Fusion diagnosis and hierarchical closed-loop control method with vibration acquisition embedded in motor controller and motor controller

Abstract

The application discloses a kind of motor controller embedded vibration acquisition fusion diagnosis and hierarchical closed-loop control method and motor controller.The method comprises: vibration acquisition circuit is embedded in the main control PCB board inside motor controller, vibration acquisition circuit includes three-axis vibration sensor and its peripheral circuit;Synchronous acquisition electrical parameter of motor and the original vibration signal obtained by vibration sensor;Based on electrical parameter, electromagnetic vibration mathematical model is established, electromagnetic vibration component is calculated, and electromagnetic vibration component is eliminated from original vibration signal, to obtain pure mechanical vibration signal;From pure mechanical vibration signal, vibration feature is extracted, and fusion analysis is carried out in combination with electrical parameter, to diagnose motor fault type;According to the fault type diagnosed and its severity, corresponding hierarchical closed-loop control strategy is executed.The application realizes the intelligent diagnosis and hierarchical closed-loop active protection of motor early fault by migrating vibration acquisition node to controller end.

Description

Technical Field

[0001] This invention belongs to the field of motor control and fault diagnosis technology, and more specifically, relates to a method for fusion diagnosis and hierarchical closed-loop control of vibration acquisition embedded in a motor controller, and a motor controller. Background Technology

[0002] With the booming development of the low-altitude economy, electric aircraft, as a new type of green transportation, are becoming a research hotspot in the aviation field. Permanent magnet synchronous motors (PMSMs), with their core advantages such as high power density, stable power output, high energy efficiency, and wide applicability, have become the core power source for electric aircraft and a key component of the low-altitude aircraft's power system, directly determining the aircraft's flight stability, range, and safety. In low-altitude flight operations, electric aircraft are subjected to harsh conditions of variable speed, variable load, and complex airflow interference, making PMSMs highly susceptible to operational abnormalities. Therefore, precise condition monitoring and fault diagnosis methods are urgently needed to ensure flight safety.

[0003] Currently, the technical solution of using permanent magnet synchronous motors as the core power component and paired with a high-voltage power distribution system has significantly improved the power response speed, transmission efficiency, and overall integration of electric aircraft, effectively reducing the aircraft's weight and manufacturing cost. This is the mainstream development trend for low-altitude electric aircraft. However, as the core power source of the entire aircraft, if the permanent magnet synchronous motor experiences hidden faults such as rotor eccentricity, bearing wear, electromagnetic anomalies, or structural loosening, it can directly cause serious problems such as power vibration, output instability, and power interruption, which can easily lead to safety accidents such as loss of control and crashes.

[0004] In the field of motor condition monitoring and fault diagnosis, existing technologies have a certain research foundation. For example, existing patents use multi-source data such as rotor surface temperature, stator coil temperature, resistance, voltage, bearing vibration frequency, and speed of permanent magnet synchronous motors for fault diagnosis; other solutions use data-driven methods to collect three-phase current and radial vibration data of the motor; still others analyze the spectral characteristics of stator or winding strain signals to achieve synchronous monitoring of vibration and temperature rise. Furthermore, some research focuses on synchronous motor fault diagnosis methods based on multi-source electromechanical information fusion, identifying faults by synchronously collecting phase voltage, rotor vibration, and stator vibration signals. In recent years, a few attempts have been made to directly integrate vibration sensors inside the motor housing to obtain more realistic mechanical vibration signals. However, such solutions are limited by the extreme high temperature, high speed, and strong alternating electromagnetic field environment inside the motor; the sensor chip is easily damaged, and the signal transmission path is difficult, resulting in very few engineering applications and failing to meet the stringent reliability requirements of aircraft.

[0005] However, the aforementioned existing technologies still have the following shortcomings: First, external sensors are susceptible to signal interference. Traditional motor monitoring methods mostly involve post-incident troubleshooting and external testing. Conventional vibration sensors need to be directly fixed to the motor housing or bearing housing, and the signals they collect contain a large amount of "electrical vibration" components caused by stator cogging pulsation and electromagnetic force waves. At the same time, external sensors require independent power supply lines and signal cables, which are prone to conducted and radiated interference in the complex electromagnetic environment of airborne applications. This results in a poor signal-to-noise ratio, making it difficult to use for accurate diagnosis. It also suffers from defects such as low accuracy, strong hysteresis, and difficulty in identifying early micro-faults.

[0006] Second, the coupling of electrical and mechanical vibrations leads to misdiagnosis of faults. A strong electromagnetic attraction exists between the stator and rotor of a permanent magnet synchronous motor. The radial component of this electromagnetic force causes stator deformation, which in turn causes mechanical vibration of the stator when the rotor rotates. In existing technologies, a single vibration signal cannot effectively isolate the coupling interference between electrical and mechanical characteristics, easily leading to misdiagnosis and missed diagnosis of bearing damage and electromagnetic faults. Furthermore, while existing data-driven fusion methods can handle multi-source signals, they often involve complex models and high computational demands, making it difficult to achieve real-time electrical vibration isolation and fault diagnosis under the limited computing resources of airborne embedded controllers, thus restricting their engineering deployment in aerospace electric propulsion systems.

[0007] Third, "post-incident alarm" protection lacks proactive intervention capabilities. Existing motor fault diagnosis methods mostly remain at the level of fault identification and alarm, lacking an active closed-loop protection mechanism that classifies and regulates according to the severity of the fault, and thus cannot proactively intervene in the early stages of a fault to avoid sudden power failure.

[0008] To address the aforementioned industry pain points, this invention proposes a fusion diagnosis and hierarchical closed-loop control method for vibration acquisition embedded in a motor controller. By migrating the vibration acquisition node to the controller, and utilizing the controller's stable electromagnetic shielding environment and power ground plane, conducted and radiated interference introduced by high-voltage circuits is fundamentally blocked, effectively eliminating electrical vibration noise. Based on this, high-precision real-time acquisition of the pure mechanical vibration signal of the permanent magnet synchronous motor is performed, combined with electrical parameters for joint analysis, achieving intelligent diagnosis of early motor faults and hierarchical closed-loop active protection. Summary of the Invention

[0009] The purpose of this invention is to propose a method for fusion diagnosis and hierarchical closed-loop control of a motor controller with embedded vibration acquisition, and to address the technical problems in existing technologies, such as external vibration sensors leading to susceptibility to electromagnetic interference, coupling of electrical and mechanical vibrations resulting in misdiagnosis of faults, and lack of hierarchical active protection capabilities. By embedding the vibration acquisition circuit inside the motor controller and establishing an electromagnetic vibration mathematical model for electrical vibration decoupling, combined with multi-dimensional fusion analysis of electrical parameters and vibration characteristics, accurate diagnosis of early motor faults and hierarchical closed-loop protection can be achieved.

[0010] To achieve the above objectives, in a first aspect, the present invention proposes a method for fusion diagnosis and hierarchical closed-loop control of vibration acquisition embedded in a motor controller, comprising: The vibration acquisition circuit is embedded in the main control PCB board inside the motor controller. The vibration acquisition circuit includes a triaxial vibration sensor and its peripheral circuit. The electrical parameters of the motor and the raw vibration signals obtained by the triaxial vibration sensor are collected synchronously. An electromagnetic vibration mathematical model is established based on the electrical parameters, electromagnetic vibration components are calculated, and electromagnetic vibration components are removed from the original vibration signal to obtain a pure mechanical vibration signal. Vibration features are extracted from the purely mechanical vibration signal and combined with the electrical parameters for fusion analysis to diagnose motor fault types; Based on the diagnosed fault type and its severity, implement the corresponding hierarchical closed-loop control strategy.

[0011] Optionally, the expression for the electromagnetic vibration mathematical model is:

[0012] in, Electromagnetic vibration components, The electromagnetic excitation frequency is the main frequency. , This represents the number of pole pairs of the motor. Provides the base frequency for stator power supply; The amplitude of the kth harmonic of the stator current is collected in real time; The electromagnetic-vibration coupling coefficient corresponding to the kth harmonic; The initial phase angle of the corresponding harmonic is given by N, the total harmonic order is given by N, and t is the time.

[0013] Optionally, the calculation expression for the purely mechanical vibration signal is:

[0014] in, It is a purely mechanical vibration signal. The original vibration signal, It is an electromagnetic vibration component.

[0015] Optionally, the vibration characteristics include at least the effective value of mechanical vibration (RMS), kurtosis (K), and peak factor (CF), which are calculated according to the following formulas:

[0016] in, The first of the purely mechanical vibration signals One sample point, The mean value of the purely mechanical vibration signal. This represents the total number of samples.

[0017] Optionally, the motor fault types include bearing failure, rotor eccentricity, base looseness, and electrical overload; The criteria for determining bearing failure are: the three-phase current balance in the electrical parameters is less than or equal to 5%, and the kurtosis in the vibration characteristics is... or peak factor And the duration of the above features is greater than or equal to 2.5s; The criteria for determining rotor eccentricity fault are: the ratio of instantaneous speed fluctuation to rated speed is greater than 3%, the ratio of low-frequency harmonic amplitude corresponding to the rotational frequency to fundamental current amplitude is greater than 5%, the rate of change of effective mechanical vibration with respect to speed is greater than the upper limit of normal rate of change, and the duration of the above characteristics is greater than or equal to 2.5s. The criteria for determining the base loosening fault are: the three-phase current balance in the electrical parameters is less than or equal to 5%, and the effective value of the mechanical vibration in the vibration characteristics is... And the absolute value of the difference between the kurtosis value at the current time and the kurtosis value at the previous time. Furthermore, the above characteristics are satisfied for a duration greater than or equal to 2.5 seconds; The criteria for determining the electrical overload fault are: the effective value of the stator current. And the effective value of mechanical vibration in the vibration characteristics And steepness And the duration of the above features is greater than or equal to 2.5s; in, The reference mechanical vibration effective value for motor calibration under no-load conditions. This is the rated current.

[0018] Optionally, the hierarchical closed-loop control strategy includes: Level 1 warning: When Record and report faults, without limiting motor power; Secondary torque limiting: when Furthermore, if all the judgment conditions for any of the aforementioned motor fault types are met, the upper limit of the output torque will be locked at 70% of the rated torque; Level 3 lockdown shutdown: When And when the duration is greater than or equal to 2.5s, the drive output of the power device is blocked, causing the motor to stop.

[0019] Optionally, the triaxial vibration sensor maintains an electromagnetic isolation distance of greater than or equal to 10 mm from the power devices and bus capacitors within the motor controller.

[0020] Optionally, the triaxial vibration sensor is powered by a two-stage LDO regulator. The first stage is stepped down to 5V by the controller's auxiliary power supply, and the second stage steps down to output a 3.3V power supply voltage.

[0021] Optionally, the X-axis of the triaxial vibration sensor is arranged parallel to the radial direction of the motor, the Y-axis is arranged perpendicular to the radial direction of the motor, and the Z-axis is arranged along the axial direction of the motor.

[0022] Secondly, the present invention provides a motor controller, comprising: Main control PCB circuit board; The vibration acquisition circuit, integrated on the main control PCB circuit board, includes a triaxial vibration sensor and its peripheral circuitry, used to acquire the vibration signal of the motor. An electrical sampling unit is used to collect the electrical parameters of the motor; The main control MCU is electrically connected to both the vibration acquisition circuit and the electrical sampling unit. The PWM drive unit, controlled by the main control MCU, is used to drive power devices; The main control MCU is configured to execute the fusion diagnosis and hierarchical closed-loop control method for embedded vibration acquisition in the motor controller as described in any of the first aspects.

[0023] The beneficial effects of this invention are as follows: By embedding the vibration acquisition circuit within the main control PCB board of the motor controller and forming a complete acquisition front end with a triaxial vibration sensor and its peripheral circuitry, the transmission path of the vibration signal is significantly shortened from a physical perspective. This eliminates radiation and conduction interference introduced by the independent power supply of external sensors and long-distance signal cables. Simultaneously, relying on the inherent electromagnetic shielding effect of the controller's metal casing, the original vibration signal possesses a high signal-to-noise ratio at its source, laying a reliable data foundation for subsequent accurate diagnosis. By synchronously acquiring motor electrical parameters and constructing an electromagnetic vibration mathematical model, the electromagnetic vibration components are calculated in reverse using real-time current harmonic amplitude and voltage phase, accurately removing them from the original vibration signal. This effectively isolates electrical vibration noise unrelated to mechanical faults, solving the long-standing problem of fault characteristics caused by the coupling of electrical and mechanical vibrations. This paper addresses the technical challenges of misjudging between confusion, bearing damage, and electromagnetic faults. By extracting multi-dimensional vibration characteristics such as effective mechanical vibration value, kurtosis, and peak factor from pure mechanical vibration signals and combining them with electrical parameters for fusion analysis, it achieves accurate identification and classification of various faults, including bearing faults, rotor eccentricity, base loosening, and electrical overload, significantly reducing the risk of misjudgment caused by monitoring a single physical quantity. By implementing a hierarchical closed-loop control strategy based on the fault type and its severity, it records alarms only for minor faults, automatically limits load reduction for moderate faults, and shuts down the aircraft for severe faults. This avoids sudden power interruptions and prevents unnecessary derating caused by false alarms, forming an integrated closed-loop system from perception and diagnosis to active protection. This significantly improves the operational safety and mission reliability of the electric aircraft's power system and has significant engineering application value and promotion prospects.

[0024] The system of the present invention has other features and advantages that will be apparent from or will be set forth in detail in the accompanying drawings and following detailed description, which together serve to explain the particular principles of the invention. Attached Figure Description

[0025] The above and other objects, features and advantages of the present invention will become more apparent from the accompanying drawings, in which like reference numerals generally denote like parts.

[0026] Figure 1 A flowchart illustrating the steps of the fusion diagnosis and hierarchical closed-loop control method for embedding vibration acquisition in a motor controller according to Embodiment 1 of the present invention is shown.

[0027] Figure 2 A schematic diagram of a motor controller according to Embodiment 2 of the present invention is shown.

[0028] Figure 3 A schematic diagram of a two-stage LDO regulated power supply circuit according to Embodiment 2 of the present invention is shown.

[0029] Figure 4 A flowchart of the electro-oscillatory decoupling and feature extraction algorithm according to Embodiment 2 of the present invention is shown.

[0030] Figure 5 A fault classification and closed-loop control logic diagram according to Embodiment 2 of the present invention is shown. Detailed Implementation

[0031] The invention will now be described in more detail with reference to the accompanying drawings. While preferred embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the invention will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

[0032] Example 1

[0033] like Figure 1 As shown, a method for fusion diagnosis and hierarchical closed-loop control of motor controller with embedded vibration acquisition includes: S1. The vibration acquisition circuit is embedded in the main control PCB board inside the motor controller. The vibration acquisition circuit includes a triaxial vibration sensor and its peripheral circuit. Specifically, embedding the vibration acquisition circuit within the main control PCB board of the motor controller means transferring the vibration sensing function from the traditional external sensor method on the surface of the motor housing to the inside of the motor controller, directly integrating it into the main control PCB board as a circuit. The vibration acquisition circuit includes a triaxial vibration sensor and its peripheral circuitry, which together constitute a complete vibration signal acquisition front end. The motor controller housing and the motor body are rigidly connected, or the motor controller and the motor body adopt an integrated structure. The mechanical vibration generated by the motor operation is transmitted to the inside of the motor controller through the rigid connection or integrated structure. For example, the controller housing is rigidly fixed to the motor housing by bolts, or the controller and the motor share the same housing structure, so that the mechanical vibration generated by the motor operation can be effectively transmitted to the main control PCB board inside the controller through a rigid transmission path, and then collected by the triaxial vibration sensor integrated on the PCB board.

[0034] In terms of overall layout, the vibration acquisition circuit is implemented as an independent sub-PCB board. This sub-PCB board centrally arranges the triaxial vibration sensor chip and all peripheral components on the same circuit board, forming a highly integrated acquisition module, which is electrically connected and mechanically fixed to the main control PCB via pin headers. The modular design brings three advantages: easy maintenance and replacement without replacing the entire main control PCB; reduced complexity of the main control PCB layout, reducing board area and wiring density; and easy future upgrades, requiring only modification of the sub-board design when changing sensor models. This sub-PCB board is installed in the non-power heat dissipation area of ​​the main control PCB, naturally separating it from power devices such as IGBTs and bus capacitors that generate a lot of heat and carry high current, thus preventing heat generated by high-power devices from being conducted to the sensor and affecting its measurement accuracy and long-term stability.

[0035] The entire vibration acquisition circuit is completely embedded within the metal housing of the motor controller. The controller uses a sealed metal casing, which structurally serves as electromagnetic shielding, forming a shielding layer against electromagnetic radiation generated by the internal circuitry and isolating it from electromagnetic interference from the external environment. The vibration acquisition circuit, housed within this metal housing, is isolated from external sources of strong electromagnetic interference such as the motor housing, power cables, and control cables, fundamentally avoiding radiation interference introduced by long cables in external sensor solutions. Furthermore, the vibration acquisition circuit requires no independent power supply or external signal cables. The controller retains only the original power input, motor drive output, and communication bus interface. The number of external wiring harnesses has not increased due to the addition of vibration monitoring functionality, contributing to the overall lightweight design and improved reliability of the aircraft.

[0036] The peripheral circuitry of the triaxial vibration sensor is a crucial component ensuring its normal operation and signal integrity. It primarily includes a power supply filter circuit and an SPI signal filter circuit. The power supply filter circuit suppresses ripple and noise in the power supply. At the output of the second-stage LDO, a combined decoupling filter network consisting of a 10µH inductor and a 1µF tantalum capacitor, a 10µF ceramic capacitor, and a 1µF ceramic capacitor is connected in series to the power supply pin. The tantalum capacitor suppresses low-frequency power supply ripple and provides transient energy reserves, while the multilayer ceramic capacitors achieve rapid decoupling of high-frequency noise. The inductor and capacitors form an LC low-pass filter to further attenuate high-frequency noise. All decoupling capacitors are placed close to the sensor's power supply pin to minimize loop area and reduce parasitic inductance. The SPI signal filtering circuit is used to suppress high-frequency noise and signal reflection on the communication line. The SPI bus (including SDI, SDO, SCLK, and chip select signal lines) is connected to the general-purpose I / O port of the main control MCU through a series 10Ω resistor. On the one hand, it forms an RC low-pass filter with the parasitic capacitance of the PCB traces to suppress high-frequency noise and spike pulses; on the other hand, it acts as a source impedance matching resistor to absorb overshoot and ringing during signal transmission, improving the edge quality of the communication signal. All SPI bus signal lines are routed away from high-current paths in the power loop to avoid introduced conducted interference through PCB trace coupling. Through the coordinated design of power supply filtering and SPI signal filtering, the peripheral circuit comprehensively ensures the working stability and data accuracy of the triaxial vibration sensor in the high-noise environment inside the motor controller from both the power supply and signal ends.

[0037] The aforementioned integrated design enables the vibration acquisition circuit to be spatially separated from the high-voltage circuit inside the controller. In terms of electrical connection, modular assembly is achieved through the use of daughterboards and pin headers. In terms of electromagnetic compatibility, embedded protection is achieved by leveraging the shielding effectiveness of the metal casing. Together, these three elements constitute a complete vibration acquisition solution that requires no external sensors, no independent power supply, and no external signal lines.

[0038] In this step, the triaxial vibration sensor maintains an electromagnetic isolation distance of greater than or equal to 10 mm from the power devices and bus capacitors in the motor controller.

[0039] Specifically, within the motor controller, power devices (such as IGBTs) and bus capacitors are the main sources of electromagnetic interference. IGBTs generate drastic voltage and current changes during high-speed switching, creating strong electric and magnetic field radiation. Bus capacitors, carrying high-voltage, high-current charging and discharging circuits, also radiate wide-spectrum electromagnetic noise. Triaxial vibration sensors (such as the ADXL312) output weak digital signals and are extremely sensitive to external electromagnetic fields. If the triaxial vibration sensor is placed too close to these strong interference sources, electromagnetic noise can couple into the sensor's signal path through spatial radiation or conduct through the PCB layers, interfering with the sensor's power supply and SPI signal lines. This results in a large amount of electromagnetic noise unrelated to mechanical vibration being mixed into the acquired raw vibration signal, severely degrading signal quality.

[0040] The establishment of the 10mm quantitative indicator is based on engineering considerations of the spatial attenuation characteristics of electromagnetic fields. The intensity of electromagnetic interference decreases approximately inversely proportional to the square with increasing distance. A physical spacing of 10mm is sufficient to attenuate the near-field radiated interference generated by power devices and bus capacitors to a level that does not affect the normal operation of the triaxial vibration sensor. In terms of specific layout implementation, this 10mm isolation spacing is ensured through the following measures: the vibration acquisition circuit is arranged in the non-power heat dissipation area of ​​the main control PCB as an independent sub-PCB, physically avoiding the heat radiation area of ​​the power devices and the strong electromagnetic field area; the main control PCB has designated no-layout areas during the layout design phase to ensure sufficient isolation between the power devices and the triaxial vibration sensor and its peripheral circuits; the power supply lines and SPI signal lines of the triaxial vibration sensor are kept away from the high-current paths of the power loop during wiring to avoid introduced conducted interference through PCB trace coupling. The entire vibration acquisition circuit is completely enclosed inside the metal housing of the motor controller, utilizing the electromagnetic shielding effectiveness of the housing itself to further attenuate spatial radiated interference.

[0041] By using the above-mentioned electromagnetic isolation spacing of ≥10mm, the interference path of the high-voltage circuit to the weak signal of the triaxial vibration sensor is physically cut off, enabling the triaxial vibration sensor to collect pure original vibration signals in a high-noise motor controller environment, providing a reliable data foundation for subsequent electro-vibration decoupling and accurate fault diagnosis.

[0042] In this step, the triaxial vibration sensor is powered by a two-stage LDO regulator. The first stage is stepped down to 5V by the controller's auxiliary power supply, and the second stage steps down to output a 3.3V power supply voltage.

[0043] Specifically, within the motor controller, the auxiliary power supply typically provides a 12V or 24V DC voltage, while the rated operating voltage of a triaxial vibration sensor (such as the ADXL312) is 3.3V. If the 12V voltage is directly stepped down to 3.3V through a single-stage LDO (low dropout linear regulator), the LDO will withstand a voltage drop of up to 8.7V. Its power supply rejection ratio will significantly decrease at high frequencies, making it easier for high-frequency noise generated by the switching action of the power devices inside the controller to penetrate the LDO and couple to the sensor power supply. Simultaneously, the single-stage LDO has limited ability to suppress input voltage ripple under high voltage drop conditions. If the abundant switching frequency harmonics on the motor controller bus side are not sufficiently suppressed before entering the sensor power supply, they will directly degrade the reference voltage accuracy of the sensor's internal ADC, thereby affecting the accuracy of vibration signal acquisition.

[0044] The two-stage LDO voltage regulation scheme effectively solves the above problems through a step-by-step voltage reduction approach. The first-stage LDO steps down the controller's auxiliary power supply (typically 12V) to 5V, undertaking the main voltage conversion task and sharing the high voltage difference of up to 7V. The second-stage LDO further steps down the 5V to 3.3V, at which point the voltage difference is only 1.7V. Operating under such a small voltage difference, the second-stage LDO can achieve its optimal power supply rejection ratio (PSRR) performance, performing secondary filtering of residual power supply ripple and noise from the previous stage. After cascading the two LDOs, the overall power supply ripple rejection capability is equivalent to the superposition of two PSRR stages, which can attenuate the wide-spectrum noise at the input to an extremely low level, providing a clean power supply environment for the sensor.

[0045] In terms of specific circuit implementation, the first-stage LDO is selected with sufficient input voltage withstand capability and output current capability to meet the requirements of all subsequent loads, converting the 12V auxiliary power supply to 5V. The second-stage LDO is selected with high PSRR, low noise, and high precision, specifically designed to power the triaxial vibration sensor, outputting a high-precision, low-temperature-drift 3.3V supply voltage to ensure that the sensor's sensitivity, bias voltage, and internal ADC reference voltage remain stable across the entire temperature range. In terms of PCB layout, both LDO stages are placed close to the sensor to shorten the power supply loop and reduce noise coupling introduced by parasitic inductance of the traces.

[0046] Furthermore, at the output of the second-stage LDO, a combined decoupling filter network consisting of a 10µH inductor, a 1µF tantalum capacitor, a 10µF ceramic capacitor, and a 1µF ceramic capacitor is connected in series to the power supply pin. The tantalum capacitor provides energy storage and ripple suppression in the low-frequency range, while the multilayer ceramic capacitor provides fast decoupling in the high-frequency range. The inductor and capacitors form an LC low-pass filter to further attenuate high-frequency noise. All decoupling capacitors are placed close to the sensor power supply pin to minimize loop area and reduce parasitic inductance.

[0047] Through the above-mentioned two-stage LDO step-by-step voltage regulation and combined decoupling filtering design, this invention cuts off the transmission path of strong electrical noise inside the controller into the sensor power supply rail from the power supply level, enabling the triaxial vibration sensor to obtain a clean and stable power supply voltage in a high-noise motor controller environment, providing a reliable power guarantee for high-precision vibration signal acquisition.

[0048] In this step, the X-axis of the triaxial vibration sensor is arranged parallel to the radial direction of the motor, the Y-axis is arranged perpendicular to the radial direction of the motor, and the Z-axis is arranged along the axial direction of the motor.

[0049] Specifically, the X-axis is arranged parallel to the motor's radial direction. The motor's radial direction refers to the direction perpendicular to the motor's rotation axis, that is, the radial direction from the motor's shaft center to the outer circumference of the casing. In actual operation, rotor mass imbalance is one of the main factors causing radial vibration. When the rotor has mass eccentricity, periodic centrifugal force is generated during rotation. This centrifugal force acts radially on the bearings and casing, causing radial vibration. By aligning the sensor's X-axis with the radial direction, this radial vibration component caused by rotor imbalance can be directly collected, providing crucial data support for subsequent rotor eccentricity fault diagnosis.

[0050] The Y-axis is arranged perpendicular to the motor's radial direction. In the motor's structure, the X-axis and Y-axis are perpendicular to each other and both lie within the motor's radial plane, together forming a two-dimensional measurement coordinate system in the radial plane. Although the X-axis covers the main radial vibration direction, the motor's radial vibration does not occur strictly in a single direction. Due to factors such as uneven rotor mass distribution, bearing clearance, and assembly errors, radial vibration often exhibits multi-directional characteristics within the radial plane. The introduction of the Y-axis allows the system to simultaneously acquire vibration components in another direction orthogonal to the X-axis within the radial plane, avoiding the omission of vibration information perpendicular to that direction due to measuring only a single radial direction, thus achieving more complete and accurate acquisition of radial vibration.

[0051] The Z-axis is arranged along the motor's axial direction. The motor's axial direction refers to the direction parallel to the motor's rotation axis. During motor operation, factors such as axial movement of bearings, axial impact during gear transmission, and fluctuations in the axial magnetic pull of the rotor under the influence of a magnetic field can all cause axial vibration. By aligning the Z-axis with the motor's axial direction, vibration signals in the axial direction can be collected independently.

[0052] With the above-described orthogonal arrangement of the three axes, the X-axis (parallel to the radial direction), Y-axis (perpendicular to the radial direction), and Z-axis (along the axial direction) form a complete spatial rectangular coordinate system. The advantages of this arrangement are: First, the physical meaning of each axis is clear, and the fault direction is strong. The X / Y axes correspond to radial vibration, which is directly related to radial faults such as rotor imbalance and bearing wear; the Z axis corresponds to axial vibration, which is related to axial faults such as bearing axial movement and thrust bearing abnormalities. The vibration data of the three axes each correspond to different physical mechanisms and fault modes, providing a clear physical basis for subsequent fault feature extraction and classification.

[0053] Secondly, the three-axis synchronous acquisition enables comprehensive perception of the motor's vibration state. Vibration data from any single axis cannot fully reflect the motor's true vibration state. Through synchronous acquisition along three orthogonal axes, the system can obtain complete vibration vector information of the motor in three-dimensional space, avoiding information loss caused by single-axis or dual-axis measurements.

[0054] Third, the arrangement method is highly compatible with the controller PCB layout. For the vibration acquisition circuit embedded in the main control PCB of the controller, the axis of the sensor chip is determined by its mounting direction. The arrangement method of X-axis parallel to the motor radial direction, Y-axis perpendicular to the radial direction, and Z-axis along the axial direction can be achieved by reasonably setting the sensor mounting angle during the PCB layout stage, without the need for additional mechanical structure adjustments, which facilitates engineering implementation.

[0055] S2. Synchronously acquire the electrical parameters of the motor and the raw vibration signals obtained by the vibration sensor; Specifically, synchronously acquiring the electrical parameters of the motor and the raw vibration signals obtained from vibration sensors is the data foundation for achieving decoupling and fusion diagnosis of electrical and vibration signals. Synchronous acquisition means that the sampling of electrical parameters and vibration signals are strictly aligned on the time axis, ensuring that each set of data corresponds to the same operating condition of the motor at the same operating moment. If there is a time deviation, the calculated value of the subsequent electromagnetic vibration component will be out of sync with the electromagnetic component in the actual vibration, making it impossible to accurately eliminate the electromagnetic components during subtraction, and severely degrading the extraction accuracy of the pure mechanical vibration signal.

[0056] To achieve synchronous data acquisition, the main control MCU employs a unified sampling clock source at the hardware level, allowing the electrical sampling unit and the vibration acquisition unit to share the same clock reference. At the software level, the MCU executes a complete sampling task sequence every 1ms, synchronously triggering electrical sampling and SPI data reading of vibration data, and uniformly appending a synchronization timestamp to all acquired data for timing alignment verification and duration timing. A sampling frequency of 1kHz is chosen, which covers the main mechanical vibration frequencies of the permanent magnet synchronous motor and the harmonics of the electromagnetic excitation main frequency (its fundamental frequency is fe=2·p·fs), satisfying the Nyquist sampling theorem, without excessively large data volumes crowding out computing resources for real-time tasks such as motor vector control. Electrical parameters include at least three-phase current (used to calculate current harmonic amplitude, three-phase current balance, and current RMS value), bus voltage (used to obtain voltage phase), and real-time motor speed (used to convert to the stator power supply fundamental frequency); the vibration signal is the raw acceleration data of the three axes (X, Y, and Z) from a triaxial vibration sensor.

[0057] Through the aforementioned synchronous acquisition mechanism, electrical parameters and vibration signals are precisely aligned in the time dimension, providing reliable data support for the electro-vibration decoupling step of modeling and calculating electromagnetic vibration components based on electrical parameters and accurately removing these components from the original vibration signals.

[0058] S3. Establish an electromagnetic vibration mathematical model based on electrical parameters, calculate the electromagnetic vibration components, and remove the electromagnetic vibration components from the original vibration signal to obtain a pure mechanical vibration signal. In this step, the mathematical model of electromagnetic vibration is expressed as follows:

[0059] in, Electromagnetic vibration components, The electromagnetic excitation frequency is the main frequency. , This represents the number of pole pairs of the motor. Provides the base frequency for stator power supply; The amplitude of the kth harmonic of the stator current is collected in real time; The electromagnetic-vibration coupling coefficient corresponding to the kth harmonic; The initial phase angle of the corresponding harmonic is given by N, the total harmonic order is given by N, and t is the time.

[0060] In this step, the calculation expression for the purely mechanical vibration signal is:

[0061] in, It is a purely mechanical vibration signal. The original vibration signal, It is an electromagnetic vibration component.

[0062] Specifically, during the operation of a permanent magnet synchronous motor, a three-phase alternating current is applied to the stator windings, generating a rotating magnetic field. This magnetic field interacts with the magnetic field of the rotor's permanent magnets, producing electromagnetic torque that drives the rotor to rotate. Simultaneously, the air gap magnetic field acts on the stator core, generating electromagnetic force waves that vary with time and space. These waves excite the stator core to produce elastic vibrations, which propagate outward through the casing, forming electromagnetic vibrations. This electromagnetic vibration is an inherent accompanying phenomenon during motor operation; its frequency and amplitude are uniquely determined by the electrical parameters of the stator power supply and are unrelated to the motor's mechanical health.

[0063] The raw vibration signal collected by the sensor In this process, electromagnetic vibration components and pure mechanical vibration components caused by mechanical faults are superimposed and mixed together. If this mixed signal is analyzed directly, the electromagnetic vibration will mask or interfere with the characteristics of the mechanical fault, leading to misdiagnosis. Therefore, it is necessary to accurately remove the electromagnetic vibration components from the original vibration signal in order to obtain a pure mechanical vibration signal that truly reflects the mechanical state of the motor.

[0064] This invention establishes a mathematical model for the electromagnetic vibration components based on the electromagnetic excitation mechanism of a three-phase permanent magnet synchronous motor. The dominant electromagnetic excitation frequency is determined by the number of pole pairs of the motor and the fundamental frequency of the stator power supply, and its calculation formula is as follows: ,in This represents the number of pole pairs of the motor. This is the fundamental frequency for stator power supply. The speed n is calculated from the real-time motor speed. For a permanent magnet synchronous motor, the speed n is related to the stator power supply base frequency. The condition n=60 is satisfied between them. Therefore, after obtaining the real-time speed through the speed detection channel of the motor controller, it can be calculated in real time as / p (r / min). .

[0065] Determining the dominant frequency of electromagnetic excitation Based on this, electromagnetic vibration components The mathematical model is constructed using the superposition of multiple harmonics, and its expression is:

[0066] in, This represents the amplitude of the k-th harmonic of the stator current, acquired in real time by the electrical sampling unit. In a permanent magnet synchronous motor powered by an inverter, the stator current inevitably contains various time harmonics. These harmonic currents interact with the fundamental magnetic field, generating electromagnetic force waves of corresponding frequencies, which in turn excite electromagnetic vibrations of those frequencies. Therefore, the amplitude of the current harmonics is a key input quantity determining the amplitude of the electromagnetic vibration.

[0067] This is the electromagnetic-vibration coupling coefficient corresponding to the k-th harmonic, characterizing the proportional relationship between the electromagnetic vibration amplitude generated by the k-th harmonic current. This coefficient comprehensively reflects the inherent properties of the motor stator structure, such as stiffness, mass distribution, damping characteristics, and the transfer function from electromagnetic force to mechanical vibration, and is related to the specific model and structural parameters of the motor. The coupling coefficients of each harmonic are obtained through calibration during the no-load test before the motor leaves the factory. That is, when the motor is in a healthy state, the excitation of each harmonic current with known amplitude is applied, and the corresponding vibration response is measured. The coupling coefficients of each harmonic can then be deduced and stored in the non-volatile memory area of ​​the MCU.

[0068] For the first The initial phase angle of the electromagnetic vibration component corresponding to the second harmonic current is obtained by real-time voltage phase synchronization. The main control MCU performs a Fast Fourier Transform (FFT) on the synchronously acquired stator current signal to extract the amplitude and phase information of each harmonic component, and recovers its continuous initial phase angle through phase dewinding processing. Substitute into the electromagnetic vibration mathematical model to participate Real-time calculation is required. The phase relationship of each harmonic current directly affects the composite vibration waveform after the superposition of multiple harmonics. Therefore, it is necessary to accurately obtain the initial phase angle through real-time voltage phase information to ensure that the calculation results of the mathematical model are strictly aligned with the actual electromagnetic vibration in phase.

[0069] Based on the above mathematical model, the MCU performs an electrical vibration decoupling operation every 1ms: first, it obtains the electrical parameters at the current moment, including the amplitude of the three-phase current harmonics. and voltage phase The stator power supply base frequency is read in real time by the electrical sampling unit. The electromagnetic-vibration coupling coefficient is obtained from the real-time rotational speed. Then, the factory calibration values ​​stored in the MCU's non-volatile memory are directly called. After substituting all parameters into the electromagnetic vibration mathematical model, the electromagnetic vibration components at the current moment are calculated point by point. Subsequently, the raw vibration signals acquired by the triaxial vibration sensor at the same time were... Subtract the calculated electromagnetic vibration components Obtaining purely mechanical vibration signals .

[0070] The effectiveness of this subtraction operation depends on two preconditions: first, the synchronous acquisition of electrical parameters and vibration signals. If there is a time discrepancy between the two, the calculated electromagnetic vibration component will be misaligned with the electromagnetic component in the actual vibration, and the subtraction operation will not be able to accurately eliminate the electromagnetic components; second, the coupling coefficient... The calibration accuracy is crucial, as calibration errors directly affect the calculation accuracy of electromagnetic vibration components, and consequently, the extraction quality of purely mechanical vibration signals.

[0071] Through the above electro-vibration decoupling operation, the electromagnetic vibration component in the original vibration signal is accurately eliminated, ultimately yielding a pure mechanical vibration signal that reflects only the mechanical state of the motor. This signal eliminates interference from electrical vibration noise, providing a clean and reliable data source for subsequent vibration feature extraction and mechanical fault diagnosis.

[0072] S4. Extract vibration features from pure mechanical vibration signals and perform fusion analysis in combination with electrical parameters to diagnose motor fault types; Specifically, extracting vibration features from purely mechanical vibration signals and combining them with electrical parameters for fusion analysis to diagnose motor fault types is the core of this invention's accurate fault classification. While purely mechanical vibration signals have eliminated electromagnetic interference, they are complex waveforms that change continuously over time and cannot be directly used for fault determination. Although electrical parameters reflect the electrical state of the motor, they are not sensitive to mechanical faults. Therefore, it is essential to organically integrate the two, extracting quantitative indicators from the vibration signals and then cross-validating them using electrical parameters as auxiliary criteria to achieve accurate fault type identification.

[0073] In this step, the vibration characteristics include at least the effective value (RMS) of mechanical vibration, kurtosis (K), and peak factor (CF), which are calculated according to the following formulas:

[0074] in, The first of the purely mechanical vibration signals One sample point, The mean value of the purely mechanical vibration signal. This represents the total number of samples.

[0075] Specifically, the RMS (Real Mean Square) of mechanical vibration is the square root of the mean square of the vibration signal, characterizing the average energy level of the vibration signal and reflecting the overall vibration intensity of the motor. When the motor experiences faults such as loose base, decreased structural stiffness, or increased mass imbalance, the overall vibration energy increases, and the RMS value increases accordingly. RMS is a global indicator, sensitive to continuous changes in vibration energy, but insensitive to short-term impacts. Even if the RMS is within the normal range, the motor may still exhibit localized early pitting or wear. Therefore, RMS is suitable for evaluating the overall trend of vibration and for threshold determination in graded control.

[0076] Kurtosis K is the ratio of the fourth-order central moment of a vibration signal to the fourth power of the RMS value. It characterizes the sharpness of the signal amplitude distribution and is the most sensitive indicator for detecting the presence of periodic impact components in a vibration signal. When pitting, spalling, or rolling element wear occurs on the raceway of a motor bearing, each ball passing through a defect point generates a transient impact pulse, which manifests as a series of large-amplitude peaks in the time-domain waveform. These peaks cause the signal amplitude distribution to deviate from a normal distribution, with a thicker tail and a sharper center, resulting in a significant increase in the kurtosis value. The core advantage of kurtosis lies in its high selectivity for impact components. Even if the RMS value changes little, as long as a weak periodic impact appears in the signal, kurtosis can capture this change, making it particularly suitable for detecting early bearing failures.

[0077] The peak value factor (CF) is the ratio of the peak value to the effective value of a vibration signal. It characterizes whether there are instantaneous spikes with large amplitudes in the signal waveform and is an effective indicator for judging rotor imbalance and periodic rubbing faults. CF reflects both the peak energy and average energy of the signal: a small CF indicates a uniform amplitude distribution and a stable waveform; a large CF indicates the presence of significant instantaneous large amplitude impacts or spikes in the signal. In rotor imbalance faults, the centrifugal force caused by the unbalanced mass generates sinusoidal vibrations at the same frequency as the rotational speed. There is a definite proportional relationship between the peak value and the effective value of this vibration (the CF of an ideal sine wave is approximately 1.414). When the imbalance worsens or rubbing occurs, the CF value will deviate significantly from the normal range. Compared with kurtosis, CF focuses on measuring the "peak-average" relationship of the waveform and is more sensitive to the overall dynamic range of the signal.

[0078] Three types of features characterize different properties of vibration from different dimensions: RMS reflects the energy magnitude, kurtosis reflects the impact intensity, and crescendo reflects the waveform sharpness. Each has its own emphasis, and when used in combination, they form a multi-dimensional description of the motor's mechanical state. The effective value of mechanical vibration and kurtosis are complementary; the effective value is sensitive to continuous energy changes, while kurtosis is sensitive to instantaneous pulses. Together, they can distinguish between "large-amplitude steady vibration" and "small-amplitude impact vibration," two different types of anomalies. Crescendo is somewhat correlated with kurtosis, but their emphases differ: kurtosis reflects the tail thickness of the probability distribution, while crescendo reflects the dynamic range of the waveform. Kurtosis is more sensitive in bearing faults, while crescendo is more directional in rotor imbalances. Simultaneous calculation and cross-validation of these three features significantly reduces the risk of misjudgment caused by a single feature. The calculated RMS, K, and CF values ​​are used as inputs to the fault diagnosis process, fused with electrical parameters, to jointly determine the final fault type.

[0079] In this step, motor fault types include bearing failure, rotor eccentricity, loose base, and electrical overload; The criteria for determining bearing failure are: the three-phase current balance in the electrical parameters is less than or equal to 5%, and the kurtosis in the vibration characteristics is... or peak factor And the duration of the above features is greater than or equal to 2.5s; The criteria for determining rotor eccentricity fault are: the ratio of instantaneous speed fluctuation to rated speed is greater than 3%, the ratio of low-frequency harmonic amplitude corresponding to the rotational frequency to fundamental current amplitude is greater than 5%, the effective value of mechanical vibration changes with speed at a rate greater than the upper limit of normal rate of change, and the duration of the above characteristics is greater than or equal to 2.5s. The criteria for determining a loose base fault are: the three-phase current balance in the electrical parameters is less than or equal to 5%, and the effective value of the mechanical vibration in the vibration characteristics is... And the absolute value of the difference between the kurtosis value at the current time and the kurtosis value at the previous time. Furthermore, the above characteristics are satisfied for a duration greater than or equal to 2.5 seconds; The criteria for determining an electrical overload fault are: the effective value of the stator current. And the effective value in the vibration characteristics And steepness Furthermore, the duration of the aforementioned characteristics is greater than or equal to 2.5 s; the effective value of the stator current is greater than 1.2 times the rated current, the effective value of the mechanical vibration is less than 1.3 times the reference effective value, and the kurtosis is less than 4, and the duration of the aforementioned characteristics is greater than or equal to 2.5 s. in, The reference mechanical vibration effective value for motor calibration under no-load conditions. This is the rated current.

[0080] Specifically, after extracting three types of time-domain features—RMS (Real Mean Squared Value), kurtosis K, and peak factor CF—from the pure mechanical vibration signal, multi-feature fusion analysis is required in conjunction with electrical parameters to ultimately achieve accurate diagnosis of motor fault types. A single physical quantity (pure vibration or pure electrical) cannot fully describe the motor's operating state. Electrical parameters are sensitive to electrical faults (such as overload and inter-turn short circuits) but not to mechanical faults. Vibration features are sensitive to mechanical faults but cannot distinguish between abnormal vibrations caused by electrical or mechanical reasons. Therefore, this invention jointly analyzes electrical parameters and vibration features, achieving accurate fault type classification through cross-validation of the two types of physical quantities. The motor fault types covered by this invention include four categories: bearing faults, rotor eccentricity, base looseness, and electrical overload. The judgment conditions for each type of fault are explained below.

[0081] The criteria for determining bearing failure are: the three-phase current balance in the electrical parameters is less than or equal to 5%, and the kurtosis in the vibration characteristics is... or peak factor Furthermore, the duration of the aforementioned characteristics is greater than or equal to 2.5 seconds.

[0082] The logic behind this judgment condition is as follows: bearing failure is a purely mechanical failure, and its occurrence and development will not cause changes in the electrical symmetry of the motor. Therefore, maintaining three-phase current balance (imbalance ≤ 5%) is a prerequisite. If the three-phase current is severely unbalanced, electrical causes should be investigated first, rather than bearings. The expression for calculating the three-phase current balance is: ;in, These are the effective current values ​​for phases A, B, and C, respectively. This is the average value of the three-phase current. Under the premise of current balance, the core symptom of bearing failure is the appearance of periodic impact pulses in the vibration signal. When pitting, spalling, or rolling element wear occurs on the bearing raceway, each ball passing through a defect point generates a transient impact, causing a large peak in the amplitude distribution of the vibration signal. At this time, both the kurtosis K and the peak factor CF will increase significantly. This invention uses kurtosis... or peak factor As the judgment threshold, kurtosis K is most sensitive to early, weak impacts and is suitable for detecting the initial stage of bearing pitting; peak factor CF is sensitive to the dynamic range of amplitude and is suitable for detecting impacts with larger amplitudes. The two are related by "OR"; exceeding either threshold triggers a bearing fault warning, avoiding missed detections due to the insensitivity of a single indicator. The 2.5s duration threshold is used to exclude instantaneous over-limits caused by occasional noise pulses, ensuring the reliability of the judgment results. This duration threshold can be adjusted according to actual conditions.

[0083] The criteria for determining rotor eccentricity fault are: the ratio of instantaneous speed fluctuation to rated speed is greater than 3%, the ratio of low-frequency harmonic amplitude corresponding to the rotational frequency to fundamental current amplitude is greater than 5%, the effective value of mechanical vibration changes with speed at a rate greater than the upper limit of normal rate of change, and the duration of the above characteristics is greater than or equal to 2.5s.

[0084] Rotor eccentricity refers to the misalignment of the rotor's rotation axis with the stator's geometric center line, resulting in uneven air gap. When rotor eccentricity occurs, the air gap magnetic permeability changes periodically with the rotor position, inducing low-frequency harmonic components corresponding to the rotational frequency in the stator current. Simultaneously, the unilateral magnetic pull generated by eccentricity increases with increasing rotational speed, causing the effective value of mechanical vibration to exceed the normal range of change with rotational speed. Furthermore, the centrifugal force caused by the eccentric mass causes periodic instantaneous fluctuations in rotational speed within each revolution. These three characteristics describe different manifestations of rotor eccentricity from three dimensions: the electrical domain (current harmonics), the mechanical domain (vibration-rotational speed change rate), and the kinematic domain (rotational speed fluctuation). All three characteristics must be satisfied simultaneously for rotor eccentricity to be determined; the presence of any single characteristic (such as large rotational speed fluctuations but normal current harmonics) is insufficient to determine eccentricity.

[0085] The criteria for determining a loose base fault are: the three-phase current balance in the electrical parameters is less than or equal to 5%, and the effective value of the mechanical vibration in the vibration characteristics is... And the absolute value of the difference between the kurtosis value at the current time and the kurtosis value at the previous time. Furthermore, the above characteristics must be satisfied for a duration greater than or equal to 2.5 seconds.

[0086] A loose base is a structural mechanical fault, and its occurrence does not affect the electrical symmetry of the motor. Therefore, the three-phase current balance is still required to be ≤5% to eliminate electrical interference. A loose base leads to a decrease in the overall support stiffness of the motor, a significant increase in vibration energy, and a substantial increase in the effective value (RMS) of mechanical vibration. However, the vibration signal generated by a loose base usually exhibits a sine wave or quasi-sine wave waveform, without the periodic impact pulses seen in bearing pitting, so the kurtosis value does not increase significantly. This invention combines the two characteristics of "increased effective value of mechanical vibration" and "no significant change in kurtosis": when (Vibrational energy significantly exceeds the health baseline) and the absolute value of the difference between the current kurtosis value and the previous kurtosis value. If the kurtosis remains stable and no impact fluctuations occur, the possibility of bearing failure is ruled out, and the problem is determined to be a loose base. The 2.5s duration threshold is required to ensure that the continuous increase in vibration energy is not caused by instantaneous load fluctuations. This duration threshold can be adjusted according to the actual situation.

[0087] The criteria for determining an electrical overload fault are: the effective value of the stator current. And the effective value of mechanical vibration in the vibration characteristics And steepness Furthermore, the duration of the aforementioned characteristics is greater than or equal to 2.5 seconds.

[0088] Electrical overload refers to a situation where the load torque on a motor exceeds its rated value, leading to a significant increase in the effective value of the stator current. At this time, the motor's electromagnetic state is abnormal (current exceeds the limit), but the mechanical structure remains intact. The main sources of vibration signals are normal electromagnetic vibration and mechanical operation vibration; there is no impact or abnormal energy release caused by mechanical damage. Therefore, the effective value and kurtosis of the mechanical vibration remain within a healthy range. This judgment condition combines electrical parameters and vibration characteristics: current exceeding the limit is the triggering condition, while the mechanical characteristics are normal (…). , This is used to rule out the possibility of "current increase due to mechanical failure." If the current exceeds the limit and is accompanied by a significant increase in the effective value of mechanical vibration or excessive kurtosis, it indicates that there may be an overload caused by mechanical failure (such as increased rotor eccentricity or severe bearing wear). In this case, the corresponding mechanical failure should be prioritized over a simple electrical overload. By combining the above conditions, the electrical overload condition can be accurately identified, while ensuring a clear distinction between it and a mechanical failure.

[0089] The judgment criteria for the above four types of faults follow the following common design rules: First, each type of fault must meet a delay requirement of ≥2.5s in duration to avoid false alarms caused by single-point instantaneous noise or disturbance; Second, each type of fault combines two types of information: electrical parameters and vibration characteristics. Cross-validation is used to distinguish fault types. Three-phase current balance is used to eliminate electrical imbalance interference, speed fluctuations and harmonic amplitude values ​​are used to capture the unique electrical and mechanical linkage characteristics of rotor eccentricity, and the effective current value is used to identify overload conditions; Third, the judgment criteria for each type of fault are mutually exclusive or complementary. Bearing faults and base loosening both require current balance but correspond to two different vibration modes: kurtosis increase and kurtosis stability, respectively; electrical overload requires normal mechanical characteristics to distinguish it from overload caused by mechanical faults; rotor eccentricity is specifically identified through the linkage characteristics of electrical and mechanical components. These four judgment criteria together constitute a complete motor fault diagnosis logic, effectively avoiding misjudgments caused by monitoring a single physical quantity.

[0090] S5. Based on the diagnosed fault type and its severity, execute the corresponding hierarchical closed-loop control strategy.

[0091] In this step, the hierarchical closed-loop control strategy includes: Level 1 warning: When Record and report faults, without limiting motor power; Secondary torque limiting: when If all the judgment conditions for any of the motor fault types are met, the upper limit of the output torque will be locked at 70% of the rated torque; Level 3 lockdown shutdown: When And when the duration is greater than or equal to 2.5s, the drive output of the power device is blocked, causing the motor to stop.

[0092] Specifically, the hierarchical closed-loop control strategy is the key link in this invention's transition from "passive alarm" to "active protection." The aforementioned electro-vibration decoupling and fault diagnosis steps complete the perception and identification of the motor's operating status. The hierarchical closed-loop control strategy, based on this, executes differentiated protection actions according to the fault type and its severity, forming a complete closed loop of "perception—diagnosis—decision—execution." This invention divides the control strategy into three levels: Level 1 warning, Level 2 torque limiting, and Level 3 shutdown. The levels are separated by the effective vibration value (RMS) relative to the no-load calibration reference value. The switching is performed progressively based on the multiples of the fault and the satisfaction of the fault determination conditions.

[0093] The triggering conditions for a Level 1 warning are: This range indicates that the motor vibration level has exceeded the health baseline but is still in a slight deterioration stage. A mechanical vibration effective value (RMS) exceeding 1.3 times the baseline value means the vibration energy has significantly deviated from the no-load calibration state, potentially indicating early wear, slight loosening, or mass imbalance. However, a value below 1.8 times the baseline indicates the vibration amplitude is still within a controllable range, and the motor can continue to operate. In this state, the system performs only two operations: first, it records the current fault type, RMS value, kurtosis value, and timestamp in the MCU's internal data storage area, forming a fault log for subsequent maintenance traceability; second, it reports an early warning message via the communication bus, sending the warning information to the aircraft's main control system or ground monitoring terminal. Level 1 warning does not restrict motor power output, and the motor can still operate at full power. The reason for not immediately taking amplitude limiting measures is that when the RMS is in this range, although the vibration level is already high, it has not yet endangered the structural safety of the motor. Premature load reduction will affect the aircraft's power performance. At the same time, slight degradation may be a momentary fluctuation caused by changes in operating conditions. Continuous monitoring can be achieved by recording and reporting. Actions can be taken only after the degradation trend is confirmed, which ensures safety and avoids unnecessary power interruption.

[0094] The trigger condition for the secondary torque limiting is: Furthermore, it must meet all the criteria for any of the motor fault types. When the effective value (RMS) of mechanical vibration exceeds 1.8 times the reference value, the vibration energy has significantly exceeded the healthy range, indicating a clear structural abnormality in the motor; simultaneously, the system must confirm that the current increase in RMS is indeed caused by a certain identifiable fault type, and only if the effective value (RMS) of mechanical vibration > However, if the fault determination conditions are not met (e.g., the RMS increase originates from external load fluctuations rather than a motor fault), the secondary load reduction will not be triggered to avoid unnecessary power limiting due to external disturbances. When both conditions are met simultaneously, the system determines that the motor has entered a moderate fault state and immediately locks the upper limit of the output torque to 70% of the rated torque. Torque limiting is achieved by the MCU adjusting the modulation ratio of the PWM drive unit. The MCU clamps the upper limit of the torque command of the current loop to 70% of the rated torque, thus limiting the current amplitude output from the inverter to the motor. The motor output torque decreases accordingly, thereby weakening the electromagnetic force amplitude and reducing the energy of the mechanical vibration excitation source. The reason for choosing 70% as the limit value is based on two considerations: First, this load reduction is sufficient to significantly reduce the vibration amplitude (RMS can usually fall back to below 1.8 times the baseline), avoiding structural damage caused by the continuous deterioration of vibration energy; second, retaining 70% of the torque output capability can maintain the basic power requirements of the aircraft, avoiding loss of control of the aircraft due to complete loss of power. In the application of electric aircraft, even if the power is downgraded, as long as the motor can still output controllable torque, the pilot or flight control system still has the opportunity to complete an emergency landing or return.

[0095] The trigger conditions for a Level 3 shutdown are: The duration must be greater than or equal to 2.5 seconds. When the effective value (RMS) of mechanical vibration exceeds 2.8 times the reference value, it indicates that the motor vibration is at an extremely high level, posing an extreme risk of impending structural damage (such as bearing breakage, rotor rubbing, and casing fracture). Continuing to operate the motor at this time would not only severely damage the motor itself but could also trigger secondary disasters such as fires and mechanical detachment, directly threatening aircraft safety. Therefore, once the system determines that the RMS is within this range and the duration reaches 2.5 seconds, it immediately executes a three-level shutdown: the MCU directly blocks the PWM drive signal of the full-bridge IGBT, all power switches of the inverter are turned off, the motor is completely disconnected from the bus power supply, and the motor stops freely due to inertia. The 2.5-second duration requirement is used to filter out extreme transient impacts (such as instantaneous vibration spikes caused by strong gusts of wind encountered by the aircraft), avoiding unnecessary in-flight shutdowns due to single-point instantaneous over-limits.

[0096] The three-level control strategies form a progressive protection chain of "monitoring-limiting-shutdown". When vibration enters the first-level warning range from the normal state, the system only records and reports, without interfering with motor operation, providing early warning information for maintenance personnel. If the vibration continues to deteriorate to the second-level range and the fault type is confirmed, the system actively reduces the load to 70% torque, suppressing further vibration deterioration by reducing the electromagnetic excitation intensity. If the vibration falls back to the first-level range due to the load reduction, the system maintains the load reduction state until the fault is manually confirmed to be eliminated before resetting. If the vibration continues to rise to the third-level range after the load reduction, the system executes a shutdown. The key design of this progressive mechanism is that different levels correspond to different fault severity and risk levels, and the intensity of the action matches the fault risk. Low-level faults are only recorded and reported, medium-level faults are limited and load reduced, and high-level faults are shut down. The three work together to avoid excessive interference to flight missions caused by over-protection of "shutting down at the first sign of an anomaly" and to prevent the breach of safety bottom lines by passive protection of "only alarming and not intervening," achieving an optimal balance between protection intensity and mission requirements.

[0097] Example 2

[0098] This embodiment provides a motor controller, including: Main control PCB circuit board; The vibration acquisition circuit, integrated on the main control PCB board, includes a triaxial vibration sensor and its peripheral circuitry, used to acquire the vibration signals of the motor. An electrical sampling unit is used to collect the electrical parameters of the motor; The main control MCU is electrically connected to both the vibration acquisition circuit and the electrical sampling unit. The PWM drive unit, controlled by the main MCU, is used to drive power devices; The main control MCU is configured to execute the fusion diagnosis and hierarchical closed-loop control method of embedded vibration acquisition in the motor controller as in Example 1.

[0099] This embodiment uses an 18kW, 8-pole-pair (p=8) permanent magnet synchronous motor applied in a light electric aircraft as the protection target, and adopts the integrated motor controller described in this embodiment (such as...). Figure 2 (As shown). The specific implementation process is as follows: I. Hardware assembly and circuit layout.

[0100] The vibration acquisition circuit is implemented as a separate sub-PCB board, integrating the ADXL312 triaxial vibration sensor chip and all peripheral components onto the same circuit board to form a highly integrated acquisition module. This sub-PCB board is mounted in the non-power heat dissipation area of ​​the main control PCB and is electrically and mechanically fixed to the main control PCB via pin headers. The X-axis of the ADXL312 is arranged parallel to the motor radial direction, the Y-axis is perpendicular to the motor radial direction, and the Z-axis is along the motor axial direction. The sub-PCB board maintains a physical distance of 10.5mm from the IGBT power transistors and high-voltage bus capacitors, physically blocking conducted and radiated interference introduced by the high-voltage circuit.

[0101] like Figure 3 The sensor power supply employs a two-stage LDO step-down voltage regulation scheme: the first stage step-down is achieved by the controller's 12V auxiliary power supply to 5V, and the second-stage LDO outputs a high-precision, low-temperature-drift 3.3V supply voltage. At the output of the second-stage LDO, a combined decoupling filter network consisting of a 10µH inductor, a 1µF tantalum capacitor, a 10µF ceramic capacitor, and a 1µF ceramic capacitor is connected in series to the power supply pin. The ADXL312 is directly connected to the general-purpose I / O port of the main control MCU via the SPI bus (including SDI, SDO, SCLK, and chip select signal lines), with a 10Ω resistor connected in series on the bus. The entire vibration acquisition circuit is completely enclosed within the metal housing of the motor controller, requiring no external leads, no external sensors, no independent power supply lines, and no external signal cables.

[0102] The controller MCU uses a unified sampling clock source. Electrical sampling and triaxial vibration sampling share the same clock reference, and the sampling frequency is uniformly 1kHz. The sampling data is accompanied by a synchronization timestamp.

[0103] II. Baseline value calibration.

[0104] The motor was run stably at 3000 r / min under no-load, normal temperature, and rated speed for 3 minutes. The controller MCU continuously collected 5000 points of decoupled mechanical vibration data and calculated the no-load baseline effective value RMS0 = 0.021g. This parameter was stored in the MCU's non-volatile memory area as the benchmark for the entire system's threshold calculation. The system preset three levels of RMS multiple thresholds: Level 1 warning coefficient 1.3, Level 2 load reduction coefficient 1.8, and Level 3 shutdown coefficient 2.8, which translates to T1 = 1.3 × RMS0 = 0.0273g, T2 = 1.8 × RMS0 = 0.0378g, and T3 = 2.8 × RMS0 = 0.0588g. The continuous delay for fault determination is uniformly fixed at 2.5s, and single-point instantaneous over-limit does not trigger a fault.

[0105] III. Online calculation of electro-oscillatory decoupling.

[0106] The main control MCU first reads the raw acceleration data from the ADXL312 triaxial vibration sensor via the SPI bus to obtain the raw vibration signal at the current moment. Since the sensor output may contain high-frequency electronic noise and quantization noise, the system performs sliding window filtering preprocessing on the raw data. The mean filtering or median filtering algorithm within the sliding window is used to smooth each sampling point and several neighboring points before and after it. This suppresses the interference of random noise and spike pulses on subsequent calculations without losing the effective frequency band components of the vibration signal.

[0107] Simultaneously, the electrical sampling unit synchronously collects three-phase current, bus voltage, and real-time motor speed. Based on the real-time speed... (r / min), MCU according to Convert the stator power supply base frequency at the current moment. ,in This is the number of pole pairs of the motor; then press Calculate the dominant frequency of electromagnetic excitation. The amplitudes of each harmonic are obtained after FFT analysis of the three-phase currents. Harmonic initial phase angle The bus voltage is used to obtain the current voltage phase reference. Substituting the above electrical parameters into the electromagnetic vibration mathematical model, the electromagnetic vibration components at the current moment are calculated point by point. :

[0108] Wherein, coupling coefficient These are fixed parameters that are calibrated and stored in the non-volatile memory area of ​​the MCU before the motor leaves the factory.

[0109] The electromagnetic vibration components at the current moment are calculated. Then, the original vibration signals preprocessed by sliding window filtering at the same time are... Subtracting the electromagnetic vibration component yields the pure mechanical vibration signal. :

[0110] This purely mechanical vibration signal eliminates the interference of electromagnetic vibration noise and truly reflects the mechanical operating state of the motor.

[0111] Subsequently, the system processed purely mechanical vibration signals. Time-domain feature extraction is performed to calculate three quantitative indicators: the effective value (RMS) of mechanical vibration, kurtosis (K), and peak factor (CF). These three feature values ​​serve as inputs to the fault diagnosis logic. Together with electrical parameters such as three-phase current balance, speed fluctuation, and effective current value, they are fed into the fault diagnosis module. The module outputs diagnostic results based on differentiated judgment conditions for four types of faults: bearing fault, rotor eccentricity, base looseness, and electrical overload.

[0112] The overall process of the above electro-oscillatory decoupling and feature extraction algorithm is as follows: Figure 4 As shown, the raw data collected by the triaxial vibration sensor is preprocessed by sliding window filtering, and then combined with the electromagnetic excitation main frequency and current harmonic coefficients calculated in real time from the three-phase current, rotational speed, and bus voltage, and substituted into the electromagnetic vibration mathematical model for calculation. Then, it is removed from the original vibration signal by subtraction. Obtaining pure mechanical vibration signals ; then on The system calculates three types of time-domain features: RMS, kurtosis K, and peak factor CF. Finally, it enters the fault determination logic and outputs the diagnostic results based on the determination conditions of the aforementioned four types of faults.

[0113] IV. Four types of fault identification.

[0114] The overall process of four types of fault classification and hierarchical closed-loop control is as follows: Figure 5 As shown.

[0115] Figure 5This demonstrates the complete decision-making chain from feature input to closed-loop control output. The input to the fault determination module consists of two parts: one part is the vibration features extracted from the pure mechanical vibration signal, namely the effective value of mechanical vibration (RMS), kurtosis K, and peak factor CF; the other part is the electrical parameters collected in real time by the electrical sampling unit, namely three-phase current, current RMS, speed, and current harmonics. Both types of features enter the fault determination logic and are matched one by one according to the differentiated determination conditions of four types of faults. In the determination logic, the mechanical vibration reference value RMS0 (the effective value of mechanical vibration reference calibrated by the motor under no-load) serves as the calculation benchmark for each level of threshold and participates in the fault level determination. The multiple relationship between the current RMS and RMS0 determines whether the fault belongs to level one warning, level two torque limiting, or level three shutdown. After the fault level determination is completed, the system outputs the corresponding closed-loop control command (record reporting, torque limiting, or shutdown), forming a complete closed loop from feature input to control execution.

[0116] Based on this, the specific judgment process for various faults is as follows: Bearing wear failure: A minor pitting defect is detected in the motor bearing, and the motor operates stably with 50% rated load. At this time, the three-phase current balance... If the mechanical kurtosis K is greater than 4.0 for 2.5s after decoupling and the effective value of mechanical vibration RMS is between T2 and T3 (e.g., 0.045g), the system determines that the bearing has a moderate fault. The controller automatically locks the upper limit of the output torque to 70% of the rated torque (111.3 N·m) and reports the bearing fault code through the CANFD bus.

[0117] Rotor eccentricity fault: When the motor is running, the mechanical vibration amplitude rises and falls synchronously with the speed. After decoupling, the effective value (RMS) of the mechanical vibration stabilizes at 0.045g (in the secondary fault range between T2 and T3). The ratio of the instantaneous speed fluctuation to the rated speed is greater than 3%. The ratio of the low-frequency harmonic amplitude corresponding to the rotational frequency to the fundamental current amplitude is greater than 5%. If the parameters meet the standard for 2.5 seconds, the system determines that the rotor is eccentric and executes 70% rated torque limiting protection.

[0118] Base loosening fault: Loosen the frame fastening bolts and run the motor at full load. At this time, the three-phase voltage, three-phase current, and speed are all within the rated normal range, with no harmonic distortion, and the three-phase current balance is good. <5%; after decoupling, the effective value of mechanical vibration RMS > 0.0378g, and the kurtosis K is stable ≤ 4.0 with no significant increase ( After 2.5 seconds, if the base is found to be loose, the system enters a reduced-load operation mode and implements 70% rated torque limiting protection. If the fault persists, the load is further reduced to 30% rated torque limiting protection.

[0119] Electrical overload condition: A short-term stall causes the three-phase current to exceed the rated value by 1.2 times. At this time, the effective value of the stator current is... After removing the electromagnetic vibration component, the effective value of mechanical vibration RMS is 0.023g < 0.0273g. All mechanical characteristics are normal. The system only executes the controller's native electrical current limiting protection, does not determine mechanical faults, and does not initiate load reduction and shutdown logic.

[0120] V. Actual machine operation of three-level closed-loop control.

[0121] Level 1 Warning: The motor has been slightly deteriorated over a long period of time. The effective value of mechanical vibration (RMS) is between 0.0273g and 0.0378g and exceeds the limit for 2.5 seconds. The system only stores the fault log and uploads the warning message through the CANFD bus. The motor torque and speed are completely unrestricted.

[0122] Level 2 load reduction: If any mechanical fault meets all the corresponding judgment conditions and RMS ≥ 0.0378g, the upper limit of the controller MCU locked output torque is 70% of the rated torque.

[0123] Level 3 shutdown: When the effective value of mechanical vibration RMS>2.8·RMS0 (i.e.>0.0588g) and the duration is ≥2.5s, and the kurtosis K>4.0 and the peak factor CF>3.5, the controller immediately blocks the PWM drive output of the full-bridge IGBT, the motor stops due to inertia, the fault code is latched, and it can only be powered on again after manual reset.

[0124] VI. Actual test results of the whole machine.

[0125] Compared with traditional external adhesive vibration monitoring solutions, the electromagnetic interference false alarm rate of this embodiment is reduced from 42% to less than 3%. It can identify early micro-faults in bearings, eccentricity, and bases in advance. The graded protection mechanism is adapted to the safe operation requirements of electric aircraft and meets the airborne use conditions of electric aircraft.

[0126] The various embodiments of the present invention have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments.

Claims

1. A method for fusion diagnosis and hierarchical closed-loop control of vibration acquisition embedded in a motor controller, characterized in that, include: The vibration acquisition circuit is embedded in the main control PCB board inside the motor controller. The vibration acquisition circuit includes a triaxial vibration sensor and its peripheral circuit. The electrical parameters of the motor and the raw vibration signals obtained by the triaxial vibration sensor are collected synchronously. An electromagnetic vibration mathematical model is established based on the electrical parameters, electromagnetic vibration components are calculated, and electromagnetic vibration components are removed from the original vibration signal to obtain a pure mechanical vibration signal. Vibration features are extracted from the purely mechanical vibration signal and combined with the electrical parameters for fusion analysis to diagnose motor fault types; Based on the diagnosed fault type and its severity, implement the corresponding hierarchical closed-loop control strategy; The mathematical model of electromagnetic vibration is expressed as follows: in, Electromagnetic vibration components, The electromagnetic excitation frequency is the main frequency. , This represents the number of pole pairs of the motor. Provides the base frequency for stator power supply; The amplitude of the kth harmonic of the stator current is collected in real time; The electromagnetic-vibration coupling coefficient corresponding to the kth harmonic; The initial phase angle of the corresponding harmonic is N, the total harmonic order is N, and t is time. The vibration characteristics include at least the effective value of mechanical vibration (RMS), kurtosis (K), and peak factor (CF). The types of motor faults include bearing failure, rotor eccentricity, loose base, and electrical overload; The criteria for determining bearing failure are: the three-phase current balance in the electrical parameters is less than or equal to 5%, and the kurtosis in the vibration characteristics is... or peak factor And the duration of the above features is greater than or equal to 2.5s; The criteria for determining rotor eccentricity fault are: the ratio of instantaneous speed fluctuation to rated speed is greater than 3%, the ratio of low-frequency harmonic amplitude corresponding to the rotational frequency to fundamental current amplitude is greater than 5%, the rate of change of effective mechanical vibration with respect to speed is greater than the upper limit of normal rate of change, and the duration of the above characteristics is greater than or equal to 2.5s. The criteria for determining the base loosening fault are: the three-phase current balance in the electrical parameters is less than or equal to 5%, and the effective value of the mechanical vibration in the vibration characteristics is... And the absolute value of the difference between the kurtosis value at the current time and the kurtosis value at the previous time. Furthermore, the above characteristics are satisfied for a duration greater than or equal to 2.5 seconds; The criteria for determining the electrical overload fault are: the effective value of the stator current. And the effective value of mechanical vibration in the vibration characteristics And steepness And the duration of the above features is greater than or equal to 2.5s; in, The reference mechanical vibration effective value for motor calibration under no-load conditions. This is the rated current.

2. The method for fusion diagnosis and hierarchical closed-loop control of vibration acquisition embedded in the motor controller according to claim 1, characterized in that, The calculation expression for the purely mechanical vibration signal is: in, It is a purely mechanical vibration signal. The original vibration signal, It is an electromagnetic vibration component.

3. The method for fusion diagnosis and hierarchical closed-loop control of vibration acquisition embedded in the motor controller according to claim 1, characterized in that, The effective value (RMS), kurtosis (K), and peak factor (CF) of the mechanical vibration are calculated according to the following formulas: in, The first of the purely mechanical vibration signals One sample point, The mean value of the purely mechanical vibration signal. This represents the total number of samples.

4. The method for fusion diagnosis and hierarchical closed-loop control of vibration acquisition embedded in the motor controller according to claim 1, characterized in that, The hierarchical closed-loop control strategy includes: Level 1 warning: When Record and report faults, without limiting motor power; Secondary torque limiting: when Furthermore, if all the judgment conditions for any of the aforementioned motor fault types are met, the upper limit of the output torque will be locked at 70% of the rated torque; Level 3 lockdown shutdown: When And when the duration is greater than or equal to 2.5s, the drive output of the power device is blocked, causing the motor to stop.

5. The method for fusion diagnosis and hierarchical closed-loop control of vibration acquisition embedded in the motor controller according to claim 1, characterized in that, The triaxial vibration sensor maintains an electromagnetic isolation distance of greater than or equal to 10 mm from the power devices and bus capacitors in the motor controller.

6. The method for fusion diagnosis and hierarchical closed-loop control of vibration acquisition embedded in the motor controller according to claim 1, characterized in that, The triaxial vibration sensor is powered by a two-stage LDO regulator. The first stage is stepped down to 5V by the controller's auxiliary power supply, and the second stage steps down to output a 3.3V power supply voltage.

7. The method for fusion diagnosis and hierarchical closed-loop control of vibration acquisition embedded in the motor controller according to claim 1, characterized in that, The X-axis of the triaxial vibration sensor is arranged parallel to the radial direction of the motor, the Y-axis is arranged perpendicular to the radial direction of the motor, and the Z-axis is arranged along the axial direction of the motor.

8. A motor controller, characterized in that, include: Main control PCB circuit board; The vibration acquisition circuit, integrated on the main control PCB circuit board, includes a triaxial vibration sensor and its peripheral circuitry, used to acquire the vibration signal of the motor. An electrical sampling unit is used to collect the electrical parameters of the motor; The main control MCU is electrically connected to both the vibration acquisition circuit and the electrical sampling unit. The PWM drive unit, controlled by the main control MCU, is used to drive power devices; The main control MCU is configured to execute the fusion diagnosis and hierarchical closed-loop control method of embedded vibration acquisition in the motor controller as described in any one of claims 1 to 7.

Images