A method for predicting power generation of photovoltaic power generation system
By preprocessing and feature extraction of monitoring data and meteorological data of the photovoltaic power generation system, and inputting the prediction model to adjust the prediction results, the problem of insufficient accuracy of photovoltaic power generation prediction in the prior art is solved, and more efficient management support is achieved.
Patent Information
- Application Number
- CN202411591762.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-08
- Publication Date
- 2025-04-08
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing photovoltaic power generation prediction methods have insufficient prediction accuracy, resulting in inadequate management efficiency.
Through the target processing center, the key data obtained by the monitoring equipment and weather station are preprocessed and featured. The power prediction model is input to adjust the prediction results to ensure the accuracy of power generation prediction.
It improves the prediction accuracy of the power generation of photovoltaic power generation systems and provides strong support for system management.
Smart Images
Figure CN119783868A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of photovoltaic power generation, and in particular to a method for predicting power generation of a photovoltaic power generation system. Background Art
[0002] In recent years, the power generation of photovoltaic power generation systems has been affected by a variety of meteorological factors, such as solar radiation intensity, temperature, and cloud coverage, resulting in low management efficiency of photovoltaic power generation systems. Although the existing power generation prediction methods can predict photovoltaic power generation to a certain extent, their prediction accuracy still needs to be improved. Therefore, how to accurately predict the power generation of photovoltaic power generation systems has become one of the current research centers.
[0003] Therefore, the present invention provides a method for predicting power generation of a photovoltaic power generation system. Summary of the invention
[0004] The present invention provides a method for predicting power generation of a photovoltaic power generation system, which is used to perform data preprocessing and feature extraction on key data obtained by monitoring a photovoltaic power generation system and a preset meteorological station using set monitoring equipment through a target processing center, so as to obtain first target data and key features; the pending prediction results obtained by inputting the first target data and the key features into an electric power prediction model are adjusted to obtain a target power generation prediction result, thereby ensuring the accuracy of power generation prediction and providing strong support for the management of the photovoltaic power generation system.
[0005] The present invention provides a method for predicting power generation of a photovoltaic power generation system, comprising:
[0006] Step 1: Use the set monitoring equipment to monitor the current photovoltaic power generation system and the preset weather station to obtain key data and transmit it to the target processing center;
[0007] Step 2: The target processing center performs data preprocessing and feature extraction on the key data to obtain first target data and key features accordingly;
[0008] Step 3: Input the acquired first target data and key features into a pre-established power prediction model, perform power generation prediction on the photovoltaic power generation system, and obtain a pending prediction result;
[0009] Step 4: Analyze and adjust the pending prediction results to determine the target power generation prediction results.
[0010] Preferably, using set monitoring equipment to monitor the current photovoltaic power generation system and the preset weather station to obtain key data and transmit it to the target processing center, including:
[0011] The power monitoring sensor installed at the preset position of the photovoltaic power generation system is used to monitor the power in real time to obtain the real-time power generation;
[0012] Using a meteorological monitoring sensor at a preset position of a preset meteorological station to monitor meteorological data in real time, to obtain real-time meteorological data;
[0013] The real-time power generation and real-time meteorological data are transmitted to a target processing center as key data.
[0014] Preferably, the real-time meteorological data includes solar radiation intensity, temperature, humidity, wind speed, wind direction and cloud coverage.
[0015] Preferably, the target processing center performs data preprocessing and feature extraction on the key data to obtain first target data and key features, including:
[0016] The key data is cleaned, converted, integrated and standardized to obtain first target data;
[0017] Extracting first meteorological features from the first target data, and determining a first correlation coefficient between each first meteorological feature and power generation by using a set similarity algorithm;
[0018] Determining a real-time similarity threshold according to the obtained first correlation coefficient;
[0019] The first correlation coefficient between each first meteorological feature and power generation is combined with the real-time similarity threshold for analysis, and the first meteorological features are screened to obtain key features.
[0020] Preferably, the first correlation coefficient between each first meteorological feature and power generation is combined with the real-time similarity threshold for analysis, and the first meteorological feature is screened to obtain key features, including:
[0021] The first meteorological feature whose current first correlation coefficient is greater than the real-time similarity threshold is regarded as the first undetermined feature;
[0022] If the current number of first undetermined features is not less than the set number of features, all the acquired first undetermined features are output as key features;
[0023] If the current number of the first undetermined features is less than the set number of features, then calculate the first number b1 which is the set number of features minus the first number of undetermined features;
[0024] All features except the first undetermined feature in the first meteorological feature are sorted from large to small according to the first correlation coefficient to obtain a feature screening list;
[0025] The first b1 first meteorological features in the feature screening list are extracted and combined with the current first pending feature to be output as key features.
[0026] Preferably, the calculation formula of the real-time similarity threshold is as follows:
[0027] Where X1 represents the real-time similarity threshold; Δs avg It is expressed as the average value of the first correlation coefficient between the current first meteorological characteristic and the power generation; s max It is expressed as the maximum value of the first correlation coefficients of all current first meteorological characteristics and power generation; s min It is represented as the minimum value of the first correlation coefficients of all current first meteorological characteristics and power generation; e is represented as the natural base.
[0028] Preferably, the first target data and key features obtained are input into a pre-established power prediction model to predict the power generation of the photovoltaic power generation system to obtain a pending prediction result, including:
[0029] Extracting historical power generation and corresponding historical meteorological data within a preset time period from a photovoltaic power generation database;
[0030] After preprocessing the acquired historical power generation and historical meteorological data, first reference data is obtained;
[0031] Extracting reference features consistent with the feature category of key features from the first reference data;
[0032] Using the first reference data and the reference feature as training data, training a neural network to obtain a first power prediction model;
[0033] Inputting the acquired first target data and key features into a first power forecasting model to obtain a first pending forecasting result;
[0034] Using the first reference data and the reference features as training data, training the time series to obtain a second power prediction model;
[0035] The acquired first target data and key features are input into a second electricity quantity prediction model to obtain a second pending prediction result.
[0036] Preferably, analyzing and adjusting the obtained pending prediction results to determine the target power generation prediction results includes:
[0037] Step 11: if the first pending prediction result is the same as the second pending prediction result, outputting the current first prediction result as the target power generation prediction result;
[0038] Step 12: if the first pending prediction result is different from the second pending prediction result, obtaining an absolute difference in power between the current first pending prediction result and the second pending prediction result;
[0039] If the absolute difference in power is not greater than the set power difference, the average power generation of the first pending prediction result and the second pending prediction result is output as the target power generation prediction result;
[0040] Step 13: If the absolute difference in power is greater than the set power difference, the set optimization strategy is used to optimize the first power prediction model and the second power prediction model;
[0041] Step 14: Re-predict the power generation using the optimized first power prediction model and the second power prediction model to obtain corresponding prediction results;
[0042] Step 15: Repeat steps 11-14 until the maximum number of iterations is reached or the target power generation prediction result is output;
[0043] Step 16: When the set maximum number of iterations is reached and the target power generation prediction result is still not output, the maintenance personnel are notified to perform abnormal maintenance.
[0044] Compared with the prior art, the present invention has the following beneficial effects:
[0045] The target processing center performs data preprocessing and feature extraction on the key data obtained by monitoring the photovoltaic power generation system and the preset meteorological station using the set monitoring equipment to obtain the first target data and key features; the pending prediction results obtained by inputting the first target data and key features into the electricity prediction model are adjusted to obtain the target power generation prediction results, which can ensure the accuracy of the power generation prediction and provide strong support for the management of the photovoltaic power generation system.
[0046] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.
[0047] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS
[0048] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:
[0049] Figure 1 The present invention is a flowchart of a photovoltaic power generation system power generation prediction method according to an embodiment of the present invention. DETAILED DESCRIPTION
[0050] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.
[0051] The embodiment of the present invention provides a method for predicting power generation of a photovoltaic power generation system. Figure 1 As shown, including:
[0052] Step 1: Use the set monitoring equipment to monitor the current photovoltaic power generation system and the preset weather station to obtain key data and transmit it to the target processing center;
[0053] Step 2: The target processing center performs data preprocessing and feature extraction on the key data to obtain first target data and key features accordingly;
[0054] Step 3: Input the acquired first target data and key features into a pre-established power prediction model, perform power generation prediction on the photovoltaic power generation system, and obtain a pending prediction result;
[0055] Step 4: Analyze and adjust the pending prediction results to determine the target power generation prediction results.
[0056] In this embodiment, the set monitoring equipment refers to power monitoring sensors and meteorological monitoring sensors, among which the power monitoring sensors are used to monitor the power generation of the photovoltaic power generation system in real time; the meteorological monitoring sensors include solar radiation sensors, temperature sensors, humidity sensors, wind speed and direction sensors, etc., which are used to monitor meteorological data in real time; the key data refers to real-time power generation and real-time meteorological data; the target processing center is used to process the data collected by the set monitoring equipment and predict the power generation of the photovoltaic power generation system.
[0057] In this embodiment, data preprocessing includes data cleaning, data conversion, data integration and standardization processing; the key feature refers to the first meteorological feature whose first correlation coefficient is greater than the set similarity threshold, wherein the first meteorological feature is extracted from the first target data, such as temperature, humidity, and wind speed; the power prediction model is a model pre-established based on a neural network or a time series, and is used to predict the power generation of a photovoltaic power generation system; the pending prediction result is determined by comprehensively analyzing the prediction results obtained by predicting the power prediction model established based on a neural network or a time series; the target power generation prediction result is the final predicted power generation obtained after analyzing and adjusting the pending prediction results.
[0058] The beneficial effects of the above technical solution are as follows: the target processing center performs data preprocessing and feature extraction on the key data obtained by monitoring the photovoltaic power generation system and the preset meteorological station using the set monitoring equipment, so as to obtain the first target data and key features; the pending prediction results obtained by inputting the first target data and key features into the electricity prediction model are adjusted to obtain the target power generation prediction results, which can ensure the accuracy of the power generation prediction and provide strong support for the management of the photovoltaic power generation system.
[0059] The embodiment of the present invention provides a method for predicting the power generation of a photovoltaic power generation system, which uses a set monitoring device to monitor the current photovoltaic power generation system and a preset weather station to obtain key data and transmit it to a target processing center, including:
[0060] The power monitoring sensor installed at the preset position of the photovoltaic power generation system is used to monitor the power in real time to obtain the real-time power generation;
[0061] Using a meteorological monitoring sensor at a preset position of a preset meteorological station to monitor meteorological data in real time, to obtain real-time meteorological data;
[0062] The real-time power generation and real-time meteorological data are transmitted to a target processing center as key data.
[0063] In this embodiment, the photovoltaic power generation system refers to a power generation system that uses the photovoltaic effect of photovoltaic cells to directly convert solar radiation energy into electrical energy, and is composed of a solar cell array (module), cables, power electronic converters (inverters), energy storage devices (batteries), loads (i.e., users), etc.; the set monitoring equipment refers to power monitoring sensors and meteorological monitoring sensors, wherein the power monitoring sensors are used to monitor the power generation of the photovoltaic power generation system in real time; the meteorological monitoring sensors include solar radiation sensors, temperature sensors, humidity sensors, wind speed and direction sensors, etc., which are used to monitor meteorological data in real time and obtain real-time meteorological data, and the real-time meteorological data include solar radiation intensity, temperature, humidity, wind speed, wind direction, cloud coverage, etc.; key data refers to real-time power generation and real-time meteorological data; the preset position is a pre-determined position for installing a power monitoring sensor or a meteorological monitoring sensor; the target processing center is used to process the data collected by the set monitoring equipment and predict the power generation of the photovoltaic power generation system.
[0064] The beneficial effect of the above technical solution is: by monitoring the current photovoltaic power generation system and the preset meteorological station with the set monitoring equipment, key data is obtained and transmitted to the target processing center, providing a data basis for the subsequent prediction of the power generation of the photovoltaic power generation system.
[0065] The embodiment of the present invention provides a method for predicting power generation of a photovoltaic power generation system. The target processing center performs data preprocessing and feature extraction on key data to obtain first target data and key features, including:
[0066] The key data is cleaned, converted, integrated and standardized to obtain first target data;
[0067] Extracting first meteorological features from the first target data, and determining a first correlation coefficient between each first meteorological feature and power generation by using a set similarity algorithm;
[0068] Determining a real-time similarity threshold according to the obtained first correlation coefficient;
[0069] The first correlation coefficient between each first meteorological feature and power generation is combined with the real-time similarity threshold for analysis, and the first meteorological features are screened to obtain key features.
[0070] In this embodiment, the first target data is obtained after data cleaning, data conversion, data integration and standardization processing of the key data; the first meteorological feature is extracted from the first target data, such as temperature, humidity, and wind speed; the similarity algorithm is set to refer to the Pearson correlation coefficient algorithm; the first correlation coefficient is used to characterize the correlation between the first meteorological feature and the power generation; the real-time similarity threshold is obtained by real-time calculation based on the first similarity coefficient of the first meteorological feature and the power generation; the key feature is screened from the first meteorological feature based on the similarity between the feature and the power generation.
[0071] The beneficial effect of the above technical solution is: after data preprocessing and feature extraction of key data, key features are screened based on the correlation coefficient between features and power generation, providing a data basis for subsequent prediction of power generation of photovoltaic power generation systems.
[0072] The embodiment of the present invention provides a method for predicting power generation of a photovoltaic power generation system, combining and analyzing a first correlation coefficient between each first meteorological feature and power generation and a real-time similarity threshold, and screening the first meteorological feature to obtain key features, including:
[0073] The first meteorological feature whose current first correlation coefficient is greater than the real-time similarity threshold is regarded as the first undetermined feature;
[0074] If the current number of first undetermined features is not less than the set number of features, all the acquired first undetermined features are output as key features;
[0075] If the current number of the first undetermined features is less than the set number of features, then calculate the first number b1 which is the set number of features minus the first number of undetermined features;
[0076] All features except the first undetermined feature in the first meteorological feature are sorted from large to small according to the first correlation coefficient to obtain a feature screening list;
[0077] The first b1 first meteorological features in the feature screening list are extracted and combined with the current first pending feature to be output as key features.
[0078] In this embodiment, the first pending feature refers to the first meteorological feature whose first correlation coefficient is greater than the set similarity threshold; the number of first pending features refers to the number of first pending features; the set number of features refers to 70% of the total number of first meteorological features, and when the obtained number of set features has a decimal point, it is rounded off as the final number of set features. For example, there are currently 7 first meteorological features, and the initial calculated number of set features is 4.9, which has a decimal point. At this time, the final number of set features is determined to be 5.
[0079] In this embodiment, the first number b1 refers to the number of set features minus the number of first undetermined features; the feature screening list is obtained by sorting all features in the first meteorological feature except the first undetermined feature from large to small according to the first correlation coefficient; the key features are screened from the first meteorological features based on the similarity between the features and the power generation.
[0080] The beneficial effect of the above technical solution is: by analyzing the correlation between the first meteorological feature and the power generation, key features are screened from the first meteorological feature, providing a data basis for subsequent prediction of the power generation of the photovoltaic power generation system, thereby ensuring the accuracy of the power generation prediction.
[0081] The embodiment of the present invention provides a method for predicting power generation of a photovoltaic power generation system. The calculation formula of the real-time similarity threshold is as follows:
[0082] Where X1 represents the real-time similarity threshold; Δs avg It is expressed as the average value of the first correlation coefficient between the current first meteorological characteristic and the power generation; s max It is expressed as the maximum value of the first correlation coefficients of all current first meteorological characteristics and power generation; s min It is represented as the minimum value of the first correlation coefficients of all current first meteorological characteristics and power generation; e is represented as the natural base.
[0083] The beneficial effect of the above technical solution is that by calculating the real-time similarity threshold, a strong data basis can be extracted for accurately and quickly screening key features, thereby providing a data basis for the subsequent prediction of the power generation of the photovoltaic power generation system.
[0084] The embodiment of the present invention provides a method for predicting the power generation of a photovoltaic power generation system, which inputs the acquired first target data and key features into a pre-established power prediction model, predicts the power generation of the photovoltaic power generation system, and obtains a pending prediction result, including:
[0085] Extracting historical power generation and corresponding historical meteorological data within a preset time period from a photovoltaic power generation database;
[0086] After preprocessing the acquired historical power generation and historical meteorological data, first reference data is obtained;
[0087] Extracting reference features consistent with the feature category of key features from the first reference data;
[0088] Using the first reference data and the reference feature as training data, training a neural network to obtain a first power prediction model;
[0089] Inputting the acquired first target data and key features into a first power forecasting model to obtain a first pending forecasting result;
[0090] Using the first reference data and the reference features as training data, training the time series to obtain a second power prediction model;
[0091] The acquired first target data and key features are input into a second electricity quantity prediction model to obtain a second pending prediction result.
[0092] In this embodiment, the photovoltaic power generation database is used to store historical system data of the photovoltaic power generation system and corresponding historical meteorological data, wherein the historical system data includes historical power generation, solar panel temperature, inverter efficiency, etc.; the preset time period is predetermined; the types of historical meteorological data include historical solar radiation intensity, historical temperature, historical humidity, historical wind speed, historical wind direction, and historical cloud coverage, etc.
[0093] In this embodiment, the first reference data is obtained by preprocessing the acquired historical power generation and historical meteorological data; the reference feature refers to a feature extracted from the first reference data that is consistent with the feature category of the key feature, wherein the feature category of the key feature includes solar radiation intensity, temperature, humidity, wind speed, cloud coverage, etc.; the first power prediction model is a model obtained by training a neural network using the first reference data and the reference feature as training data; the first pending prediction result is a predicted power generation predicted by the first power prediction model based on the first target data and the key feature; the second power prediction model is a time series model obtained by training the time series using the first reference data and the reference feature as training data; the second pending prediction result is a predicted power generation predicted by the second power prediction model based on the first target data and the key feature.
[0094] The beneficial effect of the above technical solution is: by using the two types of electricity prediction models established to predict the power generation of the photovoltaic power generation system, two prediction results are obtained, and the two prediction results are comprehensively analyzed to obtain the pending prediction result, which can improve the accuracy and reliability of the prediction.
[0095] The embodiment of the present invention provides a method for predicting power generation of a photovoltaic power generation system, which analyzes and adjusts the obtained pending prediction results to determine the target power generation prediction results, including:
[0096] Step 11: if the first pending prediction result is the same as the second pending prediction result, outputting the current first prediction result as the target power generation prediction result;
[0097] Step 12: if the first pending prediction result is different from the second pending prediction result, obtaining an absolute difference in power between the current first pending prediction result and the second pending prediction result;
[0098] If the absolute difference in power is not greater than the set power difference, the average power generation of the first pending prediction result and the second pending prediction result is output as the target power generation prediction result;
[0099] Step 13: If the absolute difference in power is greater than the set power difference, the set optimization strategy is used to optimize the first power prediction model and the second power prediction model;
[0100] Step 14: Re-predict the power generation using the optimized first power prediction model and the second power prediction model to obtain corresponding prediction results;
[0101] Step 15: Repeat steps 11-14 until the maximum number of iterations is reached or the target power generation prediction result is output;
[0102] Step 16: When the set maximum number of iterations is reached and the target power generation prediction result is still not output, the maintenance personnel are notified to perform abnormal maintenance.
[0103] In this embodiment, for example, there is a first pending prediction result q1, a second pending prediction result q2, and q1≠q2. At this time, the absolute difference in power between the first pending prediction result q1 and the second pending prediction result q2 is obtained, and the absolute difference in power |q1-q2| is not greater than the set power difference ΔQ=
[0105] At this time, the average power generation of the first pending prediction result q1 and the second pending prediction result q2 is Output as target power generation prediction result.
[0106] In this embodiment, setting the electricity difference refers to the value of the maximum power generation in the first pending prediction result and the second pending prediction result minus the average power generation of the first pending prediction result and the second pending prediction result; setting the optimization strategy is a predetermined strategy for optimizing the electricity prediction model, such as adjusting model parameters; setting the maximum number of iterations is predetermined; maintenance personnel use it for system maintenance, troubleshooting and repair, and model optimization and updating; the target power generation prediction result is the power generation obtained by analyzing the first pending prediction result and the second pending prediction result.
[0107] The beneficial effect of the above technical solution is: by analyzing the prediction results obtained based on two different power prediction models, the target power generation prediction results are obtained, which can effectively improve the accuracy and reliability of the prediction, reduce the prediction error, and thus facilitate the efficient management of the photovoltaic power generation system.
[0108] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.
Claims
1. A method for predicting power generation of a photovoltaic power generation system, characterized in that: include: Step 1: Use the set monitoring equipment to monitor the current photovoltaic power generation system and the preset weather station to obtain key data and transmit it to the target processing center; Step 2: The target processing center performs data preprocessing and feature extraction on the key data to obtain first target data and key features accordingly; Step 3: Input the acquired first target data and key features into a pre-established power prediction model, perform power generation prediction on the photovoltaic power generation system, and obtain a pending prediction result; Step 4: Analyze and adjust the pending prediction results to determine the target power generation prediction results.
2. A photovoltaic power generation system power generation prediction method according to claim 1, characterized in that: Use the set monitoring equipment to monitor the current photovoltaic power generation system and the preset weather station to obtain key data and transmit it to the target processing center, including: The power monitoring sensor installed at the preset position of the photovoltaic power generation system is used to monitor the power in real time to obtain the real-time power generation; Using a meteorological monitoring sensor at a preset position of a preset meteorological station to monitor meteorological data in real time, to obtain real-time meteorological data; The real-time power generation and real-time meteorological data are transmitted to a target processing center as key data.
3. A photovoltaic power generation system power generation prediction method according to claim 2, characterized in that: Real-time meteorological data include solar radiation intensity, temperature, humidity, wind speed, wind direction and cloud coverage.
4. A photovoltaic power generation system power generation prediction method according to claim 1, characterized in that: The target processing center performs data preprocessing and feature extraction on the key data to obtain first target data and key features, including: The key data is cleaned, converted, integrated and standardized to obtain first target data; Extracting first meteorological features from the first target data, and determining a first correlation coefficient between each first meteorological feature and power generation by using a set similarity algorithm; Determining a real-time similarity threshold according to the obtained first correlation coefficient; The first correlation coefficient between each first meteorological feature and power generation is combined with the real-time similarity threshold for analysis, and the first meteorological features are screened to obtain key features.
5. A photovoltaic power generation system power generation prediction method according to claim 4, characterized in that: The first correlation coefficient between each first meteorological feature and power generation is combined with the real-time similarity threshold for analysis, and the first meteorological feature is screened to obtain key features, including: The first meteorological feature whose current first correlation coefficient is greater than the real-time similarity threshold is regarded as the first undetermined feature; If the current number of first undetermined features is not less than the set number of features, all the acquired first undetermined features are output as key features; If the current number of the first undetermined features is less than the set number of features, then calculate the first number b1 which is the set number of features minus the first number of undetermined features; All features except the first undetermined feature in the first meteorological feature are sorted from large to small according to the first correlation coefficient to obtain a feature screening list; The first b1 first meteorological features in the feature screening list are extracted and combined with the current first pending feature to be output as key features.
6. A photovoltaic power generation system power generation prediction method according to claim 4, characterized in that: The calculation formula of the real-time similarity threshold is as follows: Where X1 represents the real-time similarity threshold; Δs avg It is expressed as the average value of the first correlation coefficient between the current first meteorological characteristic and the power generation; sma x It is expressed as the maximum value of the first correlation coefficients of all current first meteorological characteristics and power generation; smi n It is represented as the minimum value of the first correlation coefficients of all current first meteorological characteristics and power generation; e is represented as the natural base.
7. A photovoltaic power generation system power generation prediction method according to claim 1, characterized in that: The first target data and key features obtained are input into a pre-established power prediction model to predict the power generation of the photovoltaic power generation system and obtain a pending prediction result, including: Extracting historical power generation and corresponding historical meteorological data within a preset time period from a photovoltaic power generation database; After preprocessing the acquired historical power generation and historical meteorological data, first reference data is obtained; Extracting reference features consistent with the feature category of key features from the first reference data; Using the first reference data and the reference feature as training data, training a neural network to obtain a first power prediction model; Inputting the acquired first target data and key features into a first power forecasting model to obtain a first pending forecasting result; Using the first reference data and the reference features as training data, training the time series to obtain a second power prediction model; The acquired first target data and key features are input into a second electricity quantity prediction model to obtain a second pending prediction result.
8. A photovoltaic power generation system power generation prediction method according to claim 1, characterized in that: Adjust the obtained pending forecast results to determine the target power generation forecast results, including: Step 11: if the first pending prediction result is the same as the second pending prediction result, outputting the current first prediction result as the target power generation prediction result; Step 12: if the first pending prediction result is different from the second pending prediction result, obtaining an absolute difference in power between the current first pending prediction result and the second pending prediction result; If the absolute difference in power is not greater than the set power difference, the average power generation of the first pending prediction result and the second pending prediction result is output as the target power generation prediction result; Step 13: If the absolute difference in power is greater than the set power difference, the set optimization strategy is used to optimize the first power prediction model and the second power prediction model; Step 14: Re-predict the power generation using the optimized first power prediction model and the second power prediction model to obtain corresponding prediction results; Step 15: Repeat steps 11-14 until the maximum number of iterations is reached or the target power generation prediction result is output; Step 16: When the set maximum number of iterations is reached and the target power generation prediction result is still not output, the maintenance personnel are notified to perform abnormal maintenance.