A WSN Data Fusion Algorithm Based on Artificial Fish Swarm Algorithms Optimizing Neural Networks

Abstract

Wireless sensor network (WSN) usually has many nodes and high data redundancy. The traditional BP neural network (BPNN) data fusion method based on random weights and thresholds falling into local extremum is easy, which causing poor accuracy of fusion results. A new method is presented to optimize the weights and thresholds of neural networks and improve the quality of WSN data fusion - artificial fish swarm (AFS) algorithm back propagation (AFSABP) neural network data fusion. The results of simulation and comparison show that the speed of convergence and optimization accuracy of the improved fish swarm algorithm is obviously improved. Compared with the traditional BP data fusion method, the improved Artificial Fish Swarm BP (AFSBP) data fusion method can reduce the relative error by 3.06% and the root mean square error by 3.74%.