Hybrid Improved Differential Evolution and Wavelet Neural Network with load forecasting problem of air conditioning

Abstract

Air-conditioning load forecasting accuracy precludes high efficacy air-conditioning operation, and is also a key advantage in developing Smart Microgrid (SMG) power generating system control. The Wavelet Neural Network (WNN) was adopted as a principle element in air-conditioning load forecasting with Improved Differential Evolution Algorithm (IDEA) as an optimizing method for adjusting WNN parameters. This approach has replaced the formal feedback method used in solving network parameters. IDEA is an optimizing technique with simple calculation and fewer adjustable parameters, allowing the optimum solution for the entire system to be acquired more accurately and rapidly. After solving the optimum parameters WNN is further applied to accomplish air-conditioning load forecasting. A Fuzzy Expert System is adopted as an adjustment measure for special conditions, allowing ideal forecasting results to be reached. This study made practical comparisons among the generally applied methods for optimizing air-conditioning forecasting, such as the Artificial Neural Network (ANN), Evolutionary Programming-Artificial Neural Network (EP-ANN), Genetic Algorithm-Artificial Neural Network (GA-ANN), Ant Colony Optimization-Artificial Neural Network (ACO-ANN) and Particle Swarm Optimization-Artificial Neural Network (PSO-ANN), to prove the advantages and applicability of the proposed method.