Data Regression Prediction Based on PSO-CR-GWO Hybrid Optimized BP Neural Network

In this paper, a hybrid optimization back propagation (BP) neural network based on Particle Swarm Optimization (PSO) and Chaotic Random Grey Wolf Optimization (CR-GWO) is proposed and regression prediction is performed on the data. Aiming at the shortcomings of the swarm optimization algorithm such as low global search ability, poor local search accuracy and long search time, this paper develops a new hybrid optimization swarm optimization algorithm by integrating the population position update strategy in PSO algorithm and the population initialization strategy and position weight factor update strategy in CR-GWO algorithm. Meanwhile, the developed new swarm optimization algorithm is applied to data regression prediction. The data comparison results show that the accuracy of PSO-CR-GWO is improved by about 70.085% compared with GWO data regression processing; PSO-CR-GWO is improved by about 83.805% compared with CR-GWO, and PSO-CR-GWO is improved by about 54.568% compared with PSO-GWO.