Enhanced Convolutional Neural Network Based Load Forecasting in Smart Grid

Abstract

In this thesis, an improved Deep Learning (DL) based technique is introduced to forecast the electricity load accurately. Energy shortage is one of the main issue in today’s world. So, an efficient mechanism is required to solve aforementioned issue. For this purpose, moving towards Smart Grids (SG) from Traditional Grids (TG) is required. Electricity load is a factor, which plays a major role in forecasting. For the solution of this problem, we proposed a model which is based on selection, extraction and classification of historical data. Grey Correlation Analysis (GCA) based Random Forest (RF) and Mutual Information (MI) is performed for feature selection; Kernel Principle Component Analysis (KPCA) is presents for extraction of important feature and enhanced Combine Feature Selection Convolutional Neural Network (CFSCNN) is proposed for electricity load forecasting. We use an Independent System Operator New England Control Area (ISO-NECA) dataset for simulations in our scenario. Half hourly and hourly load data is used in our model for prediction. The presented model performs well on provided dataset and gives accurate results. Accuracy is also checked by four performance metrics: Mean Square Error (MSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE). The presented model is also cross checked with four existing classifiers. The comparison is performed to check the model’s performance. Simulation results prove that the presented technique beats existing schemes. Our model improves the prediction accuracy of electricity load than existing techniques. Moreover, computational time is also decreased in our scenario.M