关键词: 井下定位, 容积卡尔曼滤波, BP神经网络, LANDMARC, 智能矿山

Abstract: In order to solve the problem of low accuracy and large fluctuation of positioning results caused by the traditional LANDMARC algorithm due to the complex environmental environment of the mine,an improved LANDMARC underground positioning algorithm based on adaptive CKF is proposed. Firstly,the algorithm combines the traditional LANDMARC positioning algorithm to establish a downhole three-dimensional spatial model and solve the target location state estimation. Secondly,using the generalization mapping ability of BP neural network,neuronal parameters are introduced to optimize the volumetric Kalman filter (CKF),which fully combines the characteristics of BP neural network iterative learning and volumetric Kalman filter (CKF) to maintain stability in a strong nonlinear system,and improve the adaptive ability of the positioning algorithm. Finally,the location state estimation is used as an observation measurement for adaptive CKF filtering processing,and the optimized result is used as the true value output of the target location coordinates,which improves the accuracy of downhole positioning. The experimental results show that the introduction of adaptive CKF for filtering can greatly improve the stability of the traditional LANDMARC positioning algorithm,and the distribution of localization deviation is more concentrated,and the deviation below 1 m accounts for more than 90%. The positioning deviation of the proposed algorithm reaches 60% for labels below 0. 612 m,which can meet the high stability requirements of the complex dynamic environment of downhole,and has better applicability to downhole positioning compared with the traditional LANDMARC positioning algorithm and the LANDMARC positioning algorithm filtered by HIF.

Key words: downhole positioning,volumetric Kalman filtering,BP neural networks,LANDMARC,intelligent mine