Prediction of electrical energy demand by hybridization of Particle Swarm Optimization and Noise filtering

In the past decades, there has been an implicit explosion of absorption in Artificial Intelligence (AI) field. AI supplies robust and flexible means for acquiring solutions to a diversity of problems that often cannot be solved by other, more traditional and orthodox methods. Nowadays, its use is increasing rapidly in many sectors of complicated practical problems. Meanwhile, electrical energy demand prediction is one of the important concerns of energy systems so development of intelligent prediction methods and algorithms for performing accurate predictions is essential. Various techniques have been proposed for electrical energy demand prediction over different time intervals of short term, medium term and long term. This study presents an intelligent hybrid approach based on Particle Swarm Optimization (PSO) and Noise filtering technique (which is one of Data Mining and Data Analysis concepts) for modeling and predicting long term electrical energy demand with high accuracy. At the first stage the hybrid approach applies noise filtering technique on raw data that consists of influential socio-economic factors affecting electricity demand, and as its consequence the raw data will be cleaned and smoothed. At the next stage it performs feature selection technique to determine the most influential factors and eliminate low impact factors. Subsequently selected filtered factors will be fed into PSO in order to build the prediction model (Filtered-PSO). For the purpose of investigating and validating influence of noise filtering on improvement of PSO, it compares results of the hybrid approach with Simple-PSO model which is built using raw data. It carries out the comparisons by means of paired t-test to check if the improvement is statistically significant or not. Results show that the Filtered-PSO model significantly outperforms Simple-PSO and we may consider noise filtering as an effective concept of data mining to be integrated with intelligent approaches such as - - PSO in order to improve accuracy of electrical energy demand predictions.