Research Of Prediction Model On Airport's Operation Energy Demand

Recent years, the energy consumption of airport is increased as the infrastructure construction increasing rapidly. Under this background, the research work on energy demand forecasting model of airport's operation is carried out, and the experimental results show that the proposed method has a high accuracy in the prediction of airport's energy data. Giving an accurately prediction of the trend in airport's energy demand can be a good guidence for the development of airport.Firstly, the current situation of the airport's energy demand was analyzed, and the airport's energy data was processed. Because of the complexity in the data structure of airport's energy, poor accuracy and long statistical cycle in manual recording which would lead an interference existing in demand forecast. The SQL database was adopted to establish the backstage database of the prediction model of the airport's energy demand, moreover, airport's energy data was arranged and storage in a unified way. The results showed that the time and cost was greatly saved, the data format was standard, consistent and rational.Secondly, better generalization ability and anti noise ability were analyzed then discussed when using fuzzy support vector regression algorithm in small sample regression. Since the airport's data has the outliers, the algorithm was improved, and the selecting method of membership function also the parameters of forecasting model were optimized. Then, a forecasting model of airport's energy demand was established later verified. The results showed that: comparing with BP neural network method, the prediction accuracy of FSVR was improved by 2.66%, and outlier identification rate increased by 3.72%.Finally, a comprehensive management platform for airport was built by using this forecast model. Data interaction among the database, the prediction model and Matlab was realized. The results showed that the method acting as a forecasting method of airport's energy demand is feasible, and it can improve the analysis efficiency of system in real work effectively, also provide a better result for system's energy analysis and optimization.