage status of the photovoltaic inverter.

[0069] In one embodiment, when the computer device obtains the target real-time usage status of the photovoltaic inverter, it may:

[0070] Obtain the initial real-time usage status of the photovoltaic inverter;

[0071] When it is detected that the initial real-time usage state quantity is a standard quantity, a preset data conversion algorithm is used to convert the initial real-time usage state quantity into a target real-time usage state quantity, where the target real-time usage state quantity is supervised data.

[0072] In this embodiment, the initial real-time usage state quantity of the photovoltaic inverter may include data on the phase voltage, phase current, line voltage, line current and temperature of each phase IGBT (Insulated Gate Bipolar Transistor) of the photovoltaic inverter at a certain moment; in this embodiment, the initial real-time usage state quantity of the photovoltaic inverter is usually the initial real-time usage state quantity of a photovoltaic inverter; in this embodiment, the computer device can obtain the initial real-time usage state quantity of the photovoltaic inverter through real-time data collected by sensors in real time.

[0073] In this embodiment, it can be understood that the standard quantity is a quantity that meets the numerical standard; when the computer device detects that the initial real-time usage state quantity is a standard quantity, the initial real-time usage state quantity can be converted into the target real-time usage state quantity using a sliding window method; in this embodiment, when the computer device detects that the initial real-time usage state quantity is a non-scalar quantity, the computer device can choose to abandon the use of the non-scalar initial real-time usage state quantity.

[0074] In one embodiment, when the initial real-time usage state quantity is detected to be a standard quantity through the above method, the initial real-time usage state quantity of the photovoltaic inverter is converted into a target real-time usage state quantity, which can improve the accuracy of the predicted next usage state quantity, thereby making the online training effect of the preset data prediction model better and better updating the initial weights of the data prediction model.

[0075] In this embodiment, when the computer device obtains the target real-time usage status of the photovoltaic inverter, it may also:

[0076] When it is detected that the initial real-time usage state quantity is a non-scalar quantity, a data prediction model is used to predict a predicted usage state quantity corresponding to the time of the initial real-time usage state quantity;

[0077] Obtaining the previous usage state quantity of the initial real-time usage state quantity, and obtaining an updated initial real-time usage state quantity based on the previous usage state quantity and the predicted usage state quantity;

[0078] The updated initial real-time usage state quantity is converted into the target real-time usage state quantity using a preset data conversion algorithm, and the target real-time usage state quantity is supervised data.

[0079] In this embodiment, when the computer device uses the data prediction model to predict the predicted usage state quantity corresponding to the time of the initial real-time usage state quantity, it can use the real-time usage state quantity at the moment before the initial real-time usage state quantity (that is, the previous usage state quantity) and the data prediction model to predict the predicted usage state quantity corresponding to the time of the initial real-time usage state quantity; in this embodiment, when the computer device obtains the updated initial real-time usage state quantity based on the previous usage state quantity and the predicted usage state quantity, the updated initial real-time usage state quantity is obtained by taking the weighted average of the previous usage state quantity and the predicted usage state quantity.

[0080] In this embodiment, the computer device may use a sliding window method to convert the updated initial real-time usage state quantity into the target real-time usage state quantity.

[0081] In this embodiment, each real-time usage status quantity can be fully utilized through the above-mentioned method, which can make the training effect of the preset data prediction model better and better update the initial weight of the data prediction model.

[0082] Step S140 , using the target real-time usage state quantity and the data prediction model, predict the next usage state quantity.

[0083] In one embodiment, the next usage state quantity may include data of the phase voltage, phase current, line voltage, line current, and temperature of each phase IGBT (Insulated Gate Bipolar Transistor) of the photovoltaic inverter at the next moment.

[0084] Step S150 : updating the initial weight of the data prediction model according to the next usage state quantity and the target real-time usage state quantity.

[0085] In one embodiment, when the computer device uses the target real-time usage state quantity and the data prediction model to predict the next usage state quantity, it also updates the initial weight of the data prediction model online based on the next usage state quantity and the target real-time usage state quantity.

[0086] In one embodiment, according to the first-order Markov law, the data at time t is highly correlated with the data at time t-1. Therefore, in order to improve the accuracy of model prediction, the error function used by the data prediction model is as follows:

[0087]

[0088] in, Use state quantities for the target in real time, It is the prediction result of the usage state quantity at time t using time t-1 and the data prediction model.

[0089] In this embodiment, when the initial weights of the data prediction model are updated online according to the next usage state quantity and the target real-time usage state quantity, the epoch of each update (indicating that all data are sent to the network and a forward calculation and back propagation process is completed) can be set to 100. At time t, the epoch of online weight update is set to 100 times, and the adaptive learning rate is set to ∈ = 0.1, the decay rate ρ = 0.09, and the constant δ is 10 -6 , set the initial value of cumulative momentum r(0) = 0, the weight matrix T(i) is the matrix value in the i-th epoch, and when i = 0, T(0) is the weight matrix W(t-1) at the previous moment; set the upper and lower limit functions of the learning rate as follows:

[0090]

[0091]

[0092] In the i-th epoch, you can:

[0093] First, calculate the gradient matrix of each layer input and output, Next, calculate the cumulative amount r(i) = ρr(i-1) + (1-ρ)g(i)⊙g(i), where r(0) is 0;

[0094] In order to prevent the second-order momentum from changing too much, the momentum clipping part is introduced:

[0095]

[0096] Weight Update Where T(0) is W(t-1);

[0097] In order to prevent overfitting, the L1 regularization constraint weight matrix and hyperparameter λ are further introduced. The formula is as follows:

[0098]

[0099] The above process is repeated until epoch 100 is reached, and the final updated weight W(t) is T(i=100).

[0100] The above-mentioned data prediction model training method first performs preliminary training on the preset data prediction model through the obtained target historical usage state quantity of the photovoltaic inverter to obtain the initial weight of the data prediction model; then, the next usage state quantity is predicted through the target real-time usage state quantity of the photovoltaic inverter and the data prediction model each time subsequently obtained, and the initial weight of the data prediction model is continuously updated online according to the next usage state quantity and the target real-time usage state quantity, thereby effectively improving the generalization ability of the model, and the weight can be adaptively adjusted during model training, thereby effectively improving the accuracy of the predicted data, making the abnormal warning of the photovoltaic power generation system more accurate.

[0101] Example 2

[0102] In order to execute the methods corresponding to the above embodiments and achieve corresponding functions and technical effects, a data prediction model training device is provided below.

[0103] See also Figure 2 , Figure 2 It is a structural block diagram of the data prediction model training device provided by an embodiment of the present invention.

[0104] In one embodiment, the data prediction model training device of the present invention includes:

[0105] A first acquisition module 210 is configured to acquire a target historical usage state quantity of the photovoltaic inverter;

[0106] The training module 220 is used to train a preset data prediction model using the target historical usage state quantity to obtain the initial weight of the data prediction model;

[0107] The second acquisition module 230 is used to obtain the target real-time usage state of the photovoltaic inverter;

[0108] The data prediction module 240 is used to predict the next usage state quantity using the target real-time usage state quantity and the data prediction model;

[0109] The online updating module 250 is used to update the initial weight of the data prediction model according to the next usage state quantity and the target real-time usage state quantity.

[0110] The above-mentioned data prediction model training device first performs preliminary training on the preset data prediction model through the obtained target historical usage state quantity of the photovoltaic inverter to obtain the initial weight of the data prediction model; then, through the target real-time usage state quantity of the photovoltaic inverter and the data prediction model obtained each time subsequently, the next usage state quantity is predicted, and the initial weight of the data prediction model is continuously updated online according to the next usage state quantity and the target real-time usage state quantity, thereby effectively improving the generalization ability of the model, and the weight can be adaptively adjusted during model training, thereby effectively improving the accuracy of the predicted data, making the abnormal warning of the photovoltaic power generation system more accurate.

[0111] In one embodiment, the first acquisition module 210 may be specifically configured to:

[0112] Obtain the initial historical usage status of the photovoltaic inverter;

[0113] Using a preset data update algorithm to update the non-scalar data in the initial historical usage state;

[0114] The updated initial historical usage state quantity is converted into the target historical usage state quantity using a preset data conversion algorithm, and the target historical usage state quantity is supervised data.

[0115] In one embodiment, the second acquisition module 230 may be specifically configured to:

[0116] Obtain the initial real-time usage status of the photovoltaic inverter;

[0117] When it is detected that the initial real-time usage state quantity is a standard quantity, a preset data conversion algorithm is used to convert the initial real-time usage state quantity into a target real-time usage state quantity, where the target real-time usage state quantity is supervised data.

[0118] In this embodiment, the second acquisition module 230 may be further configured to:

[0119] When it is detected that the initial real-time usage state quantity is a non-scalar quantity, a data prediction model is used to predict a predicted usage state quantity corresponding to the time of the initial real-time usage state quantity;

[0120] Obtaining the previous usage state quantity of the initial real-time usage state quantity, and obtaining an updated initial real-time usage state quantity based on the previous usage state quantity and the predicted usage state quantity;

[0121] The updated initial real-time usage state quantity is converted into the target real-time usage state quantity using a preset data conversion algorithm, and the target real-time usage state quantity is supervised data.

[0122] The above-mentioned data prediction model training device can implement the above-mentioned data prediction model training method. The specific limitations and other contents of the above-mentioned data prediction model training device embodiment can be found in the content of the above-mentioned data prediction model training method, and will not be repeated in the embodiment.

[0123] Example 3

[0124] See also Figure 3 , Figure 3 The figure is a flow chart of an abnormality warning method for a photovoltaic power generation system provided by an embodiment of the present invention.

[0125] The abnormality warning method for a photovoltaic power generation system described below in the embodiment of the present invention can be applied to computer equipment such as servers.

[0126] In one embodiment, the present invention provides an abnormality warning method for a photovoltaic power generation system, comprising the following steps:

[0127] Step S310: obtaining the target real-time usage state quantity of the photovoltaic inverter.

[0128] In one embodiment, when the computer device obtains the target real-time usage status of the photovoltaic inverter, it may:

[0129] Obtain the initial real-time usage status of the photovoltaic inverter;

[0130] When it is detected that the initial real-time usage state quantity is a standard quantity, a preset data conversion algorithm is used to convert the initial real-time usage state quantity into a target real-time usage state quantity, where the target real-time usage state quantity is supervised data.

[0131] In this embodiment, the initial real-time usage state quantity of the photovoltaic inverter may include data on the phase voltage, phase current, line voltage, line current and temperature of each phase IGBT (Insulated Gate Bipolar Transistor) of the photovoltaic inverter at a certain moment; in this embodiment, the initial real-time usage state quantity of the photovoltaic inverter is usually the initial real-time usage state quantity of a photovoltaic inverter; in this embodiment, the computer device can obtain the initial real-time usage state quantity of the photovoltaic inverter through real-time data collected by sensors in real time.

[0132] For other contents of step S310, please refer to the contents of the above embodiment 1, which will not be described in detail in the embodiment.

[0133] Step S320 , using the target real-time usage state quantity and the data prediction model, predict the next usage state quantity.

[0134] It can be understood that the data prediction model is the data prediction model obtained through training in the above-mentioned embodiment 1.

[0135] In one embodiment, the next usage state quantity may include data of the phase voltage, phase current, line voltage, line current, and temperature of each phase IGBT (Insulated Gate Bipolar Transistor) of the photovoltaic inverter at the next moment.

[0136] For the data prediction model, please refer to the content in the above embodiment 1, and no further details will be given in the embodiment.

[0137] Step S330: When the next usage state quantity is greater than the preset abnormal warning threshold, an abnormal warning prompt is issued.

[0138] In one embodiment, the computer device performs abnormal warning of the photovoltaic power generation system through the next usage state quantity of the photovoltaic inverter, and issues an abnormal warning prompt when the next usage state quantity of the photovoltaic inverter is greater than a preset abnormal warning threshold.

[0139] The above-mentioned abnormal warning method for the photovoltaic power generation system utilizes the acquired target real-time usage status quantity and the data prediction model with higher prediction data accuracy of the above-mentioned embodiment one to more accurately predict the next usage status quantity, thereby making the abnormal warning of the photovoltaic power generation system more accurate.

[0140] Example 4

[0141] In order to execute the methods corresponding to the above embodiments and achieve corresponding functions and technical effects, an abnormality warning device for a photovoltaic power generation system is provided below.

[0142] See also Figure 4 , Figure 4 This is a structural block diagram of an abnormality warning device for a photovoltaic power generation system provided by an embodiment of the present invention.

[0143] In one embodiment, the abnormality warning device for a photovoltaic power generation system of the present invention includes:

[0144] An acquisition module 410 is configured to acquire a target real-time usage state quantity of the photovoltaic inverter;

[0145] The data prediction module 420 is used to predict the next usage state quantity using the target real-time usage state quantity and the above-mentioned data prediction model;

[0146] The abnormality warning module 430 is used to issue an abnormality warning prompt when the next usage state quantity is greater than a preset abnormality warning threshold.

[0147] The abnormal warning device of the photovoltaic power generation system mentioned above can more accurately predict the next usage state quantity by using the acquired target real-time usage state quantity and the data prediction model of the above-mentioned embodiment one with a higher accuracy of prediction data, thereby making the abnormal warning of the photovoltaic power generation system more accurate.

[0148] The above-mentioned abnormal warning device for photovoltaic power generation system can implement the above-mentioned abnormal warning method for photovoltaic power generation system. The specific limitations and other contents of the embodiment of the above-mentioned abnormal warning device for photovoltaic power generation system can be found in the contents of the abnormal warning method for photovoltaic power generation system above, and will not be repeated in the embodiment.

[0149] Example 5

[0150] In one embodiment, the present invention provides a computer device comprising a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the computer device to execute the above-mentioned data prediction model training method, or the above-mentioned abnormality warning method for the photovoltaic power generation system.

[0151] Optionally, the above-mentioned computer device may be a server.

[0152] In one embodiment, the internal structure of the computer device of the present invention can be as follows: Figure 5 shown.

[0153] In one embodiment, the present invention 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 abnormality warning method for the photovoltaic power generation system.

[0154] In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architectures, functions, and operations of the devices, methods, and computer program products according to multiple embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or part of the code, which contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the boxes can also occur in an order different from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, as well as the combination of boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified functions or actions, or can be implemented using a combination of dedicated hardware and computer instructions.

[0155] In addition, the functional modules in the various embodiments of the present invention may be integrated together to form an independent part, or each module may exist independently, or two or more modules may be integrated to form an independent part.

[0156] 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 solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0157] The above description is merely an embodiment of the present invention and is not intended to limit the scope of protection of the present invention. For those skilled in the art, the present invention may be subject to various modifications and variations. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention shall be included in the scope of protection of the present invention. It should be noted that similar reference numerals and letters represent similar items in the following figures. Therefore, once an item is defined in one figure, it does not need to be further defined or explained in subsequent figures.

[0158] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

[0159] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply the existence of any such actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device comprising the element.

Claims

1. A data prediction model training method, characterized in that: include: Obtain the target historical usage status of the photovoltaic inverter; Using the target historical usage state quantity to train a preset data prediction model to obtain initial weights of the data prediction model; Obtaining a target real-time usage state quantity of the photovoltaic inverter; Using the target real-time usage state quantity and the data prediction model, predicting the next usage state quantity; updating the initial weight of the data prediction model according to the next usage state quantity and the target real-time usage state quantity; The obtaining of the target real-time usage state of the photovoltaic inverter includes: Obtaining an initial real-time usage state quantity of the photovoltaic inverter; When it is detected that the initial real-time usage state quantity is a standard quantity, the initial real-time usage state quantity is converted into a target real-time usage state quantity using a preset data conversion algorithm, wherein the target real-time usage state quantity is supervised data; The step of obtaining the target real-time usage state of the photovoltaic inverter further includes: When it is detected that the initial real-time usage state quantity is a non-scalar quantity, using the data prediction model to predict a predicted usage state quantity corresponding to the time of the initial real-time usage state quantity; Obtaining a previous usage state quantity of the initial real-time usage state quantity, and obtaining an updated initial real-time usage state quantity based on the previous usage state quantity and the predicted usage state quantity; Using a preset data conversion algorithm, the updated initial real-time usage state quantity is converted into a target real-time usage state quantity, wherein the target real-time usage state quantity is supervised data; When obtaining the updated initial real-time usage state quantity based on the previous usage state quantity and the predicted usage state quantity, the updated initial real-time usage state quantity is obtained by taking a weighted average of the previous usage state quantity and the predicted usage state quantity.

2. The data prediction model training method according to claim 1, characterized in that: The obtaining of the target historical usage state quantity of the photovoltaic inverter includes: Obtain the initial historical usage status of the photovoltaic inverter; Using a preset data update algorithm to update the non-scalar quantity in the initial historical usage state quantity; The updated initial historical usage state quantity is converted into a target historical usage state quantity using a preset data conversion algorithm, where the target historical usage state quantity is supervised data.

3. The data prediction model training method according to claim 1, characterized in that: The error function of the data prediction model is as follows: in, Using state quantities for said target in real time, It is the prediction result of the usage state quantity at time t using time t-1 and the data prediction model.

4. A photovoltaic power generation system abnormality warning method, characterized in that: include: Obtain the target real-time usage status of the photovoltaic inverter; Predicting the next usage state quantity using the target real-time usage state quantity and a data prediction model obtained using the data prediction model training method according to any one of claims 1 to 3; When the next usage state quantity is greater than a preset abnormal warning threshold, an abnormal warning prompt is issued.

5. A data prediction model training device, characterized in that: include: A first acquisition module is used to acquire a target historical usage state quantity of the photovoltaic inverter; A training module, configured to train a preset data prediction model using the target historical usage state quantity to obtain an initial weight of the data prediction model; A second acquisition module is used to obtain a target real-time usage state quantity of the photovoltaic inverter; A data prediction module, configured to predict a next usage state quantity using the target real-time usage state quantity and the data prediction model; An online updating module, configured to update the initial weight of the data prediction model according to the next usage state quantity and the target real-time usage state quantity; The obtaining of the target real-time usage state of the photovoltaic inverter includes: Obtaining an initial real-time usage state quantity of the photovoltaic inverter; When it is detected that the initial real-time usage state quantity is a standard quantity, the initial real-time usage state quantity is converted into a target real-time usage state quantity using a preset data conversion algorithm, wherein the target real-time usage state quantity is supervised data; The step of obtaining the target real-time usage state of the photovoltaic inverter further includes: When it is detected that the initial real-time usage state quantity is a non-scalar quantity, using the data prediction model to predict a predicted usage state quantity corresponding to the time of the initial real-time usage state quantity; Obtaining a previous usage state quantity of the initial real-time usage state quantity, and obtaining an updated initial real-time usage state quantity based on the previous usage state quantity and the predicted usage state quantity; Using a preset data conversion algorithm, the updated initial real-time usage state quantity is converted into a target real-time usage state quantity, wherein the target real-time usage state quantity is supervised data; When obtaining the updated initial real-time usage state quantity based on the previous usage state quantity and the predicted usage state quantity, the updated initial real-time usage state quantity is obtained by taking a weighted average of the previous usage state quantity and the predicted usage state quantity.

6. An abnormal warning device for a photovoltaic power generation system, characterized in that: include: An acquisition module is used to obtain the target real-time usage status of the photovoltaic inverter; a data prediction module, configured to predict the next usage state quantity using the target real-time usage state quantity and a data prediction model obtained using the data prediction model training device according to claim 5; The abnormality warning module is used to issue an abnormality warning prompt when the next usage state quantity is greater than a preset abnormality warning threshold.

7. A computer device, characterized in that: It includes a memory and a processor, the memory is used to store a computer program, and the processor runs the computer program to enable the computer device to execute the data prediction model training method according to any one of claims 1 to 3, or the abnormal warning method for the photovoltaic power generation system according to claim 4.

8. 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 according to any one of claims 1 to 3, or the abnormality early warning method for a photovoltaic power generation system according to claim 4.

Images