Research On Fault Diagnosis Method Of Micro Motor Based On Current Signal

The micro motor has simple structure,low production cost,and simple use mode,and it has been widely used in engineering practice.With the development of automation technology,the output of micro motors increases year by year,so the quality inspection of its factory is particularly important.At present,motor manufacturers mainly rely on manual fault diagnosis for motors.The diagnosis accuracy rate is not optimistic.Therefore,it is necessary to propose an intelligent micro-motor fault diagnosis method,which can realize the online detection of the motor before leaving the factory and improve the accuracy of the fault diagnosis of the micro-motor.The main work done in this article is as follows:(1)This article starts with the current signal of the micro motor,and derives the armature current expression of the motor in a steady-state and no-load condition bases on its structural characteristics.When the motor fails,the current will change abnormally.On this basis,the time-domain and frequency domain features of the motor are analyzed.Using overall empirical mode decomposition,the sample entropy features and energy features of IMF components are obtained.All the features together construct a feature set.In order to select high-quality features,and use the Laplacian score for feature selection.Experiments show that after feature selection,the accuracy of fault diagnosis has been significantly improved.(2)After determining the feature parameters used,the feature set is put into BP neural network,SVM,extreme learning machine,and KELM for training.However,experiments are still needed to verify the machine learning algorithm with the highest diagnostic accuracy under 4 different classifiers.Under the condition that the original data and experimental conditions of different experiments are consistent,KELM has outstanding performance in the fault diagnosis of micro-motors after comparing and analyzing the experimental results.(3)In terms of algorithm optimization and effect of validation,a new type of intelligent bionic algorithm--bat algorithm,which is used to optimize KELM.It is quite suitable for optimizing the kernel parameters and penalty factors of KELM.Therefore,the new algorithm is called BA-KELM,and then train it in the data method classifier.The magnitude of the original current was changed and the signal was noisy.By comparing with other models: the accuracy rate does not change with the increasing of iteration number.