Real-time vibration compensation control method for bearingless composite rotor cage type asynchronous motor

Abstract

The invention relates to the technical field of motor control, and discloses a bearingless composite rotor cage type asynchronous motor vibration real-time compensation control method, which comprises the following steps: acquiring and preprocessing a rotor displacement signal to obtain a filtered displacement signal; processing the filtered displacement signal based on a second-order generalized integrator, and extracting a fundamental frequency harmonic component synchronous with the mechanical angular velocity of the rotor; adopting a simplified LMS algorithm to dynamically optimize the amplitude weight and the phase weight of the compensation force, and generating a feedforward compensation force according to the optimized weights; a two-channel control framework is constructed, and total control force is synthesized; the suspension force winding is driven by the inverter to generate corresponding radial suspension force, so that unbalance vibration of the rotor is counteracted; according to the invention, the second-order generalized integrator harmonic observer is adopted to realize high-precision extraction of the fundamental frequency component, and the mechanical angular velocity change of the rotor can be tracked in real time in cooperation with a frequency adaptive adjustment mechanism.

Description

Technical Field

[0001] The present invention relates to the technical field of motor control, and more particularly to a real-time vibration compensation control method for a bearingless composite rotor cage-type asynchronous motor. Background Art

[0002] Currently, there are several technical solutions for vibration suppression control of bearingless composite rotor cage asynchronous motors:

[0003] A proportional-integral-differential controller is used to feedback control the motor's vibration displacement signal, adjusting the compensation force in real time based on the displacement error. This method is simple in principle and easy to implement, but parameter tuning relies on experience, making it difficult to adapt to changes in motor speed and load disturbances, and its suppression effect on fundamental frequency vibration is limited. The vibration signal is transformed into the frequency domain using an FFT algorithm to extract the main frequency components, and then a corresponding compensation force is generated for a specific frequency. This method can effectively identify the frequency characteristics of vibration, but the FFT calculation is complex and has poor real-time performance. It also suffers from windowing effects and spectral leakage, which affect the accuracy of vibration suppression. A mathematical model of rotor unbalance vibration is established, and the model parameters are identified online to generate a compensation force equal to and opposite to the vibration force. This method does not rely on feedback signals from the displacement sensor, which can reduce the impact of system delays. However, it requires high model accuracy, the parameter identification process is complex, and the anti-disturbance capability is insufficient. A bandpass filter and a notch filter are combined to extract vibration characteristics, while a combined feedforward and feedback control strategy is adopted. This method works well under specific operating conditions, but the filter parameters are fixed, making it difficult to adapt to speed fluctuations, and the control algorithm is complex.

[0004] The existing technical solutions have the following problems:

[0005] The extraction of vibration harmonic components mainly relies on methods such as FFT or bandpass filters. The FFT algorithm requires a long data window, resulting in poor real-time performance; the center frequency of the bandpass filter is fixed and it is difficult to adapt to changes in motor speed. When the rotor mechanical angular velocity changes dynamically, these methods cannot accurately extract the fundamental frequency vibration components synchronized with the speed, reducing the vibration suppression effect. The root cause of this problem is that the traditional harmonic extraction algorithm lacks a frequency adaptive mechanism and cannot track speed changes in real time; in traditional PID control and feedforward compensation control, parameter tuning mainly relies on experience or offline optimization, which is difficult to adapt to system state changes. Although the complex least mean square algorithm has adaptive capabilities, the real-time calculation of multi-dimensional weight vectors leads to high computational complexity, which is not conducive to embedded system implementation. This problem stems from the lack of an efficient parameter adaptive mechanism in the control algorithm and the inability to track speed changes in real time. High-precision parameter optimization can be achieved under conditions of limited computing resources; vibration suppression and dynamic response are difficult to balance. Existing control methods often focus on the single optimization of static vibration suppression or dynamic response performance, making it difficult to achieve an effective balance between the two. Although the simple feedforward compensation method can effectively suppress fundamental frequency vibration, it has poor adaptability to system parameter changes and load disturbances; pure feedback control is difficult to respond quickly to high-frequency vibrations due to system delay limitations. The fundamental reason for this contradiction is that the control architecture design fails to effectively coordinate the coordination mechanism of feedforward compensation and feedback correction; the existing technology has a significantly reduced vibration suppression effect when facing operating conditions such as speed fluctuations and sudden load changes. Especially under high-speed rotation conditions, the centrifugal force increases with the square of the speed, making the system more sensitive to parameter changes. The lack of effective adaptive mechanism and anti-disturbance design has become a key factor limiting the robustness of the system.

[0006] In summary, how to design a vibration suppression control method with high computational efficiency, high precision, and strong real-time performance, which can accurately identify and compensate for fundamental frequency vibration under various speed and load conditions, while taking into account the system's dynamic response performance and steady-state accuracy, and improving the system's robustness to external disturbances and parameter changes. Summary of the Invention

[0007] The present invention provides a real-time vibration compensation control method for a bearingless composite rotor cage asynchronous motor, which solves the technical problems of insufficient fundamental frequency vibration suppression accuracy, difficulty in optimizing control parameters, difficulty in balancing vibration suppression and dynamic response, and insufficient system robustness during the operation of the bearingless composite rotor cage asynchronous motor.

[0008] The present invention provides a real-time vibration compensation control method for a bearingless composite rotor cage asynchronous motor, comprising:

[0009] Collect the rotor displacement signal and pre-process it to obtain a filtered displacement signal;

[0010] The filtered displacement signal is processed based on a second-order generalized integrator to extract a fundamental harmonic component synchronized with the mechanical angular velocity of the rotor;

[0011] Based on the extracted fundamental harmonic component, the amplitude weight and the phase weight of the compensation force are dynamically optimized using a simplified LMS algorithm, and the feedforward compensation force is generated according to the optimized weights;

[0012] A double-channel control architecture is constructed, the feedforward compensation force generated based on the fundamental harmonic component is superimposed with the PID correction force output by the feedback correction channel to synthesize a total control force;

[0013] Based on the synthesized total control force, the corresponding radial suspension force is generated by driving the suspension force winding through the inverter to realize the cancellation of the rotor unbalanced vibration.

[0014] Further, the preprocessing step includes: collecting the displacement signals of the rotor in the x-axis and y-axis directions by the eddy current sensor, and filtering high-frequency noise and sensor interference by using a second-order Butterworth low-pass filter, and the cutoff frequency of the filter is set to 1.2 times the mechanical angular velocity of the rotor.

[0015] Further, the difference equation of the second-order Butterworth low-pass filter is:

[0016] v f (k1)=a1v f (k1-1))+a2v f (k1-2)+b0v(k1)+b1v(k1-1)+b2v(k1-2);

[0017] wherein v f (k1) is the filtered signal at the current time; v f (k1-1) is the filtered signal at the previous time; v f (k1-2) is the filtered signal at the previous two times; v(k1) is the original signal at the current time; v(k1-1) is the original signal at the previous time; v(k1-2) is the original signal at the previous two times; a1 is the filter autoregressive coefficient one, indicating the influence weight of the filtered signal at the previous time on the current output; a2 is the filter autoregressive coefficient two, indicating the influence weight of the filtered signal at the previous two times on the current output; b0 is the current input coefficient, indicating the influence weight of the original signal at the current time on the output; b1 is the previous input coefficient, indicating the influence weight of the original signal at the previous time on the output; b2 is the previous two input coefficients, indicating the influence weight of the original signal at the previous two times on the output.

[0018] Further, the transfer function model of the second-order generalized integrator includes in-phase component transfer function and quadrature component transfer function, and the damping coefficient is set to 0.8 to ensure the balance of system stability and response speed.

[0019] Furthermore, the second-order generalized integrator has a frequency adaptive adjustment mechanism, which updates the center frequency of the SOGI in real time according to the speed change, expressed as:

[0020] Center Frequency:

[0021]

[0022] where ω m is the rotor mechanical angular velocity; π is pi; n is the rotor speed; f m is the rotor mechanical frequency.

[0023] Furthermore, the simplified LMS algorithm only retains the amplitude weight k p and phase weight k i Two parameters, k is set when the algorithm is initialized p (0) = 0.5, k i (0)=0.1.

[0024] Furthermore, the parameter update law of the simplified LMS algorithm is:

[0025]

[0026] where k p (k3+1) is the amplitude weight at the next moment; k p (k3) is the amplitude weight at the current moment; k i (k3+1) is the phase weight at the next moment; k i (k3) is the phase weight at the current moment; μ p is the amplitude step factor; μ i is the phase step factor; e(k3) is the displacement error signal; v h (k3) is the fundamental frequency harmonic component.

[0027] Furthermore, the feedforward compensation force is generated as follows:

[0028]

[0029] Among them F comp (k3) is the feedforward compensation force; v p (k3) is the amplitude weight; k i (k3) is the phase weight; v h (k3) is the fundamental frequency harmonic component.

[0030] Furthermore, the feedback correction channel in the dual-channel control architecture uses an adaptive PID controller to process the residual error after feedforward compensation.

[0031] The present invention provides a computer storage medium comprising a memory and one or more processors, wherein the memory stores executable code, and when the one or more processors execute the executable code, they are used to execute the above-mentioned real-time vibration compensation control method for a bearingless composite rotor cage-type asynchronous motor.

[0032] The present invention has the following beneficial effects: a second-order generalized integrator is used to achieve high-precision extraction of the fundamental frequency component, and a frequency adaptive adjustment mechanism is used to track the change of the rotor's mechanical angular velocity in real time. The SOGI observer avoids the high computational complexity and window effect problems of the traditional FFT algorithm, effectively solving the technical problem of insufficient accuracy in harmonic component extraction.

[0033] By simplifying the least mean square algorithm, the multidimensional weight vector is reduced to two parameters: amplitude weight and phase weight, which reduces the computational burden. At the same time, a dynamically updated Butterworth low-pass filter is used for signal preprocessing, ensuring the efficient implementation of the algorithm in embedded systems.

[0034] This paper proposes a dual-channel control architecture that organically combines feedforward compensation with feedback correction. The feedforward channel uses SOGI harmonic observation and a simplified LMS algorithm to accurately offset the fundamental frequency vibration synchronized with the mechanical angular velocity; the feedback channel uses an adaptive PID controller to process residual errors, further improving the steady-state accuracy of the system.

[0035] The frequency adaptive adjustment mechanism of the present invention can update the center frequency of SOGI in real time according to the speed change. The adaptive PID controller responds to the system parameter changes through online parameter optimization, so that the control system maintains a stable vibration suppression effect under different speed and load conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a flow chart of a real-time vibration compensation control method for a bearingless composite rotor cage asynchronous motor in the present invention;

[0037] Figure 2 This is a CC-BIM diagram of the suspension force generation;

[0038] Figure 3 This is a schematic diagram of the CCR-BIM rotor eccentricity dynamic coordinate system;

[0039] Figure 4 It is a flow chart of harmonic component extraction and displacement error signal generation;

[0040] Figure 5 It is a flow chart of the suspension force cooperative adaptive compensation process with parameter tuning and closed-loop feedback;

[0041] Figure 6 It is the control block diagram of the CCR-BIM system;

[0042] Figure 7 is a schematic diagram of initial parameters of a control algorithm in an embodiment of the present application;

[0043] Figure 8 is a schematic diagram of parameters of a bearingless composite rotor cage asynchronous motor in an embodiment of the present application;

[0044] Figure 9 is a waveform diagram of a load speed experiment when a load is suddenly added;

[0045] Figure 10 is a comparative diagram of x-axis vibration displacement in an embodiment of the present application;

[0046] Figure 11 is a comparative diagram of y-axis vibration displacement in an embodiment of the present application;

[0047] Figure 12 is an experimental diagram of rotor shaft end trajectories under four compensation control methods. DETAILED DESCRIPTION

[0048] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that the discussion of these embodiments is merely meant to provide a better understanding of the subject matter described herein and can be changed in function and arrangement without departing from the scope of the present description. Various processes or components can be omitted, substituted, or added according to desired design. In addition, features described in some examples can be combined in other examples.

[0049] A vibration real-time compensation control method for a bearingless composite rotor cage asynchronous motor is disclosed in at least one embodiment of the present application, as shown in Figure 1 , comprising:

[0050] Step 1, collecting rotor displacement signals and pre-processing to obtain filtered displacement signals;

[0051] A bearingless composite rotor cage asynchronous motor (CCR-BIM, Composite Cage Rotor Bearingless Induction Motor) realizes the rotation and suspension functions of the rotor through the synergistic effect of torque windings and suspension force windings. As shown in Figure 2 , the CCR-BIM utilizes the similarity of the magnetic bearing and the stator structure of the asynchronous motor to embed a set of three-phase suspension windings and torque windings at the same time. The torque winding pole pair number p1=1, the suspension force winding pole pair number p2=1, and the two sets of windings satisfy the conditions p1=p2±1 and ω1=ω2, generating controllable radial suspension force.

[0052] In the operation process of the CCR-BIM, as shown in Figure 3As shown in the figure, the rotor manufacturing tolerance or assembly error causes a deviation Δr between the rotor mass center and the geometric center, causing unbalanced vibration. c The dynamic characteristics of the centrifugal force are described by the centrifugal force equation:

[0053]

[0054] Where m is the rotor mass; ω m is the rotor mechanical angular velocity; |Δr| is the rotor mass eccentricity; F c It is an unbalanced vibration force.

[0055] In the fixed coordinate system (x, y), the orthogonal components of the rotor unbalanced vibration force are:

[0056]

[0057] Among them F cx is the vibration force component in the x direction; F cy is the vibration force component in the y direction; F c is the unbalanced vibration force; ω m is the mechanical angular velocity of the rotor; t is the time; η is the angle between the line from the geometric center to the center of mass and the d-axis; cos is the cosine function; sin is the sine function.

[0058] The corresponding vibration displacement components are:

[0059]

[0060] where x c is the vibration displacement in the x direction; c is the vibration displacement in the y direction; A is the vibration amplitude; ω m is the mechanical angular velocity of the rotor; t is time; η is the angle between the line from the geometric center to the center of mass and the d-axis; γ is the phase lag angle between the vibration displacement and the vibration force; cos is the cosine function; sin is the sine function.

[0061] Based on the vibration mechanism analysis above, eddy current sensors are used to collect the rotor displacement signal v(k1) in the x- and y-axis directions. Because the raw signal contains high-frequency noise and sensor interference, it needs to be preprocessed through a second-order Butterworth low-pass filter.

[0062] The transfer function of the second-order Butterworth low-pass filter is:

[0063]

[0064] Among them H LPF (s) is the low-pass filter transfer function; ω m is the rotor mechanical angular velocity; s is the Laplace transform complex variable.

[0065] The filter cutoff frequency is set to 1.2ω m , which can suppress high-frequency noise and retain ω m The integrity of the fundamental frequency component. Discretize the transfer function into a difference equation:

[0066] v f (k1)=a1v f (k1-1)+a2v f (k1-2)+b0v(k1)+b1v(k1-1)+b2v(k1-2);

[0067] where v f (k1) is the filtered signal at the current moment; v f (k1-1) is the filtered signal at the previous moment; v f (k1-2) is the filtered signal at the previous two moments; v(k1) is the original signal at the current moment; v(k1-1) is the original signal at the previous moment; v(k1-2) is the original signal at the previous two moments; a1 is the autoregressive coefficient of the filter, which represents the weight of the influence of the filtered signal at the previous moment on the current output; a2 is the autoregressive coefficient of the filter, which represents the weight of the influence of the filtered signal at the previous two moments on the current output; b0 is the current input coefficient, which represents the weight of the influence of the original signal at the current moment on the output; b1 is the previous input coefficient, which represents the weight of the influence of the original signal at the previous moment on the output; b2 is the previous two input coefficients, which represents the weight of the influence of the original signal at the previous two moments on the output.

[0068] The filter coefficient is based on the rotor mechanical angular velocity ω m Dynamic Updates:

[0069]

[0070] Where K is the intermediate variable; tan is the tangent function; ω m is the rotor mechanical angular velocity; T is the sampling period; b0 and b2 are the current input coefficient and the previous two input coefficients, respectively, and the two values ​​are equal; b1 is the previous input coefficient, which is twice the value of b0; a1 and a2 are the filter autoregressive coefficient one and autoregressive coefficient two, respectively, which are used to construct the pole characteristics of the filter.

[0071] Where T is the sampling period, which is set to T=0.0001s.

[0072] Step 2: Process the filtered displacement signal based on a second-order generalized integrator to extract the fundamental frequency harmonic component synchronized with the rotor mechanical angular velocity;

[0073] Second-Order Generalized Integrator (SOGI) is a kind of harmonic extractor based on quadrature signal generation, which can separate the harmonic component of a specific frequency from the noisy displacement signal with high precision. The transfer function model of SOGI is:

[0074]

[0075] where H α (s) is the in-phase component transfer function; H β (s) is the quadrature component transfer function; v α (s) is the in-phase output signal; v β (s) is the quadrature output signal; v f (s) is the input displacement signal; ξ is the damping coefficient; ω m is the rotor mechanical angular velocity; s is the Laplace transform complex variable.

[0076] The preprocessed signal v f (k2) is input into the SOGI harmonic observer to extract the fundamental frequency component synchronized with the mechanical angular velocity ω m . The discrete implementation of SOGI is:

[0077]

[0078] where v α (k2) is the current in-phase output; v α (k2-1) is the previous in-phase output; v β (k2) is the current quadrature output; v β (k2-1) is the previous quadrature output; v f (k2) is the current filtered signal; v f (k2-1) is the previous filtered signal; T is the sampling period; ω m is the rotor mechanical angular velocity; ξ is the damping coefficient.

[0079] where the damping coefficient ξ = 0.8 to ensure the balance between system stability and response speed.

[0080] The amplitude and phase of the fundamental frequency harmonic component are calculated through the quadrature component:

[0081]

[0082] where v h (k2) is the fundamental frequency harmonic component; v α (k2) is the in-phase output; v β (k2) is the quadrature output; is the square root operation; cos is the cosine function; ωm is the rotor mechanical angular velocity; T is the sampling period; φ is the phase angle.

[0083] The phase angle is:

[0084]

[0085] Where φ is the phase angle; arctan is the inverse tangent function; v β (k2) is the orthogonal output; v α (k2) is the non-inverting output.

[0086] The frequency adaptive adjustment mechanism updates the center frequency of SOGI in real time according to the speed change:

[0087]

[0088] where ω m is the rotor mechanical angular velocity; π is pi; n is the rotor speed; f m is the rotor mechanical frequency.

[0089] Displacement error signal generation:

[0090] e(k2)=v ref (k2)-v h (k2);

[0091] Where e(k2) is the displacement error signal; v ref (k2) is the expected displacement value; v h (k2) is the fundamental frequency harmonic component.

[0092] where v ref (k2) is the expected displacement value, which is set to 0.

[0093] like Figure 4 As shown in the figure, the harmonic component extraction process includes four steps: signal preprocessing, SOGI processing, harmonic reconstruction and error generation, which realizes the accurate extraction of fundamental frequency components from the original displacement signal.

[0094] Step 3: Based on the extracted fundamental frequency harmonic components, the simplified LMS algorithm is used to dynamically optimize the amplitude weight and phase weight of the compensation force, and the feedforward compensation force is generated according to the optimized weights;

[0095] The traditional LMS algorithm needs to adjust the multi-dimensional weight vector, which has high computational complexity. The present invention reduces the dimension and only retains the amplitude weight k p and phase weight k i Two parameters reduce the computational burden. The objective function of the simplified LMS algorithm is:

[0096] J(k p , ki ) = e[(k3) - k p v α (k3) - k i v β (k3) 2 ];

[0097] where J(k p , k i ) is the objective function; E[·] is the expectation operation; e(k3) is the displacement error signal; k p is the amplitude weight; k i is the phase weight; v α (k3) is the in-phase output; v β (k3) is the quadrature output.

[0098] The parameter update law is:

[0099]

[0100] where k p (k3+1) is the next-time amplitude weight; k p (k3) is the current-time amplitude weight; k i (k3+1) is the next-time phase weight; k i (k3) is the current-time phase weight; μ p is the amplitude step size factor; μ i is the phase step size factor; e(k3) is the displacement error signal; v h (k3) is the fundamental harmonic component.

[0101] The amplitude weight k p and the phase weight k i of the compensation force are dynamically adjusted based on the simplified LMS algorithm.

[0102] When the simplified LMS algorithm is initialized, k p (0) = 0.5, k i (0) = 0.1, the step size factor μ p = 0.02, and μ i = 0.01.

[0103] The parameter update process is:

[0104] The displacement error e(k3) and the fundamental component v h (k3) at the current time are calculated.

[0105] The amplitude weight k p (k3+1) and the phase weight k i (k3+1) are updated.

[0106] Check the convergence condition to ensure that the step size factor satisfies:

[0107]

[0108] where μ p is the amplitude step size factor; μ i is the phase step size factor; λ max is the maximum eigenvalue operation; R p is the amplitude input signal autocorrelation matrix; R i is the phase input signal autocorrelation matrix.

[0109] Generation of feedforward compensation force:

[0110]

[0111] where F comp (k3) is the feedforward compensation force; k p (k3) is the amplitude weight; k i (k3) is the phase weight; v h (k3) is the fundamental harmonic component.

[0112] The feedforward compensation force directly cancels the fundamental vibration component synchronized with ω m , whose amplitude and phase are dynamically adjusted through k p and k i .

[0113] The feedforward compensation force directly cancels the fundamental vibration component synchronized with ω m , whose amplitude and phase are dynamically adjusted through k p and k i .

[0114] As shown in Figure 5 , the adaptive compensation process is realized through three stages:

[0115] In the initialization stage, the initial values of the compensation force parameters k p and k i are set, and the pre-trained step size factors μ p and μ i are loaded; in the real-time update stage, k p and k i are optimized online through a simplified LMS algorithm to dynamically adjust the amplitude and phase of the compensation force F comp ;

[0116] In the closed-loop feedback stage, the residual error after feedforward compensation is input into an adaptive PID controller to further correct the control force F PID , and finally obtain the total control force.

[0117] Step 4, build the dual-channel control architecture, superimpose the feedforward compensation force generated based on the fundamental harmonic component with the PID correction force of the feedback correction channel output to synthesize the total control force;

[0118] The dual-channel control architecture includes a feedforward compensation channel and a feedback correction channel. The feedforward channel has generated the compensation force A in step 3 by simplifying the LMS algorithm. The feedback channel uses an adaptive PID controller to process the residual error e PID (k4) to generate the correction force F PID (k4):

[0119]

[0120] Where F PID (k4) is the PID correction force; K p is the proportional coefficient; K i is the integral coefficient; K d is the differential coefficient; e PID (k4) is the residual error; and ∫ is the integral operation.

[0121] The parameter update law of the adaptive PID controller is:

[0122]

[0123] Where K p (k4+1) is the proportional coefficient at the next time; K p (k4) is the proportional coefficient at the current time; K i (k4+1) is the integral coefficient at the next time; K i (k4) is the integral coefficient at the current time; K d (k4+1) is the differential coefficient at the next time; K d (k4) is the differential coefficient at the current time; α p is the proportional learning rate; α i is the integral learning rate; α d is the differential learning rate; e PID (k4) is the residual error; v h (k4) is the fundamental harmonic component; and ∑ is the summation operation.

[0124] The total control force is the superposition of the outputs of the two channels:

[0125] F total (k4) = F comp (k4) + F PID (k4);

[0126] Where F total (k4) is the total control force; F comp (k4) is the feedforward compensation force; and FPID (k4) is the PID correction force.

[0127] The advantage of this dual-channel architecture design is that the feedforward channel precisely cancels the fundamental frequency vibration, and the feedback channel suppresses the residual error through adaptive PID, achieving the collaborative optimization of dynamic response speed and steady-state accuracy.

[0128] As shown in Figure 6 The entire CCR-BIM system control block diagram includes suspension force vibration compensation control and speed control two subsystems, which realize stable suspension and precise control of the rotor through collaborative work.

[0129] Step 5, based on the synthesized total control force, the corresponding radial suspension force is generated by driving the suspension force winding through the inverter, realizing the cancellation of the rotor unbalanced vibration;

[0130] Total control force F total (k5) The suspension force winding is driven by a three-phase full-bridge inverter to generate the corresponding radial suspension force to cancel the rotor unbalanced vibration.

[0131] System stability is verified by Lyapunov energy function:

[0132]

[0133] Where V(k5) is the Lyapunov energy function; e(k5) is the displacement error signal; is the proportional parameter error; is the integral parameter error; is the differential parameter error.

[0134]

[0135] Where is the proportional parameter error; K p is the current proportional coefficient; is the ideal value of the proportional coefficient.

[0136]

[0137] Where is the integral parameter error; K i is the current integral coefficient; is the ideal value of the integral coefficient.

[0138]

[0139] Where is the differential parameter error; K d is the current differential coefficient; is the ideal value of the differential coefficient.

[0140] When the step size factor meets the convergence condition, the energy difference ΔV(k5)=V(k5+1)-V(k5)<0, and the system approaches global stability.

[0141] A computer storage medium includes a memory and one or more processors, wherein the memory stores executable code, and when the one or more processors execute the executable code, they are used to perform the above-mentioned real-time vibration compensation control method for a bearingless composite rotor cage asynchronous motor.

[0142] Here, the present invention provides an implementation example:

[0143] Aiming at the unbalanced vibration problem of bearingless composite rotor cage asynchronous motor during high-speed rotation, a real-time compensation control method based on SOGI harmonic observation and dual-channel adaptive PID is applied. Figure 7 The key parameter settings shown in the figure, the detailed parameters of CCR-BIM are as follows Figure 8 shown.

[0144] This application scenario applies vibration suppression control to a CCR-BIM system operating in the 3000-6000 rpm range. The rotor has a mass of 2.97 kg and an eccentricity of 0.05 mm, producing significant unbalanced vibration during high-speed rotation. Four control methods were compared: uncompensated control, traditional PID compensation control, feedforward compensation control, and the proposed real-time vibration compensation control method based on SOGI harmonic observation and dual-channel adaptive PID.

[0145] The experimental conditions included both steady-state and dynamic operating conditions. The steady-state condition tested the vibration suppression effectiveness at different speeds, while the dynamic condition tested the system's dynamic response under sudden load. The experiment collected rotor displacement signals along the x- and y-axes, and used an oscilloscope to record the vibration waveforms before and after control, the rotor axis trajectory, and speed fluctuations under load disturbances.

[0146] First, the mechanism of unbalanced vibration in the CCR-BIM rotor was analyzed. The CCR-BIM utilizes the synergistic effects of the torque winding and the levitation winding to achieve rotor rotation and levitation. On the experimental platform, it was observed that manufacturing tolerances and assembly errors in the rotor caused a deviation of approximately 0.05 mm between the rotor's center of mass and its geometric center, triggering unbalanced vibration.

[0147] Eddy current sensors were used to collect rotor displacement signals along the x- and y-axes, with a sampling period of 0.0001s. Because the raw signals contain high-frequency noise and sensor interference, a second-order Butterworth low-pass filter was used for preprocessing. The cutoff frequency was set to 1.2 times the rotor's mechanical angular velocity to suppress high-frequency noise while preserving the integrity of the fundamental frequency component.

[0148] Under the condition of 3000 r / min, the peak value of the original vibration signal without compensation reaches 5.4 pm in the x-axis direction and 5.9 pm in the y-axis direction, which will affect the operation accuracy and service life of the CCR-BIM. The filtered signal retains the fundamental frequency vibration characteristics, providing a basis for subsequent harmonic component extraction.

[0149] The preprocessed signal is input into the SOGI harmonic observer to extract the fundamental frequency component synchronized with the mechanical angular velocity. The damping coefficient of SOGI is set to 0.8 to ensure the balance between system stability and response speed.

[0150] Under the condition of 3000 r / min, the SOGI harmonic observer successfully separates the fundamental frequency vibration component synchronized with the mechanical angular velocity of 50 Hz from the noisy displacement signal. Compared with the traditional band-pass filter, the SOGI observer exhibits higher extraction accuracy and better phase tracking ability, with an extraction error controlled within 3%.

[0151] When the speed increases to 6000 r / min, the frequency adaptive adjustment mechanism automatically updates the center frequency of SOGI to 100 Hz, continuing to maintain high harmonic extraction accuracy. Experiments show that within the speed range of 3000-6000 r / min, the SOGI observer can accurately track the amplitude and phase changes of the fundamental frequency component, providing accurate vibration characteristic information for subsequent compensation force generation.

[0152] The simplified LMS algorithm is used to dynamically adjust the amplitude weight and phase weight of the compensation force. When the algorithm is initialized, the initial value of the amplitude weight is set to 0.5, the initial value of the phase weight is set to 0.1, and the step factors are set to 0.02 and 0.01 respectively.

[0153] Under the condition of 3000 r / min, the simplified LMS algorithm undergoes an adaptive adjustment process of about 0.2 seconds, and the amplitude weight converges to 0.83 and the phase weight converges to 0.27, generating a compensation force matched with the vibration characteristics. Experiments show that when only the feedforward compensation channel is used, the x-axis vibration amplitude is reduced from 5.4 pm to 2.9 pm, and the y-axis vibration amplitude is reduced from 5.9 pm to 3.2 pm, with vibration suppression effects of 46.3% and 45.8% respectively.

[0154] When the speed increases to 6000 r / min, the algorithm automatically adjusts the parameters to adapt to the new vibration characteristics, converging to the new optimal value within about 0.3 seconds and maintaining good feedforward compensation effect. This shows that the simplified LMS algorithm has good adaptive ability and can effectively cope with the changes in vibration characteristics caused by changes in speed.

[0155] The residual error after feedforward compensation is input into the adaptive PID controller for further processing. The initial PID parameters are set as follows: proportional coefficient 0.5, integral coefficient 0.1, differential coefficient 0.01, and learning rates 0.01, 0.005, and 0.001 respectively.

[0156] At 3000 rpm, the dual-channel control architecture works in tandem, with the feedforward channel precisely offsetting fundamental frequency vibrations and the feedback channel suppressing residual errors through adaptive PID control. Experimental results show that dual-channel control further reduces the x-axis vibration amplitude to 1.6μm and the y-axis vibration amplitude to 1.7μm. Compared to using feedforward compensation alone, these improvements represent 44.8% and 46.9% improvements in vibration suppression, respectively. Compared to traditional PID compensation control, these improvements are 66.0% and 66.7%, respectively.

[0157] When the speed is increased to 6000r / min, the dual-channel control architecture still maintains good synergy, and the rotor vibration amplitude is controlled within an acceptable range, indicating that the architecture design has high control accuracy and stability under different operating conditions.

[0158] In practical applications, the total control force drives the suspension winding through a three-phase full-bridge inverter, generating a corresponding radial suspension force to offset the rotor's unbalanced vibration. Multiple sets of experiments verified the system's stability and robustness.

[0159] like Figure 9 As shown in the figure, under sudden load conditions, the speed fluctuation of the proposed method is significantly smaller than that of the other three methods, demonstrating its superior disturbance tolerance. Under a sudden 10N load disturbance, the speed recovery time at 3000 rpm is shortened from 27.1ms without compensation to 9.9ms, a 63.5% reduction; at 6000 rpm, the speed recovery time is shortened from 60.2ms without compensation to 29.3ms, a 51.3% reduction. Compared to feedforward compensation control, these recovery times are reduced by 27.7% and 17.9%, respectively.

[0160] like Figure 10 and Figure 11 Comparative experimental results of x- and y-axis vibration displacements demonstrate that the proposed method exhibits excellent vibration suppression under various speed and load conditions. The maximum rotor axis trajectory is also reduced: at 3000 rpm, it decreases from 81.6 μm without compensation to 19.2 μm, a 76.5% reduction; at 6000 rpm, it decreases from 251.1 μm without compensation to 84.1 μm, a 66.5% reduction. Compared to feedforward compensation control, trajectory stability is improved by 54.7% and 29.7%, respectively.

[0161] In practical applications, in addition to the stability verification mentioned earlier, the vibration suppression effect of the system at different speeds is also comprehensively evaluated. As shown in FIG. 7, the vibration suppression effects of the non-compensation control, the traditional PID compensation control, the feedforward compensation control, and the vibration real-time compensation control method based on SOGI harmonic observation and double-channel adaptive PID proposed in the present application in the speed range of 3000-6000 r / min are compared. Figure 12

[0162] The experimental results show that, as the speed increases, the vibration suppression effect of all methods decreases, but the method proposed in the present application always maintains the best vibration suppression effect in the entire speed range. In particular, in the high-speed region (5000-6000 r / min), the advantages of the present method over other control methods are more obvious, and the vibration suppression rate is maintained at more than 65%, while the vibration suppression rates of the traditional PID control and the feedforward compensation control are reduced to about 30% and 45%, respectively.

[0163] This result further verifies the adaptability and robustness of the present method under high-speed rotating conditions, and proves that the vibration real-time compensation control method based on SOGI harmonic observation and double-channel adaptive PID can effectively suppress the unbalanced vibration of CCR-BIM in a wide speed range, providing a reliable vibration control solution for high-speed precision rotating machinery.

[0164] In the long-time running test, the system always remains stable without divergence or instability, verifying the correctness of the Lyapunov stability analysis. In addition, when the speed changes in the range of 3000-6000 r / min, the algorithm can quickly adapt to the speed change, maintaining stable vibration suppression effect, the adaptive parameter adjustment mechanism effectively responds to system parameter changes, and the double-channel collaborative control ensures the balance between dynamic response speed and steady-state accuracy.

[0165] In summary, the application example shows that the vibration real-time compensation control method for bearingless compound rotor cage asynchronous motor based on SOGI harmonic observation and double-channel adaptive PID is superior to the traditional control method in terms of vibration suppression effect, dynamic response performance, and trajectory stability, providing an efficient and reliable solution for vibration control of high-speed rotating machinery. The method has the advantages of low algorithm complexity, good real-time performance, and strong robustness, and is suitable for vibration control applications in various high-speed rotating machinery, having important engineering practical value and wide application prospects.

[0166] The above describes embodiments of the present application, but the embodiments are not limited to the specific implementation described above, which is only illustrative and not limiting. Those skilled in the art can make more forms of equivalent embodiments under the inspiration of the embodiments, which are all within the protection scope of the embodiments.​

Claims

1. A real-time vibration compensation control method for a bearingless composite rotor cage asynchronous motor, characterized in that: include: Collect the rotor displacement signal and pre-process it to obtain a filtered displacement signal; The filtered displacement signal is processed based on a second-order generalized integrator to extract the fundamental frequency harmonic component synchronized with the rotor's mechanical angular velocity. Based on the extracted fundamental frequency harmonic components, a simplified LMS algorithm is used to dynamically optimize the amplitude weight and phase weight of the compensation force, and a feedforward compensation force is generated according to the optimized weights. A dual-channel control architecture is constructed to superimpose the feedforward compensation force generated based on the fundamental frequency harmonic component and the PID correction force output by the feedback correction channel to synthesize the total control force; Based on the synthesized total control force, the inverter drives the suspension winding to generate corresponding radial suspension force, thereby offsetting the unbalanced vibration of the rotor.

2. The method for real-time vibration compensation control of a bearingless composite rotor cage asynchronous motor according to claim 1, characterized in that: The preprocessing step includes: collecting the displacement signals of the rotor in the x-axis and y-axis directions through an eddy current sensor, and filtering high-frequency noise and sensor interference with a second-order Butterworth low-pass filter, with the cutoff frequency of the filter set to 1.2 times the mechanical angular velocity of the rotor.

3. The method for real-time vibration compensation control of a bearingless composite rotor cage asynchronous motor according to claim 1, characterized in that: The difference equation of the second-order Butterworth low-pass filter is: in f (k1)=a1v f (k1-1)+a2v f (k1-2)+b0v(k1)+b1v(k1-1)+b2v(k1-2); where v f (k1) is the filtered signal at the current moment; v f (k1-1) is the filtered signal at the previous moment; v f (k1-2) is the filtered signal at the previous two moments; v(k1) is the original signal at the current moment; v(k1-1) is the original signal at the previous moment; v(k1-2) is the original signal at the previous two moments; a1 is the autoregressive coefficient of the filter, which represents the weight of the influence of the filtered signal at the previous moment on the current output; a2 is the autoregressive coefficient of the filter, which represents the weight of the influence of the filtered signal at the previous two moments on the current output; b0 is the current input coefficient, which represents the weight of the influence of the original signal at the current moment on the output; b1 is the previous input coefficient, which represents the weight of the influence of the original signal at the previous moment on the output; b2 is the previous two input coefficients, which represents the weight of the influence of the original signal at the previous two moments on the output.

4. The method for real-time vibration compensation control of a bearingless composite rotor cage asynchronous motor according to claim 1, characterized in that: The transfer function model of the second-order generalized integrator includes an in-phase component transfer function and an orthogonal component transfer function, and the damping coefficient is set to 0.8 to ensure a balance between system stability and response speed.

5. The method for real-time vibration compensation control of a bearingless composite rotor cage asynchronous motor according to claim 4, characterized in that: The second-order generalized integrator has a frequency adaptive adjustment mechanism, and the center frequency of SOGI is updated in real time according to the speed change, which is expressed as: Center Frequency: where ω m is the rotor mechanical angular velocity; π is pi; n is the rotor speed; f m is the rotor mechanical frequency.

6. The method for real-time vibration compensation control of a bearingless composite rotor cage asynchronous motor according to claim 1, characterized in that: The simplified LMS algorithm only retains the amplitude weight k p and phase weight k i Two parameters, k is set when the algorithm is initialized p (0) = 0.5, k i (0)=0.

1.

7. The method for real-time vibration compensation control of a bearingless composite rotor cage asynchronous motor according to claim 6, characterized in that: The parameter update law of the simplified LMS algorithm is: where k p (k3+1) is the amplitude weight at the next moment; k p (k3) is the amplitude weight at the current moment; k i (k3+1) is the phase weight at the next moment; k i (k3) is the phase weight at the current moment; μ p is the amplitude step factor; μ i is the phase step factor; e(k3) is the displacement error signal; v h (k3) is the fundamental frequency harmonic component.

8. The method for real-time vibration compensation control of a bearingless composite rotor cage asynchronous motor according to claim 1, characterized in that: The feedforward compensation force is generated as follows: Among them F comp (k3) is the feedforward compensation force; k p (k3) is the amplitude weight; k i (k3) is the phase weight; v h (k3) is the fundamental frequency harmonic component.

9. The method for real-time vibration compensation control of a bearingless composite rotor cage asynchronous motor according to claim 1, characterized in that: The feedback correction channel in the dual-channel control architecture adopts an adaptive PID controller to process the residual error after feedforward compensation.

10. A computer storage medium, characterized in that It includes a memory and one or more processors, wherein the memory stores executable code, and when the one or more processors execute the executable code, they are used to execute the real-time vibration compensation control method for a bearingless composite rotor cage-type asynchronous motor according to any one of claims 1 to 9.

Images