Forecast Model of V-SVR Based on an Improved GA-PSO Hybrid Algorithm

This paper firstly puts forward a new improved GA-PSO algorithm, which will solve the local and global contradictions on optimization better and can ensure the diversity, simplicity and efficiency of the population of particles at the same time. Then we embed it into an improved support vector machine (V-SVR) forecasting model, with parameters adaptive and different type of sample input feasible. In the end, this paper use matlab09a and the data of GDP, GZII and EN for model training and forecast simulation, and make comparison of results with RBF, PSO-V-SVR and GA-V-SVR model. It shows that the improved GA-PSO based on V-SVR model has the most powerful forecast capability.