Optimal Decomposition Method Based On Complex Differential Operators And Its Application In Fault Diagnosis Of Gearbox

Gear box is a key mechanical driving device.Once the failure of gearbox,machinery and equipment directly affect the operation of the situation,and even will bring tremendous economic losses and casualties.So,it is very important to study the fault condition monitoring and diagnosis of gearbox.The main steps include operation condition monitoring and dignosing feature extraction from vibration acceleration data of each rotating part,which needs proper signal prodcessing methods.Nowadays,the popular digital signal processing methods include Ensemble Empirical Mode Decomposition(EEMD),Adaptive and Sparsity Time Frequency Analysis(ASTFA),etc.Theses methods have been improved,but there are still many challenges.On the basis of ASTFA Method,this paper proposes an Optimal Decomposition Method Based on Complex Differential Operators(CDOOD),and proposes an improved method to address the shortcomings of CDOOD,then the improved CDOOD is combined with fast spectral correlation and Extreme Learning Machine(ELM)methods to realize the fault diagnosis of the gearbox.This paper mainly studies the theory of CDOOD and its application in the gearbox.The main research contents of this paper are as follows:1.The CDOOD method and its improvement are proposed.ASTFA is proposed on the basis of using gauss-newton iteration method for optimization and cannot be replaced by other methods.Gauss-newton iteration method is very sensitive to the initial value.If the initial value deviates too far from the real value,the accurate components cannot be decomposed.At the same time,since the component constraint in ASTFA is that the envelope function is smoother than the cosine function,the components of the two may overlap,which cannot guarantee that the obtained component and instantaneous frequency are physical significance.Aiming at the shortcomings of ASTFA,CDOOD method is proposed.The main idea of the CDOOD method is to transform the decomposition problem of mixed signals into a nonlinear optimization problem by optimizing filter parameters.The optimization objective is to minimize the energy of the decomposition residual,and the constraint condition is to make the single component meet the conditions of local narrowband signals.Finally,the optimization is to decompose the mixed signal into several intrinsic narrowband components.The optimization objective function of CDOOD method is improved,sothat the residual energy and orthogonal coefficient of the final components are smaller.Simulation results show that the improved CDOOD method is more effective.2.A fault diagnosis method which is based on improved CDOOD and fast spectral correlation is proposed.Spectral correlation is an ideal condition monitoring tool,which can clearly show the existence of modulation.The method is applied to the actual vibration signals of the gearbox,and the results show that the method is practical.3.A method of gearbox intelligent fault diagnosis which is based on extreme learning machine and improved CDOOD is proposed.Firstly,the fault characteristic signal of the gearbox is obtained by the improved CDOOD method,and then the enhanced envelope spectrum is obtained by the fast spectral correlation method.At last,the enhanced envelope spectrum is taken as the input of the extreme learning machine to realize the intelligent diagnosis of the gearbox.Compared with the method of BP,the vibration signals of the gearbox proves the effectiveness of this method.