基于果蝇算法优化BP神经网络的齿轮箱故障诊断

A Gear Box Fault Diagnosis Method Based on Fruit Fly Optimization Algorithm to Optimize the BP Neural Network

为有效提高风电机组齿轮箱故障诊断的快速性和准确性,采用近几年出现的果蝇算法对BP神经网络进行优化,减少了BP神经网络算法陷入局部最优解的风险,显著增强了BP神经网络的泛化能力和全局寻优能力。对比发现,果蝇算法优化后的BP神经网络模型具有比较好的快速性和准确的诊断能力。测试结果表明,果蝇算法优化BP神经网络对风机齿轮箱故障诊断具有可行性和有效性。

ABSTRACT:To effectively improve the accuracy and speed of the wind turbine gearbox fault diagnosis, this paper adopts Fruit Fly Optimization Algorithm which has been developed in recent years to optimize the BP neural network,which reduces the risk of BP neural network algorithm trapped in local optimal solution, and thus significantly enhances the generalization ability of the BP network and global optimization ability. Through comparison,the author of this article finds that the optimized BP neural network model through the Fruit Fly Optimization algorithm has a better rapid and accurate diagnostic ability. The testing results show that the Fruit Fly Optimization Algorithm for optimizing the BP neural net-work is feasible and effective for the gear box fault diagnosis.