Research On Wind Turbine Gear Fault Diagnosis Based On Multi-sensor Fusion And Feature Enhancement

Wind power is one of the most promising power generation methods in the field of renewable energy,and planetary gearboxes,as an important part of wind power generation system,assume the role of power transmission and load carrying.However,because planetary gearboxes usually work in complex and harsh environments,leading to frequent failures of key components,once they fail,they will face extremely high maintenance difficulties and high repair costs,causing huge economic losses and bad social impacts.Therefore,early fault diagnosis of planetary gearboxes is of great significance to maintain normal operation of the equipment.In the process of multi-sensor and long-distance fault monitoring of planetary gearboxes,the acquired gear signals need to be compressed,transmitted,reconstructed,feature extraction and fault identification,and the process has the problems of strong signal noise interference,high data storage and transmission pressure,and low accuracy of feature identification.For the above problems this paper proposes a gear fault diagnosis method based on multi-sensor fusion technology and feature enhancement,and the main research contents of this paper are as follows.First,for the problem of strong signal noise interference,a generalized multi-sensor weighted fusion algorithm is proposed,and the concept of deviation factor is introduced.The deviation factor can make adaptive changes according to the deviation of the current value,and adjust the proportional relationship between the current value and the historical value based on this,so that the best estimated value is closer to the true value,and obtain a multi-sensor fusion method that can handle variable signals and reduce the noise component in the collected signal.The algorithm proposed in this paper is compared with the traditional random weighting algorithm and Kalman filter algorithm to verify the denoising effect of the algorithm.Second,to address the problem of high pressure of data storage and transmission,a high-precision signal reconstruction method with variable step length forward-backward matching tracking is proposed,which reconstructs the denoised signal,uses two fuzzy parameters to control the selection-in and rejection steps of atoms in the two-stage matching tracking algorithm,and performs an update operation on the observation matrix after each iteration to solve the backtracking over problem,and improve the reconstruction accuracy and speed of the algorithm under high compression ratio.Finally,to address the problem of low accuracy of feature recognition,an improved balanced binary algorithm is proposed to extract features from the reconstructed signal,which fuses the positive and negative sequences obtained from the one-dimensional threevalue pattern(1D-TP),solving the problem of loss of useful information caused by using only a single positive and negative sequence,and can effectively retain the useful information in the signal,saving time cost while obtaining a higher It can effectively retain the useful information in the signal and obtain higher fault classification accuracy while saving time and cost.