针对定性定量间转化的模糊性。并且为避免小波网络陷入局部极小,将云模型、遗传算法和小波神经网络相结合,不但得到了最优网络参数,还提升了模糊问题的辨识精度。通过对燃机发电机继电保护的故障诊断,仿真实验验证了这种新方法对机械电气保护故障诊断的有效性。
Aiming at the fuzziness between qualitative and quantitative transformation, to avoid the wavelet network trapped in local minimum, combining the cloud model, genetic algorithm with wavelet neural network could not only get the optimal network parameter, but also improve the identification accuracy of the fuzzy problems. Based on the fault diagnosis of the gas turbine generator relay protection, the simulation experiment shows that the new method is effective in the fault diagnosis of mechanical and electrical protection.