基于SVD-Unscented卡尔曼滤波的掘进机行进调度纠偏研究

针对井下悬臂式掘进机自主行进纠偏的实际需求，基于机身调度相对于规划路径的位姿偏差模型，设计并简化了机身调度位姿跟踪控制律，通过构建Lyapunov函数证明该控制律作用下位姿偏差的收敛性。以履带打滑率表征机身行进调度的主要执行误差，以位姿检测误差表征系统的主要观测误差，通过试验及验算获得误差的统计规律，并利用SVD-Unscented卡尔曼滤波估计修正控制指令，从而降低行进调度过程中误差带来的影响。仿真结果表明，提出的轨迹跟踪控制策略能高效实现机身在有限的调整周期内向目标轨迹调度，履带驱动轮参考转速数值连续变化稳定，位姿偏差收敛性良好，达到了纠偏的目的; 基于SVD-Unscented卡尔曼滤波修正控制指令有效削弱了机身调度中来自过程误差和观测误差的影响。调度过程简单高效，跟踪效果重复性好，能为井下有限空间内的掘进机自动纠偏提供参考。

Path tracking of mining boom road-header using SVD-Unscented Kalman Filtering

Facing the demand of autonomous path correction of the road-header working underground, a control law was designed and simplified based on the position and orientation deviation model of the road-header against its aimed walking trajectory.Reasonable Lyapunov function is constructed to prove the convergence of the position and orientation deviation that resulted from the designed control law.Based on a certain real time position and orientation detection strategy for the road-header, the possible errors exist during posture adjustments are summarized and classified into two parts: one is the main execution error of the road-header and it is presented by the slip rate of tracks; the other is the main observation error and it is identified as the position and orientation measurement errors.This paper proposed to take advantages of the statistical characteristics of these errors, which is obtained by a certain number of experiments or deductions instead of rigorous theoretical analysis.The SVD-unscented Kalman Filtering is used to modify the commands that are given originally by the controller so as to deduce and reduce the influence of process error and observation error in the process of position and orientation adjusting.The simulation results show that the proposed path tracking control strategy can effectively realize the transition of the road-header to the target roadway in limited adjustment period, and the corresponding recommended speeds of the driving wheels vary continuously and stably in the whole process.The negative influence of process error and observation error is effectively weakened by commands modification based on SVD-unscented Kalman Filtering.The obtained moving trajectory of the road-header coincides with the planned path very well during the adjustment, which is of great repeatability.This proves that the proposed strategy is of great potential to be applied experimentally in underground real road heading scenario.