A parallel integrated learning technique of improved particle swarm optimization and BP neural network and its application

Swarm intelligence algorithm has attracted a lot of interest since its development, which has been proven to be effective in many application areas. In this study, an enhanced integrated learning technique of improved particle swarm optimization and BPNN (Back Propagation Neural Network) is proposed. First, the theory of good point sets is used to create a particle swarm with a uniform initial spatial distribution. So a good point set adaptive particle swarm optimization (GPSAPSO) algorithm was created by using a multi-population co-evolution approach and introducing a function that dynamically changes the inertia weights with the number of iterations. Sixteen benchmark functions were used to confirm the efficacy of the algorithm. Secondly, a parallel integrated approach combining the GPSAPSO algorithm and the BPNN was developed and utilized to build a water quality prediction model. Finally, four sets of cross-sectional data of the Huai River in Bengbu, Anhui Province, China, were used as simulation data for experiments. The experimental results show that the GPSAPSO-BPNN algorithm has obvious advantages compared with TTPSO-BPNN, NSABC-BPNN, IGSO-BPNN and CRBA-BPNN algorithms, which improves the accuracy of water quality prediction results and provides a scientific basis for water quality monitoring and management.