基于遗传算法优化神经网络的定位算法

为了减小非视距(Non Line of Sight,NLOS)误差对移动台定位精度的影响,提出一种基于遗传优化后向传播(Back Propagation,BP)神经网络的无线定位算法。先利用遗传算法优化BP神经网络的初始权值,再利用优化后的GA-BP神经网络修正到达时间差(Time Difference Of Arrival,TDOA)测量值,最后使用Chan氏算法确定移动台的位置,以避免由于神经网络初始权值的随机性所带来的网络震荡,克服网络容易陷入局部解的问题。仿真结果表明,新算法能够实现移动台的静态定位,并且性能优于传统BP神经网络与最小二乘(Least Square,LS)算法。

A location algorithm based on BP neural network optimized by genetic algorithm

In order to reduce the influence of NLOS errors on mobile positioning accuracy,a loca-tion algorithm based on BP neural network optimized by genetic algorithm is proposed.The ge-netic algorithm is able to correct the initial weights of BP neural network and then the NLOS er-rors can be corrected by the optimized GA-BP neural network.Furthermore,the positions of MS can be estimated by Chan ’s algorithm.This new algorithm overcomes the networks shock brought by the randomness of the initial weights and solves the problem of network solutions easy to fall into local issues.Simulation results show that the new algorithm can achieve static positio-ning of the mobile station and the performance is better than traditional BP neural network and LS algorithm.