A fault online diagnosis method for a low-speed large-torque motor control system

Abstract

The application relates to the field of fault detection, and discloses an online fault diagnosis method for a low-speed large-torque motor control system, which comprises the following steps: collecting multi-dimensional parameters of three-phase current, DC bus voltage, motor rotor position, motor rotor speed and radiator temperature of the motor based on a sensor group; carrying out data preprocessing on the collected multi-dimensional parameters to remove noise interference; extracting fault features related to current, voltage, rotor motion and temperature from the preprocessed data; constructing a dynamic threshold based on the fault features, distinguishing non-fault parameter fluctuation from fault abnormality through multi-feature correlation analysis, realizing online fault diagnosis, and finally outputting diagnosis results containing fault types and fault warning information. Through multi-dimensional parameter collection and dynamic correlation diagnosis, the application improves the comprehensiveness and accuracy of fault diagnosis, effectively reduces the false alarm rate, and can adapt to complex working conditions such as low-speed large-torque motor load fluctuation and speed switching.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of fault detection, in particular to an online fault diagnosis method for a low-speed large-torque motor control system. BACKGROUND

[0002] As a kind of special motor, low-speed large-torque motor can directly output low-speed and high-torque without additional reduction mechanism or only with simple reduction mechanism. The core design goal of low-speed large-torque motor is to simplify the transmission system and directly match the industrial application requirements of low-speed operation and large load driving through the optimization of motor structure and control. However, the fault diagnosis of its control system faces special challenges that traditional high-speed motor systems have never encountered. The existing diagnosis methods have significant deficiencies when applied to such motors, mainly in the following aspects:

[0003] Firstly, the data collection dimension is insufficient, resulting in the omission of key fault features closely related to motor control. Existing diagnosis methods mostly collect only basic electrical quantities such as three-phase current or DC bus voltage, ignoring the multi-dimensional state parameters closely related to motor control performance under low-speed and large-torque operating modes. For example, during low-speed operation, the detection accuracy of motor rotor position directly affects the accuracy of field-oriented control. A slight deviation in the position signal will directly lead to torque fluctuations or even loss of step. If the control system does not collect and integrate the rotor position signal, it cannot distinguish whether the current anomaly is caused by winding electrical fault or position detection deviation-induced control disorder. Meanwhile, under large-torque load conditions, motor stall and overload often cause a sharp rise in temperature of power modules (such as IGBT) and heat sinks. However, traditional diagnosis strategies do not analyze temperature parameters in conjunction with electrical parameters such as current and voltage, which can easily lead to missed diagnosis of serious faults such as IGBT module burnout due to overheating, resulting in missed diagnosis. In addition, fluctuations in rotor speed signals at low frequencies often contain early mechanical fault information such as meshing faults of mechanical transmission components (such as integrated reduction mechanisms) and bearing wear. The existing single-parameter collection mode cannot effectively capture these mechanical fault features that are closely coupled with electrical operating parameters, failing to meet the needs of mechatronic diagnosis.

[0004] Secondly, the adaptability of diagnosis algorithms to complex operating conditions is poor, resulting in a high false alarm rate. Low-speed large-torque motors often work in conditions with severe load fluctuations and frequent speed switching, and their control system parameters such as current and torque will show significant non-fault fluctuations. However, existing diagnosis methods have obvious limitations: they generally use fixed threshold criteria, which cannot dynamically track changes in operating conditions, making it difficult to distinguish between transient overcurrent caused by load impact and persistent overcurrent caused by faults such as inter-turn short circuit, resulting in frequent false alarms of the control system and reducing the reliability of diagnosis and system availability.

[0005] In summary, the existing fault diagnosis method has defects in the comprehensiveness of data collection and the intelligence of diagnosis algorithm when applied to the low-speed large-torque motor control system, and it is difficult to meet the demand of high reliability control, so an online diagnosis scheme that can deeply integrate multi-dimensional state parameters and has dynamic adaptive ability is urgently needed. SUMMARY

[0006] The purpose of the present application is to provide a fault online diagnosis method for a low-speed large-torque motor control system, which solves at least one of the above technical problems.

[0007] The purpose of the present application can be achieved by the following technical solutions:

[0008] A fault online diagnosis method for a low-speed large-torque motor control system, comprising the following steps:

[0009] S1, based on a sensor group, collecting three-phase current, DC bus voltage, motor rotor position, motor rotor speed and radiator temperature of the motor;

[0010] S2, data preprocessing is performed on the multi-dimensional parameters collected in step S1 to remove noise interference;

[0011] S3, extracting fault features from the preprocessed multi-dimensional parameters, the fault features including current features, voltage features, rotor motion features and temperature features;

[0012] S4, constructing a dynamic threshold based on the fault features, and realizing fault online diagnosis through multi-feature correlation analysis;

[0013] S5, outputting fault diagnosis results, the fault diagnosis results including fault type and fault warning information.

[0014] As a further technical solution, the dynamic threshold is used to distinguish between non-fault parameter fluctuations caused by load impact and persistent parameter abnormalities caused by faults, and the multi-feature correlation analysis is used to correlate current features, voltage features, rotor motion features and temperature features to distinguish between electrical faults, mechanical faults and thermal faults.

[0015] As a further technical solution, the data preprocessing in step S2 includes filtering and dimensionless processing, the filtering processing adopts Kalman filtering or sliding average filtering to eliminate high-frequency noise in multi-dimensional parameters; the dimensionless processing adopts maximum minimum normalization or Z-score normalization to unify the numerical magnitude of multi-dimensional parameters.

[0016] As a further technical solution, the fault features in step S3 specifically include: current imbalance degree of three-phase current, voltage fluctuation coefficient of DC bus voltage, position deviation rate of motor rotor position, speed fluctuation rate of motor rotor speed, and temperature change rate of radiator temperature.

[0017] As a further technical solution, the current imbalance degree of three-phase current is calculated in the following manner: ; wherein, is the arithmetic mean of the three-phase current in one electrical cycle;

[0018] The voltage fluctuation coefficient of DC bus voltage is calculated in the following manner: ; wherein, , are the peak value and valley value of the DC bus voltage in the set time window, respectively, is the DC bus rated voltage;

[0019] The position deviation rate of motor rotor position is calculated in the following manner: ; wherein, is the actually collected rotor position, is the expected rotor position calculated from the command given by the controller;

[0020] The speed fluctuation rate of motor rotor speed is calculated in the following manner: ; is the standard deviation of the rotor speed in the set time window, is the currently given speed reference value;

[0021] The temperature change rate of radiator temperature is calculated in the following manner: ; wherein, is the temperature at the current sampling time, is the temperature at the previous sampling time, is the sampling interval.

[0022] As a further technical solution, the dynamic threshold is used to distinguish between non-faulty parameter fluctuations caused by load impact and persistent parameter abnormalities caused by faults, and the distinguishing logic is implemented in the following manner:

[0023] When any fault feature first exceeds the corresponding dynamic threshold, start a delay timer and continuously monitor the value of the fault feature;

[0024] ​If the value of the fault feature falls within the dynamic threshold before the delay timer expires, it is determined that the threshold overshoot is caused by non-fault parameter fluctuation caused by load impact, and the timer is cleared;

[0025] If the value of the fault feature continues to exceed the dynamic threshold until the delay timer expires, it is determined that the threshold overshoot is caused by persistent parameter abnormality caused by fault, and a fault diagnosis and alarm process is triggered.

[0026] As a further technical solution, the multi-feature correlation analysis in step S4 is specifically:

[0027] If the current imbalance degree is greater than the corresponding dynamic threshold, and the position deviation rate is less than or equal to the corresponding dynamic threshold, and the temperature change rate is less than or equal to the corresponding dynamic threshold, it is determined to be an electrical fault.

[0028] If the speed fluctuation rate is greater than the corresponding dynamic threshold, and the position deviation rate is greater than the corresponding dynamic threshold, and the current imbalance degree is less than or equal to the corresponding dynamic threshold, it is determined to be a mechanical fault.

[0029] If the temperature change rate is greater than the corresponding dynamic threshold, and the current imbalance degree is greater than the corresponding dynamic threshold, and the voltage fluctuation coefficient is greater than the corresponding dynamic threshold, it is determined to be a thermal fault.

[0030] As a further technical solution, the calculation formula of the dynamic threshold is: wherein, is the dynamic threshold of any fault feature, is a reference threshold, is a working condition correction factor, which is composed of key real-time working condition parameters and experimental calibration correction coefficients corresponding to the fault feature, and the specific expression is: wherein, is the number of key working condition parameters corresponding to the fault feature, , is the correction coefficient of the th working condition parameter, is the th non-dimensional key real-time working condition parameter.

[0031] The beneficial effects of the present application are:

[0032] (1) The application breaks through the single dimension limitation of traditional diagnosis data, collects motor three-phase current, DC bus voltage, rotor position, rotor speed and radiator temperature through a sensor group, covers key parameters of electrical, mechanical and thermal systems; when running at low speed, the rotor position deviation rate can distinguish whether the current anomaly is an electrical fault or a mechanical deviation; under a large torque load, the temperature change rate is linked with the electrical parameters to avoid missing diagnosis of IGBT overheating risk by only judging current overload; meanwhile, combined with current unbalance degree, speed fluctuation rate characteristics and multi-feature correlation analysis, electrical, mechanical and thermal faults can be accurately distinguished without missing key fault modes, which adapts to the multi-system coupling characteristics of the motor and ensures operation safety;

[0033] (2) The application constructs a dynamic threshold based on fault characteristics for complex working conditions such as motor load fluctuation and speed switching, integrates real-time load rate, speed and other parameters through working condition correction factors, so that the threshold is dynamically adjusted with the working condition to avoid misjudgment of fixed threshold; in combination with delay timing logic, only when the parameters continuously exceed the threshold can the fault be determined to exclude transient fluctuation interference; multi-feature correlation analysis requires multiple features to be abnormal, such as not determining a fault by single current fluctuation, which further reduces false positives. BRIEF DESCRIPTION OF DRAWINGS

[0034] The application will be further described below with reference to the accompanying drawings.

[0035] Figure 1 The method steps of the application are shown in the figure. DETAILED DESCRIPTION

[0036] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor belong to the scope of protection of the application.

[0037] Please refer to Figure 1 The application is a fault online diagnosis method for a low-speed large-torque motor control system, which comprises the following steps:

[0038] S1, based on a sensor group, collecting three-phase current, DC bus voltage, motor rotor position, motor rotor speed and radiator temperature of the motor; specifically, collecting three-phase current by a current sensor, collecting DC bus voltage by a voltage sensor, collecting rotor position by a position sensor, collecting rotor speed by a speed sensor, and collecting radiator temperature by a temperature sensor;

[0039] S2, data preprocessing is performed on the multi-dimensional parameters collected in step S1 to remove noise interference;

[0040] S3, extracting fault features from the pre-processed multi-dimensional parameters, the fault features including current features, voltage features, rotor motion features and temperature features;

[0041] S4, constructing a dynamic threshold based on the fault features, and realizing online fault diagnosis through multi-feature correlation analysis;

[0042] S5, outputting fault diagnosis results, the fault diagnosis results including fault types and fault warning information.

[0043] In this embodiment, through the steps of multi-dimensional parameter acquisition, data preprocessing, fault feature extraction, dynamic threshold diagnosis and result output, the purposes of comprehensiveness and accuracy of fault diagnosis are achieved. Specifically, in the parameter acquisition link, the limitations of traditional three-phase current or DC bus voltage acquisition are broken through, the motor rotor position, rotor speed and radiator temperature are included in the acquisition range, covering the key parameters of the electrical, mechanical and thermal three systems, providing complete data source for subsequent multi-dimensional diagnosis. When running at low speed, the rotor position signal can be associated with mechanical structure deviation, avoiding the problem that abnormal current cannot distinguish electrical or mechanical faults. Under heavy torque load, the radiator temperature can warn the risk of IGBT overheating, making up for the deficiency of traditional diagnosis in missing heat fault signals. In the data preprocessing link, noise interference is removed to ensure the reliability of subsequent feature extraction. Fault feature extraction focuses on four types of core features: current, voltage, rotor motion and temperature, providing quantitative basis for accurate diagnosis. The combination of dynamic threshold and multi-feature correlation analysis not only adapts to the changes of working conditions, but also avoids misjudgment of single feature. Finally, the fault type and warning information are output, providing clear guidance for operation and maintenance. Thus, the fault characteristics of low-speed large-torque motor electrical, mechanical and thermal coupling are effectively adapted, providing comprehensive technical support for safe and stable operation of the motor.

[0044] The dynamic threshold is used to distinguish between non-fault parameter fluctuations caused by load impact and persistent parameter abnormalities caused by faults. The multi-feature correlation analysis is used to correlate current features, voltage features, rotor motion features and temperature features to distinguish between electrical faults, mechanical faults and thermal faults.

[0045] In this embodiment, the core function of dynamic threshold and multi-feature correlation analysis is used to further improve the accuracy of fault diagnosis and the ability to distinguish fault types. For dynamic threshold, the purpose is to solve the problem that the traditional fixed threshold cannot adapt to the fluctuation of working conditions. It is clear that the goal is to distinguish between load impact non-fault fluctuation and fault persistence anomaly. When the load of the low-speed large-torque motor increases suddenly or the speed switches, the current and voltage may fluctuate instantaneously. The dynamic threshold can avoid misjudging such non-fault signals as faults and reduce unnecessary shutdown. For multi-feature correlation analysis, the corresponding relationship between current, voltage, rotor motion, temperature characteristics and fault type is established to achieve accurate division of fault types: electrical faults are only associated with current and voltage anomalies, mechanical faults are only associated with rotor motion parameter anomalies, and thermal faults are associated with temperature and electrical parameter anomalies. The problem of not being able to clearly distinguish the fault source in traditional diagnosis is solved.

[0046] The data preprocessing in step S2 includes filtering and dimensionless processing. The filtering uses Kalman filtering or moving average filtering to eliminate high-frequency noise in multi-dimensional parameters. The dimensionless processing uses maximum minimum normalization or Z-score normalization to unify the numerical magnitude of multi-dimensional parameters. In this embodiment, by limiting the specific way of data preprocessing, the key guarantee for subsequent fault feature extraction and diagnosis accuracy is provided, and the problem of diagnosis deviation caused by noise interference in the original collected data is solved.

[0047] As a further technical solution, the fault features in step S3 specifically include: current unbalance degree of three-phase current, voltage fluctuation coefficient of DC bus voltage, position deviation rate of motor rotor position, speed fluctuation rate of motor rotor speed, and temperature change rate of radiator temperature.

[0048] Among them, the three-phase current unbalance degree reflects the symmetry degree of three-phase current. The higher the unbalance degree, the more likely there is a winding inter-turn short circuit, current sensor deviation, and electrical fault. The DC bus voltage fluctuation coefficient reflects the stability of the bus voltage. If the fluctuation coefficient is too high, there may be an inverter switch fault, power supply voltage instability, and electrical fault. The motor rotor position deviation rate reflects the deviation between the actual rotor position and the ideal position. If the deviation rate is too high, there may be a gear box meshing deviation, bearing wear, and mechanical fault. The motor rotor speed fluctuation rate reflects the stability of the rotor speed. If the fluctuation rate is too high, there may be mechanical load unevenness, bearing jamming, and mechanical fault. The radiator temperature change rate reflects the rising / dropping rate of the radiator temperature. If the change rate rises sharply, there may be a radiator blockage, IGBT power module overheating, and thermal fault.

[0049] In this embodiment, by specifying the specific type of fault feature, the abstract diagnostic dimension is converted into a specific index that can be quantified and analyzed, providing a clear carrier for dynamic threshold construction and multi-feature correlation analysis, solving the problem of traditional diagnostic features being fuzzy and unable to accurately correlate faults.

[0050] Three-phase current The calculation method of the current unbalance degree of the three-phase current is as follows: ; wherein, is the arithmetic mean value of the three-phase current in one electrical cycle; specifically, , is one electrical cycle of the motor, i.e., the time for the current to complete one cycle of change;

[0051] The voltage fluctuation coefficient of the DC bus voltage The calculation method of the voltage fluctuation coefficient of the DC bus voltage is as follows: ; wherein, , are the peak value and the valley value of the DC bus voltage in the set time window, respectively, is the rated voltage of the DC bus;

[0052] The position deviation rate of the motor rotor position The calculation formula of the position deviation rate of the motor rotor position is as follows: ; wherein, is the actual collected rotor position, is the expected rotor position calculated from the command given by the controller;

[0053] The speed fluctuation rate of the motor rotor speed The calculation formula of the speed fluctuation rate of the motor rotor speed is as follows: ; is the standard deviation of the rotor speed in the set time window, is the current given speed reference value; the size of the time window needs to be matched with the motor electrical cycle or the motor speed to avoid delay in capturing the fault feature due to a too large window, or noise interference due to a too small window:

[0054] Current and voltage parameters: the time window is set to 1-2 motor electrical cycles, such as 0.02s for the motor electrical cycle, and the window is set to 0.02-0.04s to ensure complete cycle change of the current / voltage;

[0055] Rotor speed and position parameters: the time window is set to 5-10 sampling cycles to balance fluctuation capture and noise suppression;

[0056] Temperature parameters: the time window is set to 10-30s, as temperature changes slowly and a too short window cannot reflect the trend.

[0057] The temperature change rate of the radiator temperature The calculation formula is as follows:​ ; wherein, Tcurrentis the temperature at the current sampling time, Tpreviousis the temperature at the previous sampling time, Tsamplinginterval is set based on the rate of change of the parameter.

[0058] In this embodiment, by providing specific calculation formulas for each fault feature, the fault feature is converted from qualitative description to quantitative calculation index, providing standardized basis for the consistency and accuracy of the diagnosis result, solving the problem of fuzzy calculation method and large difference in results of traditional diagnostic features in different scenes; specifically, for the current unbalance degree, the difference between the maximum and minimum values of the three-phase current is divided by the periodic average value to accurately quantify the symmetry degree of the current, avoiding ignoring the influence of periodic fluctuation due to simple comparison of current values; the voltage fluctuation coefficient reflects the bus voltage stability by setting the ratio of the voltage peak-to-valley difference in the time window to the rated voltage, adapting to the diagnosis needs of motors of different voltage grades; the position deviation rate converts the position deviation into a percentage by dividing the actual and ideal position difference by the circular angle, facilitating unified judgment standard; the speed fluctuation rate eliminates the difference in speed magnitude by comparing the speed standard deviation with the reference speed, so that the speed stability under high and low speed conditions can be accurately measured; the temperature change rate reflects the temperature change trend in real time by comparing the temperature difference between adjacent sampling times with the sampling interval, avoiding missing the risk of rapid temperature rise by only looking at the absolute temperature value; the above formulas all consider the physical meaning of the parameters and the operating characteristics of the motor, and the calculation results can accurately reflect the fault state, providing accurate quantitative data for subsequent dynamic threshold adaptation and correlation analysis.

[0059] The dynamic threshold is used to distinguish between non-fault parameter fluctuations caused by load impact and persistent parameter abnormalities caused by faults, and the distinguishing logic is achieved by the following methods:

[0060] When any fault feature value exceeds the corresponding dynamic threshold for the first time, a delay timer is started, and the value of the fault feature is continuously monitored;

[0061] If the value of the fault feature falls within the dynamic threshold before the delay timer expires, it is determined that this threshold exceeding is a non-fault parameter fluctuation caused by load impact, and the timer is cleared; the falling back determination needs to meet the persistence, i.e. the fault feature needs to be stable within the threshold for 2-3 consecutive sampling periods, rather than instantaneous falling back, such as instantaneous threshold exceeding followed by immediate falling back, but only lasting for 1 sampling period, which is not determined as falling back, avoiding misjudgment caused by signal jitter;

[0062] If the value of the fault feature quantity continues to exceed the dynamic threshold until the delay timer expires, it is determined that this threshold exceeding is a persistent parameter abnormality caused by a fault, triggering the fault diagnosis and alarm process. The timeout time needs to be set based on the typical duration of low-speed high-torque motor load impact. Through experimental statistics of the maximum duration of parameter fluctuation of such motors under normal load fluctuation, such as load sudden rise from 50% to 100%, the timeout time is set to 1.2-1.5 times the maximum duration, for example, the maximum fluctuation is measured to be 2s, and the timeout time is set to 2.4-3s, ensuring that both non-fault fluctuations and fault determination are excluded.

[0063] In this embodiment, by refining the distinguishing logic of the dynamic threshold, the ability to distinguish between transient non-fault fluctuations and persistent fault abnormalities is further enhanced, the false alarm rate is reduced, and the problem of traditional diagnosis that only looks at parameter threshold exceeding and ignores fluctuation persistence is solved. The core lies in the combination of delay timing and continuous monitoring: when the fault feature quantity first exceeds the threshold, it is not immediately determined as a fault, but a delay timer is started to give the parameter a fall-back time; when the load of a low-speed high-torque motor is impacted, the current and voltage often exceed the threshold instantaneously, but will return to normal in a short time, and the above setting can avoid false triggering of alarms in such non-fault scenarios. At the same time, it is clear that the fall-back determination needs to meet the persistence to prevent misjudgment caused by signal jitter, and only when the parameter continuously exceeds the threshold until the timer expires, it is determined as a fault, ensuring that the real and persistent fault abnormality is captured; the above technical solution makes the application of dynamic threshold more operational, can accurately filter non-fault fluctuations, and only triggers alarms for real faults, ensuring that the motor operation is not disturbed by unnecessary alarms and improving the practicality and reliability of the diagnosis system.

[0064] The multi-feature correlation analysis in step S4 is specifically:

[0065] If the current imbalance degree > corresponding dynamic threshold, and the position deviation rate ≤ corresponding dynamic threshold and the temperature change rate ≤ corresponding dynamic threshold, it is determined as an electrical fault; electrical faults such as winding short circuit and inverter faults only affect the current / voltage signal and will not directly cause rotor mechanical position deviation or radiator temperature rise; if the position deviation rate exceeds the threshold, it means that the fault may be caused by mechanical structure such as bearing wear causing position deviation, and if the temperature change rate exceeds the threshold, it means that there may be a thermal fault such as IGBT overheating, so the position and temperature need to be limited to accurately determine as a pure electrical fault;

[0066] If the speed fluctuation rate > the corresponding dynamic threshold, and the position deviation rate > the corresponding dynamic threshold, and the current imbalance degree ≤ the corresponding dynamic threshold, it is determined to be a mechanical fault; a mechanical fault such as poor meshing of a gear box or bearing jamming only affects the speed / position of the rotor and does not directly cause three-phase current imbalance; if the current imbalance degree exceeds the threshold, it indicates that the fault may be caused by an electrical system such as a winding inter-turn short circuit, which causes current imbalance, so the current needs to be limited to accurately determine a pure mechanical fault;

[0067] If the temperature change rate > the corresponding dynamic threshold, and the current imbalance degree > the corresponding dynamic threshold, and the voltage fluctuation coefficient > the corresponding dynamic threshold, it is determined to be a thermal fault; thermal faults of low-speed high-torque motors such as IGBT overheating and blocked heat sinks are often caused by abnormal electrical parameters, such as increased winding loss due to current imbalance and increased inverter loss due to voltage fluctuation, so thermal faults are often accompanied by current and voltage abnormalities; if only the temperature change rate exceeds the threshold but the current and voltage are normal, it may be a temperature sensor fault, i.e., a non-thermal fault, so the current and voltage need to be limited to exclude false positives.

[0068] By clearly defining the specific determination rules of multi-feature correlation analysis, the logic of "multi-feature cooperative anomaly" is converted into executable determination conditions, the accurate differentiation of fault types is realized, and the problem of traditional diagnosis "single feature determining fault, unable to locate fault system" is solved. The determination rules closely combine the "feature correlation characteristics" of different fault types: electrical faults are only related to current abnormalities, as electrical problems such as winding short circuit and inverter failure do not directly affect the mechanical structure and temperature, so the position and temperature are limited to be normal to avoid misjudging mechanical or thermal faults as electrical faults; mechanical faults are only related to rotor motion parameter abnormalities, as bearing wear and gear box problems do not affect the electrical system, so the current is limited to be normal to exclude electrical fault interference; thermal faults require cooperative abnormalities of temperature and electrical parameters, as thermal problems such as IGBT overheating and blocked heat sinks are often caused by abnormal electrical parameters, and only temperature abnormalities may be a sensor fault, so the current and voltage are limited to exceed the threshold to avoid false positives. This rule design makes the fault type determination more rigorous, for example, only when the current imbalance degree exceeds the threshold, it is determined to be an electrical fault, and the operation and maintenance can focus on the winding and inverter; only when the speed fluctuation rate and position deviation rate exceed the threshold, it is determined to be a mechanical fault, and the bearing and gear box can be checked; when the temperature and electrical parameters exceed the threshold at the same time, it is determined to be a thermal fault, and the cooling system can be repaired. Accurate fault type differentiation greatly shortens the operation and maintenance troubleshooting time, improves fault handling efficiency, and reduces motor downtime losses.

[0069] The calculation formula of the dynamic threshold is: wherein, is the dynamic threshold of any fault feature, is the benchmark threshold, which is the maximum value of the non-fault parameter of the corresponding fault feature under standard working conditions, measured by experiments; is the working condition correction factor, which is composed of the key real-time working condition parameters of the corresponding fault feature and the correction coefficients calibrated by experiments, and the specific expression is: , is the number of key working condition parameters of the corresponding fault feature, , is the correction coefficient of the th working condition parameter, calibrated by multiple sets of corresponding working condition experiments, is the th non-dimensional key real-time working condition parameter.

[0070] , , , The specific definitions of the above are as follows:

[0071] When is the of three-phase current imbalance, , is the maximum value of the non-fault current imbalance under the rated speed and rated load, , is the load correction coefficient, , , and Tcurr is the real-time output torque, and Trated is the rated torque;

[0072] is the of DC bus voltage fluctuation, , is the maximum value of the non-fault voltage fluctuation under the rated speed and no-load fluctuation, , is the speed correction coefficient, , is the real-time rotor speed, is the rated rotor speed;

[0073] When is the of motor rotor position deviation rate, , is the maximum value of the non-fault position deviation rate under the rated torque and low-speed stable operation, , is the torque correction coefficient, , is the torque division threshold, determined by stable working condition experiments, and when ≤ , ;

[0074] When is the maximum value of the non-fault fluctuation rate of the motor rotor speed under the rated speed and without mechanical impact, , is the maximum value of the non-fault fluctuation rate of the speed under the rated speed and without mechanical impact, is the speed adaptation coefficient, , and is the low-speed base coefficient, , which are calibrated through the speed working condition experiment;

[0075] When is the maximum value of the non-fault fluctuation rate of the radiator temperature under the rated speed and without mechanical impact, , is the maximum value of the non-fault fluctuation rate of the speed under the rated speed and without mechanical impact, is the load temperature correction coefficient, is the environmental temperature correction coefficient, wherein Tenv is the real-time environmental temperature, and Tstd is the standard environmental temperature.

[0076] In the embodiment, through the design of the reference threshold value + working condition correction factor, the basic determination standard of different fault characteristics is retained, and the key parameters of the real-time load rate, speed and environmental temperature are integrated through the working condition correction factor; when the load rate changes, the correction factor of the current imbalance degree can adjust the threshold value, so as to avoid the instantaneous overcurrent misjudgment caused by the sudden increase of the load; when the speed deviation, the correction factor of the voltage fluctuation coefficient can adapt the threshold value, so as to exclude the non-fault fluctuation of the speed switching; when the environmental temperature changes, the correction factor of the temperature change rate can adjust the threshold value, so as to avoid the normal temperature fluctuation misjudgment caused by the environmental temperature rise; at the same time, the introduction of the correction coefficient and the dimensionless working condition parameter makes the formula adaptable to different types of motors, and only needs to calibrate the reference threshold value and the correction coefficient through the experiment, so as to be applied to the low-speed large-torque motors with different rated torques and speeds, without the need to redesign the threshold value logic.

[0077] It should be noted that the calculation formula and the parameters participating in the operation in the application are all pre-processed by dimensionless processing, and the process of dimensionless processing is known in the industry, which is not described here.

[0078] The above describes one embodiment of the application in detail, but the content described is only the preferred embodiment of the application, and cannot be considered as limiting the scope of the implementation of the application. Any equivalent changes and improvements made in the scope of the application should still belong to the patent coverage range of the application.​​​​​​

Claims

1. A method for on-line fault diagnosis of a low-speed high-torque motor control system, characterized in that, Comprising the following steps: S1, based on the sensor group collecting the three-phase current of the motor, the DC bus voltage, the motor rotor position, the motor rotor speed and the radiator temperature; S2, data preprocessing is performed on the multi-dimensional parameters collected in step S1 to remove noise interference; S3, extracting fault features from the preprocessed multi-dimensional parameters, the fault features including current features, voltage features, rotor motion features and temperature features; S4, constructing a dynamic threshold based on the fault features, and realizing online fault diagnosis through multi-feature correlation analysis; S5, outputting the fault diagnosis result, which includes fault type and fault warning information; The fault features in step S3 specifically include: current unbalance degree of three-phase current, voltage fluctuation coefficient of DC bus voltage, position deviation rate of motor rotor position, speed fluctuation rate of motor rotor speed and temperature change rate of radiator temperature; Three-phase current The current unbalance degree of the three-phase current is calculated as follows: ; wherein, is the arithmetic mean value of the three-phase current in one electrical cycle.​ Voltage fluctuation coefficient of direct current bus voltage The calculation method is as follows: ; wherein, , are the peak value and the valley value of the direct current bus voltage in the set time window respectively, is the direct current bus rated voltage; Position deviation rate of motor rotor position The calculation formula is: ; wherein, is the actual collected rotor position, is the expected rotor position calculated from the command given by the controller; Speed fluctuation rate of motor rotor speed The calculation formula is: ; is the standard deviation of rotor speed in the set time window, is the current given speed reference value; temperature rate of change of the temperature of the heat sink The calculation formula is: ; wherein, Tn is the temperature at the current sampling moment, Tn-1 is the temperature at the previous sampling moment, is the sampling interval.

2. The method for online fault diagnosis of low-speed high-torque motor control system according to claim 1, characterized in that, The dynamic threshold is used to distinguish between non-fault parameter fluctuations caused by load impact and persistent parameter abnormalities caused by faults, and the multi-feature correlation analysis is used to correlate current features, voltage features, rotor motion features and temperature features to distinguish between electrical faults, mechanical faults and thermal faults.

3. The method for online fault diagnosis of low-speed high-torque motor control system according to claim 1, characterized in that, The data preprocessing in step S2 includes filtering and dimensionless processing, the filtering uses Kalman filtering or moving average filtering to eliminate high-frequency noise in multi-dimensional parameters, and the dimensionless processing uses maximum minimum normalization or Z-score normalization to unify the numerical magnitude of multi-dimensional parameters.

4. The method for online fault diagnosis of low-speed high-torque motor control system according to claim 1, characterized in that, The dynamic threshold is used to distinguish between non-fault parameter fluctuations caused by load impact and persistent parameter abnormalities caused by faults, and the distinguishing logic is realized by the following way: When any fault feature value exceeds the corresponding dynamic threshold for the first time, start a delay timer and continuously monitor the value of the fault feature; If the value of the fault feature falls within the dynamic threshold before the delay timer expires, it is determined that this threshold exceeding is a non-fault parameter fluctuation caused by load impact, and the timer is cleared; If the value of the fault feature continues to exceed the dynamic threshold until the delay timer expires, it is determined that this threshold exceeding is a persistent parameter abnormality caused by fault, triggering the fault diagnosis and alarm process.

5. The method for online fault diagnosis of low-speed high-torque motor control system according to claim 1, wherein, The multi-feature correlation analysis in step S4 specifically includes: If the current unbalance degree is greater than the corresponding dynamic threshold, and the position deviation rate is less than or equal to the corresponding dynamic threshold, and the temperature change rate is less than or equal to the corresponding dynamic threshold, it is determined as an electrical fault; If the speed fluctuation rate is greater than the corresponding dynamic threshold, and the position deviation rate is greater than the corresponding dynamic threshold, and the current unbalance degree is less than or equal to the corresponding dynamic threshold, it is determined as a mechanical fault; If the temperature change rate is greater than the corresponding dynamic threshold, and the current unbalance degree is greater than the corresponding dynamic threshold, and the voltage fluctuation coefficient is greater than the corresponding dynamic threshold, it is determined as a thermal fault.

6. The method for online fault diagnosis of low-speed high-torque motor control system according to claim 4, characterized in that, The calculation formula of the dynamic threshold is: Wherein, is the dynamic threshold of any fault feature, is the baseline threshold, is the working condition correction factor, which is composed of the key real-time working condition parameters corresponding to the fault feature and the correction coefficient calibrated by experiments, and the specific expression is: Wherein, is the number of key working condition parameters corresponding to the fault feature, , is the correction coefficient of the first working condition parameter, is the first non-dimensional key real-time working condition parameter. ​

Images