基于串级变结构神经网络的机器人补偿控制

针对机器人动力学的非线性、强耦合、时变特点，为了高速、高精度地控制机器人，提出了基于神经网络的机器人智能控制策略．讨论了机器人动力学模型的特殊性，指出模型中结构和非结构不确定性因素；研究了一种新型串级变结构神经网络（ＣＶＳＮＮ）；提出了一种基于新型ＣＶＳＮＮ的机器人控制策略．以精密Ｉ号机器人为仿真对象，证明该算法能够较好地补偿结构与非结构不确定因素，收敛速度快，能够自适应地调节自身的结构．

Manipulator Compute Torque Control Based on the Cascade Variable Structure Neural Network

In order to control the robot with high speed and high precision, an intelligent control law based on the neural network was proposed. The two kinds of uncertainty structure uncertainty and non structure uncertainty were summarized. According to the cascade variable structure neural network (CVSNN), a non linear compensation control strategy was proposed. This algorithm can compensate the above uncertainties well, converge quickly and adaptively adjust its structure. This approach was adopted to our Precision No.1 Robot System and the desired simulation result was obtained.