An integrated and efficient expert system for optimization of annual household electricity demand forecasting with fuzzy and crisp inputs

摘要

This study presents an expert system based on Artificial Neural Network (ANN), Genetic Algorithm (GA) and Adaptive Network based Fuzzy Inference System (ANFIS) to forecast long-term household electricity demand with fuzzy or crisp inputs. A Graphic User Interface (GUI) is used to represent accurate information to the user with respect to ANN, GA, ANFIS and conventional regression approaches. Five major factors impacting household electricity demand forecasting in this paper are annual household electricity consumption, annual household electricity price, annual urban household size, annual urban household income and annual household electricity consumption in previous year. The integrated expert system finds the optimum numbers of training and test data and the best model parameters for ANN, GA, ANFIS and conventional regression and reports them GUI. For comparison and accuracy measurement Minimum Absolute Percentage Error (MAPE) approach is used. Also, the expert system is equipped with analysis of variance (ANOVA) for further analysis and assessment of the stated methods. This is the first study that presents an efficient and integrated expert system for optimization of household electricity demand problem with fuzzy or crisp inputs.