Method and system for predicting total load of power system based on voltage class

Abstract

The invention discloses a power system total load prediction method based on a voltage level, and the method comprises the steps: obtaining and preprocessing the historical data information of a target power system, carrying out the classification according to the voltage level, and constructing a feature vector; performing clustering analysis on the feature vectors; constructing a training data set based on fuzzification processing, similarity analysis and correlation analysis; constructing a load prediction preliminary model set and training to obtain a load prediction model set; and adopting a model in the load prediction model set to predict, gather and integrate the load of each voltage grade in the target power system, and completing the total load prediction of the target power system based on the voltage grade. The invention also discloses a system for realizing the total load prediction method of the power system based on the voltage grade. According to the invention, the prediction of the total load of the power system based on the voltage level can be realized, and the reliability and the accuracy are higher.

Description

Technical Field

[0001] The present invention belongs to the field of electrical automation, and in particular relates to a method and system for predicting the total load of an electric power system based on voltage levels. Background Art

[0002] With the development of economy and technology and the improvement of people's living standards, electricity has become an indispensable secondary energy source in people's production and life, bringing endless convenience to people's production and life. Therefore, ensuring the stable and reliable supply of electricity has become one of the most important tasks of the power system.

[0003] Load forecasting is one of the important tasks of the power system. The load forecasting results can effectively help the power system formulate operation plans and assist the power system in making decisions. Therefore, accurate and reliable load forecasting is of great significance to the power system.

[0004] At present, the traditional load forecasting schemes for power systems all adopt the total load forecasting scheme; this type of forecasting scheme considers the power system as a whole. However, there are multiple voltage levels within the power system, and the load characteristics of different voltage levels are not the same. The current holistic total load forecasting scheme does not consider the voltage level problem of the power system; this will cause the current load forecasting scheme to have problems of poor reliability and poor accuracy. Summary of the invention

[0005] One of the purposes of the present invention is to provide a method for predicting the total load of an electric power system based on voltage levels with high reliability and good accuracy.

[0006] A second object of the present invention is to provide a system for implementing the method for predicting the total load of an electric power system based on voltage levels.

[0007] The method for predicting the total load of a power system based on voltage levels provided by the present invention comprises the following steps:

[0008] S1. Obtain historical data information of the target power system;

[0009] S2. Preprocess the historical data information obtained in step S1, classify them according to voltage levels, and construct corresponding feature vectors;

[0010] S3. Perform cluster analysis on the feature vector constructed in step S2;

[0011] S4. According to the cluster analysis results obtained in step S3, a training data set is constructed based on fuzzy processing, similarity analysis and correlation analysis;

[0012] S5. Construct a preliminary model set for load forecasting;

[0013] S6. Using the training data set obtained in step S4, the model in the load forecasting preliminary model set constructed in step S5 is trained to obtain a load forecasting model set;

[0014] S7. Use the model in the load forecasting model set obtained in step S6 to forecast the load of each voltage level in the target power system, and summarize and synthesize them to complete the total load forecast of the target power system based on the voltage level.

[0015] The acquisition of historical data information of the target power system described in step S1 specifically includes the following steps:

[0016] Obtain historical data information of the target power system;

[0017] The historical data information includes historical load data information of different voltage levels, historical temperature data, historical humidity data, historical rainfall data and historical weather data;

[0018] The voltage levels include 500kV voltage level, 330kV voltage level, 220kV voltage level and 110kV voltage level; the rainfall data includes no rain, light rain, moderate rain and heavy rain; the weather data includes sunny, cloudy, overcast and rainy.

[0019] Step S2 preprocesses the historical data information obtained in step S1, classifies them according to voltage levels, and constructs corresponding feature vectors, which specifically includes the following steps:

[0020] Preprocessing the historical data information obtained in step S1;

[0021] The preprocessing includes missing value interpolation; if the number of consecutive missing data is not greater than a set value, the missing data is interpolated using linear interpolation; if the number of consecutive missing data is greater than a set value, the missing data is interpolated using load data of similar historical days;

[0022] Classify the pre-processed historical data information according to voltage levels;

[0023] Construct the corresponding feature vector:

[0024] According to the daily load data P = {P1, P2, ..., P n}Construct feature vector, P n is the nth sampling data in the daily load data; the elements in the feature vector include the load average value Maximum daily load P max =max{P1,P2,...,P n}、Daily minimum load P min=min{P1,P2,...,P n}、Load rate Peak-to-valley rate

[0025] Construct the corresponding feature vector x as

[0026] The cluster analysis of the feature vector constructed in step S2 described in step S3 specifically includes the following steps:

[0027] The fuzzy mean cluster analysis of historical load data at the same voltage level includes the following steps:

[0028] A. Set the fuzzy index;

[0029] B. Initialize the membership matrix U. The size of the membership matrix U is N×C, where N is the total number of data points and C is the total number of clusters. The membership matrix U satisfies all rows U ij For data point x i The degree of membership to cluster j;

[0030] C. Calculate the center v of each cluster based on the current membership matrix U j for m is the fuzzy factor;

[0031] D. Update the current membership matrix U. The update formula is: in is the element of the i-th row and j-th column of the updated membership matrix U, || || is the distance calculation function;

[0032] E. Repeat step C and step D until the set number of repetitions is reached or the change value of two consecutive membership matrices U is less than the set threshold; at this time, the final membership matrix U is obtained.

[0033] According to the cluster analysis results obtained in step S3, step S4 constructs a training data set based on fuzzy processing, similarity analysis and correlation analysis, which specifically includes the following steps:

[0034] a. Fuzzy processing of day type and meteorological data:

[0035] For the day type, the following formula is used for fuzzy processing:

[0036]

[0037] For rainfall data, the following formula is used for fuzzy processing:

[0038]

[0039] For relative humidity data, the following formula is used for fuzzy processing:

[0040]

[0041] Where h' is the relative humidity data after fuzzy processing; h is the relative humidity data before fuzzy processing;

[0042] The following formula is used to perform fuzzy processing on the daily maximum temperature data:

[0043]

[0044] Where t' max is the daily maximum temperature data after fuzzy processing; t max is the daily maximum temperature data before fuzzification;

[0045] The following formula is used to perform fuzzy processing on the daily minimum temperature data:

[0046]

[0047] Where t' min is the daily minimum temperature data after fuzzy processing; t min is the daily minimum temperature data before fuzzification;

[0048] For weather data, the following formula is used for fuzzy processing:

[0049]

[0050] The grey correlation analysis method is used to analyze the correlation between the fuzzy processed daily type data, rainfall data, relative humidity data, daily maximum temperature data, daily minimum temperature data and weather data and the load data:

[0051] The grey correlation coefficient between the influencing factors and the load data is calculated using the following formula. The influencing factors include the day type data, rainfall data, relative humidity data, daily maximum temperature data, daily minimum temperature data and weather data after fuzzy processing:

[0052]

[0053] Where ξ i (k) is the sequence x0 and the sequence x i The grey correlation coefficient at point k; x0(k) is the kth value of the daily maximum load time series; x i (k) is the kth value of the i-th influencing factor sequence; ρ is the set resolution coefficient; Indicates the minimum range of the second level; Indicates the maximum range of the second level;

[0054] According to the grey correlation coefficient, calculate the influencing factor sequence x i The correlation R with the daily maximum load sequence x0 i for N is the total number of elements in the sequence;

[0055] The influencing factors with the largest correlation are selected as the main influencing factors of the daily maximum load, and the first training data set is constructed;

[0056] b. Extract the load data y at time t t The time attribute features of the system include week type features, month features, and holiday features, and perform fuzzy processing:

[0057] For the week type features, the following formula is used for fuzzy processing:

[0058]

[0059] For the monthly features, the following formula is used for fuzzy processing:

[0060]

[0061] For holiday features, the following formula is used for fuzzy processing:

[0062]

[0063] Take the load y at time t t Taking the date as the reference, obtain the average temperature data z4 of the previous day, the temperature data z5 at time t of the previous day, the average temperature data z6 of the reference day, the temperature data z7 at time t of the reference day, the temperature data z8 one hour before time t of the reference day, and the average humidity data z9 of the reference day;

[0064] According to the set mode, the load y at time t t The date is used as the benchmark to extract historical load characteristics:

[0065] In general mode, the extracted load data includes: load data at time t of the previous day z 10 、The last moment load of the previous day z 11 、Load data z at time t in the previous two days 12 ;

[0066] In the new message mode, the extracted load data includes: load data z at time t of the previous day 10 、The last moment load of the previous day z 11 、Load data z at time t in the previous two days 12 , the load data z of the last new message point in the new message data13 , the load data z one hour before the last new message data point 14 ;

[0067] According to the acquired data kk , the value of kk ranges from 1 to 14, and the second training data set is constructed.

[0068] The step S5 of constructing a preliminary load forecasting model set specifically includes the following steps:

[0069] Construct a set of preliminary load forecasting models: divide the loads of the C clusters in the membership matrix U into several types, and use the following methods for forecasting: simple moving average forecasting preliminary model, weighted moving average forecasting preliminary model, similar day load extrapolation forecasting preliminary model and SVR forecasting preliminary model;

[0070] When the load change is within the set range and the weather change is within the set range, the simple moving average prediction preliminary model is used for load forecasting; the prediction formula of the simple moving average prediction preliminary model is expressed as:

[0071]

[0072] Where F t is the prediction result; P t-n is the load value predicted n days before the day;

[0073] When the load data presents a set trend change, the weighted moving average prediction preliminary model is used for load forecasting; the prediction formula of the weighted moving average prediction preliminary model is expressed as:

[0074] F t =w1P t-1 +w2P t-2 +...+w n P t-n

[0075] Where w n is the weight value of n days before the prediction date;

[0076] When the seasonal or periodic change of load is greater than the set value, the load forecasting is carried out using the preliminary model of similar day load extrapolation forecasting. The forecasting formula of the preliminary model of similar day load extrapolation forecasting is expressed as:

[0077] P fi+1 =P i (1+ΔP iav )

[0078] Where P fi+1 is the predicted load value at time i+1; P i is the load value at time i; ΔPiav is the average daily load change rate, and M is the number of days of the selected similar day samples, ΔP j,i is the sample load change rate of the i-th point data of the load curve of the j-th similar day selected and P j,i is the data of the i-th point of the load curve on the j-th similar day;

[0079] When the load data presents nonlinear characteristics, the SVR preliminary prediction model is used for load forecasting.

[0080] Step S7 uses the model in the load forecasting model set obtained in step S6 to forecast the loads of each voltage level in the target power system, and summarizes and synthesizes them to complete the total load forecast based on the voltage level of the target power system, which specifically includes the following steps:

[0081] For the loads of each voltage level in the target power system, according to the application scenarios of the loads of each voltage level, a corresponding model in the load prediction model set obtained in step S6 is selected to perform load prediction;

[0082] Finally, the load forecast results of each voltage level are summed up to obtain the total load forecast of the target power system, thus completing the total load forecast of the target power system based on voltage level.

[0083] The present invention also provides a system for implementing the method for predicting the total load of an electric power system based on voltage levels, comprising a data acquisition module, a data processing module, a cluster analysis module, a data construction module, a model construction module, a model training module and a load prediction module; the data acquisition module, the data processing module, the cluster analysis module, the data construction module, the model construction module, the model training module and the load prediction module are connected in series in sequence; the data acquisition module is used to acquire historical data information of a target electric power system, and upload the data information to the data processing module; the data processing module is used to pre-process the acquired historical data information according to the received data information, classify the data according to the voltage level, and construct a corresponding feature vector, and upload the data information to the cluster analysis module; the cluster analysis module is used to perform cluster analysis on the constructed feature vector according to the received data information, and upload the data information to the cluster analysis module The data transmission construction module is used to construct a training data set based on the received data information and the obtained clustering analysis results, based on fuzzy processing, similarity analysis and correlation analysis, and upload the data information to the model construction module; the model construction module is used to construct a preliminary load prediction model set based on the received data information, and upload the data information to the model training module; the model training module is used to train the models in the constructed preliminary load prediction model set based on the received data information, using the obtained training data set, to obtain the load prediction model set, and upload the data information to the load prediction module; the load prediction module is used to predict the load of each voltage level in the target power system based on the received data information, using the models in the obtained load prediction model set, and summarize and integrate to complete the total load prediction of the target power system based on the voltage level.

[0084] The method and system for predicting the total load of an electric power system based on voltage levels provided by the present invention take into account the load characteristics of different voltage levels, extract corresponding historical data for loads of different voltage levels, construct and train corresponding prediction models, and perform targeted predictions on loads of different voltage levels, and finally summarize the prediction results; therefore, the present invention can not only realize the prediction of the total load of an electric power system based on voltage levels, but also has higher reliability and better accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] Figure 1 The figure is a schematic diagram of the method flow of the present invention.

[0086] Figure 2 Schematic diagram of the functional modules of the system of the present invention. DETAILED DESCRIPTION

[0087] like Figure 1The method flow chart of the method of the present invention is shown as follows: the method for predicting the total load of the power system based on voltage level disclosed by the present invention comprises the following steps:

[0088] S1. Obtain historical data information of the target power system; specifically includes the following steps:

[0089] Obtain historical data information of the target power system;

[0090] The historical data information includes historical load data information of different voltage levels, historical temperature data, historical humidity data, historical rainfall data and historical weather data;

[0091] The voltage levels include 500kV, 330kV, 220kV and 110kV; the rainfall data include no rain, light rain, moderate rain and heavy rain; the weather data include sunny, cloudy, overcast and rainy;

[0092] S2. Preprocess the historical data information obtained in step S1, classify them according to voltage levels, and construct corresponding feature vectors; specifically, the steps include:

[0093] Preprocessing the historical data information obtained in step S1;

[0094] The preprocessing includes missing value interpolation; if the number of consecutive missing data is not greater than a set value, the missing data is interpolated using linear interpolation; if the number of consecutive missing data is greater than a set value, the missing data is interpolated using load data of similar historical days;

[0095] Classify the pre-processed historical data information according to voltage levels;

[0096] Construct the corresponding feature vector:

[0097] According to the daily load data P = {P1, P2, ..., P n}Construct feature vector, P n is the nth sampling data in the daily load data; the elements in the feature vector include the load average value Maximum daily load P max =max{P1,P2,...,P n}、Daily minimum load P min =min{P1,P2,...,P n}、Load rate Peak-to-valley rate

[0098] Construct the corresponding feature vector x as

[0099] S3. Perform cluster analysis on the feature vector constructed in step S2; specifically comprising the following steps:

[0100] The fuzzy mean cluster analysis of historical load data at the same voltage level includes the following steps:

[0101] A. Set the fuzzy index, usually set to 2;

[0102] B. Initialize the membership matrix U. The size of the membership matrix U is N×C, where N is the total number of data points and C is the total number of clusters. The membership matrix U satisfies all rows U ij For data point x i The membership degree to cluster j is

[0103] C. Calculate the center v of each cluster based on the current membership matrix U j for m is the fuzzy factor;

[0104] D. Update the current membership matrix U. The update formula is: in is the element in the i-th row and j-th column of the updated membership matrix U, || || is the distance calculation function (preferably the Euclidean distance);

[0105] E. Repeat step C and step D until the set number of repetitions is reached or the change value of two consecutive membership matrix U is less than the set threshold; at this time, the final membership matrix U is obtained;

[0106] S4. According to the cluster analysis results obtained in step S3, a training data set is constructed based on fuzzy processing, similarity analysis and correlation analysis; specifically comprising the following steps:

[0107] a. Fuzzy processing of day type and meteorological data:

[0108] For the day type, the following formula is used for fuzzy processing:

[0109]

[0110] For rainfall data, the following formula is used for fuzzy processing:

[0111]

[0112] For relative humidity data, the following formula is used for fuzzy processing:

[0113]

[0114] Where h' is the relative humidity data after fuzzy processing; h is the relative humidity data before fuzzy processing;

[0115] The following formula is used to perform fuzzy processing on the daily maximum temperature data:

[0116]

[0117] Where t' max is the daily maximum temperature data after fuzzy processing; t max is the daily maximum temperature data before fuzzification;

[0118] The following formula is used to perform fuzzy processing on the daily minimum temperature data:

[0119]

[0120] Where t' min is the daily minimum temperature data after fuzzy processing; t min is the daily minimum temperature data before fuzzification;

[0121] For weather data, the following formula is used for fuzzy processing:

[0122]

[0123] The grey correlation analysis method is used to analyze the correlation between the fuzzy processed daily type data, rainfall data, relative humidity data, daily maximum temperature data, daily minimum temperature data and weather data and the load data: The basic idea of ​​the grey correlation analysis method is to judge whether the connection is close based on the similarity of the geometric shapes of the sequence curves. The closer the curves are, the greater the correlation between the corresponding sequences, and vice versa. Correlation analysis is to accurately express the difference in geometric shapes between the curves with numerical values.

[0124] The grey correlation coefficient between the influencing factors and the load data is calculated using the following formula. The influencing factors include the day type data, rainfall data, relative humidity data, daily maximum temperature data, daily minimum temperature data and weather data after fuzzy processing:

[0125]

[0126] Where ξ i (k) is the sequence x0 and the sequence x i The grey correlation coefficient at point k; x0(k) is the kth value of the daily maximum load time series; x i (k) is the kth value of the i-th influencing factor sequence; ρ is the set resolution coefficient, ranging from 0 to 1, and generally taken as 0.5; Indicates the minimum range of the second level; Indicates the maximum range of the second level;

[0127] According to the grey correlation coefficient, calculate the influencing factor sequence x i The correlation R with the daily maximum load sequence x0 i for N is the total number of elements in the sequence;

[0128] The influencing factors with the largest correlation are selected as the main influencing factors of the daily maximum load, and the first training data set is constructed;

[0129] b. Extract the load data y at time t t The time attribute features of the system include week type features, month features, and holiday features, and perform fuzzy processing:

[0130] For the week type features, the following formula is used for fuzzy processing:

[0131]

[0132] For the monthly features, the following formula is used for fuzzy processing:

[0133]

[0134] For holiday features, the following formula is used for fuzzy processing:

[0135]

[0136] Take the load y at time t t Taking the date as the reference, obtain the average temperature data z4 of the previous day, the temperature data z5 at time t of the previous day, the average temperature data z6 of the reference day, the temperature data z7 at time t of the reference day, the temperature data z8 one hour before time t of the reference day, and the average humidity data z9 of the reference day;

[0137] According to the set mode, the load y at time t t The date is used as the benchmark to extract historical load characteristics:

[0138] In general mode, the extracted load data includes: load data at time t of the previous day z 10 、The last moment load of the previous day z 11 、Load data z at time t in the previous two days 12 ;

[0139] In the new message mode, the extracted load data includes: load data z at time t of the previous day 10 、The last moment load of the previous day z 11 、Load data z at time t in the previous two days 12, the load data z of the last new message point in the new message data 13 , the load data z one hour before the last new message data point 14 ;

[0140] According to the acquired data kk , the value of kk ranges from 1 to 14, and the second training data set is constructed;

[0141] S5. Construct a preliminary load forecasting model set; specifically, the steps include:

[0142] Construct a preliminary load forecasting model set: divide the loads of the C clusters in the membership matrix U into several (preferably 4) types, and use the following methods for forecasting: simple moving average forecasting preliminary model, weighted moving average forecasting preliminary model, similar day load extrapolation forecasting preliminary model and SVR forecasting preliminary model;

[0143] When the load change is within the set range and the weather change is within the set range, the simple moving average forecasting preliminary model is used for load forecasting. This method can effectively eliminate random fluctuations in the forecast; the forecasting formula of the simple moving average forecasting preliminary model is expressed as:

[0144]

[0145] Where F t is the prediction result; P t-n is the load value predicted n days before the day;

[0146] When the load data presents a set trend change, the weighted moving average prediction preliminary model is used for load forecasting; the prediction formula of the weighted moving average prediction preliminary model is expressed as:

[0147] F t =w1P t-1 +w2P t-2 +...+w n P t-n

[0148] Where w n is the weight value of n days before the prediction date;

[0149] When the seasonal or periodic change of load is greater than the set value, the load forecasting is carried out using the preliminary model of similar day load extrapolation forecasting. The forecasting formula of the preliminary model of similar day load extrapolation forecasting is expressed as:

[0150] P fi+1 =P i (1+ΔP iav )

[0151] Where P fi+1is the predicted load value at time i+1; P i is the load value at time i; ΔP iav is the average daily load change rate, and M is the number of days of the selected similar day samples, ΔP j,i is the sample load change rate of the i-th point data of the load curve of the j-th similar day selected and P j,i is the data of the i-th point of the load curve on the j-th similar day;

[0152] When the load data presents nonlinear characteristics, the SVR prediction preliminary model is used for load forecasting; SVR aims to find a function that minimizes the difference between the training data and the actual value while allowing a certain error range; this is achieved by minimizing the norm of the weight vector w, while also taking into account the prediction bias and some slack variables;

[0153] S6. Using the training data set obtained in step S4, the model in the load forecasting preliminary model set constructed in step S5 is trained to obtain a load forecasting model set;

[0154] S7. Using the model in the load forecasting model set obtained in step S6, forecast the loads of each voltage level in the target power system, and summarize and synthesize to complete the total load forecast based on the voltage level of the target power system; specifically comprising the following steps:

[0155] For the loads of each voltage level in the target power system, according to the application scenarios of the loads of each voltage level, a corresponding model in the load prediction model set obtained in step S6 is selected to perform load prediction;

[0156] Finally, the load forecast results of each voltage level are summed up to obtain the total load forecast of the target power system, thus completing the total load forecast of the target power system based on voltage level.

[0157] like Figure 2The figure shows a schematic diagram of the functional modules of the system of the present invention: the system disclosed in the present invention for implementing the method for predicting the total load of the power system based on voltage levels comprises a data acquisition module, a data processing module, a clustering analysis module, a data construction module, a model construction module, a model training module and a load prediction module; the data acquisition module, the data processing module, the clustering analysis module, the data construction module, the model construction module, the model training module and the load prediction module are connected in series in sequence; the data acquisition module is used to acquire the historical data information of the target power system and upload the data information to the data processing module; the data processing module is used to pre-process the acquired historical data information according to the received data information, classify the data according to the voltage level, construct the corresponding feature vector, and upload the data information to the clustering analysis module; the clustering analysis module is used to perform cluster analysis on the constructed feature vector according to the received data information , and upload the data information to the data construction module; the data construction module is used to construct a training data set based on the received data information, the obtained clustering analysis results, based on fuzzy processing, similarity analysis and correlation analysis, and upload the data information to the model construction module; the model construction module is used to construct a preliminary load forecasting model set according to the received data information, and upload the data information to the model training module; the model training module is used to train the models in the constructed preliminary load forecasting model set according to the received data information, using the obtained training data set, to obtain the load forecasting model set, and upload the data information to the load forecasting module; the load forecasting module is used to predict the load of each voltage level in the target power system according to the received data information, using the models in the obtained load forecasting model set, and summarize and integrate to complete the total load forecast of the target power system based on the voltage level.

Claims

1. A method for predicting the total load of a power system based on voltage levels, comprising the following steps: S1. Obtain historical data information of the target power system; S2. Preprocess the historical data information obtained in step S1, classify them according to voltage levels, and construct corresponding feature vectors; S3. Perform cluster analysis on the feature vector constructed in step S2; S4. According to the cluster analysis results obtained in step S3, a training data set is constructed based on fuzzy processing, similarity analysis and correlation analysis; S5. Construct a preliminary model set for load forecasting; S6. Using the training data set obtained in step S4, the model in the load forecasting preliminary model set constructed in step S5 is trained to obtain a load forecasting model set; S7. Use the model in the load forecasting model set obtained in step S6 to forecast the load of each voltage level in the target power system, and summarize and synthesize them to complete the total load forecast of the target power system based on the voltage level.

2. The method for predicting the total load of a power system based on voltage level according to claim 1 is characterized in that The acquisition of historical data information of the target power system described in step S1 specifically includes the following steps: Obtain historical data information of the target power system; The historical data information includes historical load data information of different voltage levels, historical temperature data, historical humidity data, historical rainfall data and historical weather data; The voltage levels include 500kV voltage level, 330kV voltage level, 220kV voltage level and 110kV voltage level; the rainfall data includes no rain, light rain, moderate rain and heavy rain; the weather data includes sunny, cloudy, overcast and rainy.

3. The method for predicting the total load of a power system based on voltage level according to claim 2 is characterized in that Step S2 preprocesses the historical data information obtained in step S1, classifies them according to voltage levels, and constructs corresponding feature vectors, which specifically includes the following steps: Preprocessing the historical data information obtained in step S1; The preprocessing includes missing value interpolation; if the number of consecutive missing data is not greater than a set value, linear interpolation is used to interpolate the missing data; If the number of consecutive missing data is greater than the set value, the load data of similar historical days are used to interpolate the missing data; Classify the pre-processed historical data information according to voltage levels; Construct the corresponding feature vector: According to the daily load data P = {P1, P2, ..., P n }Construct feature vector, P n is the nth sampling data in the daily load data; the elements in the feature vector include the load average value Maximum daily load P max =max{P1,P2,...,P n }、Daily minimum load P min =min{P1,P2,...,P n }、Load rate Peak-to-valley rate Construct the corresponding feature vector x as 4. The method for predicting the total load of a power system based on voltage level according to claim 3 is characterized in that The cluster analysis of the feature vector constructed in step S2 described in step S3 specifically includes the following steps: The fuzzy mean cluster analysis of historical load data at the same voltage level includes the following steps: A. Set the fuzzy index; B. Initialize the membership matrix U. The size of the membership matrix U is N×C, where N is the total number of data points and C is the total number of clusters. The membership matrix U satisfies all rows U ij For data point x i The degree of membership to cluster j; C. Calculate the center v of each cluster based on the current membership matrix U j for m is the fuzzy factor; D. Update the current membership matrix U. The update formula is: in is the element of the i-th row and j-th column of the updated membership matrix U, || || is the distance calculation function; E. Repeat step C and step D until the set number of repetitions is reached or the change value of two consecutive membership matrices U is less than the set threshold; at this time, the final membership matrix U is obtained.

5. The method for predicting the total load of a power system based on voltage level according to claim 4 is characterized in that According to the cluster analysis results obtained in step S3, step S4 constructs a training data set based on fuzzy processing, similarity analysis and correlation analysis, which specifically includes the following steps: a. Fuzzy processing of day type and meteorological data: For the day type, the following formula is used for fuzzy processing: For rainfall data, the following formula is used for fuzzy processing: For relative humidity data, the following formula is used for fuzzy processing: Where h' is the relative humidity data after fuzzy processing; h is the relative humidity data before fuzzy processing; The following formula is used to perform fuzzy processing on the daily maximum temperature data: Where t' max is the daily maximum temperature data after fuzzy processing; t max is the daily maximum temperature data before fuzzification; The following formula is used to perform fuzzy processing on the daily minimum temperature data: Where t' min is the daily minimum temperature data after fuzzy processing; t min is the daily minimum temperature data before fuzzification; For weather data, the following formula is used for fuzzy processing: The grey correlation analysis method is used to analyze the correlation between the fuzzy processed daily type data, rainfall data, relative humidity data, daily maximum temperature data, daily minimum temperature data and weather data and the load data: The grey correlation coefficient between the influencing factors and the load data is calculated using the following formula. The influencing factors include the day type data, rainfall data, relative humidity data, daily maximum temperature data, daily minimum temperature data and weather data after fuzzy processing: Where ξ i (k) is the sequence x0 and the sequence x i The grey correlation coefficient at point k; x0(k) is the kth value of the daily maximum load time series; x i (k) is the kth value of the i-th influencing factor sequence; ρ is the set resolution coefficient; Indicates the minimum range of the second level; Indicates the maximum range of the second level; According to the grey correlation coefficient, calculate the influencing factor sequence x i The correlation R with the daily maximum load sequence x0 i for N is the total number of elements in the sequence; The influencing factors with the largest correlation are selected as the main influencing factors of the daily maximum load, and the first training data set is constructed; b. Extract the load data y at time t t The time attribute features of the system include week type features, month features, and holiday features, and perform fuzzy processing: For the week type features, the following formula is used for fuzzy processing: For the monthly features, the following formula is used for fuzzy processing: For holiday features, the following formula is used for fuzzy processing: Take the load y at time t t Taking the date as the reference, obtain the average temperature data z4 of the previous day, the temperature data z5 at time t of the previous day, the average temperature data z6 of the reference day, the temperature data z7 at time t of the reference day, the temperature data z8 one hour before time t of the reference day, and the average humidity data z9 of the reference day; According to the set mode, the load y at time t t The date is used as the benchmark to extract historical load characteristics: In general mode, the extracted load data includes: load data at time t of the previous day z 10 、The last moment load of the previous day z 11 、Load data z at time t in the previous two days 12 ; In the new message mode, the extracted load data includes: load data z at time t of the previous day 10 、The last moment load of the previous day z 11 、Load data z at time t in the previous two days 12 , the load data z of the last new message point in the new message data 13 , the load data z one hour before the last new message data point 14 ; According to the acquired data kk , the value of kk ranges from 1 to 14, and the second training data set is constructed.

6. The method for predicting the total load of a power system based on voltage level according to claim 5 is characterized in that The step S5 of constructing a preliminary load forecasting model set specifically includes the following steps: Construct a set of preliminary load forecasting models: divide the loads of the C clusters in the membership matrix U into several types, and use the following methods for forecasting: simple moving average forecasting preliminary model, weighted moving average forecasting preliminary model, similar day load extrapolation forecasting preliminary model and SVR forecasting preliminary model; when the load change is within the set range and the weather change is within the set range, the simple moving average forecasting preliminary model is used for load forecasting; the forecasting formula of the simple moving average forecasting preliminary model is expressed as: Where F t is the prediction result; P t-n is the load value predicted n days before; When the load data presents a set trend change, the weighted moving average prediction preliminary model is used for load forecasting; the prediction formula of the weighted moving average prediction preliminary model is expressed as: F t =w1P t-1 +w2P t-2 +...+w n P t-n Where w n is the weight value of n days before the prediction date; When the seasonal or periodic change of the load is greater than the set value, the load forecasting is carried out using the preliminary model of similar day load extrapolation forecasting. The forecasting formula of the preliminary model of similar day load extrapolation forecasting is expressed as: P fi+1 =P i (1+ΔP iav ) Where P fi+1 is the predicted load value at time i+1; P i is the load value at time i; ΔP iav is the average daily load change rate, and M is the number of days of the selected similar day samples, ΔP j,i is the sample load change rate of the i-th point data of the load curve of the j-th similar day selected and P j,i is the data of the i-th point of the load curve on the j-th similar day; When the load data presents nonlinear characteristics, the SVR preliminary prediction model is used for load forecasting.

7. The method for predicting the total load of a power system based on voltage level according to claim 6 is characterized in that Step S7 uses the model in the load forecasting model set obtained in step S6 to forecast the loads of each voltage level in the target power system, and summarizes and synthesizes them to complete the total load forecast based on the voltage level of the target power system, which specifically includes the following steps: For the loads of each voltage level in the target power system, according to the application scenarios of the loads of each voltage level, a corresponding model in the load prediction model set obtained in step S6 is selected to perform load prediction; Finally, the load forecast results of each voltage level are summed up to obtain the total load forecast of the target power system, thus completing the total load forecast of the target power system based on the voltage level.

8. A system for implementing the method for predicting the total load of a power system based on voltage levels as claimed in any one of claims 1 to 7, characterized in that It includes a data acquisition module, a data processing module, a cluster analysis module, a data construction module, a model construction module, a model training module and a load forecasting module; the data acquisition module, the data processing module, the cluster analysis module, the data construction module, the model construction module, the model training module and the load forecasting module are connected in series in sequence; the data acquisition module is used to obtain historical data information of the target power system and upload the data information to the data processing module; The data processing module is used to pre-process the historical data information obtained according to the received data information, classify it according to the voltage level, construct the corresponding feature vector, and upload the data information to the cluster analysis module; The cluster analysis module is used to perform cluster analysis on the constructed feature vector according to the received data information, and upload the data information to the data construction module; The data construction module is used to construct a training data set based on the received data information and the obtained cluster analysis results, based on fuzzy processing, similarity analysis and correlation analysis, and upload the data information to the model construction module; The model building module is used to build a preliminary load forecasting model set based on the received data information and upload the data information to the model training module; The model training module is used to train the model in the constructed load forecasting preliminary model set according to the received data information and the obtained training data set, obtain the load forecasting model set, and upload the data information to the load forecasting module; The load forecasting module is used to forecast the load of each voltage level in the target power system based on the received data information and adopt the model in the load forecasting model set, and summarize and integrate them to complete the total load forecast of the target power system based on the voltage level.

Images