Coal Mining Machine Localization Method Based on Non-Gaussian Summation Parallel Kalman Filter Group

Abstract

Coal mining machine positioning technology is the key to realizing unmanned and intelligent mining of the comprehensive mining zone. Based on the traditional Strapdown Inertial Navigation System combined with Kalman-filtering coal mining machine positioning technology, non-integrity constraints are introduced, and the error of the output of the above system is filtered by an optimized Kalman filtering method proposed in this paper: non-Gaussian summation and a parallel Kalman filter bank. This method decomposes the non-Gaussian system into a linear combination of multiple Gaussian systems through the parallel Kalman filter group, then fuses the states occupying different weight coefficients and designs a method of Gaussian-term number trimming to solve the problem of parameter explosion in the filtering process, and ultimately obtains the optimal estimation of the positioning information of the coal mining machine. Experiments show that, for the coal mining machine positioning issue in the complex noise interference environment of intelligent mines, the non-Gaussian summation and parallel Kalman filter group method in this paper, compared with the traditional particle filtering method, greatly reduces the three-dimensional attitude error, three-dimensional velocity error, three-dimensional position error in the nine dimensional parameters of the estimation error, and the average estimation error. The average estimation error is reduced by 49%, 52%, 50%, 53%, 51%, 48.8%, 50.1%, 54%, and 51.3%, respectively, which significantly improves the positioning accuracy of coal mining machines, and has stronger real-time performance, stability, and accuracy in the coal mining machine positioning system.