第51卷2023年第2期编辑严瑾采·掘1基于RBF神经网络凿岩台车钻臂逆解分析姜天优，杨聚辉，邢亚伟，陈锐，杨浩中铁工程装备集团有限公司河南郑州450016摘要：凿岩台车钻臂运动学逆解是钻臂轨迹规划的关键，采用人工神经网络是实现钻臂运动学逆解的有效途径。通过D-H法对钻臂进行正运动学分析，建立各级连杆坐标系，求解钻头位置相对于钻臂底座的位姿关系矩阵，基于蒙特卡洛法利用MATLAB计算出钻臂工作空间及掌子面覆盖范围，采用RBF神经网络逼近钻臂姿态与钻头位置的复杂映射关系，利用梯度下降法计算构建神经网络参数，完成钻臂逆解。结果表明：RBF神经网络计算转动关节角度最大误差为0.082°，误差均方根为0.0072，最大误差占比为0.91%；伸缩关节位移最大误差为1.07mm，误差均方根为0.083，最大误差占比为0.042%，能够较为精准地计算凿岩台车钻臂逆解，为规划钻臂轨迹实现自动化与智能化凿岩提供理论基础。关键词：凿岩台车；钻臂；正运动学分析；RBF神经网络；运动学逆解中图分类号：TD421.2+4文献标志码：A文章编号：1001-3954(2023)02-0001-06AnalysisoninversekinematicssolutionofdrillingarmofrockdrillingjumbobasedonRBFneuralnetworkJIANGTianyou,YANGJuhui,XINGYawei,CHENRui,YANGHaoChinaRailwayEngineeringEquipmentGroupCo.,Ltd.,Zhengzhou450016,Henan,ChinaAbstract：Theinversekinematicssolutionofthedrillingarmoftherockdrillingjumboisthekeytoplanthetrajectoryofthedrillingarm,andtheartificialneuralnetworkisaneffectivewaytorealizetheinversekinematicssolutionofthedrillingarm.D-Hmethodwasappliedtoconducttheforwardkinematicsanalysisonthedrillingarm,andthecoordinatesystemoflinkageatvariouslevelswasestablishedtosolvethematrixindicatingtheposturerelationshipofthedrillingbitandthepedestalofthedrillingarm.Inaddition,theworkspaceofthedrillingarmandthecoveragescopeofheadingfacewerecalculatedbyMATLBbasedonMonteCarlomethod,andRBFneuralnetworkwasappliedtofitthecomplexmappingrelationshipbetweenthepostureofthedrillingarmandthepositionofthedrillingbit.Afterthat,thegradientdescentmethodwasappliedtocalculatetheparametersofconstructingtheneuralnetwork,soastocompletetheinversekinematicssolutionofthedrillingarm.Resultsshowed:themaximumerrorofrotationaljointanglepredicte...
