An online dynamic balancing method for a magnetic levitation molecular pump based on a rotating electromagnetic vector field

Abstract

The present invention relates to an online dynamic balancing method for a magnetic levitation molecular pump based on a rotating electromagnetic vector field. The method comprises the following steps: constructing a rotor dynamic model based on a running magnetic levitation molecular pump, and establishing an imbalance identification model based on the dynamic model; performing correction and optimization on the imbalance identification model to obtain an optimized imbalance identification model; and identifying a correction mass based on the optimized imbalance identification model, incorporating the correction mass into the rotor's operation, thereby completing the online dynamic balancing of the magnetic levitation molecular pump. The method can accurately reflect the rotor's imbalance under conventional control modes and effectively address the issues of synchronous response interference and model parameter errors during the calculation process, thereby achieving accurate identification of rotor imbalance.

Description

Technical Field

[0001] The present invention relates to the technical field of active magnetic bearing control, in particular to an online dynamic balancing method of a magnetic suspension molecular pump based on a rotating electromagnetic vector field. Background Art

[0002] A molecular pump utilizes a high-speed rotating turbine rotor, coupled with a multi-stage turbine structure, to achieve directional velocity, transfer energy to gas molecules, and compress and drive them toward the exhaust port. It is widely used in fields such as space science, materials science, the electric vacuum industry, and the microelectronics and semiconductor industries. Compared to mechanical bearing molecular pumps, magnetic levitation molecular pumps offer significant advantages in terms of service life, oil-free lubrication, and noise reduction, making them an inevitable trend in the development of molecular pump technology.

[0003] Due to various factors, including material inhomogeneity, errors in machining and assembly, and even asymmetric rotor designs, the rotor's principal axis of inertia and geometric axis may not fully align, inevitably leading to rotor imbalance. Even after long-term operation, even a dynamically balanced molecular pump system can experience residual unbalance due to issues like rotor wear or blade corrosion, leading to a deterioration in dynamic balancing. Because magnetic levitation molecular pumps operate at high speeds, this accumulated rotor mass imbalance, if not promptly detected, can induce significant synchronous vibration, causing mechanical vibration and even collision between the rotor and the grounded bearing, potentially leading to serious accidents. Excessive system vibration can cause the power amplifier to operate in its saturation region for extended periods, resulting in poor rotor levitation accuracy and hindering stable system control. Therefore, residual unbalance is a key indicator for health monitoring of magnetic levitation rotating machinery products. Efficient rotor unbalance detection and high-precision online dynamic balancing at actual operating speeds are crucial prerequisites for the application and maintenance of molecular pump systems.

[0004] Rotor balancing methods are primarily categorized as offline and online. The fundamental principle behind both offline and online balancing is to achieve alignment between the rotor's principal axis of inertia and its geometric axis by physically adding or removing weight. Compared to traditional offline balancing, online balancing utilizes the rotor system's own sensors and algorithms, offering advantages such as eliminating the need for disassembly and achieving better balancing results. Therefore, it holds broad application prospects. Traditional online balancing methods, such as modal balancing and the influence coefficient method, rely on multiple trial weight additions to achieve the desired result, resulting in very low balancing efficiency.

[0005] Compared to patent CN106153256A, the innovation of this invention lies in: considering the control method under the actual operating conditions of the magnetic levitation molecular pump, a conventional control-based imbalance identification model is constructed. Furthermore, high-precision imbalance identification is achieved simply by adding a rotating electromagnetic field, without the need for adding a test weight. Compared to patent CN103604563A, this invention introduces a stiffness coefficient correction optimization step, which improves the accuracy of imbalance identification. Furthermore, only a single application of a rotating electromagnetic field at the same frequency is required, improving dynamic balancing efficiency. Summary of the Invention

[0006] The purpose of the present invention is to provide an online dynamic balancing method for a magnetic levitation molecular pump based on a rotating electromagnetic vector field. The method does not require any restrictions on the operating state of the rotor or the addition of test weights. Only a rotating electromagnetic field needs to be added once to achieve correction and optimization of the stiffness coefficient. The method can identify the imbalance amount at a single start under actual rotor operating conditions. The method has the advantages of high imbalance amount identification accuracy, high operating efficiency, and no impact on the stability of the molecular pump system.

[0007] To achieve the above object, the present invention provides the following solutions:

[0008] An online dynamic balancing method for a magnetic levitation molecular pump based on a rotating electromagnetic vector field, comprising:

[0009] A dynamic model of the rotor is constructed based on the running magnetic levitation molecular pump, and an imbalance identification model is established according to the dynamic model;

[0010] Correcting and optimizing the imbalance identification model to obtain an optimized imbalance identification model;

[0011] The correction mass is identified based on the optimized unbalance identification model, and the correction mass is added to the operation of the rotor to complete the online dynamic balancing of the magnetic levitation molecular pump.

[0012] Furthermore, establishing the imbalance identification model includes:

[0013] An unbalance amount identification model including rotor unbalance disturbance force is constructed based on the operating state of the rotor in a conventional control mode, and the unbalance amount identification model is generated by eliminating the rotor unbalance disturbance force in the unbalance amount identification model including the rotor unbalance disturbance force.

[0014] Furthermore, the unbalance disturbance force includes: rotor static unbalance disturbance force and rotor dynamic unbalance disturbance force, wherein the rotor static unbalance disturbance force is eliminated by bearing control, and the rotor dynamic unbalance disturbance force is eliminated by a dual-plane correction method.

[0015] Furthermore, the imbalance identification model is:

[0016] M=TV

[0017] Where M is the correction mass matrix, T is the conversion coefficient matrix, and V is the synchronization response.

[0018] Furthermore, correcting and optimizing the imbalance identification model to obtain the optimized imbalance identification model includes:

[0019] Using a sine-cosine sequence correlation operation to extract a signal synchronization response of the rotor of the magnetic levitation molecular pump to obtain a first speed and frequency signal; adding a rotating electromagnetic vector field to the first speed and frequency signal to obtain a second speed and frequency signal;

[0020] The unbalance identification model is optimized for stiffness coefficient correction based on the first speed and frequency signal and the second speed and frequency signal to generate the optimized unbalance identification model.

[0021] Furthermore, extracting the signal synchronization response of the magnetic levitation molecular pump by using a sine-cosine sequence correlation operation includes:

[0022] A rotor radial displacement signal and a bearing control current signal are obtained, wherein both the rotor radial displacement signal and the bearing control current signal include a speed-same-frequency signal and other low- and high-frequency interference signals; the rotor radial displacement signal and the bearing control current signal are respectively subjected to a sine-cosine sequence correlation operation to eliminate the other low- and high-frequency interference signals, and a first speed-same-frequency signal is extracted.

[0023] Furthermore, the rotor radial displacement signal and the bearing control current signal are respectively subjected to a sine-cosine sequence correlation operation to eliminate other low- and high-frequency interference signals, and the first speed same-frequency signal is extracted, comprising:

[0024] According to the decomposition principle of periodic signals, the rotor radial displacement signal and the bearing control current signal collected by the sensor can be decomposed into the sum of frequency signals with different amplitudes and initial phases, and the average is calculated with the sine and cosine sequence of the same frequency:

[0025]

[0026] Among them, u in (t) is the input signal sequence; ω i For different frequency values, including rotation frequency ω0 and other frequencies ω i (i≠0); A i Indicates frequency ω i The amplitude of the signal; θ i is the frequency T i The initial phase of the signal; the first speed frequency signal is u out(t) = A0sin(ω0t+θ0).

[0027] Furthermore, generating the optimized imbalance identification model includes:

[0028] By comparing the changes of the first speed same-frequency signal and the second speed same-frequency signal, a correction and optimization conversion coefficient matrix is ​​calculated, and the stiffness coefficient of the imbalance identification model is corrected and optimized based on the correction and optimization conversion coefficient matrix to generate the optimized imbalance identification model.

[0029] Furthermore, the mathematical model for calculating the correction-optimized conversion coefficient matrix is:

[0030] T2(V1-V v )=M v

[0031] Among them, T2 is the optimized conversion coefficient matrix, V1 and V v are the first speed same-frequency signal matrix and the second speed same-frequency signal matrix, M v is the equivalent centrifugal mass matrix of the rotating electromagnetic field.

[0032] Furthermore, the calculation model for correcting and optimizing the stiffness coefficient of the imbalance identification model is:

[0033] K2=-[A1(V v -V1), A2(V v -V1)][B1(V v -V1)+C1,B2(V v -V1)+C2] -1

[0034] Among them, K2 is the stiffness coefficient matrix, and A1, B1, C1, A2, B2, C2 are process operation matrices.

[0035] The beneficial effects of the present invention are:

[0036] The present invention performs dynamic modeling based on the operating state of the rotor under the conventional control mode, constructs a mathematical model for identifying the imbalance amount that meets the actual operating conditions of the rotor, and can accurately reflect the imbalance amount of the rotor during actual operation; on the basis of taking into account the parameter correction accuracy, the present invention proposes a parameter correction method that does not require adding trial weights and rotor shutdown, designs a rotating electromagnetic field, and calculates the correction optimization value of the system stiffness coefficient by comparing the rotor synchronous response matrix before and after adding. There is no need to start and stop the rotor multiple times, which can reduce the shutdown cost and obtain higher parameter correction accuracy, effectively solves the problems of synchronous response mixed interference and model parameter errors in the calculation process, and realizes accurate identification of the rotor imbalance amount. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0038] Figure 1 This is a flow chart of an online dynamic balancing method for a magnetic levitation molecular pump based on a rotating electromagnetic vector field according to an embodiment of the present invention;

[0039] Figure 2 This is a diagram showing the structure and model coordinate definition of an active magnetic levitation rotor according to an embodiment of the present invention;

[0040] Figure 3 Schematic diagram of rotor imbalance according to an embodiment of the present invention, wherein FIG (a) is a schematic diagram of static rotor imbalance, and FIG (b) is a schematic diagram of dynamic rotor imbalance;

[0041] Figure 4 Schematic diagram of the rotor structure of an embodiment of the present invention;

[0042] Figure 5 This is a structural block diagram of the same-frequency extractor according to an embodiment of the present invention. DETAILED DESCRIPTION

[0043] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0044] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0045] This embodiment provides an online dynamic balancing method for a magnetic levitation molecular pump based on a rotating electromagnetic vector field. Figure 1 Shown, including:

[0046] S1. Construct an imbalance identification model based on the rotor motion state under conventional control mode.

[0047] Based on the rigid rotor dynamics, an unbalance identification model including the rotor unbalance disturbance force is constructed. The rotor unbalance disturbance force includes the rotor static unbalance disturbance force and the dynamic unbalance disturbance force. Since the inertia axis and the rotation axis do not coincide with each other, centrifugal force and centrifugal torque are generated. By using the force decomposition principle, such as Figure 2As shown, it is decomposed into centrifugal force and centrifugal torque generated by the offset of the inertia principal axis-geometric axis and the geometric axis-rotation axis. The centrifugal force and centrifugal torque generated by the offset of the geometric axis-rotation axis can be eliminated by bearing control, and the centrifugal force and centrifugal torque generated by the offset of the inertia principal axis-geometric axis can be eliminated by the centrifugal force and centrifugal torque of the correction mass of the dual-plane correction method; the correction mass reflected on the two counterweight disks is obtained by the dual-plane correction method, wherein the imbalance identification includes the correction mass reflected on the two counterweight disks A and B by the dual-plane correction method, and the centrifugal force / torque of the correction mass is offset with the static imbalance / dynamic imbalance disturbance force; the first signal is obtained based on the rotor radial displacement sensor embedded in the inner side of the magnetic levitation bearing coil and the current sensor in the coil. The rotor structure is shown in FIG. Figure 4 shown.

[0048] The specific process of S1 is as follows:

[0049] Define the geometric axis g, the principal axis of inertia w, and the axis of rotation r. The center of mass is denoted by C, and the radial plane perpendicular to the axis of rotation r is defined as U. Obviously, the center of mass C is the intersection of the principal axis of inertia and plane U. The intersections of the geometric axis g and the axis of rotation w with plane U are denoted as the geometric center O and π, respectively.

[0050] like Figure 3 As shown in (a), the ПXY coordinate system is constructed with the rotation center П as the origin, and the coordinates of O in the inertial fixed coordinate system ΠXY are defined as (x, y). Since the Y coordinate phase value leads the X direction phase by π / 2, the vector from Π to O can be expressed as R = x + y × j using the complex coefficient method. Where j is an imaginary unit. Similarly, the distance from O to C, that is, the eccentricity of the center of mass e l It can be expressed as:

[0051]

[0052] Among them, e l and They represent the initial relative distance and angle between the rotor mass center and the geometric center, which are determined by the static balance quality characteristics of the rotor itself and do not change with the rotor speed; Ω is the rotor rotational angular velocity.

[0053] Therefore, in the ΠXY coordinate system, the rotor center of mass position can be expressed as:

[0054] C=R+e l

[0055] The equation of motion about the center of mass C is characterized by:

[0056]

[0057] Where m represents the rotor mass, is the second derivative of the displacement vector from Π to C, and f is the electromagnetic force acting on the rotor.

[0058] Combining the above equations we get:

[0059]

[0060] According to Euler's equation, the tilting motion of the principal axis of inertia ω relative to the geometric axis g is:

[0061]

[0062] Among them, e gω and They represent the deflection angles between the geometric axis and the principal axis of inertia in the rαβ coordinate system, that is, the initial angular modulus and initial phase angle of dynamic imbalance.

[0063] like Figure 3 As shown in (b), in the rαβ coordinate system, the rotor inertia principal axis angle can be expressed as:

[0064] w=γ+e gω

[0065] Using the Euler equations of rigid body dynamics, the tilt motion of the inertial principal axis can be expressed by its angular displacement vector w relative to the rotation axis as:

[0066]

[0067] Among them, J r and J z represents the lateral and polar moments of inertia of the rotor respectively, w is the angular displacement vector from the rotation axis r to the principal axis of inertia ω, and P represents the electromagnetic torque acting on the rotor.

[0068] The equivalent rotational motion equation with dynamic imbalance is obtained as follows:

[0069]

[0070] in, and They represent the centrifugal moment and gyroscopic moment caused by the rotation axis not being parallel to the geometric axis, J r e gω Ω 2 With J z e gω Ω 2 It represents the moment caused by the principal axis of inertia offsetting the geometric axis.

[0071] According to the above equations, the distance and deflection angle between the principal axis of inertia and the geometric axis are:

[0072]

[0073] Among them, J s Meet J s =J r -J z .

[0074] The electromagnetic force of the magnetic levitation for unbalance identification can be expressed as:

[0075]

[0076] Among them, F a , F b They represent the bearing force acting on the rotor at the AB end respectively; i a ,i b The control current of the bearing coils at the AB ends respectively; s a , s b are the radial displacements of the rotor at the bearing coils at the AB ends; k ia , k ib are the electromagnetic force-current stiffness at the AB end respectively; k sa , k sb are the electromagnetic force-displacement stiffness at ends AB respectively.

[0077] The radial bearing force calculation equation is:

[0078]

[0079] Among them, F ax , F bx , F ay , F by Represents the radial magnetic bearing forces of the four sets of bearing coils; i ax ,i ay ,i bx and i by is the current in the corresponding coil winding; s ax , s ay , s bx and s by They represent the radial displacement of the rotor relative to the geometric center of the magnetic bearing on the AB end coil bearing plane.

[0080] Using the complex coefficient method, each physical quantity can be expressed as:

[0081]

[0082] Therefore, the calculation equations for the resultant force F and resultant torque P generated by the electromagnetic force acting on the rotor are:

[0083]

[0084] Among them, L ma , L mbRepresents the AB end bearing coil B a , B b The vertical distance from the rotor geometric center O.

[0085] The centrifugal force and torque generated by the electromagnetic force and torque and the geometric axis-rotation axis offset are constructed as follows:

[0086]

[0087] Where R represents the deviation between the geometric axis and the rotation axis, w represents the inclination angle between the geometric axis and the rotation axis, and J r and J z Represents the lateral and polar moments of inertia of the rotor.

[0088] like Figure 4 As shown, due to the bearing space position limitation, the displacement sensor cannot be installed on the coil bearing B a , B b The displacement sensor measures the radial displacement of the rotor on the Sa and Sb planes and is recorded as x. ax , x ay , x bx and x by .

[0089]

[0090] Among them, L sa , L sb They represent the vertical distances between the sensor mounting planes Sa and Sb at the AB ends and the geometric center O, respectively, and L ca , L cb Represents the AB end counterweight plate P a , P b The distance from the geometric center O, r a and r b They represent the radial distances of the correction mass installation positions on the AB end counterweight plates relative to the rotation axis.

[0091] Using the displacement information collected by the sensor, the radial offset R and angular offset γ between the geometric axis and the rotation axis can be obtained:

[0092]

[0093] The calculation formula for the radial displacement of the rotor at the bearing coil is obtained as follows:

[0094]

[0095] Therefore, the calculation equations for the electromagnetic resultant force and torque of the magnetic bearing are:

[0096]

[0097] Among them, k s =k sa +k sb , Γ=k sb L mb -k sa L ma , Λ=L ma 2 k sa +L mb 2 k sb .

[0098] The correction quality calculated by the dual-plane correction method can be expressed as:

[0099]

[0100] Among them, m cax , m cay is the component of the correction mass required at end A of the balancing disk in the x and y directions, m cbx , m cby is the component of the correction mass required at the B end of the balancing disk in the x and y directions, j is the imaginary unit, m ca , m cb It is the complex coefficient expression of the correction mass at the A and B ends of the balancing disk.

[0101] The rotor imbalance is eliminated by adding a weight plate P a , p b Add correction mass implementation, add correction mass m of specific size and angle respectively ca and m cb , the centrifugal force generated can be expressed as:

[0102]

[0103] Among them, f ca , f cb is the centrifugal force generated by the correction mass, r a , r b is the distance between the correction mass installation position and the shaft, and Ω is the rotor equilibrium speed.

[0104] The dynamic balancing of the rotor is to offset the rotor imbalance caused by the offset between the inertial principal axis and the geometric axis by correcting the mass centrifugal force. It is manifested as the force and torque acting on the entire rotor are zero, which can be expressed as:

[0105]

[0106] Where m is the rotor mass, J r is the rotor polar moment of inertia, el is the deviation moment between the geometric axis and the principal axis of inertia, e gw is the inclination angle between the geometric axis and the principal axis of inertia, L ca , L cb are the distances between the counterweight surface at ends AB and the center of mass respectively.

[0107] Theoretically, due to the influence of unbalanced mass, the rigid rotor only responds synchronously at the balanced speed. Therefore, according to the dynamic laws of circular motion, the hypothetical equations for generalized displacement and inclination are obtained as follows:

[0108]

[0109] Using this assumption, the correction mass can be calculated directly using the coil current and rotor displacement response in the magnetic bearing system, without having to solve physical quantities such as the acceleration, angular velocity, and angular acceleration of the rotor center of mass, which greatly simplifies the imbalance solution process. The calculation equation for the unbalanced mass is:

[0110]

[0111] In order to intuitively express the relationship between sensor measurement data and unbalanced mass, the unbalanced mass calculation formula is expressed in matrix form as follows:

[0112] M=TV

[0113] Where M = [m ca , m cb ] T Represents the unbalanced mass, V=[x a , x b ,i a ,i b ] T represents the synchronous response of the rotor, Represents the correction mass transformation coefficient matrix.

[0114] S2. Accelerate the rotor to the target speed, perform synchronous response extraction, and obtain a first speed and frequency signal.

[0115] Construct a signal synchronization response extraction using the sine-cosine sequence correlation operation, wherein the signal synchronization response extraction using the sine-cosine sequence correlation operation includes the rotor radial displacement signal and the bearing control current signal, and the rotor radial displacement signal and the bearing control current signal both include the speed frequency signal and other low-high frequency interference signals; the rotor radial displacement signal and the bearing control current signal are respectively subjected to correlation operation to eliminate the low-high frequency interference signals, thereby extracting the speed frequency signal, which is the synchronization response. The structural block diagram of the frequency extractor is shown in FIG. Figure 5 shown.

[0116] The specific process of S2 is as follows:

[0117] Considering the input signal u in (t) is mixed with the synchronous response of the speed frequency ω0 and other frequency ω i According to the decomposition principle of periodic signals, the input signal after zero bias removal can be decomposed into:

[0118]

[0119] Among them, u in (t) is the input signal sequence, ω i For different frequency values, including rotation frequency ω0 and other frequencies ω i (i≠0), A i Representative frequency ω i The amplitude of the signal, θ i Represents the initial phase of the signal, and the speed frequency can be expressed as u out (t) = A0sin(ω0t+θ0).

[0120] To calculate A0 and θ0, the same frequency quantity can be decomposed into sine and cosine components:

[0121] u out (t) = u sin sin(ω0t)+u cos cos(ω0t)

[0122] Based on the constructed sine and cosine reference sequence, it is only necessary to calculate the amplitude u of the sine and cosine components sin with u cos That's it. sin For example, the signal within the time range of 0 to T is calculated:

[0123]

[0124] Since the rotor radial displacement and coil control current are limited, when the signal acquisition time is long enough, it can be considered that T>>A i The signal is divided into the speed frequency sequence A0sin(ω0t+θ0)sin(ω0t) and the interference frequency signal A i sin(ω i t+θ i )sin(ω0t). According to the trigonometric function integration law, the amplitude of each frequency signal after processing can be simplified to:

[0125]

[0126] So we can calculate u sin =A0cos(θ0). Similarly, we can get u cos =A0sin(θ0), thus obtaining the output signal u out(t)=A0sin(ω0t+θ0), thereby realizing the extraction of the same frequency signal.

[0127] S3. Add a rotating electromagnetic vector field to the rotor in the target speed operating state, obtain a second speed synchronous frequency signal, optimize the stiffness coefficient and unbalance identification model, and identify the correction quality.

[0128] Construct a system stiffness coefficient correction optimization process based on a rotating electromagnetic field. The system stiffness coefficient correction optimization based on a rotating electromagnetic field includes the current-force stiffness coefficient and the displacement-force stiffness coefficient. The actual rotor radial displacement signal and bearing control current signal are obtained, and the signal synchronization response extraction is realized based on the sine-cosine sequence correlation operation. The rotating electromagnetic vector field is added, and the front and rear rotor synchronization responses are compared to realize the numerical correction optimization of the system stiffness coefficient. The specific process is as follows:

[0129] The added rotating electromagnetic field can be expressed as:

[0130]

[0131] Among them, U 1x , U 1y , U 2x , U 2y They represent the components of the rotating voltage generated by the bearing coils at the AB ends relative to the Hall reference angular position in the x and y directions respectively.

[0132] In order to minimize the impact of the electromagnetic field on the rotor's operating stability, the electromagnetic force generated by the rotating voltage is set to 0, that is:

[0133]

[0134] Assuming that the resistance and inductance of the bearing coil are R and L respectively, the rotating electromagnetic field control voltage U v The same frequency current I in the bearing coil v satisfy:

[0135] U v =RI v +Li v =(R+LΩj)I v

[0136] To simplify the expression, I v =[I va ;I vb ] represents the added same-frequency current, and the resulting same-frequency electromagnetic force f on the AB end bearing va =k ia I va , f vb =k ib I vb Where fva , f vb Represent the electromagnetic force of the same frequency added to the AB end, I va , I vb They respectively represent the same-frequency currents that need to be applied.

[0137] According to the equivalence law between bearing electromagnetic force and mass centrifugal force, the equivalent centrifugal mass M of the rotating electromagnetic field can be obtained. v =[M va ;M vb ] is:

[0138]

[0139] The equivalent centrifugal mass generated by the rotating electromagnetic field is expanded according to the real and imaginary parts as follows:

[0140] M vr =[real(M v ), imag(M v )].

[0141]

[0142] The online dynamic balancing process of the correction mass transformation coefficient matrix optimization correction process is as follows: first, accelerate the rotor to the balanced speed to obtain the synchronous response V1 under the normal control mode, and use the initial transformation coefficient matrix T1 calculated by the design parameters to calculate the unbalanced mass, which is recorded as M1:

[0143] M1=T1V1Add the same-frequency control voltage in the control program, compare the changes in the synchronous response of the rotor before and after, and calculate the principle equation of the correction optimization conversion coefficient matrix:

[0144] T2(V1-V v )=M v

[0145] Among them, T2 is the optimized conversion coefficient matrix, V1 and V v is the rotor synchronous response matrix before and after the addition of the rotating electromagnetic field, M v is the equivalent centrifugal mass matrix of the rotating electromagnetic field.

[0146] Specifically, keep the rotor running at the equilibrium speed and add the corresponding equivalent centrifugal mass M vr The rotating electromagnetic field U v , get the synchronous response V in the new state v . The synchronous response matrix V1, V v Expanded by real and imaginary parts

[0147] V 1r =[real(V1),imag(V1)],V vr=[real(V v ), imag(V v )].

[0148] Combining the synchronous responses of the rotor in the two equilibrium states, the calculation principle equation of the state transition matrix is ​​obtained as follows:

[0149] T2(V 1r -V vr )=M vr

[0150] In order to directly solve the stiffness coefficient by using matrix operations, the stiffness coefficient matrix K is defined as [k sa , k sb , k ia , k ib ].

[0151] Split the correction mass transform coefficient matrix into two row matrices:

[0152] T=[T up ;T down ]

[0153] Among them, T up =[C a , D a , E a , F a ],T down =[C b , D b , E b , F b ].

[0154] The stiffness coefficient matrix is ​​extracted as follows:

[0155]

[0156] Among them, A1 and A2 are 1×4 matrices, B1 and B2 are 4×4 square matrices, and their expressions are:

[0157]

[0158]

[0159]

[0160] Similarly, M vr The stiffness coefficient matrix in is extracted, and the equation can be obtained:

[0161]

[0162] Among them, the expressions of C1 and C2 are:

[0163]

[0164]

[0165] The correction optimization principle equation of the correction quality transformation matrix can be split into two upper and lower row matrices. Using the matrix calculation rules, the correction equations of the two transformation coefficient row matrices are obtained as follows:

[0166]

[0167] Combining the above equations, we can get:

[0168]

[0169] Among them, B1(V v -V1)+C1 and B2(V v -V1)+C2 are both matrices with 4 rows and 2 columns.

[0170] To simplify the calculation process, the two equations are recombined as follows:

[0171] 4×4 square matrix K B =[B1(V v -V1)+C1,B2(V v -V1)+C2];

[0172] 1×4 matrix K A =[A1(V v -V1), A2(V v -V1)].

[0173] Therefore, the calculation equation of the stiffness coefficient matrix is ​​obtained as follows:

[0174] K2=-[A1(V v -V1), A2 (V v -V1)][B1(V v -V1)+C1,B2(V v -V1)+C2] -1

[0175] Among them, K2 is the stiffness coefficient matrix, A1, B1, C1 are process operation matrices.

[0176] Simplified to:

[0177]

[0178] The actual stiffness coefficient value under the rotor balance speed condition is obtained by calculation, thereby obtaining the corrected mass transformation coefficient matrix T2, and the actual corrected mass calculation equation is obtained as follows:

[0179] M2=T2V1.

[0180] S4. Stop the rotor, add the obtained correction mass to the rotor, and determine whether the residual unbalance meets the balancing accuracy level. If so, end the balancing process; if not, repeat steps S2 to S4 until the residual unbalance meets the balancing accuracy level.

[0181] The present invention performs dynamic modeling based on the operating state of the rotor under the conventional control mode, constructs a mathematical model for identifying the imbalance amount that meets the actual operating conditions of the rotor, and can accurately reflect the imbalance amount of the rotor during actual operation; on the basis of taking into account the parameter correction accuracy, the present invention proposes a parameter correction method that does not require adding trial weights and rotor shutdown, designs a rotating electromagnetic field, and calculates the correction optimization value of the system stiffness coefficient by comparing the rotor synchronous response matrix before and after adding. There is no need to start and stop the rotor multiple times, which can reduce the shutdown cost and obtain higher parameter correction accuracy, effectively solves the problems of synchronous response mixed interference and model parameter errors in the calculation process, and realizes accurate identification of the rotor imbalance amount.

[0182] The embodiments described above are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by persons skilled in the art should fall within the scope of protection defined by the claims of the present invention.

Claims

1. A method for online dynamic balancing of a magnetic levitation molecular pump based on a rotating electromagnetic vector field, characterized in that: include: A dynamic model of the rotor is constructed based on the running magnetic levitation molecular pump, and an imbalance identification model is established according to the dynamic model; Correcting and optimizing the imbalance identification model to obtain an optimized imbalance identification model; Identifying a correction mass based on the optimized imbalance identification model, adding the correction mass to the operation of the rotor, and completing the online dynamic balancing of the magnetic levitation molecular pump; Correcting and optimizing the imbalance identification model to obtain an optimized imbalance identification model includes: Using a sine-cosine sequence correlation operation to extract a signal synchronization response of the rotor of the magnetic levitation molecular pump to obtain a first speed and frequency signal; adding a rotating electromagnetic vector field to the first speed and frequency signal to obtain a second speed and frequency signal; performing stiffness coefficient correction optimization on the imbalance identification model based on the first speed synchronous frequency signal and the second speed synchronous frequency signal to generate the optimized imbalance identification model; Extracting the signal synchronization response of the magnetic levitation molecular pump by using the sine-cosine sequence correlation operation includes: Acquire a rotor radial displacement signal and a bearing control current signal, wherein both the rotor radial displacement signal and the bearing control current signal include a rotational speed frequency signal and other low- and high-frequency interference signals; perform a sine- and cosine-sequence correlation operation on the rotor radial displacement signal and the bearing control current signal, eliminate the other low- and high-frequency interference signals, and extract a first rotational speed frequency signal; The rotor radial displacement signal and the bearing control current signal are respectively subjected to a sine-cosine sequence correlation operation to eliminate other low- and high-frequency interference signals, and the first speed same-frequency signal is extracted, which includes: According to the decomposition principle of periodic signals, the rotor radial displacement signal and the bearing control current signal collected by the sensor can be decomposed into the sum of frequency signals with different amplitudes and initial phases, and the average is calculated with the sine and cosine sequence of the same frequency: Among them, u in (t) is the input signal sequence; ω i For different frequency values, including rotation frequency ω0 and other frequencies ω i (i≠0); A i Indicates frequency ω i The amplitude of the signal; θ i is the frequency ω i The initial phase of the signal; the first speed frequency signal is u out (t) = A0sin(ω0t+θ0); Generating the optimized imbalance identification model includes: By comparing changes in the first speed and frequency signal and the second speed and frequency signal, a correction and optimization conversion coefficient matrix is ​​calculated, and based on the correction and optimization conversion coefficient matrix, the stiffness coefficient of the imbalance identification model is corrected and optimized to generate the optimized imbalance identification model; The mathematical model for calculating the conversion coefficient matrix of the correction optimization is: T2(V1-V v )=M v Among them, T2 is the optimized conversion coefficient matrix, V1 and V v are the first speed same-frequency signal matrix and the second speed same-frequency signal matrix, M v is the equivalent centrifugal mass matrix of the rotating electromagnetic field; The calculation model for correcting and optimizing the stiffness coefficient of the imbalance identification model is: K2=-[A1(V v -V1), A2(V v -V1][B1(V v -V1)+C1,B2(V v -V1)+C2] -1 Among them, K2 is the stiffness coefficient matrix, and A1, B1, C1, A2, B2, C2 are process operation matrices.

2. The online dynamic balancing method of a magnetic levitation molecular pump based on a rotating electromagnetic vector field according to claim 1, characterized in that: Establishing the imbalance identification model includes: An unbalance amount identification model including rotor unbalance disturbance force is constructed based on the operating state of the rotor in a conventional control mode, and the unbalance amount identification model is generated by eliminating the rotor unbalance disturbance force in the unbalance amount identification model including the rotor unbalance disturbance force.

3. The online dynamic balancing method of a magnetic levitation molecular pump based on a rotating electromagnetic vector field according to claim 2, characterized in that: The unbalance disturbance force includes: a rotor static unbalance disturbance force and a rotor dynamic unbalance disturbance force, wherein the rotor static unbalance disturbance force is eliminated by bearing control, and the rotor dynamic unbalance disturbance force is eliminated by a dual-plane correction method.

4. The online dynamic balancing method of a magnetic levitation molecular pump based on a rotating electromagnetic vector field according to claim 3 is characterized in that: The imbalance identification model is: M=TV Where M is the correction mass matrix, T is the conversion coefficient matrix, and V is the synchronization response.

Images