A photovoltaic power generation power prediction error evaluation method, system, device and medium

Abstract

The application discloses a kind of photovoltaic power generation power prediction error evaluation method, system, equipment and medium, and the present application includes based on target Euclidean distance, the comprehensive score value of photovoltaic power prediction scheme is calculated;According to the comprehensive score value of each photovoltaic power prediction scheme is sorted according to pre-set sorting, and prediction error evaluation result is generated.Solve the individual difference between the prediction model of the existing technology without considering index proximity, resulting in the technical problem that the evaluation effect of photovoltaic prediction scheme is poor.The present application effectively evaluates photovoltaic power prediction scheme, provides reference for whether to select certain prediction scheme;At the same time, since the influence of cloud cluster movement on photovoltaic power characteristics and the individual difference between the prediction model of index proximity are comprehensively considered, the evaluation result is more scientific and reasonable, so that the optimal prediction method can be determined, and the prediction accuracy is improved.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic power generation prediction error assessment technology, and in particular to a method, system, device and medium for photovoltaic power generation prediction error assessment. Background Technology

[0002] Photovoltaic power forecasting refers to using a predictive model to forecast the active power generated by a photovoltaic power station over a future period. Because photovoltaic power generation is affected by terrain, topography, weather, cloud movement, and the operating status of the photovoltaic system, prediction algorithms struggle to accurately reflect the fluctuating characteristics of power generation, inevitably resulting in prediction errors.

[0003] However, current assessments of short-term photovoltaic power forecasting errors primarily focus on the mean error, with evaluation indicators including root mean square error, mean absolute error, average absolute error, and average relative error. These methods, however, only utilize one or more of these indicators in combination, judging the quality of the forecasting method solely by comparing numerical values. They fail to consider individual differences between forecasting models with similar indicators, resulting in poor evaluation of photovoltaic forecasting schemes. Summary of the Invention

[0004] This invention provides a method, system, device, and medium for evaluating photovoltaic power generation prediction errors, which solves the technical problem that existing technologies fail to consider individual differences between prediction models with similar indicators, resulting in poor evaluation of photovoltaic prediction schemes.

[0005] The first aspect of this invention provides a method for evaluating photovoltaic power generation prediction errors, comprising:

[0006] In response to the received prediction error assessment request, the system obtains the prediction data and actual data of each photovoltaic power prediction scheme corresponding to the prediction error assessment request, and generates a sample dataset.

[0007] Calculate the index values ​​for each sample in the sample dataset to generate an index dataset;

[0008] Using the indicator values ​​in the aforementioned indicator dataset, a target evaluation matrix is ​​constructed;

[0009] Calculate the ideal solution of the index matrix in the target evaluation matrix, and calculate the target Euclidean distance between the target evaluation matrix and the ideal solution;

[0010] Based on the target Euclidean distance, calculate the comprehensive score of the photovoltaic power prediction scheme;

[0011] The comprehensive scores of each photovoltaic power prediction scheme are sorted according to a preset order to generate prediction error evaluation results.

[0012] Optionally, the step of calculating the indicator values ​​for each sample in the sample dataset and generating the indicator dataset includes:

[0013] The index values ​​of each sample in the sample dataset are calculated using the primary index in the preset photovoltaic power prediction error evaluation index system.

[0014] A sample dataset is generated using the values ​​of each of the aforementioned indicators.

[0015] Optionally, the primary indicators in the photovoltaic power prediction error assessment index system specifically include:

[0016] Error evaluation indicators are generated using root mean square error, mean absolute error, and mean relative error.

[0017] The forecast assessment indicators are generated using the indicator pass rate, indicator accuracy rate, and indicator correlation coefficient.

[0018] A volatility index is generated by using the drop prediction time and the root mean square error of the drop.

[0019] The error evaluation index, the forecast assessment index, and the volatility index are used to generate the primary index in the photovoltaic power prediction error evaluation index system.

[0020] Optionally, the step of constructing the target evaluation matrix using the indicator values ​​in the indicator dataset includes:

[0021] The indicator values ​​in the indicator dataset are positiveized to generate an updated indicator dataset;

[0022] The initial evaluation matrix is ​​constructed using the updated index dataset;

[0023] Based on the vectorization normalization method, the initial evaluation matrix is ​​normalized to generate the target evaluation matrix.

[0024] Optionally, the ideal solution includes a positive ideal solution and a negative ideal solution; the step of calculating the ideal solution of the index matrix in the target evaluation matrix and calculating the target Euclidean distance between the target evaluation matrix and the ideal solution includes:

[0025] The positive ideal solution is calculated using the maximum value of the index matrix in each column of the target evaluation matrix; wherein the calculation formula for the positive ideal solution is as follows:

[0026]

[0027] The negative ideal solution is calculated using the minimum value of the index matrix in each column of the target evaluation matrix; wherein the calculation formula for the negative ideal solution is as follows:

[0028]

[0029] In the formula, z + For the positive ideal solution, z is the maximum value of the index matrix in each column of the target evaluation matrix; - For a negative ideal solution, The minimum value of the index matrix in each column of the target evaluation matrix;

[0030] Calculate the target Euclidean distance between the target evaluation matrix and the ideal solution.

[0031] Optionally, the step of calculating the target Euclidean distance between the target evaluation matrix and the ideal solution includes:

[0032] Calculate the first Euclidean distance between the target evaluation matrix and the positive ideal solution;

[0033] Calculate the second Euclidean distance between the target evaluation matrix and the negative ideal solution;

[0034] Calculate the sum between the first Euclidean distance and the second Euclidean distance to generate the target Euclidean distance.

[0035] Optionally, the step of calculating the comprehensive score of the photovoltaic power prediction scheme based on the target Euclidean distance includes:

[0036] Calculate the ratio between the second Euclidean distance and the target Euclidean distance to generate a comprehensive score for the photovoltaic power prediction scheme.

[0037] A second aspect of the present invention provides a photovoltaic power generation prediction error assessment system, comprising:

[0038] The sample dataset module is used to respond to the received prediction error assessment request, obtain the prediction data and actual data of each photovoltaic power prediction scheme corresponding to the prediction error assessment request, and generate a sample dataset.

[0039] The indicator dataset module is used to calculate the indicator values ​​of each sample in the sample dataset and generate the indicator dataset.

[0040] The target evaluation matrix module is used to construct a target evaluation matrix using the indicator values ​​in the indicator dataset.

[0041] The target Euclidean distance module is used to calculate the ideal solution of the index matrix in the target evaluation matrix, and to calculate the target Euclidean distance between the target evaluation matrix and the ideal solution;

[0042] The comprehensive scoring module is used to calculate the comprehensive scoring value of the photovoltaic power prediction scheme based on the target Euclidean distance;

[0043] The prediction error assessment result module is used to sort the comprehensive score values ​​of each photovoltaic power prediction scheme according to a preset sorting and generate prediction error assessment results.

[0044] A third aspect of the present invention provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the photovoltaic power generation prediction error assessment method as described in any of the preceding claims.

[0045] The fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed, implements the photovoltaic power generation prediction error assessment method as described in any of the preceding claims.

[0046] As can be seen from the above technical solutions, the present invention has the following advantages:

[0047] This invention, in response to a received prediction error assessment request, acquires the prediction and actual data of each photovoltaic power prediction scheme corresponding to the request, generating a sample dataset; calculates the index values ​​of each sample in the sample dataset, generating an index dataset; constructs a target evaluation matrix using the index values ​​in the index dataset; calculates the ideal solution of the index matrix in the target evaluation matrix, and calculates the target Euclidean distance between the target evaluation matrix and the ideal solution; calculates the comprehensive score of the photovoltaic power prediction scheme based on the target Euclidean distance; and sorts the comprehensive scores of each photovoltaic power prediction scheme according to a preset sorting to generate the prediction error assessment result. This solves the technical problem of existing technologies failing to consider the individual differences between prediction models with similar indices, resulting in poor evaluation performance of photovoltaic prediction schemes.

[0048] This invention effectively evaluates photovoltaic power prediction schemes and provides a reference for deciding whether to select a particular scheme. At the same time, by comprehensively considering the impact of cloud movement on photovoltaic power characteristics and the individual differences between prediction models with similar indicators, the evaluation results are more scientific and reasonable, thereby determining the optimal prediction method and improving prediction accuracy. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of the present 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 only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 This is a flowchart of the steps in a photovoltaic power generation prediction error assessment method provided in Embodiment 1 of the present invention;

[0051] Figure 2 This is a flowchart of the steps in a photovoltaic power generation prediction error assessment method provided in Embodiment 2 of the present invention;

[0052] Figure 3 This is a structural block diagram of a photovoltaic power generation prediction error assessment system provided in Embodiment 3 of the present invention. Detailed Implementation

[0053] This invention provides a method, system, device, and medium for evaluating photovoltaic power generation prediction errors, which solves the technical problem that existing technologies fail to consider individual differences between prediction models with similar indicators, resulting in poor evaluation of photovoltaic prediction schemes.

[0054] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0055] Please see Figure 1 , Figure 1 The flowchart illustrates the steps of a photovoltaic power generation prediction error assessment method provided in Embodiment 1 of the present invention.

[0056] This invention provides a method for evaluating photovoltaic power generation prediction errors, comprising the following steps:

[0057] Step 101: Respond to the received prediction error assessment request, obtain the prediction data and actual data of each photovoltaic power prediction scheme corresponding to the prediction error assessment request, and generate a sample dataset.

[0058] It should be noted that a prediction error assessment request refers to a request to assess the prediction errors of multiple photovoltaic power prediction schemes.

[0059] In the specific implementation, operational data from a photovoltaic power station from 00:00 on January 1, 2022 to 23:55 on November 11, 2022 were selected, with a time resolution of one point every 5 minutes, totaling 90,720 points. Using the station's irradiance, meteorological factors, and historical power from the operational data as input data, and the total power of the entire power station as output power, photovoltaic power prediction models based on linear regression, neural networks, decision trees, limit trees, random forests, and bagging methods were constructed. The photovoltaic power prediction data output from each of these models, along with actual data, were used to generate a sample dataset.

[0060] Step 102: Calculate the index values ​​of each sample in the sample dataset to generate the index dataset.

[0061] It should be noted that a comprehensive evaluation index system for photovoltaic prediction errors that takes into account cloud movement needs to be constructed.

[0062] In practice, the index values ​​of each sample in the sample dataset are calculated using the comprehensive evaluation index system for photovoltaic prediction errors, and the index dataset is obtained by combining all the index data.

[0063] Step 103: Construct the target evaluation matrix using the indicator values ​​in the indicator dataset.

[0064] It should be noted that the target evaluation matrix refers to the evaluation matrix constructed using the processed index values.

[0065] In practice, the indicator values ​​in the indicator dataset are positively oriented to obtain positively oriented indicator values, and the target evaluation matrix is ​​constructed using the positively oriented indicator values.

[0066] Step 104: Calculate the ideal solution of the index matrix in the target evaluation matrix, and calculate the target Euclidean distance between the target evaluation matrix and the ideal solution.

[0067] It should be noted that ideal solutions include positive ideal solutions and negative ideal solutions.

[0068] In practice, the positive and negative ideal solutions of the index values ​​in the target evaluation matrix are calculated, and then the Euclidean distance between the target evaluation matrix and the positive ideal solution, as well as the Euclidean distance between the target evaluation matrix and the negative ideal solution, are calculated.

[0069] Step 105: Calculate the comprehensive score of the photovoltaic power prediction scheme based on the target Euclidean distance.

[0070] It should be noted that the overall score is relative fit.

[0071] In practical implementation, the Euclidean distance between the target evaluation matrix and the negative ideal solution is calculated, and then the ratio of this Euclidean distance to the target Euclidean distance is calculated to obtain the relative fit of the photovoltaic power prediction scheme.

[0072] Step 106: Sort the comprehensive scores of each photovoltaic power prediction scheme according to the preset sorting and generate prediction error evaluation results.

[0073] It should be noted that the prediction error assessment result refers to the ranking of the photovoltaic power prediction schemes from highest to lowest based on all comprehensive evaluation values.

[0074] In practice, by arranging all comprehensive evaluation values ​​from highest to lowest, the merits of the corresponding photovoltaic power prediction schemes can be determined. Then, the photovoltaic power prediction model corresponding to the optimal photovoltaic power prediction scheme is used to predict the photovoltaic power.

[0075] This invention, in response to a received prediction error assessment request, acquires the prediction and actual data of each photovoltaic power prediction scheme corresponding to the request, generating a sample dataset; calculates the index values ​​of each sample in the sample dataset, generating an index dataset; constructs a target evaluation matrix using the index values ​​in the index dataset; calculates the ideal solution of the index matrix in the target evaluation matrix, and calculates the target Euclidean distance between the target evaluation matrix and the ideal solution; calculates the comprehensive score of the photovoltaic power prediction scheme based on the target Euclidean distance; and sorts the comprehensive scores of each photovoltaic power prediction scheme according to a preset sorting to generate the prediction error assessment result. This solves the technical problem of existing technologies failing to consider the individual differences between prediction models with similar indices, resulting in poor evaluation performance of photovoltaic prediction schemes.

[0076] This invention effectively evaluates photovoltaic power prediction schemes and provides a reference for deciding whether to select a particular scheme. At the same time, by comprehensively considering the impact of cloud movement on photovoltaic power characteristics and the individual differences between prediction models with similar indicators, the evaluation results are more scientific and reasonable, thereby determining the optimal prediction method and improving prediction accuracy.

[0077] Please see Figure 2 , Figure 2 The flowchart illustrates the steps of a photovoltaic power generation prediction error assessment method provided in Embodiment 2 of the present invention.

[0078] This invention provides a method for evaluating photovoltaic power generation prediction errors, comprising the following steps:

[0079] Step 201: Respond to the received prediction error assessment request, obtain the prediction data and actual data of each photovoltaic power prediction scheme corresponding to the prediction error assessment request, and generate a sample dataset.

[0080] In this embodiment of the invention, the specific implementation process of step 201 is similar to that of step 101, and will not be repeated here.

[0081] Step 202: Calculate the index values ​​of each sample in the sample dataset using the first-level index in the preset photovoltaic power prediction error evaluation index system.

[0082] It should be noted that the primary indicators refer to error evaluation indicators, forecast assessment indicators, and volatility indicators.

[0083] In practical implementation, based on the existing photovoltaic prediction error assessment indicators, it is also necessary to consider the photovoltaic power plant forecast assessment indicators and the impact of cloud movement on photovoltaic power generation. It is necessary to construct a photovoltaic power prediction error assessment indicator system to calculate the indicator data of each sample in the sample dataset, which can realize a comprehensive and accurate assessment of photovoltaic power prediction schemes and has wide applicability.

[0084] Optionally, the primary indicators in the photovoltaic power prediction error evaluation index system in step 202 specifically include the following steps S11-S14:

[0085] S11. Using root mean square error, mean absolute error and mean relative error, generate error evaluation indicators;

[0086] S12. Use the indicator pass rate, indicator accuracy rate and indicator correlation coefficient to generate forecast assessment and evaluation indicators;

[0087] S13. Using the drop prediction time and the root mean square error of the drop, a volatility index is generated;

[0088] S14. Using error evaluation indicators, forecast assessment indicators, and volatility indicators, generate the primary indicators in the photovoltaic power prediction error assessment indicator system.

[0089] It should be noted that the primary and secondary indicators in the photovoltaic power prediction error assessment index system are shown in Table 1:

[0090]

[0091] Table 1

[0092] Error evaluation indicators include root mean square error, mean absolute error, and mean relative error.

[0093] In practical implementation, the formula for calculating the root mean square error is:

[0094]

[0095] The formula for calculating the mean absolute error is:

[0096]

[0097] The formula for calculating the average relative error is:

[0098]

[0099] In the formula, E RMSE E is the root mean square error. MAE E represents the mean absolute error. MRE The average relative error, g is the number of samples (i.e., the photovoltaic power prediction scheme samples), and P Pp Let P be the predicted power at time p. Mp C is the actual power at time p. p Let p be the power-on capacity at time p.

[0100] It should be noted that the forecast assessment indicators include the indicator pass rate, indicator accuracy rate, and indicator correlation coefficient.

[0101] The formula for calculating the pass rate of the indicators is:

[0102]

[0103]

[0104] The formula for calculating the accuracy of the indicator is:

[0105]

[0106] The formula for calculating the correlation coefficient of an indicator is:

[0107]

[0108] In the formula, Q R For the indicator pass rate, B p A value of 1 indicates that the MAE index is better than the acceptable value. p A value of 0 indicates that the MAE index is worse than the acceptable value. ACC For the accuracy of the indicator, the power rationing time P Mp Let P be the available power at time p, and let P be the power at time when there is no power restriction. Mp Let p be the actual power at time p, and Cap be the total installed capacity of the photovoltaic power station for the current month. r is the correlation coefficient. This represents the average actual power during the error statistics period. This is the average value of the predicted power at the time of error statistics.

[0109] It should be noted that volatility indicators include the predicted drop time and the root mean square error of the drop.

[0110] The formula for calculating the predicted fall time is:

[0111]

[0112] In the formula, t drop The moment when the photovoltaic panels fell due to the influence of cloud formations, t fore To predict when photovoltaic panels will fall due to cloud cover.

[0113] The formula for calculating the root mean square error of the drop is:

[0114]

[0115] In the formula, RMSE drop The root mean square error of the drop prediction. and y drop,h These are the predicted and actual depth values ​​for the h-th drop in photovoltaic power output due to cloud cover, respectively. drop This refers to the frequency of power drop within the call cycle. Based on the rotating door algorithm's fluctuation segment division method, the formula for calculating the depth of photovoltaic power output drop due to cloud influence is as follows:

[0116] y drop =max(P[t drop ,t drop +T drop ])-min(P[t drop ,t drop +T drop (10)

[0117] In the formula, P[t drop ,t drop +T drop ] represents the power sequence of the fluctuating segment, t drop The moment when the photovoltaic output began to decline due to the influence of cloud clusters, T drop For the duration of the fall, max(P[t]) drop ,t drop +T drop ]) is the maximum power during the fluctuation segment, min(P[t) drop ,t drop +T drop ]) represents the minimum power during the fluctuation segment, y drop The depth of the fall.

[0118] Step 203: Use the values ​​of each indicator to generate a sample dataset.

[0119] In practice, a sample dataset is generated by combining all the indicator values.

[0120] Step 204: Construct the target evaluation matrix using the indicator values ​​in the indicator dataset.

[0121] Optionally, step 204 includes the following steps S21-S23:

[0122] S21. Perform positive transformation on the indicator values ​​in the indicator dataset to generate an updated indicator dataset.

[0123] S22. Construct an initial evaluation matrix using the updated indicator dataset;

[0124] S23. Based on the vectorization normalization method, the initial evaluation matrix is ​​normalized to generate the target evaluation matrix.

[0125] It should be noted that in the above steps, in order to eliminate the influence of each indicator value, the indicator values ​​are positively normalized. The calculation formula for the positive normalization is as follows:

[0126] For example, let X be all the data for an indicator, and let x be the elements within it. ij There are n photovoltaic power prediction schemes to be evaluated, i = 1, 2, ..., n, and the set of photovoltaic power prediction schemes is M = [M1, M2, ..., Mn]. n If there are m indicator values, then the indicator dataset is D = [D1, D2, ..., Dm]. m ], j = 1, 2, ..., m.

[0127] 1) Transformation of extremely small indicators into extremely large indicators:

[0128]

[0129] In the formula, For the transformed indicator, x jmax x represents the maximum value of the index before the transformation. ij These are the indicators before the conversion.

[0130] 2) Transformation of intermediate indicators into extremely large indicators:

[0131] If the optimal value is x jbest Then:

[0132]

[0133] 3) Range-type indicators are converted into extremely large indicators:

[0134] For interval-type indicators, assuming the optimal interval is [a, b], the positive transformation formula is:

[0135]

[0136] Where M = max{a - min{x} ij},max{x ij}-b}.

[0137] In practice, it is necessary to construct an initial evaluation matrix from the positively processed indicator dataset (i.e., the updated indicator dataset).

[0138]

[0139] In the formula, This represents the value of the m-th evaluation index for the n-th photovoltaic power prediction scheme.

[0140] In practical implementation, to eliminate the influence of dimensions among the data of various indicators, a vector normalization method is used to normalize the initial evaluation matrix. Normalization yields the target evaluation matrix Z:

[0141]

[0142]

[0143] In the formula, n represents the photovoltaic power prediction scheme to be evaluated, m represents the index value, and z ij Let z be the normalized evaluation index value of the j-th evaluation index for the i-th photovoltaic power prediction scheme to be evaluated. nm This is the normalized evaluation index value of the m-th evaluation index for the n-th photovoltaic power prediction scheme to be evaluated.

[0144] Step 205: Calculate the ideal solution of the index matrix in the target evaluation matrix, and calculate the target Euclidean distance between the target evaluation matrix and the ideal solution.

[0145] Optionally, the ideal solution includes a positive ideal solution and a negative ideal solution; step 205 includes the following steps S31-S33:

[0146] S31. Calculate the positive ideal solution using the maximum value of the index matrix in each column of the target evaluation matrix; the formula for calculating the positive ideal solution is as follows:

[0147]

[0148] S32. Calculate the negative ideal solution using the minimum value of the index matrix in each column of the target evaluation matrix; the formula for calculating the negative ideal solution is as follows:

[0149]

[0150] In the formula, z + For the positive ideal solution, z is the maximum value of the index matrix in each column of the target evaluation matrix; - For a negative ideal solution, The minimum value of the index matrix in each column of the target evaluation matrix;

[0151] S33. Calculate the target Euclidean distance between the target evaluation matrix and the ideal solution.

[0152] In practice, the above steps are followed to calculate the target Euclidean distance between the target evaluation matrix and the ideal solution.

[0153] Optionally, step S33 further includes the following steps S41-S43:

[0154] S41. Calculate the first Euclidean distance between the target evaluation matrix and the positive ideal solution;

[0155] S42. Calculate the second Euclidean distance between the target evaluation matrix and the negative ideal solution;

[0156] S43. Calculate the sum between the first Euclidean distance and the second Euclidean distance to generate the target Euclidean distance.

[0157] It should be noted that both the first Euclidean distance and the second Euclidean distance are Euclidean distances, but they are referred to as the first Euclidean distance and the second Euclidean distance for the purpose of distinction.

[0158] The target Euclidean distance is the sum of the two Euclidean distances.

[0159] In practical implementation, the first Euclidean distance between each index matrix and the positive ideal solution is calculated. The formula for calculating the first Euclidean distance is as follows:

[0160]

[0161] Calculate the second Euclidean distance between each index matrix and the negative ideal solution. The formula for calculating the second Euclidean distance is as follows:

[0162]

[0163] In the formula, Let be the Euclidean distance between the i-th object to be evaluated and the optimal solution. Let be the Euclidean distance between the i-th object to be evaluated and the worst solution.

[0164] Specifically, the two Euclidean distances are added together to obtain the target Euclidean distance.

[0165] Step 206: Calculate the comprehensive score of the photovoltaic power prediction scheme based on the target Euclidean distance.

[0166] Optionally, step 206 includes the following step S51:

[0167] S51. Calculate the ratio between the second Euclidean distance and the target Euclidean distance to generate a comprehensive score for the photovoltaic power prediction scheme.

[0168] It should be noted that the comprehensive score of each photovoltaic power prediction scheme is the relative fit S.

[0169] In practical implementation, the formula for calculating the comprehensive score of each photovoltaic power prediction scheme, i.e., the relative fit S, is as follows:

[0170]

[0171] In the formula, S i Let be the score value of the i-th photovoltaic power prediction scheme to be evaluated.

[0172] In practice, the higher the comprehensive score, the better the photovoltaic power prediction scheme of the evaluated object. Therefore, after solving the comprehensive score of all photovoltaic power prediction schemes, it is necessary to sort all photovoltaic power prediction schemes from high to low according to the comprehensive score to obtain the optimal photovoltaic power prediction scheme.

[0173] Step 207: Sort the comprehensive scores of each photovoltaic power prediction scheme according to the preset sorting and generate prediction error evaluation results.

[0174] In practice, after calculating the comprehensive score of all photovoltaic power prediction schemes, the schemes are ranked from highest to lowest according to their comprehensive scores to obtain the optimal photovoltaic power prediction scheme.

[0175] Specifically, the results of the evaluation of eight photovoltaic power prediction errors based on actual photovoltaic power plant data and six photovoltaic power prediction models are shown in Table 2:

[0176]

[0177] Table 2

[0178] The prediction error evaluation results based on the above six photovoltaic power prediction models are shown in Table 3:

[0179]

[0180] Table 3

[0181] In practical implementation, based on the above comprehensive evaluation results, the bagging method model showed the best predictive performance, followed by the random forest model, while the linear regression model showed the worst predictive performance. The decision tree and limit tree models also achieved a comprehensive evaluation score greater than 0.9, indicating good predictive performance as well.

[0182] This invention, in response to a received prediction error assessment request, acquires the prediction and actual data of each photovoltaic power prediction scheme corresponding to the request, generating a sample dataset; calculates the index values ​​of each sample in the sample dataset, generating an index dataset; constructs a target evaluation matrix using the index values ​​in the index dataset; calculates the ideal solution of the index matrix in the target evaluation matrix, and calculates the target Euclidean distance between the target evaluation matrix and the ideal solution; calculates the comprehensive score of the photovoltaic power prediction scheme based on the target Euclidean distance; and sorts the comprehensive scores of each photovoltaic power prediction scheme according to a preset sorting to generate the prediction error assessment result. This solves the technical problem of existing technologies failing to consider the individual differences between prediction models with similar indices, resulting in poor evaluation performance of photovoltaic prediction schemes.

[0183] This invention effectively evaluates photovoltaic power prediction schemes and provides a reference for deciding whether to select a particular scheme. At the same time, by comprehensively considering the impact of cloud movement on photovoltaic power characteristics and the individual differences between prediction models with similar indicators, the evaluation results are more scientific and reasonable, thereby determining the optimal prediction method and improving prediction accuracy.

[0184] Please see Figure 3 , Figure 3 This is a structural block diagram of a photovoltaic power generation prediction error assessment system provided in Embodiment 3 of the present invention.

[0185] This invention provides a photovoltaic power generation prediction error assessment system, comprising:

[0186] The sample dataset module 301 is used to respond to the received prediction error assessment request, obtain the prediction data and actual data of each photovoltaic power prediction scheme corresponding to the prediction error assessment request, and generate a sample dataset.

[0187] The indicator dataset module 302 is used to calculate the indicator values ​​of each sample in the sample dataset and generate the indicator dataset.

[0188] The target evaluation matrix module 303 is used to construct a target evaluation matrix using the indicator values ​​in the indicator dataset;

[0189] The target Euclidean distance module 304 is used to calculate the ideal solution of the index matrix in the target evaluation matrix, and to calculate the target Euclidean distance between the target evaluation matrix and the ideal solution;

[0190] The comprehensive scoring module 305 is used to calculate the comprehensive scoring value of the photovoltaic power prediction scheme based on the target Euclidean distance;

[0191] The prediction error assessment result module 306 is used to sort the comprehensive score values ​​of each photovoltaic power prediction scheme according to a preset sorting and generate prediction error assessment results.

[0192] Optionally, the indicator dataset module 302 includes:

[0193] The index value submodule is used to calculate the index value of each sample in the sample dataset using the first-level index in the preset photovoltaic power prediction error evaluation index system.

[0194] The sample dataset submodule is used to generate a sample dataset using the values ​​of various indicators.

[0195] Optionally, the primary indicators in the photovoltaic power prediction error assessment index system specifically include:

[0196] The error evaluation index submodule is used to generate error evaluation indices using root mean square error, mean absolute error, and mean relative error.

[0197] The forecast assessment and evaluation index submodule is used to generate forecast assessment and evaluation indicators by using the indicator pass rate, indicator accuracy rate and indicator correlation coefficient.

[0198] The volatility indicator submodule is used to generate volatility indicators by using the predicted drop time and the root mean square error of the drop.

[0199] The primary indicator submodule is used to generate primary indicators in the photovoltaic power prediction error assessment indicator system by using error evaluation indicators, forecast assessment indicators, and volatility indicators.

[0200] Optionally, the target evaluation matrix module 303 includes:

[0201] The Update Indicator Dataset submodule is used to positively process the indicator values ​​in the indicator dataset and generate an updated indicator dataset.

[0202] The initial evaluation matrix submodule is used to construct the initial evaluation matrix using the updated indicator dataset.

[0203] The target evaluation matrix submodule is used to normalize the initial evaluation matrix based on the vector normalization method to generate the target evaluation matrix.

[0204] Optionally, the ideal solution includes a positive ideal solution and a negative ideal solution; the target Euclidean distance module 304 includes:

[0205] The positive ideal solution submodule is used to calculate the positive ideal solution by taking the maximum value of the index matrix in each column of the target evaluation matrix; the calculation formula for the positive ideal solution is as follows:

[0206]

[0207] The negative ideal solution submodule is used to calculate the negative ideal solution by taking the minimum value of the index matrix in each column of the target evaluation matrix; the calculation formula for the negative ideal solution is as follows:

[0208]

[0209] In the formula, z + For the positive ideal solution, z is the maximum value of the index matrix in each column of the target evaluation matrix; - For a negative ideal solution, The minimum value of the index matrix in each column of the target evaluation matrix;

[0210] The target Euclidean distance calculation submodule is used to calculate the target Euclidean distance between the target evaluation matrix and the ideal solution.

[0211] Optionally, the submodule for calculating the target Euclidean distance includes:

[0212] The first Euclidean distance submodule is used to calculate the first Euclidean distance between the target evaluation matrix and the positive ideal solution;

[0213] The second Euclidean distance submodule is used to calculate the second Euclidean distance between the target evaluation matrix and the negative ideal solution;

[0214] The target Euclidean distance submodule is used to calculate the sum between the first Euclidean distance and the second Euclidean distance to generate the target Euclidean distance.

[0215] Optionally, the comprehensive score module 305 includes:

[0216] The comprehensive score submodule is used to calculate the ratio between the second Euclidean distance and the target Euclidean distance to generate a comprehensive score for the photovoltaic power prediction scheme.

[0217] Embodiment 4 of the present invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program; when the computer program is executed by the processor, the processor performs the steps of the photovoltaic power generation prediction error assessment method as described in any of the above embodiments.

[0218] Embodiment 5 of the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed, it implements the photovoltaic power generation prediction error assessment method as described in any of the above embodiments.

[0219] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0220] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0221] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0222] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0223] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0224] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for evaluating the prediction error of photovoltaic power generation, characterized in that, include: In response to the received prediction error assessment request, the system obtains the prediction data and actual data of each photovoltaic power prediction scheme corresponding to the prediction error assessment request, and generates a sample dataset. Calculate the index values ​​for each sample in the sample dataset to generate an index dataset; Using the indicator values ​​in the aforementioned indicator dataset, a target evaluation matrix is ​​constructed; Calculate the ideal solution of the index matrix in the target evaluation matrix, and calculate the target Euclidean distance between the target evaluation matrix and the ideal solution; Based on the target Euclidean distance, calculate the comprehensive score of the photovoltaic power prediction scheme; The comprehensive score values ​​of each photovoltaic power prediction scheme are sorted according to a preset sorting to generate prediction error evaluation results; The step of calculating the indicator values ​​of each sample in the sample dataset and generating the indicator dataset includes: The index values ​​of each sample in the sample dataset are calculated using the primary index in the preset photovoltaic power prediction error evaluation index system. A sample dataset is generated using the values ​​of each of the aforementioned indicators; The primary indicators in the photovoltaic power prediction error assessment index system specifically include: Error evaluation indicators are generated using root mean square error, mean absolute error, and mean relative error. The forecast assessment indicators are generated using the indicator pass rate, indicator accuracy rate, and indicator correlation coefficient. A volatility index is generated by using the drop prediction time and the root mean square error of the drop. Using the error evaluation index, the forecast assessment index, and the volatility index, the primary index in the photovoltaic power prediction error evaluation index system is generated; The volatility indicators include the predicted drop time and the root mean square error of the drop. The formula for calculating the fall prediction time is as follows: In the formula, This refers to the moment when solar panels fell due to the influence of cloud formations. To predict when photovoltaic panels will fall due to cloud cover; The formula for calculating the root mean square error of the drop is: In the formula, The root mean square error of the drop prediction. and These represent the predicted and actual depth values ​​of the h-th drop in photovoltaic power output due to cloud cover. The frequency of drops within the call cycle.

2. The photovoltaic power generation prediction error assessment method according to claim 1, characterized in that, The step of constructing the target evaluation matrix using the indicator values ​​in the indicator dataset includes: The indicator values ​​in the indicator dataset are positiveized to generate an updated indicator dataset; The initial evaluation matrix is ​​constructed using the updated index dataset; Based on the vectorization normalization method, the initial evaluation matrix is ​​normalized to generate the target evaluation matrix.

3. The photovoltaic power generation prediction error assessment method according to claim 1, characterized in that, The ideal solution includes a positive ideal solution and a negative ideal solution; the step of calculating the ideal solution of the index matrix in the target evaluation matrix and calculating the target Euclidean distance between the target evaluation matrix and the ideal solution includes: The positive ideal solution is calculated using the maximum value of the index matrix in each column of the target evaluation matrix; wherein the calculation formula for the positive ideal solution is as follows: The negative ideal solution is calculated using the minimum value of the index matrix in each column of the target evaluation matrix; wherein the calculation formula for the negative ideal solution is as follows: In the formula, For the positive ideal solution, This represents the maximum value of the index matrix in each column of the target evaluation matrix; For a negative ideal solution, The minimum value of the index matrix in each column of the target evaluation matrix; Calculate the target Euclidean distance between the target evaluation matrix and the ideal solution.

4. The photovoltaic power generation prediction error assessment method according to claim 3, characterized in that, The step of calculating the target Euclidean distance between the target evaluation matrix and the ideal solution includes: Calculate the first Euclidean distance between the target evaluation matrix and the positive ideal solution; Calculate the second Euclidean distance between the target evaluation matrix and the negative ideal solution; Calculate the sum between the first Euclidean distance and the second Euclidean distance to generate the target Euclidean distance.

5. The photovoltaic power generation prediction error assessment method according to claim 4, characterized in that, The step of calculating the comprehensive score of the photovoltaic power prediction scheme based on the target Euclidean distance includes: Calculate the ratio between the second Euclidean distance and the target Euclidean distance to generate a comprehensive score for the photovoltaic power prediction scheme.

6. A photovoltaic power generation prediction error assessment system, characterized in that, include: The sample dataset module is used to respond to the received prediction error assessment request, obtain the prediction data and actual data of each photovoltaic power prediction scheme corresponding to the prediction error assessment request, and generate a sample dataset. The indicator dataset module is used to calculate the indicator values ​​of each sample in the sample dataset and generate the indicator dataset. The target evaluation matrix module is used to construct a target evaluation matrix using the indicator values ​​in the indicator dataset. The target Euclidean distance module is used to calculate the ideal solution of the index matrix in the target evaluation matrix, and to calculate the target Euclidean distance between the target evaluation matrix and the ideal solution; The comprehensive scoring module is used to calculate the comprehensive scoring value of the photovoltaic power prediction scheme based on the target Euclidean distance; The prediction error assessment result module is used to sort the comprehensive score values ​​of each photovoltaic power prediction scheme according to a preset sorting and generate prediction error assessment results. The indicator dataset module includes: The index value submodule is used to calculate the index value of each sample in the sample dataset using the first-level index in the preset photovoltaic power prediction error evaluation index system. The sample dataset submodule is used to generate a sample dataset using the values ​​of each of the aforementioned indicators; The primary indicators in the photovoltaic power prediction error assessment index system specifically include: The error evaluation index submodule is used to generate error evaluation indices using root mean square error, mean absolute error, and mean relative error. The forecast assessment and evaluation index submodule is used to generate forecast assessment and evaluation indicators by using the indicator pass rate, indicator accuracy rate and indicator correlation coefficient. The volatility indicator submodule is used to generate volatility indicators by using the predicted drop time and the root mean square error of the drop. The first-level indicator submodule is used to generate the first-level indicators in the photovoltaic power prediction error assessment indicator system by using the error evaluation indicator, the forecast assessment indicator, and the volatility indicator. The volatility indicators include the predicted drop time and the root mean square error of the drop. The formula for calculating the fall prediction time is as follows: In the formula, This refers to the moment when solar panels fell due to the influence of cloud formations. To predict when photovoltaic panels will fall due to cloud cover; The formula for calculating the root mean square error of the drop is: In the formula, The root mean square error of the drop prediction. and These represent the predicted and actual depth values ​​of the h-th drop in photovoltaic power output due to cloud cover. The frequency of drops within the call cycle.

7. An electronic device, characterized in that, The device includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the photovoltaic power generation prediction error assessment method as described in any one of claims 1-5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed, it implements the photovoltaic power generation prediction error assessment method as described in any one of claims 1-5.

Images