A method for intelligent monitoring and fault early warning of a hair dryer motor operating state

Abstract

The present application relates to the technical field of motor control, and particularly relates to a hair dryer motor operation state intelligent monitoring and fault early warning method. Phase current and bus voltage of a brushless direct current motor are collected to calculate a commutation interval; a stator current dynamic equation based on a back electromotive force observer is constructed in a non-commutation interval; a wind tunnel aerodynamic load torque is introduced as an unknown disturbance term, and a sliding mode observer is designed to dynamically decouple the unknown disturbance term and the back electromotive force; a low-frequency asymmetric distortion component of the back electromotive force and a high-frequency ripple variance of the phase current in the non-commutation interval are extracted; when both exceed a dynamic reference threshold and the duration meets a time window constraint, an early warning signal is output and the motor drive pulse is locked out. The present application decouples and separates the aerodynamic load disturbance, eliminates the interference of wind tunnel working condition changes on electrical characteristic extraction, realizes the identification of early inter-turn short circuit faults, and avoids false positives and false negatives.

Description

Technical Field

[0001] This invention relates to the field of motor control technology, specifically to a method for intelligent monitoring and fault early warning of the operating status of a hair dryer motor. Background Technology

[0002] Current methods for monitoring the operating status and providing fault warnings for brushless DC motors in hair dryers primarily employ a bus current amplitude threshold comparison method. This method collects the motor bus circuit current in real time and compares the sampled value with a preset fixed current threshold. When the bus current exceeds this fixed threshold, the motor is determined to be in an overload or fault state, and the motor drive signal is then cut off for protection. In such solutions, the average current during steady-state operation is typically used as a benchmark, and the motor's operating status is assessed by determining whether the current fluctuation amplitude exceeds the limit.

[0003] The aforementioned monitoring method based on bus current amplitude threshold comparison suffers from a core technical problem: it cannot distinguish between fluctuations in the pneumatic load of the air duct and early electrical faults in the motor. During operation, the air inlet of the blower is easily blocked or experiences drastic changes in air pressure, causing significant fluctuations in the pneumatic load torque and resulting in a corresponding increase in the bus current. Similarly, when an early inter-turn short circuit occurs in the motor stator winding, it also leads to an increase in current, but the resulting subtle changes in electrical characteristics are completely masked by the current changes caused by the pneumatic load fluctuations. Because existing technology does not separate pneumatic load disturbances from electrical characteristics, it is highly prone to triggering false overcurrent alarms during sudden load changes, while failing to detect early inter-turn short circuits when their characteristics are weak, thus failing to reliably identify early electrical faults. Summary of the Invention

[0004] The purpose of this invention is to provide a method for intelligent monitoring and fault early warning of the operating status of a hair dryer motor, which can solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A method for intelligent monitoring and fault early warning of the operating status of a hair dryer motor includes: acquiring phase current and bus voltage signals of the brushless DC motor of the hair dryer; calculating commutation intervals and non-commutation intervals based on the phase current and bus voltage signals; constructing a stator current dynamic equation based on a back EMF observer within the non-commutation interval of the commutation interval; introducing the aerodynamic load torque of the air duct of the hair dryer as an unknown disturbance term, and designing a sliding mode observer to dynamically decouple the unknown disturbance term from the back EMF; based on the result of dynamic decoupling, extracting the low-frequency asymmetric distortion component of the back EMF and the high-frequency ripple variance of the phase current in the non-commutation interval; when the low-frequency asymmetric distortion component and the high-frequency ripple variance exceed a dynamic reference threshold, and the duration of exceeding the dynamic reference threshold satisfies a preset time window constraint, outputting an early warning signal for inter-turn short circuit and blocking the motor drive pulse of the brushless DC motor.

[0007] Preferably, the step of calculating the commutation interval and non-commutation interval based on the phase current and bus voltage signals includes: detecting the zero-crossing point of the phase current and the slope change point of the bus voltage; aligning and cross-validating the zero-crossing point and the slope change point on the time axis to determine the commutation start time of the brushless DC motor; after the commutation start time, monitoring the moment when the rising edge of the phase current reaches the steady-state peak value as the commutation end time; defining the interval between two adjacent commutation start times as the commutation interval, defining the interval between the commutation end time and the next commutation start time as the non-commutation interval, and obtaining the commutation overlap angle by calculating the time difference between the commutation start time and the commutation end time.

[0008] Preferably, the step of constructing the stator current dynamic equation based on the back EMF observer in the non-commutation interval of the commutation interval includes: transforming the phase voltage and phase current in the three-phase stationary coordinate system of the brushless DC motor to the two-phase rotating coordinate system to obtain the d-axis voltage component, q-axis voltage component, d-axis current component, and q-axis current component; decomposing the back EMF into harmonic components on the d-axis voltage component and the q-axis voltage component; introducing a stator resistance dynamic compensation term that varies with temperature; and combining the differential relationship between the d-axis current component and the q-axis current component to establish the stator current dynamic equation in the non-commutation interval with the harmonic components and the stator resistance dynamic compensation term as state variables.

[0009] Preferably, the step of introducing the aerodynamic load torque of the blower duct as an unknown disturbance term and designing a sliding mode observer to dynamically decouple the unknown disturbance term from the back electromotive force includes: extending the aerodynamic load torque of the blower duct and its derivative with respect to time into an augmented state variable of the stator current dynamic equation, constructing an extended state equation containing the augmented state variable; designing a sliding surface based on the error between the phase current and the output current of the extended state equation, calculating a correction term for the augmented state variable according to the sign function of the sliding surface and a preset sliding gain; and using the correction term to perform negative feedback compensation on the augmented state variable, so that the observed value of the back electromotive force and the observed value of the aerodynamic load torque of the blower duct are dynamically decoupled.

[0010] Preferably, based on the results of dynamic decoupling, the steps of extracting the low-frequency asymmetric distortion component of the back electromotive force in the non-commutation interval and the high-frequency ripple variance of the phase current include: inputting the back electromotive force obtained by dynamic decoupling into a phase-locked loop, extracting the fundamental component of the back electromotive force, subtracting the back electromotive force from the fundamental component to obtain a residual signal, performing low-pass filtering on the residual signal to extract the low-frequency asymmetric distortion component; performing discrete wavelet decomposition on the phase current, extracting the detail coefficients of the phase current at a preset high-frequency scale, calculating the mean of the sum of squares of the detail coefficients in the non-commutation interval, and obtaining the high-frequency ripple variance of the phase current.

[0011] Preferably, the step of outputting an early warning signal for inter-turn short circuit and blocking the motor drive pulse of the brushless DC motor when the low-frequency asymmetric distortion component and the high-frequency ripple variance exceed the dynamic reference threshold and the duration of exceeding the dynamic reference threshold meets the preset time window constraint includes: collecting historical low-frequency asymmetric distortion components and historical high-frequency ripple variance of the brushless DC motor under normal fault-free operating conditions, and fitting a two-dimensional reference surface of the dynamic reference threshold with the rotor speed and bus voltage of the brushless DC motor as independent variables; starting a counter when the low-frequency asymmetric distortion component and the high-frequency ripple variance are both greater than the threshold corresponding to the two-dimensional reference surface; and outputting the early warning signal for inter-turn short circuit and blocking the motor drive pulse when the count value of the counter reaches the duration threshold of the preset time window constraint.

[0012] Preferably, the step after obtaining the commutation overlap angle by calculating the time difference between the commutation start time and the commutation end time further includes: injecting a square wave test voltage pulse into the non-commutation interval when the rotor speed of the brushless DC motor exceeds a preset high-speed threshold; acquiring the response polarity reversal delay time of the phase current under the action of the square wave test voltage pulse, calculating the commutation delay compensation angle based on the response polarity reversal delay time and the injection time of the square wave test voltage pulse; correcting the commutation overlap angle using the commutation delay compensation angle, and redefining the start and end times of the non-commutation interval based on the corrected commutation overlap angle to eliminate the interference of the commutation freewheeling process on the stator current dynamic equation.

[0013] Preferably, the step of establishing the stator current dynamic equation with the harmonic components and the stator resistance dynamic compensation term as state variables in the non-commutation interval further includes: obtaining the mapping relationship between the eddy current loss of the stator core of the brushless DC motor and the rotor speed, and converting the mapping relationship into an additional damping current component in phase with the back electromotive force; superimposing the additional damping current component into the differential term of the q-axis current component in the stator current dynamic equation, and using the additional damping current component to perform feedforward compensation and cancellation on the back electromotive force coupling term in the stator current dynamic equation, thereby eliminating the attenuation effect of the stator core eddy current effect on the extraction accuracy of the low-frequency asymmetric distortion component.

[0014] Preferably, the step of using the correction term to perform negative feedback compensation on the augmented state variable to dynamically decouple the observed value of the back EMF from the observed value of the duct aerodynamic load torque further includes: calculating the high-frequency disturbance energy of the phase current in the non-commutation interval in real time, inputting the high-frequency disturbance energy into the fuzzy inference rule, and outputting the dynamic adjustment coefficient of the preset sliding mode gain according to the magnitude of the high-frequency disturbance energy; when the high-frequency disturbance energy increases with the early characteristics of the inter-turn short circuit, reducing the preset sliding mode gain through the dynamic adjustment coefficient to suppress the high-frequency chattering of the sliding mode observer during the fault transient process and maintain the decoupling stability between the observed value of the duct aerodynamic load torque and the observed value of the back EMF.

[0015] Preferably, the steps of low-pass filtering the residual signal to extract the low-frequency asymmetric distortion component and calculating the mean of the sum of squares of the detail coefficients in the non-commutation interval to obtain the high-frequency ripple variance of the phase current further include: obtaining the physical resonant frequency band corresponding to the inter-turn short circuit of the stator winding of the brushless DC motor; setting the target decomposition scale of the discrete wavelet decomposition based on the physical resonant frequency band; extracting the detail coefficients of two adjacent commutation cycles in the non-commutation interval under the target decomposition scale; calculating the cross-correlation difference of the detail coefficients of two adjacent commutation cycles; using the cross-correlation difference as a weighting factor to weight and correct the high-frequency ripple variance; and using the corrected high-frequency ripple variance to update the judgment conditions for the early warning signal of the inter-turn short circuit.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0017] 1. This invention constructs a stator current dynamic equation based on a back EMF observer in the non-commutation interval, introduces the aerodynamic load torque of the duct as an unknown disturbance term, designs a sliding mode observer to dynamically decouple the unknown disturbance term from the back EMF, extracts the low-frequency asymmetric distortion component of the back EMF and the high-frequency ripple variance of the phase current, and combines a dynamic reference threshold and time window constraints to output early warnings, thus solving the problem of being unable to distinguish between duct load fluctuations and early electrical faults. Decoupling and separating the aerodynamic load as an unknown disturbance eliminates the interference of duct operating condition changes on electrical feature extraction, allowing the extracted low-frequency asymmetric distortion component and high-frequency ripple variance to directly reflect the stator winding state, achieving the identification of early inter-turn short-circuit faults and avoiding false alarms caused by sudden load changes and missed alarms caused by weak fault characteristics.

[0018] 2. This invention corrects the commutation overlap angle by injecting a square wave test voltage pulse during the high-speed phase and collecting the response polarity reversal delay time to calculate the commutation delay compensation angle, thus eliminating the interference of the commutation freewheeling process on the stator current dynamic equation. It also equates the stator core eddy current loss to an additional damping current component superimposed on the dynamic equation for feedforward compensation and cancellation, eliminating the attenuation effect of eddy current on the extraction accuracy of low-frequency asymmetric distortion components. Furthermore, it dynamically adjusts the sliding mode gain of the sliding mode observer based on the high-frequency disturbance energy of the phase current using fuzzy inference rules, suppressing high-frequency chattering during fault transients and maintaining the decoupling stability between the aerodynamic load torque and the back EMF observation value. Finally, it sets the target decomposition scale of the discrete wavelet decomposition according to the physical resonant frequency band of the inter-turn short circuit, and uses the cross-correlation difference of the detail coefficients of adjacent commutation cycles as a weighting factor to weight and correct the high-frequency ripple variance, thereby improving the signal-to-noise ratio of fault feature extraction. Attached Figure Description

[0019] Figure 1The flowchart below shows the overall implementation process of the intelligent monitoring and fault early warning of the brushless DC motor operation status of the hair dryer according to the present invention.

[0020] Figure 2 This is a flowchart of the calculation of the commutation interval and non-commutation interval of the brushless DC motor and the acquisition of the commutation overlap angle according to the present invention.

[0021] Figure 3 This is a flowchart illustrating the construction of the stator current dynamic equation and the feedforward compensation of stator core eddy current loss in this invention.

[0022] Figure 4 This is a flowchart illustrating the design of the sliding mode observer and the dynamic decoupling of unknown disturbances and back EMF in this invention.

[0023] Figure 5 This is a flowchart of the extraction and weighted correction of the back electromotive force distortion component and phase current ripple variance in this invention.

[0024] Figure 6 The flowchart shows the dynamic reference threshold determination, inter-turn short circuit early warning, and motor drive pulse interlocking of the present invention. Detailed Implementation

[0025] refer to Figure 1 and Figure 2 In one embodiment, the three-phase current signal and bus voltage signal of the brushless DC motor of the hair dryer are acquired at a preset sampling frequency. The sampling frequency meets the requirements of the Nyquist sampling theorem, which can completely preserve the effective frequency components in the signal. Based on the phase current and bus voltage signals, the commutation interval and non-commutation interval are calculated. The zero-crossing point of any phase current changing from positive to negative or from negative to positive is detected, and the slope abrupt change point where the absolute value of the first derivative of the bus voltage exceeds a preset tolerance threshold is detected. The zero-crossing point and slope abrupt change point detected in the same commutation cycle are aligned on the time axis, and the time difference between the two is calculated. When the time difference is less than the preset time tolerance, the average time of the two is taken as the commutation start time of the commutation cycle. After the commutation start time, the rising edge change of the phase current is continuously monitored. When the rate of change of the phase current in multiple consecutive sampling points is less than the preset steady-state threshold, it is determined that the phase current has reached the steady-state peak value, and this moment is recorded as the commutation end time. The interval between two adjacent commutation start times is defined as the commutation interval, and the interval between the commutation end time and the next commutation start time is defined as the non-commutation interval. The commutation overlap angle is obtained by calculating the time difference between the commutation start time and the commutation end time.

[0026] refer to Figure 3 Within the non-commutation interval of the commutation interval, a dynamic equation for the stator current based on a back EMF observer is constructed. The phase voltages and currents in the three-phase stationary coordinate system are transformed to the two-phase stationary coordinate system using the Clarke transformation. The transformation process satisfies the following mathematical relationships:

[0027]

[0028] in, , , These represent the current components of phase a, phase b, and phase c in a three-phase stationary coordinate system. , They are respectively in two stationary coordinate systems axis, Axial current components. The voltage and current components in a two-phase stationary coordinate system are transformed to a two-phase rotating coordinate system using the Park transformation. The transformation process satisfies the following mathematical relationships:

[0029]

[0030] in, This refers to the rotor position angle of the brushless DC motor. , These represent the d-axis and q-axis current components in a two-phase rotating coordinate system. The d-axis voltage component is obtained through the same transformation process. With q-axis voltage component The back electromotive force is decomposed into d-axis voltage components and harmonic components on the q-axis voltage components. A dynamic compensation term for stator resistance that varies with temperature is introduced. The relationship between stator resistance and temperature satisfies the following mathematical relationship:

[0031]

[0032] in, Reference temperature The stator resistance value below, The temperature coefficient of resistance of the stator winding. The current is the real-time temperature of the stator winding. Based on the differential relationship between the d-axis and q-axis current components, a dynamic equation for the stator current in the non-commutation region is established, with harmonic components and stator resistance dynamic compensation terms as state variables:

[0033]

[0034] in, , These are the d-axis and q-axis inductances of a brushless DC motor. The rotor electrical angular velocity of the brushless DC motor. , These are the harmonic components of the back electromotive force on the d-axis and q-axis, respectively.

[0035] refer to Figure 4The aerodynamic load torque of the blower duct is introduced as an unknown disturbance term, and a sliding mode observer is designed to dynamically decouple the unknown disturbance term from the back electromotive force. The aerodynamic load torque of the blower duct is then used as an unknown disturbance term. and its derivative with respect to time Extending the dynamic equations of the stator current to include augmented state variables, we construct extended state equations containing these augmented state variables. The state vector is defined as follows: The extended state equations satisfy the following mathematical relations:

[0036]

[0037] in, The system state matrix, For the input matrix, Here is the perturbation matrix. The system input voltage vector, To address system process noise, a sliding surface is designed based on the error between the actual phase current and the output current in the extended state equation. The sliding surface satisfies the following mathematical relationship:

[0038]

[0039] in, , These are the observed values ​​of the d-axis and q-axis currents, respectively. Based on the sign function of the sliding surface and the preset sliding gain, the correction terms for the augmented state variables are calculated. The state update equation of the sliding observer satisfies the following mathematical relationship:

[0040]

[0041] in, To augment the observed values ​​of the state variables, To preset the sliding mode gain matrix, The sign function is used. The augmented state variables are compensated using a correction term with negative feedback, when the sliding surface... When the current approaches zero, the current observation error approaches zero, and the observed value of the augmented state variable converges to the actual value, thus achieving dynamic decoupling between the observed value of the back electromotive force and the observed value of the aerodynamic load torque of the duct.

[0042] refer to Figure 5 Based on the results of dynamic decoupling, the low-frequency asymmetric distortion component of the back electromotive force (EMF) and the high-frequency ripple variance of the phase current in the non-commutation region are extracted. The back EMF signal obtained by dynamic decoupling is input into the phase-locked loop (PLL), and the phase update equation of the PLL satisfies the following mathematical relationship:

[0043]

[0044] in, The phase angle of the phase-locked loop output. The center angular frequency of the phase-locked loop. , These are the proportional gain and integral gain of the phase-locked loop, respectively. This is the phase error signal. The fundamental component of the back electromotive force output by the phase-locked loop. The residual signal is obtained by subtracting the original back electromotive force signal from the fundamental component. The residual signal is subjected to a low-pass filter with a cutoff frequency of 3 times the fundamental frequency to extract the low-frequency asymmetric distortion component. Discrete wavelet decomposition is performed on the phase current signal, using the db4 wavelet as the mother wavelet, decomposing it to a preset high-frequency scale, and extracting the detail coefficients at this scale. Calculate the mean of the sum of squares of the detail coefficients in the non-commutation interval to obtain the high-frequency ripple variance of the phase current:

[0045]

[0046] in, This represents the number of sampling points within the non-commutation interval. The preset high-frequency scale for discrete wavelet decomposition.

[0047] refer to Figure 6 When the low-frequency asymmetric distortion component and the high-frequency ripple variance exceed the dynamic reference threshold, and the duration of exceeding the dynamic reference threshold meets the preset time window constraint, an early warning signal for inter-turn short circuit is output, and the motor drive pulse of the brushless DC motor is blocked. Historical low-frequency asymmetric distortion components and historical high-frequency ripple variance of the brushless DC motor under normal, fault-free operating conditions are collected, with the rotor speed of the brushless DC motor as the reference value. and bus voltage Using the least squares method as the independent variable, a two-dimensional reference surface for the dynamic reference threshold is generated through fitting.

[0048]

[0049] in, to These are the fitting coefficients for the two-dimensional reference surface. Two-dimensional reference surfaces for the low-frequency asymmetric distortion component and the high-frequency ripple variance are generated separately. When both the low-frequency asymmetric distortion component and the high-frequency ripple variance are simultaneously greater than the threshold corresponding to the two-dimensional reference surface, a counter is started. The counter increments in units of sampling periods. If the counter value reaches the duration threshold of the preset time window constraint, an early warning signal for inter-turn short circuit is output, and the motor drive pulse is blocked.

[0050] In this embodiment, the comparison of characteristic parameters under normal operating conditions and early inter-turn short circuit conditions is shown in Table 1.

[0051] Table 1 Comparison of characteristic parameters under normal operating conditions and early inter-turn short circuit conditions

[0052]

[0053] In Table 1, under normal operating conditions, both the low-frequency asymmetric distortion component and the high-frequency ripple variance are below the corresponding dynamic reference thresholds, and no warning is triggered. Under early inter-turn short circuit conditions, both characteristic parameters exceed the dynamic reference thresholds simultaneously, and a warning signal is triggered when the duration meets the preset time window constraint. This embodiment dynamically decouples the aerodynamic load torque of the duct as an unknown disturbance term, thus eliminating the influence of load fluctuations on electrical characteristics. This allows the extracted characteristic parameters to directly reflect the operating state of the stator winding, enabling the identification of early inter-turn short circuit faults.

[0054] In a preferred embodiment, after obtaining the commutation overlap angle by calculating the time difference between the commutation start and end times, a square wave test voltage pulse is injected into the non-commutation interval when the rotor speed of the brushless DC motor exceeds a preset high-speed threshold. The width of the square wave test voltage pulse is set to 1 / 10 of the duration of the non-commutation interval, and the amplitude is the same as the current bus voltage. The injection time is selected at the beginning of the non-commutation interval. The phase current response under the action of the square wave test voltage pulse is collected, and the time interval from the injection time of the square wave test voltage pulse to the reversal of the phase current polarity is recorded as the response polarity reversal delay time. The commutation delay compensation angle is calculated based on the response polarity reversal delay time and the injection time of the square wave test voltage pulse. The commutation delay compensation angle satisfies the following mathematical relationship:

[0055]

[0056] in, The current rotor electrical angular velocity is given. The commutation overlap angle is corrected using the commutation delay compensation angle. The corrected commutation overlap angle is the difference between the original commutation overlap angle and the commutation delay compensation angle. Based on the corrected commutation overlap angle, the start and end times of the non-commutation interval are redefined. The start time of the non-commutation interval is delayed by the time length corresponding to the commutation delay compensation angle to eliminate the interference of the commutation freewheeling process on the stator current dynamic equation. The commutation freewheeling process will cause a peak in the phase current for a short time after the commutation ends. This peak is not a steady-state component of the stator current and will affect the accuracy of the stator current dynamic equation. By correcting the commutation overlap angle and redefining the non-commutation interval, the influence of the commutation freewheeling peak can be avoided.

[0057] In establishing the dynamic equation of the stator current in the non-commutation interval, with harmonic components and stator resistance dynamic compensation terms as state variables, the mapping relationship between the eddy current loss of the stator core of the brushless DC motor and the rotor speed is obtained. The eddy current loss of the stator core is proportional to the square of the rotor speed, satisfying the following mathematical relationship:

[0058]

[0059] in, The eddy current loss coefficient of the stator core is obtained through the motor's factory no-load test. The eddy current loss mapping relationship is equivalent to an additional damped current component in phase with the back electromotive force. This additional damped current component satisfies the following mathematical relationship:

[0060]

[0061] in, Let be the fundamental amplitude of the back electromotive force. By superimposing the additional damping current component into the differential term of the q-axis current component in the stator current dynamic equation, the corrected stator current dynamic equation is:

[0062]

[0063] By using an additional damped current component to feedforward compensate and cancel the back electromotive force coupling term in the stator current dynamic equation, the attenuation effect of the stator core eddy current effect on the extraction accuracy of the low-frequency asymmetric distortion component is eliminated. The stator core eddy current effect causes phase shift and amplitude attenuation in the back electromotive force, resulting in a deviation between the extracted low-frequency asymmetric distortion component and the actual value. By using a feedforward compensated additional damped current component, the back electromotive force change caused by the eddy current effect can be offset, thereby improving the extraction accuracy of the low-frequency asymmetric distortion component.

[0064] In this embodiment, the comparison of the commutation delay compensation angle and commutation overlap angle correction values ​​at different rotor speeds is shown in Table 2.

[0065] Table 2 Comparison of Commutation Delay Compensation Angle and Commutation Overlap Angle Correction Values ​​at Different Rotor Speeds

[0066]

[0067] Table 2 shows the correction values ​​of commutation delay compensation angle and commutation overlap angle at different rotor speeds. As the rotor speed increases, the commutation delay compensation angle gradually increases, and the difference between the corrected commutation overlap angle and the original commutation overlap angle also gradually increases. In this embodiment, the commutation overlap angle is corrected by injecting a square wave test voltage pulse during the high-speed stage, eliminating the interference of the commutation freewheeling process. At the same time, the influence of the stator core eddy current effect is eliminated by adding damping current component through feedforward compensation, thereby improving the accuracy of the stator current dynamic equation and the extraction accuracy of characteristic parameters.

[0068] In another preferred embodiment, a correction term is used to perform negative feedback compensation on the augmented state variable, enabling dynamic decoupling between the observed back EMF and the observed aerodynamic load torque of the duct. During this process, the high-frequency disturbance energy of the phase current in the non-commutation range is calculated in real time. The high-frequency disturbance energy is obtained by calculating the sum of squares after bandpass filtering the phase current signal; the passband range of the bandpass filter is 1kHz to 10kHz. The high-frequency disturbance energy is input into the fuzzy inference rule, which outputs a dynamic adjustment coefficient for the preset sliding mode gain based on the magnitude of the high-frequency disturbance energy. The fuzzy inference rule divides the high-frequency disturbance energy into three fuzzy subsets (low, medium, and high) and the dynamic adjustment coefficient into three fuzzy subsets (large, medium, and small). A triangular membership function is used to describe the membership degree of each fuzzy subset.

[0069] In this embodiment, the fuzzy inference rules for dynamic adjustment of sliding mode gain are shown in Table 3.

[0070] Table 3. Fuzzy Inference Rules for Dynamic Adjustment of Sliding Mode Gain

[0071]

[0072] In Table 3, a large dynamic adjustment coefficient is output when the high-frequency disturbance energy is low; a medium dynamic adjustment coefficient is output when the high-frequency disturbance energy is medium; and a small dynamic adjustment coefficient is output when the high-frequency disturbance energy is high. The preset sliding mode gain is adjusted based on the dynamic adjustment coefficient output by the fuzzy inference rules. The adjusted sliding mode gain satisfies the following mathematical relationship:

[0073]

[0074] in, To preset the sliding mode gain, This is a dynamic adjustment coefficient. When the high-frequency disturbance energy increases with the early characteristics of the inter-turn short circuit, the preset sliding mode gain is reduced by the dynamic adjustment coefficient to suppress high-frequency chattering of the sliding mode observer during the fault transient process, maintaining the decoupling stability between the observed values ​​of the aerodynamic load torque and the back EMF. High-frequency chattering of the sliding mode observer causes fluctuations in the observed values ​​of back EMF and load torque, affecting the accuracy of characteristic parameter extraction. By dynamically adjusting the sliding mode gain, the generation of high-frequency chattering can be suppressed while ensuring the convergence speed of the observer.

[0075] In the process of extracting low-frequency asymmetric distortion components by low-pass filtering the residual signal and calculating the mean of the sum of squares of the detail coefficients in the non-commutation interval to obtain the high-frequency ripple variance of the phase current, the physical resonant frequency band corresponding to the inter-turn short circuit of the stator winding of the brushless DC motor is obtained. The physical resonant frequency of the stator winding inter-turn short circuit is determined by the inductance and distributed capacitance of the winding, and is usually in the 1kHz to 5kHz frequency band. Based on this physical resonant frequency band, the target decomposition scale of discrete wavelet decomposition is set so that the frequency range corresponding to the target decomposition scale matches the physical resonant frequency band. At the target decomposition scale, the detail coefficients of two adjacent commutation cycles in the non-commutation interval are extracted. and Calculate the cross-correlation difference of detail coefficients between two adjacent commutation cycles:

[0076]

[0077] in, This represents the number of sampling points within the non-commutation interval. The cross-correlation difference is used as a weighting factor to weight and correct the high-frequency ripple variance. The corrected high-frequency ripple variance satisfies the following mathematical relationship:

[0078]

[0079] in, The weighting coefficients are used to update the criteria for early warning signals of inter-turn short circuits. When the corrected high-frequency ripple variance and the low-frequency asymmetric distortion component simultaneously exceed the dynamic reference threshold, and the duration meets the preset time window constraint, an early warning signal for inter-turn short circuits is output, and the motor drive pulse is blocked. The cross-correlation difference of the detail coefficients of adjacent commutation cycles reflects the degree of change in high-frequency ripple. When an inter-turn short circuit occurs, the high-frequency ripple will show obvious periodic changes, and the cross-correlation difference will increase. Weighted correction can improve the sensitivity of high-frequency ripple variance to early inter-turn short circuit faults. In this embodiment, the high-frequency chattering of the sliding mode observer is suppressed by dynamically adjusting the sliding mode gain, and the signal-to-noise ratio of the fault characteristics is improved by weighted correction of the high-frequency ripple variance, further enhancing the reliability of early warning signals for inter-turn short circuits.

Claims

1. A method for intelligent monitoring and fault early warning of the operating status of a hair dryer motor, characterized in that, include: The phase current and bus voltage signals of the brushless DC motor of the blower are collected, and the commutation interval and non-commutation interval are calculated based on the phase current and bus voltage signals. Within the non-commutation interval of the commutation interval, a stator current dynamic equation based on a back EMF observer is constructed. The aerodynamic load torque of the blower duct is introduced as an unknown disturbance term, and a sliding mode observer is designed to dynamically decouple the unknown disturbance term from the back electromotive force. Based on the results of dynamic decoupling, the low-frequency asymmetric distortion component of the back electromotive force in the non-commutation interval and the high-frequency ripple variance of the phase current are extracted. When the low-frequency asymmetric distortion component and the high-frequency ripple variance exceed the dynamic reference threshold, and the duration of exceeding the dynamic reference threshold satisfies the preset time window constraint, an early warning signal for inter-turn short circuit is output, and the motor drive pulse of the brushless DC motor is blocked.

2. The intelligent monitoring and fault early warning method for the operating status of a hair dryer motor according to claim 1, characterized in that, The steps of calculating the commutation interval and non-commutation interval based on the phase current and bus voltage signals include: detecting the zero-crossing point of the phase current and the slope change point of the bus voltage, aligning and cross-validating the zero-crossing point and the slope change point on the time axis, and determining the commutation start time of the brushless DC motor. After the commutation start time, the moment when the rising edge of the phase current reaches the steady-state peak value is used as the commutation end time. The interval between two adjacent commutation start times is defined as the commutation interval, and the interval between the commutation end time and the next commutation start time is defined as the non-commutation interval. The commutation overlap angle is obtained by calculating the time difference between the commutation start time and the commutation end time.

3. The intelligent monitoring and fault early warning method for the operating status of a hair dryer motor according to claim 1, characterized in that, In the non-commutation interval of the commutation interval, the step of constructing the stator current dynamic equation based on the back EMF observer includes: transforming the phase voltage and phase current in the three-phase stationary coordinate system of the brushless DC motor to the two-phase rotating coordinate system to obtain the d-axis voltage component, q-axis voltage component, d-axis current component and q-axis current component. The back electromotive force is decomposed into d-axis voltage components and harmonic components on the q-axis voltage components. A dynamic compensation term for stator resistance that varies with temperature is introduced. Combining the differential relationship between the d-axis current components and the q-axis current components, the dynamic equation of the stator current in the non-commutation interval is established with the harmonic components and the dynamic compensation term for stator resistance as state variables.

4. The intelligent monitoring and fault early warning method for the operating status of a hair dryer motor according to claim 1, characterized in that, The steps of introducing the air duct aerodynamic load torque of the blower as an unknown disturbance term and designing a sliding mode observer to dynamically decouple the unknown disturbance term from the back electromotive force include: extending the air duct aerodynamic load torque and its derivative with respect to time into an augmented state variable of the stator current dynamic equation, and constructing an extended state equation containing the augmented state variable. A sliding surface is designed based on the error between the phase current and the output current of the extended state equation, and the correction term of the augmented state variable is calculated based on the sign function of the sliding surface and the preset sliding gain. The augmented state variable is compensated by negative feedback using the correction term, so that the observed value of the back electromotive force is dynamically decoupled from the observed value of the aerodynamic load torque of the duct.

5. The intelligent monitoring and fault early warning method for the operating status of a hair dryer motor according to claim 4, characterized in that, Based on the results of dynamic decoupling, the steps of extracting the low-frequency asymmetric distortion component of the back electromotive force in the non-commutation interval and the high-frequency ripple variance of the phase current include: inputting the back electromotive force obtained by dynamic decoupling into a phase-locked loop, extracting the fundamental component of the back electromotive force, subtracting the back electromotive force from the fundamental component to obtain a residual signal, and performing low-pass filtering on the residual signal to extract the low-frequency asymmetric distortion component. Discrete wavelet decomposition is performed on the phase current to extract the detail coefficients of the phase current at a preset high-frequency scale. The mean of the sum of squares of the detail coefficients in the non-commutation interval is calculated to obtain the high-frequency ripple variance of the phase current.

6. The intelligent monitoring and fault early warning method for the operating status of a hair dryer motor according to claim 1, characterized in that, When the low-frequency asymmetric distortion component and the high-frequency ripple variance exceed the dynamic reference threshold, and the duration of exceeding the dynamic reference threshold meets the preset time window constraint, the step of outputting an early warning signal for inter-turn short circuit and blocking the motor drive pulse of the brushless DC motor includes: collecting the historical low-frequency asymmetric distortion component and historical high-frequency ripple variance of the brushless DC motor under normal fault-free operating conditions, and using the rotor speed and bus voltage of the brushless DC motor as independent variables to fit and generate a two-dimensional reference surface of the dynamic reference threshold; When the low-frequency asymmetric distortion component and the high-frequency ripple variance are both greater than the threshold corresponding to the two-dimensional reference surface, the counter is started. If the count value of the counter reaches the duration threshold of the preset time window constraint, the early warning signal of the inter-turn short circuit is output and the motor drive pulse is blocked.

7. The intelligent monitoring and fault early warning method for the operating status of a hair dryer motor according to claim 2, characterized in that, The step of obtaining the commutation overlap angle by calculating the time difference between the commutation start time and the commutation end time further includes: injecting a square wave test voltage pulse in the non-commutation interval when the rotor speed of the brushless DC motor exceeds a preset high-speed threshold. The response polarity reversal delay time of the phase current under the action of the square wave test voltage pulse is collected, and the commutation delay compensation angle is calculated based on the response polarity reversal delay time and the injection time of the square wave test voltage pulse. The commutation delay compensation angle is used to correct the commutation overlap angle, and the start and end times of the non-commutation interval are redefined based on the corrected commutation overlap angle to eliminate the interference of the commutation freewheeling process on the stator current dynamic equation.

8. The intelligent monitoring and fault early warning method for the operating status of a hair dryer motor according to claim 3, characterized in that, The step of establishing the stator current dynamic equation in the non-commutation interval with the harmonic components and the stator resistance dynamic compensation term as state variables further includes: obtaining the mapping relationship of the eddy current loss of the stator core of the brushless DC motor as a function of the rotor speed, and converting the mapping relationship into an additional damping current component in phase with the back electromotive force. The additional damping current component is superimposed on the differential term of the q-axis current component in the stator current dynamic equation. The additional damping current component is used to feedforward compensate and cancel the back electromotive force coupling term in the stator current dynamic equation, thereby eliminating the attenuation effect of the stator core eddy current effect on the extraction accuracy of the low-frequency asymmetric distortion component.

9. The intelligent monitoring and fault early warning method for the operating status of a hair dryer motor according to claim 4, characterized in that, The step of using the correction term to perform negative feedback compensation on the augmented state variable to dynamically decouple the observed value of the back electromotive force from the observed value of the aerodynamic load torque of the duct, further includes: calculating the high-frequency disturbance energy of the phase current in the non-commutation interval in real time, inputting the high-frequency disturbance energy into the fuzzy inference rule, and outputting the dynamic adjustment coefficient of the preset sliding mode gain according to the magnitude of the high-frequency disturbance energy. When the high-frequency disturbance energy increases with the early characteristics of the inter-turn short circuit, the preset sliding mode gain is reduced by the dynamic adjustment coefficient to suppress the high-frequency chattering of the sliding mode observer during the fault transient process, thereby maintaining the decoupling stability between the observed value of the duct aerodynamic load torque and the observed value of the back electromotive force.

10. The intelligent monitoring and fault early warning method for the operating status of a hair dryer motor according to claim 5, characterized in that, The steps of low-pass filtering the residual signal to extract the low-frequency asymmetric distortion component and calculating the mean of the sum of squares of the detail coefficients in the non-commutation interval to obtain the high-frequency ripple variance of the phase current further include: obtaining the physical resonant frequency band corresponding to the inter-turn short circuit of the stator winding of the brushless DC motor, and setting the target decomposition scale of the discrete wavelet decomposition based on the physical resonant frequency band. At the target decomposition scale, the detail coefficients of two adjacent commutation cycles within the non-commutation interval are extracted, and the cross-correlation difference of the detail coefficients of two adjacent commutation cycles is calculated. The cross-correlation difference is used as a weighting factor to correct the high-frequency ripple variance, and the corrected high-frequency ripple variance is used to update the judgment criteria for the early warning signal of the inter-turn short circuit.

Images