A method for combined online detection of electrical faults and demagnetization of permanent magnets of an electric machine

Abstract

The application discloses a kind of motor electrical fault and permanent magnet demagnetization combined online detection method, it is related to motor fault diagnosis technical field, the method uses sequential diagnosis strategy to construct unified joint detection framework: first, four fault sub-item detections of three-phase current unbalance degree, hall sensor failure, interphase resistance consistency, motor insulation state are executed in parallel, when it is directly output corresponding fault code and skips demagnetization detection after comprehensive determination exists electrical fault or insulation serious anomaly;Confirm that electrical circuit and insulation state satisfy demagnetization detection condition, then extract fundamental wave and third harmonic component from phase voltage signal or reconstructed phase counter electromotive force waveform, complete demagnetization evaluation with third harmonic and fundamental wave amplitude ratio as demagnetization characteristic index, finally output unified diagnosis message.The application does not need additional sensor, does not inject test signal, can realize online detection in motor normal operation, solve the problem of electrical and demagnetization two kinds of fault characteristics cross coupling misjudgment, reduce system resource consumption.

Description

Technical Field

[0001] This invention relates to the field of motor fault diagnosis technology, and in particular to an online detection method for combined electrical faults and permanent magnet demagnetization in motors, applicable to multi-fault integrated online monitoring scenarios of permanent magnet synchronous drive motors for electric two-wheeled vehicles. Background Technology

[0002] Permanent magnet synchronous motors are widely used in electric two-wheeler drive systems due to their high efficiency and high power density. In actual operation, these motors are exposed to complex outdoor environments, facing conditions such as humidity, heat, vibration, and frequent start-stop cycles, making them prone to two types of core faults: electrical faults, including phase wire burnout, cable breakage, Hall sensor failure, inter-turn short circuits in the windings, and insulation degradation; and magnetic degradation faults, i.e., demagnetization of the permanent magnets. Failure to detect either type of fault in a timely manner can lead to reduced motor power, overheating and burnout, and even safety issues for the vehicle.

[0003] Existing detection solutions for the above two types of faults have the following shortcomings: First, electrical fault detection and demagnetization detection are independent of each other and lack a unified online joint diagnostic framework. The two detection methods each occupy system resources, resulting in low system resource utilization and high hardware and computing costs.

[0004] Secondly, existing demagnetization detection solutions have obvious limitations in terms of application scenarios: solutions based on vibration sensors require the addition of an accelerometer, increasing hardware costs and installation complexity; solutions based on detection coils are invasive modifications and are not suitable for mass-produced motors; solutions based on inductance identification require shutdown for detection, interfering with the normal drive control of the motor and making it impossible to achieve continuous online monitoring during operation; and demagnetization algorithms based on complex time-frequency analysis have a large computational load, making it difficult to run in real time on the limited computing power of low-cost controllers for electric two-wheelers.

[0005] Third, electrical faults and demagnetization faults exhibit cross-coupling characteristics in current / voltage signals: electrical faults such as burnt-out phase lines and broken cables can cause three-phase current imbalances, leading to back electromotive force waveform distortion via armature reaction, which may be misidentified as demagnetization characteristics; conversely, magnetic field distortion caused by demagnetization can also interfere with the accuracy of electrical fault detection. Existing solutions do not have a dedicated decoupling mechanism designed to address the cross-interference between these two types of faults, posing a risk of false alarms and missed alarms.

[0006] In summary, existing technologies lack a unified online detection solution that can integrate electrical faults and permanent magnet demagnetization faults into a single framework and achieve cross-decoupling without interfering with the normal operation of the motor. This makes it unsuitable for the low-cost, high-reliability online diagnostic requirements of electric two-wheelers. Summary of the Invention

[0007] To address the aforementioned problems and technical requirements, the inventors have proposed a combined online detection method for motor electrical faults and permanent magnet demagnetization. The technical solution of this invention is as follows: A combined online detection method for electrical faults in motors and demagnetization of permanent magnets includes the following steps: Acquire three-phase current, three-phase phase voltage, and Hall sensor signals during motor operation; Multiple electrical fault sub-item detections are performed in parallel based on signals, including three-phase current imbalance detection, Hall sensor failure detection, phase-to-phase resistance consistency assessment, and motor insulation status assessment. Based on the comprehensive judgment of the test results of each sub-item, if there is an electrical fault or serious insulation abnormality that reaches the preset fault level, the corresponding fault code will be directly output and the demagnetization test will be skipped; if it is confirmed that the electrical circuit and insulation condition meet the demagnetization test conditions, the demagnetization test will be entered. In the demagnetization detection process, demagnetization characteristic indicators are calculated based on the phase voltage signal or the reconstructed back electromotive force waveform. By comparing the demagnetization characteristic indicators with the baseline values, the degree of demagnetization of the permanent magnet is confirmed. This information is then integrated with the electrical fault results to output a motor fault diagnosis report.

[0008] A further technical solution involves calculating demagnetization characteristic indicators based on the phase voltage signal or the reconstructed back electromotive force waveform, including: Perform a Fast Fourier Transform or Goertzel algorithm on the three-phase phase voltage signal or back electromotive force waveform to extract the fundamental component and the third harmonic component in order to calculate the demagnetization characteristic index, which is the ratio of the third harmonic amplitude to the fundamental amplitude.

[0009] A further technical solution is that the method for obtaining the three-phase phase voltage signal includes: The three-phase winding terminal voltage is acquired using PWM synchronous sampling technology. After DC bus voltage normalization and digital bandpass filtering, the phase voltage signal is obtained. The center frequency of the digital bandpass filter is dynamically adjusted according to the current electric base frequency of the motor. Methods for obtaining the back electromotive force waveform include: Based on a sliding mode observer, the waveforms of the three-phase opposite electromotive forces are reconstructed online in a stationary coordinate system using the three-phase current and the DC bus voltage.

[0010] Its further technical solution is that the Hall sensor failure detection adopts a three-level progressive verification mechanism, including: The first level is edge continuity detection, which monitors whether the number of signal transitions in each Hall channel within each electrical cycle conforms to the normal pattern, in order to identify Hall sensor signal loss or jamming faults. The second level is the state combination legality detection, which verifies whether the state combination sequence of the signal traversal of each Hall channel in each electrical cycle conforms to the legal sequence of forward or reverse rotation, so as to identify multiple channel simultaneous abnormal faults. The third level is timing consistency detection, which calculates the coefficient of variation of the time interval of each sector within multiple consecutive electrical cycles to identify Hall sensor installation misalignment or timing disorder faults. Based on the combined results of the three-level verification, a Level 1 fault is output when only one channel is abnormal, and a Level 2 fault, which is of a higher level, is output when two or more channels are abnormal.

[0011] Its further technical solution is that the method for timing consistency detection includes: Using the six state transitions of the Hall signal within one electrical cycle as the boundary, each electrical cycle is divided into six sectors, and the duration of each sector is recorded as the sector time interval. Collect sector time interval data for no less than twelve electrical cycles continuously, and form a time series for each sector location; Calculate the mean and standard deviation of the time series for each sector to obtain the coefficient of variation for each sector. The coefficient of variation is the ratio of the standard deviation to the mean. When the coefficient of variation of any sector exceeds a preset threshold, or the ratio of the time interval between adjacent sectors exceeds a preset threshold, it is determined to be a Hall sensor installation misalignment or timing disorder fault.

[0012] A further technical solution is that the method for evaluating the consistency of phase-to-phase resistance includes: During the motor power-on self-test phase or low-speed standby condition, short-time DC test current is injected between each pair of the three-phase windings of the motor in sequence, and the corresponding line voltage and steady-state current values ​​are collected simultaneously. The DC resistance values ​​between each phase are then calculated. Calculate the average value of the three-phase interphase resistance. The maximum relative deviation is used to calculate the resistance asymmetry as an evaluation index, and the corresponding fault level is output by comparing it with the set threshold. The resistance asymmetry is expressed as:

[0013] In the formula, , , These are the calculated DC resistance values ​​between each phase.

[0014] The further technical solution is that the motor insulation condition assessment adopts a multi-dimensional joint judgment, which includes at least three dimensions: high-frequency leakage current detection, insulation resistance estimation, and shell potential difference detection. Each dimension independently sets three-level fault level judgment thresholds. High-frequency leakage current detection refers to extracting the high-frequency leakage current component of the motor winding to the ground terminal of the casing during the zero vector stage of PWM modulation in motor operation, and detecting the trend of its amplitude change compared with the baseline. Insulation resistance estimation refers to applying a DC test excitation to the motor windings during the zero-vector phase of PWM modulation in motor operation, and calculating the insulation resistance value of the windings to ground by measuring the leakage current of the windings to the grounding terminal of the casing. Housing potential difference detection refers to detecting the potential offset between the motor housing and the reference ground; When any one dimension reaches level three fault, it is judged as a serious insulation abnormality; when two or more dimensions reach level two fault at the same time, it is judged as insulation degradation of concern.

[0015] Its further technical solution is that the logic for making a comprehensive judgment based on the test results of each sub-item includes: When any of the sub-items in three-phase current imbalance detection, Hall sensor failure detection, or phase-to-phase resistance consistency assessment reaches a level two or above fault level, the corresponding electrical fault code is output and the demagnetization detection is terminated. When the insulation condition is assessed as severely abnormal, an insulation fault code is output and the demagnetization detection is terminated. When the insulation condition assessment indicates a concern about deterioration, demagnetization testing continues, but a low-confidence flag is added to the demagnetization assessment results; When all sub-items are normal or only have a level one fault, the demagnetization test conditions are confirmed to be met, and the demagnetization test phase begins.

[0016] The further technical solution is that the demagnetization detection process also includes: The amplitude and phase information of the three-phase fundamental current are extracted from the three-phase current signal by frequency domain transformation, and the positive sequence current component and negative sequence current component of the three-phase fundamental current are calculated based on symmetrical component transformation. Calculate the proportion of the negative sequence current component to the positive sequence current component; When the demagnetization characteristic indicators exceed the normal range and the negative sequence current ratio exceeds the preset threshold, it is judged as a suspected inter-turn short circuit, and no permanent magnet demagnetization fault conclusion is output. When the demagnetization characteristic indicators exceed the normal range and the negative sequence current ratio does not exceed the preset threshold, it is determined to be a permanent magnet demagnetization fault, and the corresponding demagnetization degree is output.

[0017] A further technical solution involves an adaptive normalization compensation step for operating conditions before calculating the demagnetization characteristic indicators. Using the current motor speed and load current as input, the original third harmonic amplitude is corrected by a preset compensation coefficient; The preset compensation coefficient is achieved through one of the following two methods: Offline calibration lookup method: Before the motor leaves the factory, the compensation coefficient matrix is ​​calibrated under different speed and load current combinations. During operation, the corresponding compensation coefficient is obtained by looking up the table according to the current operating conditions. Polynomial online fitting method: Perform two-dimensional polynomial surface fitting on the calibrated compensation coefficient matrix to obtain the analytical compensation expression, and calculate the compensation coefficient in real time by inputting the current working condition during runtime.

[0018] The beneficial technical effects of this invention are: 1. Sensorless online demagnetization detection: Using only the controller's existing three-phase current sensor and DC bus voltage detection circuit, the ratio of the third harmonic of the back electromotive force to the fundamental frequency amplitude is used as a demagnetization characteristic indicator. No additional vibration sensor or motor structure modification is required, which significantly reduces system cost and installation complexity.

[0019] 2. Integrated framework for joint detection of electrical faults and demagnetization: It covers both electrical faults (phase wire burnout, cable breakage, Hall sensor failure, inter-turn short circuit, insulation degradation) and permanent magnet demagnetization faults in a unified diagnostic process. The detection paths for the two types of faults share signal acquisition and preprocessing resources, improving diagnostic efficiency and reducing system resource consumption.

[0020] 3. Sequential diagnostic strategy for decoupling cross-interference: First, complete the electrical anomaly judgment and output electrical fault code through the electrical fault detection path. Then, perform demagnetization assessment on the premise that there is no electrical fault. This avoids the current distortion caused by electrical faults (such as phase line imbalance) being misjudged as demagnetization characteristics, effectively reducing the false alarm rate.

[0021] 4. Continuous online operation without interruption of normal drive: The third harmonic component of back electromotive force can be extracted online from the phase voltage signal during normal motor operation without stopping the machine, injecting test signals, or interfering with normal PWM drive, thus realizing true continuous online monitoring during motor operation.

[0022] 5. Adaptive Normalization Processing: The amplitude of the third harmonic of the back electromotive force is normalized and compensated based on the current speed and load current to eliminate the natural variation of the harmonic amplitude caused by the fluctuation of the operating conditions, so that the demagnetization characteristic index after compensation is stably distributed around the baseline value in the entire operating condition range.

[0023] 6. Controllable computational complexity and adaptability to low-cost hardware: Back EMF harmonic extraction based on FFT or Goertzel algorithms can be completed on an ARM Cortex-M4 level MCU in less than 50ms computation time. Combined with time-sharing scheduling within a 100ms diagnostic cycle, it does not encroach on the computational resources of real-time motor control. Attached Figure Description

[0024] Figure 1This is a complete schematic diagram of the online detection method for combined electrical faults in motors and demagnetization of permanent magnets provided in this application; Figure 2 This is a flowchart of the cross-decoupling of electrical faults and demagnetization faults based on a sequential diagnostic strategy provided in this application; Figure 3 This is a flowchart of the calculation of demagnetization characteristic indicators based on the back electromotive force detection link provided in this application, which includes direct sampling path A and sliding mode observer (SMO) reconstruction path B; Figure 4 This is a schematic diagram comparing the adaptive normalization compensation effect provided in this application, where (a) shows the change of the demagnetization characteristic index with the fluctuation of the working condition (speed, load) before compensation, and (b) shows the change of the demagnetization characteristic index with the fluctuation of the working condition (speed, load) after compensation. Figure 5 This is a schematic diagram of the four-level quantitative grading standard for demagnetization and maintenance recommendations provided in this application. Detailed Implementation

[0025] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0026] This embodiment provides a method for joint online detection of electrical faults in motors and demagnetization of permanent magnets, applied to the on-board online diagnostic scenario of permanent magnet synchronous direct drive motors in electric two-wheeled vehicles. The overall approach adopts a sequential diagnostic strategy to construct a unified joint detection framework: first, full-dimensional detection of electrical faults is completed, and after confirming that the electrical circuit and insulation status meet the demagnetization detection conditions, demagnetization assessment is performed, thus achieving cross-decoupling of the two types of fault characteristics at the process level. The entire detection process is continuously executed during normal motor operation, and demagnetization detection does not require machine shutdown, injection of additional test signals, or interference with normal PWM drive control.

[0027] Combination Figure 1 , Figure 2 As shown, the detection method is executed cyclically according to the following six steps, and the diagnostic cycle can be configured to be on the order of hundreds of milliseconds. Specifically, it includes the following steps: S1 Signal Acquisition and Preprocessing. This involves acquiring various raw signals during motor operation, standardizing and preprocessing them for subsequent use by other detection pathways. Acquisition items include: three-phase current. , , Three-phase phase voltage , , (Or reconstructed via an observer), three-phase Hall sensor signals HA, HB, and HC. Optionally, DC bus voltage may also be included. motor speed ω Rotor position θ And motor temperature, etc.

[0028] Preprocessing includes common operations such as signal filtering, zero-point calibration, and DC bias removal; among which, the DC bus voltage... It is also used for phase voltage normalization calibration and observer reconstruction input reference, which can eliminate the influence of bus voltage fluctuations on subsequent harmonic measurements.

[0029] S2 Parallel Electrical Fault Detection. Based on the pre-processed signal, four electrical fault sub-items are detected in parallel, with each sub-item calculated and output independently. In this embodiment, the detection sub-items include three-phase current imbalance detection, Hall sensor failure detection, phase-to-phase resistance consistency assessment, and motor insulation status assessment. The detection process for each item will be described in detail later.

[0030] S3 Comprehensive Electrical Fault Judgment (Core of Sequential Strategy). This involves a comprehensive judgment based on the detection results of the four sub-items mentioned above to determine the subsequent process flow. It is the core step in achieving cross-decoupling between electrical faults and demagnetization faults: If an electrical fault or severe insulation abnormality reaches a preset fault level, the corresponding fault code is directly output and demagnetization detection is skipped; if the electrical circuit and insulation condition are confirmed to meet the demagnetization detection conditions, the demagnetization detection process begins.

[0031] S4 executes the demagnetization detection step. In this step, demagnetization characteristic indicators are calculated based on the phase voltage signal or the reconstructed back electromotive force waveform. In this embodiment, the acquired three-phase back electromotive force signals (three-phase phase voltage signals or back electromotive force waveforms) are analyzed in the frequency domain using Fast Fourier Transform (FFT) or the Goertzel algorithm to extract the fundamental component amplitude. With the amplitude of the third harmonic component To calculate the demagnetization characteristic index, this index is defined as the ratio of the third harmonic amplitude to the fundamental amplitude:

[0032] In this embodiment, the preferred option is... and As a demagnetizing characteristic, it is taken into account that under normal permanent magnet conditions, the air gap magnetic field has an almost sinusoidal distribution. It stabilizes near the baseline value; when demagnetization occurs, the air gap magnetic field waveform is distorted, and the third harmonic component relatively increases. The temperature subsequently rises. Therefore, continuous monitoring is necessary. The rate of change relative to the factory baseline allows for online quantitative assessment of the degree of demagnetization.

[0033] S5 Demagnetization Degree Assessment and Grading. The demagnetization characteristic indicators are compared with baseline values ​​to confirm the degree of permanent magnet demagnetization. In this embodiment, the demagnetization characteristic indicator values ​​measured under the motor's factory rated operating conditions or during the initial operation phase are used. The statistical mean was used as the baseline value. Based on the multiples of the demagnetization characteristic indicators relative to the baseline value, the degree of demagnetization was divided into four levels: normal, mild, moderate, and severe demagnetization. The rate of change of the demagnetization characteristic indicators over time was calculated simultaneously for degradation trend prediction and remaining service life assessment.

[0034] S6 motor fault diagnosis results output. Integrating electrical fault detection results and demagnetization assessment results, the system outputs a motor fault diagnosis message. The message content includes fault type code, fault level, current value and trend of demagnetization characteristic indicators, insulation confidence level marker, timestamp, and operating condition snapshot information, which can be accessed by the vehicle controller, cloud platform, or terminal equipment.

[0035] The entire detection method shares the same set of signal acquisition and preprocessing resources. Electrical fault detection and demagnetization detection are scheduled in a time-sharing manner. The computational complexity is adapted to low-cost vehicle controllers, without crowding out the computing power for real-time motor control, thus achieving true continuous online joint diagnosis during operation.

[0036] In S2, the three-phase current imbalance detection specifically includes the following: After low-pass filtering the three-phase current signals acquired in S1, the effective value of the fundamental period of each phase current is calculated. , , Then calculate the three-phase average current. Calculate the percentage deviation of the effective value of each phase current from the average value, and take the maximum value of the three as the three-phase current imbalance. The fault levels are divided into four categories based on the degree of imbalance: normal, level one, level two, and level three. Level two and above trigger S3 jump.

[0037] Optional, Corresponding to the normal state, Corresponding to Level 1 fault, Corresponding to level two faults, This corresponds to a level three fault.

[0038] In S2, the Hall sensor failure detection adopts a three-level progressive verification mechanism, which includes the following: Level 1: Edge Continuity Detection. This monitors the number of signal transitions in the three Hall channels within each electrical cycle. Under normal circumstances, each phase Hall HA, HB, and HC generates one rising edge and one falling edge per electrical cycle, totaling six transitions per cycle. In this embodiment, two consecutive electrical cycles are monitored. If the number of transitions in any channel is less than the normal number (i.e., lacking a rising edge or falling edge), it is determined that the Hall sensor signal is lost or stuck.

[0039] Level 2: State Combination Legality Detection. This verifies the state combination sequence of the signals traversed by the three Hall channels within each electrical cycle. During normal forward rotation, the state combination is traversed sequentially using the 6-combination method: {001→011→010→110→100→101→001}. During reverse rotation, the state combination is traversed in the reverse sequence: {001→101→100→110→010→011→001}. Furthermore, if an illegal state such as 000 or 111 appears and its duration exceeds one sector switching cycle (i.e., 1 / 6 of the electrical cycle), it is determined to be a simultaneous multi-channel abnormality fault.

[0040] Level 3: Timing Consistency Detection. This involves statistically analyzing the coefficient of variation of the time intervals between sectors over multiple consecutive electrical cycles to identify latent faults such as Hall sensor installation misalignment and timing disorders. In one possible implementation, the specific process is as follows: (1) Taking the six state transitions of Hall signals HA, HB, and HC within one electrical cycle as the boundary, each electrical cycle is divided into six sectors, and the duration of each sector is recorded as the sector time interval. .

[0041] (2) Collect sector time interval data for no less than twelve electrical cycles continuously, and form a time series for each sector location.

[0042] (3) Calculate the mean of the time series for each sector. with standard deviation Then, the coefficient of variation corresponding to each sector is calculated. Under normal circumstances, the time of each sector should be approximately equal (at steady speed). ); Simultaneously calculate the ratio of the time intervals between adjacent sectors, i.e., the jump ratio. , The time interval between two adjacent sectors. It is used to measure the abrupt change in duration between two adjacent sectors within the same electrical cycle.

[0043] (4) When the coefficient of variation of any sector exceeds the preset threshold (e.g., greater than 30%), or the jump ratio of adjacent sectors exceeds the preset threshold (e.g., greater than 3:1), it is determined to be a Hall sensor installation offset or timing disorder fault.

[0044] Based on the combined results of the three-level verification, the fault is located to a specific phase (A / B / C). A level-one fault is output when only one channel is abnormal, and a level-two fault, which is higher, is output when two or more channels are abnormal. A level-two fault triggers the S3 jump.

[0045] In step S2, the phase-to-phase resistance consistency assessment specifically includes the following: During the motor's power-on self-test phase or low-speed standby mode (without interfering with normal motor operation), inject short-duration DC test current (e.g., 10%-20% of the rated current, lasting 100ms) sequentially between phases AB, BC, and CA. After the current reaches a steady state, synchronously collect the corresponding line voltage and steady-state current values, and calculate the DC resistance between each phase according to Ohm's law. , , This reflects the impedance of the copper wire in the winding body.

[0046] Calculate the average value of the three-phase interphase resistance. The resistance asymmetry is defined by the maximum relative deviation. :

[0047] Based on the magnitude of resistance asymmetry, there are four fault levels: normal, level one, level two, and level three. Level two and above trigger S3 jump. The greater the asymmetry, the worse the consistency of the three-phase winding circuit.

[0048] Optional, Corresponding to the normal state, Corresponding to Level 1 fault, Corresponding to level two faults, This corresponds to a level three fault.

[0049] In S2, the motor insulation condition assessment adopts a three-dimensional joint judgment mechanism, which includes three dimensions: high-frequency leakage current detection, insulation resistance estimation, and casing potential difference detection. Each dimension independently sets a three-level fault level judgment threshold.

[0050] (1) High-frequency leakage current detection: During the zero-vector stage of PWM modulation in motor operation, the high-frequency leakage current component of the motor winding to the ground terminal of the casing is extracted by the bus current sensor, and the ratio of its amplitude to the baseline (i.e., the increase factor) is calculated to reflect the high-frequency characteristic degradation of the insulation medium. Among them, the high-frequency leakage current amplitude measured under the rated operating conditions of the motor at the factory or the statistical average value of the initial operation stage at the factory is used as the baseline value.

[0051] Optionally, a rise of less than 2 times is considered a Level 1 normal state, a rise of 2-5 times is considered a Level 2 alarm state, indicating that the motor has slight water ingress / dampness, and a rise of more than 5 times is considered a Level 3 severe state, indicating that the motor has significant water ingress / severe insulation aging.

[0052] In this embodiment, the zero-vector stage detection is selected because the potential of the three-phase windings is clamped simultaneously during this stage, the winding-to-ground potential is stable, the measurement results are not affected by switch tripping, and the detection results are more accurate.

[0053] (2) Insulation resistance estimation: Similarly, during the zero vector stage of PWM modulation in the motor operation, a DC test excitation of no more than 5V U_test is applied to the motor winding. By measuring the leakage current I_leak of the winding to the ground terminal of the casing, the insulation resistance value of the winding to ground R_ins = U_test / I_leak is calculated to reflect the DC steady-state insulation level.

[0054] Optionally, R_ins>20MΩ (typical value in a dry environment) is a Level 1 normal state; 5MΩ≤R_ins≤20MΩ is a Level 2 alarm, indicating that the motor has slight water ingress / moisture; R_ins<5MΩ is a Level 3 severe state, indicating that the motor has significant water ingress / severe insulation aging.

[0055] (3) Housing potential difference detection: Detect the potential deviation between the motor housing and the reference ground to help characterize the degree of insulation deterioration and water ingress.

[0056] Optionally, a potential deviation of <20mV is considered a Level 1 normal state; a potential deviation between 20-50mV is considered a Level 2 alarm state, indicating that the motor has slight water ingress / dampness; a potential deviation >50mV is considered a Level 3 severe state, indicating that the motor has significant water ingress / severe insulation aging.

[0057] The three-dimensional joint judgment rule is as follows: when any dimension reaches level three fault, it is judged as a serious insulation abnormality; when two or more dimensions reach level two fault at the same time, it is judged as insulation deterioration concern; and in other cases, it is judged as normal insulation.

[0058] In step S3, the specific logic for making a comprehensive judgment based on the detection results of each sub-item is as follows: (1) When any of the sub-items in the three-phase current imbalance detection, Hall sensor failure detection, or phase-to-phase resistance consistency assessment reaches a level two or above fault, the corresponding electrical fault code is directly output, and the process jumps to the diagnostic result output step S6 to terminate the demagnetization detection, thus avoiding false judgments of demagnetization caused by waveform distortion due to electrical faults. Among them, the electrical fault codes include phase wire burnout, cable breakage, Hall sensor failure, and suspected inter-turn short circuit. Inter-turn short circuits can be further confirmed in S5 using the negative sequence current-assisted decoupling mechanism.

[0059] Optionally, when the three-phase current imbalance detection outputs a three-level fault level ( Furthermore, the phase-to-phase resistance consistency assessment output is normal or at level one fault (resistance asymmetry). If the Hall sensor remains at a continuous zero level, it indicates a broken output cable fault; if the three-phase current imbalance detection output is not lower than the second-level fault level (…), then… ), and the phase-to-phase resistance consistency assessment outputs a three-level fault level ( If the Hall signal is intermittent, the output phase line may be burnt out.

[0060] (2) When the insulation condition is assessed as severely abnormal, output the insulation fault code, jump to the diagnostic result output step S6, and terminate the demagnetization detection, because leakage current coupling during severe insulation degradation will make the demagnetization characteristic indicators unreliable.

[0061] (3) When the insulation condition assessment is of concern for deterioration, continue to execute the demagnetization test step S4, but add a low confidence mark to the final demagnetization assessment result given in S5, and simultaneously indicate the impact of insulation condition when outputting the diagnostic message. If the "motor is damp" mark is added, it is recommended to retest after dehumidification.

[0062] (4) When all sub-items are normal or only have a minor level one fault, confirm that the electrical circuit and insulation condition meet the demagnetization detection conditions, and proceed to demagnetization detection step S4. A high confidence mark can be added to the final demagnetization assessment result given in S5. It should be noted that when the three-phase current imbalance detection and the phase-to-phase resistance consistency assessment both reach the level one fault, it is judged as a suspected inter-turn short circuit (minor), and the fault code is output. However, S4 and S5 are still executed, and the negative sequence current-assisted decoupling mechanism is used for further confirmation.

[0063] In step S4, the back electromotive force can be obtained through either direct sampling or observer reconstruction. The two methods can be switched according to the actual motor configuration, and the subsequent harmonic extraction process is universal. For example... Figure 3 As shown, this embodiment provides a direct phase voltage sampling path A, specifically including: using PWM synchronous sampling technology to sample the three-phase winding terminal voltage at the peak or valley of the triangular carrier wave, at which time the switching action is minimal and the sampling noise is lowest; the sampled value is transmitted through the DC bus voltage. Normalization eliminates the influence of bus voltage fluctuations and improves the accuracy of FFT / Goertzel extraction. Further digital bandpass filtering suppresses PWM switching harmonic interference. The center frequency of the digital bandpass filter dynamically adjusts with the motor's current electrical fundamental frequency, always aligning with the target frequency band. In this embodiment, the center frequency of the digital bandpass filter is locked at 3 times the electrical fundamental frequency, with a bandwidth of ±10%.

[0064] like Figure 3 As shown, this embodiment also provides an online reconstruction path B for a sliding mode observer (SMO), which requires no additional phase voltage sampling circuit and utilizes only the three-phase current. , , With DC bus voltage The three-phase back electromotive force waveforms are reconstructed online using SMO in the α-β stationary coordinate system. The specific implementation principle is as follows: The voltage equation of the permanent magnet synchronous motor in the α-β stationary coordinate system is:

[0065] in , Obtained from three-phase current via Clarke transformation. , It is obtained by converting the PWM duty cycle and the DC bus voltage. , Let be the back electromotive force component to be estimated.

[0066] Using the current observation error as the sliding mode surface, a sign function-type switching control law is designed, and the current observer equation is constructed. After the observer enters the sliding mode steady state, according to the equivalent control principle, the equivalent average value of the sliding mode control quantity is equal to the back electromotive force (EMF). A low-pass filter is applied to the sliding mode control output to remove high-frequency chattering, resulting in a smooth estimate of the back EMF. , Finally, the back electromotive force components in the α-β coordinate system are subjected to the inverse Clarke transform to obtain the three-phase back electromotive force waveforms. , , .

[0067] Due to the amplitude of the third harmonic Due to natural fluctuations caused by operating parameters such as rotational speed and load, demagnetization characteristic indicators are calculated in step S4. Beforehand, adaptive normalization compensation is required: using the current motor speed ω and load current Iq as inputs, the original third harmonic amplitude A is adjusted using a preset compensation coefficient. 3rd _raw is corrected: A 3rd _norm =A 3rd _raw × f_comp(ω, Iq) is used to eliminate the natural variation in harmonic amplitude caused by operating condition fluctuations, thus ensuring that the compensated value is... It is stably distributed around the baseline value across the entire operating range.

[0068] The preset compensation coefficient f_comp(ω, Iq) can be implemented in one of the following two ways: Offline calibration lookup method: Motors under different speed and load current combinations (ω) before leaving the factory. i , Iq j ) Measure the A of a normal motor 3rd _raw(i,j), construct the compensation coefficient matrix M[i][j]= A 3rd _ref / A 3rd _raw(i,j), where A 3rd_ref is used to continuously collect the amplitude values ​​of three harmonics under stable motor operation and stable working conditions, and take the statistical average value as the initial reference. During online operation, the corresponding compensation coefficient is obtained by looking up a table according to the current working condition (ω, Iq).

[0069] Polynomial online fitting method: A two-dimensional polynomial surface fitting is performed on the calibrated compensation coefficient matrix M[i][j] to obtain the analytical compensation expression, such as: f_comp(ω, Iq) = a0+ a1·ω + a2·Iq + a3·ω·Iq + a4·ω² + a5·Iq², where a0~a5 are calibration coefficients. During online operation, the current operating condition (ω, Iq) is input to calculate the compensation coefficients in real time, saving table lookup storage space.

[0070] Table 1 shows the simulation data of adaptive normalized compensation based on the characteristics of a typical IPMSM motor, which aims to illustrate the processing logic of normalized compensation.

[0071] Table 1. Simulation test data of adaptive normalized compensation under operating conditions

[0072] like Figure 4 As shown in (a), the original demagnetization characteristic indicators (third harmonic amplitude or...) of the same healthy motor under different speed and load combinations are... The demagnetizing signal fluctuates significantly with changes in operating conditions, and is easily masked. After the above-mentioned adaptive normalization compensation for operating conditions, as follows: Figure 4 As shown in (b), the index values ​​of a healthy motor consistently converge to near the baseline value across the entire operating range (typical fluctuation range less than ±3%). Taking a typical operating point as an example: under operating condition P1, the original third harmonic amplitude is too high, but after compensation, it falls back to the baseline value; under operating condition P3, the original third harmonic amplitude is too low, but after compensation, it rises to the same baseline value. The fluctuation amplitude after compensation is controlled within a small tolerance band across the entire operating range, eliminating the natural fluctuation of harmonic amplitude caused by changes in speed and load. When the permanent magnet demagnetizes, the third harmonic amplitude after compensation at the same operating point P5 will be significantly higher than the baseline value of the healthy state, and this deviation does not disappear with the change of operating condition, clearly distinguishing it from the fluctuation of normal operating condition. This ensures that the demagnetization characteristic index has a unified judgment benchmark across the entire operating range, avoiding false alarms and false negatives under varying operating conditions, and improving the consistency and reliability of diagnostic results.

[0073] In this embodiment, to further distinguish between demagnetization faults and inter-turn short-circuit faults, a fundamental negative-sequence current-assisted decoupling mechanism is set in step S5 as a supplementary protection to the sequential strategy, specifically including the following: The amplitudes of the three-phase fundamental currents are extracted from the three-phase current signal by frequency domain transformation. With phase information Subsequently, the positive sequence current components of the three-phase fundamental current were calculated based on the symmetrical component transformation. With negative sequence current component , respectively represented as:

[0074]

[0075] Where the rotation operator .

[0076] It is the amplitude of the symmetrical component (A→B→C phase sequence) in the three-phase fundamental current, corresponding to the normal symmetrical operating state; It is the amplitude of the asymmetrical component (A→C→B phase sequence) in the three-phase fundamental current, reflecting the asymmetry of the three-phase impedance, and is the direct physical manifestation of the inter-turn short circuit.

[0077] In this embodiment, the proportion of the negative sequence current component to the positive sequence current component is calculated: When demagnetization characteristics exceed the normal range and the proportion of negative sequence current exceeds a preset threshold, such as... If a suspected inter-turn short circuit is detected, no permanent magnet demagnetization fault conclusion will be output. If the demagnetization characteristic indicators exceed the normal range and the negative sequence current ratio does not exceed the preset threshold, such as... If the condition is found to be a permanent magnet demagnetization fault, the corresponding demagnetization level will be output. Additionally, if the demagnetization characteristic indicators are within the normal range but the negative sequence current ratio exceeds a preset threshold, an electrical abnormality will be marked for observation, and the demagnetization level will not be output.

[0078] This mechanism effectively eliminates the interference of mild inter-turn short circuits on demagnetization determination, further improving the distinguishability between the two types of faults. In this embodiment, the sequential strategy in step S3 and the fundamental negative sequence current-assisted decoupling mechanism in step S5 are combined to avoid cross-coupling interference between electrical fault characteristics and demagnetization fault characteristics from the process design level, thereby improving the accuracy of motor fault diagnosis. Table 2 shows the three-phase current imbalance and demagnetization characteristic indicators corresponding to different motor fault types. Summary diagram of negative sequence current ratio and sequential strategy determination logic: Table 2. Correspondence between different motor fault types and various indicators and sequential strategies

[0079] Refer to Table 3 and Figure 5 As shown in the figure, this embodiment provides a schematic diagram of the four-level quantization of the demagnetization degree in step S5, with each level corresponding to a specific... Threshold ranges and maintenance recommendations facilitate after-sales maintenance decisions.

[0080] Table 3. Schematic diagram of four levels of demagnetization and maintenance recommendations.

[0081] The above descriptions are merely preferred embodiments of this application, and the present invention is not limited to the above embodiments. It is understood that other improvements and variations directly derived or conceived by those skilled in the art without departing from the spirit and concept of the present invention should be considered to be included within the protection scope of the present invention.

Claims

1. A method for combined online detection of electrical faults in motors and demagnetization of permanent magnets, characterized in that, The method includes: Acquire three-phase current, three-phase phase voltage, and Hall sensor signals during motor operation; Based on the signal, multiple electrical fault sub-item detections are performed in parallel, including three-phase current imbalance detection, Hall sensor failure detection, phase-to-phase resistance consistency assessment, and motor insulation status assessment. Based on the comprehensive judgment of the test results of each sub-item, if there is an electrical fault or serious insulation abnormality that reaches the preset fault level, the corresponding fault code will be directly output and the demagnetization test will be skipped; if it is confirmed that the electrical circuit and insulation condition meet the demagnetization test conditions, the demagnetization test will be entered. In the demagnetization detection process, demagnetization characteristic indicators are calculated based on the phase voltage signal or the reconstructed back electromotive force waveform. The demagnetization characteristic index is compared with the baseline value to confirm the degree of permanent magnet demagnetization. This is then integrated with the electrical fault results to output a motor fault diagnosis message.

2. The online detection method according to claim 1, characterized in that, The calculation of demagnetization characteristic indicators based on the phase voltage signal or the reconstructed back electromotive force waveform includes: A fast Fourier transform or Goertzel algorithm is performed on the three-phase phase voltage signal or back electromotive force waveform to extract the fundamental component and the third harmonic component in order to calculate the demagnetization characteristic index, which is the ratio of the third harmonic amplitude to the fundamental amplitude.

3. The online detection method according to claim 1 or 2, characterized in that, The method for obtaining the three-phase phase voltage signal includes: The three-phase winding terminal voltage is acquired using PWM synchronous sampling technology. After DC bus voltage normalization and digital bandpass filtering, the phase voltage signal is obtained. The center frequency of the digital bandpass filter is dynamically adjusted according to the current electric base frequency of the motor. The method for obtaining the back electromotive force waveform includes: Based on a sliding mode observer, the waveforms of the three-phase opposite electromotive forces are reconstructed online in a stationary coordinate system using the three-phase current and the DC bus voltage.

4. The online detection method according to claim 1, characterized in that, The Hall sensor failure detection employs a three-level progressive verification mechanism, including: The first level is edge continuity detection, which monitors whether the number of signal transitions in each Hall channel within each electrical cycle conforms to the normal pattern, in order to identify Hall sensor signal loss or jamming faults. The second level is the state combination legality detection, which verifies whether the state combination sequence of the signal traversal of each Hall channel in each electrical cycle conforms to the legal sequence of forward or reverse rotation, so as to identify multiple channel simultaneous abnormal faults. The third level is timing consistency detection, which calculates the coefficient of variation of the time interval of each sector within multiple consecutive electrical cycles to identify Hall sensor installation misalignment or timing disorder faults. Based on the combined results of the three-level verification, a Level 1 fault is output when only one channel is abnormal, and a Level 2 fault, which is of a higher level, is output when two or more channels are abnormal.

5. The online detection method according to claim 4, characterized in that, The method for timing consistency detection includes: Using the six state transitions of the Hall signal within one electrical cycle as the boundary, each electrical cycle is divided into six sectors, and the duration of each sector is recorded as the sector time interval. Collect sector time interval data for no less than twelve electrical cycles continuously, and form a time series for each sector location; The mean and standard deviation of the time series for each sector are calculated separately to obtain the coefficient of variation for each sector. The coefficient of variation is the ratio of the standard deviation to the mean. When the coefficient of variation of any sector exceeds a preset threshold, or the ratio of the time interval between adjacent sectors exceeds a preset threshold, it is determined that the Hall sensor is installed off-center or has a timing disorder fault.

6. The online detection method according to claim 1, characterized in that, The method for assessing the phase-to-phase resistance consistency includes: During the motor power-on self-test phase or low-speed standby condition, short-time DC test current is injected between each pair of the three-phase windings of the motor in sequence, and the corresponding line voltage and steady-state current values ​​are collected simultaneously. The DC resistance values ​​between each phase are then calculated. Calculate the average value of the three-phase interphase resistance. The maximum relative deviation is used to calculate the resistance asymmetry as an evaluation index, and the corresponding fault level is output by comparing it with the set threshold. The resistance asymmetry is expressed as: In the formula, , , These are the calculated DC resistance values ​​between each phase.

7. The online detection method according to claim 1, characterized in that, The motor insulation status assessment adopts a multi-dimensional joint judgment, including at least three dimensions: high-frequency leakage current detection, insulation resistance estimation, and shell potential difference detection. Each dimension independently sets a three-level fault level judgment threshold. The high-frequency leakage current detection refers to extracting the high-frequency leakage current component of the motor winding to the ground terminal of the casing during the zero vector stage of PWM modulation in motor operation, and detecting the trend of its amplitude change compared with the baseline. The insulation resistance estimation refers to applying a DC test excitation to the motor winding during the zero vector phase of PWM modulation during motor operation, and calculating the insulation resistance value of the winding to ground by measuring the leakage current of the winding to the grounding terminal of the casing. The housing potential difference detection refers to detecting the potential offset between the motor housing and the reference ground; When any one dimension reaches level three fault, it is judged as a serious insulation abnormality; when two or more dimensions reach level two fault at the same time, it is judged as insulation degradation of concern.

8. The online detection method according to claim 1, characterized in that, The logic for making a comprehensive judgment based on the test results of each sub-item includes: When any of the sub-items in the three-phase current imbalance detection, Hall sensor failure detection, or phase-to-phase resistance consistency assessment reaches a level two or above fault, the corresponding electrical fault code is output and the demagnetization detection is terminated. When the insulation condition is assessed as severely abnormal, an insulation fault code is output and the demagnetization detection is terminated. When the insulation condition assessment indicates a concern about deterioration, demagnetization testing continues, but a low-confidence flag is added to the demagnetization assessment results; When all sub-items are normal or only have a level one fault, the demagnetization test conditions are confirmed to be met, and the demagnetization test phase begins.

9. The online detection method according to claim 1, characterized in that, The demagnetization detection process also includes: The amplitude and phase information of the three-phase fundamental current are extracted from the three-phase current signal by frequency domain transformation, and the positive sequence current component and negative sequence current component of the three-phase fundamental current are calculated based on symmetrical component transformation. Calculate the proportion of the negative sequence current component to the positive sequence current component; When the demagnetization characteristic index exceeds the normal range and the negative sequence current ratio exceeds the preset threshold, it is judged as a suspected inter-turn short circuit, and no permanent magnet demagnetization fault conclusion is output. When the demagnetization characteristic index exceeds the normal range and the negative sequence current ratio does not exceed the preset threshold, it is determined to be a permanent magnet demagnetization fault, and the corresponding demagnetization degree is output.

10. The online detection method according to claim 2, characterized in that, Before calculating the demagnetization characteristic index, an adaptive normalization compensation step based on operating conditions is also included: Using the current motor speed and load current as input, the original third harmonic amplitude is corrected by a preset compensation coefficient; The preset compensation coefficient is implemented in one of the following two ways: Offline calibration lookup method: Before the motor leaves the factory, the compensation coefficient matrix is ​​calibrated under different speed and load current combinations. During operation, the corresponding compensation coefficient is obtained by looking up the table according to the current operating conditions. Polynomial online fitting method: Perform two-dimensional polynomial surface fitting on the calibrated compensation coefficient matrix to obtain the analytical compensation expression, and calculate the compensation coefficient in real time by inputting the current working condition during runtime.

Images