The Research On Rotor Rubbing And Fault Diagnosis Technology

As the core component of rotating machinery,the rotor system studies the nonlinear behavior and dynamic characteristics under fault conditions,which has far-reaching significance for learning and adjusting the fault of the rotor system.Based on several nonlinear forces such as rubbing force and nonlinear oil film force,a dynamic model of single-span double-disc rotor is established.The dynamic response of the parameter change to the system is analyzed by numerical simulation.The experimental research was carried out by constructing the rotor test bench,and the results were consistent with the simulation results.The accuracy of the model was verified,which provided an important reference for system design and engineering practice.In this paper,the rotor vibration signal is taken as the research object,and a rotor fault diagnosis method based on the combination of variational mode decomposition and optimized support vector machine model is proposed.The diagnosis of rotor rubbing fault is to collect vibration signals,feature extraction of faults and pattern recognition and classification.The key point is to extract sensitive fault features from nonlinear,non-stationary fault vibration signals.Therefore,this paper chooses a new adaptive signal decomposition method-variational mode decomposition(VMD)to extract the features of fault signals.The basic principles and steps of the VMD algorithm are introduced in detail.The acquired data is VMD decomposed and compared with the results of the local mean decomposition(LMD)method.The results show that the variational mode decomposition method is better than the local mean decomposition and VMD The method can effectively and accurately extract the fault features and is suitable for fault diagnosis of the rotor system.As the discriminant basis of rotor rub-impact fault,the method of particle swarm optimization is used to optimize the support vector machine model.By selecting the parameters in the particle swarm optimization algorithm,the optimal penalty factor in the support vector machine model is found.C and kernel function parameters.The eigenmode signals obtained by the variational mode decomposition are reconstructed,the sample set is obtained some of them are selected as the training set,the classifier model is trained,and the rest is used as the test set and input into the optimized classifier.Calculate the accuracy of this model.The result proves that the particle swarm optimization based support vector machine parameter method not only accelerates the optimization process of penalty factor C and kernel function parameters,but also has high diagnostic accuracy.Therefore,the particle swarm optimization support vector machine parameter method is suitable for pattern recognition of rotor rubbing faults,and has certain effectiveness and practicability.