Research On Accurate Positioning Technology For Moving Target In Complex Coal Mine Environment

With the development of positioning system and sensor technology,location service has become an indispensable basic data service for digital production and life.Many guiding documents indicate that coal mining has been gradually advanced from mechanization and automation to intelligent direction.The development goal of intelligent mine and unmanned mine has become an industry consensus.As the front line of resource mining,coal mine automation construction and digital mine construction urgently need the support of high-precision location services.However,under the special and complex environment such as coal mine roadway,the single positioning technology is severely limited,and the fusion of various sensors to improve the positioning accuracy is imperative.Therefore,this thesis takes sensor information fusion as the research goal,combines two different types of positioning technologies,UWB radio frequency positioning and IMU incremental positioning,fuses sensor positioning data of different dimensions through machine learning and innovation of traditional filtering algorithms,so as to suppress errors and improve positioning accuracy.The main work of this thesis includes the following aspects:(1)Positioning scheme design and positioning signal acquisition.First of all,in order to realize the synchronous acquisition of all kinds of positioning data,the data acquisition and transmission system is designed while implementing the basic algorithms of radio frequency positioning and incremental positioning respectively.Then,use a variety of devices in different scenarios to carry out real experiment verification and basic data collection.Finally,the data are cleaned and labeled to form an experimental data set.(2)Range correction method based on NLOS range jump detection.The RF positioning error starts from the ranging error,and the range fluctuation caused by the non-line-of-sight phenomenon is the most significant.Therefore,this method starts from the detection of the range positive jump when the non-line-of-sight occurs,and first uses the machine learning method to learn the jump characteristics and detect whether the non-line-of-sight range occurs.Then combined with IMU sensor data and historical ranging data,non-line-of-sight ranging is corrected.Finally,the correction of ranging information is completed through the non-line-of-sight end feature.By correcting the ranging information,the trilateral positioning error is reduced and the positioning accuracy is improved.(3)An adaptive Kalman filter fusion localization method based on motion inertia estimation.This method introduces the concepts of heading memory and inertial confidence parameters to participate in the data fusion of Kalman filter.First of all,this method uses the long linear characteristics of the underground roadway to mine and extract the target motion inertia law,and forms a comparison standard for the new time UWB and IMU positioning data.Then,with the help of data inversion and normalization operation,the coefficient pairs that can act on the Kalman filter error matrix are formed,and the two positioning fusion weights are adjusted by expanding or reducing the error matrix to achieve adaptive Kalman filter.(4)Design and implementation of underground positioning system.Based on the object-oriented design idea,adopting the hierarchical and modular design idea,combined with the research results of this project,the underground positioning system based on B/S architecture is realized.The system visualizes the target location information and includes all kinds of analysis data,providing a basic framework for further practical application and modular function addition.