遗传算法和模糊神经网络在齿轮传动中的应用

摘要

基于遗传算法优化模糊神经网络齿轮传动机构优化的新模型，首先将各参数用二进制串表示，用适合度函数衡量算法的收敛状况。然后寻找最优模糊隶属函数参数，按适值选取最后一代群体中N个可能具有全局性的进化解，分别以该进化解为初始权值，用BP神经网络进行求解，比较N个由神经网络求得最优解，从而获得全局最优解。Matlab仿真结果表明所构造的识别模型预测误差非常小。%The paper proposes a new model of gear transmission mechanism based on genetic algorithm to optimize the fuzzy neural network. The various parameters are expressed by binary string, and theconvergence is evaluated using fitness function. The optimal fuzzy membership function parameters can be therefore found. According to fitness N-solutions are selected based on the overall evolution of the last generation in population. Then the evolutionary solution ia regarded as the initial weights, the final solution can be found by using BP neural network. By comparing these solutions from neural network, the global optimal solution is obtained. Matlab simulation results show that the constructed model has a very small prediction error.