Power demand prediction method and related device

Abstract

The invention discloses a power demand prediction method and a related device. The method comprises the following steps: acquiring a plurality of influence factors influencing a power demand; determining an original data set corresponding to each influence factor in the plurality of influence factors in a preset area; performing correlation analysis on the plurality of original data sets through a grey correlation analysis method to screen the plurality of influence factors to obtain target key influence factors; determining data corresponding to the target key influence factor in the plurality of original data sets to obtain a sample data set; training a preset power demand model according to a preset expert experience set and the sample data set to obtain a target power demand prediction model; and inputting a target data set corresponding to the target key influence factor in the preset area into the target power demand prediction model to obtain a target power prediction result. According to the invention, the method achieves the accurate prediction of the power demand.

Description

Technical Field

[0001] The present application relates to the technical field of power demand forecasting, and in particular to a power demand forecasting method and related devices. Background Art

[0002] Electricity demand forecasting refers to the prediction of electricity demand in the future by calculating through models based on historical electricity data and multiple related factors. Accurate electricity demand forecasting is extremely important for the planning, operation and economic benefits of the power industry. As a complex large system closely related to many other areas of society, the long-term change of electricity demand in the power system is a dynamic process that is affected by social, economic and other factors. Therefore, long-term electricity demand forecasting is the basis for formulating power development plans.

[0003] Since the 1980s, scholars have begun to apply modern forecasting theories to the medium- and long-term forecasting of electricity demand. These methods include time series method, grey forecasting method, and fuzzy forecasting method based on fuzzy set theory. The introduction of these theories has made the electricity demand forecasting model more scientific and accurate. However, although these methods have been developed and applied in the academic field, in actual work, there may be certain limitations in forecasting only through models or algorithms.

[0004] Therefore, how to improve the accuracy of electricity demand forecasting has become an urgent problem to be solved. Summary of the invention

[0005] The embodiment of the present application provides a method and related devices for predicting electricity demand, which improves the accuracy of electricity demand prediction by combining an expert experience set with a electricity demand model to construct an electricity demand prediction model.

[0006] In a first aspect, an embodiment of the present application provides a method for predicting power demand, the method comprising:

[0007] Acquire multiple influencing factors affecting electricity demand; the multiple influencing factors include: historical electricity, economy, population, and climate;

[0008] Within a first preset time period, determining an original data set corresponding to each of the multiple influencing factors in a preset area to obtain multiple original data sets;

[0009] Performing correlation analysis on the multiple original data sets by using a grey correlation analysis method to screen the multiple influencing factors and obtain target key influencing factors;

[0010] Determine the data corresponding to the target key influencing factor in the multiple original data sets to obtain a sample data set;

[0011] Obtain preset expert experience set;

[0012] Training a preset power demand model according to the preset expert experience set and the sample data set to obtain a target power demand prediction model;

[0013] Within a second preset time period, a target data set corresponding to the target key influencing factors in the preset area is obtained, and the target data set is input into the target power demand prediction model to obtain a target power prediction result; the start time of the second preset time period is later than the end time of the first preset time period.

[0014] In a second aspect, an embodiment of the present application provides a power demand prediction device, the device comprising: a data acquisition unit, a correlation analysis unit, a model training unit, and a power prediction unit, wherein:

[0015] The data acquisition unit is used to acquire multiple influencing factors that affect power demand; the multiple influencing factors include: historical power, economy, population, and climate;

[0016] The data acquisition unit is further used to determine, within a first preset time period, an original data set corresponding to each of the multiple influencing factors in a preset area to obtain multiple original data sets;

[0017] The association analysis unit is used to perform association analysis on the multiple original data sets by using a grey association analysis method to screen the multiple influencing factors and obtain target key influencing factors;

[0018] The data acquisition unit is further used to determine the data corresponding to the target key influencing factor in the multiple original data sets to obtain a sample data set;

[0019] The data acquisition unit is also used to acquire a preset expert experience set;

[0020] The model training unit is used to train the preset power demand model according to the preset expert experience set and the sample data set to obtain a target power demand prediction model;

[0021] The power prediction unit is used to obtain a target data set corresponding to the target key influencing factors in the preset area within a second preset time period, input the target data set into the target power demand prediction model, and obtain a target power prediction result; the start time of the second preset time period is later than the end time of the first preset time period.

[0022] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a processor, a memory, a communication interface, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the processor, and the program includes instructions for executing the steps in the first aspect of the embodiment of the present application.

[0023] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program for electronic data exchange, wherein the computer program enables a computer to execute part or all of the steps described in the first aspect of the embodiment of the present application.

[0024] In a fifth aspect, an embodiment of the present application provides a computer program product, wherein the computer program product includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to execute some or all of the steps described in the first aspect of the embodiment of the present application. The computer program product may be a software installation package.

[0025] By implementing the embodiments of the present application, multiple influencing factors that affect power demand are obtained; within a first preset time period, the original data set corresponding to each of the multiple influencing factors in a preset area is determined to obtain multiple original data sets; the multiple original data sets are subjected to correlation analysis by the grey correlation analysis method to screen the multiple influencing factors and obtain target key influencing factors; the data corresponding to the target key influencing factors in the multiple original data sets are determined to obtain a sample data set; a preset expert experience set is obtained; a preset power demand model is trained according to the preset expert experience set and the sample data set to obtain a target power demand prediction model; within a second preset time period, a target data set corresponding to the target key influencing factors in the preset area is obtained, and the target data set is input into the target power demand prediction model to obtain a target power prediction result. It can be seen that by combining the expert experience set and the power demand model to construct a power demand prediction model, the accuracy of power demand prediction is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the background technology, the drawings required for use in the embodiments of the present application or the background technology will be described below.

[0027] Figure 1 It is a flow chart of a method for predicting power demand provided in an embodiment of the present application;

[0028] Figure 2 It is a structural schematic diagram of a power demand prediction device provided in an embodiment of the present application;

[0029] Figure 3 It is a structural schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0030] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0031] The terms "first", "second", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices.

[0032] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0033] The following is an explanation of the relevant contents, concepts, meanings, technical issues, technical solutions, beneficial effects, etc. involved in the embodiments of the present application.

[0034] First, some terms involved in this application are explained:

[0035] Grey Relation Analysis: Grey Relation Analysis (GRA) is a statistical method used to evaluate the degree of correlation between various factors in a system. It determines which factors have the greatest impact on a specific result by comparing the similarities or differences between data sequences.

[0036] ARIMA model: ARIMA model is the Autoregressive Integrated Moving Average Model, which is a statistical model used for time series data analysis and prediction. It combines three main elements: autoregression (AR), difference (I) and moving average (MA) to describe the characteristics of time series data. It can be specifically expressed as ARIMA (p, d, q), where p is the order of the autoregressive term, indicating that the current value is related to several past values, d is the difference order, indicating how many differences are needed to obtain a stationary series, and q is the order of the moving average term, indicating that the current value is related to several past forecast errors. The ARIMA model is mainly used to process and predict non-stationary time series data with trend and seasonal characteristics.

[0037] Ant Colony Algorithm: Ant Colony Optimization (ACO) is a computational method that simulates the foraging behavior of ants in nature. It is used to solve optimization problems. It simulates the group behavior of ants to find the optimal solution to the problem. The ant colony algorithm can also solve combinatorial optimization problems.

[0038] See also Figure 1 , Figure 1 : is a flow chart of a method for predicting power demand provided by an embodiment of the present application, the method includes but is not limited to the following steps:

[0039] S101. Obtain multiple influencing factors affecting power demand.

[0040] In an embodiment of the present application, the power demand forecast characterizes the determination of future power consumption. Before predicting the power demand, it is necessary to determine the influencing factors that affect the power demand forecast, wherein the influencing factors may include: historical power, economy, population, climate, power system structure, social living habits, etc., which are not limited here. By collecting and analyzing the data corresponding to these influencing factors, and predicting future power demand based on these data. For example, historical power data (power consumption or power generation) corresponding to historical power influencing factors, past power usage patterns can be determined based on historical power data, thereby predicting future power demand, and economic influencing factors correspond to economic growth data. The increase in economic activities is usually accompanied by an increase in power demand. Therefore, future power demand can also be predicted based on economic influencing factors, etc.

[0041] In a specific embodiment, multiple influencing factors affecting electricity demand are obtained, wherein the multiple influencing factors may include: historical electricity, economy, population, and climate.

[0042] S102: Within a first preset time period, determine an original data set corresponding to each of the multiple influencing factors in a preset area to obtain multiple original data sets.

[0043] Among them, the first preset time period can be a default time period or a pre-set time period. The first preset time period can be the past 5 years or the past 10 years. It can be determined based on prediction efficiency and prediction accuracy. That is, the longer the first preset time period is, the more data points are obtained, the better the quality of the data will be, and the higher the accuracy of the prediction will be. However, the prediction efficiency will also decrease as the number of data points involved in the prediction increases. Therefore, the first preset time period can be determined according to specific prediction requirements. The preset area represents the geographical area where power demand prediction is required. The preset area can be a city, a district, etc., which is not limited here.

[0044] In a specific embodiment, the first preset time period is the past 10 years. During the first preset time period, the original data sets corresponding to each of the multiple influencing factors in the preset area can be collected from the meteorological reports provided by the local power company, data statistics website or meteorological bureau to obtain multiple original data sets, wherein each original data set includes multiple original data, and the original data may include historical power data, economic growth data, demographic data, and climate data. The historical power data represents the monthly power generation data in the past 10 years, the economic growth data represents the quarterly economic growth rate in the past 10 years, the demographic data represents the annual population growth rate in the past 10 years, and the climate data represents the average temperature and precipitation per month in the past 10 years.

[0045] S103 , performing correlation analysis on the multiple original data sets by using a grey correlation analysis method to screen the multiple influencing factors and obtain target key influencing factors.

[0046] In the embodiment of the present application, the target key influencing factor represents a factor that has a significant impact on electricity demand, and the target key influencing factor can be one or more.

[0047] In a specific embodiment, only one influencing factor may be specifically analyzed to determine the power demand predicted under the influencing factor. Specifically, a plurality of original data sets may be subjected to correlation analysis by a grey correlation analysis method to screen a plurality of influencing factors, and an influencing factor may be obtained. The influencing factor is used as a target key influencing factor, and the power demand is predicted according to the target key influencing factor, so that the changing trend of the power demand can be captured more accurately.

[0048] Optionally, the above step S103, performing correlation analysis on the multiple original data sets by using a grey correlation analysis method to screen the multiple influencing factors to obtain target key influencing factors, may include the following steps:

[0049] A31. Acquire electricity consumption data in the preset area within the first preset time period to obtain a reference electricity consumption data set;

[0050] A32. Construct a reference series according to the reference electricity consumption data set;

[0051] A33. Constructing comparison series respectively according to the multiple original data sets to obtain multiple comparison series; each comparison series corresponds to one original data set;

[0052] A34, performing dimensionless processing on the reference number series and the multiple comparison number series respectively to obtain a standard reference number series and multiple standard comparison number series; the dimensionless processing includes one of the following: mean normalization, first row normalization and MIN-MAX normalization;

[0053] A35, determining a grey relational coefficient set corresponding to each standard comparison number series in the multiple standard comparison number series according to the standard reference number series and the multiple standard comparison number series, to obtain multiple grey relational coefficient sets; each grey relational coefficient set includes multiple grey relational coefficients;

[0054] A36, determining the correlation degree corresponding to each standard comparison series in the plurality of standard comparison series according to the plurality of grey correlation coefficient sets, and obtaining a plurality of correlation degrees;

[0055] A37, determining a correlation degree among the multiple correlation degrees that is greater than a preset correlation degree threshold, and obtaining at least one correlation degree;

[0056] A38. Select a maximum correlation degree from the at least one correlation degree, and use the influencing factor corresponding to the maximum correlation degree as the target key influencing factor.

[0057] In a specific embodiment, electricity consumption data of a preset area in a first preset time period is collected to obtain a reference electricity consumption data set, which may be actual electricity consumption data of each month in the preset area, and a reference series is constructed based on the reference electricity consumption data set, the reference series represents a data sequence reflecting the system behavior characteristics (power demand) at each moment. Next, a comparison series is constructed based on a plurality of original data sets to obtain a plurality of comparison series, the comparison series represents a data sequence composed of factors affecting the system behavior at each moment, wherein each comparison series corresponds to an original data set, and the number of data points in each comparison series is equal to the number of data points in the reference series.

[0058] Furthermore, since the physical meanings of the influencing factors in the system are different, the dimensions of the data may not be the same. In order to eliminate this difference, the reference series and the multiple comparison series can be dimensionally processed to obtain a standard reference series and multiple standard comparison series, wherein the dimensionless processing includes one of the following: mean normalization, first row normalization, and MIN-MAX normalization. In a specific embodiment, MIN-MAX normalization can be used to perform dimensionless processing on the reference series and the multiple comparison series.

[0059] Then, according to the standard reference series and multiple standard comparison series, the grey correlation coefficient set corresponding to each standard comparison series in the multiple standard comparison series can be determined to obtain multiple grey correlation coefficient sets, wherein the grey correlation coefficient is an indicator used to measure the degree of correlation between the comparison series and the reference series at a certain moment, and each grey correlation coefficient set includes multiple grey correlation coefficients.

[0060] The calculation formula of grey correlation coefficient is as follows:

[0061]

[0062] In the above formula, ξ(i,j) is the grey correlation coefficient of the jth data point of the i-th standard comparison series; x0 represents the standard reference series; x i represents the i-th standard comparison series; x0(j) represents the j-th data point in the standard reference series; x i (j) represents the jth data point in the i-th standard comparison series; ρ is the resolution coefficient, which ranges from (0 to 1) and is usually 0.5; min i min j |x0(j)-x i (j)| is the minimum difference between the two levels, that is, the difference between the comparison standard series and the standard reference series at each moment, taking the minimum value; max i max j |x0(j)-x i (j)| is the maximum difference between the two levels, that is, the difference between the standard comparison series and the standard reference series at each moment, and the maximum value is taken.

[0063] Next, for each standard comparison series, the average value of the grey correlation coefficient between it and the standard reference series at all data points is calculated to obtain the correlation degree corresponding to the standard comparison series. The correlation degree is a measure of the degree of correlation between the standard comparison series and the standard reference series in the entire time period. Specifically, the correlation degree corresponding to each standard comparison series in multiple standard comparison series can be determined according to multiple grey correlation coefficient sets to obtain multiple correlation degrees. The correlation degree calculation formula is as follows:

[0064]

[0065] In the above formula, r i is the correlation degree of the i-th standard comparison series; n is the number of data points in the standard comparison series; ξ(i,j) is the grey correlation coefficient of the j-th data point of the i-th standard comparison series.

[0066] Further, a correlation degree greater than a preset correlation degree threshold value among multiple correlation degrees is determined to obtain at least one correlation degree, wherein the preset correlation degree threshold value is preset or system default, and the preset correlation degree threshold value is used to evaluate the importance of the influencing factor to the power demand forecast. If. The maximum correlation degree is selected from at least one correlation degree, and the influencing factor corresponding to the maximum correlation degree is taken as the target key influencing factor, the influencing factor with the most significant impact on the power demand can be identified through the grey correlation analysis method, thereby providing a basis for the construction and optimization of the power demand forecasting model, and improving the accuracy and reliability of the forecasting model.

[0067] S104: Determine the data corresponding to the target key influencing factor in the multiple original data sets to obtain a sample data set.

[0068] In a specific embodiment, multiple original data sets are subjected to correlation analysis by grey correlation analysis to screen multiple influencing factors, and target key influencing factors are obtained. Next, the data corresponding to the target key influencing factors in multiple original data sets are determined to obtain a sample data set. If the target key influencing factor is a climate influencing factor, the sample data set may include the average temperature data and corresponding power consumption data for each month in the past 10 years. Based on the sample data set, the power demand forecasting model can be trained and verified to ensure that the model can accurately represent the impact of climate change on power demand.

[0069] S105: Obtain a preset expert experience set.

[0070] The preset expert experience set is a pre-established database including the experience of power planning experts and economic experts. The preset expert experience set includes the experts' understanding of the factors affecting power demand, the judgment and selection of forecasting methods, the experience of parameter adjustment, etc. Based on the preset expert experience set, a power demand forecasting model can be established more accurately.

[0071] It should be noted that before establishing the electricity demand forecasting model according to the preset expert experience set, the preset expert experience set can be converted into specific parameter ranges, weight distribution, forecasting strategies, etc., where the parameter range refers to the possible value intervals of each parameter in the forecasting model. For example, these parameters can be the smoothing coefficient of the time series model, the slope of the trend term, the amplitude of the seasonal component, etc., or the hyperparameters of the model training, etc. The reasonable value range of these parameters can be determined according to the parameter range of the preset expert experience set. For example, if the ARIMA model is used for electricity demand forecasting, the value range of the smoothing coefficient of the ARIMA model can be determined to be (0.2, 0.6) according to the preset expert experience set. Among them, weight allocation refers to the allocation of different weights to different influencing factors in the multi-factor prediction model to reflect the extent of their influence on electricity demand. For example, when constructing an electricity demand prediction model that includes multiple influencing factors such as economy, population and climate, it is determined according to the preset expert experience set that the economic influencing factor is the most important factor affecting electricity demand, so a higher weight (such as 0.5) is assigned to the economic influencing factor, while population and climate also have an impact on electricity demand prediction, but the degree of influence is relatively small, so lower weights (such as 0.3 and 0.2) are assigned respectively. Among them, the prediction strategy refers to the specific methods and steps used in electricity demand prediction, including data preprocessing, model selection, parameter tuning, result verification and other aspects. For example, before predicting electricity demand, it can be determined according to the preset expert experience set that seasonal adjustment and stationarity test of data are the first steps, and then the appropriate prediction model can be selected according to the characteristics of the data. Then, the model parameters need to be tuned using optimization algorithms, and finally, the prediction results are verified and evaluated by methods such as cross-validation.

[0072] In a specific embodiment, expert experience can be collected from relevant literature, reports and case studies, or from existing expert systems and databases, and an expert experience set can be constructed based on the expert experience to provide guidance and support for the development of the power demand forecasting model.

[0073] S106: training a preset power demand model according to the preset expert experience set and the sample data set to obtain a target power demand prediction model.

[0074] In the embodiment of the present application, the preset power demand model may include one of the following: a time series model, a neural network model, a regression model, etc., which is not limited here.

[0075] In a specific embodiment, a time series model is selected as the preset electricity demand model, wherein the preset electricity demand model can specifically be an ARIMA model. Specifically, the model parameters of the preset electricity demand model can be optimized according to a preset expert experience set, and the preset electricity demand model can be trained according to a sample data set until the prediction accuracy and stability of the preset electricity demand model reach the expected goals, thereby obtaining a target electricity demand prediction model.

[0076] Optionally, as the preset expert experience set and the sample data set are updated, the target power demand prediction model may be updated and optimized based on the updated preset expert experience set and the sample data set to maintain the prediction performance of the target power demand prediction model.

[0077] Optionally, the preset expert experience set includes: experience parameters and judgment criteria. The above step S106, training the preset power demand model according to the preset expert experience set and the sample data set to obtain the target power demand prediction model, may include the following steps:

[0078] A61. Perform data preprocessing on the sample data set to obtain a first sample data set; the data preprocessing includes: removing abnormal data, filling missing data, and seasonal adjustment;

[0079] A62. Determine a model parameter value range of the preset power demand model according to the empirical parameters;

[0080] A63. Determine target model parameters of the preset power demand model according to the model parameter value range and the first sample data set;

[0081] A64. Adjusting the initial model parameters of the preset power demand model to the target model parameters to obtain a first prediction model;

[0082] A65. Optimize the first prediction model according to the judgment criterion to obtain the target power demand prediction model.

[0083] Among them, the preset expert experience set may include experience parameters and judgment criteria, the experience parameters include specific parameter ranges, weight distribution, etc., and the judgment criteria include prediction strategies, accuracy determination, etc.

[0084] In a specific embodiment, data preprocessing is performed on a sample data set to obtain a first sample data set, wherein the data preprocessing may include: removing abnormal data, filling missing data, seasonal adjustment, etc. Specifically, a statistical analysis method is used to identify and remove abnormal data points in the sample data set that are far from the average value, and interpolation is used to fill in the missing data points to ensure the integrity of the sample data set. It should be noted that the data in the sample data set are sorted in chronological order. Therefore, if there are missing data points in the sample data set, the filled data can be determined based on the analysis of adjacent data points. Then, the sample data set can be seasonally adjusted to eliminate the impact of seasonal changes on electricity demand, so that the preset electricity demand model can more accurately predict the changing trend of electricity demand.

[0085] Furthermore, the model parameter value range of the preset power demand model can be determined according to the empirical parameters. For example, the reasonable value range of the autoregressive term (p), the difference order (d) and the sliding average term (q) in the ARIMA model can be determined. The parameter value range of different model parameters can be determined according to the empirical parameters. Different parameter combinations are found in these parameter value ranges for testing and verification to find the best parameter combination, that is, the optimal model parameters. Then, the target model parameters of the preset power demand model are determined according to the model parameter value range and the first sample data set. The target model parameters are the optimal model parameters of the preset power demand model. The initial model parameters of the preset power demand model are adjusted to the target model parameters to obtain the first prediction model. Finally, the first prediction model can be optimized according to the judgment criteria to obtain the optimized first prediction model, that is, the target power demand prediction model. The power demand can be accurately predicted according to the target power demand prediction model.

[0086] Optionally, the above step A63, determining the target model parameters of the preset power demand model according to the model parameter value range and the first sample data set, may include the following steps:

[0087] B61. Constructing a parameter search space according to the model parameter value range; the parameter search space includes multiple dimensions, each dimension corresponding to a model parameter;

[0088] B62, searching in the parameter search space by using an ant colony algorithm to obtain multiple model parameter combinations;

[0089] B63. Determine the target model parameters according to the first sample data set and the plurality of model parameter combinations.

[0090] In a specific embodiment, a parameter search space can be constructed according to the value range of the model parameters, wherein the parameter search space includes multiple dimensions, each dimension corresponds to a model parameter, for example, for the p, d, and q parameters in the ARIMA model, the possible value range of p is set to 1 to 5, and there is obvious autocorrelation at these orders, the possible value range of the difference order d is set to 0 to 2 to ensure data stability, and the possible value range of the sliding average term q is set to 0 to 3. Therefore, the three dimensions of the parameter search space are constructed according to these value ranges.

[0091] Furthermore, according to the ant colony algorithm, a plurality of model parameter combinations can be obtained by searching in the parameter search space, and then, the best parameter combination can be determined according to the first sample data set and the plurality of model parameter combinations, and the best parameter combination can be used as the target model parameters of the preset power demand model. For example, for the ARIMA model, when p=2, d=1, q=1, the prediction error of the model is the smallest, that is, the model has the best prediction performance.

[0092] Optionally, the above step B63, determining the target model parameters according to the first sample data set and the plurality of model parameter combinations, may include the following steps:

[0093] C61. Divide the first sample data set into a first training set and a first validation set;

[0094] C62, obtain the target fitness function;

[0095] C63. Determine multiple basic power demand forecasting models according to the multiple model parameter combinations and the preset power demand model; each model parameter combination corresponds to a basic power demand forecasting model;

[0096] C64. Training each of the multiple basic power demand forecasting models according to the first training set to obtain multiple reference power demand forecasting models;

[0097] C65. Determine the performance indicator values ​​of the multiple reference power demand prediction models according to the first verification set to obtain multiple performance indicator values;

[0098] C66. Determine the fitness of the multiple performance indicator values ​​according to the target fitness function to obtain multiple fitnesses; each fitness corresponds to a performance indicator value;

[0099] C67. Select the maximum fitness among the multiple fitnesses, and determine the target model parameters according to the model parameter combination corresponding to the maximum fitness.

[0100] In a specific embodiment, the first sample data set is a sample data set after data preprocessing, which includes data corresponding to the target key influencing factors and power consumption data corresponding to the target key influencing factors. Then, the first sample data set can be divided into a first training set and a first validation set. The first training set is used for model training, and the first validation set is used to evaluate the prediction performance of the model. In order to evaluate and compare the performance of different models, a target fitness function can be defined. For example, the target fitness function can be the inverse of the mean square error. The smaller the calculated fitness, the better the prediction performance of the model. It can be seen that the stability and computational efficiency of the prediction results are evaluated by the target fitness function.

[0101] Furthermore, multiple basic power demand forecasting models can be trained based on multiple model parameter combinations and preset power demand models, and each model parameter combination corresponds to a basic power demand forecasting model. Each of the multiple basic power demand forecasting models is trained based on the first training set to obtain multiple reference power demand forecasting models, and then the performance index values ​​of the multiple reference power demand forecasting models are evaluated based on the first validation set to obtain multiple performance index values. In the embodiment of the present application, the performance index value can be the mean square error, stability, prediction time, etc. of the model, which is not limited here.

[0102] In a specific embodiment, the target fitness function is the inverse of the mean square error, and the fitness of multiple performance index values ​​can be determined according to the target fitness function to obtain multiple fitnesses, wherein each fitness corresponds to a performance index value. The maximum fitness among the multiple fitnesses is selected, and the target model parameters can be determined according to the model parameter combination corresponding to the maximum fitness.

[0103] Optionally, the above step A65, optimizing the first prediction model according to the judgment criterion to obtain the target power demand prediction model, may include the following steps:

[0104] D61, using a cross validation method to validate the first prediction model using the first sample data set to obtain a target evaluation value;

[0105] D62, determining a target evaluation threshold according to the judgment criteria;

[0106] D63. When the target evaluation value is greater than or equal to the target evaluation threshold, determining the first prediction model as the target power demand prediction model;

[0107] D64, when the target evaluation value is less than the target evaluation threshold, determining a target adjustment coefficient corresponding to a model parameter of the first prediction model according to the target evaluation value;

[0108] D65. Adjust the model parameters of the first prediction model according to the target adjustment coefficient to obtain the target power demand prediction model.

[0109] Among them, the cross validation method is a statistical analysis method, which is used to evaluate and improve the prediction performance and generalization ability of the model. In the embodiment of the present application, the cross validation method may include one of the following: leave-out cross validation, K-fold cross validation, leave-one-out cross validation, time series cross validation, etc., which are not limited here.

[0110] In a specific embodiment, a time series cross-validation method is used to verify the first prediction model through a first sample data set to obtain a target evaluation value, wherein the target evaluation value may be an evaluation index such as prediction accuracy and accuracy. In the preset expert experience, a target evaluation threshold may be determined according to a judgment criterion. For example, based on the judgment criterion, it may be determined that the accuracy of the model should be greater than 95%. When the target evaluation value is greater than or equal to the target evaluation threshold, the first prediction model may be determined to be a target power demand prediction model.

[0111] Optionally, when the target evaluation value is less than the target evaluation threshold, the target adjustment coefficient corresponding to the model parameters of the first prediction model can be determined according to the target evaluation value, and the model parameters of the first prediction model can be adjusted according to the target adjustment coefficient to obtain the target power demand prediction model. For example, for the ARIMA model, the target evaluation value is obtained by the cross-validation method, that is, the accuracy is 94%, which is lower than the target evaluation threshold of 95%. The reason for the accuracy being lower than the target evaluation threshold may be that the p value in the ARIMA model is too small. In this case, the p value can be increased accordingly, and retraining can be performed to obtain the target power demand prediction model that can accurately predict power demand.

[0112] S107. Within a second preset time period, obtain a target data set corresponding to the target key influencing factor in the preset area, input the target data set into the target power demand prediction model, and obtain a target power prediction result.

[0113] Among them, the start time of the second preset time period is later than the end time of the first preset time period, the second preset time period can be 1 year or 2 years, the data collected according to the first preset time period is used to establish and train the prediction model, and the data collected according to the second preset time period is used for actual prediction of the prediction model.

[0114] In a specific embodiment, a target data set corresponding to a target key influencing factor in a preset area in a second preset time period is collected. For example, if the target power demand forecasting model is a forecasting model constructed based on climate influencing factors, climate change data in the second preset time period is collected. The target data set is input into the target power demand forecasting model to obtain a target power forecasting result. In the embodiment of the present application, the target power forecasting result may include at least one of the following: a forecast value, a forecast interval, a trend analysis, etc. The power demand required to be forecasted for a certain period of time may be determined based on the target power forecasting result.

[0115] Optionally, the target power forecast results may be presented in the form of charts or reports, and an interactive visualization interface may be developed to present the target power forecast results in an intuitive manner, thereby facilitating viewing and analyzing the forecast results.

[0116] Optionally, the following steps may also be included:

[0117] A71. Obtain data corresponding to the target key influencing factor in the target data set to obtain a first test data set;

[0118] A72. Input the first test data set into the target power demand prediction model to obtain a first power prediction result;

[0119] A73. Adjust the first test data set based on a preset data step size to obtain a second test data set;

[0120] A74. Inputting the second test data set into the target power demand prediction model to obtain a second power prediction result;

[0121] A75. Determine a sensitivity coefficient corresponding to the target key influencing factor according to the preset data step, the first power prediction result, and the second power prediction result;

[0122] A76. Determine the target reliability degree of the target power demand forecasting model according to the sensitivity coefficient; the target reliability degree is used to evaluate the reliability of the target power demand forecasting model under the target key influencing factors; the sensitivity coefficient is negatively correlated with the target reliability degree;

[0123] A77. Determine a target reliability level corresponding to the target reliability degree;

[0124] A78. Determine a target interference coefficient corresponding to the target reliability level;

[0125] A79. Adjust the target power prediction result according to the target interference coefficient to obtain an actual power prediction result.

[0126] Among them, the sensitivity test is to evaluate the impact of changes in the input parameters of the target power demand forecasting model on the model output, so as to determine the sensitivity and importance of these parameters to the forecast results. Through the sensitivity test, the forecast results of the target power demand forecasting model can be further optimized and improved.

[0127] The preset data step size is pre-set or system default, and is used to increase or decrease the value of the model input parameter to change the model output.

[0128] In a specific embodiment, the data corresponding to the target key influencing factor in the target data set is obtained to obtain a first test data set. Specifically, a section of data points in the target data set where the data trend is stable is analyzed, and the first test data set is constructed based on these data points. The first test data set is input into the target power demand forecasting model to obtain a first power forecasting result, the first test data set is adjusted based on a preset data step size to obtain a second test data set, and the second test data set is input into the target power demand forecasting model to obtain a second power forecasting result.

[0129] Furthermore, the sensitivity coefficient corresponding to the target key influencing factor can be determined according to the preset data step size, the first power forecast result and the second power forecast result, that is, based on the step size adjusted by the data, the first power forecast result and the second power forecast result are compared, and the impact of the change in data on the power demand forecast is calculated to obtain the sensitivity coefficient. For example, when the target key influencing factor is a climate influencing factor, the sensitivity coefficient reflects the sensitivity of the power demand forecast result to climate change. The target reliability of the target power demand forecast model can be determined according to the sensitivity coefficient. The target reliability is used to evaluate the reliability of the target power demand forecast model under the target key influencing factor, and the sensitivity coefficient is negatively correlated with the target reliability. The higher the sensitivity coefficient, the lower the reliability of the model.

[0130] Next, determine the target reliability level corresponding to the target reliability degree, where the target reliability level may include: low, medium, and high. According to the target reliability level, the corresponding target interference coefficient may be determined, where the target interference coefficient represents the adjustment range of the power demand forecast result, which is used to adjust the power forecast result to make the power forecast result more in line with the actual forecast situation. By adjusting the target power forecast result according to the target interference coefficient, the actual power forecast result can be obtained. It can be seen that by further optimizing and adjusting the target power forecast result, a more accurate power forecast result can be obtained.

[0131] In summary, by implementing the embodiments of the present application, multiple influencing factors affecting electricity demand are obtained; within a first preset time period, the original data set corresponding to each of the multiple influencing factors in a preset area is determined to obtain multiple original data sets; the multiple original data sets are subjected to correlation analysis by the grey correlation analysis method to screen the multiple influencing factors to obtain target key influencing factors; the data corresponding to the target key influencing factors in the multiple original data sets are determined to obtain a sample data set; a preset expert experience set is obtained; a preset electricity demand model is trained according to the preset expert experience set and the sample data set to obtain a target electricity demand prediction model; within a second preset time period, a target data set corresponding to the target key influencing factors in the preset area is obtained, and the target data set is input into the target electricity demand prediction model to obtain a target electricity prediction result. It can be seen that by combining the expert experience set and the electricity demand model to construct an electricity demand prediction model, the accuracy of electricity demand prediction is improved.

[0132] See also Figure 2 , Figure 2 is a schematic diagram of the structure of a power demand prediction device provided in an embodiment of the present application. The power demand prediction device 200 includes: a data acquisition unit 201, a correlation analysis unit 202, a model training unit 203, and a power prediction unit 204, wherein:

[0133] The data acquisition unit 201 is used to acquire multiple influencing factors that affect power demand; the multiple influencing factors include: historical power, economy, population, and climate;

[0134] The data acquisition unit 201 is further configured to determine, within a first preset time period, an original data set corresponding to each of the multiple influencing factors in a preset area, to obtain multiple original data sets;

[0135] The association analysis unit 202 is used to perform association analysis on the multiple original data sets by using a grey association analysis method to screen the multiple influencing factors and obtain target key influencing factors;

[0136] The data acquisition unit 201 is further used to determine the data corresponding to the target key influencing factor in the multiple original data sets to obtain a sample data set;

[0137] The data acquisition unit 201 is also used to acquire a preset expert experience set;

[0138] The model training unit 203 is used to train the preset power demand model according to the preset expert experience set and the sample data set to obtain a target power demand prediction model;

[0139] The power prediction unit 204 is used to obtain a target data set corresponding to the target key influencing factors in the preset area within a second preset time period, input the target data set into the target power demand prediction model, and obtain a target power prediction result; the start time of the second preset time period is later than the end time of the first preset time period.

[0140] Optionally, in the aspect of performing correlation analysis on the multiple original data sets by using the grey correlation analysis method to screen the multiple influencing factors and obtain target key influencing factors, the correlation analysis unit 202 is further specifically used for:

[0141] Acquiring power consumption data in the preset area within the first preset time period to obtain a reference power consumption data set;

[0142] Constructing a reference series according to the reference electricity consumption data set;

[0143] Constructing comparison series respectively according to the multiple original data sets to obtain multiple comparison series; each comparison series corresponds to an original data set;

[0144] The reference number series and the multiple comparison number series are dimensionally normalized to obtain a standard reference number series and multiple standard comparison number series; the dimensionally normalized processing includes one of the following: mean normalization, first row normalization, and MIN-MAX normalization;

[0145] Determine a grey relational coefficient set corresponding to each standard comparison series in the plurality of standard comparison series according to the standard reference series and the plurality of standard comparison series, and obtain a plurality of grey relational coefficient sets; each grey relational coefficient set includes a plurality of grey relational coefficients;

[0146] Determine the correlation degree corresponding to each standard comparison series in the plurality of standard comparison series according to the plurality of grey correlation coefficient sets to obtain a plurality of correlation degrees;

[0147] Determine a correlation degree among the multiple correlation degrees that is greater than a preset correlation degree threshold, and obtain at least one correlation degree;

[0148] A maximum degree of association is selected from the at least one degree of association, and an influencing factor corresponding to the maximum degree of association is used as the target key influencing factor.

[0149] Optionally, the preset expert experience set includes: experience parameters and judgment criteria; in the aspect of training the preset power demand model according to the preset expert experience set and the sample data set to obtain the target power demand prediction model, the model training unit 203 is further specifically used for:

[0150] Performing data preprocessing on the sample data set to obtain a first sample data set; the data preprocessing includes: removing abnormal data, filling missing data, and seasonal adjustment;

[0151] Determining a model parameter value range of the preset power demand model according to the empirical parameters;

[0152] Determining target model parameters of the preset power demand model according to the model parameter value range and the first sample data set;

[0153] Adjusting the initial model parameters of the preset power demand model to the target model parameters to obtain a first prediction model;

[0154] The first prediction model is optimized according to the judgment criterion to obtain the target power demand prediction model.

[0155] Optionally, in determining the target model parameters of the preset power demand model according to the model parameter value range and the first sample data set, the model training unit 203 is further specifically used to:

[0156] Constructing a parameter search space according to the model parameter value range; the parameter search space includes multiple dimensions, each dimension corresponding to a model parameter;

[0157] Searching in the parameter search space by using an ant colony algorithm to obtain multiple model parameter combinations;

[0158] The target model parameters are determined according to the first sample data set and the plurality of model parameter combinations.

[0159] Optionally, in determining the target model parameters according to the first sample data set and the plurality of model parameter combinations, the model training unit 203 is further specifically configured to:

[0160] Dividing the first sample data set into a first training set and a first validation set;

[0161] Get the target fitness function;

[0162] Determine a plurality of basic power demand forecasting models according to the plurality of model parameter combinations and the preset power demand model; each model parameter combination corresponds to a basic power demand forecasting model;

[0163] Training each of the multiple basic power demand forecasting models according to the first training set to obtain multiple reference power demand forecasting models;

[0164] Determine the performance indicator values ​​of the multiple reference power demand prediction models according to the first verification set to obtain multiple performance indicator values;

[0165] Determine the fitness of the multiple performance indicator values ​​according to the target fitness function to obtain multiple fitnesses; each fitness corresponds to a performance indicator value;

[0166] A maximum fitness among the multiple fitnesses is selected, and the target model parameters are determined according to a combination of model parameters corresponding to the maximum fitness.

[0167] Optionally, in optimizing the first prediction model according to the judgment criterion to obtain the target power demand prediction model, the model training unit 203 is further specifically used for:

[0168] Using a cross validation method to validate the first prediction model using the first sample data set to obtain a target evaluation value;

[0169] Determine a target evaluation threshold according to the judgment criteria;

[0170] When the target evaluation value is greater than or equal to the target evaluation threshold, determining the first prediction model as the target power demand prediction model;

[0171] When the target evaluation value is less than the target evaluation threshold, determining a target adjustment coefficient corresponding to a model parameter of the first prediction model according to the target evaluation value;

[0172] The model parameters of the first prediction model are adjusted according to the target adjustment coefficient to obtain the target power demand prediction model.

[0173] Optionally, the power demand prediction device 200 is further specifically used for:

[0174] Acquire data corresponding to the target key influencing factor in the target data set to obtain a first test data set;

[0175] Inputting the first test data set into the target power demand prediction model to obtain a first power prediction result;

[0176] Adjusting the first test data set based on a preset data step size to obtain a second test data set;

[0177] Inputting the second test data set into the target power demand prediction model to obtain a second power prediction result;

[0178] Determining a sensitivity coefficient corresponding to the target key influencing factor according to the preset data step size, the first power prediction result, and the second power prediction result;

[0179] Determining a target reliability degree of the target power demand forecasting model according to the sensitivity coefficient; the target reliability degree is used to evaluate the reliability of the target power demand forecasting model under the target key influencing factors; the sensitivity coefficient is negatively correlated with the target reliability degree;

[0180] Determining a target reliability level corresponding to the target reliability degree;

[0181] Determining a target interference coefficient corresponding to the target reliability level;

[0182] The target power prediction result is adjusted according to the target interference coefficient to obtain an actual power prediction result.

[0183] The power demand forecasting device 200 described in the present application can obtain multiple influencing factors that affect power demand; within a first preset time period, determine the original data set corresponding to each of the multiple influencing factors in a preset area to obtain multiple original data sets; perform correlation analysis on the multiple original data sets by the gray correlation analysis method to screen the multiple influencing factors to obtain target key influencing factors; determine the data corresponding to the target key influencing factors in the multiple original data sets to obtain a sample data set; obtain a preset expert experience set; train a preset power demand model based on the preset expert experience set and the sample data set to obtain a target power demand forecasting model; within a second preset time period, obtain a target data set corresponding to the target key influencing factors in the preset area, input the target data set into the target power demand forecasting model, and obtain a target power forecasting result. It can be seen that by combining the expert experience set and the power demand model to construct a power demand forecasting model, the accuracy of power demand forecasting is improved.

[0184] See also Figure 3 , Figure 3 : is a structural diagram of an electronic device provided in an embodiment of the present application, the electronic device may include a processor, a memory, a communication interface and one or more programs, the processor, the memory and the communication interface may be interconnected through a bus; the one or more programs are stored in the memory and configured to be executed by the processor; in the embodiment of the present application, the program includes instructions for executing the following steps:

[0185] Acquire multiple influencing factors affecting electricity demand; the multiple influencing factors include: historical electricity, economy, population, and climate;

[0186] Within a first preset time period, determining an original data set corresponding to each of the multiple influencing factors in a preset area to obtain multiple original data sets;

[0187] Performing correlation analysis on the multiple original data sets by using a grey correlation analysis method to screen the multiple influencing factors and obtain target key influencing factors;

[0188] Determine the data corresponding to the target key influencing factor in the multiple original data sets to obtain a sample data set;

[0189] Obtain preset expert experience set;

[0190] Training a preset power demand model according to the preset expert experience set and the sample data set to obtain a target power demand prediction model;

[0191] Within a second preset time period, a target data set corresponding to the target key influencing factors in the preset area is obtained, and the target data set is input into the target power demand prediction model to obtain a target power prediction result; the start time of the second preset time period is later than the end time of the first preset time period.

[0192] The electronic device described in the present application can obtain multiple influencing factors that affect power demand; within a first preset time period, determine the original data set corresponding to each of the multiple influencing factors in a preset area to obtain multiple original data sets; perform correlation analysis on the multiple original data sets by gray correlation analysis to screen the multiple influencing factors and obtain target key influencing factors; determine the data corresponding to the target key influencing factors in the multiple original data sets to obtain a sample data set; obtain a preset expert experience set; train a preset power demand model based on the preset expert experience set and the sample data set to obtain a target power demand prediction model; within a second preset time period, obtain a target data set corresponding to the target key influencing factor in the preset area, input the target data set into the target power demand prediction model, and obtain a target power prediction result. It can be seen that by combining the expert experience set and the power demand model to construct a power demand prediction model, the accuracy of power demand prediction is improved.

[0193] An embodiment of the present application also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program for electronic data exchange, wherein the computer program enables a computer to execute part or all of the steps of any method described in the above method embodiments, and the above computer includes an electronic device.

[0194] The embodiment of the present application also provides a computer program product, the computer program product includes a non-transitory computer-readable storage medium storing a computer program, the computer program is operable to cause a computer to execute some or all of the steps of any method described in the method embodiment. The computer program product may be a software installation package, and the computer includes an electronic device.

[0195] Those skilled in the art can understand that to implement all or part of the processes in the above-mentioned embodiments, the processes can be completed by computer programs to instruct related hardware, and the programs can be stored in computer-readable storage media. When the programs are executed, they can include the processes of the above-mentioned method embodiments. The aforementioned storage media include: ROM or random access memory RAM, magnetic disk or optical disk and other media that can store program codes.

[0196] The steps of the method or algorithm described in the embodiments of the present application can be implemented in hardware or by executing software instructions by a processor. The software instructions can be composed of corresponding software modules, and the software modules can be stored in RAM, flash memory, ROM, EPROM, electrically erasable programmable read-only memory (electrically EPROM, EEPROM), registers, hard disks, mobile hard disks, read-only compact disks (CD-ROMs) or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can be located in an ASIC. In addition, the ASIC can be located in a terminal device or a management device. Of course, the processor and the storage medium can also be present in a terminal device or a management device as discrete components.

[0197] Those skilled in the art should be aware that in one or more of the above examples, the functions described in the embodiments of the present application can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiments of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, computer, server, or data center to another website site, computer, server, or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (digital subscriber line, DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrations. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a digital video disc (DVD)), or a semiconductor medium (eg, a solid state disk (SSD)).

[0198] The modules / units included in the devices and products described in the above embodiments may be software modules / units or hardware modules / units, or may be partially software modules / units and partially hardware modules / units. For example, for the devices and products applied to or integrated in the chip, the modules / units included therein may all be implemented in the form of hardware such as circuits, or at least some of the modules / units may be implemented in the form of software programs, which run on the processor integrated inside the chip, and the remaining (if any) modules / units may be implemented in the form of hardware such as circuits; for the devices and products applied to or integrated in the chip module, the modules / units included therein may all be implemented in the form of hardware such as circuits, and different modules / units may be located in the same component (such as a chip, circuit module, etc.) or in different components of the chip module, or at least some of the modules / units may be implemented in the form of software programs. The software programs run on the processor integrated inside the chip, and the remaining (if any) modules / units may be implemented in the form of hardware such as circuits. It is implemented in the form of a software program, which runs on a processor integrated inside the chip module, and the remaining (if any) modules / units can be implemented in hardware such as circuits; for various devices and products applied to or integrated in the terminal equipment, the various modules / units contained therein can be implemented in hardware such as circuits, and different modules / units can be located in the same component (for example, chip, circuit module, etc.) or in different components in the terminal equipment, or, at least some modules / units can be implemented in the form of a software program, which runs on a processor integrated inside the terminal equipment, and the remaining (if any) modules / units can be implemented in hardware such as circuits.

[0199] The specific implementation methods described above further illustrate the purpose, technical solutions and beneficial effects of the embodiments of the present application. It should be understood that the above description is only the specific implementation method of the embodiments of the present application and is not intended to limit the protection scope of the embodiments of the present application. Any modifications, equivalent substitutions, improvements, etc. made on the basis of the technical solutions of the embodiments of the present application should be included in the protection scope of the embodiments of the present application.

Claims

1. A method for predicting power demand, characterized in that: The method comprises: Acquire multiple influencing factors affecting electricity demand; the multiple influencing factors include: historical electricity, economy, population, and climate; Within a first preset time period, determining an original data set corresponding to each of the multiple influencing factors in a preset area to obtain multiple original data sets; Performing correlation analysis on the multiple original data sets by using a grey correlation analysis method to screen the multiple influencing factors and obtain target key influencing factors; Determine the data corresponding to the target key influencing factor in the multiple original data sets to obtain a sample data set; Obtain preset expert experience set; Training a preset power demand model according to the preset expert experience set and the sample data set to obtain a target power demand prediction model; Within a second preset time period, a target data set corresponding to the target key influencing factors in the preset area is obtained, and the target data set is input into the target power demand prediction model to obtain a target power prediction result; the start time of the second preset time period is later than the end time of the first preset time period.

2. The method according to claim 1, characterized in that The grey correlation analysis method is used to perform correlation analysis on the multiple original data sets to screen the multiple influencing factors to obtain target key influencing factors, including: Acquiring power consumption data in the preset area within the first preset time period to obtain a reference power consumption data set; Constructing a reference series according to the reference electricity consumption data set; Constructing comparison series respectively according to the multiple original data sets to obtain multiple comparison series; each comparison series corresponds to an original data set; The reference number series and the multiple comparison number series are dimensionally normalized to obtain a standard reference number series and multiple standard comparison number series; the dimensionally normalized processing includes one of the following: mean normalization, first row normalization, and MIN-MAX normalization; Determine a grey relational coefficient set corresponding to each standard comparison series in the plurality of standard comparison series according to the standard reference series and the plurality of standard comparison series, and obtain a plurality of grey relational coefficient sets; each grey relational coefficient set includes a plurality of grey relational coefficients; Determine the correlation degree corresponding to each standard comparison series in the plurality of standard comparison series according to the plurality of grey correlation coefficient sets to obtain a plurality of correlation degrees; Determine a correlation degree among the multiple correlation degrees that is greater than a preset correlation degree threshold, and obtain at least one correlation degree; A maximum degree of association is selected from the at least one degree of association, and an influencing factor corresponding to the maximum degree of association is used as the target key influencing factor.

3. The method according to claim 1, characterized in that The preset expert experience set includes: experience parameters and judgment criteria; the preset power demand model is trained according to the preset expert experience set and the sample data set to obtain a target power demand prediction model, including: Performing data preprocessing on the sample data set to obtain a first sample data set; the data preprocessing includes: removing abnormal data, filling missing data, and seasonal adjustment; Determining a model parameter value range of the preset power demand model according to the empirical parameters; Determining target model parameters of the preset power demand model according to the model parameter value range and the first sample data set; Adjusting the initial model parameters of the preset power demand model to the target model parameters to obtain a first prediction model; The first prediction model is optimized according to the judgment criterion to obtain the target power demand prediction model.

4. The method according to claim 3, characterized in that The determining the target model parameters of the preset power demand model according to the model parameter value range and the first sample data set includes: Constructing a parameter search space according to the model parameter value range; the parameter search space includes multiple dimensions, each dimension corresponding to a model parameter; Searching in the parameter search space by using an ant colony algorithm to obtain multiple model parameter combinations; The target model parameters are determined according to the first sample data set and the plurality of model parameter combinations.

5. The method according to claim 4, characterized in that The determining the target model parameter according to the first sample data set and the plurality of model parameter combinations includes: Dividing the first sample data set into a first training set and a first validation set; Get the target fitness function; Determine a plurality of basic power demand forecasting models according to the plurality of model parameter combinations and the preset power demand model; each model parameter combination corresponds to a basic power demand forecasting model; Training each of the multiple basic power demand forecasting models according to the first training set to obtain multiple reference power demand forecasting models; Determine the performance indicator values ​​of the multiple reference power demand prediction models according to the first verification set to obtain multiple performance indicator values; Determine the fitness of the multiple performance indicator values ​​according to the target fitness function to obtain multiple fitnesses; each fitness corresponds to a performance indicator value; A maximum fitness among the multiple fitnesses is selected, and the target model parameters are determined according to a combination of model parameters corresponding to the maximum fitness.

6. The method according to any one of claims 3 to 5, characterized in that: The step of optimizing the first prediction model according to the judgment criterion to obtain the target power demand prediction model includes: Using a cross validation method to validate the first prediction model using the first sample data set to obtain a target evaluation value; Determine a target evaluation threshold according to the judgment criteria; When the target evaluation value is greater than or equal to the target evaluation threshold, determining the first prediction model as the target power demand prediction model; When the target evaluation value is less than the target evaluation threshold, determining a target adjustment coefficient corresponding to a model parameter of the first prediction model according to the target evaluation value; The model parameters of the first prediction model are adjusted according to the target adjustment coefficient to obtain the target power demand prediction model.

7. The method according to claim 1, characterized in that The method further comprises: Acquire data corresponding to the target key influencing factor in the target data set to obtain a first test data set; Inputting the first test data set into the target power demand prediction model to obtain a first power prediction result; Adjusting the first test data set based on a preset data step size to obtain a second test data set; Inputting the second test data set into the target power demand prediction model to obtain a second power prediction result; Determining a sensitivity coefficient corresponding to the target key influencing factor according to the preset data step size, the first power prediction result, and the second power prediction result; Determining a target reliability degree of the target power demand forecasting model according to the sensitivity coefficient; the target reliability degree is used to evaluate the reliability of the target power demand forecasting model under the target key influencing factors; the sensitivity coefficient is negatively correlated with the target reliability degree; Determining a target reliability level corresponding to the target reliability degree; Determining a target interference coefficient corresponding to the target reliability level; The target power prediction result is adjusted according to the target interference coefficient to obtain an actual power prediction result.

8. A power demand forecasting device, characterized in that: The device comprises: a data acquisition unit, a correlation analysis unit, a model training unit, and a power prediction unit, wherein: The data acquisition unit is used to acquire multiple influencing factors that affect electricity demand; the multiple influencing factors include: historical electricity, economy, population, and climate; The data acquisition unit is further configured to determine, within a first preset time period, an original data set corresponding to each of the multiple influencing factors in a preset area, to obtain multiple original data sets; The association analysis unit is used to perform association analysis on the multiple original data sets by using a grey association analysis method to screen the multiple influencing factors and obtain target key influencing factors; The data acquisition unit is further used to determine the data corresponding to the target key influencing factor in the multiple original data sets to obtain a sample data set; The data acquisition unit is also used to acquire a preset expert experience set; The model training unit is used to train the preset power demand model according to the preset expert experience set and the sample data set to obtain a target power demand prediction model; The power prediction unit is used to obtain a target data set corresponding to the target key influencing factors in the preset area within a second preset time period, input the target data set into the target power demand prediction model, and obtain a target power prediction result; the start time of the second preset time period is later than the end time of the first preset time period.

9. An electronic device, characterized in that: include: A processor, a memory, a communication interface, and one or more programs; The one or more programs are stored in the memory and configured to be executed by the processor, the programs comprising instructions for executing the steps in the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the processor is caused to perform the method according to any one of claims 1 to 7.

Images