Time series prediction using RBF neural networks with a nonlinear time-varying evolution PSO algorithm

The time series prediction of a practical power system is investigated in this paper. The radial basis function neural network (RBFNN) with a nonlinear time-varying evolution particle swarm optimization (NTVE-PSO) algorithm is developed. When training RBFNNs, the NTVE-PSO method is adopted to determine the optimal structure of the RBFNN to predict time series, in which the NTVE-PSO algorithm is a dynamically adaptive optimization approach using the nonlinear time-varying evolutionary functions for adjusting inertia and acceleration coefficients. The proposed PSO method will expedite convergence toward the global optimum during the iterations. To compare the performance of the proposed NTVE-PSO method with existing PSO methods, the different practical load types of Taiwan power system (Taipower) are utilized for time series prediction of one-day ahead and five-days ahead. Simulation results illustrate that the proposed NTVE-PSO-RBFNN has better forecasting accuracy and computational efficiency for different electricity demands than the other PSO-RBFNNs.