Research On Vibration Signal Processing Method For Dynamic Balancing Of Micro-motor Armature

Due to errors and uncertainties in the manufacture,assembly and machining of micro-motor armatures,the armature inertia spindle can deviate from the actual axis of rotation,which creates a dynamic imbalance during operation.This causes the armature to generate vibration and noise during operation,which affects the operating efficiency,service life and safety performance of the whole machine.By extracting the armature unbalance signal,the armature unbalance can be reduced by using dynamic balancing technique.The accuracy of the micro-motor armature requires high precision,and the existing methods have large errors in the extraction accuracy.The accurate extraction of micro-motor armature unbalance signal has an important significance for micro-motor armature dynamic balancing.Therefore,this paper investigates the techniques to accurately extract the unbalance signal from the vibration signal of micro-motor armature.The methods which can accurately extract the unbalance signal of micro-motor armature is proposed to improve the accuracy of micro-motor armature dynamic balancing.The main work is as follows:The basic theory of dynamic balancing is researched,and the two-sided influence coefficient method is analyzed to lay the foundation for subsequent dynamic balancing experiments.The simulated signal is used to simulate and analyze three existing unbalance signal extraction methods,namely DFT,mutual correlation and least squares.The unbalance signal extracted by DFT has a large loss in amplitude.The unbalanced signal extracted by the mutual correlation method has a phase shift.The least-squares method has a better extraction effect compared with the first two methods,but there are still errors in the amplitude and more than 10° deviation in the phase.Aiming at the problem that the micro-motor armature industrial frequency and the dynamic balancing system support frame inherent frequency are similar.A method of extracting the unbalance signal of micro-motor armature based on variational modal decomposition(VMD)is proposed.With the sample entropy convergence as the goal,the parameters K and α of VMD are optimized by SSA.And the threshold value of the reference alignment entropy of the unbalanced signal is used as a filter to select the desired modal components.After two VMD decompositions and reconstructing the effective signals,the unbalanced signal features can be extracted accurately.Simulation experiments are performed with 100 groups of simulated signals and the results are compared with the least squares extraction method.The proposed method has higher extraction accuracy for amplitude and phase,with an average amplitude difference of 0.72μm and an average phase difference of 2.5°.Ten micro-motor armatures are selected and the feasibility of the method is verified by experimental studies.The extracted armature unbalance signals were used for rotor dynamic balancing.The experimental results show that the residual unbalance of the armature is smaller after the dynamic balancing by the method,which proves the accuracy of the quadratic SSA-VMD method in extracting the unbalance signal of the micromotor rotor.The LSTM-ZPF method for micro-motor rotor unbalance signal extraction is proposed based on long and short time memory neural network(LSTM)and zero phase filtering(ZPF).Based on the adaptability of LSTM in timing prediction,a multilayer LSTM neural network structure is constructed to extract the unbalance signal using LSTM units.ZPF is also introduced to correct the phase of the extracted unbalance signal.5000 groups of simulated signals are constructed to train and test the proposed method,and the phase of the obtained unbalanced signal has almost no offset,and the average absolute error of the phase is 2.4°.Compared with the ideal frequency spectrum curve,the amplitude attenuation obtained with this method is smaller,and the coefficient of determination of the amplitude is 0.9999.Different signal-to-noise ratios were set to research the noise-resistance performance of the method,and it was concluded that the method can stably extract the unbalanced signal in different noises.The method was used in the experiments to perform dynamic balancing tests on different models of micro-motor armatures.The results show that the method has high accuracy in extracting unbalance signals for different types of micro-motor armatures,and the remaining unbalance after balancing is very small.The method also takes into account the fidelity and denoising of the unbalance signal,and solves the problem of amplitude loss when extracting the phase accurately,and achieves the accurate extraction of the unbalance signal.