A multi-signal fusion permanent magnet motor inter-turn short circuit fault diagnosis method

Abstract

The application discloses a kind of multi-signal fusion's permanent magnet motor turn-to-turn short-circuit fault diagnosis method, to solve the prior art from motor current to diagnose turn short, especially for early turn short diagnosis, the probability of false negative is big technical problem, belongs to motor turn-to-turn short-circuit fault diagnosis technical field.The application respectively real-time acquisition motor three-phase current, three-phase line voltage, bearing end cap load area radial vibration acceleration signal in the process of motor operation, specific method includes the following steps Step one, current fault feature extraction;Step two, vibration fault feature extraction;S21, vibration spectrum fault feature extraction;S22, vibration envelope spectrum fault feature extraction;S23, vibration fault feature flag calculation;Step three, turn short online feature flag calculation;Step four, judge whether to open counter electromotive force feature calculation;Step five, counter electromotive force fault feature extraction;Step six, final diagnosis of motor turn short.

Description

Technical Field

[0001] This invention belongs to the field of motor inter-turn short circuit fault diagnosis technology, and specifically relates to a multi-signal fusion method for diagnosing inter-turn short circuit faults in permanent magnet motors. Background Technology

[0002] Permanent magnet motors (PMMs) possess advantages such as high efficiency, high power density, wide speed range, and low noise, making them widely used in various industrial and civil equipment. Especially in the rail transportation industry, their green, energy-saving, and lightweight advantages have garnered significant attention in recent years. However, during operation, due to their unique structure and working principle, when a PMM experiences an inter-turn insulation fault, even after cutting off the power to the faulty motor, other motors still need to operate to traction the train. The rotor of the faulty motor continues to rotate, generating a magnetic field, and the motor still retains a high back electromotive force, causing the stator insulation fault to continue to expand rapidly. Therefore, inter-turn short circuits are one of the most serious faults in PMMs. Accurate diagnosis of inter-turn short circuits and fault-tolerant handling are crucial aspects of the widespread application of PMMs today.

[0003] Inter-turn short circuits are a common fault. An inter-turn short circuit occurs when adjacent coils of a motor short circuit. This circuit generates a large circulating current, producing significant localized heat and severely damaging phase-to-phase and main insulation, potentially leading to motor burnout. For low-power motors, the progression from an early inter-turn short circuit to insulation burnout and grounding typically takes 20-60 seconds. For high-power, high-speed permanent magnet motors, this process is even shorter. Insulation faults often begin with inter-turn faults. If early inter-turn short circuits are diagnosed promptly and the motor is stopped or protected, only stator rewinding and other repairs are needed. However, waiting until insulation grounding or stator core burnout results in more severe damage and significantly increased repair costs. Therefore, diagnosing early inter-turn insulation faults is crucial.

[0004] Introduction to existing technologies:

[0005] Patent CN115792700A, "Method and Related Equipment for Detecting Inter-turn Short Circuit Faults in Traction Motors," provides a method for detecting inter-turn short circuit faults in traction motors. This method converts the three-phase current in the traction motor into a negative-sequence current through signal transformation and analyzes the negative-sequence current waveform using a waveform recognition model based on a long short-term memory network. This enables real-time detection and diagnosis of inter-turn short circuit faults in the traction motor. This method combines signal transformation and artificial intelligence algorithms, improving the accuracy and efficiency of fault detection and effectively ensuring the safe operation of electric locomotive traction motors.

[0006] Patent CN111722145B, "A Method for Diagnosing Minor Inter-turn Short Circuit Faults in the Excitation Winding of a Synchronous Motor," combines a Convolutional Network (CNN) and a Deep Belief Network (DBN) to diagnose minor inter-turn short circuit faults in the excitation winding of a synchronous motor. This invention collects historical data from both normal and minor fault conditions of the synchronous motor, and trains the CNN and DBN after normalization. The invention uses 12 physical quantities as input to the CNN to extract data features, uses these extracted features as input to the DBN, and uses fault and normal conditions as outputs of the DBN. Field-collected data is input to the trained network; a fault is determined when 20 consecutive data samples are classified as faults.

[0007] Patent CN111060844A, "A Method and Apparatus for Diagnosing Inter-turn Short Circuit Faults in High-Speed ​​Train Traction Drive Systems," discloses a method and apparatus for diagnosing inter-turn short circuit faults in high-speed train traction drive systems. This invention relates to the field of high-speed train fault diagnosis and can solve the problem of diagnosing inter-turn short circuit faults in high-speed train traction systems, improving the accuracy of fault diagnosis. The invention includes: collecting steady-state operating data of high-speed trains and preprocessing offline data; performing wavelet packet transform on the preprocessed data and extracting time-domain and time-frequency domain features; using the extracted features to train a random forest model; preprocessing online data and extracting features; and using the obtained random forest model to classify the online data to achieve fault diagnosis. This invention is applicable to the diagnosis of inter-turn short circuit faults in high-speed train traction motors.

[0008] Deficiencies of existing technology:

[0009] The above patents all rely on data-driven intelligent diagnostic algorithms based on deep learning or machine learning, such as Long Short-Term Memory networks, combinations of Convolutional Networks (CNN) and Deep Belief Networks (DBN), and Random Forests. These algorithms not only require a large amount of fault data for training the model but also demand high computing power, which increases the difficulty and reduces the efficiency of fault diagnosis. Furthermore, the third patent, which collects steady-state operating data from high-speed trains and preprocesses offline data, raises questions about its applicability to complex online operating conditions.

[0010] Furthermore, most existing short-turn fault diagnosis technologies are based on motor current. Since motor control algorithms largely employ closed-loop current control, their basic principle is to measure the difference (error) between the motor current and the setpoint, and then use proportional, integral, and derivative operations to obtain the control signal, thereby adjusting the motor's operating state to ensure stable operation of the motor current according to the setpoint. In current closed-loop control systems, to reduce motor power harmonic distortion and improve motor performance and efficiency, the controller typically filters and compensates the current signal, making the output current as close as possible to the fundamental frequency, thus reducing harmonic components. Especially for the third harmonic, the control algorithm can use precise current sensors to monitor the current signal in real time and quickly respond to system changes, thereby reducing the third harmonic content. However, this approach is undoubtedly very unfriendly for diagnosing short-turn faults in motors. Therefore, diagnosing short turns solely based on motor current is insufficient, especially for early-stage short-turn fault diagnosis, where the probability of missed detection is high. Summary of the Invention

[0011] The purpose of this invention is to provide a multi-signal fusion method for diagnosing inter-turn short circuit faults in permanent magnet motors, which solves the technical problem that existing technologies rely solely on motor current to diagnose inter-turn short circuits, especially for early-stage inter-turn short circuit diagnoses, resulting in a high probability of missed detections.

[0012] This invention is achieved using the following technical solution:

[0013] A multi-signal fusion method for diagnosing inter-turn short-circuit faults in permanent magnet motors involves real-time acquisition of the motor's three-phase current, three-phase line voltage, and radial vibration acceleration signals of the bearing end cover load-bearing area during motor operation. The specific method includes the following steps:

[0014] Step 1: Extraction of current fault features;

[0015] Step 2: Vibration fault feature extraction;

[0016] S21. Vibration spectrum fault feature extraction;

[0017] S22, Fault feature extraction from vibration envelope spectrum;

[0018] S23. Calculation of vibration fault characteristic indicators;

[0019] Step 3: Calculate the online characteristic markers for short turns;

[0020] Step 4: Determine whether to enable back EMF characteristic calculation;

[0021] When the short-circuit characteristic flag f d_in =0, directly determining that the motor has no short-turn fault, the motor is running normally, and this short-turn diagnosis process ends.

[0022] Otherwise, the short-circuit online characteristic flag f d_in =1, the motor control algorithm issues a command to disconnect the three-phase input contactor of the motor, the motor is reverse-driven, the back EMF signal of the motor is collected, and the back EMF fault feature extraction calculation is continued.

[0023] Step 5: Extraction of back EMF fault features;

[0024] Step Six: Final Diagnosis of Short Motor Turns;

[0025] When the back EMF fault characteristic flag is equal to 0, it is directly determined that the motor has no short-turn fault. The motor control algorithm re-executes the motor three-phase input contactor closing command, the motor runs normally, and the short-turn diagnosis process ends.

[0026] Otherwise, if the back EMF fault characteristic flag equals 1, a short-circuit warning for the motor turns is directly triggered. The motor control algorithm issues a short-circuit command to the three-phase input contactors, and the motor is protected against a three-phase short circuit. This completes the short-circuit diagnosis process. The motor can continue to run in reverse until the conditions for removing the motor from the system are met, allowing for subsequent inter-turn insulation measurement and maintenance.

[0027] More preferably, in step one, the current signal is normalized to [-1, 1], and the normalized current signal is then subjected to an FFT (Fast Fourier Transform) to calculate the spectrum function F of the current signal. i (f), calculate the fundamental frequency f of the current. i The third harmonic feature is extracted in the current spectrum according to the following formula (1):

[0028]

[0029] When F d_i When the current exceeds the fault threshold, the current fault characteristic flag f d_i =1, otherwise f d_i =0.

[0030] More preferably, S21 specifically involves normalizing the vibration acceleration signal to [0, 1], and then performing an FFT (Fast Fourier Transform) on the normalized vibration acceleration signal to obtain the spectrum function F of the vibration acceleration signal. v (f) Based on the fundamental frequency f of the permanent magnet synchronous motor current i Calculate the motor's rotational frequency fr, extract the 2nd, 4th, 6th, and 8th harmonic characteristics from the vibration spectrum, and extract the vibration spectrum characteristics from the vibration spectrum according to the following formula (2):

[0031]

[0032] When F d_vf When the value exceeds the fault threshold, the vibration spectrum fault characteristic flag f d_vf=1, otherwise f d_vf =0;

[0033] S22 specifically involves performing a Hilbert transform on the normalized vibration acceleration signal calculated in step S21, calculating the envelope signal of the vibration acceleration signal, and then performing a Fast Fourier Transform (FFT) on the envelope signal to calculate the vibration envelope spectrum function F. v_h (f), based on the fundamental frequency f of the current i Calculate the motor rotation frequency fr, and extract the vibration envelope spectrum features in the vibration envelope spectrum according to the following formula (3);

[0034]

[0035] When F d_vh When the fault threshold is exceeded, the vibration envelope spectrum fault characteristic flag f d_vh =1, otherwise f d_vh =0;

[0036] S23 specifically refers to the vibration spectrum fault characteristic marker f calculated above. d_vf Vibration envelope spectrum fault characteristic marker f d_vh According to the following formula (4), the vibration fault characteristic indicator f is obtained by performing an AND operation. d_v :

[0037] f d_v =f d_vf ∩f d_vh Formula (4).

[0038] In a further preferred embodiment, step three specifically involves calculating the vibration characteristic indicator f based on the above-mentioned parameters. d_v and current fault characteristic mark f d_i According to the following formula (5), the short-circuit characteristic flag f is obtained by performing an OR operation. d_in :

[0039] f d_in =f d_i ∪f d_v Formula (5).

[0040] In a further preferred embodiment, step five specifically involves normalizing the back EMF to [-1, 1], and then performing an FFT (Fast Fourier Transform) on the normalized back EMF signal to obtain the spectral function F of the back EMF signal. u (f) Calculate the fundamental frequency f of the back electromotive force. u The third harmonic distortion rate feature is extracted in the back potential spectrum according to the following formula (6):

[0041]

[0042] When F d_uWhen the back EMF is greater than the fault threshold, the fault characteristic flag f d_u =1, otherwise f d_u =0.

[0043] This invention patent integrates and analyzes signals such as motor current, back EMF, and vibration, comprehensively considering the impact of inter-turn short circuits on various aspects of the motor. Using signal processing methods and feature calculations, it can accurately extract and locate the features of inter-turn short circuits. Furthermore, the algorithm can be implemented in general embedded systems, offering high timeliness and accuracy. Attached Figure Description

[0044] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0045] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 This is a flowchart illustrating the process of this invention. Detailed Implementation

[0047] Numerous specific details are set forth in the following description to provide a thorough understanding of the invention; however, the invention may also be practiced in other ways different from those described herein. Clearly, the embodiments described are only a portion, not all, of the embodiments of the invention. Specific embodiments of the invention will now be described in detail with reference to the accompanying drawings.

[0048] Example 1

[0049] A multi-signal fusion method for diagnosing inter-turn short-circuit faults in permanent magnet motors involves real-time acquisition of the motor's three-phase current, three-phase line voltage, and radial vibration acceleration signals of the bearing end cover load-bearing area during motor operation. The specific method includes the following steps:

[0050] Step 1: Extraction of current fault features;

[0051] The current signal is normalized to [-1, 1], and then the frequency spectrum function F is calculated by performing a Fast Fourier Transform (FFT) on the normalized current signal. i (f), calculate the fundamental frequency f of the current. i The third harmonic feature is extracted in the current spectrum according to the following formula (1):

[0052]

[0053] When F d_i When the current exceeds the fault threshold, the current fault characteristic flag f d_i =1, otherwise f d_i =0.

[0054] Step 2: Vibration fault feature extraction;

[0055] S21. Vibration spectrum fault feature extraction;

[0056] S21 specifically involves normalizing the vibration acceleration signal to [0, 1], and then performing an FFT (Fast Fourier Transform) on the normalized vibration acceleration signal to obtain the spectrum function F of the vibration acceleration signal. v (f) Based on the fundamental frequency f of the permanent magnet synchronous motor current i Calculate the motor's rotational frequency fr, extract the 2nd, 4th, 6th, and 8th harmonic characteristics from the vibration spectrum, and extract the vibration spectrum characteristics from the vibration spectrum according to the following formula (2):

[0057]

[0058] When F d_vf When the value exceeds the fault threshold, the vibration spectrum fault characteristic flag f d_vf =1, otherwise f d_vf =0;

[0059] S22, Fault feature extraction from vibration envelope spectrum;

[0060] S22 specifically involves performing a Hilbert transform on the normalized vibration acceleration signal calculated in step S21, calculating the envelope signal of the vibration acceleration signal, and then performing a Fast Fourier Transform (FFT) on the envelope signal to calculate the vibration envelope spectrum function F. v_h (f), based on the fundamental frequency f of the current i Calculate the motor rotation frequency fr, and extract the vibration envelope spectrum features in the vibration envelope spectrum according to the following formula (3);

[0061]

[0062] When F d_vh When the fault threshold is exceeded, the vibration envelope spectrum fault characteristic flag f d_vh =1, otherwise f d_vh =0;

[0063] S23. Calculation of vibration fault characteristic indicators;

[0064] S23 specifically refers to the vibration spectrum fault characteristic marker f calculated above. d_vf Vibration envelope spectrum fault characteristic marker f d_vhAccording to the following formula (4), the vibration fault characteristic indicator f is obtained by performing an AND operation. d_v :

[0065] f d_v =f d_vf ∩f d_vh Formula (4).

[0066] Step 3: Calculate the online characteristic markers for short turns;

[0067] Based on the vibration characteristic indicator f calculated above d_v and current fault characteristic mark f d_i According to the following formula (5), the short-circuit characteristic flag f is obtained by performing an OR operation. d_in :

[0068] f d_in =f d_i ∪f d_v Formula (5).

[0069] Step 4: Determine whether to enable back EMF characteristic calculation;

[0070] If the short-turn online feature flag is equal to 0, it is directly determined that the motor has no short-turn fault and the motor runs normally, and the short-turn diagnosis process ends; if the short-turn online feature flag is equal to 1, the motor control algorithm issues a command to disconnect the three-phase input contactor of the motor, the motor is reverse-driven, the back EMF signal of the motor is collected, and the back EMF fault feature extraction calculation continues, that is, the following steps continue.

[0071] Step 5: Extraction of back EMF fault features;

[0072] The back EMF is normalized to [-1, 1], and the spectrum function F of the back EMF signal is obtained by performing an FFT (Fast Fourier Transform) on the normalized back EMF signal. u (f) Calculate the fundamental frequency f of the back electromotive force. u The third harmonic distortion rate feature is extracted in the back potential spectrum according to the following formula (6):

[0073]

[0074] When F d_u When the back EMF is greater than the fault threshold, the fault characteristic flag f d_u =1, otherwise f d_u =0.

[0075] Step Six: Final Diagnosis of Short Motor Turns;

[0076] When the back EMF fault characteristic flag f d_u=0, directly determining that there is no short-turn fault in the motor, the motor control algorithm re-executes the motor three-phase input contactor closing command, the motor runs normally, and this short-turn diagnosis process ends.

[0077] Otherwise, when the back EMF fault characteristic flag f d_u =1, directly determining a short-circuit warning for the motor turns. The motor control algorithm issues a short-circuit command to the three-phase input contactors of the motor, and the motor is protected against three-phase short circuit. This concludes the short-circuit diagnosis process. The motor can continue to run in reverse until the conditions for removing the motor from the system are met, allowing for subsequent inter-turn insulation measurement and maintenance.

[0078] During implementation, normal data is typically collected over a period of time, and the fault threshold is set based on the baseline of normal conditions, which is generally more than 5 times the normal value.

[0079] The working principle is as follows:

[0080] (1) When a short circuit occurs between turns in a phase of the stator winding of a permanent magnet motor, the short-circuit current characteristic mechanism is as follows:

[0081] ① Electromagnetic symmetry disruption: When a short-circuit fault occurs in a motor, its electromagnetic symmetry is disrupted. This disruption prevents the third harmonic component in the short-circuit current from being canceled out, thus making it significantly visible in the current.

[0082] ② Changes in short-circuit current composition: When a short circuit occurs between motor turns, a local short-circuit loop is generated in the winding. This affects not only the fundamental current but also higher-order harmonic currents, especially the third harmonic current. A short circuit leads to an increase in the local electromagnetic field, further affecting the harmonic components of the current. The relationship between fault severity and the third harmonic: As the severity of the short-circuit fault increases, i.e., the number of short-circuited turns increases, the proportion of the third harmonic in the short-circuit current increases significantly. However, since most motor control algorithms employ closed-loop current control technology, they reduce the harmonic content of the current, especially the third characteristic harmonic.

[0083] ③ When a negative sequence component is generated in the motor current, three-phase amplitude imbalance and phase imbalance will occur; ④ When the motor turns are shortened, the impedance changes and the effective value of the fundamental current increases.

[0084] ⑤ Among the above characteristics, the third harmonic current is the most prominent.

[0085] (2) When the motor is de-energized and in open-loop state, the back EMF of the motor is measured. When a short circuit occurs between turns in a phase of the stator winding of the permanent magnet motor, the back EMF short-turn characteristic mechanism is similar to that of the motor current short-turn characteristic mechanism, but it is more obvious than the characteristics of the motor current: ① The back EMF of the motor generates increased harmonic components, and the third harmonic belongs to the increased fault harmonic component.

[0086] ② The negative sequence component generated in the back EMF of the motor will cause three-phase amplitude imbalance, phase imbalance, etc.

[0087] ③ Due to the change in impedance after the motor turns are shortened, the magnetic field balance is disrupted, and the back EMF will decrease due to the influence of the short-circuit current.

[0088] ④ Among the above characteristics, the third harmonic of the back EMF increases while the fundamental wave decreases, thus the third harmonic / fundamental wave characteristics of the back EMF are obvious.

[0089] (3) When a short circuit occurs between turns in a phase of the stator winding of a permanent magnet motor, the magnetic field distribution changes, the air gap magnetic flux density changes, and the resultant force of the magnetic pull per unit area is no longer zero. The additional magnetic pull causes abnormal vibration of the motor. Therefore, the fault characteristic mechanism of motor vibration is as follows:

[0090] ① The effective value of motor vibration increases;

[0091] ② It will generate electromagnetic forces with even harmonic frequencies such as 2f, 4f, 6f, and 8f, and the stator vibration change at frequency 2f is relatively large;

[0092] ③ The motor's frequency vibration increases, and it is modulated by the frequency.

[0093] ④Among the above features, features ② and ③ are more obvious.

[0094] The current signal of the motor in the power supply state is affected by the power supply; the vibration acceleration signal is affected by the operating conditions and will have some interference. However, after the three-phase isolating contactor is disconnected, the permanent magnet motor is in a no-load generating mode, and the normal back EMF is an ideal sine wave, so the fault back EMF is relatively obvious and easy to identify. Theoretically, the motor should always be in traction state. Only under this fault condition is a secondary judgment made in the generating state to determine whether it is a short-turn fault.

[0095] Example 2

[0096] During traction motor operation, motor current and vibration acceleration signals are collected in real time, and short-turn fault characteristics in both signals are extracted. When either signal reaches the warning condition (fault flag is 1), the motor is isolated using a three-phase isolating contactor. With other normal motors running in the same vehicle, this motor is in a three-phase open-circuit reverse-dragging condition. At this time, the short-turn fault characteristics of the motor's back EMF are extracted in real time. When the back EMF reaches the warning condition, the motor is protected against a three-phase short circuit using the three-phase isolating contactor, and the driver is prompted to take appropriate action. Otherwise, the three-phase isolating contactor is closed, the motor is reconnected, and the motor resumes normal operation. The operating status is then continuously monitored. (Specific procedures follow.) Figure 1 As shown below, specific examples will be used to illustrate this.

[0097] A short-circuit test platform for permanent magnet motors was constructed. Two turns of the stator winding of permanent magnet motor A were short-circuited via a copper busbar or a high-power isolating contactor. Another permanent magnet motor B of the same model, in normal condition, was used as the test machine. Motor A was then started to drive motor B, with motor A operating at its rated speed. During operation, the three-phase current, three-phase line voltage, and radial vibration acceleration signals of the bearing end cover load-bearing area of ​​motor A were collected in real time, and the following calculations and analyses were performed:

[0098] ① Current fault feature extraction: Normalize the current signal of any phase to [-1, 1], and perform FFT (Fast Fourier Transform) on the normalized current signal to obtain the spectrum function F of the current signal. i (f), calculate the fundamental frequency f of the current. i According to formula (1), the third harmonic characteristic value F is obtained in the current spectrum. d_i Calculate F d_i The fault current characteristic flag f is greater than the fault threshold. d_i =1. ② Vibration spectrum fault feature extraction: The vibration acceleration signal is normalized to [0,1], and the frequency spectrum function F of the vibration acceleration signal is obtained by performing FFT (Fast Fourier Transform) on the normalized vibration acceleration signal. v (f) Calculate the motor rotation frequency fr based on the fundamental frequency fi of the permanent magnet synchronous motor current, and perform vibration spectrum characteristic value F in the vibration spectrum according to formula (2). d_vf Calculate F d_vf The vibration spectrum fault characteristic flag f is greater than the fault threshold. d_vf =1.

[0099] ③ Vibration envelope spectrum fault feature extraction: Perform Hilbert transform on the normalized vibration acceleration signal calculated in step ②, and calculate the envelope signal of the vibration acceleration signal. Perform FFT (Fast Fourier Transform) on the envelope signal to calculate the vibration envelope spectrum function F. v_h (f), based on the fundamental frequency f of the current i Calculate the motor speed f r According to formula (3), the eigenvalues ​​of the vibration envelope spectrum are calculated in the vibration envelope spectrum, F d_vh The vibration envelope spectrum fault characteristic indicator f is greater than the fault threshold. d_vh =1.

[0100] ④ Calculation of vibration fault characteristic indicators: Based on the vibration spectrum fault characteristic indicators f calculated above... d_vf Vibration envelope spectrum fault characteristic marker f d_vh According to formula (4), the vibration fault characteristic indicator f is obtained by performing an AND operation. d_v =1.

[0101] ⑤ Calculation of short-circuit characteristic indicators: Based on the vibration characteristic indicator f calculated above... d_v and current fault characteristic mark f d_i According to formula (5), the short-circuit characteristic indicator f is obtained by performing an OR operation. d_in =1.

[0102] ⑥ Determine whether the back EMF characteristic calculation is enabled: Turn short online characteristic flag f d_in =1, the test platform motor control algorithm issues a three-phase input contactor disconnect command, the three-phase power supply of motor A is cut off, the test platform simultaneously controls motor B to run in traction mode, motor A is dragged in reverse, and the calculation of fault features of motor A line voltage (back EMF) continues.

[0103] ⑦ Back EMF Fault Feature Extraction: The back EMF is normalized to [-1, 1], and the normalized back EMF signal is calculated using FFT (Fast Fourier Transform) to obtain the spectrum function F of the back EMF signal. u (f) Calculate the fundamental frequency f of the back electromotive force. u According to formula (6), the characteristic value of the third harmonic distortion rate F is obtained in the back potential spectrum. d_u Calculate F d_u The back EMF fault characteristic flag f is greater than the fault threshold. d_u =1.

[0104] ⑧ Final diagnosis of short motor turns: Back EMF fault characteristic indicator f d_u =1, directly determine the motor short circuit warning, the test platform motor control algorithm issues a three-phase input contactor short circuit command, motor A is protected by three-phase short circuit, and this short circuit diagnosis process ends.

[0105] When the number of turns involved in an early short circuit in a permanent magnet motor is small, the fault characteristics are often not obvious. Diagnosis and identification based on the characteristics of a single signal often leads to false alarms, easily misdiagnosing other motor faults as short-circuit faults. Since short-circuit faults typically result in three-phase short circuits or complete shutdown, this can lead to performance degradation, reduced power and efficiency in multi-motor systems, and even system-wide shutdown in single-motor systems. False alarms can cause significant economic losses. Increasing the fault characteristic threshold to improve diagnostic accuracy can lead to missed or delayed early warnings. Even a minor inter-turn short circuit generates a large circulating current in the short-circuit coil, producing significant heat and causing insulation damage. Delaying optimal diagnosis can quickly lead to multi-turn short circuits, or even phase-to-phase short circuits and grounding, severely impacting the entire system. Furthermore, since motor control algorithms often employ closed-loop current control technology, reducing current harmonic content, especially short-circuit characteristic harmonics, diagnosing short-circuit faults solely based on motor current has a high rate of missed and delayed diagnosis. Therefore, when a short circuit occurs between turns in a motor, by extracting the multi-dimensional features of the motor's electrical and mechanical signals and performing fusion diagnosis, the short circuit can be diagnosed and located in a timely and accurate manner, and an early warning can be issued. This reduces the probability of missed, false, or delayed alarms, and reminds the system to take timely fault safety guidance measures such as stopping operation or activating three-phase short circuit protection to prevent the fault from escalating further and minimize the impact of the short circuit fault to the greatest extent possible. This is of great importance.

[0106] The beneficial effects of this technical solution are as follows: 1. Real-time monitoring: This invention can monitor the traction motor in real time. The algorithm execution cycle is at the millisecond level, with strong timeliness, timely warning of short-turn insulation and fault protection processing, avoiding fault escalation. 2. Precise warning: This invention considers the fault characteristics of short-turn insulation from multiple dimensions and performs fusion diagnosis. It can not only warn of short-turn insulation faults in the motor, but also accurately determine the fault location, thereby more effectively preventing the expansion of the fault and providing precise maintenance guidance. 3. Reduced maintenance costs and lower fault handling difficulty: If a warning is issued and corresponding measures are taken when short-turn insulation occurs, the damage to the motor is minimized. Only stator winding repair is required, reducing maintenance costs. If the short-turn insulation develops to the point of large-area insulation burnout or stator core burnout before treatment, the motor will suffer more severe damage, and the maintenance workload will increase significantly. 4. Improved motor service life: By detecting and handling inter-turn insulation faults early, motor burnout due to decreased insulation performance can be avoided, thereby improving the motor's service life. Therefore, early diagnosis of inter-turn insulation faults is of great value.

[0107] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the present invention. Although detailed descriptions have been provided with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments, and they should all be covered within the protection scope of the claims.

Claims

1. A method for diagnosing inter-turn short-circuit faults in a permanent magnet motor using multi-signal fusion, characterized in that: During motor operation, the three-phase current, three-phase line voltage, and radial vibration acceleration signals of the bearing end cover load-bearing area are collected in real time. The specific method includes the following steps: Step 1: Extraction of current fault features; Step 2: Vibration fault feature extraction; S21. Vibration spectrum fault feature extraction; S22, Fault feature extraction from vibration envelope spectrum; S23. Calculation of vibration fault characteristic indicators; Step 3: Calculation of short-circuit online characteristic indicators; specifically, based on the vibration fault characteristic indicator f d_v and current fault characteristic mark f d_i According to the following formula (5), the short-circuit characteristic flag f is obtained by performing an OR operation. d_in : Official (5); Step 4: Determine whether to enable back EMF characteristic calculation; If the short-turn online feature flag is equal to 0, it is directly determined that the motor has no short-turn fault and the motor runs normally, and the short-turn diagnosis process ends; if the short-turn online feature flag is equal to 1, the motor control algorithm issues a command to disconnect the three-phase input contactor of the motor, the motor is reverse-driven, the back EMF signal of the motor is collected, and the back EMF fault feature extraction calculation continues, that is, the following steps continue. Step 5: Extraction of back EMF fault features; Step Six: Final Diagnosis of Short Motor Turns; When the back EMF fault characteristic flag is equal to 0, it is directly determined that the motor has no short-turn fault. The motor control algorithm re-executes the motor three-phase input contactor closing command, the motor runs normally, and the short-turn diagnosis process ends. Otherwise, when the back EMF fault characteristic flag is equal to 1, a short-circuit warning for the motor turns is directly determined. The motor control algorithm issues a short-circuit command for the three-phase input contactor of the motor, and the motor is protected against three-phase short circuit. This short-circuit diagnosis process ends, and the motor continues to run in reverse until the conditions for removing the motor from the system are met, and subsequent motor turn insulation measurement and maintenance work is carried out.

2. The method for diagnosing inter-turn short-circuit faults in a permanent magnet motor by multi-signal fusion according to claim 1, characterized in that: In step one, the current signal is normalized to [-1, 1], and the frequency spectrum function F is obtained by performing a Fast Fourier Transform (FFT) on the normalized current signal. i (f), calculate the fundamental frequency f of the current. i The third harmonic feature is extracted in the current spectrum according to the following formula (1): Official (1) When F d_i When the current exceeds the fault threshold, the current fault characteristic flag f d_i =1, otherwise f d_i =0.

3. The method for diagnosing inter-turn short-circuit faults in a permanent magnet motor by multi-signal fusion according to claim 2, characterized in that: S21 specifically involves normalizing the vibration acceleration signal to [0, 1], and then performing an FFT (Fast Fourier Transform) on the normalized vibration acceleration signal to obtain the spectrum function F of the vibration acceleration signal. v (f) Based on the fundamental frequency f of the permanent magnet synchronous motor current i Calculate the motor's rotational frequency fr, extract the 2nd, 4th, 6th, and 8th harmonic characteristics from the vibration spectrum, and extract the vibration spectrum characteristics from the vibration spectrum according to the following formula (2): Official (2) When F d_vf When the value exceeds the fault threshold, the vibration spectrum fault characteristic flag f d_vf =1, otherwise f d_vf =0; S22 specifically involves performing a Hilbert transform on the normalized vibration acceleration signal calculated in step S21, calculating the envelope signal of the vibration acceleration signal, and then performing a Fast Fourier Transform (FFT) on the envelope signal to calculate the vibration envelope spectrum function F. v_h (f), based on the fundamental frequency f of the current i Calculate the motor rotation frequency fr, and extract the vibration envelope spectrum features in the vibration envelope spectrum according to the following formula (3); Official (3) When F d_vh When the fault threshold is exceeded, the vibration envelope spectrum fault characteristic flag f d_vh =1, otherwise f d_vh =0; S23 specifically refers to the vibration spectrum fault characteristic marker f calculated above. d_vf Vibration envelope spectrum fault characteristic marker f d_vh According to the following formula (4), the vibration fault characteristic indicator f is obtained by performing AND operation. d_v : Official (4).

4. The multi-signal fusion method for diagnosing inter-turn short-circuit faults in a permanent magnet motor according to claim 3, characterized in that: Step five specifically involves normalizing the back EMF to [-1, 1], and then performing an FFT (Fast Fourier Transform) on the normalized back EMF signal to obtain the spectral function F of the back EMF signal. u (f) Calculate the fundamental frequency f of the back electromotive force. u The third harmonic distortion rate feature is extracted in the back potential spectrum according to the following formula (6): Official (6) When F d_u When the back EMF is greater than the fault threshold, the fault characteristic flag f d_u =1, otherwise f d_u =0.

Images