Research On Fusion Positioning Method Of Drilling Robot For Rockburst Prevention Based On UWB And IMU

Rockburst is one of the most severe dynamic hazards in deep coal mining in our country.Currently,pressure relief operations mainly rely on manual intervention by workers entering hazardous areas,which can lead to significant accidents and injuries in the event of a rockburst.Therefore,there is an urgent need to automate drilling and pressure relief operations.The key technological challenge in robotic operations within the drilling and pressure relief area is the precise positioning of the drilling robot.However,due to the complex mine environment,single sensors such as vision and Li DAR have certain limitations in positioning technology and cannot fully meet the positioning requirements of the drilling robot for rockburst prevention.As a result,combined positioning has become the mainstream solution.In light of this,this thesis proposes a fusion positioning method for the drilling robot for rockburst prevention based on redundant Inertial Measurement Units(IMUs)and Ultra-wideband(UWB)technology.It utilizes the array layout of redundant IMUs and the positioning information from UWB to suppress the accumulation of errors in IMUs,thereby achieving precise positioning during the operation of the drilling robot for rockburst prevention.The main work and research achievements of this thesis are as follows:(1)This thesis analyzes and establishes the error models for the attitude,velocity,and position of the drilling robot for rockburst prevention.Addressing the deterministic drift and non-deterministic drift of IMUs,six layout models for redundant IMUs are proposed,and an algorithm for solving the redundant IMUs is designed.Through simulation analysis and motion experiments with a wheeled robot,it is verified that the tetrahedral layout model of redundant IMUs exhibits better error elimination capability,leading to higher positioning accuracy.(2)The main factors causing positioning errors in Ultra-wideband(UWB)technology are analyzed.A polynomial fitting model for ranging error is established,and the layout of UWB base stations is optimized and analyzed.To address the issue of initial value selection in the Taylor algorithm,a hybrid solution method combining Bacterial Foraging Optimization Algorithm(BFOA)and Taylor series expansion(BFOA-Taylor)is proposed.This method effectively reduces the impact of harsh environments on UWB performance.Through simulation analysis and motion experiments with a wheeled robot,the feasibility and superiority of the BFOA-Taylor solution method are verified.(3)The relevant definitions of Factor Graph(FG)and the associated characteristics between factor nodes and variable nodes are analyzed.The fusion principle of the factor graph model is studied.Based on this,a UWB credibility evaluation method based on BP Neural Network(BPNN)is proposed.By using the residual discrimination method and BP neural network,multiple model recognition and switching of UWB under different environmental conditions are achieved.Subsequently,a UWB/IMU data fusion positioning model based on BPNN-FG is established.Simulation results verify that the proposed fusion positioning method still exhibits superior performance in different line-of-sight environments.(4)The experimental platforms for motion experiments with a wheeled robot and a tracked drilling robot are designed and constructed.Motion experiments are conducted in both line-of-sight and non-line-of-sight environments.The results show that after fusing UWB and IMU data using the proposed method,the accuracy of positioning solutions is significantly improved.The maximum absolute errors in direction are 0.1857 m,0.1954 m,and 0.2424 m,respectively.This validates the practicality of the fusion positioning method for the drilling robot for rockburst prevention.The thesis includes 47 figures,23 tables,and references to 106 sources.