基于PSO-BP神经网络的煤矿井下自适应定位算法

摘要

提出了一种基于PSO-BP神经网络的煤矿井下自适应定位算法.针对传统的基于测距模型的定位算法易受煤矿井下环境干扰、测距误差大的问题,选择指纹匹配定位模型.针对煤矿井下环境强时变性,易增大实时采集的指纹信息与离线阶段建立的静态指纹数据库信息的匹配误差问题,将信标节点作为参考点的校准节点,以更好地反映参考点随环境变化的情况,避免增加额外的校准节点;在不增加硬件成本的同时,通过动态补偿法实时修正目标节点指纹数据,解决了指纹匹配定位模型自适应差的问题.匹配定位阶段采用PSO优化BP神经网络权值,以加速BP神经网络收敛,提高学习速度.实验结果表明,该算法更加适应随时间变化的煤矿井下环境,满足井下自适应定位要求.%A kind of underground adaptive positioning algorithm based on PSO-BP neural network was proposed.In view of the problem that traditional positioning algorithm based on ranging model is sensitive to coal mine environment disturbance and ranging error is large,a fingerprint matching positioning model is selected for positioning.In view of the problem that strong time-varying nature of coal mine environment is easy to increase matching error between the fingerprint information collected in real time and the static fingerprint database information established in offline phase,the beacon node is used as calibration node to better reflect the condition of reference point changes with environment,and avoid adding additional calibration nodes.The dynamic compensation method is used to correct the fingerprint data of the target node in real time without increasing the hardware cost,which solves the problem of poor adaptation of the fingerprint matching positioning model.At the matching positioning stage,PSO is used to optimize weight of BP neural network to accelerate convergence of BP neural network and improve learning speed.The experimental results show that the algorithm is more adapted to the coal mine environment varies with time,and meets the requirement of adaptive underground positioning.