A photovoltaic power generation fault intelligent prediction and early warning method and system

Abstract

The application discloses a kind of photovoltaic power generation fault intelligent prediction early warning method and system, it is related to photovoltaic power generation technical field, the method includes: by obtaining the health monitoring category set of photovoltaic power generation, and determine health monitoring index set;Real-time index parameter set is collected, and real-time health index is calculated;If not reach preset health threshold, call preset fault prediction scheme;Obtain historical photovoltaic power generation fault record, and analysis obtains photovoltaic power generation fault tree;Obtain first fault type and match first fault factor, collect first state characteristic parameter;Establish the first state analysis result of photovoltaic power generation simulation model;Photovoltaic power generation is fault prediction early warning.It solves the problem that the work efficiency is low, and the fault is not found in time in the prior art by using traditional artificial inspection and the like to photovoltaic module for operation check.The abnormal identification accuracy of photovoltaic module is improved, and then the system intelligent early warning response is improved, and the effect of guaranteeing photovoltaic power generation normal and stable is achieved.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic power generation technology, and in particular to a method and system for intelligent prediction and early warning of photovoltaic power generation faults. Background Technology

[0002] With the increasing demand for energy and the continuous development of the photovoltaic industry, solar energy has become an important energy source, and the importance of photovoltaic power generation systems in energy supply has gradually become apparent. In existing photovoltaic power generation processes, the core component, the photovoltaic module, is affected by various factors in the actual working environment, leading to abnormal power generation or malfunctions. If regulatory personnel cannot take timely measures to address these emergencies, the normal operation of the entire photovoltaic power generation system will be affected to varying degrees, and may even cause uncontrollable events such as fires. For example, the operation of photovoltaic modules is affected by environmental factors such as light intensity, ambient temperature, wind conditions, and relative humidity, resulting in changes in the power generation efficiency, active power, reactive power, apparent power, and power factor of the photovoltaic power generation system, as well as changes in the amplitude and phase of harmonic voltage and current. Therefore, research on using computer science and technology to dynamically monitor the operating status of each component in the photovoltaic power generation system and to provide intelligent fault prediction and early warning, thereby promptly identifying fault risks and improving the overall efficiency of the photovoltaic power generation system, is urgently needed.

[0003] However, in the existing technology, photovoltaic modules are affected by various factors in the actual working environment during the photovoltaic power generation process, which may lead to abnormal power generation status or failure events. The operation inspection of photovoltaic modules by traditional manual inspection methods is inefficient, the faults are not detected in time, and it is impossible to take corresponding measures in time for sudden accidents. Ultimately, the normal operation of the entire photovoltaic power generation system will be affected to varying degrees. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for intelligent prediction and early warning of photovoltaic power generation faults, in order to solve the technical problems in the prior art where photovoltaic modules are affected by various factors in the actual working environment during the photovoltaic power generation process, resulting in abnormal power generation status or fault events. The traditional manual inspection and other methods for checking the operation of photovoltaic modules are inefficient, faults are not detected in time, and it is impossible to take corresponding measures in time for sudden accidents, which ultimately affects the normal operation of the entire photovoltaic power generation system to varying degrees.

[0005] In view of the above problems, the present invention provides a method and system for intelligent prediction and early warning of photovoltaic power generation faults.

[0006] In a first aspect, the present invention provides a method for intelligent prediction and early warning of photovoltaic power generation faults. The method is implemented through a photovoltaic power generation fault intelligent prediction and early warning system. The method includes: acquiring a health monitoring category set for photovoltaic power generation and determining a set of health monitoring indicators based on the health monitoring category set; collecting data from the health monitoring indicator set of the photovoltaic power generation to obtain a real-time indicator parameter set, and calculating a real-time health index of the photovoltaic power generation using weighted calculations; if the real-time health index does not reach a preset health threshold, invoking a preset fault prediction scheme; acquiring historical photovoltaic power generation fault records based on the preset fault prediction scheme and analyzing the historical photovoltaic power generation fault records to obtain a photovoltaic power generation fault tree; obtaining a first fault type, matching a first fault factor with the photovoltaic power generation fault tree, and collecting a first state characteristic parameter based on the first fault factor; performing principle and technical analysis on the photovoltaic power generation and establishing a photovoltaic power generation simulation model, and obtaining a first state analysis result based on the first state characteristic parameter; and performing fault prediction and early warning for the photovoltaic power generation based on the first state analysis result.

[0007] Secondly, the present invention also provides a photovoltaic power generation fault intelligent prediction and early warning system, used to execute a photovoltaic power generation fault intelligent prediction and early warning method as described in the first aspect, wherein the system includes: an index determination module, used to acquire a health monitoring category set of photovoltaic power generation and determine a health monitoring index set based on the health monitoring category set; a health calculation module, used to collect the health monitoring index set of photovoltaic power generation, obtain a real-time index parameter set, and calculate the real-time health index of photovoltaic power generation by weighted calculation; a scheme invocation module, used to invoke a preset fault prediction scheme if the real-time health index does not reach a preset health threshold; a first fault analysis module, used to acquire historical photovoltaic power generation fault records based on the preset fault prediction scheme and analyze the historical photovoltaic power generation fault records to obtain a photovoltaic power generation fault tree; a second fault analysis module, used to obtain a first fault type, match a first fault factor with the photovoltaic power generation fault tree, and collect a first state feature parameter based on the first fault factor; a fault prediction module, used to perform principle and technical analysis on photovoltaic power generation and establish a photovoltaic power generation simulation model, and obtain a first state analysis result by combining the first state feature parameter; and a fault early warning module, used to perform fault prediction and early warning on photovoltaic power generation based on the first state analysis result.

[0008] One or more technical solutions provided in this invention have at least the following technical effects or advantages:

[0009] By acquiring a health monitoring category set for photovoltaic (PV) power generation and determining a set of health monitoring indicators based on the category set, the system collects data from the PV power generation health monitoring indicator set to obtain a real-time indicator parameter set, and calculates the real-time health index of the PV power generation using weighted calculation. If the real-time health index does not reach a preset health threshold, a preset fault prediction scheme is invoked. Based on the preset fault prediction scheme, historical PV power generation fault records are acquired, and the historical PV power generation fault records are analyzed to obtain a PV power generation fault tree. A first fault type is obtained, and a first fault factor is matched with the PV power generation fault tree. A first state characteristic parameter is collected based on the first fault factor. The system performs principle and technical analysis on the PV power generation and establishes a PV power generation simulation model, and obtains a first state analysis result based on the first state characteristic parameter. Fault prediction and early warning are performed on the PV power generation based on the first state analysis result. This achieves the technical goal of improving the intelligence level of PV module anomaly identification in PV power generation, improving the accuracy of PV module operation anomaly identification, thereby improving the system's intelligent early warning response, and ultimately ensuring the normal and stable operation of the PV power generation system.

[0010] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0012] Figure 1 This is a flowchart illustrating a method for intelligent prediction and early warning of photovoltaic power generation faults according to the present invention.

[0013] Figure 2 This is a flowchart illustrating the process of calculating the first category health index in a photovoltaic power generation fault intelligent prediction and early warning method of the present invention.

[0014] Figure 3 This is a flowchart illustrating the process of drawing the photovoltaic power generation fault tree in the intelligent prediction and early warning method for photovoltaic power generation faults of the present invention.

[0015] Figure 4 This is a flowchart illustrating the process of obtaining the photovoltaic power generation simulation model in the intelligent prediction and early warning method for photovoltaic power generation faults of the present invention.

[0016] Figure 5 This is a schematic diagram of the structure of a photovoltaic power generation fault intelligent prediction and early warning system according to the present invention.

[0017] Attached reference numerals: Indicator determination module M100, health calculation module M200, scheme calling module M300, first fault analysis module M400, second fault analysis module M500, fault prediction module M600, fault early warning module M700. Detailed Implementation

[0018] This invention provides a method and system for intelligent prediction and early warning of photovoltaic (PV) power generation faults. It addresses the technical problems of existing technologies where PV modules, affected by various factors in the actual working environment, may experience abnormal power generation or malfunctions. Traditional manual inspections of PV modules suffer from low efficiency, delayed fault detection, and an inability to respond promptly to emergencies, ultimately impacting the normal operation of the entire PV power generation system. This invention achieves the technical goal of improving the intelligence of PV module anomaly identification, increasing the accuracy of anomaly identification, thereby enhancing the system's intelligent early warning response and ultimately ensuring the normal and stable operation of the PV power generation system.

[0019] The acquisition, storage, use, and processing of data in the technical solution of this invention all comply with the relevant provisions of national laws and regulations.

[0020] The technical solutions of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be understood that the present invention is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. It should also be noted that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, not all of them.

[0021] Example 1

[0022] Please see the appendix Figure 1 This invention provides a method for intelligent prediction and early warning of photovoltaic power generation faults, wherein the method is applied to a photovoltaic power generation fault intelligent prediction and early warning system, and the method specifically includes the following steps:

[0023] Step S100: Obtain the health monitoring category set for photovoltaic power generation, and determine the health monitoring indicator set based on the health monitoring category set;

[0024] Specifically, the aforementioned intelligent prediction and early warning method for photovoltaic power generation faults is applied to a photovoltaic power generation intelligent prediction and early warning system. It can realize intelligent monitoring of the real-time operating status of photovoltaic modules in the photovoltaic power generation system, thereby identifying and predicting abnormal conditions of photovoltaic modules and providing timely and targeted early warnings, thus ensuring the normal and stable operation of the photovoltaic power generation system.

[0025] The health monitoring category set was determined by relevant technical personnel based on historical operation and maintenance data, fault analysis, and their own subjective experience. This set includes system categories prone to faults and anomalies in photovoltaic power generation systems. Exemplary categories include power generation efficiency, active power, reactive power, apparent power and power factor, voltage deviation and frequency deviation under grid connection, etc., within a photovoltaic power generation system. After determining the health monitoring category set, relevant experts and technical personnel analyze each category sequentially to identify environmental factors that affect photovoltaic power generation, thus forming the health monitoring indicator set. Exemplary environmental parameters include temperature, humidity, and irradiance. Furthermore, the operating parameters of the photovoltaic power generation system itself also reflect its operational health, such as electrical parameters like voltage, current, power, and frequency. Therefore, relevant electrical parameters are also added to the health monitoring indicator set. The indicators in this health monitoring indicator set provide theoretical guidance and preference references for the intelligent predictive early warning system.

[0026] Step S200: Collect the health monitoring index set of the photovoltaic power generation to obtain the real-time index parameter set, and calculate the real-time health index of the photovoltaic power generation by weighting.

[0027] Further details are attached. Figure 2 As shown, step S200 of the present invention includes:

[0028] Step S210: Extract the first monitoring category from the health monitoring category set;

[0029] Step S220: Based on the first monitoring category, filter the health monitoring indicator set to obtain the first monitoring indicator set;

[0030] Furthermore, step S220 of the present invention includes:

[0031] Step S221: Take each health monitoring indicator in the health monitoring indicator set as the independent variable and the first monitoring category as the dependent variable;

[0032] Step S222: Draw a scatter plot based on the independent variable, the dependent variable, and their mapping relationship, wherein the scatter plot includes multiple target scatter plots;

[0033] Step S223: Analyze the scatter plots of the multiple targets to obtain the maximum information coefficients of the multiple targets, and filter to obtain the first set of maximum information coefficients of the targets;

[0034] Step S224: Back-match the first monitoring index set based on the first target maximum information coefficient set.

[0035] Step S230: Collect parameters of each monitoring indicator in the first monitoring indicator set in real time to obtain the first indicator parameter set;

[0036] Step S240: Normalize the first index parameter set to obtain the first parameter processing result, and obtain the corresponding first index weight coefficient set;

[0037] Step S250: Based on the processing result of the first parameter and the first indicator weight coefficient set, calculate the first category health index.

[0038] Specifically, after analyzing the environmental factors affecting the operation of photovoltaic modules and the electrical factors reflecting the operating status of the photovoltaic power generation system during the photovoltaic power generation process, and obtaining the set of health monitoring indicators, relevant intelligent devices are used to dynamically monitor and collect the set of health monitoring indicators during the photovoltaic power generation process, thereby obtaining the set of real-time indicator parameters. The set of real-time indicator parameters and the set of health monitoring indicators have a one-to-one correspondence.

[0039] After dynamically acquiring the real-time indicator parameter set, the actual status of each category in the health monitoring category set is objectively analyzed and evaluated based on the real-time acquired parameter information. First, any monitoring category in the health monitoring category set is extracted and designated as the first monitoring category. Then, based on the first monitoring category, the health monitoring indicator set is subjected to correlation screening, retaining only electrical and environmental indicators that affect the first monitoring category, thus obtaining the first monitoring indicator set. Parameters of each monitoring indicator in the first monitoring indicator set are acquired in real-time to obtain the first indicator parameter set. Next, the first indicator parameter set is normalized to obtain the first parameter processing result. Normalization is used to eliminate the dimensional influence between different indicator factors, facilitating subsequent calculations. Furthermore, a first indicator weight coefficient set is obtained corresponding to the first indicator parameter set. For example, combining the acquired first indicator parameter set, the weight coefficients of each indicator in the first monitoring indicator set can be calculated using the coefficient of variation method. Finally, based on the first parameter processing result and the first indicator weight coefficient set, a weighted calculation is performed to obtain the first category health index. In other words, the first set of monitoring indicators consists of indicator factors that will affect the first monitoring category. By calculating and analyzing the parameters of each indicator in the first set of monitoring indicators, the actual health status of the first monitoring category can be obtained.

[0040] When performing relevance screening on the health monitoring indicator set based on the first monitoring category, to improve screening efficiency and ensure the scientific validity and rationality of the resulting first monitoring indicator set, the maximum information coefficient method is used to calculate the correlation degree between each monitoring indicator and its corresponding monitoring category. Indicators with high correlation are then retained, while other indicators with little impact on the first monitoring category are removed. Specifically, each health monitoring indicator in the health monitoring indicator set is first used as an independent variable, and the first monitoring category as a dependent variable. Then, a scatter plot is drawn based on the independent variables, the dependent variables, and their mapping relationships. This scatter plot includes multiple target scatter plots, meaning each health monitoring indicator forms a scatter plot with the first monitoring category. Next, the multiple target scatter plots are analyzed to obtain multiple target maximum information coefficients. Based on the magnitude of these maximum information coefficients, a first target maximum information coefficient set is obtained. For example, the multiple target maximum information coefficients are sorted in descending order, and the top 50% of health indicators in the descending order are retained as monitoring indicators for the first health category. Finally, the first monitoring indicator set is back-matched based on the first target maximum information coefficient set. By using correlation analysis, the goal of quantifying the correlation between each health monitoring indicator in the health monitoring indicator set and the first monitoring category was achieved. This resulted in improving the scientific rigor and rationality of indicator selection, while also enhancing the intelligence of the selection process.

[0041] Step S300: If the real-time health index does not reach the preset health threshold, invoke the preset fault prediction scheme;

[0042] Step S400: Obtain historical photovoltaic power generation fault records based on the preset fault prediction scheme, and analyze the historical photovoltaic power generation fault records to obtain a photovoltaic power generation fault tree;

[0043] Further details are attached. Figure 3 As shown, step S400 of the present invention includes:

[0044] Step S410: Construct a power generation fault set based on the historical photovoltaic power generation fault records, which includes multiple power generation faults;

[0045] Step S420: Sequentially treat the multiple power generation failures as top events;

[0046] Step S430: Analyze the accident factors of the top event;

[0047] Step S440: Based on the top event and the accident factor, draw the photovoltaic power generation fault tree.

[0048] Specifically, after sequentially analyzing and calculating the health indices of each monitoring category in the health monitoring category set, a weighted calculation is performed to obtain the overall health status of the photovoltaic power generation, i.e., the real-time health index. When the real-time health index fails to reach a preset health threshold, it indicates that the current operating status of the photovoltaic power generation system is unhealthy and a fault is highly likely to occur. Therefore, the intelligent prediction and early warning system automatically invokes a preset fault prediction scheme to intelligently analyze and predict potential anomalies and faults in the photovoltaic power generation system. The preset health threshold is a range of health indices for the healthy operating status of the photovoltaic power generation system, determined by relevant personnel through comprehensive analysis of the system's actual operating years, operating environment, and operating load, combined with relevant operating experience. The preset fault prediction scheme is pre-stored in the intelligent prediction and early warning system and serves as the logic for fault monitoring and analysis of the photovoltaic power generation.

[0049] Furthermore, historical photovoltaic (PV) power generation fault records are obtained based on the preset fault prediction scheme. These records include information on various faults that occurred during the PV power generation process. Therefore, a power generation fault set is constructed based on these historical records, comprising multiple power generation faults. Next, these multiple power generation faults are sequentially treated as top events, and their accident factors are analyzed. Finally, based on the top events, the accident factors, and their correspondences, a PV power generation fault tree is constructed. By analyzing the historical PV power generation fault records to obtain the PV power generation fault tree, a personalized monitoring indicator scheme is provided for subsequent targeted fault prediction of various items in the PV power generation system, thereby improving the targeting and efficiency of fault prediction and early warning.

[0050] Step S500: Obtain the first fault type, match the first fault factor in combination with the photovoltaic power generation fault tree, and collect the first state feature parameters based on the first fault factor;

[0051] Specifically, the first fault type refers to any fault for which the intelligent prediction and early warning system will perform intelligent prediction and early warning. An example is predicting whether the power generation of the current photovoltaic power generation system is normal. After determining a specific fault type to be analyzed, the system matches the first fault factor of that fault type using the photovoltaic power generation fault tree. Then, based on the first fault factor, relevant state indicators of the photovoltaic power generation system are collected in a targeted manner, i.e., first state characteristic parameters are obtained. This achieves the technical effect of targeted monitoring and analysis of the photovoltaic power generation system, improving the personalization of fault prediction and early warning.

[0052] Step S600: Perform principle and technical analysis on the photovoltaic power generation and establish a photovoltaic power generation simulation model, and obtain the first state analysis results by combining the first state characteristic parameters;

[0053] Further details are attached. Figure 4 As shown, step S600 of the present invention includes:

[0054] Step S610: Obtain the simulation training dataset and randomly divide the simulation training dataset to obtain the division result;

[0055] Step S620: Wherein, the partitioning result includes a first dataset, a second dataset, and a third dataset;

[0056] Step S630: Train a first model based on the first dataset, train a second model based on the second dataset, and train a third model based on the third dataset;

[0057] Step S640: Integrate and fuse the first model, the second model, and the third model to obtain the photovoltaic power generation simulation model.

[0058] Furthermore, step S640 of the present invention includes:

[0059] Step S641: Integrate and fuse the first model, the second model, and the third model to obtain multiple integrated and fused models;

[0060] Step S642: Randomly extract any one of the multiple integrated fusion models;

[0061] Step S643: Obtain the primary learner and meta-learner of any one of the ensemble fusion models;

[0062] Step S644: Perform verification analysis on the primary learner and the meta learner to obtain verification record information;

[0063] Step S645: Determine the photovoltaic power generation simulation model based on the verification record information.

[0064] Furthermore, step S645 of the present invention also includes:

[0065] Step S6451: The verification record information includes the verification status analysis results and the verification duration;

[0066] Step S6452: Obtain the preset state analysis results;

[0067] Step S6453: Compare the verification state analysis results with the preset state analysis results to obtain the verification deviation;

[0068] Step S6454: Calculate the comprehensive verification index based on the verification duration and the verification deviation.

[0069] Specifically, the photovoltaic power generation simulation model is used for intelligent simulation analysis of the photovoltaic power generation process. It also combines actual data to achieve simulation monitoring in a virtual space, and then uses the simulation monitoring results for intelligent fault prediction and early warning. First, a simulation training dataset is obtained based on big data, and this dataset is randomly divided into a first dataset, a second dataset, and a third dataset, forming the division result. Then, intelligent models are trained using each of the divided datasets: a first model is trained based on the first dataset, a second model is trained based on the second dataset, and a third model is trained based on the third dataset. The intelligent models trained on each dataset are based on different computer-supervised learning principles, thus each intelligent model possesses different advantages. Examples include training neural network models to improve prediction accuracy and training gray wolf optimization models to improve prediction efficiency. Finally, the first model, the second model, and the third model are integrated and fused to obtain the photovoltaic power generation simulation model. An example is using the Stacking algorithm, where two models are used as primary learners, and another model is used as a meta-learner for model integration and fusion. The prediction results from multiple primary learners serve as the training basis for the meta-learner, which then performs new learning and prediction. Therefore, by integrating and fusing the first model, the second model, and the third model, multiple different integrated and fused models can be obtained.

[0070] Next, any one of the multiple integrated fusion models is randomly selected, and its primary learner and meta-learner are obtained. The primary learner and meta-learner correspond to intelligent models trained using different training principles. Then, the primary learner and meta-learner are validated and analyzed, and relevant data during the validation process is recorded, resulting in validation record information, including validation state analysis results and validation duration. Further, preset state analysis results are obtained, and the validation state analysis results are compared with these preset state analysis results to obtain the validation deviation of any one integrated fusion model. Next, based on the validation duration and validation deviation, a validation comprehensive index is calculated, and the photovoltaic power generation simulation model is determined based on this index. For example, the integrated fusion model with the highest validation comprehensive index is selected as the final photovoltaic power generation simulation model, and the first state analysis result is obtained by combining the photovoltaic power generation simulation model with the first state feature parameters. By determining the photovoltaic power generation simulation model through integrated fusion analysis, the technical effect of ensuring the accuracy of model fault prediction is achieved.

[0071] Step S700: Perform fault prediction and early warning for the photovoltaic power generation based on the first state analysis results.

[0072] Specifically, after the photovoltaic power generation simulation model obtains the first state analysis result by combining the first state characteristic parameters, the intelligent prediction and early warning system performs fault prediction and early warning for the photovoltaic power generation based on the first state analysis result. An exemplary approach is for relevant technical personnel to analyze and preset the threshold values ​​of indicator factors for the monitoring category to determine whether a fault exists. After the model's intelligent analysis obtains the first state analysis result, it is compared with the preset parameter thresholds. If any indicator parameter is outside the preset range, it indicates that the parameter is abnormal. This achieves the technical goal of improving the intelligence level of photovoltaic module anomaly identification in photovoltaic power generation.

[0073] In summary, the intelligent prediction and early warning method for photovoltaic power generation faults provided by this invention has the following technical effects:

[0074] By acquiring a health monitoring category set for photovoltaic (PV) power generation and determining a set of health monitoring indicators based on the category set, the system collects data from the PV power generation health monitoring indicator set to obtain a real-time indicator parameter set, and calculates the real-time health index of the PV power generation using weighted calculation. If the real-time health index does not reach a preset health threshold, a preset fault prediction scheme is invoked. Based on the preset fault prediction scheme, historical PV power generation fault records are acquired, and the historical PV power generation fault records are analyzed to obtain a PV power generation fault tree. A first fault type is obtained, and a first fault factor is matched with the PV power generation fault tree. A first state characteristic parameter is collected based on the first fault factor. The system performs principle and technical analysis on the PV power generation and establishes a PV power generation simulation model, and obtains a first state analysis result based on the first state characteristic parameter. Fault prediction and early warning are performed on the PV power generation based on the first state analysis result. This achieves the technical goal of improving the intelligence level of PV module anomaly identification in PV power generation, improving the accuracy of PV module operation anomaly identification, thereby improving the system's intelligent early warning response, and ultimately ensuring the normal and stable operation of the PV power generation system.

[0075] Example 2

[0076] Based on the intelligent prediction and early warning method for photovoltaic power generation faults described in the foregoing embodiments, and using the same inventive concept, this invention also provides an intelligent prediction and early warning system for photovoltaic power generation faults. Please refer to the appendix. Figure 5 The system includes:

[0077] The indicator determination module M100 is used to obtain the health monitoring category set of photovoltaic power generation and determine the health monitoring indicator set based on the health monitoring category set;

[0078] The health calculation module M200 is used to collect the health monitoring index set of the photovoltaic power generation, obtain the real-time index parameter set, and calculate the real-time health index of the photovoltaic power generation by weighting.

[0079] The scheme invocation module M300 is used to invoke a preset fault prediction scheme if the real-time health index does not reach a preset health threshold.

[0080] The first fault analysis module M400 is used to obtain historical photovoltaic power generation fault records based on the preset fault prediction scheme, and analyze the historical photovoltaic power generation fault records to obtain a photovoltaic power generation fault tree.

[0081] The second fault analysis module M500 is used to obtain the first fault type, match the first fault factor in combination with the photovoltaic power generation fault tree, and collect the first state feature parameters based on the first fault factor.

[0082] The fault prediction module M600 is used to perform principle and technical analysis of the photovoltaic power generation and establish a photovoltaic power generation simulation model, and obtain the first state analysis result by combining the first state characteristic parameters.

[0083] The fault warning module M700 is used to predict and warn of faults in the photovoltaic power generation based on the first state analysis results.

[0084] Furthermore, the health calculation module M200 in the system is also used for:

[0085] Extract the first monitoring category from the health monitoring category set;

[0086] The first monitoring indicator set is obtained by filtering the health monitoring indicator set based on the first monitoring category;

[0087] The parameters of each monitoring indicator in the first monitoring indicator set are collected in real time to obtain the first indicator parameter set;

[0088] The first indicator parameter set is normalized to obtain the first parameter processing result, and the corresponding first indicator weight coefficient set is obtained.

[0089] Based on the processing results of the first parameter and the first set of indicator weight coefficients, the health index of the first category is calculated.

[0090] Furthermore, the health calculation module M200 in the system is also used for:

[0091] Each health monitoring indicator in the health monitoring indicator set is used as the independent variable, and the first monitoring category is used as the dependent variable.

[0092] Scatter plots are drawn based on the independent variables, the dependent variables, and their mapping relationships, wherein the scatter plots include multiple target scatter plots;

[0093] The maximum information coefficients of multiple targets are obtained by analyzing the scatter plots of the multiple targets, and the first set of maximum information coefficients of the targets is obtained by filtering.

[0094] The first monitoring indicator set is back-matched based on the first target maximum information coefficient set.

[0095] Furthermore, the first fault analysis module M400 in the system is also used for:

[0096] A set of power generation faults is constructed based on the historical photovoltaic power generation fault records, which includes multiple power generation faults;

[0097] The multiple power generation failures are sequentially treated as top events;

[0098] Analyze the accident factors of the aforementioned top event;

[0099] Based on the top event and the accident factors, the photovoltaic power generation fault tree is plotted.

[0100] Furthermore, the fault prediction module M600 in the system is also used for:

[0101] Obtain the simulation training dataset and randomly divide the simulation training dataset to obtain the division result;

[0102] The partitioning results include a first dataset, a second dataset, and a third dataset;

[0103] A first model is trained based on the first dataset, a second model is trained based on the second dataset, and a third model is trained based on the third dataset.

[0104] The first model, the second model, and the third model are integrated and fused to obtain the photovoltaic power generation simulation model.

[0105] Furthermore, the fault prediction module M600 in the system is also used for:

[0106] The first model, the second model, and the third model are integrated and fused to obtain multiple integrated and fused models;

[0107] Randomly select any one of the multiple integrated fusion models;

[0108] Obtain the primary learner and meta-learner of any one of the integrated fusion models;

[0109] The primary learner and the meta-learner are validated and analyzed to obtain validation record information;

[0110] The photovoltaic power generation simulation model is determined based on the verification record information.

[0111] Furthermore, the fault prediction module M600 in the system is also used for:

[0112] The verification record information includes the verification status analysis results and the verification duration;

[0113] Obtain the preset state analysis results;

[0114] By comparing the verification state analysis results with the preset state analysis results, the verification deviation is obtained;

[0115] Based on the verification duration and the verification deviation, a comprehensive verification index is calculated.

[0116] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Figure 1 The intelligent prediction and early warning method and specific examples of photovoltaic power generation faults in Embodiment 1 are also applicable to the intelligent prediction and early warning system of photovoltaic power generation faults in this embodiment. Through the foregoing detailed description of the intelligent prediction and early warning method of photovoltaic power generation faults, those skilled in the art can clearly understand the intelligent prediction and early warning system of photovoltaic power generation faults in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here. As for the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant details can be found in the method section.

[0117] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0118] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of this invention and its equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for intelligent prediction and early warning of photovoltaic power generation faults, characterized in that, include: Obtain the health monitoring category set for photovoltaic power generation, and determine the health monitoring indicator set based on the health monitoring category set; The health monitoring index set of the photovoltaic power generation is collected to obtain a real-time index parameter set, and the real-time health index of the photovoltaic power generation is calculated by weighted calculation. If the real-time health index does not reach the preset health threshold, the preset fault prediction scheme is invoked. Based on the preset fault prediction scheme, historical photovoltaic power generation fault records are obtained, and the historical photovoltaic power generation fault records are analyzed to obtain a photovoltaic power generation fault tree; The first fault type is obtained, the first fault factor is matched in combination with the photovoltaic power generation fault tree, and the first state feature parameters are collected based on the first fault factor. The principle and technology of photovoltaic power generation are analyzed and a photovoltaic power generation simulation model is established. The first state analysis results are obtained by combining the first state characteristic parameters. Based on the first state analysis results, fault prediction and early warning are performed for the photovoltaic power generation. The process of collecting the set of health monitoring indicators for the photovoltaic power generation to obtain a real-time indicator parameter set, and then calculating the real-time health index of the photovoltaic power generation using weighted methods, includes: Extract the first monitoring category from the health monitoring category set; The first monitoring indicator set is obtained by filtering the health monitoring indicator set based on the first monitoring category; The parameters of each monitoring indicator in the first monitoring indicator set are collected in real time to obtain the first indicator parameter set; The first indicator parameter set is normalized to obtain the first parameter processing result, and the corresponding first indicator weight coefficient set is obtained. Based on the processing results of the first parameter and the first set of indicator weight coefficients, the health index of the first category is calculated. The first monitoring indicator set is obtained by filtering the health monitoring indicator set based on the first monitoring category, including: Each health monitoring indicator in the health monitoring indicator set is used as the independent variable, and the first monitoring category is used as the dependent variable. Scatter plots are drawn based on the independent variables, the dependent variables, and their mapping relationships, wherein the scatter plots include multiple target scatter plots; The maximum information coefficients of multiple targets are obtained by analyzing the scatter plots of the multiple targets, and the first set of maximum information coefficients of the targets is obtained by filtering. The first monitoring indicator set is back-matched based on the first target maximum information coefficient set.

2. The intelligent prediction and early warning method according to claim 1, characterized in that, The step of obtaining historical photovoltaic power generation fault records based on the preset fault prediction scheme and analyzing the historical photovoltaic power generation fault records to obtain a photovoltaic power generation fault tree includes: A set of power generation faults is constructed based on the historical photovoltaic power generation fault records, which includes multiple power generation faults; The multiple power generation failures are sequentially treated as top events; Analyze the accident factors of the aforementioned top event; Based on the top event and the accident factors, the photovoltaic power generation fault tree is plotted.

3. The intelligent prediction and early warning method according to claim 1, characterized in that, The process of performing a principle and technical analysis of photovoltaic power generation and establishing a photovoltaic power generation simulation model includes: Obtain the simulation training dataset and randomly divide the simulation training dataset to obtain the division result; The partitioning results include a first dataset, a second dataset, and a third dataset; A first model is trained based on the first dataset, a second model is trained based on the second dataset, and a third model is trained based on the third dataset. The first model, the second model, and the third model are integrated and fused to obtain the photovoltaic power generation simulation model.

4. The intelligent prediction and early warning method according to claim 3, characterized in that, The process of integrating and fusing the first model, the second model, and the third model to obtain the photovoltaic power generation simulation model includes: The first model, the second model, and the third model are integrated and fused to obtain multiple integrated and fused models; Randomly select any one of the multiple integrated fusion models; Obtain the primary learner and meta-learner of any one of the integrated fusion models; The primary learner and the meta-learner are validated and analyzed to obtain validation record information; The photovoltaic power generation simulation model is determined based on the verification record information.

5. The intelligent prediction and early warning method according to claim 4, characterized in that, The process of determining the photovoltaic power generation simulation model based on the verification record information includes: The verification record information includes the verification status analysis results and the verification duration; Obtain the preset state analysis results; By comparing the verification state analysis results with the preset state analysis results, the verification deviation is obtained; Based on the verification duration and the verification deviation, a comprehensive verification index is calculated.

6. A photovoltaic power generation fault intelligent prediction and early warning system, characterized in that, The intelligent prediction and early warning system includes: The indicator determination module is used to obtain the health monitoring category set of photovoltaic power generation and determine the health monitoring indicator set based on the health monitoring category set; The health calculation module is used to collect the set of health monitoring indicators of the photovoltaic power generation, obtain a set of real-time indicator parameters, and calculate the real-time health index of the photovoltaic power generation by weighting. The scheme invocation module is used to invoke a preset fault prediction scheme if the real-time health index does not reach a preset health threshold. The first fault analysis module is used to obtain historical photovoltaic power generation fault records based on the preset fault prediction scheme, and analyze the historical photovoltaic power generation fault records to obtain a photovoltaic power generation fault tree. The second fault analysis module is used to obtain the first fault type, match the first fault factor in combination with the photovoltaic power generation fault tree, and collect the first state feature parameters based on the first fault factor. The fault prediction module is used to perform principle and technical analysis of the photovoltaic power generation and establish a photovoltaic power generation simulation model, and obtain the first state analysis result by combining the first state characteristic parameters. A fault early warning module is used to predict and warn of faults in the photovoltaic power generation based on the first state analysis results. The health calculation module in the system is also used for: Extract the first monitoring category from the health monitoring category set; The first monitoring indicator set is obtained by filtering the health monitoring indicator set based on the first monitoring category; The parameters of each monitoring indicator in the first monitoring indicator set are collected in real time to obtain the first indicator parameter set; The first indicator parameter set is normalized to obtain the first parameter processing result, and the corresponding first indicator weight coefficient set is obtained. Based on the processing results of the first parameter and the first set of indicator weight coefficients, the health index of the first category is calculated. Each health monitoring indicator in the health monitoring indicator set is used as the independent variable, and the first monitoring category is used as the dependent variable. Scatter plots are drawn based on the independent variables, the dependent variables, and their mapping relationships, wherein the scatter plots include multiple target scatter plots; The maximum information coefficients of multiple targets are obtained by analyzing the scatter plots of the multiple targets, and the first set of maximum information coefficients of the targets is obtained by filtering. The first monitoring indicator set is back-matched based on the first target maximum information coefficient set.

Images