Terminal Cooling Load Forecasting Model Based on Particle Swarm Optimization

Abstract

With the development of the civil aviation industry, the passenger throughput of airports has increased explosively, and they need to carry a large number of passengers every day and maintain operations for a long time. These factors cause the large space buildings in the airport to have higher energy consumption than ordinary buildings and have energy-saving potential. In practical engineering, there are problems such as low accuracy of prediction results due to inability to provide accurate building parameters and design parameters, some scholars oversimplify the large space building load forecasting model, and the prediction results have no reference significance. Therefore, establishing a load forecasting model that is closer to the actual operating characteristics and laws of large space buildings has become a research difficulty. This paper analyzes and compares the building and load characteristics of airport large space buildings, which are different from general large space buildings. The factors influencing large space architecture are divided into time characteristics and space characteristics, and the influencing reasons and characteristics of each factor are discussed. The Pearson analysis method is used to eliminate the influence parameters that have a very low connection with the cooling load, and then the historical data that affect the cooling load parameters are input. The MATLAB software is used to select a variety of neural network models for training and prediction. On this basis, the particle swarm optimization algorithm is used to optimize the prediction model. The results show that the prediction effect of the gated recurrent neural network based on particle swarm optimization algorithm is the best, the average absolute percentage error is only 0.7%, and the prediction accuracy is high.