SUPPORT VECTOR MACHINE BASED MODEL DESIGN FOR SHORT TERM LOAD FORECAST IN DISTRIBUTION SYSTEM by Manmeet Kaur
Abstract
Future electric load, historical electric load data, weather information, and predicted weather information is predicted and the process is called Electric load Forecasting. Various models have been developed to predict the electric load more precisely. It is divided in three types Long-
term load forecasting
Medium-term forecasting
Short-term forecasting
Long-term electric load forecasting: - it is used to predict the requirement for spreading out the system, purchasing new machinery, and staff requirement for smooth conduct. Medium-term forecasting:- it is used for the function of arrangement of fuel supplies and division repairs. Short-term forecasting:- in this predication method important information for the smooth conduct of system
management is given. In the present work, the center of attention is on short-term load forecasting (STLF). In this type of prediction method time horizon is predicted from 1 hour to 1 week. Short-term load forecasting has various other benefits for e.g it helps cut the spinning reserve power and make the proper plan for instrument maintenance. A new loom will be developed by this study for the the prediction of load in power grid system. Data is divided into various parts by using KNN and after that SVM classifier is put into practice for dividing, instructing checking the data. Accuracy, Mean square error, mean absolute error, mean absolute percentage error and error rate is enhanced by the permutation of KNN and SVM. The examination of normal SVM based load forecasting system is performed in the present study with help of MATLAB. Three scale is set to illustrate the proposed method is important and efficient to various other methods. The precision of work proposed is 95% to 100% . It is very high than the traditional system of work
