Method and device for observing rotating speed of magnetic bearing

Abstract

The invention discloses a magnetic bearing rotating speed observation method and device, and belongs to the field of magnetic bearing rotating speed measurement. The method comprises the following steps: generating an error signal according to an input signal and an internal reconstruction signal; in the amplitude estimation loop, performing synchronous demodulation and integration on the error signal to generate an estimated amplitude; and in a frequency / phase estimation loop, synchronously demodulating the error signal, normalizing the demodulation result by using the estimated amplitude, and processing by a loop filter and an integrator serving as a voltage-controlled oscillator to generate estimated frequency and phase. According to the method, decoupling of amplitude and frequency estimation is achieved through normalization processing, processing of the alternating current signals is converted into integration of the direct current signals, the system reliability is improved, and the method is particularly suitable for serving as a redundancy observation means when the rotating speed sensor breaks down.

Description

Technical Field

[0001] This invention belongs to the field of magnetic bearing speed measurement technology, specifically relating to a method and device for observing the speed of a magnetic bearing. Background Technology

[0002] Magnetic levitation bearings are high-performance bearings that use electromagnetic force to levitate a rotating shaft in space. Compared with traditional mechanical bearings, they have advantages such as no friction, no need for lubrication, low wear, high speed, and active controllability. They are widely used in precision rotating machinery fields such as flywheel energy storage, aerospace, and high-speed motors.

[0003] In the control system of magnetic levitation bearings, accurate real-time rotational speed information is crucial. For example, suppressing vibrations that are in sync with the rotational speed caused by rotor mass imbalance (affected by factors such as machining accuracy, material uniformity, installation errors, and thermal deformation), and decoupling control of the gyroscopic effect generated during high-speed rotation, all require accurate rotational speed as a prerequisite.

[0004] Currently, there are two main methods for obtaining speed information. The first is to install a dedicated speed sensor, such as a Hall sensor or photoelectric encoder. While this method is direct, it has several drawbacks: First, in some compact or harsh application environments, there may not be enough space to install the sensor; second, the sensor itself and its wiring are potential points of failure, and damage or poor contact will lead to a decrease in the performance of the entire system or even failure. The second method is to obtain speed information by communicating with the frequency converter of the drive system, but this also carries the risk of communication interruption or data delay, making reliability difficult to guarantee.

[0005] To address the reliance on physical sensors, sensorless observation techniques have emerged. One common algorithm is the adaptive observer based on the generalized integrator (GI). This method aims to extract frequencies from vibration signals containing rotational speed information. However, the GI-based approach has inherent limitations in digital implementation: as rotational speed increases (i.e., signal frequency increases), the GI struggles to generate two strictly orthogonal (90° out of phase) signals relative to a fixed controller sampling rate. This non-orthogonality leads to steady-state errors in rotational speed estimation, with the error increasing at higher speeds. Furthermore, its convergence speed is slow when rotational speed changes abruptly.

[0006] Therefore, how to provide a speed observation method that does not require physical sensors, maintains high accuracy at high speeds, has fast dynamic response, and high reliability is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0007] Purpose of the invention: The present invention aims to solve the problems of existing technologies that rely on physical sensors for speed observation, have insufficient reliability, and have low accuracy and slow convergence at high frequencies, as well as existing sensorless algorithms. Therefore, it provides a high-performance magnetic bearing speed observation method and device.

[0008] Technical Solution: The magnetic bearing speed observation method of the present invention is applied to a magnetic levitation bearing system. The method uses a rotor displacement signal (u) with the same frequency as the rotational speed to observe the rotational speed, and includes the following steps: S1: Generate an error signal (e) based on the rotor displacement signal (u) and the internally reconstructed displacement signal (VO); S2: Apply an amplitude estimation loop to compare the error signal (e) with the estimated phase (e) The generated in-phase reference signal is multiplied, and the result is integrated to generate an estimated amplitude (V1). S3: Apply a frequency / phase estimation loop to compare the error signal (e) with the estimated phase (e) based on the frequency / phase estimation loop. The orthogonal reference signal generated is multiplied, and the result is normalized using the estimated amplitude (V1); the normalized signal is filtered to generate an angular frequency adjustment, and an estimated frequency is generated based on the angular frequency adjustment. ), and then the estimated frequency ( Integrate to generate the estimated phase ( ); S4: The estimated amplitude (V1) is multiplied by the in-phase reference signal to generate the internally reconstructed displacement signal (VO), which is then provided to step S1 for closed-loop calculation.

[0009] To further improve the above technical solution, the in-phase reference signal and the quadrature reference signal are respectively based on the estimated phase ( The generated sine and cosine signals.

[0010] Furthermore, the amplitude estimation loop includes an integrator for integration operations and a configurable amplitude estimation gain (K1).

[0011] Furthermore, the frequency / phase estimation loop includes a PI controller as a loop filter and an integrator as a voltage-controlled oscillator; the normalized signal is processed by the loop filter to generate the estimated frequency. ), and the estimated frequency ( The input is given to the voltage-controlled oscillator to generate the estimated phase. ).

[0012] Furthermore, the frequency / phase estimation loop also includes a preset initial frequency ( The initial frequency ( The estimated frequency is generated by adding the signal processed by the loop filter to the signal processed by the loop filter. ).

[0013] Furthermore, the rotor displacement signal (u) is obtained by high-pass filtering the original displacement signal to extract the vibration component with the same frequency as the rotational speed.

[0014] Furthermore, the method is applied to provide redundant speed information when the speed sensor of the magnetic levitation bearing system malfunctions or communication with the frequency converter is interrupted.

[0015] The present invention also provides a magnetic bearing speed observation device, comprising: The error generation unit is configured to generate an error signal (e) based on the input rotor displacement signal (u) and the internal reconstruction signal. The amplitude estimation unit is configured to estimate the amplitude by comparing the error signal (e) with the estimated phase (e) The generated in-phase reference signal is multiplied, and the result is integrated to generate an estimated amplitude (V1). The frequency / phase estimation unit is configured to estimate the frequency / phase by comparing the error signal (e) with the estimated phase (e) according to the frequency / phase estimation signal (e). The generated orthogonal reference signal is multiplied, and the result is normalized using the estimated amplitude (V1) to obtain a normalized signal; the normalized signal is then filtered to generate an angular frequency adjustment, and an estimated frequency (V1) is generated based on the angular frequency adjustment. ), and then the estimated frequency ( Integrate to generate the estimated phase ( ); The signal reconstruction unit is configured to generate the internally reconstructed displacement signal (VO) by multiplying the estimated amplitude (V1) with the in-phase reference signal and provide it to the error generation unit.

[0016] Beneficial effects: Compared with the prior art, the advantages of the present invention are as follows: The dual-ring decoupled digital phase-locked loop (PLL) structure employed in this invention converts the AC error signal into a DC component for processing through synchronous demodulation, fundamentally avoiding the accuracy loss problem of digital integration of high-frequency AC signals. When a step change occurs in frequency, the observer of this invention can instantly and accurately lock onto the new frequency without steady-state error, and its convergence speed is much faster than that of the generalized integration method.

[0017] The response curve of this invention is fast and smooth, without any overshoot or oscillation, demonstrating excellent dynamic performance and system stability. This is due to the successful decoupling of the amplitude and frequency loops, which avoids mutual interference between the two during the dynamic process.

[0018] During operation, especially when passing through the resonance region such as the critical speed, the amplitude of the rotor displacement signal in a magnetic bearing system undergoes drastic changes. For traditional coupled observers, such amplitude fluctuations severely interfere with frequency estimation. In this invention, the frequency / phase estimation loop uses the real-time amplitude (V1) estimated by the amplitude loop to normalize the error signal (e / V1 ≈ e / A), thereby completely eliminating the influence of input signal amplitude changes on frequency estimation.

[0019] This invention, through an innovative dual-loop decoupling design, not only overcomes the accuracy and speed bottlenecks of existing technologies at high frequencies, but also endows it with strong robustness, completely immune to changes in the amplitude of the input signal. The method is simple to calculate and easy to implement on a digital controller, providing fast, accurate, and highly reliable speed information over a wide speed range, offering strong redundancy protection for magnetic levitation bearing systems under abnormal conditions such as sensor failure.

[0020] This invention achieves independent operation without relying on external physical sensors or communication, serving as a reliable redundant observation method. When the main sensor or communication fails, this method can seamlessly take over, ensuring uninterrupted and safe operation of the entire magnetic levitation system, greatly improving the system's reliability and fault tolerance. Attached Figure Description

[0021] Figure 1 This is a block diagram of an adaptive rotational speed observer based on generalized integrals in the prior art.

[0022] Figure 2 This is a block diagram of the rotational speed observer according to an embodiment of the present invention.

[0023] Figure 3 This is a comparison chart of the frequency locking performance of the observer of this invention and the observer based on generalized integral.

[0024] Figure 4 This is a graph showing the frequency sweep test results of the observer of this invention.

[0025] Figure 5 This is a step response diagram of the observer of the present invention. Detailed Implementation

[0026] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings, but the scope of protection of the present invention is not limited to the embodiments described.

[0027] Example 1: Figure 1The diagram shows a block diagram of an adaptive speed observer based on generalized integrals, where K1 and K2 represent the gains of the generalized integrator (GI) and the frequency estimator, respectively. Signal Va has the same phase as the fundamental component of the input signal, while signal Vb has a 90° phase delay. The transfer function between the error signal and the input signal can be expressed as: .

[0028] The input signal U is obtained by extracting the vibration displacement at the same frequency as the rotational speed through high-pass filtering. Then, the error signal e and the two-phase quadrature output signals Va and Vb are calculated, and finally, the estimated rotational speed is obtained. However, when the fundamental frequency increases while the sampling frequency remains unchanged, discretization cannot provide an accurate 90° phase at the fundamental frequency, resulting in non-orthogonality between the Va and Vb outputs, thus causing errors in frequency estimation.

[0029] This invention proposes a method for observing the rotational speed of a magnetic bearing, which constructs as follows: Figure 2 The illustrated dual-loop decoupled digital phase-locked loop (PLL) system shows u as the input rotor displacement signal, VO as the reconstructed signal, and Vout as the output. When phase locking is successful, the output reconstructed signal VO ≈ the input u. This indicates the estimated frequency of the signal, corresponding to the rotational speed of the magnetic bearing.

[0030] Assuming the input signal Under steady-state phase-locked loop (PLL) conditions, the output signal V1 = A (input signal amplitude), and e represents the error signal between the input signal and the estimated signal. Since the input and output signals are not amplitude normalized, the expression for the error signal e includes the voltage amplitude coefficient A, leading to frequency... and phase It is easily affected by the amplitude of the input signal. To decouple this relationship, the characteristic of the amplitude estimation loop to accurately track the amplitude of the input signal is utilized. Before the error signal e is input into the phase-locked loop, the error signal e is divided by V1 to make it e / V1 ( ), which normalizes the amplitude of the error signal e.

[0031] The normalized error signal is processed by a loop filter (i.e., a PI controller) and a voltage-controlled oscillator (VCO, i.e., an integrator) to generate an estimated frequency. and estimated phase ;in, This indicates initial frequency-assisted locking. In the amplitude estimation algorithm, the error signal e and the estimated phase... sine function After performing multiplication and adjusting the gain K1, the intermediate amplitude estimate V1 is generated by the voltage-controlled oscillator (VCO, i.e., the integrator). V1 is further processed to obtain the final estimated amplitude VO.

[0032] The system dynamically adjusts the proportional coefficient KP and integral coefficient KI of the PI controller through a preset lookup table. K1 represents the gain of the amplitude estimation loop and is numerically the same as the KP coefficient in the PI controller. ,in, Equal to the angular frequency of the target signal, the coefficient K is adjustable. A smaller K value corresponds to a slower / smoother response, while a larger K value results in a faster response but may be accompanied by a larger oscillation amplitude. The value range is 0.5 ≤ K ≤ 1.5. (Integral coefficient in a PI controller) Therefore, once K is determined, other parameters are also determined accordingly.

[0033] Example 2: In this example, the method uses an amplitude estimation loop and a frequency / phase estimation loop working together. The specific steps are as follows: S1: Error generation.

[0034] like Figure 2 As shown, the externally input rotor displacement signal u and the reconstructed displacement signal VO fed back from the system are subtracted in the subtractor to generate an error signal e. The magnitude of this error signal e reflects the deviation between the current estimated value and the true value, and serves as the common drive signal for the two parallel loops.

[0035] S2: Amplitude estimation.

[0036] The error signal e is fed into the amplitude estimation loop ( Figure 2 (Upper half). In this loop, e is related to the estimated phase. The generated in-phase reference signal ( The signal is then multiplied (synchronously demodulated). The demodulated signal is then amplified by a configurable gain module K1 and then input to an integrator, which acts as a voltage-controlled oscillator (VCO), for integration. The output of the integrator is a real-time estimate of the amplitude of the input signal u, V1. V1 is further processed to generate the reconstructed displacement signal VO.

[0037] S3: Frequency / phase estimation.

[0038] In parallel with the amplitude loop, the error signal e is also fed into the frequency / phase estimation loop ( Figure 2 (Lower half)

[0039] Error signal e and the estimated phase The generated orthogonal reference signal ( Multiplication is performed to extract phase error information. The result of the multiplication is divided by the estimated amplitude V1 obtained in step S2, so that the frequency / phase estimation process is no longer affected by the magnitude of the input signal u, thus achieving decoupling of the two loops.

[0040] The normalized signal is then fed into a PI controller PI(z) acting as a loop filter for filtering. The filtered signal is then compared with a preset initial frequency. Add them together to get the final estimated angular frequency. , This setting helps the system lock onto the target frequency faster during startup. It will estimate the angular frequency. The sample is fed into another integrator, which acts as a voltage-controlled oscillator (VCO), for integration to generate an estimated phase. .

[0041] S4: Signal reconstruction.

[0042] The estimated amplitude V1 obtained in step S2 is compared with the in-phase reference signal. Multiplying these signals generates an internally reconstructed displacement signal VO. This signal VO is fed back to the subtractor in step S1, forming a complete negative feedback closed loop. This drives the error signal e to approach zero, thereby causing the estimated value V1 to... , It converges to the true value.

[0043] contrast Figure 1 The observer shown in this invention uses an integrator that operates on DC signals, while the two integrators in the generalized integral require a higher sampling rate when processing AC signals. Figure 3 To compare the performance of the two velocity observers, the observer based on generalized integrals exhibits frequency errors when processing high-frequency signals and has a slow convergence speed during frequency jumps. The observer of this invention converges quickly during frequency jumps, locks onto the new target frequency almost instantaneously, and has no errors after stabilization.

[0044] Figure 4 The results of the frequency sweep test by the speed observer show that, with the bearing in a static suspension state, the controller provides a frequency sweep signal from 1 to 700 Hz. The upper part shows the rotor displacement response, displaying the actual vibration displacement of the rotor during the frequency sweep process. It can be seen that due to system resonance and other reasons, the amplitude of the displacement signal fluctuates drastically at different frequencies. The lower part shows the observation effect of the observer, which is a very smooth and linear sloping line that accurately tracks from about 48 Hz to about 689 Hz, completely reproducing the input frequency sweep signal.

[0045] Figure 5The observed rotational speed step response is such that when the target rotational speed undergoes a step change, the observed rotational speed value rises at an extremely rapid rate and stabilizes immediately after reaching the target value without any overshoot or oscillation. This demonstrates that the method provided by this invention has significant accuracy advantages, a simple calculation process, and is easy to implement with a digital controller.

[0046] As described above, although the invention has been shown and described with reference to specific preferred embodiments, it should not be construed as limiting the invention itself. Various changes in form and detail may be made without departing from the spirit and scope of the invention as defined in the appended claims.

Claims

1. A method for observing the rotational speed of a magnetic bearing, applied to a magnetic levitation bearing system, wherein the method utilizes a rotor displacement signal (u) with the same frequency as the rotational speed for rotational speed observation, characterized in that... Includes the following steps: S1: Generate an error signal (e) based on the rotor displacement signal (u) and the internally reconstructed displacement signal (VO); S2: Apply an amplitude estimation loop to compare the error signal (e) with the estimated phase (e) The generated in-phase reference signal is multiplied, and the result is integrated to generate an estimated amplitude (V1). S3: Apply a frequency / phase estimation loop to compare the error signal (e) with the estimated phase (e) based on the frequency / phase estimation loop. The orthogonal reference signal generated is multiplied, and the result is normalized using the estimated amplitude (V1); the normalized signal is filtered to generate an angular frequency adjustment, and an estimated frequency is generated based on the angular frequency adjustment. ), and then the estimated frequency ( Integrate to generate the estimated phase ( ); S4: The estimated amplitude (V1) is multiplied by the in-phase reference signal to generate the internally reconstructed displacement signal (VO), which is then provided to step S1 for closed-loop calculation.

2. The method for observing the rotational speed of a magnetic bearing according to claim 1, characterized in that, The in-phase reference signal and the quadrature reference signal are respectively based on the estimated phase ( The generated sine and cosine signals.

3. The method for observing the rotational speed of a magnetic bearing according to claim 1, characterized in that, The amplitude estimation loop includes an integrator for integration operations and a configurable amplitude estimation gain (K1).

4. The method for observing the rotational speed of a magnetic bearing according to claim 1, characterized in that, The frequency / phase estimation loop includes a PI controller as a loop filter and an integrator as a voltage-controlled oscillator; the normalized signal is processed by the loop filter to generate the estimated frequency. ), and the estimated frequency ( The input is given to the voltage-controlled oscillator to generate the estimated phase. ).

5. The method for observing the rotational speed of a magnetic bearing according to claim 4, characterized in that, The frequency / phase estimation loop also includes a preset initial frequency ( The initial frequency ( The estimated frequency is generated by adding the signal processed by the loop filter to the signal processed by the loop filter. ).

6. The method for observing the rotational speed of a magnetic bearing according to claim 1, characterized in that, The rotor displacement signal (u) is obtained by high-pass filtering the original displacement signal to extract the vibration component with the same frequency as the rotational speed.

7. The method for observing the rotational speed of a magnetic bearing according to claim 1, characterized in that, The method is used to provide redundant speed information when the speed sensor in a magnetic levitation bearing system malfunctions or communication with the frequency converter is interrupted.

8. A magnetic bearing speed observation device, characterized in that, include: The error generation unit is configured to generate an error signal (e) based on the input rotor displacement signal (u) and the internal reconstruction signal. The amplitude estimation unit is configured to estimate the amplitude by comparing the error signal (e) with the estimated phase (e) The generated in-phase reference signal is multiplied, and the result is integrated to generate an estimated amplitude (V1). The frequency / phase estimation unit is configured to estimate the frequency / phase by comparing the error signal (e) with the estimated phase (e) according to the frequency / phase estimation signal (e). The generated orthogonal reference signal is multiplied, and the result is normalized using the estimated amplitude (V1) to obtain a normalized signal; the normalized signal is then filtered to generate an angular frequency adjustment, and an estimated frequency (V1) is generated based on the angular frequency adjustment. ), and then the estimated frequency ( Integrate to generate the estimated phase ( ); The signal reconstruction unit is configured to generate the internally reconstructed displacement signal (VO) by multiplying the estimated amplitude (V1) with the in-phase reference signal and provide it to the error generation unit.

Images