A preview of the PDF is not available
Efficient energy consumption and generation in smart grids with scheduling the appliances and forecasting the load
Abstract and Figures
Energy is the most needed commodity of the current era. Recently, the energy demand is far higher than the available energy. Moreover, generation and consumption of energy shows fluctuating behavior. By the incorporation of Demand Side Management (DSM) with the Smart Grid (SG) and forecasting the load results in the solution of this problem. Different techniques are utilized in SG to provide the optimum solution for scheduling of household appliances at Home Energy Management System (HEMS), which leads to minimize the cost of electricity and manage the load in residential areas, industrial and commercial areas. It also decreases in the waiting time of appliances and Peak to Average Ratio (PAR), which leads to maximize User Comfort (UC). Forecasting of electric load can decrease the fluctuation behavior between energy generation and consumption. In this thesis, we proposed six meta heuristic techniques for the scheduling of appliances in HEMS; Firefly Algorithm (FA), Bacterial foraging Algorithm (BFA), Earth Worm Optimization Algorithm (EWA), Genetic Algorithm (GA), Hybrid of Genetic and Bacterial foraging (HBG), and Harmony Search Algorithm (HSA). We also proposed a model for forecasting of electricity load, which consists of a two step process; feature engineering (feature selection and extraction) and classification. By combining Extreme Gradient Boosting (XGBoost) and Decision Tree (DT) techniques, we proposed a hybrid feature selector to minimize the feature redundancy. Furthermore, Recursive Feature Elimination (RFE) technique is applied for dimension reduction and improve feature selection. To forecast electric load, we applied Support Vector Machine (SVM) set tuned with three super parameters, i.e., kernel parameter, cost penalty and incentive loss function parameter. Electricity market data of New England Control Area Independent System Operator (ISO-NE) is taken as input in our proposed forecasting model. After extensive simulations, we achieved a reduction in PAR, electric cost and user comfort maximization through appliances scheduling using the optimization techniques. The BFA technique performs better than other techniques in terms of electricity cost and HBG beats other techniques in terms of PAR, while the HSA outperforms in terms of UC. Using the proposed forecasting model, weekly and monthly ahead forecasting experiments are conducted. Forecasting performance is assessed by using RMSE and MAPE. The error value of MAPE and RMSE is 1.682 and 12. SVM classifier shows better accuracy i.e. 98% in terms of load forecasting.
Content may be subject to copyright.