Load prediction method and system for intelligent power distribution system

Abstract

The invention discloses a load prediction method and system for an intelligent power distribution system, relates to the technical field of load prediction of the power distribution system, and solves the technical problem that the model prediction accuracy is affected by taking numerous influence factor data of the load of the power distribution system as training data of a machine learning model for training in the prior art. According to the method, the abnormality of load data and influence factor data of the power distribution system in a historical evaluation period is processed; after processing, analyzing the correlation between the load of the power distribution system and each influence factor in the historical evaluation period, and obtaining a strong correlation influence factor; predicting a load change condition based on the prediction data change condition of the strong correlation influence factors to obtain a prediction load, and analyzing a demand load; comparing the predicted load with the demand load, and optimizing power system distribution; according to the method, the strong correlation influence factors are analyzed, and the load change condition is further predicted, so that the accuracy and reliability of load prediction are greatly improved.

Description

Technical Field

[0001] The present invention belongs to the field of load prediction of power distribution systems, and in particular to a load prediction method and system for intelligent power distribution systems. Background Art

[0002] In the power system, the distribution network is a power network that receives electric energy from the transmission network and distributes it to various power users. The distribution network is directly connected to the power supply users and is an important part of the power system. Load forecasting is an important part of distribution network planning and the basis of distribution network planning. It is closely related to construction investment and is also an important condition for the power system to maintain high-quality and economical operation.

[0003] At present, by taking various factors that affect the load data of the distribution system, such as seasonal changes, weather conditions and population growth, as input data and load data as output data, training linear models such as time series models and machine learning models such as artificial neural network models, a model for predicting load is obtained to predict the future change trend of load. However, since the load is affected by many factors, and the degree of influence of each factor on the load is different, using all of them as training data for the machine learning model not only increases the complexity and volume of the data, but also weakens the training effect and prediction accuracy of the model. Therefore, the present invention provides a load prediction method and system for an intelligent distribution system. Summary of the invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes a load prediction method and system for an intelligent power distribution system, which is used to solve the technical problem that the prior art uses data of numerous influencing factors of the power distribution system load as training data for a machine learning model, thereby affecting the accuracy of model prediction.

[0005] To achieve the above-mentioned object, a first aspect of the present invention provides a load prediction system for an intelligent power distribution system, comprising a data processing module and a prediction module;

[0006] Data analysis module: used to analyze whether there are any abnormalities in the load data and influencing factor data of the power distribution system within the historical evaluation period; if yes, analyze the abnormality type and process the abnormal data based on the abnormality type; if no, do not process it; the historical evaluation period includes day, week and month;

[0007] After abnormal data processing, the correlation between the load of the distribution system and various influencing factors during the historical evaluation period is analyzed to obtain the strongly correlated influencing factors;

[0008] Prediction module: Predict load changes based on changes in forecast data of strongly correlated influencing factors, obtain predicted load, and analyze demand load; compare predicted load with demand load to optimize power system distribution.

[0009] Preferably, the load data and influencing factor data of the power distribution system within the historical evaluation period include:

[0010] Extracting a number of data of various influencing factors of the load in the historical assessment period from the historical data to obtain a number of influencing factor data sequences; and,

[0011] A number of load data corresponding to the influencing factor data sequence within the historical assessment period are extracted to obtain a number of load data sequences; wherein the influencing factor data sequence / load data sequence both include a number of daily data sequences, a number of weekly data sequences and a number of monthly data sequences.

[0012] The present invention can obtain information of different dimensions by extracting data series at three different time scales: daily, weekly and monthly. For example, the daily data series can capture short-term load fluctuations, such as the impact of weather changes, special events, etc. The weekly data series is helpful in discovering the periodic laws of load, such as the difference between weekdays and weekends. The monthly data series can reflect the long-term trend of load, such as seasonal changes, economic growth, etc. Combining and analyzing the data at these three scales can provide a more comprehensive understanding of the changing characteristics of load.

[0013] Preferably, the analyzing whether there are any abnormalities in the load data and influencing factor data of the power distribution system during the historical evaluation period includes:

[0014] Plotting several influencing factor data series and several load data series in the coordinate system respectively, and obtaining the change curves within the historical evaluation period; wherein the types of change curves include daily change curves, weekly change curves and monthly change curves;

[0015] If there is a sharp drop point on the change curve and the value is 0, it means data is missing; when there is a slope change trend in the change curve, it means a brief error occurred in the short-term data collection process; when the change curve is a straight line or there are several fluctuations on the straight line, the distribution system is faulty or under maintenance; the fluctuation range is within the preset range.

[0016] The present invention plots several influencing factor data sequences and load data sequences in a coordinate system respectively to form daily, weekly and monthly variation curves, and identifies data anomalies by observing the trend variation characteristics of the variation curves, so that the data fluctuations, periodic laws and long-term trends are clear at a glance, providing a clear perspective for data anomaly detection and data analysis, helping analysts to quickly locate problems and correct the extracted data, providing accurate data for subsequent analysis of the correlation between influencing factors and loads.

[0017] Preferably, the data in the daily data sequence is the mean of the daily influencing factor / load data changes;

[0018] The data in the weekly data series are the mean values ​​of the weekly influencing factors / load data changes;

[0019] The data in the monthly data series are the mean values ​​of the monthly influencing factors / load data changes.

[0020] Preferably, the processing of abnormal data based on the abnormal type includes:

[0021] When data is missing or there is a temporary error, the mean value between the previous moment and the next moment is used to supplement the data;

[0022] When the fluctuation range of the change curve is within the preset fluctuation range, the data corresponding to the change curve is deleted, and the data of the influencing factors at the corresponding moment is deleted.

[0023] Preferably, the analysis of the correlation between the load of the power distribution system and various influencing factors within the historical evaluation period to obtain strongly correlated influencing factors includes:

[0024] Calculate the Pearson correlation coefficient between the daily / weekly / monthly data series of the influencing factors and the daily / weekly / monthly data series of the corresponding load to obtain the daily / weekly / monthly correlation coefficient, and perform mean processing on several daily / weekly / monthly correlation coefficients to obtain the daily average / weekly average / monthly average correlation coefficient;

[0025] The daily average correlation coefficient, weekly average correlation coefficient and monthly average correlation coefficient are processed by averaging to obtain the positive correlation index between each influencing factor and load;

[0026] When the positive correlation index is greater than the preset threshold, there is a positive correlation between the influencing factor and the load, that is, the influencing factor is a strongly correlated influencing factor.

[0027] The present invention calculates the Pearson correlation coefficient between daily, weekly and monthly data sequences. This method can accurately quantify the strength and direction of the linear relationship between the influencing factors and the load, and reflects the degree of linear correlation between the two variables. The value is between -1 and 1. The larger the value, the stronger the correlation. The factors that have the greatest impact on load changes are obtained through analysis, and the changes in load with strongly correlated influencing factors are effectively reflected, avoiding the interference of irrelevant or less correlated influencing factors, which not only reduces the amount of data processed, but also improves the accuracy and efficiency of the analysis.

[0028] Preferably, the load change condition is predicted based on the predicted data change condition of the strongly correlated influencing factors, including:

[0029] The prediction data of strongly correlated influencing factors are input into the load forecasting model to predict future load changes; among them, the load forecasting model is built based on the LSTM model.

[0030] The present invention makes full use of the information of strongly correlated influencing factors and the positive correlation between the influencing factors and the load determined through preliminary analysis, thereby ensuring that the data input into the load forecasting model has a high degree of correlation and predictive value, which helps to improve the accuracy and reliability of the forecasting model and make the forecast results closer to the actual load changes; secondly, the load forecasting model constructed based on the LSTM model has powerful time series processing capabilities, which is particularly suitable for processing and predicting time series data, can capture long-term dependencies in the data, and effectively deal with the nonlinear and volatility characteristics in the load data, thereby improving the generalization ability and robustness of the forecasting model.

[0031] Preferably, by inputting the forecast data of strongly correlated influencing factors of different forecast time lengths into the load forecasting model, the load changes corresponding to the different forecast time lengths are predicted; wherein the different forecast time lengths include short-term forecast, medium-term forecast and long-term forecast.

[0032] The present invention can more finely capture the short-term fluctuations, medium-term trends and long-term laws of load changes by respectively inputting the prediction data of strongly correlated influencing factors of different time lengths, thereby generating more accurate prediction results, which is of great significance for the real-time scheduling, resource allocation and fault prevention of the power system; and is helpful for the optimal allocation of resources. Through long-term prediction, the construction of power facilities can be reasonably planned to ensure the sustainability of power supply; medium-term prediction helps to adjust the power production plan and improve operational efficiency; and short-term prediction can adjust power distribution in real time to ensure the stable operation of the power grid.

[0033] Preferably, comparing the predicted load with the demand load to optimize the power system distribution includes:

[0034] Extracting a number of electricity consumption data from historical data, calculating the electricity consumption data of an average period, the demand load is the average daily electricity consumption data multiplied by the forecast time length; wherein the unit of the average period is determined based on the unit of the forecast time length;

[0035] If the demand load is greater than the total of the predicted load and the currently stored load, the power generation will be increased so that the load for the predicted time length meets the demand.

[0036] Preferably, a load forecasting method for an intelligent power distribution system comprises the following steps:

[0037] Step 1: Perform anomaly detection on the load data and influencing factor data of the distribution system within the historical evaluation period, and process the abnormal data;

[0038] Step 2: After abnormal data processing, analyze the correlation between the load of the distribution system and various influencing factors during the historical evaluation period to obtain strongly correlated influencing factors;

[0039] Step 3: Predict the load change based on the predicted data change of the strongly correlated influencing factors, obtain the predicted load, and analyze the demand load;

[0040] Step 4: Compare the forecast load with the demand load and optimize the power system distribution.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] 1. The present invention performs anomaly detection on the extracted influencing factor data and load data. Through anomaly detection, it can more accurately determine which data needs to be processed and which data can be retained, which can significantly improve the accuracy and reliability of the data, thereby ensuring the data quality of the input model and improving the accuracy of the prediction;

[0043] 2. The present invention obtains the strongly correlated influencing factors of the load by analyzing the positive correlation index between each influencing factor and the load. The introduction of the strongly correlated influencing factors makes the prediction model more robust when facing complex and changeable load changes, ensuring that the model focuses on the most critical information during the training process and avoids being disturbed by minor or irrelevant factors. This not only improves the training efficiency of the model, but also enables the model to learn the inherent laws of load changes more quickly, thereby enhancing the prediction ability;

[0044] 3. The present invention performs anomaly detection on the load data and influencing factor data of the distribution system within the historical evaluation period; after the abnormal data is processed, the correlation between the load of the distribution system and the influencing factors within the historical evaluation period is analyzed to obtain the strongly correlated influencing factors; based on the predicted data changes of the strongly correlated influencing factors, the load changes are predicted to obtain the predicted load, and the demand load is analyzed to optimize the power system distribution; the present invention improves the accuracy of load prediction by screening the input data of the model, which is conducive to the early and reasonable allocation and efficient utilization of power resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 This is a schematic diagram of abnormal data types of the present invention;

[0047] Figure 2 It is a schematic diagram of the flow chart of the positive correlation index analysis method between the influencing factors and the load of the present invention;

[0048] Figure 3 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0049] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] A first aspect of the present invention provides a load prediction system for an intelligent power distribution system, comprising a data processing module and a prediction module;

[0051] The data analysis module analyzes whether there are any abnormalities in the load data and influencing factor data of the power distribution system during the historical evaluation period; if yes, the abnormality type is analyzed and the abnormal data is processed based on the abnormality type; if no, no processing is performed;

[0052] Specifically, extracting several data of various influencing factors of the load in the historical evaluation period from the historical data to obtain several influencing factor data sequences; and,

[0053] Extract a number of load data corresponding to the influencing factor data sequence within the historical assessment period to obtain a number of load data sequences; wherein the influencing factor data sequence / load data sequence both include a number of daily data sequences, a number of weekly data sequences and a number of monthly data sequences;

[0054] Among them, the data in the daily data series is the mean of the daily changes in influencing factors / load data; the data in the weekly data series is the mean of the weekly changes in influencing factors / load data; and the data in the monthly data series is the mean of the monthly changes in influencing factors / load data.

[0055] The historical evaluation period includes days, weeks and months; it provides a higher time resolution and allows for a more detailed observation of the relationship between variables;

[0056] Daily cycle data analyzes daily load peaks and troughs to ensure adequate generation capacity during peak hours; suitable for short-term forecasts within a few hours to a day;

[0057] Periodic data is used to analyze load patterns within a week, especially the differences between weekends and weekdays, so as to make power generation plans in advance; it is suitable for medium-term forecasts within a few days to a few weeks;

[0058] Monthly cycle data is used to analyze the overall load trend of a month in order to carry out long-term grid planning and equipment investment; it is suitable for long-term forecasts within a few months to a few years.

[0059] The factors affecting the power system load mainly include weather, seasonal changes, holidays and other factors.

[0060] Plotting several influencing factor data series and several load data series in the coordinate system respectively, and obtaining the change curves within the historical evaluation period; wherein the types of change curves include daily change curves, weekly change curves and monthly change curves;

[0061] See also Figure 1 (a) If the change curve shows a sharp drop point and the value is 0, the data is missing; please refer to Figure 1 (b) When a slope trend appears in the change curve, it means that a short-term error occurred during the short-term data collection process; please refer to Figure 1 (c) When the variation curve is a straight line or there are several fluctuations on the straight line, the power distribution system fails or is under maintenance; wherein the fluctuation range is within a preset range.

[0062] When data is missing or there is a temporary error, the mean value between the previous moment and the next moment is used to supplement the data;

[0063] When the change curve is a straight line or there are several fluctuations on the straight line, the data corresponding to the change curve is deleted, and the data of the influencing factors at the corresponding moment are deleted.

[0064] After abnormal data processing, the correlation between the load of the distribution system and various influencing factors during the historical evaluation period is analyzed to obtain the strongly correlated influencing factors;

[0065] See also Figure 2 Specifically, the Pearson correlation coefficient between the daily data series of the influencing factors and the daily data series of the corresponding load is calculated to obtain the daily correlation coefficient, and the average value of several daily correlation coefficients is processed to obtain the daily average correlation coefficient;

[0066] Calculate the Pearson correlation coefficient between the weekly data series of the influencing factors and the weekly data series of the corresponding load to obtain the weekly correlation coefficient, and perform mean processing on several weekly correlation coefficients to obtain the weekly average correlation coefficient;

[0067] Calculate the Pearson correlation coefficient between the monthly data series of the influencing factors and the monthly data series of the corresponding load to obtain the monthly correlation coefficient, and perform mean processing on several monthly correlation coefficients to obtain the monthly average correlation coefficient;

[0068] The daily average correlation coefficient, weekly average correlation coefficient and monthly average correlation coefficient are processed by averaging to obtain the positive correlation index between each influencing factor and load;

[0069] For example: to analyze the correlation between temperature and load on a certain day, assuming that the daily data sequence of temperature on a certain day is Aj=[a0,a1,…,ai,…,an], and the corresponding daily data sequence of load is Fj=[f0,f1,…,fi,…,fn], the Pearson correlation coefficient between the two is WFj=Σ[(ai-EA)×(fi-Ef) / (σA×σF)] / (n-1); where EA is the mean of the daily temperature data sequence, Ef is the mean of the daily load data sequence, σA is the variance of the daily temperature data sequence, and σF is the variance of the daily load data sequence; Σ represents the sum of i; j represents the daily data sequence of temperature and load of the jth group; the Pearson correlation coefficients between several groups of temperature and load are averaged, that is, the daily average correlation coefficient between temperature and load is obtained RWF=(ΣWFj) / j max , Σ represents the summation over j;

[0070] Obtain several sets of weekly / monthly data series of temperature and load respectively, and calculate the weekly / monthly Pearson correlation coefficient between temperature and load according to the above method, and then obtain the weekly correlation coefficient ZWF and the monthly correlation coefficient YWF;

[0071] Based on the daily average correlation coefficient RWF, the weekly correlation coefficient ZWF and the monthly correlation coefficient YWF, the positive correlation index ZXS between temperature and load is calculated to be ZXS=(RWF+ZWF+YWF) / 3.

[0072] When the positive correlation index is greater than the preset threshold, there is a positive correlation between the influencing factor and the load, that is, the influencing factor is a strongly correlated influencing factor.

[0073] It should be noted that the Pearson correlation coefficient is between -1 and 1, and the closer it is to 1, the higher the correlation between the two elements.

[0074] The present invention evaluates the correlation between the influencing factors and the load through a positive correlation index; the correlation degree is mainly closely related to the daily average correlation coefficient, weekly average correlation coefficient and monthly average correlation coefficient between the influencing factors and the load; if the daily average correlation coefficient, weekly average correlation coefficient and monthly average correlation coefficient are all large, it indicates that the influencing factor has a high degree of influence on the load, that is, the obtained positive correlation coefficient is large.

[0075] The prediction module predicts the load change based on the change of the prediction data of the strongly related influencing factors to obtain the predicted load;

[0076] Specifically: the prediction data of strongly correlated influencing factors are input into the load forecasting model to predict future load changes; wherein, the load forecasting model is constructed based on the LSTM model.

[0077] It should be noted that the forecast data of strongly correlated influencing factors can be obtained from meteorological departments, government departments and other channels; for example, future weather conditions can be obtained from the meteorological department, and future holiday conditions can be obtained from government departments, etc.

[0078] Among them, by inputting the prediction data of strongly correlated influencing factors of different prediction time lengths into the load prediction model, the load changes corresponding to different prediction time lengths are predicted; among them, the different prediction time lengths include short-term prediction, medium-term prediction and long-term prediction.

[0079] For example: extract several data of strongly correlated factors and corresponding load data from historical data; group the extracted data according to different time lengths to obtain several groups, including short-term data group, medium-term data group and long-term data group; use the data of strongly correlated factors in each data group as the standard input data of the LSTM model, and the corresponding load data as the standard output data of the LSTM model, train the LSTM model with the standard input data and standard output data to obtain a load forecasting model; according to the forecasting requirements, the load forecasting model can perform short-term, medium-term or long-term forecasts.

[0080] And, analyze demand load; compare predicted load with demand load and optimize power system distribution.

[0081] Extract some electricity consumption data from historical data, calculate the electricity consumption data of the average period, and the demand load is the average daily electricity consumption data multiplied by the forecast time length;

[0082] It should be noted that the unit of the average period is determined based on the unit of the forecast time length; if the forecast time length is one day, the unit of the average period is hourly; if the forecast time length is one week, the unit of the average period is daily; if the forecast time length is one month, the unit of the average period is weekly; that is, the unit of the average unit period is one unit level lower than the forecast time length.

[0083] If the demand load is greater than the total of the predicted load and the currently stored load, that is, the predicted load cannot meet the future load demand, the power generation needs to be increased so that the load within the predicted time length can meet the future load demand.

[0084] See also Figure 3 The second aspect of the present invention provides a load forecasting method for an intelligent power distribution system, comprising the following steps:

[0085] Step 1: Perform anomaly detection on the load data and influencing factor data of the distribution system within the historical evaluation period, and process the abnormal data;

[0086] Step 2: After abnormal data processing, analyze the correlation between the load of the distribution system and various influencing factors during the historical evaluation period to obtain strongly correlated influencing factors;

[0087] Step 3: Predict the load change based on the predicted data change of the strongly correlated influencing factors, obtain the predicted load, and analyze the demand load;

[0088] Step 4: Compare the forecast load with the demand load and optimize the power system distribution.

[0089] Part of the data in the above formula is calculated by removing the dimension and taking its numerical value. The formula is a formula closest to the actual situation obtained by software simulation of a large amount of collected data; the preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained through simulation of a large amount of data.

[0090] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A load forecasting system for an intelligent power distribution system, characterized in that: Including data processing module and prediction module; Data analysis module: used to analyze whether there are any abnormalities in the load data and influencing factor data of the power distribution system during the historical evaluation period; If yes, the abnormal type is analyzed and the abnormal data is processed based on the abnormal type; if no, no processing is performed; the historical evaluation cycle includes day, week and month; After abnormal data processing, the correlation between the load of the distribution system and various influencing factors during the historical evaluation period is analyzed to obtain the strongly correlated influencing factors; Prediction module: Predict load changes based on changes in forecast data of strongly correlated influencing factors, obtain predicted load, and analyze demand load; compare predicted load with demand load to optimize power system distribution.

2. A load forecasting system for an intelligent power distribution system according to claim 1, characterized in that: The load data and influencing factor data of the power distribution system during the historical evaluation period include: Extracting a number of data of various influencing factors of the load in the historical assessment period from the historical data to obtain a number of influencing factor data sequences; and, A number of load data corresponding to the influencing factor data sequence within the historical assessment period are extracted to obtain a number of load data sequences; wherein the influencing factor data sequence / load data sequence both include a number of daily data sequences, a number of weekly data sequences and a number of monthly data sequences.

3. A load forecasting system for an intelligent power distribution system according to claim 2, characterized in that: The analysis of whether there are abnormalities in the load data and influencing factor data of the power distribution system during the historical evaluation period includes: Plotting several influencing factor data series and several load data series in the coordinate system respectively, and obtaining the change curves within the historical evaluation period; wherein the types of change curves include daily change curves, weekly change curves and monthly change curves; If there is a sharp drop point on the change curve and the value is 0, it means data is missing; when there is a slope change trend in the change curve, it means a brief error occurred in the short-term data collection process; when the change curve is a straight line or there are several fluctuations on the straight line, the distribution system is faulty or under maintenance; the fluctuation range is within the preset range.

4. A load forecasting system for an intelligent power distribution system according to claim 3, characterized in that: The data in the daily data series are the mean values ​​of the daily influencing factors / load data changes; The data in the weekly data series are the mean values ​​of the weekly influencing factors / load data changes; The data in the monthly data series are the mean values ​​of the monthly influencing factors / load data changes.

5. A load forecasting system for an intelligent power distribution system according to claim 3, characterized in that: The processing of abnormal data based on the abnormal type includes: When data is missing or there is a temporary error, the mean value between the previous moment and the next moment is used to supplement the data; When the fluctuation range of the change curve is within the preset fluctuation range, the data corresponding to the change curve is deleted, and the data of the influencing factors at the corresponding moment is deleted.

6. A load forecasting system for an intelligent power distribution system according to claim 5, characterized in that: The analysis of the correlation between the load of the power distribution system and various influencing factors during the historical evaluation period to obtain strongly correlated influencing factors includes: Calculate the Pearson correlation coefficient between the daily / weekly / monthly data series of the influencing factors and the daily / weekly / monthly data series of the corresponding load to obtain the daily / weekly / monthly correlation coefficient, and perform mean processing on several daily / weekly / monthly correlation coefficients to obtain the daily average / weekly average / monthly average correlation coefficient; The daily average correlation coefficient, weekly average correlation coefficient and monthly average correlation coefficient are processed by averaging to obtain the positive correlation index between each influencing factor and load; When the positive correlation index is greater than the preset threshold, there is a positive correlation between the influencing factor and the load, that is, the influencing factor is a strongly correlated influencing factor.

7. A load forecasting system for an intelligent power distribution system according to claim 6, characterized in that: The load change condition is predicted based on the predicted data change condition of the strongly correlated influencing factors, including: The prediction data of strongly correlated influencing factors are input into the load forecasting model to predict future load changes; among them, the load forecasting model is built based on the LSTM model.

8. A load forecasting system for an intelligent power distribution system according to claim 7, characterized in that: By inputting the forecast data of strongly correlated influencing factors of different forecast time lengths into the load forecasting model, the load changes corresponding to different forecast time lengths are predicted; wherein the different forecast time lengths include short-term forecast, medium-term forecast and long-term forecast.

9. A load forecasting system for an intelligent power distribution system according to claim 7, characterized in that: The method of comparing the predicted load with the demand load and optimizing the power system distribution includes: Extracting a number of electricity consumption data from historical data, calculating the electricity consumption data of an average period, the demand load is the average daily electricity consumption data multiplied by the forecast time length; wherein the unit of the average period is determined based on the unit of the forecast time length; If the demand load is greater than the total of the predicted load and the currently stored load, the power generation will be increased so that the load for the predicted time length meets the demand.

10. A load forecasting method for an intelligent power distribution system, based on the operation of a load forecasting system for an intelligent power distribution system according to any one of claims 1 to 9, characterized in that: The following steps are involved: Step 1: Perform anomaly detection on the load data and influencing factor data of the distribution system within the historical evaluation period, and process the abnormal data; Step 2: After abnormal data processing, analyze the correlation between the load of the distribution system and various influencing factors during the historical evaluation period to obtain strongly correlated influencing factors; Step 3: Predict the load change based on the predicted data change of the strongly correlated influencing factors, obtain the predicted load, and analyze the demand load; Step 4: Compare the forecast load with the demand load and optimize the power system distribution.

Images