Demagnetization Fault Recognition And Classification Research Of Double Stator Air-core Permanent Magnet Synchronous Linear Motor

Double stator air-core permanent magnet synchronous linear motor(DSA-PMSLM)has the advanages of direct displacement method without intermediate conversion mechanism,high thrust density and high efficiency,which satisfy high speed large stroke motion and micro level dynamic accuracy.It has wide application prospect in laser cutting machine,metal 3D printers,chip lithography machine,and liquid crystal visual inspection such as precision fields.As the core part of DSA-PMSLM,the permanent magnet(PM)can provide a constant magnetic field environment for the air gap space in motor,and it can ensure the stability of the motor's power density and thrust density.However,due to the timeliness of PM materials,the processing process,and the factors such as chemistry,temperature and external magnetic field in the actual working environment of the motor.It is easy to cause irreversible demagnetization of the PM and result in motor demagnetization.Demagnetization fault of DSA-PMSLM will cause uneven distribution of air gap magnetic field,destroy the consistency of magnetic field inside the motor,and then cause the increase of linear motor thrust fluctuation,resulting in the reduction of positioning accuracy of the motor.In addition,demagnetization fault may cause the electrical fault and mechanical fault in motor,and even cause the motor to stop and damage.Serious motor demagnetization fault may cause the damage and scrap of high precision machine tool equipment.Therefore,in order to improve the stability and reliability of DSA-PMSLM,and the security of related equipment,the recognition and diagnosis of DSA-PMSLM demagnetization fault is particularly important.This paper takes a DSA-PMSLM applied in laser cutting machine as the research object,and demagnetization fault recognition and classification method is systematically studied for the industrial application scenarios of batch quality inspection,motor magnetic field consistency detection and regular product maintenance before the motor is delivered to the factory,so as to realize accurate recognition and classification of demagnetization fault.According to the characteristics of the topological structure of the double-permanent magnet stator motor,from the aspects as the establishment of the demagnetization electromagnetic calculation model of the motor,the modeling of the thermal demagnetization of the permanent magnet body,the acquisition of the effective signal of the demagnetization fault in motor,the extraction and enhancement of the fault features and the demagnetization fault identification classifier were studied.And the proposed method was simulated and verified by prototype experiments.It lays an important technical foundation for further improving the reliability of permanent magnet linear motor and promoting the development of high precision machining industry,and provides a new way for the demagnetization fault diagnosis of other motors.The main research results of this paper include the following aspects:1.A DSA-PMSLM motor demagnetization fault analysis and numerical electromagnetic calculation model was established to calculate and analyze the effects of different demagnetization faults on the performance of the motor in multiple physical fields.The main magnetic field parameters affecting the performance of the motor are analyzed qualitatively by the equivalent magnetization method.Based on double permanent magnet stator symmetry and the periodic array topology structure,to preset different PM demagnetization fault,and conduct simulation to analyze the air gap flux density distribution,no-load induction emf harmonic content and thrust performance in normal and fault statements,which can establish the model basis for subsequent demagnetization fault signal acquisition and feature extraction.2.A method for modeling the thermal demagnetization of NdFeB permanent magnet based on extreme learning machine is presented,and the high accuracy thermal demagnetization model of PM is established.The thermal demagnetization mechanism of NdFeB permanent magnet was deeply analyzed,the mapping relationship between temperature and demagnetization degree of permanent magnet was explored,and the extreme learning machine method was introduced to establish the thermal demagnetization regression calculation model of permanent magnet,which provided a basis for the subsequent acquisition of specific fault permanent magnets.The effectiveness of the method was verified by experiments.3.In this paper,a three line fault signal extraction method for magnetic density in air gap space is studied,and DSA-PMSLM demagnetization fault signal is obtained effectively.This paper compares and analyzes the possibility,advantages and disadvantages of current signal,voltage signal and vibration signal application in DSA-PMSLM demagnetization fault diagnosis.A three line magnetic density signal acquisition method is proposed,which is suitable for demagnetization fault recognition and classification of linear motor with double PM stator topological structure,which can lay foundation of signal source for fault feature extraction.The simulation and prototype experiments verify the effectiveness of tproposed method.4.A Teager-Hanning energy operator method is proposed to extract and enhance the fault features effectively.Different signal processing methods were studied to extract the characteristics of demagnetization fault.Compared with simulation experiments,the shortcomings of wavelet transform,EMD decomposition,TT transform and other methods in feature extraction of demagnetization fault were verified.A new method of Teager-Hanning energy operator is proposed to enhance the extraction of fault features,which provides a data source for the following research on demagnetization fault recognition and classification.5.The random forest machine learning classifier is established to realize accurate recognition of dsa-pmslm demagnetization fault.In-depth analysis of DSA-PMSLM demagnetization fault classification principle,the establishment of sample database.A variety of machine learning classifier methods,such as feature weighted k-nearest neighbor,decision tree and random forest,were studied,and a dsa-pmslm demagnetization fault classifier was designed.6.The fault permanent magnet was obtained by experiment and DSA-PMSLM prototype was made.The demagnetization fault detection experimental platform based on three-channel gaussimeter was built to provide the software and hardware platform for the verification.Simulation experiments and prototype experiments show that the method can identify the type of demagnetization fault of DSA-PMSLM,which can determine the specific location and severity of demagnetization fault,and the recognition accuracy is higher than 95%.It provides a new technical means for the mass off-line detection and regular maintenance in the production and manufacture of this kind of motor,and also provides a new idea for the demagnetization fault diagnosis of other kind of motor.