CAJ | 학술논문

提出一种基于蚁群和微分进化优化BP神经网络的风电机组齿轮箱故障诊断方法。将蚁群算法的信息素更新机制用于微分进化算法当中，提高微分进化算法的收敛速度，并利用微分进化个体更新方式改善蚁群算法的早熟问题，利用AC-DE算法对BP神经网络的权值和阈值进行优化，改善了BP神经网络算法陷入局部最优解的缺点，提高了神经网络的训练效率和收敛速度。经测试该方法诊断结果正确且精度高，表明了AC-DE优化BP神经网络用于风电机组齿轮箱故障诊断的有效性。
제출일충기우의군화미분진화우화BP신경망락적풍전궤조치륜상고장진단방법。장의군산법적신식소경신궤제용우미분진화산법당중，제고미분진화산법적수렴속도，병이용미분진화개체경신방식개선의군산법적조숙문제，이용AC-DE산법대BP신경망락적권치화역치진행우화，개선료BP신경망락산법함입국부최우해적결점，제고료신경망락적훈련효솔화수렴속도。경측시해방법진단결과정학차정도고，표명료AC-DE우화BP신경망락용우풍전궤조치륜상고장진단적유효성。
A method based on BP neural networks trained by Ant Colony and Differential Evolution(AC-DE)algorithm is presented for fault diagnosis of wind turbine gearbox. The ant colony algorithm pheromone update mechanism for differ-ential evolution algorithm which improves the convergence speed of differential evolution algorithm and using differential evolution individual ways to improve the ant colony algorithm update premature problem, it can reduce the risk of BP neural network algorithm falling into local minimum, improve the training efficiency, and speed up convergence by using AC-DE algorithm to optimize the weights and bias of BP neural network. The new algorithm is applied to wind turbine gearbox fault diagnosis forecast, the method is tested and results of fault diagnosis are right. The validity and practicability of BP neural network algorithm trained by AC-DE algorithm for the wind turbine gearbox fault diagnosis are proved.