A method and system for predicting power load during extreme weather events

Abstract

This invention provides a method and system for predicting power load under extreme weather conditions. It acquires historical data and determines the base load under extreme weather conditions based on this data. A total load decomposition model is constructed, and based on this model, the base load and random load are initially estimated and separated from the total load. The median meteorological load is obtained through multivariate regression fitting. Based on the median meteorological load, core meteorological factors are screened using improved grey relational analysis, combined with stepwise regression screening and extreme value adaptation correction to obtain the accurate meteorological load. A comprehensive prediction factor set is constructed based on the accurate meteorological load. Each prediction factor in the comprehensive prediction factor set is input into a trained improved BP neural network model, which outputs the predicted power load value under extreme weather conditions, effectively improving the accuracy of meteorological load separation and extreme weather load prediction.

Description

Technical Field

[0001] This invention belongs to the field of power load forecasting technology, and specifically relates to a method and system for forecasting power load during extreme weather. Background Technology

[0002] Power load forecasting is a core foundation for power grid dispatching, power planning, and ensuring power supply reliability. Especially under extreme weather conditions (such as extreme heat, heavy rain, and blizzards), power load is significantly affected by meteorological factors and is prone to nonlinear abrupt changes and extreme fluctuations, greatly increasing the difficulty of forecasting. Currently, existing technologies mainly use correlation analysis to analyze the relationship between power load characteristics and meteorological factors, and multiple regression, stepwise regression analysis, and BP neural network models are commonly used to establish daily maximum meteorological load forecasting models.

[0003] However, existing methods have significant technical shortcomings: First, the accuracy of meteorological load separation is low. Traditional methods often use single regression analysis to separate meteorological load from the total load, ignoring the small fluctuations in the baseline load under extreme weather conditions, the interference of random loads, and the nonlinear mutation characteristics of meteorological load. This results in the separated meteorological load being mixed with a large number of non-meteorological components, which in turn affects the accuracy of subsequent predictions. Second, the prediction deviation of extreme load values ​​under extreme weather conditions is large. Although the prediction model based on BP neural network can fit the overall trend of the load, the inaccuracy of the input meteorological load data and the lack of optimization for extreme load characteristics lead to the model's simulation values ​​of extreme values ​​being generally too large. Third, existing methods lack a closed-loop correction mechanism and cannot optimize the meteorological load separation parameters in reverse based on the prediction error, making it difficult to adapt to different types and intensities of extreme weather. Summary of the Invention

[0004] Based on this, the present invention provides a method and system for predicting power load in extreme weather, which aims to accurately separate meteorological load from power load, thereby improving the accuracy of power load prediction in extreme weather.

[0005] A first aspect of this invention provides a method for predicting power load during extreme weather events, the method comprising:

[0006] Acquire historical data, which includes at least daily / hourly total power load data, meteorological factor data, extreme weather level data, and power grid emergency dispatch records;

[0007] Based on the historical data, determine the baseline load under extreme weather conditions;

[0008] A total load decomposition model is constructed. Based on the total load decomposition model, the basic load and random load in the total load are preliminarily estimated and fitted by multiple regression to obtain the median meteorological load.

[0009] Based on the median meteorological load, core meteorological factors are screened through improved grey relational analysis, and then combined with stepwise regression screening and extreme value adaptation correction to obtain the accurate meteorological load.

[0010] Based on the precise meteorological load, a comprehensive predictive factor set is constructed;

[0011] Each predictor in the comprehensive predictor set is input into the trained improved BP neural network model, which outputs the predicted power load under extreme weather conditions.

[0012] Furthermore, the meteorological factor data includes at least extreme high / low temperatures, precipitation intensity, wind speed, and relative humidity.

[0013] Furthermore, the step of determining the base load under extreme weather conditions based on the historical data includes:

[0014] Calculate the correlation coefficients between the total power load and each meteorological factor, and determine whether the absolute value of the correlation coefficient is less than a first threshold.

[0015] If so, it is determined to be a non-meteorologically sensitive period, and the non-meteorologically sensitive periods are combined to form a sample set;

[0016] An adaptive sliding window mean algorithm is used to calculate the baseline load value, and abnormal loads are removed by using a variance threshold to obtain the corrected baseline load value.

[0017] Based on the revised baseline load value, and combined with the duration of extreme weather and emergency dispatch coefficient, the final baseline load under extreme weather conditions is obtained.

[0018] Furthermore, the steps of constructing the total load decomposition model, separating the basic load and random load from the total load based on the total load decomposition model, and obtaining the median meteorological load through multiple regression fitting include:

[0019] A total load decomposition model is constructed, and the total load decomposition model is deformed to obtain a rough value calculation model for meteorological load.

[0020] Based on the aforementioned meteorological load coarse value calculation model, wavelet denoising is used to extract random load, and the meteorological load coarse value is calculated based on the random load;

[0021] Based on the quadratic and interaction terms of the meteorological factors, a multiple regression model is constructed, and the coarse value of the meteorological load is fitted to obtain the median value of the meteorological load.

[0022] Furthermore, the step of obtaining the accurate meteorological load by screening core meteorological factors through improved grey relational analysis based on the median meteorological load, combined with stepwise regression screening and extreme value adaptation correction, includes:

[0023] Based on the extreme weather level data, determine the extreme weather level weighting coefficient;

[0024] An extreme weather level weighting coefficient is introduced, and the median meteorological load is used as the reference sequence, while meteorological factors are used as the comparison sequence to perform an improved grey relational analysis.

[0025] Determine whether the improved grey relational degree calculation result is greater than the second threshold;

[0026] If so, the corresponding meteorological factor is determined to be the core meteorological factor;

[0027] Using core meteorological factors as independent variables and the median meteorological load as the dependent variable, an optimal regression model is constructed by introducing and removing factors through stepwise regression. Based on the optimal regression model, the fine-tuning value of the meteorological load is determined.

[0028] Based on historical extreme weather meteorological load extreme value data, extreme value constraints are set, threshold verification is performed on the meteorological load fine correction value, and exponential smoothing method is used to correct extreme values ​​exceeding the threshold to obtain accurate meteorological load.

[0029] Furthermore, in the step of constructing a comprehensive prediction factor set based on the precise meteorological load, the comprehensive prediction factor set includes precise meteorological load, optimal combination of meteorological factors, extreme weather level data, and historical total power load for the same period.

[0030] Furthermore, the improved BP neural network model includes an input layer, a hidden layer, an extremum fitting layer, and an output layer, wherein the activation function of the hidden layer is:

[0031] ;

[0032] The activation function of the output layer is:

[0033] ;

[0034] in, Let x be the output layer's computation result at time t, and w be the hidden layer's output. i x represents the weights from the extreme value adaptation layer to the output layer. i is the input to the extremum adaptation layer, b is the output layer bias, k is the extremum correction coefficient, and sign() is the sign function. To accurately calculate the average meteorological load, For the precise meteorological load at time t, This represents the number of nodes in the extreme value adaptation layer.

[0035] A second aspect of this invention provides a power load forecasting system for extreme weather, used to implement the power load forecasting method for extreme weather described in the first aspect, the system comprising:

[0036] The acquisition module is used to acquire historical data, which includes at least daily / hourly total power load data, meteorological factor data, extreme weather level data, and power grid emergency dispatch record data.

[0037] The determination module is used to determine the base load under extreme weather conditions based on the historical data.

[0038] The first construction module is used to construct a total load decomposition model. Based on the total load decomposition model, the basic load and random load in the total load are initially estimated and then fitted by multiple regression to obtain the median meteorological load.

[0039] The screening module is used to screen core meteorological factors based on the median meteorological load by improving grey relational analysis, and then combine stepwise regression screening and extreme value adaptation correction to obtain accurate meteorological load.

[0040] The second construction module is used to construct a comprehensive prediction factor set based on the precise meteorological load;

[0041] The input module is used to input each predictor of the comprehensive predictor set into the trained improved BP neural network model and output the predicted power load value under extreme weather conditions.

[0042] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the power load forecasting method for extreme weather provided in the first aspect.

[0043] A fourth aspect of the present invention provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the power load forecasting method for extreme weather provided in the first aspect.

[0044] This invention provides a method and system for predicting power load under extreme weather conditions. The method involves acquiring historical data and determining the base load under extreme weather conditions. A total load decomposition model is constructed, and based on this model, the base load and random load are initially estimated and separated from the total load. A median meteorological load is obtained through multivariate regression fitting. Based on the median meteorological load, core meteorological factors are screened using improved grey relational analysis, combined with stepwise regression screening and extreme value adaptation correction to obtain an accurate meteorological load. A comprehensive prediction factor set is constructed based on the accurate meteorological load. Each prediction factor in the comprehensive prediction factor set is input into a trained improved BP neural network model to output the predicted power load under extreme weather conditions, effectively improving the accuracy of meteorological load separation and extreme weather load prediction. Attached Figure Description

[0045] Figure 1 The flowchart illustrates the implementation of a power load forecasting method for extreme weather conditions, as provided in Embodiment 1 of the present invention.

[0046] Figure 2 This is a structural block diagram of an extreme weather power load forecasting system provided in Embodiment 2 of the present invention;

[0047] Figure 3 This is a structural block diagram of an electronic device provided in Embodiment 3 of the present invention. Detailed Implementation

[0048] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Several embodiments of the invention are illustrated in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

[0049] It should be noted that when a component is said to be "fixed to" another component, it can be directly on the other component or there may be an intervening component. When a component is said to be "connected to" another component, it can be directly connected to the other component or there may be an intervening component. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0050] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0051] Example 1

[0052] According to an embodiment of the present invention, an embodiment of a power load forecasting method for extreme weather is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0053] This first embodiment provides a method for predicting power load during extreme weather events, which can be used in electronic devices, such as computers. Please refer to... Figure 1 , Figure 1 The flowchart of an extreme weather power load forecasting method provided in Embodiment 1 of the present invention is shown, specifically including steps S01 to S06.

[0054] Step S01: Obtain historical data, which includes at least daily / hourly total power load data, meteorological factor data, extreme weather level data, and power grid emergency dispatch record data.

[0055] Among them, meteorological factor data include at least extreme high / low temperature T(t), precipitation intensity R(t), wind speed V(t), and relative humidity H(t).

[0056] Step S02: Determine the base load under extreme weather conditions based on the historical data.

[0057] Specifically, the correlation coefficients between the total power load and each meteorological factor are calculated, and it is determined whether the absolute value of the correlation coefficient is less than a first threshold. The expression for calculating the correlation coefficient is as follows:

[0058] ;

[0059] Let n be the correlation coefficient between total electricity load and a certain meteorological factor X, where n is the total number of samples. Let t be the total electrical load. Let X(t) be the average total electricity load, and X(t) be the value of a certain meteorological factor at time t. This is the average value of the meteorological factor;

[0060] If so, it is determined to be a non-meteorologically sensitive period, and the non-meteorologically sensitive periods are combined to form a sample set. In the embodiments of the present invention, if If time t is determined to be a non-meteorologically sensitive period, then all non-meteorologically sensitive periods are selected to form a sample set. ;

[0061] An adaptive sliding window mean algorithm is used to calculate the baseline load value. Abnormal loads are then removed using a variance threshold to obtain the corrected baseline load value. It should be noted that the size of the adaptive sliding window is first determined, expressed as follows:

[0062] ;

[0063] W represents the size of the sliding window. The standard deviation of the load during non-meteorologically sensitive periods. This represents the average load during non-meteorologically sensitive periods, and `round()` is the rounding function. A preliminary calculation of the baseline load value is then performed, expressed as:

[0064] ;

[0065] This is the preliminary result of the baseline value of the base load at time t. Let be the total power load at time i. Further, abnormal loads are identified, and corrections are made based on the identification results. It should be noted that the identification expression is:

[0066] ;

[0067] The expression for the correction process is:

[0068] ;

[0069] in, The variance of the load during non-meteorologically sensitive periods. This is the corrected baseline value of the base load at time t under non-extreme weather conditions;

[0070] Based on the revised baseline load value, and considering the duration of extreme weather and emergency dispatch coefficients, a correction process is performed to obtain the final baseline load under extreme weather conditions. The expression for this correction process is as follows:

[0071] ;

[0072] This represents the final base load value at time t under extreme weather conditions, i.e., the final base load under extreme weather conditions. The baseline load value at time t0 under extreme weather conditions is given by D(t), which is the extreme weather duration coefficient (D(t) = duration of extreme weather / historical average duration of this type of extreme weather, with a value of 0-1), and S(t) is the emergency dispatch coefficient (obtained by fitting the power grid emergency dispatch record data in historical data, S(t) = 0.01~0.05, the greater the dispatch intensity, the greater S(t)); k1 and k2 are the correction weights (obtained by fitting historical data using the least squares method, k1∈[0.02,0.05], k2∈[0.5,0.8]).

[0073] Step S03: Construct a total load decomposition model. Based on the total load decomposition model, extract the basic load and random load from the total load for preliminary estimation, and obtain the median meteorological load through multiple regression fitting.

[0074] It should be noted that the total load decomposition model is constructed, and then deformed to obtain a coarse meteorological load calculation model. In this embodiment of the invention, the total load decomposition model is represented as follows:

[0075] ;

[0076] After deformation, the coarse calculation model for meteorological load is expressed as follows:

[0077] ;

[0078] Let t be the actual meteorological load. Let t be the random load. This is a preliminary result of the rough values ​​of meteorological load. This is a preliminary estimate of the random load;

[0079] Based on the aforementioned coarse meteorological load calculation model, wavelet denoising is used to extract random loads, and based on these random loads, coarse meteorological load values ​​are calculated. Specifically, Decomposed into low-frequency component A m (Dominated by meteorological load) and high-frequency components D1, D2, ..., D m (Dominated by random load), the decomposition formula is:

[0080] ;

[0081] Subsequently, the high-frequency components D1, D2, ..., D... m Soft thresholding is used, with the threshold formula as follows: Noise estimation ;

[0082] The preliminary estimate of the random load is expressed as follows:

[0083] ;

[0084] The high-frequency components after noise reduction. It is a low-pass wavelet filter. Here, λ is a high-pass wavelet filter, m is the wavelet decomposition level (set according to the data sampling frequency; m=5 for hourly data and m=3 for daily data), λ is the noise reduction threshold, and median() is the median function. Understandably, substituting the preliminary estimate of the random load into the coarse meteorological load calculation model yields the coarse meteorological load value, i.e., the preliminary result of the coarse meteorological load value. ;

[0085] Based on the quadratic and interaction terms of the meteorological factors, a multiple regression model is constructed, and the coarse value of the meteorological load is fitted to obtain the median meteorological load. The multiple regression model is expressed as follows:

[0086] ;

[0087] Let t be the median meteorological load. For constant terms, For regression coefficients, They are T(t), R(t), V(t), and H(t), respectively. , For meteorological factor interaction terms, For the random error term ( );

[0088] It should be noted that the regression coefficients are calculated using the least squares method, expressed as:

[0089] ;

[0090] in, For the regression coefficient estimates, X is the independent variable matrix (dimension n×11, where n is the total number of samples, and each row corresponds to the combination of independent variables at time t), and Y is the dependent variable vector (dimension n×1, where each row corresponds to the preliminary results of the coarse meteorological load). ).

[0091] Step S04: Based on the median meteorological load, core meteorological factors are screened through improved grey relational analysis, and combined with stepwise regression screening and extreme value adaptation correction to obtain accurate meteorological load.

[0092] Specifically, based on the extreme weather level data, the weighting coefficient for the extreme weather level is determined, and the calculation formula is as follows:

[0093] ;

[0094] Let t be the extreme weather weighting coefficient. The extreme weather level at time t;

[0095] An improved grey relational analysis is performed by introducing extreme weather level weighting coefficients, using the median meteorological load as the reference sequence and meteorological factors as the comparison sequence. The reference sequence is represented as follows:

[0096] ;

[0097] The comparison sequence is represented as:

[0098] ;

[0099] This is the minimum value of the meteorological load median. It is the maximum value of the median meteorological load. Let j be the meteorological factor data at time t. Let j be the minimum value of the meteorological factor data. The maximum value of the j-th meteorological factor data;

[0100] Furthermore, the improved expression for calculating grey relational degree is as follows:

[0101] ;

[0102] ;

[0103] The resolution coefficient, Let r be the correlation coefficient between the j-th meteorological factor and the meteorological load at time t. j Let j be the correlation degree of the j-th meteorological factor. It is the minimum difference between two levels. The maximum difference between the two levels;

[0104] Determine whether the improved grey relational degree calculation result is greater than the second threshold;

[0105] If so, the corresponding meteorological factor is determined to be the core meteorological factor;

[0106] Using core meteorological factors as independent variables and the median meteorological load as the dependent variable, an optimal regression model is constructed through stepwise regression by introducing and removing factors. Based on the optimal regression model, a fine-tuning value for the meteorological load is determined. In this embodiment of the invention, a factor threshold F is introduced. in =3.84, removal factor threshold F out =2.71, introduce the factor F1 statistic:

[0107] ;

[0108] F2 statistic of elimination factor:

[0109] ;

[0110] SSE0 is the sum of squared residuals without the factor, SSE1 is the sum of squared residuals with the factor, and SSE2 is the sum of squared residuals before the factor was removed. 3为 The sum of squared residuals after removing this factor , The number of independent variables before and after the introduction of the factor, where n is the total number of samples. Given an F-distribution (degrees of freedom a, b), it's understandable that we first set a threshold F to introduce the factor. in =3.84, removal factor threshold F out The F-test criterion is 2.71. Starting from the empty model, we first introduce the F1 statistic as greater than the F-statistic. in Significant factors were identified, and then factors that caused the explanatory power to be covered due to collinearity or other reasons, as well as factors with an F2 statistic less than F, were removed from the model. out The insignificant factors are then iterated and looped in a "forward introduction-backward elimination" operation until no new factors can be introduced and no introduced factors can be eliminated.

[0111] Furthermore, the optimal regression model is expressed as:

[0112] ;

[0113] This is the fine-corrected value of the meteorological load at time t. This is the optimal combination of factors selected through stepwise regression. , The optimal regression coefficients are obtained by the least squares method. For residual terms ( );

[0114] Based on historical extreme weather meteorological load extreme value data, extreme value constraints are set, and threshold verification is performed on the meteorological load fine correction value. Exponential smoothing is then used to correct extreme values ​​exceeding the threshold, resulting in an accurate meteorological load. It should be noted that the upper limit of the meteorological load extreme value is:

[0115] ;

[0116] Lower limit of extreme meteorological load:

[0117] ;

[0118] This represents the average of the refined values ​​of meteorological loads during historical extreme weather periods. The standard deviation of the refined values ​​of meteorological loads during historical extreme weather periods. , These are the upper and lower limits of extreme meteorological loads;

[0119] The exponential smoothing method is used to correct extreme values ​​that exceed the threshold, which can be expressed as:

[0120]

[0121] Let be the precise meteorological load at time t, and α be the smoothing coefficient.

[0122] Step S05: Construct a comprehensive prediction factor set based on the precise meteorological load.

[0123] The comprehensive forecasting factor set includes precise meteorological load, optimal combination of meteorological factors, extreme weather level data, and historical total power load for the same period. , For the precise meteorological load at time t, The optimal combination of meteorological factors at time t. This represents the final base load value at time t under extreme weather conditions. Let t be the extreme weather level at time t. This represents the total power load for the same historical period (T is the time interval between periods).

[0124] Step S06: Input each predictor of the comprehensive predictor set into the trained improved BP neural network model and output the predicted power load value under extreme weather conditions.

[0125] Specifically, the improved BP neural network model includes an input layer, hidden layers, an extreme value fitting layer, and an output layer. The number of nodes in the input layer is: (Dimensions of comprehensive predictive factors), Number of hidden layer nodes (adaptively determined): ( Number of output layer nodes (i.e., power load forecast value); number of extreme value adaptation layer nodes: (Input is) , Number of output layer nodes: (The output is the predicted power load at time t) The activation function of the hidden layer is:

[0126] ;

[0127] The activation function of the output layer is:

[0128] ;

[0129] in, The output layer at time t represents the calculation result (i.e., the intermediate output of the power load forecast). Further processing will yield the final load forecast. x represents the hidden layer output, and w... i x represents the weights from the extreme value adaptation layer to the output layer. i is the input to the extremum adaptation layer, b is the output layer bias, k is the extremum correction coefficient, and sign() is the sign function. To accurately calculate the average meteorological load, For the precise meteorological load at time t, The number of nodes in the extreme value adaptation layer is given. Additionally, the adaptive learning rate (dynamically adjusted based on prediction error) is expressed as:

[0130] ;

[0131] Let be the learning rate at time t. Let E(t) be the initial learning rate, and E(t) be the model prediction error (mean squared error) at time t.

[0132] In this embodiment of the invention, the loss function during model training is expressed as:

[0133] ;

[0134] E represents the total loss of the model. Let t be the predicted power load value. Let be the actual power load value at time t. Furthermore, the parameter update uses the gradient descent method, and the weight update is expressed as:

[0135] ;

[0136] Bias update is represented as:

[0137] ;

[0138] when Training will stop when the number of iterations reaches the preset maximum value (1000 times).

[0139] In summary, the extreme weather power load forecasting method in the above embodiments of the present invention obtains historical data and determines the base load under extreme weather conditions based on the historical data; constructs a total load decomposition model, and based on the total load decomposition model, separates the base load and random load from the total load for preliminary estimation, and obtains the meteorological load median through multivariate regression fitting; based on the meteorological load median, core meteorological factors are screened through improved grey relational analysis, combined with stepwise regression screening and extreme value adaptation correction to obtain accurate meteorological load; based on the accurate meteorological load, a comprehensive prediction factor set is constructed; and the prediction factors of the comprehensive prediction factor set are input into a trained improved BP neural network model to output the power load forecast value under extreme weather conditions, effectively improving the meteorological load separation accuracy and the extreme weather load forecasting accuracy.

[0140] Example 2

[0141] Please see Figure 2 , Figure 2 This is a structural block diagram of an extreme weather power load forecasting system 200 provided in Embodiment 2 of the present invention. This extreme weather power load forecasting system 200 is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the apparatus described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0142] Specifically, the extreme weather power load forecasting system 200 includes: an acquisition module 21, a determination module 22, a first construction module 23, a filtering module 24, a second construction module 25, and an input module 26, wherein:

[0143] The acquisition module 21 is used to acquire historical data, which includes at least daily / hourly total power load data, meteorological factor data, extreme weather level data, and power grid emergency dispatch record data. The meteorological factor data includes at least extreme high / low temperatures, precipitation intensity, wind speed, and relative humidity.

[0144] The determination module 22 is used to determine the base load under extreme weather conditions based on the historical data.

[0145] The first construction module 23 is used to construct a total load decomposition model. Based on the total load decomposition model, the basic load and random load in the total load are initially estimated and then fitted by multiple regression to obtain the median meteorological load.

[0146] The screening module 24 is used to screen core meteorological factors based on the median meteorological load by improving grey relational analysis, and combine stepwise regression screening and extreme value adaptation correction to obtain accurate meteorological load.

[0147] The second construction module 25 is used to construct a comprehensive prediction factor set based on the accurate meteorological load. The comprehensive prediction factor set includes accurate meteorological load, optimal combination of meteorological factors, extreme weather level data, and historical total power load for the same period.

[0148] Input module 26 is used to input each predictor of the comprehensive predictor set into the trained improved BP neural network model, and output the predicted power load value under extreme weather conditions. The improved BP neural network model includes an input layer, a hidden layer, an extreme value adaptation layer, and an output layer. The activation function of the hidden layer is:

[0149] ;

[0150] The activation function of the output layer is:

[0151] ;

[0152] in, Let x be the output layer's computation result at time t, and w be the hidden layer's output. i x represents the weights from the extreme value adaptation layer to the output layer. i is the input to the extremum adaptation layer, b is the output layer bias, k is the extremum correction coefficient, and sign() is the sign function. To accurately calculate the average meteorological load, For the precise meteorological load at time t, This represents the number of nodes in the extreme value adaptation layer.

[0153] Furthermore, in some other embodiments of the present invention, the determining module 22 includes:

[0154] The first judgment unit is used to calculate the correlation coefficient between the total power load and each meteorological factor, and to determine whether the absolute value of the correlation coefficient is less than a first threshold.

[0155] The combination unit is used to determine a non-meteorologically sensitive period when the absolute value of the correlation coefficient is less than a first threshold, and to combine the non-meteorologically sensitive periods to form a sample set.

[0156] The first calculation unit is used to calculate the baseline load value using an adaptive sliding window mean algorithm, and at the same time remove abnormal loads by using a variance threshold to obtain the corrected baseline load value.

[0157] The correction unit is used to perform correction processing based on the corrected baseline load value, combined with the duration of extreme weather and emergency dispatch coefficient, to obtain the final baseline load under extreme weather conditions.

[0158] Furthermore, in some other embodiments of the present invention, the first building module 23 includes:

[0159] A construction unit is used to construct a total load decomposition model and deform the total load decomposition model to obtain a rough calculation model for meteorological load.

[0160] The second calculation unit is used to extract random loads by wavelet denoising according to the meteorological load coarse value calculation model, and to calculate the meteorological load coarse value according to the random loads.

[0161] The fitting unit is used to construct a multiple regression model based on the quadratic and interaction terms of the meteorological factors, and to fit the coarse value of the meteorological load to obtain the median value of the meteorological load.

[0162] Furthermore, in some other embodiments of the present invention, the screening module 24 includes:

[0163] The first determining unit is used to determine the extreme weather level weighting coefficient based on the extreme weather level data.

[0164] The third calculation unit is used to introduce the extreme weather level weight coefficient, with the meteorological load median as the reference sequence and the meteorological factors as the comparison sequence, to perform improved grey relational degree calculation.

[0165] The second judgment unit is used to determine whether the improved gray relation degree calculation result is greater than the second threshold.

[0166] The determination unit is used to determine that the corresponding meteorological factor is the core meteorological factor when the result of the improved grey relational degree calculation is greater than the second threshold.

[0167] Using core meteorological factors as independent variables and the median meteorological load as the dependent variable, an optimal regression model is constructed by introducing and removing factors through stepwise regression. Based on the optimal regression model, the fine-tuning value of the meteorological load is determined.

[0168] The second determining unit is used to set extreme value constraints based on historical extreme weather meteorological load extreme value data, perform threshold verification on the meteorological load fine correction value, and use the exponential smoothing method to correct extreme values ​​exceeding the threshold to obtain accurate meteorological load.

[0169] Example 3

[0170] In another aspect, the present invention also proposes an electronic device, please refer to [link to relevant documentation]. Figure 3 The image shows an electronic device according to Embodiment 3 of the present invention, including a memory 20, a processor 10, and a computer program 30 stored in the memory and executable on the processor. When the processor 10 executes the computer program 30, it implements the power load forecasting method for extreme weather as described above.

[0171] In some embodiments, the processor 10 may be a central processing unit (CPU), controller, microcontroller, microprocessor or other data processing chip, used to run program code stored in memory 20 or process data, such as executing access restriction programs.

[0172] The memory 20 includes at least one type of readable storage medium, such as flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 20 can be an internal storage unit of an electronic device, such as the hard disk of the electronic device. In other embodiments, the memory 20 can also be an external storage device of the electronic device, such as a plug-in hard disk, SmartMediaCard (SMC), SecureDigital (SD) card, FlashCard, etc., equipped on the electronic device. Furthermore, the memory 20 can include both internal and external storage units of the electronic device. The memory 20 can be used not only to store application software and various types of data of the electronic device, but also to temporarily store data that has been output or will be output.

[0173] It should be pointed out that, Figure 3 The structure shown does not constitute a limitation on the electronic device. In other embodiments, the electronic device may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0174] This invention also proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for predicting power load during extreme weather.

[0175] Those skilled in the art will understand that the logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can mean any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0176] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0177] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0178] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0179] The above embodiments merely illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this patent should be determined by the appended claims.

Claims

1. A method for predicting power load during extreme weather, characterized in that, The method includes: Acquire historical data, which includes at least daily / hourly total power load data, meteorological factor data, extreme weather level data, and power grid emergency dispatch records; Based on the historical data, determine the baseline load under extreme weather conditions; A total load decomposition model is constructed. Based on the total load decomposition model, the basic load and random load in the total load are initially estimated and then fitted using multiple regression to obtain the median meteorological load. Specifically, this includes: A total load decomposition model is constructed, and the total load decomposition model is deformed to obtain a rough value calculation model for meteorological load. Based on the aforementioned meteorological load coarse value calculation model, wavelet denoising is used to extract random load, and the meteorological load coarse value is calculated based on the random load; Based on the quadratic and interaction terms of the meteorological factors, a multiple regression model is constructed, and the coarse value of the meteorological load is fitted to obtain the median value of the meteorological load. Based on the median meteorological load, core meteorological factors are screened using improved grey relational analysis, and then combined with stepwise regression screening and extreme value fitting correction to obtain the accurate meteorological load, specifically including: Based on the extreme weather level data, determine the extreme weather level weighting coefficient; An extreme weather level weighting coefficient is introduced, and the median meteorological load is used as the reference sequence, while meteorological factors are used as the comparison sequence to perform an improved grey relational analysis. Determine whether the improved grey relational degree calculation result is greater than the second threshold; If so, the corresponding meteorological factor is determined to be the core meteorological factor; Using core meteorological factors as independent variables and the median meteorological load as the dependent variable, an optimal regression model is constructed by introducing and removing factors through stepwise regression. Based on the optimal regression model, the fine-tuning value of the meteorological load is determined. Based on historical extreme weather meteorological load extreme value data, extreme value constraints are set, threshold verification is performed on the meteorological load fine correction value, and exponential smoothing method is used to correct extreme values ​​that exceed the threshold to obtain accurate meteorological load. Based on the precise meteorological load, a comprehensive predictive factor set is constructed; Each predictor in the comprehensive predictor set is input into the trained improved BP neural network model, which outputs the predicted power load under extreme weather conditions.

2. The method for predicting power load during extreme weather according to claim 1, characterized in that, The meteorological data include at least extreme high / low temperatures, precipitation intensity, wind speed, and relative humidity.

3. The method for predicting power load during extreme weather according to claim 2, characterized in that, The step of determining the base load under extreme weather conditions based on the historical data includes: Calculate the correlation coefficients between the total power load and each meteorological factor, and determine whether the absolute value of the correlation coefficient is less than a first threshold. If so, it is determined to be a non-meteorologically sensitive period, and the non-meteorologically sensitive periods are combined to form a sample set; An adaptive sliding window mean algorithm is used to calculate the baseline load value, and abnormal loads are removed by using a variance threshold to obtain the corrected baseline load value. Based on the revised baseline load value, and combined with the duration of extreme weather and emergency dispatch coefficient, the final baseline load under extreme weather conditions is obtained.

4. The method for predicting power load during extreme weather according to claim 3, characterized in that, In the step of constructing a comprehensive prediction factor set based on the accurate meteorological load, the comprehensive prediction factor set includes accurate meteorological load, optimal combination of meteorological factors, extreme weather level data, and historical total power load for the same period.

5. The method for predicting power load during extreme weather according to claim 4, characterized in that, The improved BP neural network model includes an input layer, hidden layers, an extremum fitting layer, and an output layer. The activation function of the hidden layer is: ； The activation function of the output layer is: ； in, Let x be the output layer's computation result at time t, and w be the hidden layer's output. i x represents the weights from the extreme value adaptation layer to the output layer. i is the input to the extremum adaptation layer, b is the output layer bias, k is the extremum correction coefficient, and sign() is the sign function. To accurately calculate the average meteorological load, For the precise meteorological load at time t, This represents the number of nodes in the extreme value adaptation layer.

6. A power load forecasting system for extreme weather, characterized in that, For implementing the power load forecasting method for extreme weather as described in any one of claims 1-5, the system comprises: The acquisition module is used to acquire historical data, which includes at least daily / hourly total power load data, meteorological factor data, extreme weather level data, and power grid emergency dispatch record data. The determination module is used to determine the base load under extreme weather conditions based on the historical data. The first construction module is used to construct a total load decomposition model. Based on the total load decomposition model, the basic load and random load in the total load are initially estimated and then fitted by multiple regression to obtain the median meteorological load. The screening module is used to screen core meteorological factors based on the median meteorological load by improving grey relational analysis, and then combine stepwise regression screening and extreme value adaptation correction to obtain accurate meteorological load. The second construction module is used to construct a comprehensive prediction factor set based on the precise meteorological load; The input module is used to input each predictor of the comprehensive predictor set into the trained improved BP neural network model and output the predicted power load value under extreme weather conditions.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the power load forecasting method for extreme weather as described in any one of claims 1-5.

8. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the extreme weather power load forecasting method as described in any one of claims 1-5.

Images