一种神经网络方法在机械手控制中的应用

采用多层前向神经网络建立机械手逆运动学模型。提出了一种改进遗传算法来学习网络的权系数，其交叉概率根据解的适应度来自适应调整，变异概率根据迭代次数来动态调整。这样可以有效地克服传统的反向传播算法求解精度低、搜索速度慢、易陷于局部极小的缺点。仿真结果表明，所提方法大大提高了机械手逆运动学解的精度，确保快速达到全局收敛。

A neural network method applying to robot manipulator control

The multilayer forward neural networks are used to establish the inverse kinematic models for robot manipulator. An improved genetic algorithm is presented to update the weights of the networks. In the improved genetic algorithm, the crossover probability is adapted by the fitness values of the solutions and the mutation probability is adjusted by the iteration times. The motivation of this approach is to overcome the shortcomings of traditional back propagation algorithm, such as the low precision of the solutions, the slow search speed and easy convergence to the local minimum points. Simulations show that the proposed method improves considerably the inverse kinematic solutions for robot manipulator and guarantees a rapid global convergence.