Wind power gearbox fault diagnosis of BP neural network based on improved Sparrow algorithm

It is difficult to identify the faults on the surface and root of the fan gearbox, resulting in low diagnostic accuracy. As an important part of the wind turbine, the abnormal vibration of the gear is easy to cause great losses, in order to improve the specific fault identification rate and reduce the fault. The following aspects are used to improve the sparrow algorithm. First, the homogeneous distribution Tent chaotic model is introduced to initialize the population distribution to improve the uniformity of the population distribution, then the Gaussian variation is used to add the random distribution to the newly generated individuals, and finally the adaptive weight is used to determine the value of the adjustment weight from the adaptation according to the population search state. The optimized sparrow algorithm can avoid falling into the local optimal solution and improve the convergence speed. By using the final conclusion of the improved sparrow algorithm, the accuracy of fault detection can be improved.