Tower air temperature correction method and device based on weather station forecast and computer equipment

Abstract

The invention relates to a tower air temperature correction method and device based on weather station forecast and computer equipment. The method comprises the following steps: acquiring monitored air temperature data and air temperature forecast data of a target power grid in a set historical time period, and acquiring a to-be-trained air temperature prediction model related to a power grid tower at the same time; using the monitored air temperature data and the forecast data as a training set input model, training the model, and generating a trained air temperature prediction model; and evaluating the model in the aspects of a low-temperature interval, a whole-section temperature interval and low-temperature forecasting performance so as to obtain a low-temperature interval index, a whole-section temperature index and a low-temperature forecasting performance index. Based on the evaluation results, different indexes of the model are endowed with combination weights, and various adjusted air temperature prediction models are generated. And selecting the adjustment model with the maximum weight index total amount in all the weight combinations as a final target air temperature prediction model applied to the power grid tower. By adopting the method, the accuracy of predicting the icing of the power grid can be effectively improved.

Description

Technical Field

[0001] The present application relates to the field of smart grid technology, and in particular to a method, device and computer equipment for correcting tower temperature based on weather station forecasts. Background Art

[0002] Refined meteorological forecast data is a prerequisite for accurate icing prediction, and is of great significance for preventing emergency events such as tower collapse, wire breakage, and insulator flashover of transmission lines caused by icing of conductors in winter due to low temperature and freezing weather. In recent years, with the continuous improvement of the accuracy of numerical weather forecasts, it has become the main technical support for meteorological forecasting agencies in various countries. However, there are still large errors in the forecast of meteorological elements such as precipitation and temperature and the actual business application. At present, the research on temperature correction has been relatively rich, such as the model output statistics post-processing (MOS) method, algorithm correction based on machine learning and deep learning, linear regression method, etc. The corrected temperature forecast data has lower root mean square error and higher correlation coefficient than the original forecast.

[0003] Current studies on temperature correction mostly consider the correction of maximum and minimum temperatures, the evaluation of the overall correction and the selection method. There are fewer studies on low temperature conditions (below 0°C) that are the focus of icing prediction. In addition, it has been evaluated that the winter temperature forecast deviation of the power grid icing monitoring terminal is large, which leads to deviations in the prediction of power grid icing. Summary of the invention

[0004] Based on this, it is necessary to provide a tower temperature correction method, device, computer equipment, computer-readable storage medium and computer program product based on meteorological station forecasts, which can effectively improve the accuracy of prediction of power grid icing in response to the above technical problems.

[0005] In a first aspect, the present application provides a tower temperature correction method based on weather station forecasts. The method comprises:

[0006] Acquire the monitored temperature data and temperature forecast data of the target power grid in a preset historical time period, and obtain the to-be-trained temperature prediction model corresponding to the towers in the target power grid;

[0007] Using the monitored temperature data and the temperature forecast data, the temperature prediction model to be trained is trained to obtain a trained temperature prediction model;

[0008] Evaluate the low temperature interval state, the full temperature state and the low temperature forecast performance of the trained temperature prediction model to obtain low temperature interval index information, full temperature index information and low temperature forecast performance index information;

[0009] Assigning different combinations of weights to the low temperature interval index information, the full temperature index information, and the low temperature forecast performance index information, respectively, to obtain a temperature prediction model adjusted by each index;

[0010] The index-adjusted temperature prediction model with the maximum total weight index among the index-adjusted temperature prediction models is selected as the target temperature prediction model corresponding to the tower.

[0011] In a second aspect, the present application also provides a tower temperature correction device based on weather station forecasts. The device comprises:

[0012] A tower data acquisition module is used to obtain the monitored temperature data and temperature forecast data of the target power grid in a preset historical time period, and to obtain the to-be-trained temperature prediction model corresponding to the tower in the target power grid;

[0013] A prediction model training module, used to train the temperature prediction model to be trained using the monitored temperature data and the temperature forecast data to obtain a trained temperature prediction model;

[0014] An index information evaluation module is used to evaluate the low temperature interval state, the full temperature state and the low temperature forecast performance of the trained temperature prediction model, and obtain the low temperature interval index information, the full temperature index information and the low temperature forecast performance index information;

[0015] A prediction model adjustment module is used to assign different combinations of weights to the low temperature interval index information, the full temperature index information and the low temperature forecast performance index information, so as to obtain a temperature prediction model for each index adjustment;

[0016] The prediction model selection module is used to select the index-adjusted temperature prediction model with the maximum total weight index among the index-adjusted temperature prediction models as the target temperature prediction model corresponding to the tower.

[0017] In a third aspect, the present application further provides a computer device. The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:

[0018] Acquire the monitored temperature data and temperature forecast data of the target power grid in a preset historical time period, and obtain the to-be-trained temperature prediction model corresponding to the towers in the target power grid;

[0019] Using the monitored temperature data and the temperature forecast data, the temperature prediction model to be trained is trained to obtain a trained temperature prediction model;

[0020] Evaluate the low temperature interval state, the full temperature state and the low temperature forecast performance of the trained temperature prediction model to obtain low temperature interval index information, full temperature index information and low temperature forecast performance index information;

[0021] Assigning different combinations of weights to the low temperature interval index information, the full temperature index information, and the low temperature forecast performance index information, respectively, to obtain a temperature prediction model adjusted by each index;

[0022] The index-adjusted temperature prediction model with the maximum total weight index among the index-adjusted temperature prediction models is selected as the target temperature prediction model corresponding to the tower.

[0023] In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0024] Acquire the monitored temperature data and temperature forecast data of the target power grid in a preset historical time period, and obtain the to-be-trained temperature prediction model corresponding to the towers in the target power grid;

[0025] Using the monitored temperature data and the temperature forecast data, the temperature prediction model to be trained is trained to obtain a trained temperature prediction model;

[0026] Evaluate the low temperature interval state, the full temperature state and the low temperature forecast performance of the trained temperature prediction model to obtain low temperature interval index information, full temperature index information and low temperature forecast performance index information;

[0027] Assigning different combinations of weights to the low temperature interval index information, the full temperature index information, and the low temperature forecast performance index information, respectively, to obtain a temperature prediction model adjusted by each index;

[0028] The index-adjusted temperature prediction model with the maximum total weight index among the index-adjusted temperature prediction models is selected as the target temperature prediction model corresponding to the tower.

[0029] In a fifth aspect, the present application further provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0030] Acquire the monitored temperature data and temperature forecast data of the target power grid in a preset historical time period, and obtain the to-be-trained temperature prediction model corresponding to the towers in the target power grid;

[0031] Using the monitored temperature data and the temperature forecast data, the temperature prediction model to be trained is trained to obtain a trained temperature prediction model;

[0032] Evaluate the low temperature interval state, the full temperature state and the low temperature forecast performance of the trained temperature prediction model to obtain low temperature interval index information, full temperature index information and low temperature forecast performance index information;

[0033] Assigning different combinations of weights to the low temperature interval index information, the full temperature index information, and the low temperature forecast performance index information, respectively, to obtain a temperature prediction model adjusted by each index;

[0034] The index-adjusted temperature prediction model with the maximum total weight index among the index-adjusted temperature prediction models is selected as the target temperature prediction model corresponding to the tower.

[0035] The above-mentioned tower temperature correction method, device, computer equipment, storage medium and computer program product based on meteorological station forecast obtains the monitored temperature data and temperature forecast data of the target power grid in a preset historical time period, and obtains the to-be-trained temperature prediction model corresponding to the tower in the target power grid; uses the monitored temperature data and temperature forecast data to train the to-be-trained temperature prediction model to obtain a trained temperature prediction model; evaluates the trained temperature prediction model in the low temperature interval state, the full-section temperature state and the low temperature forecast performance, and obtains low temperature interval index information, full-section temperature index information and low temperature forecast performance index information; assigns different combinations of weights to the low temperature interval index information, the full-section temperature index information and the low temperature forecast performance index information, respectively, to obtain each index-adjusted temperature prediction model; selects the index-adjusted temperature prediction model with the maximum total weight index in each index-adjusted temperature prediction model as the target temperature prediction model corresponding to the tower.

[0036] By introducing monitored temperature data and forecast data, dynamic and high-precision forecasting capabilities are provided for different temperature model requirements of towers in the power grid. Through multi-dimensional evaluation of low temperature intervals, full-section temperature states, and low temperature forecasting performance, different weight combinations are assigned to each interval, so that the model can be adaptively adjusted according to the actual needs of different temperature scenarios. The target temperature forecast model finally selected can more accurately adapt to the different temperature conditions of the target power grid. This multi-index, combined weighted optimization model selection strategy ensures the prediction accuracy and robustness under special climatic conditions such as extreme low temperatures. It can not only effectively improve the accuracy of power grid icing prediction, but also improve the response capability and control efficiency of the power grid system under temperature fluctuations. In addition, a statistical correction model is established based on the historical forecast of the meteorological station and the temperature monitoring data of the icing monitoring terminal. It has the advantages of easy acquisition of training data, stable forecast effect of the meteorological station, small sample size and low computational energy consumption. At the same time, the optimal temperature correction model is established for each icing monitoring terminal. Compared with the establishment of a unified correction model for all terminals, the correction effect is better and can be effectively applied in business. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] FIG1 is an application environment diagram of a tower temperature correction method based on weather station forecasts in one embodiment;

[0038] FIG2 is a flow chart of a tower temperature correction method based on weather station forecasts in one embodiment;

[0039] FIG3 is a schematic flow chart of a method for obtaining an index-adjusted temperature prediction model in one embodiment;

[0040] FIG4 is a schematic diagram of a flow chart of a method for obtaining evaluation index weights in one embodiment;

[0041] FIG5 is a schematic flow chart of a method for obtaining an indicator difference coefficient in one embodiment;

[0042] FIG6 is a schematic flow chart of a method for obtaining a standardized matrix of evaluation indicators in one embodiment;

[0043] FIG7 is a schematic diagram of a flow chart of a method for obtaining indicator information in one embodiment;

[0044] FIG8 is a schematic diagram of a technical route for correcting tower temperature based on weather station forecasts in one embodiment;

[0045] FIG9 is a schematic diagram of an overall effect evaluation result of forecast correction in one embodiment;

[0046] FIG10 is a schematic diagram of the terminal temperature correction effect in one embodiment;

[0047] FIG11 is a structural block diagram of a tower temperature correction device based on weather station forecasts in one embodiment;

[0048] FIG. 12 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0050] A tower temperature correction method based on weather station forecast provided by an embodiment of the present application can be applied in an application environment as shown in FIG. 1 . Among them, the terminal 102 communicates with the server 104 through a network. The data storage system can store data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on a cloud or other network server. The server 104 obtains the monitored temperature data and temperature forecast data of the target power grid in a preset historical time period from the terminal 102, and obtains the to-be-trained temperature prediction model corresponding to the tower in the target power grid; the to-be-trained temperature prediction model is trained using the monitored temperature data and the temperature forecast data to obtain a trained temperature prediction model; the trained temperature prediction model is evaluated in the low temperature interval state, the full-section temperature state, and the low temperature forecast performance to obtain low temperature interval index information, full-section temperature index information, and low temperature forecast performance index information; different combinations of weights are assigned to the low temperature interval index information, the full-section temperature index information, and the low temperature forecast performance index information, respectively, to obtain each index adjustment temperature prediction model; the index adjustment temperature prediction model with the maximum total weight index in each index adjustment temperature prediction model is selected as the target temperature prediction model corresponding to the tower. The server 104 may be implemented as an independent server or a server cluster consisting of multiple servers.

[0051] In one embodiment, as shown in FIG2 , a tower temperature correction method based on weather station forecast is provided, which is described by taking the method applied to the server in FIG1 as an example, and includes the following steps:

[0052] Step 202, obtaining the monitored temperature data and temperature forecast data of the target power grid in a preset historical time period, and obtaining the to-be-trained temperature prediction model corresponding to the towers in the target power grid.

[0053] Among them, the monitored temperature data can be the actual temperature information collected in real time by the temperature sensors installed on site, reflecting the temperature changes in the target power grid area during a specific period of time.

[0054] Among them, the temperature forecast data can be the temperature forecast information for a period of time in the future obtained based on meteorological services, which is used to reflect the possible changing trends of temperature in the short term.

[0055] Among them, the temperature prediction model to be trained corresponding to the tower can be the temperature prediction model corresponding to each tower, which has not yet been trained with data of specific regions and climate characteristics.

[0056] Specifically, the temperature monitoring data of ice-covered terminals in the area where the target power grid is located in the past four winters are collected, and the abnormal values ​​(-20℃~30℃) and null values ​​are removed; at the same time, the temperature forecast data of each meteorological station in the area where the target power grid is located in the past four winters are collected for the next three hours (linearly interpolated into hourly) and the monitoring temperature data and temperature forecast data of the three meteorological stations near the tower are interpolated using the inverse distance weighting method to generate the overall monitoring data of the meteorological point corresponding to the tower. The first three years are the training period, and the last year is the verification period. And the temperature prediction model to be trained corresponding to the tower in the target power grid is extracted from the storage medium of the server.

[0057] Step 204, using the monitored temperature data and the temperature forecast data, the temperature prediction model to be trained is trained to obtain a trained temperature prediction model.

[0058] Among them, the trained temperature prediction model can be a model that has been trained with monitored temperature data and forecast data, and whose model parameters and logic have been optimized, and has the ability to predict the future temperature of the target power grid area.

[0059] Specifically, the median of the overall monitoring data of the meteorological points is found (taking the median of 4.7℃ as an example), and the training samples are divided into five thresholds of 0℃, 1℃, 2℃, 3℃, 4℃, and 30℃, and the temperature prediction models to be trained are trained respectively. Since there is only one temperature column in the data features, the commonly used linear regression method is selected. At the same time, due to the small amount of algorithm calculation, each tower is trained separately for its own temperature prediction model to be trained, so that the trained temperature prediction model is more targeted and the correction effect is better.

[0060] Step 206, evaluate the trained temperature prediction model in the low temperature interval state, the full temperature state and the low temperature forecast performance, and obtain the low temperature interval index information, the full temperature index information and the low temperature forecast performance index information.

[0061] Among them, the low temperature interval index information can be the root mean square error of the temperature prediction model's forecast of some temperature intervals under specific low temperature conditions (0°C and below).

[0062] The whole-segment temperature index information may be the root mean square error of the temperature prediction model in all temperature intervals under specific low temperature conditions (0°C and below).

[0063] Among them, the low temperature forecast performance index information can be the accuracy of the temperature prediction model in predicting future low temperature changes under specific low temperature conditions (0°C and below).

[0064] Specifically, the trained temperature prediction model obtained through training is used to evaluate the trained temperature prediction model. In the forecast error of the trained temperature prediction model, RMSE is used to evaluate the part of the monitored temperature at 0°C and below (low temperature interval state) and the full data part (full temperature state), and two indicators of RMSE low temperature and RMSE full amount (low temperature interval indicator information and full temperature indicator information) are obtained, where the formula of RMSE is formula 1.1. At the same time, the effect of low temperature forecast is considered, that is, whether the low temperature situation can be predicted, and the TS score is used for evaluation to obtain the low temperature forecast performance indicator information, where the formula of TS score is shown in formula 1.2.

[0065] (1.1)

[0066] (1.2)

[0067] In the formula, is the total number of samples, and Respectively Temperature forecast data and monitored temperature data; The number of times when both the temperature forecast data and the monitored temperature data are less than 0℃; It is the number of times when the temperature forecast data is less than 0℃ and the monitored temperature data is greater than 0℃ in the low temperature forecast, indicating the number of false alarms; It is the number of times when the temperature forecast data in the low temperature forecast is greater than 0℃ and the monitored temperature data is less than 0℃, indicating the number of missed reports.

[0068] Step 208, assign different combinations of weights to the low temperature interval index information, the full temperature index information, and the low temperature forecast performance index information, to obtain a temperature prediction model for each index adjustment.

[0069] Among them, the indicator-adjusted temperature prediction model can be multiple model versions generated by assigning specific weight combinations to the performance indicators of the trained model under different temperature states, where each model has different priority performances in different temperature ranges or states.

[0070] Specifically, by assigning weights to the calculated low temperature interval index information, the whole section temperature index information, and the low temperature forecast performance index information, the final index information is calculated to be the index of the optimal model corresponding to the tower. This paper gives 6 (m) algorithm models (linear regression methods of 0℃, 1℃, 2℃, 3℃, 4℃, and 30℃) and 3 (n) evaluation indicators to form the original data matrix ,in It represents the value of the i-th algorithm model under the j-th evaluation index. The weight of the index is calculated using the entropy weight method:

[0071] ① Normalization of evaluation index data. For positive indicators such as TS (the larger the index value, the better), use formula 1.3 for normalization, and for negative indicators such as RMSE (the smaller the index value, the better), use formula 1.4 for normalization to obtain the normalized matrix .

[0072] (1.3)

[0073] (1.4)

[0074] ② Calculate the proportion of the next ith sample value of the jth indicator, and calculate the entropy value and difference coefficient of the jth indicator, as shown in formula 1.5, formula 1.6, and formula 1.7 respectively:

[0075] (1.5)

[0076] (1.6)

[0077] (1.7)

[0078] ③Finally calculate the weight of the jth indicator:

[0079] (1.8)

[0080] ④ The final indicators of the temperature prediction model after adjusting each indicator are:

[0081] (1.9)

[0082] Step 210: Select the index-adjusted temperature prediction model with the maximum total weight index among the index-adjusted temperature prediction models as the target temperature prediction model corresponding to the tower.

[0083] Among them, the total amount of weighted indicators can be the sum of the weight values ​​of different temperature states in the temperature prediction model adjusted by each indicator, which represents the total score obtained by the model after comprehensively considering the low temperature range, the full temperature and the low temperature forecast performance.

[0084] Among them, the target temperature prediction model can be the final temperature prediction model selected after optimization and evaluation, which has the best overall prediction ability and regional adaptability.

[0085] Specifically, according to the calculated final indicators, the indicator with the highest final indicator is selected to adjust the temperature prediction model as the target temperature prediction model of the tower as the correction algorithm, and then the business temperature forecast correction is performed. The schematic diagram of the tower temperature correction technology route based on the meteorological station forecast is shown in Figure 8.

[0086] In the above-mentioned tower temperature correction method based on meteorological station forecast, the monitored temperature data and temperature forecast data of the target power grid in a preset historical time period are obtained, and the to-be-trained temperature prediction model corresponding to the tower in the target power grid is obtained; the to-be-trained temperature prediction model is trained using the monitored temperature data and the temperature forecast data to obtain a trained temperature prediction model; the trained temperature prediction model is evaluated in the low temperature interval state, the full-section temperature state and the low temperature forecast performance to obtain low temperature interval index information, full-section temperature index information and low temperature forecast performance index information; different combinations of weights are assigned to the low temperature interval index information, the full-section temperature index information and the low temperature forecast performance index information to obtain each index-adjusted temperature prediction model; the index-adjusted temperature prediction model with the maximum total weight index in each index-adjusted temperature prediction model is selected as the target temperature prediction model corresponding to the tower.

[0087] By introducing monitored temperature data and forecast data, dynamic and high-precision forecasting capabilities are provided for different temperature model requirements of towers in the power grid. Through multi-dimensional evaluation of low temperature intervals, full-section temperature states, and low temperature forecasting performance, different weight combinations are assigned to each interval, so that the model can be adaptively adjusted according to the actual needs of different temperature scenarios. The target temperature forecast model finally selected can more accurately adapt to the different temperature conditions of the target power grid. This multi-index, combined weighted optimization model selection strategy ensures the prediction accuracy and robustness under special climatic conditions such as extreme low temperatures. It can not only effectively improve the accuracy of power grid icing prediction, but also improve the response capability and control efficiency of the power grid system under temperature fluctuations. In addition, a statistical correction model is established based on the historical forecast of the meteorological station and the temperature monitoring data of the icing monitoring terminal. It has the advantages of easy acquisition of training data, stable forecast effect of the meteorological station, small sample size and low computational energy consumption. At the same time, the optimal temperature correction model is established for each icing monitoring terminal. Compared with the establishment of a unified correction model for all terminals, the correction effect is better and can be effectively applied in business.

[0088] In one embodiment, as shown in FIG3 , different combinations of weights are assigned to the low temperature interval index information, the full temperature index information, and the low temperature forecast performance index information, respectively, to obtain a temperature prediction model for each index adjustment, including:

[0089] Step 302, combining several index evaluation algorithm models, low temperature interval index information, full-segment temperature index information and low temperature forecast performance index information to construct an evaluation index raw data matrix.

[0090] Among them, the index evaluation algorithm model can be an algorithm method used to analyze and quantify the performance of the temperature prediction model. Various performance indicators are extracted by processing and evaluating the prediction data of different temperature ranges (such as low temperature range and full temperature range).

[0091] Among them, the original data matrix of the evaluation index can be a matrix containing the evaluation algorithm models of each index, which records the performance of the temperature prediction model in multiple dimensions such as low temperature range, full temperature range and low temperature forecast.

[0092] Specifically, based on the trained temperature prediction model, the system combines several indicator evaluation algorithm models to quantify and organize the low temperature interval indicator information, the full temperature index information, and the low temperature forecast performance indicator information into a matrix format. In order to achieve data standardization, the system converts these indicator information into specific evaluation data through the algorithm model. For example, the indicators of the low temperature interval may include the minimum temperature deviation rate and the extreme low temperature response time, the full temperature state covers the accuracy and stability in different temperature ranges, and the low temperature forecast performance includes the low temperature forecast lead time and error rate, etc. The final raw data matrix carefully covers the multi-dimensional performance of the model, laying a structured foundation for the subsequent weight calculation.

[0093] Step 304: Use the entropy weight method to solve the original data matrix of the evaluation index to obtain the weight of each evaluation index.

[0094] Among them, the evaluation index weight can be the numerical weight assigned to the importance of different performance indicators through methods such as the entropy weight method, which is used to quantify the relative importance of each indicator in the overall model evaluation.

[0095] Specifically, based on the constructed original data matrix of the evaluation indicators, the system applies the entropy weight method to calculate the evaluation indicator weights of each indicator. The entropy weight method calculates the information entropy of each indicator by analyzing the distribution of the indicator data, and assigns weights according to the discreteness (i.e., variability) of the indicator. Indicators with higher discreteness represent greater unique information for model evaluation, and therefore obtain higher weights. For example, if the error rate data of low temperature forecast performance varies greatly, it will receive a higher weight in the construction of the model. The application of the entropy weight method enables the weight allocation to be based on objective data characteristics, so that the weight allocation accurately reflects the importance of each indicator in the model evaluation.

[0096] Step 306, based on the evaluation algorithm model of each indicator as the accumulation condition, according to the weight of each evaluation indicator, each low temperature interval indicator information, the whole section temperature indicator information and the low temperature forecast performance indicator information, obtain the temperature prediction model adjusted by each indicator.

[0097] Among them, the cumulative condition can be a standard used when constructing multiple weighted combination models.

[0098] Specifically, after the weight calculation of each indicator is completed, the system uses each indicator evaluation algorithm model as the accumulation condition, and weightedly calculates each low temperature interval indicator, full-section temperature indicator and low temperature forecast performance indicator by multiplying the weights one by one, and generates an indicator adjustment temperature prediction model with different weight combinations. The weight combination of each model is designed to highlight one or more specific performance priorities. For example, one model may perform best in low temperature interval prediction, while another has higher stability in the full temperature range. Since each candidate model has different performance characteristics, it can be flexibly selected according to the temperature requirements of the actual power grid environment to ensure the best prediction effect.

[0099] In this embodiment, by combining multiple indicator evaluation algorithm models, the indicator information of low temperature interval, full-section temperature and low temperature forecast performance is constructed as an evaluation indicator original data matrix, and the weight is assigned by the entropy weight method, so that the influence of each indicator is objectively determined based on its information volume. On this basis, different weight combinations are used to generate multiple indicators to adjust the temperature prediction model so that the model can perform best in different temperature scenarios. This method ensures the adaptability and accuracy of the temperature prediction model, especially in low temperature extreme weather, and can more accurately predict temperature changes, thereby improving the safety and stability of the power grid under variable climate conditions, and providing scientific support for the efficient operation and maintenance of the power system.

[0100] In one embodiment, as shown in FIG4 , the entropy weight method is used to solve the original data matrix of the evaluation index to obtain the weight of each evaluation index, including:

[0101] Step 402, normalize the data of the original data matrix of the evaluation index to obtain a standardized matrix of the evaluation index.

[0102] The evaluation index standardization matrix may be a matrix obtained by normalizing the index data in the original data matrix.

[0103] Specifically, the data in the original data matrix of the evaluation index is normalized to convert the index data of different dimensions into a relatively uniform numerical range (usually 0 to 1) for effective comparison. The normalization process often uses methods such as range standardization or Z-score standardization to adjust the differences in the values ​​of various indicators to the same order of magnitude, thereby eliminating the impact of differences in measurement units and numerical values ​​between different indicators. The standardized matrix is ​​called the evaluation index standardization matrix, which provides a standardized data basis for subsequent difference analysis.

[0104] Step 404, calculating the difference between any designated evaluation index information and any other evaluation index information in the evaluation index standardization matrix to obtain the difference coefficient of each index.

[0105] Specifically, after obtaining the standardized matrix of evaluation indicators, the indicator difference coefficient of each evaluation indicator will be calculated. Among them, the indicator difference coefficient is used to quantify the degree of difference between each indicator in the standardized matrix and different samples, reflecting the dispersion of the indicator in the data. For example, for each evaluation indicator, its relative difference value relative to the indicator entropy value of each indicator can be calculated. The larger the indicator difference coefficient, the stronger the ability of the indicator to distinguish between different samples, the greater the amount of information contained, and the contribution of the indicator to the model distinction.

[0106] Step 406, calculating the weight of each evaluation indicator according to the difference coefficient of each indicator.

[0107] Specifically, using the index difference coefficient, the system further calculates the weight of each evaluation index. The entropy weight method assigns higher weights to the indexes with larger difference coefficients, because these indexes have larger information content in the index entropy value of each index and have higher evaluation importance. When calculating, all index difference coefficients are added together to obtain the sum of the index difference coefficients, and then the difference coefficient of each index is divided by the sum of the index difference coefficients to achieve normalization, so as to calculate the evaluation index weight of each evaluation index. The weight set allocated in this way not only reflects the objective importance of each index, but also makes the overall evaluation more data-driven, and finally forms a set of weight system to provide a basis for the subsequent optimization of the multi-index model.

[0108] In this embodiment, the original data matrix of the evaluation index is normalized to construct a standardized matrix, ensuring that different indicators are compared within the same magnitude range, eliminating the impact of unit and scale differences. Then, by calculating the difference coefficient between each indicator, the distinguishing effect of each indicator is quantified, and the indicators with large differences between samples are identified, so that the weight allocation is more data-driven by allocating weights through the difference coefficient. The final indicator weights more accurately reflect the relative importance of each indicator in the model evaluation, thereby improving the scientificity and objectivity of the overall evaluation, making the model optimization more in line with actual needs, and providing stronger support for the accurate prediction of the model and the stable operation of the power grid.

[0109] In one embodiment, as shown in FIG5 , the difference between any specified evaluation index information and any other evaluation index information in the evaluation index standardization matrix is ​​calculated to obtain the difference coefficient of each index, including:

[0110] Step 502, calculating the weight of any designated evaluation index information in the evaluation index standardization matrix in all evaluation index information to obtain weight information of each index.

[0111] The indicator weight information may be the relative weight of a certain evaluation indicator in each sample, reflecting the distribution of the indicator in different samples.

[0112] Specifically, the proportion of any indicator in the evaluation indicator standardization matrix in each sample is accurately calculated. The specific method is to divide the standardized value of the indicator in a specific evaluation indicator by the sum of the standardized values ​​of the indicator in all evaluation indicators to obtain the relative proportion of the evaluation indicator in the total of the evaluation indicators. The result of repeating this step is a weight matrix composed of the weight information of each indicator, which shows the relative proportion of each sample in different indicators.

[0113] Step 504, calculating the evaluation index entropy value corresponding to any evaluation index according to the weight information of any index in the evaluation index standardization matrix.

[0114] The entropy value of the evaluation index can be a value calculated based on the index weight information, which is used to measure the information distribution and dispersion of the index among different samples. The higher the entropy value, the more evenly the index is distributed among samples, and the lower the discrimination; the lower the entropy value, the greater the difference between different samples, and the more useful information it contains.

[0115] Specifically, the weight p of the jth indicator in the evaluation index standardization matrix in the i-th sample is ij Take the natural logarithm and multiply it by p ij Multiply them together to get the information entropy contribution value of the proportion. Add the weight entropy contribution values ​​of all samples to get the total information entropy of the jth indicator. For standardization, use −1 / ln(m) as the normalization factor (where m is the number of samples) and multiply the result by this factor to ensure that the entropy value is between 0 and 1. The final e j It reflects the degree of dispersion of the indicator among different samples and is used for the subsequent calculation of the coefficient of difference and weight. The higher the entropy value, the more uniform the distribution of the indicator, that is, the smaller the variation among all samples, and the relatively low amount of information contained; the lower the entropy value, the more significant the difference between the indicator and the greater the amount of information contained.

[0116] Step 506, obtaining the difference coefficient of each indicator according to the evaluation indicator entropy value corresponding to each evaluation indicator.

[0117] Specifically, the entropy value e of the indicator j Subtract from 1, which is g j =1−e j . Index difference coefficient g j It indicates the degree of dispersion of the indicator between different samples, reflecting its information content or differentiation effect. The larger the difference coefficient, the more obvious the difference between samples of the indicator, and the greater its contribution to distinguishing samples. This difference coefficient will be used for subsequent weight allocation, so that indicators with higher information content will receive higher weights.

[0118] In this embodiment, the weight of each specified indicator in the evaluation indicator standardization matrix among all indicators is first calculated to ensure that the relative impact of different indicators on the overall model evaluation is quantified. Then, the entropy value of each indicator is calculated based on the weight information to evaluate the discrete degree of its information distribution, reflecting the distinguishing effect of the indicator between samples. The difference coefficient is further calculated through the entropy value to identify key indicators with high information content and good distinguishing effect. The final difference coefficient can scientifically allocate weights, making the weight allocation more objective and accurate, thereby optimizing the accuracy and reliability of model evaluation and improving the actual application effect of the model in power grid temperature forecasting, especially the response capability to extreme weather.

[0119] In one embodiment, as shown in FIG6 , the data of the original data matrix of the evaluation index is normalized to obtain a standardized matrix of the evaluation index, including:

[0120] Step 602, forward normalize the low temperature interval index information and the full temperature index information in the original data matrix of the evaluation index to obtain the standardized low temperature interval index information and the standardized full temperature index information.

[0121] Among them, the standardized low-temperature interval index information and the standardized full-section temperature index information are the data after the low-temperature interval index and the full-section temperature index are forward standardized, respectively, indicating the performance of the temperature prediction model in a low-temperature environment and at all temperatures.

[0122] Specifically, the low temperature interval index information and the full temperature index information in the original data matrix of the evaluation index are forward standardized. Forward standardization is applicable to the index of "the larger the value, the better" to ensure that higher index values ​​are mapped to higher standardized values. The specific calculation method is: for a certain index value in each sample, subtract the minimum value of the index in all samples, and then divide it by the value range of the index (that is, the maximum value minus the minimum value), so that the standardized value obtained is between 0 and 1. This method scales the values ​​in different samples to the same interval, so that the high index value in each sample also retains a higher relative weight after standardization.

[0123] Step 604, negatively normalize the low temperature forecast performance index information in the original data matrix of the evaluation index to obtain standardized low temperature forecast performance index information.

[0124] Among them, the standardized low temperature forecast performance index information can be the data after negative standardization processing of the low temperature forecast performance index, which reflects the accuracy and reliability of the temperature prediction model in low temperature forecasting.

[0125] Specifically, the low temperature forecast performance indicators are negatively normalized because these indicators usually have the characteristic of "the smaller the value, the better" (such as prediction error). The calculation method of negative normalization is: for a certain indicator value in each sample, the maximum value of the indicator in all samples minus the indicator value of the sample, and then divided by the value range of the indicator (maximum value minus minimum value). Through this process, the original lower indicator value will be mapped to a higher standardized value, thereby maintaining the logical consistency of "the smaller the value, the better", so that the low error of each sample reflects a higher performance level after normalization.

[0126] Step 606, obtaining a standardized matrix of evaluation indicators according to the standardized low temperature interval indicator information, the standardized full-segment temperature indicator information and the standardized low temperature forecast performance indicator information.

[0127] Specifically, after completing the positive and negative normalization of the low temperature interval, the whole temperature and the low temperature forecast performance index, the obtained standardized low temperature interval index information, the standardized whole temperature index information and the standardized low temperature forecast performance index information are integrated into a unified evaluation index standardization matrix. This matrix contains the standardized data of all indicators, unifies the magnitude and direction of different indicators in the interval of 0 to 1, and eliminates the scale difference and directional contradiction in the original data.

[0128] In this embodiment, by positively standardizing the low temperature interval and full-section temperature index information and negatively standardizing the low temperature forecast performance index information, the directionality of different types of indicators is ensured to be consistent, so that the characteristics of "the larger the value, the better" or "the smaller the value, the better" are retained. The standardized matrix of evaluation indicators formed after such processing not only eliminates the dimensional differences between different indicators, but also makes it easier to compare each indicator on the same scale, thereby improving the accuracy and scientificity of the model evaluation. Ultimately, the standardized matrix provides a clear and unified evaluation basis, making the performance of the temperature prediction model in different temperature ranges more reliable, and helping to improve the adaptability and safety of the power grid system under complex climatic conditions.

[0129] In one embodiment, as shown in FIG7 , the trained temperature prediction model is evaluated in the low temperature interval state, the full temperature state, and the low temperature forecast performance to obtain low temperature interval index information, full temperature index information, and low temperature forecast performance index information, including:

[0130] Step 702, based on the monitored temperature data, determine the low temperature range and full temperature range of the trained temperature prediction model.

[0131] Specifically, the temperature range is divided into a low temperature range and a full temperature range using monitored temperature data. Specifically, the low temperature range refers to the temperature range when the temperature is below a certain threshold (such as a set low temperature standard), which is used to evaluate the performance of the model under cold conditions; the full temperature range covers the entire temperature range, from the lowest temperature to the highest temperature, including all seasons and weather conditions. For example, the full temperature range is all temperatures below zero, and the low temperature range is the range from minus 20 degrees to minus 30 degrees in the full temperature range. This division process ensures that the performance of the model can be evaluated in a targeted manner under different temperature conditions, and provides a clear temperature range basis for subsequent steps, so as to more deeply analyze the prediction accuracy of the model under low temperature and overall temperature conditions.

[0132] Step 704, in the low temperature section, the low temperature section state and the low temperature forecast performance of the trained temperature prediction model are evaluated to obtain low temperature section index information and low temperature forecast performance index information.

[0133] Specifically, within the divided low-temperature interval, the system conducts a detailed evaluation of the low-temperature interval state and low-temperature forecast performance of the trained temperature prediction model. The low-temperature interval state evaluation includes calculating the prediction error, deviation range, and deviation fluctuation between the predicted value and the actual value of the model within this temperature interval, focusing on whether the model performs accurately and stably under extreme low temperatures. The low-temperature forecast performance evaluation focuses on the model's ability to predict low-temperature events in advance, such as the prediction accuracy and deviation of the arrival time of low temperatures, and evaluates the model's response speed and prediction accuracy before the actual low-temperature event arrives. The low-temperature interval index information and low-temperature forecast performance index information obtained in this step comprehensively quantify the performance of the model under cold conditions and provide a reference for the safe operation of the power grid under low-temperature conditions.

[0134] Step 706, in the whole temperature range, evaluate the whole temperature state of the trained temperature prediction model to obtain the whole temperature index information.

[0135] Specifically, within the full temperature range, a comprehensive evaluation is conducted on the full temperature state of the trained temperature prediction model, covering the prediction performance within the entire temperature range. This evaluation process includes calculating the average prediction error of the model in different temperature ranges, the prediction stability of each range, and the adaptability of the model under changing temperature conditions. For example, in the temperature changes of spring, summer, autumn and winter, it is analyzed whether the model can maintain consistent prediction accuracy and low volatility to ensure that it can adapt to the challenges brought by different temperature changes. The full-segment temperature index information obtained through this process not only reveals the performance of the model under normal and high temperature conditions, but also provides a comprehensive reference for the temperature prediction of the power grid during the year-round operation, ensuring that the model has sufficient reliability over a wide temperature range.

[0136] In this embodiment, the temperature data is scientifically divided into low-temperature intervals and full-temperature intervals by monitoring the temperature data, so that the performance of the model under different temperature conditions can be accurately evaluated. In the low-temperature interval, the low-temperature interval status and low-temperature forecast performance of the model are evaluated to ensure its prediction accuracy and response speed under extreme low temperatures; in the full-temperature interval, the overall temperature prediction capability of the model is evaluated to ensure that it remains stable in temperature changes throughout the year. This evaluation system effectively improves the adaptability and reliability of the temperature prediction model, especially providing precise support for the operation of the power grid in cold climates and temperature fluctuations throughout the year, thereby enhancing the safety and stability of the power system in dealing with complex climatic conditions.

[0137] In a specific embodiment, Figure 9 is a schematic diagram of the overall effect evaluation results of the correction of the low temperature and full temperature forecast in winter 2023. Taking the four icing processes in winter 2023 as the validation set, the correction effect of the optimal temperature correction model of each terminal is evaluated. It can be seen that the deviation of the low temperature forecast below 0°C is significantly reduced, and the TS score is significantly improved.

[0138] Figure 10 is a schematic diagram of the temperature correction effect at a certain terminal during the four rounds of cold wave processes in 2023. Taking a certain ice monitoring device as an example, the low temperature forecast effect for the next 24 hours during the 2023 winter ice process is analyzed. The average root mean square errors of the low temperature (0d) and full amount (30d) correction algorithms are 0.8℃ and 2.1℃, respectively, which are 3.7℃ and 2.4℃ lower than the original forecast errors, respectively, verifying the effectiveness of the temperature correction.

[0139] It should be understood that, although the steps in the flowcharts involved in the above embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0140] Based on the same inventive concept, the embodiment of the present application also provides a tower temperature correction device based on weather station forecasts for implementing the tower temperature correction method based on weather station forecasts involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more embodiments of the tower temperature correction device based on weather station forecasts provided below can refer to the limitations of the tower temperature correction method based on weather station forecasts above, and will not be repeated here.

[0141] In one embodiment, as shown in FIG11 , a tower temperature correction device based on weather station forecast is provided, comprising: a tower data acquisition module 1102, a prediction model training module 1104, an index information evaluation module 1106, a prediction model adjustment module 1108 and a prediction model selection module 1110, wherein:

[0142] The tower data acquisition module 1102 is used to obtain the monitored temperature data and temperature forecast data of the target power grid in a preset historical time period, and obtain the to-be-trained temperature prediction model corresponding to the tower in the target power grid;

[0143] The prediction model training module 1104 is used to train the temperature prediction model to be trained by using the monitored temperature data and the temperature forecast data to obtain a trained temperature prediction model;

[0144] The index information evaluation module 1106 is used to evaluate the low temperature interval state, the full temperature state and the low temperature forecast performance of the trained temperature prediction model, and obtain the low temperature interval index information, the full temperature index information and the low temperature forecast performance index information;

[0145] The prediction model adjustment module 1108 is used to assign different combinations of weights to the low temperature interval index information, the full temperature index information and the low temperature forecast performance index information, and obtain the temperature prediction model adjusted by each index;

[0146] The prediction model selection module 1110 is used to select the index-adjusted temperature prediction model with the maximum total weight index among the index-adjusted temperature prediction models as the target temperature prediction model corresponding to the tower.

[0147] In one embodiment, the prediction model adjustment module 1108 is also used to combine several index evaluation algorithm models, low temperature interval index information, full-section temperature index information and low temperature forecast performance index information to construct an evaluation index original data matrix; use the entropy weight method to solve the evaluation index original data matrix to obtain the weight of each evaluation index; based on each index evaluation algorithm model as a cumulative condition, according to each evaluation index weight, each low temperature interval index information, full-section temperature index information and low temperature forecast performance index information, obtain each index adjustment temperature prediction model.

[0148] In one embodiment, the prediction model adjustment module 1108 is also used to normalize the data of the evaluation index original data matrix to obtain the evaluation index standardized matrix; calculate the difference between any specified evaluation index information and any other evaluation index information in the evaluation index standardized matrix to obtain the difference coefficient of each indicator; calculate the weight of each evaluation index according to the difference coefficient of each indicator.

[0149] In one embodiment, the prediction model adjustment module 1108 is also used to calculate the proportion of any specified evaluation index information in the evaluation index standardization matrix in all evaluation index information to obtain the proportion information of each index; calculate the evaluation index entropy value corresponding to any evaluation index according to the proportion information of any index in the evaluation index standardization matrix; and obtain the difference coefficient of each index according to the evaluation index entropy value corresponding to each evaluation index.

[0150] In one embodiment, the prediction model adjustment module 1108 is also used to perform positive standardization on the low temperature interval index information and the full-section temperature index information in the original data matrix of the evaluation index to obtain standardized low temperature interval index information and standardized full-section temperature index information; perform negative standardization on the low temperature forecast performance index information in the original data matrix of the evaluation index to obtain standardized low temperature forecast performance index information; and obtain the evaluation index standardized matrix based on the standardized low temperature interval index information, the standardized full-section temperature index information and the standardized low temperature forecast performance index information.

[0151] In one embodiment, the index information evaluation module 1106 is also used to determine the low temperature section and the full temperature section of the trained temperature prediction model based on the monitored temperature data; in the low temperature section, the low temperature section state and the low temperature forecast performance of the trained temperature prediction model are evaluated to obtain the low temperature section index information and the low temperature forecast performance index information; in the full temperature section, the full temperature state of the trained temperature prediction model is evaluated to obtain the full temperature index information.

[0152] Each module in the above-mentioned tower temperature correction device based on weather station forecast can be implemented in whole or in part by software, hardware and their combination. Each of the above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the corresponding operations of each of the above modules.

[0153] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as shown in FIG12. The computer device includes a processor, a memory, and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store server data. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a tower temperature correction method based on a weather station forecast is implemented.

[0154] Those skilled in the art will understand that the structure shown in FIG. 12 is merely a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0155] In one embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiments when executing the computer program.

[0156] In one embodiment, a computer-readable storage medium is provided, storing a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0157] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program includes computer instructions, the computer instructions are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device performs the steps in the above-mentioned method embodiments.

[0158] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0159] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0160] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0161] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A tower temperature correction method based on weather station forecast, characterized in that: The method comprises: Acquire the monitored temperature data and temperature forecast data of the target power grid in a preset historical time period, and obtain the to-be-trained temperature prediction model corresponding to the towers in the target power grid; Using the monitored temperature data and the temperature forecast data, the temperature prediction model to be trained is trained to obtain a trained temperature prediction model; Evaluate the low temperature interval state, the full temperature state and the low temperature forecast performance of the trained temperature prediction model to obtain low temperature interval index information, full temperature index information and low temperature forecast performance index information; Assigning different combinations of weights to the low temperature interval index information, the full temperature index information, and the low temperature forecast performance index information, respectively, to obtain a temperature prediction model adjusted by each index; The index-adjusted temperature prediction model with the maximum total amount of weighted indicators among the index-adjusted temperature prediction models is selected as the target temperature prediction model corresponding to the tower.

2. The method according to claim 1, characterized in that The method of assigning different combinations of weights to the low temperature interval index information, the full temperature index information, and the low temperature forecast performance index information to obtain a temperature prediction model adjusted by each index includes: Combining several index evaluation algorithm models, the low temperature interval index information, the full-segment temperature index information, and the low temperature forecast performance index information, constructing an evaluation index raw data matrix; Use the entropy weight method to solve the original data matrix of the evaluation index to obtain the weight of each evaluation index; Based on each of the indicator evaluation algorithm models as a cumulative condition, according to each of the evaluation indicator weights, each of the low temperature interval indicator information, the full-section temperature indicator information and the low temperature forecast performance indicator information, each of the indicator adjustment temperature prediction models is obtained.

3. The method according to claim 2, characterized in that The entropy weight method is used to solve the original data matrix of the evaluation index to obtain the weight of each evaluation index, including: Normalizing the data of the original data matrix of the evaluation index to obtain a standardized matrix of the evaluation index; Calculate the difference between any specified evaluation index information and any other evaluation index information in the evaluation index standardization matrix to obtain the difference coefficient of each index; The weight of each evaluation indicator is calculated according to the difference coefficient of each indicator.

4. The method according to claim 3, characterized in that The calculating the difference between any specified evaluation index information and any other evaluation index information in the evaluation index standardization matrix to obtain the difference coefficient of each index includes: Calculate the weight of any specified evaluation index information in the evaluation index standardization matrix in all evaluation index information to obtain weight information of each index; Calculate the evaluation index entropy value corresponding to any of the evaluation indicators according to the weight information of any of the indicators in the evaluation index standardization matrix; According to the evaluation index entropy value corresponding to each of the evaluation indexes, the difference coefficient of each of the indexes is obtained.

5. The method according to claim 3, characterized in that: The normalization process of the original data matrix of the evaluation index to obtain a standardized matrix of the evaluation index includes: Performing positive normalization on the low temperature interval index information and the whole temperature index information in the original data matrix of the evaluation index to obtain the standardized low temperature interval index information and the standardized whole temperature index information; Negatively normalizing the low temperature forecast performance index information in the original data matrix of the evaluation index to obtain standardized low temperature forecast performance index information; The evaluation index standardization matrix is ​​obtained according to the standardized low temperature interval index information, the standardized full-section temperature index information and the standardized low temperature forecast performance index information.

6. The method according to any one of claims 1 to 5, characterized in that: The low temperature interval state, the full temperature state and the low temperature forecast performance of the trained temperature prediction model are evaluated to obtain low temperature interval index information, full temperature index information and low temperature forecast performance index information, including: Determine the low temperature segment and the full temperature segment of the trained temperature prediction model according to the monitored temperature data; In the low temperature section, the low temperature section state and the low temperature forecast performance of the trained temperature prediction model are evaluated to obtain the low temperature section index information and the low temperature forecast performance index information; In the full-temperature section, the full-section temperature state of the trained air temperature prediction model is evaluated to obtain the full-section temperature index information.

7. A tower temperature correction device based on weather station forecast, characterized in that: The device comprises: A tower data acquisition module is used to obtain the monitored temperature data and temperature forecast data of the target power grid in a preset historical time period, and to obtain the to-be-trained temperature prediction model corresponding to the tower in the target power grid; A prediction model training module, used to train the temperature prediction model to be trained using the monitored temperature data and the temperature forecast data to obtain a trained temperature prediction model; An index information evaluation module is used to evaluate the trained temperature prediction model in the low temperature interval state, the full temperature state and the low temperature forecast performance, and obtain the low temperature interval index information, the full temperature index information and the low temperature forecast performance index information; A prediction model adjustment module is used to assign different combinations of weights to the low temperature interval index information, the full temperature index information and the low temperature forecast performance index information, so as to obtain an air temperature prediction model adjusted by each index; The prediction model selection module is used to select the index-adjusted temperature prediction model with the maximum total weight index among the index-adjusted temperature prediction models as the target temperature prediction model corresponding to the tower.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images