the method for removing abnormal day in system for predicting electric power demand

摘要

The present invention selects a similar day suitable for power demand prediction by measuring the pattern similarity of power demand through the Scaled RMSE technique, (1) an electric power calculation step of obtaining the daily average electric power amount and the electric power amount for each time period of the data to be compared with the pattern of the electric power demand; (2) a power amount scaling step of selecting a standard for the daily average power amount as a value of '1' and scaling the power amount for each time period according to the standard; (3) an RMSE calculation step of measuring the similarity of the pattern of power demand through the RMSE calculation of the amount of power for each time period between the scaled data; (4) a similar day selection step for selecting a reference date suitable for power demand prediction through the RMSE calculation; It provides a method of selecting a similar date through a similarity analysis of a pattern comprising And in the power demand prediction system for predicting the power demand of the future forecast day with the amount of power on the reference days selected as similar days, In order to eliminate abnormal reference days with significantly low electricity among past reference days that reduce the accuracy of the predicted future electricity demand, The average daily power consumption of the past reference days ( ) and one-day standard deviation ( ) by using (1) Centerline ( ) and the lower management line ( ) to find and write control chart writing step; (2) the lower management line ( ) a first abnormal reference date removal step of removing the abnormal reference date deviated to the outside; (3) performing step (1) with data in a state in which the abnormal reference date has been removed control chart rewriting step; (4) center line ( ) and the lower management line ( ) to find and write control chart writing step; (5) the lower management line ( ) a second abnormal reference date removal step of removing the abnormal reference date deviated to the outside; (6) performing the step (4) with the data in a state in which the abnormal reference date has been removed control chart rewriting step; It provides a method of removing the abnormal reference date of the power demand forecasting system, characterized in that it comprises a. In addition, the same day of the week selected as a similar day in the community unit is used as the past reference date to predict the electricity demand of the forecasted day, and to improve the prediction accuracy by correcting the average electric energy on the forecasted day through the average electric energy correction factor calculated by considering the temperature factor As an electricity demand forecasting system that (1) calculating the amount of electricity for each time period of each of the reference days and the maximum and minimum amount of electricity for each of the reference days; (2) Using the exponential smoothing method with the numerical value obtained in step (1) above, the maximum power demand ( ) and the minimum power demand ( ), the maximum and minimum amount of electricity to predict; (3) Select the standard as a value of '1' for the maximum amount of power from the value obtained in step (1) above, and scale the amount of power for each time period according to the standard and normalize ( ) input data normalization step; (4) Using exponential smoothing method to predict the normalized power demand on the forecast day, which is the same day of the future, for each time period ( ) and the normalized forecast day power demand forecasting and renormalization step of renormalizing so that the maximum value is 1; (5) With the numerical value obtained in step (2) and the normalized power demand of the forecast day obtained in step (4), the predicted power demand of the forecast day to be predicted ( ) the electricity demand forecasting step to obtain; (6) Correction coefficient according to seasonal temperature and daily average power ( , ) calculating the average wattage correction coefficient calculating step; (7) the correction factor ( , ) by correcting the average daily power consumption of the forecasted day using ), the final corrected forecasted electric power demand by time of the same day of the week in the future ( ) the final power demand prediction step to obtain; It provides an improved power demand prediction system of the correction of the daily average power amount in consideration of the temperature factor, characterized in that it comprises a. It is also a power demand prediction system that attempts to predict the power demand on the same day in the future with the amount of electricity on the same days in the past. (1) calculating the amount of electricity for each time period of each of the reference days and the maximum and minimum amount of electricity for each of the reference days; (2) Correction coefficient according to the seasonal temperature and the amount of electricity for each time period obtained in step (1) above ( , , , ) calculating the maximum and minimum wattage correction coefficient calculation step; (3) the correction factor ( , , , ), the corrected maximum power demand ( ) and the corrected minimum power demand ( ) a power amount correction step to obtain; (4) Using the exponential smoothing method, the maximum power demand ( ) and the minimum power demand ( ) a maximum and minimum amount of electricity to predict; (5) Normalized by selecting the maximum power amount as a value of '1' from the value obtained in step (1) above, and scaling the power amount for each time period according to the standard ( ) input data normalization step; (6) Normalized electricity demand forecasting on the forecast day, which predicts the normalized power demand on the forecast day, the same day of the week, for each time period, and renormalizes it so that the maximum value is 1. and renormalization steps; (7) With the numerical value obtained in step (4) and the normalized future demand for the same day of the week obtained in step (6), the predicted power demand by time of the same day of the future to be predicted ( ) the final power demand prediction step to obtain; It provides an improved power demand prediction system of correcting the maximum and minimum wattage per day in consideration of the temperature factor, characterized in that it comprises a.