基于改进粒子群的BP神经网络WSN数据融合算法

针对无线传感器网络（WSN）数据融合算法中传统反向传播（BP）神经网络收敛速度慢、对初值敏感和易陷入局部最优解的问题，提出基于改进粒子群的BP神经网络WSN数据融合算法（BSO-BP）。用天牛须搜索（BAS）算法对粒子群算法进行改进，利用改进后的粒子群算法优化BP神经网络权值和阈值，引入WSN数据融合中，簇首节点通过优化训练后的BP神经网络对采集数据进行特征提取，将融合后的数据发送至汇聚节点。仿真实验表明，BSO-BP算法能有效地提高融合精度和收敛速度，减少冗余数据传输，延长网络生命周期。相较于传统BP数据融合算法和PSO-BP算法，BSO-BP算法减少了至少11%的平均相对误差和13.89%的均方根误差。

Information fusion algorithm based on improved particle swarm BP neural network in WSN

Back-propagation (BP) neural network has low convergence speed, is sensitive to the initial value, and easily falls into the local optimal solution in data fusion algorithms in wireless sensor network (WSN). To solve these problems, a data fusion algorithm based on improved particle swarm optimization BP neural network in WSN (BSO-BP) is proposed. The beetle antennae search(BAS) algorithm is used to improve the particle swarm optimization. Then the imporved particle swarm optimization is used to optimize the BP neural network weights and thresholds, which are applied to WSN data fusion. The cluster head nodes extract the feature of the collected data by optimizing the trained BP neural network, and send the merged data to the sink node. Simulation results show that BSO-BP algorithm effectively improves the fusion accuracy and convergence speed, decreases the redundant data communication and prolongs network lifetime. BSO-BP algorithm reduces the relative error by 12.4% and the root-mean-square error by 11%, compared to BP and PSO-BP algorithms.