A compensating device and method for radial runout component of magnetic suspension rotor position

Abstract

The present invention discloses a device and method for compensating radial runout components of a magnetically suspended rotor position. These devices relate to the technical field of high-speed magnetically suspended motors and include a hardware-in-the-loop (HIL) simulation system. The HIL simulation system comprises a high-speed magnetically suspended motor and its control system, which together constitute the physical "hardware" system of the HIL simulation system. Furthermore, the HIL system comprises a dynamic mathematical model of the high-speed magnetically suspended motor and its control system, which forms a simulation system within a computer digital space. The high-speed magnetically suspended motor comprises a magnetically suspended rotor assembly, radial / axial electromagnetic bearings, radial / axial displacement sensors for the rotor assembly, a three-phase motor stator, and an angular displacement sensor for the magnetically suspended rotor assembly. The present invention utilizes HIL simulation and artificial intelligence computer optimization to determine the Fourier coefficient parameters of each harmonic of the radial runout component in a virtual digital space. The obtained Fourier coefficient parameters are then used in the actual system to reversely reconstruct the radial runout to offset the adverse effects of the actual radial runout component.

Description

Technical Field

[0001] The present invention relates to the technical field of high-speed magnetic levitation motors, and in particular to a device and method for compensating a radial runout component of a magnetic levitation rotor position. Background Art

[0002] The rotor of a high-speed magnetic levitation motor rotates in a suspended state under the electromagnetic force of a magnetic bearing. This results in zero mechanical wear, no lubricant contamination, low energy consumption, low noise, and a long lifespan. Consequently, it is increasingly replacing traditional motors with mechanical bearings in applications requiring high speed, high efficiency, vacuum, and ultra-clean operation. One of the primary applications of high-speed magnetic levitation motors is in various types of high-speed turbomachinery, such as centrifugal air compressors and centrifugal refrigerant compressors.

[0003] High-speed centrifugal compressors using high-speed magnetic levitation motor drive technology are primarily characterized by their high efficiency. Compared to conventional compressors, they typically achieve 20-40% energy savings. A key factor in achieving these significant energy savings is the high-precision position control of the magnetically suspended rotor assembly (a rotating body comprising the turbine wheel, motor rotor, and magnetic bearing rotor magnetic circuit) while in suspension. A key requirement for achieving this high-precision position control is the position detection accuracy of the rotor assembly's radial position sensor. In practical magnetic levitation systems, the rotor position signal output by the magnetically suspended rotor assembly's radial displacement sensor often contains a radial runout component.

[0004] The main factors causing radial runout are: manufacturing errors causing the circumference of the electromagnetic target ring surface of the rotor assembly radial displacement sensor to be non-concentric with the geometric rotation axis of the rotor assembly, deformation of the target ring surface, and uneven electromagnetic properties of the material used to manufacture the target ring.

[0005] The radial runout component manifests as a harmonic component with a frequency of one or several times the motor speed. Ideally, the rotor assembly of a magnetic levitation motor is non-contactly suspended in the motion space within a magnetic bearing under closed-loop position control, rotating at high speed. The rotor assembly's rotation axis coincides with the geometric centerline of the magnetic bearing's electromagnet. However, the presence of radial runout disrupts this balance, causing radial oscillations in the rotor assembly's motion and causing the rotor's actual rotation axis to deviate from the designed rotation axis. When the load equipment of a high-speed magnetic levitation motor is a high-speed turbine machine, such as a centrifugal air compressor or a centrifugal refrigerant compressor, the design requires increasing the clearance between the turbine wheel and the volute to prevent friction between the turbine wheel and the volute due to axis of rotation offset. However, this increased clearance reduces compressor efficiency.

[0006] In addition, the radial oscillation of the rotor assembly will also cause the temperature of the magnetic bearing electromagnet to be too high, and the rotor assembly material will suffer fatigue damage or even fracture due to internal periodic stress, etc., which will lead to a series of problems.

[0007] Therefore, reducing the jitter in the radial position detection of the rotor assembly is an issue that must be considered during the manufacturing process of the magnetic levitation motor.

[0008] In addition to ensuring the accuracy of rotor assembly processing and assembly, conventional methods for reducing radial runout also require obtaining radial runout parameters through testing during the product manufacturing process, or adding radial runout parameter identification or adaptive functions to the control algorithm to compensate for the residual radial runout component.

[0009] However, the above-mentioned radial runout compensation process often needs to be carried out in a trial-and-error manner during the operation of the equipment, so there is a risk of equipment damage caused by improper parameter selection. Summary of the Invention

[0010] The purpose of the present invention is to provide a compensation device and method for the radial runout component of the magnetic levitation rotor position, which has the characteristics of first determining the Fourier coefficient parameters of each harmonic of the radial runout component in a virtual digital space through hardware-in-the-loop simulation and artificial intelligence computer optimization, and then using the obtained Fourier coefficient parameters in the actual system to reversely reconstruct the radial runout to offset the adverse effects of the actual radial runout component. Therefore, the method proposed in the present invention does not require trial and error during equipment operation, can greatly reduce the risk of equipment damage and improve the degree of automation of the production process, solving the problems raised in the above-mentioned background technology.

[0011] To achieve the above-mentioned object, the present invention provides the following technical solution: a device for compensating for radial runout component of magnetically suspended rotor position, comprising: a hardware-in-the-loop simulation system;

[0012] The hardware-in-the-loop simulation system includes: a high-speed magnetic levitation motor and its control system, wherein the high-speed magnetic levitation motor and the control system constitute the physical "hardware" system of the hardware-in-the-loop simulation system; a high-speed magnetic levitation motor dynamic mathematical model, a magnetic levitation control simulation system, and a computer optimization algorithm for radial runout compensation parameters constitute the simulation system of the hardware-in-the-loop simulation system existing in the computer digital space;

[0013] The high-speed magnetic levitation motor includes: a magnetic levitation rotor assembly, a radial electromagnetic bearing, a rotor assembly radial displacement sensor, a three-phase motor stator, an axial electromagnetic bearing and an axial displacement sensor, and a magnetic levitation rotor assembly angular displacement sensor.

[0014] Optionally, the rotor assembly radial displacement sensor is used to detect the position of the magnetically suspended rotor assembly in the air gap of the axial electromagnetic bearing.

[0015] Optionally, the control system is a high-speed magnetic levitation motor controller;

[0016] The high-speed magnetic levitation motor controller includes a rotor assembly radial / axial displacement sensor signal processing circuit, an electromagnetic bearing current PWM power amplifier, a magnetic levitation rotor assembly angular displacement signal processing circuit, a magnetic levitation position closed-loop controller, and a motor PWM drive and controller.

[0017] Optionally, the magnetic levitation rotor assembly includes:

[0018] Turbine impeller, radial magnetic bearing rotor magnetic circuit, motor rotor, rotor assembly radial displacement sensor electromagnetic target ring and axial magnetic bearing rotor;

[0019] The magnetic levitation rotor assembly is a rigid rotating body with the flexible mode ignored.

[0020] Optionally, the radial electromagnetic bearing, axial electromagnetic bearing, rotor assembly radial / axial displacement sensor, magnetic suspension position closed-loop controller and associated sensor signal processing circuit and power amplification circuit constitute a magnetic suspension control system of the magnetic suspension rotor assembly.

[0021] Optionally, the high-speed magnetic levitation motor dynamic mathematical model includes:

[0022] Position jitter component in the output signal of a radial position sensor with adjustable parameters;

[0023] Functional module for simulating the dynamic characteristics of the magnetic bearing current control loop and the dynamic characteristics of the rotor assembly radial position sensor.

[0024] Optionally, the hardware-in-the-loop simulation system further includes:

[0025] An inverse dynamics control algorithm is a radial position closed-loop control algorithm for the magnetically suspended rotor assembly.

[0026] A method for using a device for compensating a radial runout component of a magnetically suspended rotor position comprises the following steps:

[0027] S1: determining characteristic parameters of a position jitter component in an output signal of a radial position sensor of a magnetic levitation motor rotor by running the hardware-in-the-loop simulation system;

[0028] S2: then storing the obtained characteristic parameters of the position jump component in a digital storage unit of the control system of the high-speed magnetic levitation motor;

[0029] S3: During the operation of the control system of the high-speed magnetic levitation motor, the position runout component in the radial position sensor is reconstructed in real time using the stored characteristic parameters of the position runout component;

[0030] S4: and injecting it back into the radial displacement feedback signal of the control system to compensate for the jitter component in the feedback signal.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] 1. The present invention determines the Fourier coefficient parameters of each harmonic of the radial runout component in a virtual digital space through hardware-in-the-loop simulation and artificial intelligence computer optimization methods. The obtained Fourier coefficient parameters are then used to reversely reconstruct the radial runout in the actual system to offset the adverse effects of the actual radial runout components. Therefore, the method proposed by the present invention does not require trial and error during equipment operation, can significantly reduce the risk of equipment damage, and improve the degree of automation of the production process.

[0033] 2. The present invention utilizes the mechanical, electromagnetic and control system structures of the high-speed magnetic levitation motor, and determines the radial position jitter harmonic components present in the output signal of the high-speed magnetic levitation motor rotor radial position sensor through signal acquisition, system simulation and computer artificial intelligence parameter optimization methods, and eliminates the adverse effects of radial position jitter on the control of the high-speed magnetic levitation motor through reverse compensation; thereby achieving the purpose of reducing the loss of the magnetic levitation bearing electromagnet, improving the rotor position control accuracy, and improving the efficiency of the magnetic levitation motor drive equipment and extending the service life of the equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a front view of the structure of the present invention;

[0035] Figure 2 Schematic diagram of a high-speed magnetic levitation motor and its controller according to the present invention;

[0036] Figure 3 Schematic diagram of the magnetic levitation rotor assembly of the present invention;

[0037] Figure 4 Figure a) is a schematic diagram of the coordinate system involved in the radial motion control of the magnetically suspended rotor assembly in the electromagnetic bearing;

[0038] b) The figure is a schematic diagram of the six degrees of freedom, or motion state variables, of the rotor assembly in the suspended state relative to the inertial coordinate system;

[0039] c) The figure is a schematic diagram of the coordinate vector of the intersection of the geometric center axis of the rotor assembly and the sensor coordinate system plane, and a vector diagram of the bearing electromagnetic force in the bearing coordinate system plane;

[0040] Figure 5Figure (a) shows the state-space mathematical model of the magnetic levitation motor and the control system block diagram consisting of the closed-loop radial position control of the motor rotor assembly using an inverse dynamics algorithm. It does not include the effects of rotor radial position runout and dynamic balance imbalance.

[0041] Figure b) shows a control system block diagram consisting of a state-space mathematical model of the magnetic levitation motor and a closed-loop control of the motor rotor assembly's radial position using an inverse dynamics algorithm, including the effects of rotor radial position runout and dynamic imbalance.

[0042] Figure 6 This is a structural block diagram of the dynamic simulation of a magnetic levitation motor in the hardware-in-the-loop simulation system of the present invention;

[0043] Figure 7 This is a block diagram of an implementation of a hardware-in-the-loop simulation compensation device for radial runout of a magnetic levitation motor rotor according to the present invention.

[0044] Figure: 1. High-speed magnetic levitation motor; 2. High-speed magnetic levitation motor controller; 3. High-speed magnetic levitation motor dynamic mathematical model; 4. Magnetic levitation control simulation system; 5. Computer optimization algorithm for radial runout compensation parameters;

[0045] The high-speed magnetic levitation motor 1 includes: 1_1, magnetic levitation rotor assembly; 1_2, radial electromagnetic bearing; 1_3, rotor assembly radial displacement sensor; 1_4, three-phase motor stator; 1_5, axial electromagnetic bearing; 1_6, magnetic levitation rotor assembly angular displacement sensor, such as Figure 2 As shown;

[0046] The high-speed magnetic levitation motor controller 2 includes: 2-1, rotor assembly radial displacement sensor signal processing circuit; 2_2, electromagnetic bearing PWM power amplifier; 2_3, magnetic levitation rotor assembly angular displacement signal processing circuit; 2_4, magnetic levitation position closed-loop controller; 2_5, motor PWM drive and controller;

[0047] The magnetic suspension rotor assembly 1_1 includes: 1_1_1, turbine impeller; 1_1_2, radial magnetic suspension bearing rotor magnetic circuit; 1_1_3, motor rotor; 1_1_4, rotor assembly radial displacement sensor electromagnetic target ring; 1_1_5, axial magnetic suspension bearing rotor. DETAILED DESCRIPTION

[0048] 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.

[0049] See also Figures 1 to 7 ,This embodiment provides a technical solution: a compensation device for radial runout component of magnetic levitation rotor position, comprising: a hardware-in-the-loop simulation system;

[0050] The hardware-in-the-loop simulation system includes: a high-speed magnetic levitation motor 1 and its control system 2. The high-speed magnetic levitation motor 1 and the control system 2 constitute the physical "hardware" system of the hardware-in-the-loop simulation system;

[0051] The high-speed magnetic levitation motor dynamic mathematical model 3, the magnetic levitation control simulation system 4 and the radial runout compensation parameter computer optimization algorithm 5 constitute a simulation system existing in the computer digital space in the hardware-in-the-loop simulation system.

[0052] More specifically, in this embodiment: this device utilizes the mechanical, electromagnetic, and control system structures of the high-speed magnetic levitation motor 1 and its control system 2 to determine the radial position runout harmonic components present in the output signal of the high-speed magnetic levitation motor rotor radial position sensor through signal acquisition, system simulation, and computer artificial intelligence parameter optimization methods. Then, it eliminates the adverse effects of radial position runout on the control of the high-speed magnetic levitation motor through reverse compensation, thereby achieving the goals of reducing the electromagnetic losses of the magnetic levitation bearing, improving the rotor position control accuracy, improving the efficiency of the magnetic levitation motor drive equipment, and extending the service life of the equipment.

[0053] Specifically, in the hardware-in-the-loop simulation system consisting of a high-speed magnetic levitation motor 1 and its control system 2, the magnetic levitation rotor assembly 1_1 is controlled by the magnetic levitation motor controller 2 to rotate around its geometric center axis x r o r Rotate at a constant speed Ω, and the geometric center axis of the rotor is aligned with the geometric center line x of the electromagnet. i o i coincide;

[0054] Computer optimization algorithm for radial runout compensation parameters 5:

[0055] First, the frequency spectrum characteristics of the position jitter component in the output signal of the rotor assembly radial displacement sensor 1_3 are determined by fast Fourier transform; the main harmonic components of the position jitter component, i.e., the pilot spectrum, are determined based on the frequency spectrum characteristics; and computer parameter optimization of the characteristic parameters of the position jitter component is carried out in a targeted manner based on the harmonics in the pilot spectrum;

[0056] A method including but not limited to Fourier series expansion is used to extract characteristic parameter vectors of the disturbance component caused by the position jitter in the output signals of the rotor assembly radial displacement sensor 1_3 of the actual physical system and the computer simulation system, respectively, and the vector difference between the two constitutes the objective function of the computer optimization algorithm;

[0057] Using but not limited to the Hooke-Jeeves Algorithm as a computer optimization search engine;

[0058] The linear system superposition principle is used to optimize the characteristic parameter vectors of each harmonic in the guidance spectrum one by one, and then the compensation value of the total position jitter component is calculated according to the superposition principle.

[0059] Furthermore, in this embodiment: the high-speed magnetic levitation motor 1 includes:

[0060] Magnetic suspension rotor assembly 1_1, radial electromagnetic bearing 1_2, rotor assembly radial displacement sensor 1_3, three-phase motor stator 1_4, axial electromagnetic bearing 1_5 and axial displacement sensor, and magnetic suspension rotor assembly angular displacement sensor 1_6;

[0061] The rotor assembly radial displacement sensor 1_3 is used to detect the position of the magnetic suspension rotor assembly 1_1 in the air gap of the radial electromagnetic bearing 1_2.

[0062] Furthermore, in this embodiment: the control system is a high-speed magnetic levitation motor controller 2;

[0063] High-speed magnetic levitation motor controller 2 includes a rotor assembly radial displacement sensor signal processing circuit 2_1, an electromagnetic bearing current PWM power amplifier 2_2, a magnetic levitation rotor assembly angular displacement signal processing circuit 2_3, a magnetic levitation position closed-loop controller 2_4, and a motor PWM drive and controller 2_5. The rotor assembly axial position control system is also an integral part of high-speed magnetic levitation motor control, but due to its limited relevance to the present invention, it will not be described in detail.

[0064] Furthermore, in this embodiment: the magnetic suspension rotor assembly 1_1 includes:

[0065] Turbine impeller 1_1_1, radial magnetic bearing rotor magnetic circuit 1_1_2, motor rotor 1_1_3, rotor assembly radial displacement sensor electromagnetic target ring 1_1_4 and axial magnetic bearing rotor 1_1_5;

[0066] The magnetic levitation rotor assembly 1_1 is a rigid rotating body with the flexible mode ignored.

[0067] Furthermore, in this embodiment, the radial electromagnetic bearing 1_2, the rotor assembly radial displacement sensor 1_3, the magnetic suspension position closed-loop controller 2_4 and the associated sensor signal processing circuit and power amplifier circuit constitute a magnetic suspension radial position control system of the magnetic suspension rotor assembly 1_1.

[0068] More specifically, in this embodiment, the magnetic levitation rotor assembly 1_1 is supported contactlessly in the space within the magnetic levitation bearing, and the geometric center line of the magnetic levitation rotor assembly 1_1 is made to coincide with the geometric center line of the electromagnet. The magnetic levitation rotor assembly 1_1 rotates at a constant speed in a suspended state under the control of the motor PWM drive and the controller 2_5.

[0069] Furthermore, in this embodiment: the high-speed magnetic levitation motor dynamic mathematical model 3 includes:

[0070] Position jitter component in the output signal of a radial position sensor with adjustable parameters;

[0071] More specifically, it can be used to perform computer simulation on the response of the high-speed magnetic levitation motor 1 and its control system 2 to the jitter component;

[0072] Functional module to simulate the dynamic characteristics of the magnetic bearing current control loop and the dynamic characteristics of the rotor assembly radial position sensor;

[0073] More specifically, the functional module adopts but is not limited to an inertial filter structure, and its filter time constant is adjusted according to the bandwidth and transmission delay of the magnetic bearing current control loop and the rotor assembly radial position sensor in the actual system to make the dynamic characteristics of the simulation system closer to the actual system.

[0074] Furthermore, in this embodiment: the hardware-in-the-loop simulation system further includes:

[0075] Inverse dynamics control algorithm, the inverse dynamics control algorithm is a radial position closed-loop control algorithm of the magnetic suspension rotor assembly 1_1.

[0076] More specifically, in this embodiment: a control structure of "proportional differential control + rotor gyroscopic effect compensation + magnetic bearing negative stiffness characteristic compensation" is designed based on the radial motion mathematical model of the rotor assembly;

[0077] In order to detect the position jitter component in the output signal of the rotor radial position sensor, the control setting of the radial position closed loop is to make the geometric center axis of the rotor assembly coincide with the geometric center line of the magnetic bearing electromagnet, and to rotate at a low speed so that the centrifugal force generated by the dynamic imbalance of the rotor assembly can be ignored.

[0078] A method for compensating a radial runout component of a magnetically suspended rotor position comprises the following steps:

[0079] S1: Determine the characteristic parameters of the position runout component in the output signal of the radial position sensor of the magnetic levitation motor rotor by running the hardware-in-the-loop simulation system;

[0080] S2: then storing the obtained characteristic parameters of the radial position runout component in a digital storage unit of the control system of the high-speed magnetic levitation motor;

[0081] S3: During the operation of the control system of the high-speed magnetic levitation motor, the radial position runout component in the radial position sensor is reconstructed in real time using the stored characteristic parameters of the radial position runout component;

[0082] S4: and injecting it back into the radial displacement feedback signal of the control system to compensate for the jitter component in the feedback signal.

[0083] Specific description and operating principle:

[0084] 1) The radial motion control of the magnetically suspended rotor assembly 1_1 in the electromagnetic bearing involves four coordinate systems, such as Figure 4 As shown in Figure a), where o i _x i y i z i The coordinate system is an inertial coordinate system fixed to the electromagnet of the magnetic bearing, or a stationary coordinate system; r _x r y r z r is the motion coordinate system fixed to the rotor assembly, o r Ideally, it coincides with the center of gravity COG of the rotor assembly. r _x r y r z r Also known as COG coordinate system; a _y a z a and o b _y b z b The combined bearing coordinate system forms the plane where the electromagnetic force vector of the magnetic bearing is located; o as _y as z as and o bs _y bs z bs The sensor coordinate system of the plane where the rotor radial position sensor is located is jointly formed. The magnetic suspension rotor assembly 1_1 has 6 degrees of freedom of motion, or motion state variables, relative to the inertial coordinate system when in suspension state, such as Figure 4 As shown in Figure b), [x r y r z r ]and are the relative inertial coordinate system o of the rotor assembly i _x i y i z iThe motion state variables of translation and rotation, among the six motion state variables mentioned above, represent the motion state variables x of axial translation and rotation r (t) and Under normal circumstances, they can be controlled independently and the influence of the rotor radial motion can be ignored. Therefore, the implementation of axial rotation control and displacement control will not be analyzed in depth in the following text description of the present invention. The remaining four motion state variables y of the rotor assembly are r , z r , ψ, θ constitute the radial motion state variable q=[y r z r ψ θ] T .

[0085] 2) The radial electromagnetic force vector of the magnetic bearing is in the bearing coordinate system o a _y a z a and o b _y b z b It can be expressed as a vector

[0086] U F =[f ay f by f az f bz ] T

[0087] 3) Radial electromagnetic force U F The linear mathematical model is

[0088] U F =-K g B T q+K i I MB

[0089] in is the negative stiffness matrix of the magnetic bearing position, K ga , K gb are the negative stiffness coefficients of the front and rear radial magnetic bearings,

[0090] Its transpose B T is the transformation matrix from the COG coordinate system state variables to the bearing coordinate system displacement,

[0091] is the current stiffness matrix of the magnetic bearing, K ia , K ib are the current stiffness coefficients of the front and rear radial magnetic bearings respectively,

[0092] IMB =[i ay i by i az i bz ] T , is the radial magnetic bearing current vector, and is also the control output of the radial magnetic suspension control system.

[0093] 4) The radial electromagnetic force vector U of the magnetic bearing can be transformed into F Convert to rotor COG coordinate system

[0094] F coc =BU F

[0095] Where B is the bearing coordinate system o a _y a z a and o b _y b z b Force transformation matrix to COG coordinate system.

[0096] 5) Radial motion state variables of COG coordinate system

[0097] q=[y r z r ψ θ] T

[0098] The geometric center axis of the rotor assembly and the sensor coordinate plane o can be equivalently as _y as z as and o bs _y bs z bs The intersection o ras and o rbs The coordinate vector

[0099] q s =[y as z as y bs z bs ] T

[0100] To express, such as Figure 4 As shown in Figure c). Equivalent motion state variable q s The measurement is in the sensor coordinate system o as _y as z as and o bs _y bs z bs In the implementation, the measured state variable q sIt can be mapped to the rotor COG coordinate system through the following linear transformation

[0101] q=C -1 q s

[0102] in

[0103] Its inverse matrix C -1 is the sensor coordinate system o as _y as z as and o bs _y bs z bs The transformation matrix from the state variables in to the COG coordinate system.

[0104] 6) According to the coordinate system, state variables and electromagnetic force vector of the magnetic levitation bearing defined above, the linearized dynamic mathematical model of radial motion in the state space form of the magnetic levitation motor can be derived.

[0105]

[0106]

[0107] in

[0108] is the state variable of radial motion state space,

[0109] is the state matrix,

[0110] is the generalized mass matrix of the rotor assembly, m is the mass of the rotor assembly, I yy is the rotor assembly y r Shaft moment of inertia,

[0111] Gyroscopic effect matrix, where I xx is the rotor assembly x r The shaft moment of inertia, Ω is the rotor assembly speed,

[0112] K gCOG =BK g B T , is the coordinate system from the bearing o a _y a z a and o b _y b z b The negative stiffness matrix of the magnetic bearing position transformed to the COG coordinate system is:

[0113] is the input matrix,

[0114] is the output matrix.

[0115] 7) In the present invention, in order to realize the magnetic suspension rotor assembly 1_1 around its geometric center axis x under the control of the magnetic suspension control system, r o r Rotate, and the geometric center axis of the rotor is aligned with the geometric center line x of the magnetic bearing electromagnet i o i The magnetic suspension position closed-loop controller adopts the inverse dynamics algorithm, that is, the structure of COG coordinate system PD control + rotor gyro effect compensation + magnetic suspension bearing negative stiffness characteristic compensation is adopted. The magnetic suspension controller output corresponding to this algorithm is I MB It can be expressed as follows

[0116]

[0117] K p =(BK i ) -1 MP COG C -1 is the proportional control gain of PD control, where

[0118] P par , P con are translation and rotation proportional control gains respectively, K d =(BK i ) -1 MD COG C -1 is the differential control gain of PD control, where D par , D con are the translation and rotation differential control gains, C S (BK i ) -1 K gCOG C -1 is the compensation coefficient of the negative stiffness characteristic of the magnetic bearing position,

[0119] C G =(BK i ) -1 GC -1 is the rotor gyroscopic effect compensation coefficient.

[0120] 8) Figure 5 Figure a) shows a control system block diagram consisting of a mathematical model of the state space form of the magnetic levitation motor and a closed-loop control of the radial position of the motor rotor assembly using an inverse dynamics algorithm;

[0121] Ideally, the magnetically suspended rotor assembly 1_1 is non-contactly suspended in the motion space within the magnetic bearing under closed-loop position control, rotating at high speed. The rotation axis of the magnetically suspended rotor assembly 1_1 coincides with the geometric centerline of the magnetic bearing electromagnet. The rotor position signal output by the rotor assembly radial displacement sensor 1_3 has an average value of zero and does not contain a periodic component with a frequency equal to the rotational speed Ω or an integer multiple of Ω. However, in an actual system, due to manufacturing errors, such as the circumference of the surface of the electromagnetic target ring 1_1_4 of the rotor assembly radial displacement sensor being non-concentric with the geometric rotational axis of the rotor assembly, deformation of the target ring surface, and uneven electromagnetic properties of the material used to manufacture the target ring, the rotor position signal output by the rotor assembly radial displacement sensor 1_3 often contains a periodic component with a frequency equal to the rotational speed Ω or an integer multiple of Ω, namely, a radial runout component, represented by a vector R.

[0122] R=[r ay r az r by r bz ] T

[0123] in

[0124] o as _y as The runout component of the coordinate axis position sensor;

[0125] α r =Ωt is o r _y r With o i _y i The rotation angle between the two coordinate axes is measured by the magnetic suspension rotor assembly angular displacement sensor 1_6;

[0126] Because o as _z as The position of the coordinate axis position sensor lags behind the direction of rotation of the rotor assembly. as _y as Coordinate axis position sensor 90°,o as _z as The axis position sensor runout component can be expressed as

[0127] o bs _y bs The runout component of the coordinate axis position sensor;

[0128] With r az Similarly bz It can be expressed as

[0129]

[0130] Since the elements of vector R are the Fourier series of the harmonics of the beating components, the characteristic parameters of R can be represented by a 4-dimensional vector consisting of the Fourier coefficients of the harmonics.

[0131] ξ i =[A ai B ai A bi B bi I T

[0132] i=1, 2, ..., n is the harmonic order.

[0133] 9) In ξ i If R is known, it can be restored by the following formula

[0134]

[0135] in

[0136] R i =[r ayi r azi r byi r bzi ] T =Γ i ξ i

[0137]

[0138] That is R i Represents the i-th harmonic in R after harmonic decomposition of R.

[0139] 10) In addition to the aforementioned factors causing radial displacement jitter, the magnetic levitation rotor assembly 1_1 inevitably experiences a certain degree of dynamic imbalance during the manufacturing process. When the magnetic levitation rotor assembly 1_1 experiences dynamic imbalance, it will generate radial centrifugal force during rotation, forcing the geometric centerline of the rotor assembly to deviate from the geometric centerline of the electromagnet of the magnetic levitation bearing. This will also generate a periodic component with the same frequency as the rotational speed Ω in the output signal of the rotor assembly radial displacement sensor 1_3. The centrifugal force caused by dynamic imbalance can be equivalently expressed in the bearing coordinate system o a _y a z a and o b _y b z b Vector F in u To express.

[0140] 11) Considering the beating component R in the output of the rotor assembly radial displacement sensor 1_3 and the centrifugal force F caused by the dynamic balance imbalance u After the influence of the radial motion linear dynamic mathematical model of the magnetic levitation motor in the state space form becomes the following form

[0141]

[0142]

[0143] q se =q s +R

[0144] in is the centrifugal force F u The input matrix, q se is the output of the radial displacement sensor containing the runout component R, q se =[y ase z ase y bse z bse ] T .

[0145] 12) Figure 5 Figure b) shows the consideration of the radial displacement sensor output's jitter disturbance component R and the centrifugal force disturbance F generated by the rotor assembly's dynamic imbalance. u Block diagram of the radial position closed-loop control system of the motor rotor assembly in the future.

[0146] contrast Figure 5 From the diagrams a) and b) in Figure 1, we can see that the system output q that really needs to be controlled at point a is s At point b, the disturbance of the jump component R becomes unmeasurable, while the measurable feedback variable q se It contains not only the direct influence of R, but also the influence of R and F. u The closed loop formed by the magnetic suspension position controller and the magnetic suspension motor is s The response caused by se The vibration component R and centrifugal force F u The displacement generated by the disturbance is also a periodic component of the frequency of Ω or an integer multiple of Ω, which can be expressed as follows:

[0147] Q=[s ay s az s by s bz ] T

[0148] in

[0149] o as _yas Coordinate axis position sensor disturbance component;

[0150] o as _z as Coordinate axis position sensor disturbance component;

[0151] o bs _y bs Coordinate axis position sensor disturbance component;

[0152] o bs _z bs Coordinate axis position sensor disturbance component.

[0153] 13) and the vector ξ formed by Fourier coefficients i Similar to the jitter component R, the disturbance component Q can also be represented by a vector consisting of the Fourier coefficients of its harmonics.

[0154] γ i =[C ai D ai C bi D bi l T

[0155] i=1, 2, ..., n is the harmonic order.

[0156] γ i The functional relationship with Q is

[0157]

[0158] Q i =Γ i γ i

[0159] The jitter component R in the output of the radial displacement sensor 1_3 of the rotor assembly will cause radial oscillation in the movement of the magnetic levitation rotor assembly 1_1, causing the actual rotation axis of the rotor to deviate from the designed rotation axis. When the load equipment of the high-speed magnetic levitation motor is a high-speed turbine machine, such as a centrifugal air compressor or a centrifugal refrigerant compressor, in order to avoid the turbine wheel and the volute from rubbing against each other due to the deviation of the rotating shaft, the design needs to increase the gap between the turbine wheel and the volute. However, the increase in the gap will lead to a decrease in the efficiency of the compressor. In addition, the radial oscillation of the rotor assembly will also cause the temperature of the magnetic levitation bearing electromagnet to be too high, and the rotor assembly material will suffer fatigue damage or even fracture due to internal periodic stress, etc. Therefore, it is very necessary to compensate for the disturbance component in the control algorithm to eliminate the adverse effects caused by it in order to improve the control performance, efficiency, service life and reliability of the high-speed magnetic levitation motor. Due to R and qs For properties that cannot be directly measured, the relevant parameters of R are usually obtained through indirect testing during product manufacturing and use, or radial runout parameter identification or adaptive functions are added to the control algorithm to compensate for the radial runout component. The above-mentioned radial runout compensation process often needs to be carried out in a trial-and-error manner during the operation of the equipment, so there is a risk of equipment damage caused by improper parameter selection. The compensation method proposed in the present invention is to obtain the relevant parameters of the radial runout component R in a virtual digital space through hardware-in-the-loop simulation and artificial intelligence computer optimization, according to Figure 5 Figure b) and the relationship between R and q se From the analysis of the composition, we can know that q se The periodic disturbance component Q in the equation can be represented by the vector γ consisting of Fourier coefficients i To represent, R can be represented by the vector ξ consisting of Fourier coefficients i to represent, and γ i Yes i function.

[0160] 14) Therefore, the γ obtained in the actual system and the digital simulation is used in the hardware-in-the-loop simulation system of the present invention. i and (The number ^ indicates a variable or model in the simulation system, the same below) Composition The objective function is a variable, and the objective function is optimized by computer artificial intelligence to converge to the minimum extreme value, and the objective function variable corresponding to the extreme value is Reconstructed jump component This is the jump component R in the actual system i The optimal approximation of can be used to compensate for the jump component Ri in the actual system, that is, Replace q se To achieve the offset q se The radial runout component R in i Purpose. The process is carried out in a virtual digital space and does not require changing the operating state of the actual system. Therefore, the present invention can greatly reduce the risk of equipment damage during the radial runout compensation process and improve the degree of automation of the production process.

[0161] 15) It should be pointed out that if Figure 5 As shown in Figure b), in addition to the beating component R in the output of the radial displacement sensor, the radial position feedback signal q of the rotor assembly will be affected. se Causes disturbance, as before, the centrifugal force vector F generated by the dynamic imbalance u Also in q se Since F uThe frequency is synchronized with the motor speed Ω, so q se Corresponding to F u The component will have the same frequency as the fundamental component R1 in R, thus interfering with the acquisition of the relevant parameters of R1. As we all know, F u The amplitude is proportional to the square of Ω. According to the mechanism of the runout component in the output of the radial displacement sensor, the parameter change of R is minimally affected by the change of Ω. Therefore, in the process of obtaining the relevant parameters of the radial runout component R through hardware-in-the-loop simulation and artificial intelligence computer optimization, the motor speed Ω should be controlled in the lowest possible range to meet the stability requirements in order to minimize the dynamic imbalance centrifugal force F. u impact.

[0162] 16) The structural block diagram of the dynamic simulation of the magnetic levitation motor in the hardware-in-the-loop simulation system is as follows Figure 6 The magnetic levitation controller in the simulation system uses the same structure and control parameters as the controller in the actual system. and They are used to simulate the dynamic characteristics of the magnetic bearing current control loop and the rotor assembly radial position sensor respectively. and The present invention adopts but is not limited to the structure of the inertial filter, and its filtering time constant needs to be adjusted according to the bandwidth and transmission delay of the magnetic bearing current control loop and the rotor assembly radial position sensor in the actual system so that the dynamic characteristics of the simulation system are closer to the actual system.

[0163] 17) In the process of artificial intelligence computer optimization, the input of the magnetic levitation motor dynamic simulation system is the radial runout given in the optimization path. The Fourier coefficient vector After calculation formula Restore Generation Then it is superimposed on the simulation input of the magnetic suspension position closed-loop controller; the output of the magnetic suspension motor dynamic simulation system relative to the computer optimization system is Through The Fourier series expansion is performed and used as the input of the artificial intelligence computer optimization system to calculate the optimization objective function.

[0164] In the process of artificial intelligence computer optimization, the speed of magnetic levitation motor Ω will be controlled to make the dynamic balance unbalance centrifugal force F u The influence of can be ignored in the low speed range operation, so in Figure 6 F can be set in u =0.

[0165] 18) If Figure 7The figure shows an implementation block diagram of the hardware-in-the-loop simulation compensation device for radial runout of the magnetic levitation motor rotor position of the present invention. The specific implementation process is described as follows:

[0166] In the hardware-in-the-loop simulation system consisting of a high-speed magnetic levitation motor 1 and its control system 2, the rotor assembly is controlled by the magnetic levitation motor controller to rotate around its geometric center axis x. r o r Rotate at a constant speed Ω, and the geometric center axis of the rotor is aligned with the geometric center line x of the electromagnet. i o i In this state, if there is a radial runout component R with a frequency of Ω or several times of Ω in the output of the rotor position sensor, the feedback input variable q of the rotor assembly position closed-loop control se A periodic component of the same frequency will appear in .

[0167] 19) First, we need to determine q by fast Fourier transform se The guided spectrum in q se The necessity of determining the guide spectrum is to carry out subsequent computer parameter optimization in a targeted manner. The guide spectrum is often q se The frequency of the fundamental wave in the harmonic sequence is Ω. Without losing the general meaning, we can assume that these harmonics are

[0168] i=n1,n2,…,n N

[0169] Where N is the number of harmonics in the pilot spectrum, and n1, n2, ... are the orders of the harmonics.

[0170] 20) After the guide spectrum is determined, an optional method for optimizing the elements of the characteristic parameter vector of R is to optimize all the harmonics in the guide spectrum at the same time, that is, to construct the optimization variable parameter vector

[0171]

[0172] and the objective function

[0173]

[0174] The optimal value of ξ is searched by computer opt , so that J(ξ) converges to the minimum extreme value J(ξ opt ).

[0175] The amount of computation required for computer optimization increases exponentially with the dimension of the variable parameters. iis a 4-dimensional vector, and the dimension of vector ξ is 4N. Therefore, the above method of simultaneously optimizing all harmonics in the pilot spectrum will greatly increase the amount of calculation and the uncertainty of the optimization result.

[0176] 21) On the other hand, the mathematical model of the magnetic levitation motor and its control system near its operating point is a linearized model that satisfies the superposition principle of linear systems. Therefore, the method proposed in the present invention for optimizing the characteristic parameter vector of R is to utilize the characteristic that the system satisfies the superposition principle of linear systems, firstly, the characteristic parameter vectors of each harmonic in the pilot spectrum are optimized. Optimize the parameters one by one, and then calculate the radial position jump corresponding to the guidance spectrum according to the superposition principle

[0177]

[0178] Below is i=n k Computer optimization as an example process.

[0179] set up The corresponding computer optimization objective function is

[0180]

[0181] in That is, the radial displacement sensor output q of the actual system se Corresponding to n k The harmonic parameter vector of the subharmonic component is obtained by se Perform Fourier series expansion to obtain; The actual system is simulated in the digital space of the computer. The operating conditions of the magnetic levitation motor and its control system in the simulation system are consistent with the actual system, and the radial displacement sensor simulation output is Injection

[0182] 22) The process of computer optimization is The four-dimensional space Search for an "optimal" path to approximate the objective function The harmonic parameter vector corresponding to the minimum value of The computer optimization algorithm used to implement this process falls within the realm of artificial intelligence. Given the rapid development of artificial intelligence, there are numerous options for achieving the objectives of this invention. The term "optimal" refers to the selected optimization algorithm. In this invention, the Hooke-Jeeves algorithm is selected as one specific implementation of the computer optimization algorithm, but is not limited to it.

[0183] 23) The optimization path of the Hooke-Kieves algorithm consists of two movement processes: detection movement and pattern movement, so it is also called pattern search method. Through the iterative process of detection movement → pattern movement → detection movement..., the target function extreme value is approached one by one. The j+1th detection movement starts from the current parameter variable vector Start, respectively Each component of the harmonic component is perturbed one by one in both the positive and negative directions with the set harmonic component amplitude increment step h, and each parameter variable perturbation is obtained through simulation (l=1,2,…,8,represents The objective function corresponding to the 8 perturbations on the 4 components of Change to determine the optimal moving direction of the perturbation component, that is, increase, decrease or remain unchanged. After perturbation of the four components, the parameter variable vector can be obtained The incremental vector and as a result of detecting movement.

[0184] 24) Pattern movement is the result of detection movement If it is not zero, Along Continuously move in the direction of

[0185]

[0186] After each mode move, the new Perform simulation and calculate the corresponding objective function The condition for the pattern movement to be completed is

[0187] J(j, m)<J(j, m+1)

[0188] That is, the target function is obtained by moving along the probe The direction will no longer converge. The parameter variable vector obtained at this time will be used as the new current value

[0189]

[0190] Perform a new detection move.

[0191] 24) If the result of detecting movement Equal to zero, that is, the result of detecting movement is Keep still, it means It is close to the optimal value. At this time, the detection moving step length h can be reduced to As the current parameter variable vector, the detection movement phase continues to search for the best value to improve the accuracy of radial runout compensation. If h is less than the set minimum step length, the computer search process ends. That is, for n in the guided spectrum k Optimal target value of parameter variables by computer optimization of subharmonics

[0192] 26) After repeatedly applying the Hooke-Kieves algorithm to complete the computer optimization of all harmonic components in the guided spectrum, we can obtain

[0193]

[0194] The characteristic parameters of the position runout component obtained by running the hardware-in-the-loop simulation system k = 1, 2, ..., N will be stored in the digital storage unit of the high-speed magnetic levitation motor control system. During the actual operation of the high-speed magnetic levitation motor system, the position jump component characteristic parameters in the digital storage unit are used. k=1, 2, ..., N to reconstruct the jitter component in the radial position sensor in real time, that is,

[0195]

[0196] and will Back-injection radial displacement sensor feedback input q se This can be achieved for q se Compensation for the jitter component R in .

Claims

1. A method for compensating the radial runout component of a magnetically suspended rotor position, characterized by: include: Hardware-in-the-loop simulation system; The hardware-in-the-loop simulation system comprises: a high-speed magnetic levitation motor (1) and a control system (2) thereof, wherein the high-speed magnetic levitation motor (1) and the control system (2) constitute a physically existing "hardware" system of the hardware-in-the-loop simulation system; The high-speed magnetic levitation motor dynamic mathematical model (3), the magnetic levitation control simulation system (4) and the radial runout compensation parameter computer optimization algorithm (5) constitute the simulation system existing in the computer digital space in the hardware-in-the-loop simulation system; The high-speed magnetic levitation motor (1) comprises: a magnetic levitation rotor assembly (1_1), a radial electromagnetic bearing (1_2), a rotor assembly radial displacement sensor (1_3), a three-phase motor stator (1_4), an axial electromagnetic bearing (1_5) and an axial displacement sensor, and a magnetic levitation rotor assembly angular displacement sensor (1_6); The following steps are involved: S1: determining characteristic parameters of a position jitter component in an output signal of a radial position sensor of a magnetic levitation motor rotor by running the hardware-in-the-loop simulation system; S2: then storing the obtained characteristic parameters of the position jump component in a digital storage unit of the control system of the high-speed magnetic levitation motor; S3: During the operation of the control system of the high-speed magnetic levitation motor, the position runout component in the radial position sensor is reconstructed in real time using the stored characteristic parameters of the position runout component; Determining the frequency spectrum characteristics of the position jitter component in the output signal of the rotor assembly radial displacement sensor (1_3) by fast Fourier transform; determining the main harmonic components of the position jitter component, i.e., the guide spectrum, based on the frequency spectrum characteristics; and performing computer parameter optimization of characteristic parameters of the position jitter component in a targeted manner based on the harmonics in the guide spectrum; The characteristic parameter vectors of the disturbance component caused by position runout in the output signals of the rotor assembly radial displacement sensor (1_3) of the actual physical system and the computer simulation system are extracted by using the Fourier series expansion method, and the vector difference between the two constitutes the objective function of the computer optimization algorithm (5) for radial runout compensation parameters; The linear system superposition principle is used to optimize the characteristic parameter vectors of each harmonic in the guidance spectrum one by one, and the compensation value of the total position jitter component is calculated according to the superposition principle; S4: and injecting it back into the radial displacement feedback signal of the control system to compensate for the jitter component in the feedback signal.

2. The method for compensating the radial runout component of the magnetic levitation rotor position according to claim 1, characterized in that: The rotor assembly radial displacement sensor (1_3) is used to detect the position of the magnetic suspension rotor assembly (1_1) in the air gap of the axial electromagnetic bearing (1_5).

3. The method for compensating the radial runout component of the magnetic levitation rotor position according to claim 1, characterized in that: The control system is a high-speed magnetic levitation motor controller (2); The high-speed magnetic levitation motor controller (2) comprises a rotor assembly radial displacement sensor signal processing circuit (2_1), an electromagnetic bearing current PWM power amplifier (2_2), a magnetic levitation rotor assembly angular displacement signal processing circuit (2_3), a magnetic levitation position closed-loop controller (2_4), and a motor PWM drive and controller (2_5).

4. The method for compensating the radial runout component of the magnetically suspended rotor position according to claim 1, characterized in that: The magnetic suspension rotor assembly (1_1) comprises: A turbine impeller (1_1_1), a radial magnetic bearing rotor magnetic circuit (1_1_2), a motor rotor (1_1_3), a rotor assembly radial displacement sensor electromagnetic target ring (1_1_4), and an axial magnetic bearing rotor (1_1_5); The magnetic levitation rotor assembly (1_1) is a rigid rotating body with the flexible mode ignored.

5. The method for compensating the radial runout component of the magnetic levitation rotor position according to claim 3, characterized in that: The radial electromagnetic bearing (1_2), the rotor assembly radial displacement sensor (1_3), the magnetic suspension position closed-loop controller (2_4), and the associated sensor signal processing circuit and power amplification circuit constitute a magnetic suspension radial position control system of the magnetic suspension rotor assembly (1_1).

6. The method for compensating the radial runout component of the magnetically suspended rotor position according to claim 1, characterized in that: The high-speed magnetic levitation motor dynamic mathematical model (3) includes: Position jitter component in the output signal of a radial position sensor with adjustable parameters; Functional module for simulating the dynamic characteristics of the magnetic bearing current control loop and the dynamic characteristics of the rotor assembly radial position sensor.

7. The method for compensating the radial runout component of the magnetically suspended rotor position according to claim 1, characterized in that: The hardware-in-the-loop simulation system also includes: An inverse dynamics control algorithm is provided, wherein the inverse dynamics control algorithm is a radial position closed-loop control algorithm of the magnetic suspension rotor assembly (1_1).

Images