Magnetic bearing vibration suppression and test method

Abstract

According to the vibration suppression and test method for the magnetic bearing, vibration suppression of the magnetic bearing is achieved through the steps of off-line frequency sweeping identification, on-line compensation signal generation, off-line calibration of compensation current gain Ka and the like. According to the method, while the suppression precision of the same-frequency vibration force of the magnetic suspension rotor is improved, the control precision is improved, the vibration force is reduced, and the system reliability is improved; moreover, the method is small in calculation amount, effectively improves the real-time performance of system control, and is very high in practicability and wide in applicability.

Description

Technical Field

[0001] This invention specifically relates to a method for suppressing and testing vibrations in magnetic bearings, belonging to the field of magnetic bearing technology. Background Technology

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

[0003] Due to factors such as machining accuracy, material uniformity, installation errors, and thermal deformation, an eccentric force with the same frequency as the rotational speed is generated, causing the rotor to vibrate with the same frequency as the rotational speed. This is generally optimized using algorithms such as repetitive control, notch filters, and resonant controllers. However, this only eliminates the current stiffness force caused by the same-frequency current; the same-frequency displacement stiffness force generated by the negative displacement stiffness still exists. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to provide a method for suppressing and testing magnetic bearing vibration.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A method for suppressing and testing magnetic bearing vibration includes offline frequency sweep identification, online compensation signal generation, and offline calibration of the compensation current gain Ka. The steps are as follows:

[0007] Step 1: Offline frequency scanning and identification

[0008] The controller outputs a sinusoidal voltage signal ASin(ωt), which is amplified by a power amplifier and applied to the magnetic levitation bearing to generate an electromagnetic force of the same frequency, causing the rotor to generate a displacement signal of the same frequency. The displacement signal is converted into a voltage signal by a displacement sensor and the difference is calculated with the bias voltage to obtain the displacement error signal X*Sin(ωt+φ) of the rotor relative to the center position. This displacement error signal includes the accumulated value of the hysteresis phase after the control signal passes through the power amplifier, magnetic levitation bearing, rotor, and displacement sensor.

[0009] Construct sinusoidal and cosine signals Sin(ωt) and Cos(ωt) of the same frequency, and multiply the displacement error signal by the sinusoidal and cosine signals respectively to obtain:

[0010] X*Sin(ωt+φ)Sin(ωt)=A / 2[Cosφ-Cos(2ωt+φ)],

[0011] XSin(ωt+φ)*Cos(ωt)=A / 2[Sinφ+Sin(2ωt+φ)];

[0012] The multiplication result is low-pass filtered to obtain the DC components I = XSinφ and Q = XCosφ. According to the formula X = √(I... 2 +Q 2 ), φ=arctan(I / Q), calculate the amplitude X of the displacement error signal and the hysteresis phase φ at the excitation frequency ω;

[0013] Based on the 180° phase difference between the electromagnetic compensation force and the rotor's own unbalance force, the phase angle φ0 that the controller needs to compensate at frequency ω is determined to be 180°-φ. An offline lookup table corresponding to different frequencies ω and compensation angle φ0 is established by frequency sweeping.

[0014] Step 2: Online compensation signal generation

[0015] The notch filter algorithm is used to remove the same frequency current in the power amplifier and suppress the current stiffness force with the same frequency as the rotation speed. At this time, the displacement error caused by the same frequency displacement stiffness force is denoted as e(t)=D*sin(ωt+φ).

[0016] Using the orthogonal demodulation method in step 1, calculate the real-time I and Q values, and obtain the coefficients Cosφ0 and Sinφ0 from the offline lookup table in conjunction with the real-time rotational speed ω; calculate the intermediate coefficients:

[0017] W1=D SinφCosφ0+DCosφSinφ0,

[0018] W2=D CosφCosφ0–D SinφSinφ0,

[0019] The compensation signal is then obtained as W1 COS(ωt) + W2 Sin(ωt), which is a signal with the same frequency but opposite phase as the displacement error, i.e., D sin[ωt+(φ+180°-φ)].

[0020] Step 3: Offline calibration of the compensation current gain Ka

[0021] Set the initial value of Ka to the ratio of displacement stiffness to current stiffness, and make the system run stably at a certain speed ω, and record the vibration amplitude of the rotor at this time.

[0022] The compensation algorithm is activated, the Ka coefficient is gradually adjusted, the change in rotor vibration amplitude is observed, and the Ka value that minimizes vibration is determined as the optimal Ka value at that speed.

[0023] Repeat the above operation for multiple different speed points to establish a lookup table corresponding to the speed and the optimal Ka value, and complete the offline calibration of the compensation current gain Ka.

[0024] In step 1 above, when constructing the sinusoidal signals Sin(ωt) and Cos(ωt) of the same frequency, a lookup table is used to ensure that the signal frequency is consistent with the frequency of the sinusoidal voltage signal output by the controller.

[0025] The process of calculating the real-time I and Q values ​​through quadrature demodulation in step 2 above is the same as the quadrature demodulation principle in offline frequency sweep identification in step 1. Both are achieved by multiplying the displacement error signal with the sine and cosine signals and obtaining the DC component through low-pass filtering.

[0026] When performing offline calibration of the compensation current gain Ka in step 3 above, the selected speed points must cover the commonly used speed range in the actual operation of the magnetic bearing, and the interval between adjacent speed points must be such that the error of the established Ka lookup table is controlled within a preset range during interpolation calculation.

[0027] During the online compensation process, the real-time rotational speed ω is obtained through the rotational speed detection module in the magnetic bearing system, and the detection accuracy of the rotational speed detection module must meet the accuracy requirements for finding the compensation phase angle φ0 and generating the compensation signal.

[0028] In step 1, during the low-pass filtering process, the filter cutoff frequency must be lower than 2ω to effectively filter out the 2ω high-frequency component in the multiplication result and accurately obtain the DC components I and Q.

[0029] The measurement accuracy of the aforementioned displacement sensor must meet the requirement of capturing minute changes in rotor displacement, ensuring that the converted voltage signal can accurately reflect the error of the rotor relative to the center position.

[0030] The compensation signal generated in step 2 needs to be input to the controller. After being superimposed with the original control signal, it is amplified by the power amplifier and applied to the magnetic levitation bearing to cancel the stiffness force of the displacement at the same frequency.

[0031] The advantages of this invention are:

[0032] The present invention provides a magnetic bearing vibration suppression and testing method, which improves the suppression accuracy of the same frequency vibration force of the magnetic levitation rotor, enhances the control accuracy, reduces the vibration force, and improves the system reliability. Moreover, the method involves a small amount of calculation, effectively improves the real-time performance of system control, and has strong practicality and wide applicability. Attached Figure Description

[0033] Figure 1 This is a block diagram of the control system.

[0034] Figure 2 The principle block diagram for compensation.

[0035] Figure 3 This is a simulation diagram of the displacement and vibration force suppression effect of the present invention.

[0036] Figure 4 This is the result of orthogonal demodulation. Detailed Implementation

[0037] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0038] A method for suppressing and testing vibration in magnetic bearings, comprising the following steps:

[0039] Step 1, Offline frequency scanning and identification:

[0040] The controller outputs a sinusoidal voltage signal ASin(ωt). This sinusoidal voltage signal is amplified by a power amplifier and applied to the magnetic levitation bearing to generate an electromagnetic force of the same frequency. The rotor generates a displacement signal of the same frequency after being subjected to the electromagnetic force. This signal is converted into a voltage signal by a displacement sensor and then the difference between the voltage signal and the bias voltage is used to obtain the displacement error signal X*Sin(ωt+φ) of the rotor relative to the center position. This error signal includes the accumulation of the lag phase of the control signal after passing through the power amplifier, magnetic levitation bearing, rotor, and displacement sensor.

[0041] The displacement sensor must be accurate enough to capture minute changes in rotor displacement, ensuring that the converted voltage signal accurately reflects the error of the rotor relative to the center position.

[0042] We construct sine and cosine signals of the same frequency using a lookup table, denoted as Sin(ωt) and Cos(ωt).

[0043] Using orthogonal demodulation, the displacement error signal is multiplied by the constructed sine and cosine signals respectively. The result of the multiplication is:

[0044] X*Sin(ωt+φ)*Sin(ωt)=A / 2[Cosφ-Cos(2wt+φ)],

[0045] X*Sin(ωt+φ)*Cos(ωt)=A / 2[Sinφ+Sin(2wt+φ)];

[0046] After low-pass filtering, the DC components are obtained: I = XSinφ, Q = XCosφ.

[0047] When performing low-pass filtering, the filter cutoff frequency must be lower than 2ω to effectively filter out the 2ω high-frequency component in the multiplication result and accurately obtain the DC components I and Q.

[0048] According to the formula X=√(I) 2 +Q 2 ),φ=arctan(I / Q), can be used to calculate the amplitude X of the displacement error signal and the hysteresis phase φ at the excitation frequency ω.

[0049] Since the electromagnetic compensation force and the rotor's own unbalanced force are opposite in direction (180° out of phase) during online compensation, the phase angle φ0 that the controller needs to compensate for at frequency ω can be obtained as 180° - φ. An offline lookup table of compensation angle φ0 for different frequencies ω is established by frequency sweeping.

[0050] Step 2, Generation of online compensation signal

[0051] First, a notch filter algorithm is used to remove the same-frequency current in the power amplifier, thus suppressing the current stiffness force at the same frequency as the rotation speed. At this point, only the displacement error caused by the same-frequency displacement stiffness force remains, denoted as e(t)=D*sin(ωt+φ).

[0052] Based on the orthogonal demodulation in step 1, calculate the real-time I and Q, and obtain the coefficients Cosφ0 and Sinφ0 by looking up the table using the real-time rotational speed ω.

[0053] The intermediate coefficients were obtained as follows:

[0054] W1=D Sinφ*Cosφ0+D Cosφ*Sinφ0

[0055] W2=D Cosφ*Cosφ0-D Sinφ*Sinφ0;

[0056] Finally, the compensation signal is obtained as W1*COS(wt) + W2*Sin(wt), which constructs a compensation signal with the same frequency but opposite phase.

[0057] D sin[ωt+(φ+φ0)]=D sin[ωt+(φ+180-φ)].

[0058] The real-time rotational speed ω is obtained through the rotational speed detection module in the magnetic bearing system, and the detection accuracy of the rotational speed detection module must meet the accuracy requirements for finding the compensation phase angle φ0 and generating the compensation signal.

[0059] Step 3, Offline calibration of the compensation current gain Ka

[0060] The initial value of Ka can be set as the ratio of displacement stiffness to current stiffness. After stable operation at a certain speed ω, the vibration amplitude of the rotor is recorded. With the compensation algorithm enabled, the coefficient Ka is gradually adjusted, and the vibration amplitude of the rotor is observed to find the coefficient Ka that minimizes vibration. This Ka corresponds to the optimal value at that speed.

[0061] Repeat the steps at multiple speed points to build a lookup table for Ka and complete the calibration. When performing offline calibration of the compensation current gain Ka, the selected speed points must cover the commonly used speed range in the actual operation of the magnetic bearing, and the interval between adjacent speed points must ensure that the error of the established Ka lookup table is controlled within a preset range during interpolation calculation.

[0062] like Figure 1 The diagram shown is a block diagram of the control system of the present invention.

[0063] This invention first uses a notch filter to estimate the control error caused by the same-frequency vibration of the rotational speed, and then subtracts this error in the controller to suppress the same-frequency current stiffness force.

[0064] Reconstruction such as Figure 2 The compensation module shown constructs a sine and cosine signal of the same frequency based on the real-time rotation speed, demodulates it orthogonally with the residual error, calculates the amplitude and phase of the residual error, and obtains the phase compensation angle and gain coefficient Ka by looking up a table to generate a compensation current signal.

[0065] Finally, as Figure 3 The simulation diagram of displacement and vibration force suppression effect is shown. Notch filtering is turned on at 0.3S. At this time, the current at the same frequency drops to 0 and the amplitude decreases; however, there is still residual displacement stiffness force. Compensation is turned on at 1S. After 0.1S, it converges and the displacement and vibration force decrease by 90%.

[0066] Figure 3 The data was obtained through SIMULINK simulation, and the simulation data is the same as the actual motor parameters. The rotor mass is 18 kg, the magnetic bearing current stiffness is 577 N / A, the magnetic bearing displacement stiffness is 2.75e6 N / m, the displacement sensor sensitivity is 5000 V / m, and the rotational frequency is 500 Hz.

[0067] Step 1, the orthogonal demodulation results are as follows Figure 4 The formula shows that the amplitude and phase are 0.014V and 0.36rad (20.6°) respectively, so the corresponding phase compensation angle is 159°.

[0068] That is, the signal to be generated in step 2 is 0.014*sin(2*pi*500*t+159°).

[0069] Finally, the coefficient ka is adjusted (initial value 2.75E6 / 577) to obtain the actual compensation current.

[0070] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the above embodiments do not limit the present invention in any way, and all technical solutions obtained by equivalent substitution or equivalent transformation fall within the protection scope of the present invention.

Claims

1. A method for suppressing and testing vibration in magnetic bearings, characterized in that, The process includes offline frequency sweep identification, online compensation signal generation, and offline calibration of the compensation current gain Ka. The steps are as follows: Step 1: Offline frequency scanning and identification The controller outputs a sinusoidal voltage signal ASin(ωt), which is amplified by a power amplifier and applied to the magnetic levitation bearing to generate an electromagnetic force of the same frequency, causing the rotor to generate a displacement signal of the same frequency. The displacement signal is converted into a voltage signal by a displacement sensor and the difference is calculated with the bias voltage to obtain the displacement error signal X*Sin(ωt+φ) of the rotor relative to the center position. This displacement error signal includes the accumulated value of the hysteresis phase after the control signal passes through the power amplifier, magnetic levitation bearing, rotor, and displacement sensor. Construct sinusoidal and cosine signals Sin(ωt) and Cos(ωt) of the same frequency, and multiply the displacement error signal by the sinusoidal and cosine signals respectively to obtain: X*Sin(ωt+φ)Sin(ωt)=A / 2[Cosφ-Cos(2ωt+φ)], X*Sin(ωt+φ)*Cos(ωt)=A / 2[Sinφ+Sin(2ωt+φ)]; The multiplication result is low-pass filtered to obtain the DC components I = XSinφ and Q = XCosφ. According to the formula X = √(I... 2 +Q 2 ), φ=arctan(I / Q), calculate the amplitude X of the displacement error signal and the hysteresis phase φ at the excitation frequency ω; Based on the 180° phase difference between the electromagnetic compensation force and the rotor's own unbalance force, the phase angle φ0 that the controller needs to compensate at frequency ω is determined to be 180°-φ. An offline lookup table corresponding to different frequencies ω and compensation angle φ0 is established by frequency sweeping. Step 2: Online compensation signal generation The notch filter algorithm is used to remove the same frequency current in the power amplifier and suppress the current stiffness force with the same frequency as the rotation speed. At this time, the displacement error caused by the same frequency displacement stiffness force is denoted as e(t)=D*sin(ωt+φ). Using the orthogonal demodulation method in step 1, calculate the real-time I and Q values, and obtain the coefficients Cosφ0 and Sinφ0 from the offline lookup table in combination with the real-time rotational speed ω; Calculate the intermediate coefficients: W1=D SinφCosφ0+DCosφSinφ0, W2=D CosφCosφ0–D SinφSinφ0, The compensation signal is then obtained as W1 COS(ωt) + W2 Sin(ωt), which is a signal with the same frequency but opposite phase as the displacement error, i.e., D sin[ωt+(φ+180°-φ)]. Step 3: Offline calibration of the compensation current gain Ka Set the initial value of Ka to the ratio of displacement stiffness to current stiffness, and make the system run stably at a certain speed ω, and record the vibration amplitude of the rotor at this time. The compensation algorithm is activated, the Ka coefficient is gradually adjusted, the change in rotor vibration amplitude is observed, and the Ka value that minimizes vibration is determined as the optimal Ka value at that speed. Repeat the above operation for multiple different speed points to establish a lookup table corresponding to the speed and the optimal Ka value, and complete the offline calibration of the compensation current gain Ka.

2. The method according to claim 1, characterized in that, In step 1, when constructing the sinusoidal signals Sin(ωt) and Cos(ωt) of the same frequency, a lookup table is used to ensure that the signal frequency is consistent with the frequency of the sinusoidal voltage signal output by the controller.

3. The method according to claim 1, characterized in that, The process of calculating the real-time I and Q values ​​through quadrature demodulation in step 2 is the same as the quadrature demodulation principle in offline frequency sweep identification in step 1. Both are achieved by multiplying the displacement error signal with the sine and cosine signals and obtaining the DC component through low-pass filtering.

4. The method according to claim 1, characterized in that, When performing offline calibration of the compensation current gain Ka in step 3, the selected rotational speed points must cover the commonly used rotational speed range in the actual operation of the magnetic bearing, and the interval between adjacent rotational speed points must be such that the error of the established Ka lookup table is controlled within a preset range during interpolation calculation.

5. The method according to claim 1, characterized in that, During the online compensation process, the real-time rotational speed ω is obtained through the rotational speed detection module in the magnetic bearing system, and the detection accuracy of the rotational speed detection module must meet the accuracy requirements for finding the compensation phase angle φ0 and generating the compensation signal.

6. The method according to claim 1, characterized in that, In step 1, during the low-pass filtering process, the filter cutoff frequency must be lower than 2ω to effectively filter out the 2ω high-frequency component in the multiplication result and accurately obtain the DC components I and Q.

7. The method according to claim 1, characterized in that, The displacement sensor must be accurate enough to capture minute changes in rotor displacement, ensuring that the converted voltage signal accurately reflects the error of the rotor relative to the center position.

8. The method according to claim 1, characterized in that, The compensation signal generated in step 2 needs to be input to the controller. After being superimposed with the original control signal, it is amplified by the power amplifier and applied to the magnetic levitation bearing to cancel the stiffness force of the displacement at the same frequency.

Images