针对齿轮故障诊断的重要性和复杂性,基于神经网络固有的局部极小值及遗传算法良好的全局收敛能力和搜索性,在遗传算法优化初始权值和阈值的基础上构建了GA-Elman神经网络的故障诊断系统。仿真结果表明,该方法用于齿轮常见故障诊断和预测有效、可行。 Aiming at the importance and complexity of gear fault diagnosis, based on the inherent local minimum of neural network and the good global convergence ability and searching ability of genetic algorithm, GA-Elman neural network is constructed on the basis of optimized initial weight and threshold of genetic algorithm Network fault diagnosis system. Simulation results show that this method is effective and feasible for common fault diagnosis and prediction of gears.