Hybrid GOA-SVR technique for short term load forecasting during periods with substantial weather changes in North-East India

The power system load demand is extremely prejudiced by thermal inertia caused by weather parameters. This influence even became stronger when there are vast and rapid changes in the weather parameters. Therefore it is necessary to consider the weather changes for a precise short term load forecasting (STLF). This work proposes a hybrid technique for conducting STLF during periods with substantial weather changes. The proposed technique is based on support vector regressing (SVR), grasshopper optimization algorithm (GOA) and a novel Euclidean norm (EN) based data processing unit. Here GOA is utilized for accessing the appropriate SVR parameters. The EN based data processing unit is applied for fixing the training data by assessing some alike days in the recent historicized days. The proposed GOA-SVR technique identifies the substantial changes in weather (temperature) and regulates the training data accordingly. This work is done in North-East India and the proposed technique is tested on the data of load dispatch center of Assam state. Three case studies are conducted and forecasting performances of the proposed technique are compared with the conventional technique of training the system with just recent historicized day's data. The proposed technique found outperformed the conventional technique in all cases. (C) 2018 The Authors. Published by Elsevier B. V.