【摘要】 针对传统矿井下定位方法精度偏低问题，提出一种混沌粒子群优化(CPSO)Elman神经网络矿井下可见光定位系统。由于Elman神经网络在初始化时存在参数设置的随机性导致预测精度不高，采用CPSO算法优化Elman神经网络，选取适合的各层的初始权值和阈值，用于提高神经网络拓扑的稳定性。仿真结果表明：在3.6 m×3.6 m×3.6 m的环境里，本文所提的算法的平均定位误差达到3.70 cm,最大定位误差为26.54 cm,在实验阶段的平均定位误差为5.91 cm,最大定位误差为36.95 cm,能够满足煤矿井下定位需求。更多还原

【Abstract】 Aiming at problem of low precision of traditional underground mine positioning methods, a chaotic particle swarm optimization(CPSO)Elman neural network underground mine visible light positioning system is proposed.Due to the randomness of parameter setting during the initialization of Elman neural network, the prediction precision is not high.CPSO algorithm is used to optimize Elman neural network, and the appropriate initial weights and thresholds of each layer are selected to improve the stability of neural network topology.The simulation results show that in the environment of 3.6 m×3.6 m×3.6 m, the average positioning error of the proposed algorithm is 3.70 cm, and the maximum positioning error is 26.54 cm.In the experimental stage, the average positioning error is 5.91 cm, and the maximum positioning error is 36.95 cm, which can meet the positioning requirements of underground coal mine.更多还原