Abstract: When de-icing robot performs the de-icing task，it must across all kinds of obstacles.In this paper，using Kalman filter arithmetic，the Fuzzy Neural Network（FNN） controller is used for de-icing robot’s manipulator tracking controller when it crosses obstacles，solves the problem of BP algorithm.By simulation，the control method can achieve a better function for robot arm tracking control when de-icing robot crosses obstacles.Results show the verification of the control strategy and the feasibility of the theoretical analysis.

Key words: power transmission line, de-icing robot, Fuzzy Neural Network（FNN）, adaptive merits

摘要： 覆冰机器人除冰时要跨越各种障碍物。采用卡尔曼滤波学习算法，将自适应模糊神经网络控制器用于覆冰机器人越障时的机械臂轨迹跟踪控制，解决了BP算法实时性差的问题。经过仿真实验论证，该方法对覆冰机器人越障时的机械臂轨迹跟踪控制具有很好的效果，表明控制策略和理论分析的可行性。

关键词: 输电线路, 覆冰机器人, 模糊神经网络, 自适应性