论文部分内容阅读
利用遗传算法优化的神经网络,对机器人臂的重力补偿进行研究。首先,根据力学基本知识和D-H参数建模法得出机器人臂各关节转矩的重力项理论计算公式;其次,在Solid Works仿真软件中,得到个别位姿下的重力项仿真值,并验证理论公式的正确性;最后,用遗传算法优化的神经网络对重力项进行预测。实验结果表明,采用该算法得到的重力项预测值和理论值基本一致,减少了运算量,提高了效率,为进一步实时控制提供了可能。
Using the neural network optimized by genetic algorithm, the gravity compensation of robot arm is studied. First of all, based on the basic knowledge of mechanics and the DH parameter modeling method, the formula of the gravity term of each joint torque of the robot arm is derived. Secondly, in the Solid Works simulation software, the simulation results of the gravity item under individual pose are obtained, The correctness of the formula; Finally, using the genetic algorithm optimized neural network to predict the gravity. The experimental results show that the predicted value of the gravitational item obtained by this algorithm is basically the same as the theoretical value, which reduces the computation load and improves the efficiency, which provides the possibility for further real-time control.