Design Of A Novel Global Adjustable Magnetic Field Motor And Research Of Perception Methods For States

At present,the traditional permanent magnet motor has problems in that the permanent magnet magnetic field is difficult to adjust flexibly in real time and the speed regulation range is limited,and so on.Such drawbacks cannot fully meet the needs of constant voltage power generation and wide range of variable speed drive.By changing the air gap flux density of the motor,the adjustable flux motor can realize the coupling adjustment of machine,electricity,and magnetic.This improves the range and operation efficiency of the traditional permanent magnet motor to a certain extent.However,there are still problems in that the range of magnetic field adjustment precision cannot realize accurate automatic control,and the parameters of the motor body cannot be adjusted controllable.In this dissertation,the concept of "intelligent motor" is introduced into the design of an "adjustable flux motor," and a new type of Global adjustable magnetic field motor(GAMFM)is proposed to solve the problem that the magnetic field of adjustable flux motor cannot be adjusted accurately in all fields.Rotor structure parameters can not be changed quickly and flexibly;they have low intelligence and no self-sensing and self-regulation.The rotor structure of the motor contains four rotating permanent magnetic poles,an additional rotating magnetic pole control device,and a sensing device,which is convenient for realizing multiple rotating permanent magnetic poles simultaneously or separately.The motor can adjust its parameters frequently and quickly.Since the motor rotor is a rotating body structure with nested permanent magnet rotating mechanisms,its key structural component bearing plays an essential role in the stable operation of each motor mechanism.Therefore,studying the motor’s state perception and fault diagnosis and critical structural components,such as bearings,is necessary.This dissertation focuses on several fundamental problems of the main pole field adjustable motor.The main work includes the following parts.The topological structure of a new motor with a global adjustable main pole magnetic field is proposed,the main features of the whole motor and key structural components are expounded,and its assembly and connection mode are explained.In addition,the stator,rotor,permanent magnet,air gap,and rotor magnetic circuit of the motor are designed,and the torque ripple problem at low speed is optimized.Combined with the motor’s structural characteristics and functional requirements,the motor rotor control system is designed,which mainly includes a rotor self-generation device,main controller,DC servo motor,wireless communication module,magnetic pole rotation angle sensor,and so on.The simulation model of the motor is established and the working mechanism of the motor is explained.The electromagnetic field distribution,air gap magnetic density,no-load back electromotive force,and magnetic adjustment range at different magnetic adjustment working points are studied and analyzed.It is verified that the proposed motor plays an important role in solving the problems of narrow magnetic field adjustment range,low precision,no self-sensing,and low intelligence of permanent magnet motors.Simulation and experimental analysis proved that the motor has a better effect on weak magnetic speed regulation.The above results verify the rationality and feasibility of the motor design.The rotor structure of this motor is different from that of conventional motors.and there is some uncertainty in using conventional perception and feature extraction methods.To satisfy the reliable and stable operation of the motor in the running process,and to improve the perception accuracy and fault diagnosis accuracy of the motor running state,a fault diagnosis method based on the attention mechanism(SE)time convolution network(TCN)model is proposed.The original sensor signal is converted into a two-dimensional time-frequency image,and the characteristic information in the time domain and frequency domain is preserved.Then,the model framework is adopted,and the public data sets are first used in SE-TCN,TCN,and 1DCNN models to investigate the performance of perception and diagnosis of the proposed model.Then,the test data set of the experimental prototype is used to experiment on the SE-TCN model,which proves that SE-TCN can extract fault features from shorter vibration sequences for motor fault diagnosis and detection,and verifies the effectiveness of the real-time motor fault diagnosis method based on SE-TCN.To further meet the demand for fast real-time fault diagnosis for bearings and other parts of motor key structures,a new intelligent fault diagnosis method based on an improved time convolution network(TCN)and efficient channel attention(ECA)is proposed.By selecting the appropriate Dropout ratio and the coefficient of activation function,the model parameters are further determined,and the feature extraction efficiency of the model is improved.Aiming at the problem of weak generalization ability and poor stability of real-time fault diagnosis for motors in the state of changing working conditions,an improved fault diagnosis method of TCN joining the ECA attention module is proposed.The addition of the ECA module compensates for the shortcomings of inefficiency of feature extraction by the TCN null convolution,and the addition of the ECA makes the model autonomously pay attention to the feature channel related to the recognition target.By comparing the performance of the three models in a variable operating environment,ECA-TCN has the highest average accuracy of 91.32% with a short sample size.This shows that ECA-TCN has a strong generalization ability under variable operating conditions environment.In addition,through the experimental verification of rotor faults with different signal-to-noise ratios,the proposed model can still achieve good diagnostic results in noisy environments with high stability.During the operation of this type of motor,the rotor of the motor contains multiple rotating bodies running simultaneously or independently,which can adjust its parameter changes frequently and quickly,and its key structural component bearing plays an important role in the stable operation of each mechanism of the motor.To ensure and monitor the running state of the motor,a bearing fault diagnosis and classification method based on Gram Angular Field(GAF)and small-scale convolutional capsule network(Caps Net)is proposed for the most common bearing faults in key rotating mechanisms.The collected vibration signals are coded by GAF,and the corresponding signal characteristic map is generated.The characteristic map is input to the improved capsule network for training,and the GAF-Caps Net model is constructed to diagnose the motor bearing fault.The model was used to conduct experiments on the bearing fault test data set of this motor,and the results show that the model’s fault identification accuracy in the bearing fault diagnosis of this motor under mixed operating conditions reaches 94.15%,and the performance of the model generally has higher accuracy than other models compared with other coding methods and convolutional neural networks.Compared with one-dimensional convolution and other networks,the proposed model has stronger anti-noise performance.Finally,the experimental prototype is designed and manufactured,the prototype experimental platform is built,and the prototype performance is tested and analyzed.The no-load counter potential and speed range of the prototype are tested to verify the correctness of the motor magnetizing mechanism and simulation analysis.It is proved that the motor can have a wide range of magnetic adjustment and speed regulation by changing the rotation angle of magnetic poles.Through the experimental analysis and comparison of intelligent sensing and fault of the motor,the rationality of the proposed new motor design is verified,and the proposed intelligent sensing system and method have good effects in real-time sensing and fault diagnosis of the motor.