多 尺 度 复 合 稀 疏 的 齿 轮 箱 复 合 故 障 诊 断 研 究

In modern mechanical equipment transmission systems，bearings and gears are prone to local fatigue failures. Load fluctuations caused by typical failures of a single component are very likely to cause secondary fa⁃ tigue failures of other components，making the mechanical transmission system present a multi-component com⁃ posite failure state. Aiming at the problem of compound fault diagnosis of bearings and gears in gearbox transmis⁃ sion systems，this paper proposes a convolutional sparse coding（MCRCSC）separation diagnosis algorithm based on multi-scale convolution kernel matching compound regularization. First，the model assumptions are made based on the sparseness and scale characteristics of the typical composite faults of gearbox bearings and gears. Then the concepts of multi-scale convolution kernel and convolution sparse composite regularization are proposed in terms of the signal scale and distribution characteristics of different faults，and the multi-component convolution separation is accordingly established. In the model，the optimization equation after frequency do⁃ main conversion is decomposed into sub-problems by alternating direction multiplier（ADMM）optimization ar⁃ chitecture and alternately solved，and the corresponding fault frequency distribution is obtained by spectrum anal⁃ ysis of the fault signal after separation and convolution reconstruction. Simulation analysis based on model as⁃ sumptions and actual gearbox fault simulation experiments show that the proposed algorithm has excellent fault separation and diagnosis capabilities under random noise and harmonic interference.