Research On Micro Fault Diagnosis Method Of Permanent Magnet Synchronous Motor

Permanent magnet synchronous motor is a kind of synchronous motor.Because of its advantages of simple structure,small size,light weight,large overload capacity and brushless,PMSM has been widely used in manufacturing systems,electric vehicles,wind power generation,ship propulsion systems and other industrial fields.And with microelectronics,intelligent control and other technology increasingly advanced and social attention to environmental protection issues,permanent magnet synchronous motor development prospects will be more and more broad.However,because the motor works in the complex and changeable industrial environment,it is affected by the power supply and load conditions.During the longterm continuous operation,the motor will inevitably have various faults.Machinery of the ship in fault analysis,the mechanism and evolution of any serious failure is the accumulation of from early minor faults,and with the accumulation of time,the small fault amplitude increases gradually,if it exceed the critical value,then the small fault can quickly increase or jump into a serious fault,have a serious impact to the system and fault maintenance is expensive.In order to identify these minor faults as soon as possible,eliminate potential risks for the motor and ensure the normal sailing of the ship,The main research contents of this paper are as follows:(1)The finite element simulation model of permanent magnet synchronous motor is built based on ANSYS Electronics software platform.Firstly,the dynamic and static simulation analysis of the simulation model is carried out.Then the simulation model is run under different typical working conditions,and the running data is analyzed to verify the reliability of the model and prove the reference value of the model.Then,the model is transformed to the fault model,and the parameter changes during the fault are compared with the normal working condition,or the different degree of parameter changes of the same fault is analyzed.Through the analysis and comparison,the influence of the fault on the permanent magnet synchronous motor is understood.(2)Based on the Matlab platform,this paper groups data sets of different fault types with certain data points as time Windows,uses time domain,frequency domain and time-frequency methods to extract features,and obtains(66,1300)highdimensional feature matrix,and then carries out feature dimension reduction.In terms of dimension reduction methods,this paper comprehensively compares the dimensionality reduction effects of PCA and PLS methods,and the results show that PLS method has better effect.PLS method is used for dimension reduction of fault data,which makes the difference between feature data in different fault conditions more obvious.(3)In this paper,based on the Matlab platform,after the dimension reduction feature matrix,after data preprocessing,data normalization,through BRB method to establish the fault identification model,the data were randomly divided into training set and test set,the training set the substitution model,through the optimization algorithm to optimize BRB reference value in the model,the reliability,weight and other parameters.After the training,the test set data is substituted into the model with trained parameters to verify the reliability of the model.In terms of optimization algorithm,this paper comprehensively compares the advantages and disadvantages of genetic algorithm,particle swarm algorithm and Matlab optimization algorithm,and believes that particle swarm algorithm combined with BRB model has the best comprehensive diagnosis effect.To sum up,this paper takes permanent magnet synchronous motor as the research object.The ANSYS Electronics is used as the simulation platform to build motor simulation model and fault model.And feature extraction and feature dimension reduction are carried out for the obtained data in turn.Finally,the BRB method is used for fault diagnosis.On this basis,the diagnosis effect of different optimization algorithms combined with BRB method is compared,and the final results show that the diagnosis effect of particle swarm optimization combined with BRB method is good.this paper has some reference value for the study of permanent magnet synchronous motor fault diagnosis.