WO2024259754A1 - 六极主动磁轴承支承的高速电主轴控制系统的构造方法 - Google Patents

六极主动磁轴承支承的高速电主轴控制系统的构造方法 Download PDF

Info

Publication number
WO2024259754A1
WO2024259754A1 PCT/CN2023/106684 CN2023106684W WO2024259754A1 WO 2024259754 A1 WO2024259754 A1 WO 2024259754A1 CN 2023106684 W CN2023106684 W CN 2023106684W WO 2024259754 A1 WO2024259754 A1 WO 2024259754A1
Authority
WO
WIPO (PCT)
Prior art keywords
disturbance
radial
magnetic bearing
transfer learning
active magnetic
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2023/106684
Other languages
English (en)
French (fr)
Inventor
朱熀秋
马志豪
刁小燕
刘奕辰
华逸舟
潘伟
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jiangsu University
Original Assignee
Jiangsu University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Jiangsu University filed Critical Jiangsu University
Publication of WO2024259754A1 publication Critical patent/WO2024259754A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F16—ENGINEERING ELEMENTS AND UNITS; GENERAL MEASURES FOR PRODUCING AND MAINTAINING EFFECTIVE FUNCTIONING OF MACHINES OR INSTALLATIONS; THERMAL INSULATION IN GENERAL
    • F16C—SHAFTS; FLEXIBLE SHAFTS; ELEMENTS OR CRANKSHAFT MECHANISMS; ROTARY BODIES OTHER THAN GEARING ELEMENTS; BEARINGS
    • F16C32/00—Bearings not otherwise provided for
    • F16C32/04—Bearings not otherwise provided for using magnetic or electric supporting means
    • F16C32/0406—Magnetic bearings
    • F16C32/044—Active magnetic bearings
    • F16C32/0474—Active magnetic bearings for rotary movement
    • F16C32/048—Active magnetic bearings for rotary movement with active support of two degrees of freedom, e.g. radial magnetic bearings
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F16—ENGINEERING ELEMENTS AND UNITS; GENERAL MEASURES FOR PRODUCING AND MAINTAINING EFFECTIVE FUNCTIONING OF MACHINES OR INSTALLATIONS; THERMAL INSULATION IN GENERAL
    • F16C—SHAFTS; FLEXIBLE SHAFTS; ELEMENTS OR CRANKSHAFT MECHANISMS; ROTARY BODIES OTHER THAN GEARING ELEMENTS; BEARINGS
    • F16C32/00—Bearings not otherwise provided for
    • F16C32/04—Bearings not otherwise provided for using magnetic or electric supporting means
    • F16C32/0406—Magnetic bearings
    • F16C32/044—Active magnetic bearings
    • F16C32/0474—Active magnetic bearings for rotary movement
    • F16C32/0485—Active magnetic bearings for rotary movement with active support of three degrees of freedom
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F16—ENGINEERING ELEMENTS AND UNITS; GENERAL MEASURES FOR PRODUCING AND MAINTAINING EFFECTIVE FUNCTIONING OF MACHINES OR INSTALLATIONS; THERMAL INSULATION IN GENERAL
    • F16C—SHAFTS; FLEXIBLE SHAFTS; ELEMENTS OR CRANKSHAFT MECHANISMS; ROTARY BODIES OTHER THAN GEARING ELEMENTS; BEARINGS
    • F16C32/00—Bearings not otherwise provided for
    • F16C32/04—Bearings not otherwise provided for using magnetic or electric supporting means
    • F16C32/0406—Magnetic bearings
    • F16C32/044—Active magnetic bearings
    • F16C32/0474—Active magnetic bearings for rotary movement
    • F16C32/0493—Active magnetic bearings for rotary movement integrated in an electrodynamic machine, e.g. self-bearing motor

Definitions

  • the present invention belongs to the technical field of electrical and mechanical transmission equipment, and specifically relates to an anti-disturbance decoupling control system for a high-speed electric spindle supported by a six-pole active magnetic bearing, and is suitable for the decoupling control of a high-speed electric spindle supported by a six-pole active magnetic bearing with multivariable, nonlinear and strong coupling.
  • the high-speed electric spindle supported by six-pole active magnetic bearings is an electric spindle supported by two six-pole active radial magnetic bearings and one single-degree-of-freedom axial magnetic bearing.
  • the magnetic bearing uses the magnetic field force to suspend the rotor in space, realizing a new type of high-performance bearing without any mechanical contact between the rotor and the stator, solving the problems of thermal deformation, short life, and insufficient seismic resistance of traditional bearing spindles.
  • the radial magnetic bearing adopts a six-pole symmetrical stator structure.
  • the stator of the six-pole magnetic bearing has a symmetrical structure, small magnetic circuit coupling between the two radial degrees of freedom, large bearing capacity, and excellent performance.
  • the radial control coil is wound on the magnetic poles of the radial magnetic bearing and driven by a three-phase AC inverter. After power is turned on, a radial control magnetic flux is formed to generate a suspension force to achieve radial stable suspension. Since the radial magnetic bearings are all driven by inverters, it is inevitable that the radial magnetic flux will be coupled, affecting the precision control of the electric spindle. Therefore, an efficient decoupling controller is required to achieve high-speed, high-precision and stable operation of the magnetic bearing.
  • the methods to solve the decoupling control of magnetic suspension electric spindles mainly include the comprehensive application of approximate linearized anti-disturbance decoupling control method, matrix converter control method, inverse system decoupling control method and other intelligent algorithms.
  • Approximate linearized anti-disturbance decoupling control is a new decoupling control idea. It regards the mutual coupling between the radial magnetic circuits of the radial magnetic bearings in the five-degree-of-freedom magnetic suspension electric spindle as the internal disturbance of the system, and uses the extended state observer to estimate and compensate for the internal disturbance to achieve precise decoupling control, so that each degree of freedom of the magnetic bearing can be linearized.
  • the performance of the traditional approximate linearized anti-disturbance controller depends on the selection of internal parameters.
  • the anti-disturbance controller cannot accurately compensate for the disturbance.
  • the control system provided in the document entitled "Least Squares Support Vector Machine Optimization Control System for Five-degree-of-freedom Magnetic Suspension Electric Spindle" with Chinese patent publication number CN112532134A selects the tracking signal and differential signal of the displacement as training samples to predict and compensate for the disturbance of the electric spindle, but the selection of parameters in the least squares support vector machine will result in a slow solution speed, and each operation requires a large amount of training data, which affects the control accuracy of the system.
  • the purpose of the present invention is to provide a method for constructing a high-speed electric spindle control system supported by a six-pole active magnetic bearing that can solve the problems existing in the existing high-speed electric spindle control supported by a five-degree-of-freedom six-pole active magnetic bearing, such as reliance on internal parameter selection, imprecise compensation of disturbances, need for a large amount of training data for operation, slow solution speed, and poor control accuracy.
  • the technical solution adopted by the construction method of the high-speed electric spindle control system supported by the six-pole active magnetic bearing of the present invention is:
  • the high-speed electric spindle has first and second six-pole radial active magnetic bearings and an axial magnetic bearing, and is characterized by comprising the following steps:
  • Step 1) constructing a three-degree-of-freedom composite controlled object including a first six-pole radial active magnetic bearing and an axial magnetic bearing, and a two-degree-of-freedom composite controlled object including a second six-pole radial active magnetic bearing, wherein the inputs of the three-degree-of-freedom composite controlled object and the two-degree-of-freedom composite controlled object are four corresponding radial control current expected values and one axial control current expected value;
  • Step 2) the high-speed electric spindle is operated under rated load, and multiple groups of displacement signals and corresponding disturbance signals outputted by the high-speed electric spindle are collected to form a source domain data set, and the load size is changed within the rated load range and multiple groups of displacement signals of the high-speed electric spindle are collected again to form a target domain data set; the edge distribution distance between the source domain data set and the target domain data set is calculated, a transfer learning model is constructed based on the edge distribution distance, and the transfer learning model is trained to obtain a predicted disturbance function, and the predicted disturbance function is built into the transfer learning module (504);
  • Step 3) using the actual radial or axial displacement as the input of the transfer learning module (504), obtaining the predicted disturbance according to the predicted disturbance function, and compensating the radial or axial control current expected value with the second compensation factor (506), and then adding the predicted disturbance to obtain the second control amount;
  • the second control variable and the actual radial or axial displacement are used as inputs of an extended state observer (502), and the extended state observer (502) outputs a tracking signal, a first-order differential signal of the tracking signal, and an observed disturbance;
  • the first control amount is obtained after compensation by a first compensation factor (505);
  • a tracking signal and a differential signal of a given radial or axial displacement are outputted through a tracking differentiator (501), the tracking signal outputted by the tracking differentiator (501) is subtracted from the tracking signal outputted by an extended state observer (502), and the differential signal outputted by the tracking differentiator (501) is subtracted from the first-order differential signal outputted by the extended state observer (502), and the two errors obtained are used together as inputs of a nonlinear state error feedback control law (503), and the nonlinear state error feedback control law (503) outputs a feedback control amount;
  • the feedback control amount is subtracted from the first control amount to obtain a compensated control amount as the radial or axial control current expected value;
  • the tracking differentiator (501), the nonlinear state error feedback control law (503), the first compensation factor (505), the second compensation factor (506), the transfer learning module (504) and the extended state observer (502) together constitute an optimized active disturbance rejection controller;
  • Step 4) three of the optimized anti-disturbance control devices are connected in series to the front end of the three-degree-of-freedom composite controlled object, and two of the optimized anti-disturbance control devices are connected in series to the front end of the two-degree-of-freedom composite controlled object, together forming a high-speed electric spindle control system.
  • the present invention adopts the self-disturbance rejection control based on transfer learning optimization, optimizes the performance of the extended state observer, predicts the disturbance of the electric spindle, and uses the optimized extended state observer to estimate and compensate the total disturbance of the system, thereby realizing the precise decoupling control of the radial degrees of freedom of the two six-pole radial active magnetic bearings in the high-speed electric spindle.
  • This method can decouple the multi-input multi-output strongly coupled and nonlinear system under non-ideal conditions into a multi-input multi-output uncoupled linear system, thereby improving the system control performance.
  • the present invention adopts a transfer learning algorithm to obtain predicted disturbances, and adopts the displacement signal and disturbance signal collected under the control of the disturbance rejection controller as the source domain data set, which reduces the data demand. After the load is replaced, the displacement signal and the corresponding disturbance signal are collected as the target domain data set, and a transfer learning model is constructed according to the edge distance distribution, and the source domain data is used to train and test the transfer learning model. Since the transfer learning has obtained the characteristics of the source domain data set, it can avoid selecting too many internal parameters and reduce the demand for a large amount of training data. Therefore, the calculation speed of the control system is improved, and the system has better rapidity.
  • transfer learning uses training data to obtain data features, which can avoid overfitting and improve the convergence speed and control accuracy of the control system.
  • the radial magnetic bearings in the high-speed electric spindle supported by the six-pole active magnetic bearings are equipped with two radial magnetic bearings and one axial magnetic bearing to achieve stable suspension of the five degrees of freedom of the electric spindle shaft.
  • the radial and axial magnetic circuits are independent of each other, and the coupling is reduced.
  • the radial magnetic bearings are driven and controlled by a three-phase inverter.
  • Five anti-disturbance controllers are used to decouple the high-speed electric spindle, without the need for an accurate mathematical model.
  • the controllers have the characteristics of small overshoot, fast response, and strong anti-interference ability, and can greatly improve the control accuracy of the electric spindle control system.
  • FIG1 is a schematic structural diagram of a high-speed electric spindle supported by a six-pole active magnetic bearing
  • FIG2 is a schematic diagram of the radial structure and magnetic circuit of the first six-pole radial active magnetic bearing in FIG1 ;
  • FIG3 is a schematic diagram of the axial structure and magnetic circuit of the first six-pole radial active magnetic bearing in FIG1 ;
  • FIG4 is an equivalent structural block diagram of a three-degree-of-freedom composite controlled object constructed based on the first six-pole active radial magnetic bearing and the axial magnetic bearing in FIG1 ;
  • FIG5 is an equivalent structural block diagram of a two-degree-of-freedom composite controlled object constructed based on the second six-pole active radial magnetic bearing in FIG1 ;
  • FIG6 is a block diagram of a high-speed electric spindle control system supported by a six-pole active magnetic bearing
  • FIG7 is a control structure block diagram of the first optimized ADRC controller in FIG6 ;
  • FIG8 is a flow chart of a construction method of the transfer learning module in FIG7 ;
  • FIG. 9 is a control structure block diagram of the conventional active disturbance rejection controller shown in FIG. 7 .
  • the fourth optimized auto-disturbance rejection controller 24. the fifth optimized auto-disturbance rejection controller; 25. the first displacement sensor; 26. the second displacement sensor; 27. the third displacement sensor; 28. the fourth displacement sensor; 29. the fifth displacement sensor; 31. the first Clark inverse transformation; 32 the second Clark inverse transformation; 33. the first current tracking inverter; 34. the second current tracking inverter; 35. the axial power amplifier;
  • first radial rotor 101. first radial rotor; 102. first radial stator; 103. first radial control coil; 104. first radial stator magnetic pole; 105. bias magnetic flux; 106. control magnetic flux; 107. radial air gap;
  • Axial stator 202. Axial rotor; 203. Axial control coil; 204. Axial stator magnetic pole; 210. Three-degree-of-freedom composite controlled object; 211. Two-degree-of-freedom composite controlled object;
  • Second radial rotor 302.
  • Second radial rotor 302.
  • Second radial rotor 303.
  • Second radial control coil
  • First tracking differentiator 502. First extended state observer; 503. First nonlinear feedback control law; 504. Transfer learning module; 505. First compensation factor; 506. Second compensation factor.
  • the electric spindle supported by the six-pole magnetic bearing shown in FIG1 includes a rotating shaft 5, a high-speed spindle motor 4, a first six-pole radial active magnetic bearing 1, an axial magnetic bearing 2, a second six-pole radial active magnetic bearing 3, a steel cylinder outer sleeve 14, a steel cylinder inner sleeve 15, radial displacement sensors 6, 7, 8, 9, an axial displacement sensor 10, an auxiliary bearing 11, and end covers 12, 13.
  • the high-speed spindle motor 4 includes a high-speed spindle motor stator 16 and a high-speed spindle motor rotor 17.
  • the high-speed spindle motor rotor 17 is coaxially fixedly sleeved on the outside of the rotating shaft 5 to drive the rotating shaft 5 to rotate.
  • On the left side of the high-speed spindle motor 4 is the second six-pole radial active magnetic bearing 14, a steel cylinder inner sleeve 15, radial displacement sensors 6, 7, 8, 9, an axial displacement sensor 10, an auxiliary bearing 11, and end covers 12, 13.
  • the first six-pole radial active magnetic bearing 3 is on the right, and the axial magnetic bearing 2 and the first six-pole radial active magnetic bearing 1 are on the right.
  • the first six-pole radial active magnetic bearing 1 and the second six-pole radial active magnetic bearing 3 have the same structure.
  • a steel cylinder is wrapped around the outside of the high-speed spindle motor 4, the first six-pole radial active magnetic bearing 1, the second six-pole radial active magnetic bearing 3 and the axial magnetic bearing 2.
  • the steel cylinder is composed of a steel cylinder outer sleeve 14 and a steel cylinder inner sleeve 15.
  • a spiral channel for water cooling is provided between the steel cylinder outer sleeve 14 and the steel cylinder inner sleeve 15.
  • An end cap 12, 13 is fixedly connected to the left and right end faces of the steel cylinder outer sleeve 14 and the steel cylinder inner sleeve 15, respectively.
  • the two end caps 12, 13 are each supported on the outside of the rotating shaft 5 by an auxiliary bearing 11.
  • the auxiliary bearing 11 is used to support the rotating shaft 5 when the magnetic bearing is shut down or in a fault state.
  • the rotating shaft 5 coaxially passes through the first six-pole radial active magnetic bearing 1, the axial magnetic bearing 2, the high-speed spindle motor 4 and the second six-pole radial active magnetic bearing 3.
  • the second six-pole radial active magnetic bearing 3 is located between the high-speed spindle motor 4 and the left end cover 13, and an axial distance is left between the high-speed spindle motor 4 and the left end cover 13.
  • a third radial displacement sensor 8 and a fourth radial displacement sensor 9 are installed between the left end cover 13 and the second six-pole radial active magnetic bearing 3.
  • the third radial displacement sensor 8 and the fourth radial displacement sensor 9 are symmetrically installed face to face in the radial direction to detect the displacement of the two degrees of freedom in the radial direction of the left side of the rotating shaft 5.
  • the axial magnetic bearing 2 is composed of an axial rotor 201, an axial stator 202 and an axial control coil 203.
  • the axial magnetic bearing 2 is located between the high-speed spindle motor 4 and the first six-pole radial active magnetic bearing 1.
  • two annular limiting sleeves are respectively placed between the first six-pole radial active magnetic bearing 1 and the axial magnetic bearing 2, and between the axial magnetic bearing 2 and the high-speed spindle motor 4, to fix the axial positions of the three.
  • the first six-pole radial active magnetic bearing 1 is located between the axial magnetic bearing 2 and the right end cover 12, and an axial distance is left between the axial magnetic bearing 2 and the right end cover 12.
  • a first radial displacement sensor 6 and a second radial displacement sensor 7 are provided between the right end cover 12 and the first six-pole radial active magnetic bearing 1.
  • the first radial displacement sensor 6 and the second radial displacement sensor 7 are symmetrically installed face to face in the radial direction to detect the displacement of the two degrees of freedom of the right radial direction of the rotating shaft 5.
  • An axial displacement sensor 10 is installed on the right end cover 12 to detect the displacement of the axial degree of freedom of the rotating shaft 5.
  • the probes of the first, second, third and fourth radial displacement sensors 6, 7, 8, 9 and the axial displacement sensor 10 are all eddy current sensors.
  • the first six-pole radial active magnetic bearing 1 shown in FIG. 2 and FIG. 3 it is composed of a first radial rotor 101, a first radial stator 102 and a first radial control coil 103.
  • the first radial rotor 101 is cylindrical and coaxially fixedly sleeved outside the rotating shaft 5 to form an integral body with the rotating shaft 5.
  • the outer wall of the first radial stator 102 is fixedly connected to the inner wall of the steel cylinder inner sleeve 15.
  • the first radial stator 102 is coaxially sleeved outside the first radial rotor 101, and a radial air gap 107 exists between the two.
  • the first radial stator 102 is composed of an annular stator yoke and six convex first radial stator poles 104 evenly arranged along the circumference of the inner wall of the stator yoke.
  • the side surface of the first radial stator pole 104 is a cylindrical surface.
  • Each first radial stator pole 104 The first radial control coils 103 on the two opposite magnetic poles are wound with the same direction, and the first radial control coils 103 on the two opposite magnetic poles are connected in series into one phase to form a three-phase coil.
  • the three-phase coils are connected in a star shape, and a three-phase inverter is used to control the first radial control coils 103 to control the size and direction of the suspension force.
  • the first radial control coil 103 provides a bias flux 105 and a control flux 106 for the first six-pole radial active magnetic bearing 1.
  • the circuit of the bias flux 105 is: from a first radial stator magnetic pole 104 and a radial air gap 107 to the first radial rotor 101, then through the opposite radial air gap 107 to the opposite first radial stator magnetic pole 104, and finally through the stator yoke to return to the first radial stator magnetic pole 104 at the beginning.
  • the bias flux 105 circuit mainly functions to balance the gravity of the rotor with the suspension force generated by the bias magnetomotive force.
  • the control flux 106 circuit is: from a first radial stator magnetic pole 104 and a radial air gap 107 to the first radial rotor 101, then through the opposite radial air gap 107 to the opposite first radial stator magnetic pole 104.
  • the bias flux 105 and the control flux 106 interact with each other to generate a radial two-degree-of-freedom suspension force on the first radial rotor 101 of the first six-pole radial active magnetic bearing 1.
  • the second six-pole radial active magnetic bearing 3 is composed of a second radial stator 302, a second radial rotor 301 and a second radial control coil 303. Its structure and assembly method are exactly the same as those of the first radial stator 102, the first radial rotor 101 and the first radial control coil 103.
  • the size of the first radial air gap 107 between the first radial stator 102 and the first radial rotor 101 is the same as the size of the second radial air gap between the second radial stator 302 and the second radial rotor 301, both of which are 0.5 mm.
  • bias flux and the control flux of the second six-pole radial active magnetic bearing interact with each other to generate a radial two-degree-of-freedom suspension force on the second radial rotor 301 of the second six-pole radial active magnetic bearing 3.
  • the axial magnetic bearing 2 is composed of an axial stator 201, an axial rotor 202 and an axial control coil 203.
  • the axial rotor 202 is cylindrical and coaxially fixedly sleeved on the rotating shaft 5 and becomes one with the rotating shaft 5.
  • the axial control coil 203 is respectively wound on the left and right magnetic poles of the axial stator 202, and the axial control magnetic flux and the axial bias magnetic flux are generated by the bipolar DC power amplifier.
  • the axial control magnetic flux and the axial bias magnetic flux interact with each other to generate an axial suspension force on the axial rotor 202 of the axial magnetic bearing 2.
  • the first radial stator 102, the first radial rotor 101, the second radial stator 302, the second radial rotor 301, the axial stator 202 and the axial rotor 201 are all made of laminated silicon steel sheets.
  • the first radial control coil 103, the second radial control coil 303 and the axial control coil 203 are all made of 0.67 mm diameter copper wire with insulating paint.
  • the auxiliary bearing 11 is a self-aligning ball bearing.
  • a three-degree-of-freedom composite controlled object 210 including the first six-pole radial active magnetic bearing 1 and the axial magnetic axis 2 is constructed.
  • the three-degree-of-freedom composite controlled object 210 is composed of a first Clark inverse transform 31, a first current tracking inverter 33, a first six-pole radial active magnetic bearing 1, and an axial power amplifier 35 and an axial magnetic bearing 2 connected in series in sequence.
  • the three-degree-of-freedom composite controlled object 210 The inputs are the first radial control current expected value i ax * , the second radial control current expected value i ay * and the axial control current expected value i az * , and the outputs of the three-degree-of-freedom composite controlled object 210 are the actual radial displacements x a , ya and the actual axial displacement z a .
  • the first radial control current expected value i ax * and the second radial control current expected value i ay * are transformed by the first Clark inverse transformation 31 to obtain currents i au * , i av * and i aw * , and the currents i au * , i av * and i aw * are output as radial control currents i au , i av and i aw via the first current tracking inverter 33.
  • the first radial control coil 103 in the first six-pole radial active magnetic bearing 1 is controlled by the radial control currents i au , i av and i aw , and the first six-pole radial active magnetic bearing 1 generates actual radial displacements x a and ya .
  • the first and second radial displacement sensors 6 and 7 detect the actual radial displacements x a and ya respectively.
  • the expected value of the axial control current i az * is output as the axial control current i az after passing through the axial power amplifier 35.
  • the axial control current i az controls the axial control coil 203 in the axial magnetic bearing 2.
  • the axial magnetic bearing 2 generates an actual axial displacement za .
  • the axial displacement sensor 10 detects the actual axial displacement za , thereby forming a three-degree-of-freedom composite controlled object 210 including the first six-pole radial active magnetic bearing 1 and the axial
  • a two-degree-of-freedom composite controlled object 211 including the second six-pole radial active magnetic bearing 3 is constructed.
  • the two-degree-of-freedom composite controlled object 211 is composed of the second Clark inverse transform 32, the second current tracking inverter 34, and the second six-pole radial active magnetic bearing 3 connected in series in sequence.
  • the input of the two-degree-of-freedom composite controlled object 211 is the third and fourth radial control current expected values i bx * , i by *
  • the output of the two-degree-of-freedom composite controlled object 211 is the actual radial displacements x b , y b .
  • the third and fourth radial control current expected values i bx * and i by * are input into the second Clark inverse transform 32, and the second Clark inverse transform 32 outputs currents i bu * , i bv * , and i bw * .
  • the currents i bu * , i bv * , and i bw * are output to the second six-pole radial active magnetic bearing 3 via the second current tracking inverter 34 as radial control currents i bu , i bv , and i bw, and the second radial control coil 302 is controlled therein.
  • the second six-pole radial active magnetic bearing 3 outputs actual radial displacements x b and y b , and a two-degree-of-freedom composite controlled object 211 including the second six-pole radial active magnetic bearing 3 is constructed.
  • the third and fourth radial displacement sensors 8 and 9 detect the actual radial displacements x b and y b .
  • the five optimized anti-disturbance controllers are used to form a control system, and the five optimized anti-disturbance controllers control the five degrees of freedom respectively.
  • the five optimized anti-disturbance controllers are the first, second, third, fourth, and fifth optimized anti-disturbance controllers 20, 21, 22, 23, and 24.
  • the five optimized anti-disturbance controllers are only different in input and output signals, and the internal structures are exactly the same.
  • the first and second radial displacement sensors 6 and 7 collect the actual radial displacements xa and ya of the three-degree-of-freedom composite controlled object 210, and the actual radial displacements xa and ya are input into the corresponding first optimized ADRC controller 20 and second optimized ADRC controller 21 one by one.
  • the axial displacement sensor 10 collects the actual axial displacement za of the three-degree-of-freedom composite controlled object 210, and the actual axial displacement za is input into the third optimized ADRC controller 22.
  • the corresponding given radial displacement xa * is input into the first optimized ADRC controller 20, and the first optimized ADRC controller 20 processes the input actual radial displacement xa and given radial displacement xa * , and outputs the first radial control current expected value iax * to the three-degree-of-freedom composite controlled object 210.
  • the corresponding given radial displacement ya * is input to the second optimized auto-disturbance rejection controller 21, and the second optimized auto-disturbance rejection controller 21 outputs the second radial control current expected value iay * to the three-degree-of-freedom composite controlled object 210;
  • the given axial displacement za * is input to the third optimized auto-disturbance rejection controller 22, and the third optimized auto-disturbance rejection controller 22 outputs the axial control current expected value iaz * to the three-degree-of-freedom composite controlled object 210.
  • the first radial control current expected value iax * , the second radial control current expected value iay * , and the axial control current expected value iaz * are collectively used as inputs of the three-degree-of-freedom composite controlled object 210.
  • the third and fourth radial displacement sensors 8 and 9 collect the actual radial displacements x b and y b of the two-degree-of-freedom composite controlled object 211.
  • the actual radial displacements x b and y b are input into the corresponding fourth optimized ADRC controller 23 and the fifth optimized ADRC controller 24 one by one.
  • the corresponding given radial displacement x b * is input into the fourth optimized ADRC controller 23.
  • the fourth optimized ADRC controller 23 processes the input actual radial displacement x b and given radial displacement x b * , and outputs the third radial control current expected value i bx * to the two-degree-of-freedom composite controlled object 211.
  • the corresponding given radial displacement y b * is input into the fifth optimized ADRC controller 24.
  • the fifth optimized ADRC controller 24 outputs the fourth radial control current expected value i by * to the two-degree-of-freedom composite controlled object 211.
  • the third radial control current expected value i bx * and the fourth radial control current expected value i by * are used together as the input of the two-degree-of-freedom composite controlled object 211.
  • the construction process of the optimized ADRC controller is described in detail below by taking the first optimized ADRC controller 20 as an example:
  • the first optimized ADRC 20 is composed of a tracking differentiator 501 , a nonlinear state error feedback control law 503 , a first compensation factor 505 , a second compensation factor 506 , a transfer learning module 504 and an extended state observer 502 .
  • the given radial displacement x a * is outputted via the tracking differentiator 501 as a tracking signal v a1x of the given radial displacement x a * and a differential signal v a2x of the given radial displacement x a * .
  • the mathematical model of the tracking differentiator 501 is:
  • T is the fast tracking factor, with a value range of [0,1].
  • the value of T determines the system tracking performance; ⁇ is the integration step; h is the system sampling period; For a given displacement The value at time k; ⁇ a1x (k) and ⁇ a1x (k+1) are the values of the given radial displacement tracking signal va1x at time k and k+1 respectively; ⁇ a2x (k) and ⁇ a2x (k+1) are the values of the given radial displacement differential signal va2x at time k and k+1 respectively.
  • the actual radial displacement x a is outputted through the extended state observer 502 as a tracking signal za1x , a first-order differential signal za2x of the tracking signal za1x , and an observed disturbance za3x of the system.
  • the two system state errors e a1x and e a2x serve as inputs of the nonlinear state error feedback control law 503, and the nonlinear state error feedback control law 503 outputs the feedback control quantity u a0x .
  • k 1 , k 2 , ⁇ 4 , ⁇ 5 and ⁇ 2 are adjustable parameters of the nonlinear feedback control law.
  • ⁇ 4 is 0.5; ⁇ 5 is 0.25; the range of k 1 and k 2 is [0,100], the range of ⁇ 2 is [0,1], and the nonlinear function
  • e is a natural constant, ⁇ is a constant between 0 and 1, and ⁇ is a custom constant that affects the filtering effect.
  • the feedback control amount u a0x is compensated by the first compensation factor 505 to obtain the compensated control amount u ax :
  • 1/d is the first compensation factor
  • f ax is the predicted disturbance output by the transfer learning module 504.
  • the value range of d is [10, 200].
  • the compensated control amount u ax is also used as the input of the second compensation factor 506, the second compensation factor is d, and the compensated control amount du ax is obtained after compensation by the second compensation factor 506. Then the control amount du ax is added to the predicted disturbance f ax to obtain the second control amount (f ax +du ax ), and the second control amount (f ax +du ax ) after addition is used as the second input of the extended state observer 502.
  • the discrete mathematical model of the extended state observer 502 is:
  • fal is a nonlinear function, and its expression is: Wherein, e is a natural constant, ⁇ is a constant between 0 and 1, and ⁇ is a user-defined constant that affects the filtering effect; ⁇ 1 , ⁇ 2 , ⁇ 3 , ⁇ 1 , ⁇ 2 , ⁇ 3 and ⁇ 1 are adjustable parameters of the extended state observer 502; za1x (k+1), za2x (k+1) and za3x (k+1) are the values of za1x , za2x and za3x at time k+1 respectively; h is the sampling period; ea1x is the first system state error; ea2x is the second system state error; xa (k) is the actual radial (axial) displacement at time k, uax (k) is the compensated control amount uax at time k; fax (k) is the predicted disturbance fax output by the transfer learning module (504) at time k, ⁇ 1 is usually
  • the actual radial displacement xa collected by the first radial displacement sensor 6 is used as the first input of the expanded state observer 502, and the second control quantity ( fax + duax ) is used as the second input of the expanded state observer 502.
  • the expanded state observer 502 outputs the tracking signal za1x of the actual radial displacement xa , the first-order differential signal za2x of the tracking signal za1x , and the observed disturbance za3x of the system.
  • the actual radial displacement x a , the source domain data set D s and the target domain data set D t are used as inputs of the transfer learning module 504.
  • the transfer learning module 504 outputs the predicted disturbance f ax of the system.
  • the predicted disturbance f ax is added to the observed disturbance za3x output by the extended state observer 502.
  • the sum of the disturbances ( za3x + f ax ) is used as the input of the first compensation factor 505.
  • the feedback control quantity u a0x output by the nonlinear state error feedback control law 503 is related to the first control quantity
  • the compensated control variable u ax is obtained by subtracting the compensation value u ax , which is the radial control current expected value iax * .
  • the construction methods of the second, third, fourth and fifth optimized anti-disturbance rejection controllers 20, 21, 22, 23 and 24 are exactly the same as that of the first optimized anti-disturbance rejection controller 20. They output corresponding control current expected values for different given radial or axial displacements respectively. Therefore, the other four optimized anti-disturbance rejection controllers are not described in detail.
  • the method of constructing the transfer learning module 504 is as follows:
  • the high-speed electric spindle supported by the six-pole active magnetic bearing shown in Figure 1 is connected to a traditional anti-disturbance control controller, which is a controller that can detect the disturbance signal of the electric spindle, such as the support vector machine optimized anti-disturbance control controller in Figure 9 (using the anti-disturbance control controller with patent publication number CN112532134A), or other controllers that detect the disturbance signal of the electric spindle, including a tracking differentiator 501, a nonlinear state error feedback control law 503, a first compensation factor 505, a second compensation factor 506, an extended state observer 502 and a least squares support vector machine module.
  • a tracking differentiator 501 a tracking differentiator 501
  • a nonlinear state error feedback control law 503 a first compensation factor 505, a second compensation factor 506, an extended state observer 502
  • a least squares support vector machine module Compared with the first optimized anti-disturbance control 20 in the present invention shown in Figure 7, only the predicted
  • the present invention collects n groups of displacement signals xi and disturbance signals, where n is the number of source domain data samples, 1 ⁇ i ⁇ n.
  • Step S2 Edge distribution distance calculation
  • the key to transfer learning is to quantify the data difference between the source domain and the target domain to ensure the effectiveness of transfer learning.
  • the source domain dataset Ds and the target domain dataset Dt are mapped to the reproducing kernel Hilbert space by the mapping function ⁇ (x), and the maximum mean discrepancy (MMD) is used to characterize the marginal distribution distance DH between the source domain and the target domain.
  • the marginal distribution distance DH between the source domain dataset Ds and the target domain dataset Dt is calculated by the following formula:
  • H represents the reproducing kernel Hilbert space
  • xi is the displacement signal in the source domain dataset
  • zj is the displacement signal in the target domain dataset
  • n is the number of samples in the source domain dataset
  • m is the number of samples in the target domain dataset
  • ⁇ ( xi ) and ⁇ ( zj ) are the mapping functions ⁇ (x) of the displacement signals in the source domain dataset and the target domain dataset in the reproducing kernel Hilbert space, respectively.
  • the initial function of the mapping function ⁇ (x) is defined as the softmax function.
  • the perturbation signal yi in the source domain dataset can realize the transformation of the original data in the reproducing kernel Hilbert space through the mapping function. Then the perturbation signal yi in the source domain dataset is:
  • Step S3 Construct a transfer learning model based on the marginal distribution distance DH :
  • f(y i ) represents the transfer learning model
  • argmin represents minimizing the objective function
  • L is the loss function that measures the accuracy of the source domain dataset.
  • L selects the cross entropy loss function.
  • ⁇ is a correction coefficient, which is selected in the range of [0, 1], and the best value in the present invention is 0.5.
  • Step S4 Train the transfer learning model f(y i )
  • Step S4.1 To train the transfer learning model, call the deep learning toolbox in the MATLAB software environment. Sample the ⁇ group from the source domain dataset Ds , ⁇ n, and use half of the ⁇ group as the training sample set The other half is used as a test sample set Considering the training accuracy and training time, the value range of ⁇ is [100,600].
  • the displacement signal xi in the ⁇ /2 group of training samples is used as the input of the transfer learning model f( yi ), and the disturbance signal yi is used as the output of the transfer learning model f( yi ).
  • Step S4.2 Use the training module to train the ⁇ /2 training sample set D train .
  • Step S4.3 Distribute the edge distances of the ⁇ /2 groups Substituting into the transfer learning model f(y i ), we get the corresponding value of the ⁇ /2 group of transfer learning models f(y i ), that is, Select the mapping function ⁇ min (x) corresponding to the minimum value among the corresponding values of the ⁇ /2 group transfer learning model.
  • the mapping function ⁇ min (x) is the predicted disturbance function f ax (x), and the training is completed.
  • Step S5 Test the predicted disturbance function f ax (x)
  • y outp , y testp represent the pth disturbance signal in the predicted output set D out and the test sample set D test respectively, 0 ⁇ p ⁇ /2.
  • the mean square error threshold ⁇ is set according to the actual accuracy requirement.
  • the present invention sets the range of ⁇ to be [10 -5 ,10 -3 ].
  • the mean square error ⁇ is compared with the set error threshold ⁇ . If the mean square error ⁇ is less than the error threshold ⁇ , it is considered that the prediction disturbance function f ax (x) obtained in step S4.3 meets the requirement; otherwise, if the mean square error ⁇ is greater than or equal to the error threshold ⁇ , it is necessary to readjust the correction coefficient ⁇ , and then reconstruct the transfer learning model f(y i ) based on the adjusted new correction coefficient ⁇ , and repeat steps S3-S5, that is, it is necessary to reconstruct the transfer learning model f(y i ) based on the adjusted new correction coefficient ⁇ , and train the new
  • the transfer learning model f(y i ) is constructed and tested until the mean square error is less than ⁇ , at which point a prediction disturbance function f ax (x) that meets the requirements is obtained
  • Step S6 Constructing an optimized ADRC controller
  • the predicted disturbance function f ax (x) that meets the requirements after testing is built into the transfer learning module 504, and the transfer learning module 504 is connected to the first optimized ADAC 20.
  • the transfer learning module 504 reduces the burden of the extended state observer 502 and improves the stability of the overall control system.
  • the predicted disturbance f ax is added to du ax output by the second compensation factor 506, and the obtained second control quantity (f ax + du ax ) is used as the second input of the extended state observer 502.
  • the predicted disturbance f ax is added to the observed disturbance za3x output by the extended state observer 502 to obtain the disturbance quantity (f ax + za3x ), and the disturbance quantity (f ax + za3x ) is compensated by the first compensation factor 505 to obtain the first control quantity
  • the feedback control quantity u a0x output by the nonlinear state error feedback control law 503 is combined with the first control quantity Make a difference and get the control amount after compensation
  • the current radially controls the current desired value, thus forming an optimized active disturbance rejection controller.
  • the second, third, fourth and fifth optimized auto-disturbance rejection controllers 21, 22, 23, 24 are also connected to a transfer learning module to predict disturbances in the same way as the first optimized auto-disturbance rejection controller 20, so it is not repeated here.
  • five optimized auto-disturbance rejection controllers 20, 21, 22, 23, 24 with the same structure are constructed, three of which are connected in series to the front end of the three-degree-of-freedom composite controlled object 210, and two optimized auto-disturbance rejection controllers 23, 24 are connected in series to the front end of the two-degree-of-freedom composite controlled object 211, together forming a high-speed electric spindle control system.
  • the high-speed electric spindle supported by the six-pole active magnetic bearing adopts a three-phase inverter to drive and control the first six-pole radial active magnetic bearing 1 and the second six-pole radial active magnetic bearing 3 during operation, and adopts a bipolar power amplifier to drive the axial magnetic bearing 2 to achieve stable suspension of the five degrees of freedom of the electric spindle shaft. Since the three-phase AC inverter is used to drive and control the radial magnetic bearing, there is coupling between the radial magnetic circuits, which affects the high-precision control of the electric spindle shaft. Five self-anti-disturbance controllers are used to decouple the high-speed electric spindle supported by the six-pole active magnetic bearing.
  • the transfer learning algorithm is used to obtain the predicted disturbance, and the displacement signal and disturbance signal of the electric spindle rotor under the rated load are collected as the source domain data set. After the load is changed, the displacement signal and disturbance signal are collected as the target domain data set.
  • the transfer learning model is constructed according to the edge distance distribution, and the transfer learning model is trained and tested using the source data. Since the transfer learning has obtained the source data set The features reduce the need for large amounts of data, improve the calculation speed of the control system, and make the system more rapid.
  • the trained transfer learning module is connected to the anti-disturbance controller, and the source domain dataset Ds , the target domain dataset Dt , and the real-time displacement signal of the high-speed electric spindle supported by the six-pole active magnetic bearing are used as the input signals of the trained transfer learning model.
  • the predicted disturbance is used as the output of the transfer learning module to realize the operation of the control system.

Landscapes

  • Engineering & Computer Science (AREA)
  • General Engineering & Computer Science (AREA)
  • Mechanical Engineering (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Magnetic Bearings And Hydrostatic Bearings (AREA)

Abstract

本发明公开一种六极主动磁轴承支承的高速电主轴控制系统的构造方法，由源域数据集与目标域数据集的边缘分布距离构建迁移学习模型并训练得到预测扰动函数，将预测扰动函数内置于迁移学习模块中，将实际位移作为迁移学习模块的输入得到预测扰动，由跟踪微分器、非线性状态误差反馈控制律、第一补偿因子、第二补偿因子、迁移学习模块以及扩张状态观测器共同构成优化自抗扰控制器；三个优化自抗扰控制器串接在三自由度复合被控对象前端以及两个优化自抗扰控制器串接在二自由度复合被控对象前端共同构成控制系统；采用基于迁移学习优化的自抗扰控制对扰动进行预测并使用优化的扩张状态观测器对系统总扰动进行估计和补偿，提高系统控制性能。

Description

六极主动磁轴承支承的高速电主轴控制系统的构造方法 技术领域
本发明属于电气和机械传动设备技术领域，具体涉及六极主动磁轴承支承的高速电主轴的自抗扰解耦控制系统，适用于多变量、非线性、强耦合的六极主动磁轴承支承的高速电主轴的解耦控制。
背景技术
六极主动磁轴承支承的高速电主轴是一种由两个六极主动径向磁轴承和一个单自由度轴向磁轴承共同支承的电主轴。磁轴承是利用磁场力将转子悬浮于空间中，实现转子与定子间没有任何机械接触的一种新型高性能轴承，解决了传统轴承主轴热变形、寿命短、抗震性不足等问题。径向磁轴承采用六极式对称定子结构，六极磁轴承的定子拥有对称结构，径向两个自由度之间的磁路耦合性小，承载力大，性能优秀。径向控制线圈缠绕在径向磁轴承的磁极上，利用三相交流逆变器驱动，通电后形成径向控制磁通产生悬浮力，实现径向的稳定悬浮。由于径向磁轴承均由逆变器驱动，不可避免地会引起径向磁通产生耦合性，影响电主轴的精密控制，因此需要通过高效的解耦控制器来实现磁轴承的高速、高精度稳定运行。
目前，解决磁悬浮电主轴的解耦控制的方法主要包括近似线性化自抗扰解耦控制方法、矩阵变换器控制方法、逆系统解耦控制方法和其他智能算法的综合应用。近似线性化自抗扰解耦控制是一种新的解耦控制思路，它将五个自由度的磁悬浮电主轴中径向磁轴承径向磁路之间的相互耦合作用看作系统内部扰动，采用扩张状态观测器来估计和补偿内部扰动，实现精确解耦控制，使得磁轴承各自由度实现线性化。然而，传统的近似线性化自抗扰控制器性能依赖于内部参数的选取，当六极主动磁轴承支承的高速电主轴工作状况发生变化时，自抗扰控制器无法对扰动做出精确的补偿。
中国专利公开号为CN103076740B、名称为“交流磁悬浮电主轴控制器的构造方法”文献中提供的构造方法，采用模糊神经网络选取静态实验数据离线训练控制对象的逆系统，但采用模糊控制，控制精度较低，无法满足电主轴高精要求。中国专利公开号为CN112532134A、名称为“五自由度磁悬浮电主轴最小二乘支持向量机优化控制系统”文献中提供的控制系统，选取位移量的跟踪信号和微分信号作为训练样本，对电主轴的扰动预测并补偿，但最小二乘支持向量机中参数的选取会导致求解速度慢，并且每次运行都需要大量的训练数据，影响系统控制精度。
发明内容
本发明的目的是针对现有五自由度六极主动磁轴承支承的高速电主轴控制存在的依赖于内部参数选取、补偿扰动欠精确、运行需要大量训练数据以及求解速度慢、控制精度差等问题，提供一种能解决上述问题的六极主动磁轴承支承的高速电主轴控制系统的构造方法。
为实现上述目的，本发明六极主动磁轴承支承的高速电主轴控制系统的构造方法采用的技术方案是：
所述的高速电主轴具有第一、第二六极径向主动磁轴承和轴向磁轴承，其特征是包括以下步骤：
步骤1)：构造包含有第一六极径向主动磁轴承和轴向磁轴承的三自由度复合被控对象，以及包含有第二六极径向主动磁轴承的二自由度复合被控对象，所述的三自由度复合被控对象和二自由度复合被控对象的输入是对应的四个径向控制电流期望值以及一个轴向控制电流期望值；
步骤2)：所述的高速电主轴在额定负载下运行，采集其输出的多组位移信号及对应的扰动信号组成源域数据集，在额定负载范围内更改负载的大小再采集多组高速电主轴的位移信号组成目标域数据；计算所述的源域数据集与所述的目标域数据集的边缘分布距离，基于边缘分布距离构建迁移学习模型并训练该迁移学习模型得到预测扰动函数，将预测扰动函数内置于迁移学习模块(504)中；
步骤3)：将实际径向或轴向位移作为所述的迁移学习模块(504)的输入，根据所述的预测扰动函数得到预测扰动，所述的径向或轴向控制电流期望值经第二补偿因子(506)补偿后，再与所述的预测扰动相加后得到第二控制量；
将所述的第二控制量与所述的实际径向或轴向位移作为扩张状态观测器(502)的输入，扩张状态观测器(502)输出跟踪信号、跟踪信号的一阶微分信号以及观测扰动；
将所述的预测扰动与所述的观测扰动相加后，再经第一补偿因子(505)的补偿后得到第一控制量；
给定的径向或轴向位移经跟踪微分器(501)输出其跟踪信号和微分信号，将跟踪微分器(501)输出的跟踪信号与扩张状态观测器(502)输出的跟踪信号作差，跟踪微分器(501)输出的微分信号与扩张状态观测器(502)输出的一阶微分信号作差，得到的两个误差共同作为非线性状态误差反馈控制律(503)的输入，非线性状态误差反馈控制律(503)输出反馈控制量；
所述的反馈控制量与所述的第一控制量作差，得到补偿后的控制量作为所述的径向或轴向控制电流期望值；
由所述的跟踪微分器(501)、非线性状态误差反馈控制律(503)、第一补偿因子(505)、第二补偿因子(506)、迁移学习模块(504)以及扩张状态观测器(502)共同构成优化自抗扰控制器；
步骤4)：三个所述的优化自抗扰控制器串接在所述的三自由度复合被控对象前端，两个所述的优化自抗扰控制器串接在所述的二自由度复合被控对象前端，共同构成高速电主轴控制系统。
本发明采用上述技术方案后的优点在于：
(1)本发明采用基于迁移学习优化的自抗扰控制，优化扩张状态观测器的性能，对电主轴的扰动进行预测并使用优化的扩张状态观测器对系统总扰动进行估计和补偿，从而实现对高速电主轴中的两个六极径向主动磁轴承径向自由度的精确解耦控制。该方法能够将非理想情况下的多输入多输出的强耦合、非线性系统解耦成多输入多输出的无耦合线性系统，提高系统控制性能。
(2)本发明采用迁移学习算法获取预测扰动，采用收集自抗扰控制器控制下的位移信号和扰动信号作为源域数据集，减少了数据需求，更换负载后收集位移信号及相应的扰动信号作为目标域数据集，根据边缘距离分布构建迁移学习模型，并利用源域数据训练和测试迁移学习模型。由于迁移学习已经获取了源域数据集的特征，所以可避免过多选取内部参数，减少了对大量训练数据的需求，因此，提高了控制系统的计算速度，使系统具有更好的快速性。
(3)在预测扰动的输出过程中，迁移学习采用训练数据获取数据特征，可以避免过度拟合，提高了控制系统的收敛速度和控制精度。
(4)将六极主动磁轴承支承的高速电主轴中径向磁轴承采用两个径向磁轴承和一个轴向磁轴承实现电主轴转轴的五个自由度稳定悬浮，径向与轴向磁路彼此间独立，耦合性减小。用三相逆变器驱动控制径向磁轴承。采用五个自抗扰控制器对高速电主轴进行解耦控制，无需精确的数学模型，具有超调小、响应快、抗干扰能力强等特点，能够大幅提高电主轴控制系统的控制精度。
附图说明
图1是六极主动磁轴承支承的高速电主轴的结构示意图；
图2是图1中第一六极径向主动磁轴承的径向结构以及磁路示意图；
图3是图1中的第一六极径向主动磁轴承的轴向结构以及磁路示意图；
图4是基于图1中的第一六极主动径向磁轴承和轴向磁轴承构造三自由度复合被控对象的等效结构框图；
图5是基于图1中的第二六极主动径向磁轴承构造二自由度复合被控对象的等效结构框图；
图6是所构造的六极主动磁轴承支承的高速电主轴控制系统的结构框图；
图7是图6中第一优化自抗扰控制器的控制结构框图；
图8是图7中迁移学习模块的构造方法流程图；
图9是相对图7示出的传统自抗扰控制器的控制结构框图。
图中：1.第一六极径向主动磁轴承；2.轴向磁轴承；3.第二六极径向主动磁轴承；4.高速主轴电机；5.转轴；6、7、8、9.径向位移传感器；10.轴向位移传感器；11.辅助轴承；12、13.端盖；14.钢筒外套；15.钢筒内套；16.高速主轴电机定子；17.高速主轴电机转子；20.第一优化自抗扰控制器；21.第二优化自抗扰控制器；22.第三优化自抗扰控制器；23.第四优化自抗扰控制器；24.第五优化自抗扰控制器；25.第一位移传感器；26.第二位移传感器；27.第三位移传感器；28.第四位移传感器；29.第五位移传感器；31.第一Clark逆变换；32第二Clark逆变换；33.第一电流跟踪型逆变器；34.第二电流跟踪型逆变器；35.轴向功率放大器；
101.第一径向转子；102.第一径向定子；103.第一径向控制线圈；104.第一径向定子磁极；105.偏置磁通；106.控制磁通；107.径向气隙；
201.轴向定子；202.轴向转子；203.轴向控制线圈；204.轴向定子磁极；210.三自由度复合被控对象；211.二自由度复合被控对象；
301.第二径向转子；302.第二径向转子；303.第二径向控制线圈；
501.第一跟踪微分器；502.第一扩张状态观测器；503.第一非线性反馈控制律；504.迁移学习模块；505.第一补偿因子；506.第二补偿因子。
具体实施方式
参见图1所示的六极磁轴承支承的电主轴基本结构，包括转轴5、高速主轴电机4、第一六极径向主动磁轴承1、轴向磁轴承2、第二六极径向主动磁轴承3、钢筒外套14、钢筒内套15、径向位移传感器6、7、8、9、轴向位移传感器10、辅助轴承11以及端盖12、13。其中，高速主轴电机4包括高速主轴电机定子16和高速主轴电机转子17，高速主轴电机转子17同轴心地固定套在转轴5外，带动转轴5旋转。在高速主轴电机4的左侧为第二六 极径向主动磁轴承3，右侧为轴向磁轴承2和第一六极径向主动磁轴承1，第一六极径向主动磁轴承1和第二六极径向主动磁轴承3结构完全相同。在高速主轴电机4、第一六极径向主动磁轴承1、第二六极径向主动磁轴承3和轴向磁轴承2的外部共同包裹一个钢筒，该钢筒由钢筒外套14和钢筒内套15构成，钢筒外套14和钢筒内套15之间设有用于水冷散热的螺旋沟道。钢筒外套14和钢筒内套15的左右端面上分别各自固定连接一个端盖12、13，两个端盖12、13各通过一个辅助轴承11支撑在转轴5外部，辅助轴承11用于磁轴承停机或者故障状态下对转轴5进行支承。转轴5同轴穿过第一六极径向主动磁轴承1、轴向磁轴承2、高速主轴电机4以及第二六极径向主动磁轴承3。
第二六极径向主动磁轴承3位于高速主轴电机4和左侧的端盖13之间，分别与高速主轴电机4和左侧的端盖13之间都留有轴向距离。在左侧的端盖13和第二六极径向主动磁轴承3之间安装第三径向位移传感器8和第四径向位移传感器9，第三径向位移传感器8和第四径向位移传感器9沿径向上面对面地对称安装，用来检测转轴5左侧的径向两个自由度的位移。
轴向磁轴承2由轴向转子201、轴向定子202以及轴向控制线圈203构成。轴向磁轴承2位于高速主轴电机4和第一六极径向主动磁轴承1之间。在轴向上，采用两个环形限位套筒分别置于第一六极径向主动磁轴承1和轴向磁轴承2之间、轴向磁轴承2和高速主轴电机4之间，用于固定三者的轴向位置。第一六极径向主动磁轴承1位于轴向磁轴承2和右侧的端盖12之间，且与轴向磁轴承2和右侧的端盖12之间均留有轴向距离。在右侧的端盖12与第一六极径向主动磁轴承1之间设有第一径向位移传感器6和第二径向位移传感器7，第一径向位移传感器6和第二径向位移传感器7在径向上面对面对称安装，用来检测转轴5右侧径向两个自由度的位移。在右侧的端盖12上安装轴向位移传感器10，用于检测转轴5轴向自由度的位移。
第一、第二、第三、第四径向位移传感器6、7、8、9和轴向位移传感器10的探头均采用电涡流传感器。
参见图2和图3所示的第一六极径向主动磁轴承1，由第一径向转子101、第一径向定子102和第一径向控制线圈103组成，第一径向转子101为圆筒状，同轴心地固定套在转轴5外，和转轴5成为一体。第一径向定子102的外壁固定连接于钢筒内套15的内壁上。第一径向定子102同轴套在第一径向转子101外部，且二者之间存在径向气隙107。第一径向定子102由圆环形定子轭以及六个沿定子轭内壁圆周均匀布置的呈凸型的第一径向定子磁极104组成，第一径向定子磁极104的侧表面为圆柱面。每个第一径向定子磁极104 上都缠绕有方向相同的第一径向控制线圈103，相对的两个磁极上的第一径向控制线圈103串联成一相，构成三相线圈，三相线圈采用星形连接方式，采用一个三相逆变器控制第一径向控制线圈103，以控制悬浮力大小和方向。
第一径向控制线圈103为第一六极径向主动磁轴承1提供偏置磁通105和控制磁通106。偏置磁通105的回路是：自一个第一径向定子磁极104、径向气隙107进入第一径向转子101，再经对面的径向气隙107进入对面的一个第一径向定子磁极104，最后经过定子轭重新回到开始的一个第一径向定子磁极104。偏置磁通105回路主要作用为使偏置磁动势产生的悬浮力平衡转子的重力。控制磁通106回路是：自一个第一径向定子磁极104、径向气隙107进入第一径向转子101，再经对面的径向气隙107进入对面的一个第一径向定子磁极104。偏置磁通105和控制磁通106相互作用，在第一六极径向主动磁轴承1的第一径向转子101上产生径向二自由度悬浮力。
参见图1，第二六极径向主动磁轴承3由第二径向定子302、第二径向转子301和第二径向控制线圈303组成。其结构和装配方式与第一径向定子102、第一径向转子101和第一径向控制线圈103对应地完全相同。第一径向定子102和第一径向转子101之间的第一径向气隙107的大小与第二径向定子302和第二径向转子301之间的第二径向气隙的大小相同，均取0.5mm。同理，第二六极径向主动磁轴承的偏置磁通和控制磁通相互作用，在第二六极径向主动磁轴承3的第二径向转子301上产生径向二自由度悬浮力。
轴向磁轴承2由轴向定子201、轴向转子202和轴向控制线圈203构成。轴向转子202为圆筒状，同轴心地固定套在转轴5外且和转轴5成为一体。轴向控制线圈203分别缠绕在轴向定子202的左右两个磁极上，由双极性直流功率放大器通电产生轴向控制磁通和轴向偏置磁通，轴向控制磁通和轴向偏置磁通相互作用，在轴向磁轴承2的轴向转子202上产生轴向悬浮力。
第一径向定子102、第一径向转子101、第二径向定子302、第二径向转子301、轴向定子202和轴向转子201均采用硅钢片叠压而成。第一径向控制线圈103、第二径向控制线圈303和轴向控制线圈203均采用直径为0.67mm的带绝缘漆铜线制成。辅助轴承11采用调心滚珠轴承。
如图4所示，基于第一六极径向主动磁轴承1和轴向磁轴承2，构造包含有第一六极径向主动磁轴承1和轴向磁轴2的三自由度复合被控对象210，由依次串联的第一Clark逆变换31、第一电流跟踪型逆变器33、第一六极径向主动磁轴承1以及依次串联的轴向功率放大器35、轴向磁轴承2共同构成三自由度复合被控对象210，三自由度复合被控对象210 的输入是第一径向控制电流期望值iax *、第二径向控制电流期望值iay *和轴向控制电流期望值iaz *，三自由度复合被控对象210的输出是实际径向位移xa、ya和实际轴向位移za。其中，第一径向控制电流期望值iax *和第二径向控制电流期望值iay *经过第一Clark逆变换31得到电流iau *、iav *和iaw *，电流iau *、iav *和iaw *经第一电流跟踪型逆变器33输出径向控制电流iau、iav和iaw。以该径向控制电流iau、iav、iaw控制第一六极径向主动磁轴承1中的第一径向控制线圈103，第一六极径向主动磁轴承1产生实际径向位移xa、ya。第一、第二径向位移传感器6、7分别检测到实际径向位移xa、ya。轴向控制电流期望值iaz *经轴向功率放大器35后输出轴向控制电流iaz，轴向控制电流iaz控制轴向磁轴承2中的轴向控制线圈203，轴向磁轴承2产生实际轴向位移za，轴向位移传感器10检测到实际轴向位移za，构造成包含了第一六极径向主动磁轴承1和轴向磁轴承2的三自由度复合被控对象210。
如图5所示，基于第二六极径向主动磁轴承3，构造包含有第二六极径向主动磁轴承3的二自由度复合被控对象211。由依次串联的第二Clark逆变换32、第二电流跟踪型逆变器34、第二六极径向主动磁轴承3共同构成二自由度复合被控对象211，二自由度复合被控对象211的输入是第三、第四径向控制电流期望值ibx *、iby *，二自由度复合被控对象211的输出是实际径向位移xb、yb。其中，第三、第四径向控制电流期望值ibx *、iby *共同输入第二Clark逆变换32，第二Clark逆变换32输出电流ibu *、ibv *和ibw *，电流ibu *、ibv *和ibw *经第二电流跟踪型逆变器34输出径向控制电流ibu、ibv、ibw至第二六极径向主动磁轴承3，控制其中的第二径向控制线圈302，第二六极径向主动磁轴承3输出实际径向位移xb、yb，构造成包含了第二六极径向主动磁轴承3的二自由度复合被控对象211。第三、第四径向位移传感器8、9检测到该实际径向位移xb、yb。
如图6所示，由于需要对图1所示的六极主动磁轴承支承的高速电主轴五个自由度进行控制，因此采用五个优化自抗扰控制器构成控制系统，五个优化自抗扰控制器分别控制五个自由度。五个优化自抗扰控制器分别是第一、第二、第三、第四、第五优化自抗扰控制器20、21、22、23、24。这五个优化自抗扰控制器仅是输入和输出信号不同，内部结构完全相同。
第一、第二径向位移传感器6、7采集三自由度复合被控对象210的实际径向位移xa、ya，实际径向位移xa、ya分别一一对应地输入至对应的第一优化自抗扰控制器20和第二优化自抗扰控制器21中。轴向位移传感器10采集三自由度复合被控对象210的实际轴向位移za，实际轴向位移za输入至第三优化自抗扰控制器中22中。同时，将相应的给定径向位移xa *输入至第一优化自抗扰控制器20，第一优化自抗扰控制器20对输入的实际径向位移xa和给定径向位移xa *处理，输出第一径向控制电流期望值iax *至三自由度复合被控对象210中。同理，将相应的给定径向位移ya *输入至第二优化自抗扰控制器21，第二优化自抗扰控制器21输出第二径向控制电流期望值iay *至三自由度复合被控对象210中；将给定轴向位移za *输入至第三优化自抗扰控制器22，第三优化自抗扰控制器22输出轴向控制电流期望值iaz *至三自由度复合被控对象210中。由此，将第一径向控制电流期望值iax *、第二径向控制电流期望值iay *和轴向控制电流期望值iaz *共同作为三自由度复合被控对象210的输入。
第三、第四径向位移传感器8、9采集二自由度复合被控对象211的实际径向位移xb、yb，实际径向位移xb、yb分别一一对应地输入至对应的第四优化自抗扰控制器23和第五优化自抗扰控制器24中，将相应的给定径向位移xb *输入至第四优化自抗扰控制器23，第四优化自抗扰控制器23对输入的实际径向位移xb和给定径向位移xb *处理，输出第三径向控制电流期望值ibx *至二自由度复合被控对象211中。将相应的给定径向位移yb *输入至第五优化自抗扰控制器24中，第五优化自抗扰控制器24输出第四径向控制电流期望值iby *至二自由度复合被控对象211中。第三径向控制电流期望值ibx *和第四径向控制电流期望值iby *共同作为二自由度复合被控对象211的输入。
如图7所示，以下以第一优化自抗扰控制器20为例具体描述优化自抗扰控制器的构造过程：
第一优化自抗扰控制器20由跟踪微分器501、非线性状态误差反馈控制律503、第一补偿因子505、第二补偿因子506、迁移学习模块504以及扩张状态观测器502构成。
给定径向位移xa *经跟踪微分器501输出给定径向位移xa *的跟踪信号va1x和给定径向位移xa *的微分信号va2x。跟踪微分器501的数学模型为：
式中：sat为非线性函数，表达式为：其中T为快速跟踪因子，取值范围[0,1]，T的取值决定系统跟踪性能；ε为积分步长；h为系统采样周期；为给定位移在k时刻的值；νa1x(k)和νa1x(k+1)分别为给定径向位移跟踪信号va1x在k和k+1时刻的值；νa2x(k)和νa2x(k+1)分别为给定径向位移微分信号va2x在k和k+1时刻的值。
实际径向位移xa经扩张状态观测器502输出跟踪信号za1x、跟踪信号za1x的一阶微分信号za2x以及系统的观测扰动za3x。
将跟踪微分器501输出的跟踪信号va1x与扩张状态观测器502输出的跟踪信号za1x作差，得到第一系统状态误差ea1x＝va1x-za1x。将跟踪微分器501输出的微分信号va2x和扩张状态观测器502输出的一阶微分信号za2x作差得到第二系统状态误差ea2x＝va2x-za2x。两个系统状态误差ea1x和ea2x共同作为非线性状态误差反馈控制律503的输入，非线性状态误差反馈控制律503输出反馈控制量ua0x。非线性状态误差反馈控制律的模型为：
ua0x＝k1fal(ea1x,α4,δ2)+k2fal(ea2x,α5,δ2)，
式中：k1、k2、α4、α5和δ2为非线性反馈控制律的可调参数，一般α4取0.5；α5取0.25；k1、k2的范围为[0,100]，δ2的范围为[0,1]，非线性函数其中e为自然常数，α为0～1之间的常数，δ为影响滤波效果自定义的常数。
反馈控制量ua0x经第一补偿因子505补偿，得到补偿后的控制量uax：
式中：1/d为第一补偿因子；fax为迁移学习模块504输出的预测扰动，考虑综合因素，d的取值范围为[10，200]。
补偿后的控制量uax还作为第二补偿因子506的输入，第二补偿因子为d，经过第二补偿因子506的补偿得到补偿后的控制量duax，再将控制量duax与预测扰动fax相加，得到第二控制量(fax+duax)，相加之后的第二控制量(fax+duax)作为扩张状态观测器502的第二个输入。扩张状态观测器502的离散数学模型为：
式中：fal为非线性函数，其表达式为：其中e为自然常数，α为0～1之间的常数，δ为影响滤波效果的自定义常数；α1、α2、α3、β1、β2、β3和δ1为扩张状态观测器的502可调参数；za1x(k+1)、za2x(k+1)和za3x(k+1)分别为za1x、za2x和za3x在k+1时刻的值；h为采样周期；ea1x为第一系统状态误差；ea2x为第二系统状态误差；xa(k)为k时刻的实际径向(轴向)位移，uax(k)为k时刻的补偿后的控制量uax；fax(k)为在k时刻由迁移学习模块(504)输出的预测扰动fax，α1通常取0.75，α2取0.5，α3取0.25，δ1取采样周期的5～10倍，β1、β2、β3要根据系统要求的跟踪效果不断调节。
第一径向位移传感器6采集的实际径向位移xa作为扩张状态观测器502的第一个输入，第二控制量(fax+duax)作为扩张状态观测器502的第二个输入，扩张状态观测器502输出实际径向位移xa的跟踪信号za1x、跟踪信号za1x的一阶微分信号za2x以及系统的观测扰动za3x。
将实际径向位移xa、源域数据集Ds以及目标域数据集Dt作为迁移学习模块504的输入，迁移学习模块504输出系统的预测扰动fax，该预测扰动fax与扩张状态观测器502输出的观测扰动za3x相加，相加后得到扰动之和(za3x+fax)作为第一补偿因子505的输入， 经过第一补偿因子505作用后得到第一控制量非线性状态误差反馈控制律503输出的反馈控制量ua0x与第一控制量作差，得到补偿后的控制量uax，该补偿后的控制量uax即是径向控制电流期望值iax *，uax＝iax *，作为第一优化自抗扰控制器20的输出。
第二、第三、第四、第五优化自抗扰控制器20、21、22、23、24的构造方法与第一优化自抗扰控制器20完全相同，分别针对不同的给定径向或轴向位移，输出的是相应的控制电流期望值，因此，不再赘述其他四个优化自抗扰控制器。
参见图7、图8和图9，构造迁移学习模块504的方法如下：
步骤S1：数据集收集
将图1所示的六极主动磁轴承支承的高速电主轴与传统自抗扰控制器相连接，所述的传统自抗扰控制器是可以检测到电主轴扰动信号的控制器，例如图9中的支持向量机优化的自抗扰控制器(采用专利公开号为CN112532134A的自抗扰控制器)，或其他的检测到电主轴扰动信号的控制器，包括跟踪微分器501、非线性状态误差反馈控制律503、第一补偿因子505、第二补偿因子506、扩张状态观测器502以及最小二乘支持向量机模块，与图7所示的本发明中的第一优化自抗扰控制器20相比，仅是预测扰动输出模块不同，其余部分的功能和作用一样。
在图9所示的传统的支持向量机优化自抗扰控制器控制下，使图1所示的六极主动磁轴承支承的高速电主轴在额定负载下运行，采集其输出的多组位移信号xi及其对应的扰动信号yi，对位移信号xi和扰动信号yi作预处理，组成源域数据集Ds＝{(x1,y1),...,(xn,yn)}，本发明采集的是n组位移信号xi及扰动信号，n为源域数据样本数目，1≤i≤n。在额定负载范围内更改负载的大小，再采集多组高速电主轴的位移信号zj，并对位移信号zj作预处理，组成目标域数据Dt＝{z1,...,zm}，1≤j≤m，m是采集的位移信号zj数目，也是目标域数据样本数目。
步骤S2：边缘分布距离计算
迁移学习关键在于量化源域和目标域之间的数据差异，以确保迁移学习的有效性。通过映射函数θ(x)分别将源域数据集Ds与目标域数据集Dt映射至再生核希尔伯特空间，用最大均值差异(Maximum Mean Discrepancy,MMD)来表征源域和目标域的边缘分布距离DH。源域数据集Ds与目标域数据集Dt的边缘分布距离DH由以下公式进行计算：
式中：H表示再生核希尔伯特空间，xi为源域数据集中位移信号；zj为目标域数据集中的位移信号；n为源域数据集样本数；m为目标域数据集样本数；θ(xi)、θ(zj)分别为源域数据集和目标域数据集中的位移信号在再生核希尔伯特空间中的映射函数θ(x)，定义映射函数θ(x)的初始函数为softmax函数，源域数据集中的扰动信号yi可以通过该映射函数实现原始数据在再生核希尔伯特空间中的变换，则源域数据集中的扰动信号yi为：
e为自然常数。
步骤S3：基于边缘分布距离DH构建迁移学习模型：
式中：f(yi)表示迁移学习模型；argmin表示使目标函数取最小值；L为度量源域数据集精度的损失函数，此处L选择交叉熵损失函数，λ为修正系数，选取范围在[0，1]，本发明最佳为0.5。
步骤S4：训练迁移学习模型f(yi)
步骤S4.1：为了训练迁移学习模型，在MATLAB软件环境下调用深度学习工具箱。从源域数据集Ds中抽样选取其中的ω组，ω≤n，将ω组中的其中一半作为训练样本集另一半作为测试样本集考虑训练精度和训练时间，ω取值范围为[100,600]。将ω/2组训练样本集中的位移信号xi作为迁移学习模型f(yi)的输入，扰动信号yi作为迁移学习模型f(yi)的输出。
步骤S4.2：用训练模块对ω/2组训练样本集Dtrain进行训练。将ω/2组训练样本集作为迁移学习模型f(yi)的输入输出信号，通过软件拟合输出ω/2组映射函数将ω/2组训练样本集中的位移信号映射至再生核希尔伯特空间，分别将ω/2组映射函数 代入边缘分布距离的计算公式，得到ω/2组边缘分布距离
步骤S4.3：分别将ω/2组边缘分布距离代入迁移学习模型f(yi)，得到ω/2组迁移学习模型f(yi)的对应值，即选取ω/2组迁移学习模型对应值中最小值所对应的映射函数θmin(x)，该映射函数θmin(x)即为预测扰动函数fax(x)，训练完成。
步骤S5：测试预测扰动函数fax(x)
将测试样本集Dtest中的位移信号[xtest1,xtest2,...,xtestω/2]输入至训练完成的预测扰动函数fax(x)，即可得到预测扰动输出集Dout＝[yout1,yout2,…,youtω/2]。
针对测试样本集取出其中的所有扰动信号，形成扰动信号样本集[ytest1,ytest2,...,ytestω/2]，计算出该扰动信号样本集[ytest1,ytest2,...,ytestω/2]和所述的预测扰动输出集Dout＝[yout1,yout2,…,youtω/2]的均方误差：
其中，youtp，ytestp分别表示预测输出集Dout与测试样本集Dtest中第p个扰动信号，0<p<ω/2。
根据实际精度要求设定均方误差阈值γ,本发明设定γ的范围为[10-5,10-3]。将均方误差Δ与设定的误差阈值γ相比较，若均方误差Δ小于误差阈值γ，则认为步骤S4.3得到的预测扰动函数fax(x)满足要求；反之，若均方误差δ大于等于误差阈值γ，则需要重新调整修正系数λ，再根据调整后的新的修正系数λ，再重新构建迁移学习模型f(yi)，重复执行步骤S3-S5，即需要基于调整后的新的修正系数λ来重新构建迁移学习模型f(yi)，并训练新 构建迁移学习模型f(yi)且作测试，直至均方误差小于γ为止，此时便得到符合要求的预测扰动函数fax(x)。
步骤S6：构建优化自抗扰控制器
参见图7，将测试后符合要求的预测扰动函数fax(x)内置于迁移学习模块504中，将迁移学习模块504接入第一优化自抗扰控制器20，迁移学习模块504降低了扩张状态观测器502的负担，提高整体控制系统的稳定性。
六极主动磁轴承支承的高速电主轴实时位移信号xa作为迁移学习模块504的输入信号，迁移学习模块504输出预测扰动fax＝fax(x)。将预测扰动fax与第二补偿因子506输出的duax相加，得到的第二控制量(fax+duax)作为扩张状态观测器502的第二个输入。将预测扰动fax与扩张状态观测器502输出的观测扰动za3x相加，得到扰动量(fax+za3x)，扰动量(fax+za3x)经第一补偿因子505补偿后得到第一控制量将非线性状态误差反馈控制律503输出的反馈控制量ua0x与第一控制量作差，得到补偿后的控制量作为电流径向控制电流期望值，如此构成优化自抗扰控制器。
第二、第三、第四和第五优化自抗扰控制器21、22、23、24中也同样地各自连接一个迁移学习模块，预测出扰动，方法与第一优化自抗扰控制器20雷同，因此不再赘述。如此地便构造成五个结构相同的五个优化自抗扰控制器20、21、22、23、24，其中的三个优化自抗扰控制器20、21、22串接在三自由度复合被控对象210的前端，两个优化自抗扰控制器23、24串接在二自由度复合被控对象211的前端，共同构成高速电主轴控制系统。
根据以上所述，六极主动磁轴承支承的高速电主轴在工作时采用三相逆变器驱动控制第一六极径向主动磁轴承1和第二六极径向主动磁轴承3，采用双极性功率放大器驱动轴向磁轴承2即可实现电主轴转轴五个自由度的稳定悬浮。由于采用三相交流逆变器驱动控制径向磁轴承，径向磁路间存在耦合，影响电主轴转轴的高精度控制，采用五个自抗扰控制器对六极主动磁轴承支承的高速电主轴进行解耦控制，无需精确的数学模型，具有超调量小、响应快、抗干扰能力强等特点，能够大幅提高电主轴系统的控制精度。采用迁移学习算法获取预测扰动，收集额定负载下电主轴转子的位移信号和扰动信号作为源域数据集，更换负载后收集位移信号和扰动信号作为目标域数据集，根据边缘距离分布构建迁移学习模型，并利用源于数据训练和测试迁移学习模型。由于迁移学习已经获取了源于数据集的 特征，减少了对大量数据的需求，提高了控制系统的计算速度，使系统具有更好的快速性。将训练完成的迁移学习模块接入自抗扰控制器中，将源域数据集Ds、目标域数据集Dt和六极主动磁轴承支承的高速电主轴实时位移信号作为训练完成的迁移学习模型的输入信号，预测扰动做为迁移学习模块的输出，即可实现控制系统的运行。

Claims (10)

  1. 一种六极主动磁轴承支承的高速电主轴控制系统的构造方法，所述的高速电主轴具有第一、第二六极径向主动磁轴承和轴向磁轴承，其特征是包括以下步骤：
    步骤1)：构造包含有第一六极径向主动磁轴承和轴向磁轴承的三自由度复合被控对象，以及包含有第二六极径向主动磁轴承的二自由度复合被控对象，所述的三自由度复合被控对象和二自由度复合被控对象的输入是对应的四个径向控制电流期望值以及一个轴向控制电流期望值；
    步骤2)：所述的高速电主轴在额定负载下运行，采集其输出的多组位移信号及对应的扰动信号组成源域数据集，在额定负载范围内更改负载的大小再采集多组高速电主轴的位移信号组成目标域数据；计算所述的源域数据集与所述的目标域数据集的边缘分布距离，基于边缘分布距离构建迁移学习模型并训练该迁移学习模型得到预测扰动函数，将预测扰动函数内置于迁移学习模块(504)中；
    步骤3)：将实际径向或轴向位移作为所述的迁移学习模块(504)的输入，根据所述的预测扰动函数得到预测扰动，所述的径向或轴向控制电流期望值经第二补偿因子(506)补偿后，再与所述的预测扰动相加后得到第二控制量；
    将所述的第二控制量与所述的实际径向或轴向位移作为扩张状态观测器(502)的输入，扩张状态观测器(502)输出跟踪信号、跟踪信号的一阶微分信号以及观测扰动；
    将所述的预测扰动与所述的观测扰动相加后，再经第一补偿因子(505)的补偿后得到第一控制量；
    给定的径向或轴向位移经跟踪微分器(501)输出其跟踪信号和微分信号，将跟踪微分器(501)输出的跟踪信号与扩张状态观测器(502)输出的跟踪信号作差，跟踪微分器(501)输出的微分信号与扩张状态观测器(502)输出的一阶微分信号作差，得到的两个误差共同作为非线性状态误差反馈控制律(503)的输入，非线性状态误差反馈控制律(503)输出反馈控制量；
    所述的反馈控制量与所述的第一控制量作差，得到补偿后的控制量作为所述的径向或轴向控制电流期望值；
    由所述的跟踪微分器(501)、非线性状态误差反馈控制律(503)、第一补偿因子(505)、第二补偿因子(506)、迁移学习模块(504)以及扩张状态观测器(502)共同构成优化自抗扰控制器；
    步骤4)：三个所述的优化自抗扰控制器串接在所述的三自由度复合被控对象前端，两个所述的优化自抗扰控制器串接在所述的二自由度复合被控对象前端，共同构成高速电主 轴控制系统。
  2. 根据权利要求1所述的六极主动磁轴承支承的高速电主轴控制系统的构造方法，其特征是：步骤2)中，所述的边缘分布距离H表示再生核希尔伯特空间，xi、zj分别为源域数据集和目标域数据集中位移信号；n、m分别为源域数据集和目标域数据集的样本数；1≤i≤n，1≤j≤m；θ(xi)、θ(zj)分别为源域数据集和目标域数据集中的位移信号在再生核希尔伯特空间中的映射函数θ(x)，映射函数θ(x)的初始函数为softmax函数，则源域数据集中的扰动信号
  3. 根据权利要求2所述的六极主动磁轴承支承的高速电主轴控制系统的构造方法，其特征是：所述的迁移学习模型argmin表示使目标函数取最小值；λ为修正系数，取[0，1]。
  4. 根据权利要求3所述的六极主动磁轴承支承的高速电主轴控制系统的构造方法，其特征是：先从源域数据集中抽样选取ω组，ω≤n，将ω组中的其中一半作为训练样本集，另一半作为测试样本集；再将ω/2组训练样本集作为迁移学习模型f(yi)的输入输出信号，输出ω/2组映射函数，将ω/2组训练样本集中的位移信号映射至再生核希尔伯特空间，分别将ω/2组映射函数代入所述的边缘分布距离DH计算，得到ω/2组边缘分布距离DHi；最后分别将ω/2组边缘分布距离代入所述的迁移学习模型f(yi)，得到ω/2组迁移学习模型f(yi)的对应值，选取对应值中最小值所对应的映射函数，该映射函数即为训练完成的预测扰动函数。
  5. 根据权利要求4所述的六极主动磁轴承支承的高速电主轴控制系统的构造方法，其特征是：将所述的测试样本集中的位移信号输入至所述的训练完成的预测扰动函数，得到预测扰动输出集；取出测试样本集中的所有的扰动信号形成扰动信号样本集，计算出该扰动信号样本集和所述的预测扰动输出集的均方误差，将所述的均方误差与设定的误差阈值相比较，若均方误差小于误差阈值，则训练完成的预测扰动函数满足要求；反之，则调整 所述的迁移学习模型f(yi)中修正系数λ，再重新构建迁移学习模型f(yi)。
  6. 根据权利要求1所述的六极主动磁轴承支承的高速电主轴控制系统的构造方法，其特征是：步骤3)中，所述的跟踪微分器(501)的数学模型为：
    非线性函数T为快速跟踪因子，ε为积分步长；h为采样周期；为给定位移在k时刻的值；νa1x(k)和νa1x(k+1)分别为给定位移跟踪信号va1x在k和k+1时刻的值；νa2x(k)和νa2x(k+1)分别为给定位移微分信号va2x在k和k+1时刻的值。
  7. 根据权利要求6所述的六极主动磁轴承支承的高速电主轴控制系统的构造方法，其特征是：步骤3)中，所述的反馈控制量ua0x＝k1fal(ea1x,α4,δ2)+k2fal(ea2x,α5,δ2)，k1、k2、α4、α5和δ2为非线性反馈控制律的可调参数，α4取0.5；α5取0.25；k1、k2的范围为[0,100]，δ2的范围为[0,1]，e为自然常数，α为0～1之间的常数，δ为影响滤波效果自定义的常数；ea1x＝va1x-za1x；ea2x＝va2x-za2x；za1x、za2x分别为扩张状态观测器(502)输出的跟踪信号和一阶微分信号。
  8. 根据权利要求7所述的六极主动磁轴承支承的高速电主轴控制系统的构造方法，其特征是：步骤3)中，扩张状态观测器(502)的离散数学模型为：
    α1、α2、α3、β1、β2、β3和δ1为可调参数；za1x(k+1)、za2x(k+1)和za3x(k+1)分别为za1x、za2x和观测扰动za3x在k+1时刻的值；xa(k)为k时刻的实际位移，uax(k)为k时刻的补偿后的控制量uax；fax(k)为在k时刻由迁移学习模块(504)输出的预测扰动fax。
  9. 根据权利要求1所述的六极主动磁轴承支承的高速电主轴控制系统的构造方法，其 特征是：步骤3)中，所述的补偿后的控制量ua0x为反馈控制量，1/d为第一补偿因子，取值为[10，200]；za3x为扩张状态观测器(502)输出的观测扰动；fax为预测扰动。
  10. 根据权利要求1所述的六极主动磁轴承支承的高速电主轴控制系统的构造方法，其特征是：步骤1)中，由依次串联的第一Clark逆变换、第一电流跟踪型逆变器、所述的第一六极径向主动磁轴承以及依次串联的轴向功率放大器、所述的轴向磁轴承共同构成所述的三自由度复合被控对象；由依次串联的第二Clark逆变换、第二电流跟踪型逆变器和所述的第二六极径向主动磁轴承共同构成所述的二自由度复合被控对象。
PCT/CN2023/106684 2023-06-19 2023-07-11 六极主动磁轴承支承的高速电主轴控制系统的构造方法 Ceased WO2024259754A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202310727637.XA CN116557421B (zh) 2023-06-19 2023-06-19 六极主动磁轴承支承的高速电主轴控制系统的构造方法
CN202310727637.X 2023-06-19

Publications (1)

Publication Number Publication Date
WO2024259754A1 true WO2024259754A1 (zh) 2024-12-26

Family

ID=87502068

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/106684 Ceased WO2024259754A1 (zh) 2023-06-19 2023-07-11 六极主动磁轴承支承的高速电主轴控制系统的构造方法

Country Status (2)

Country Link
CN (1) CN116557421B (zh)
WO (1) WO2024259754A1 (zh)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN119717547A (zh) * 2025-02-27 2025-03-28 深圳市高川自动化技术有限公司 一种主从轴耦合控制方法、系统、终端设备及介质
CN120217899A (zh) * 2025-05-21 2025-06-27 南昌航空大学 基于ddpg算法的电磁轴承转子系统不平衡控制方法
CN120447350A (zh) * 2025-05-07 2025-08-08 北京品德技术有限公司 一种线性自抗扰控制器及其设计方法和参数整定方法
CN120507992A (zh) * 2025-07-18 2025-08-19 中北大学 一种基于efdt-eso的显模型跟踪飞行控制方法
CN120729123A (zh) * 2025-08-28 2025-09-30 华能国际电力股份有限公司上海石洞口第二电厂 无人值守螺旋卸船机的电流控制方法
CN120928776A (zh) * 2025-10-16 2025-11-11 江苏美弛智造科技有限公司 一种非标自动化检具的检测精度动态校准控制方法
CN121356340A (zh) * 2025-12-16 2026-01-16 江西水利电力大学 基于线性扩张状态观测器和mpc的降压变换器控制方法

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118011788A (zh) * 2023-12-21 2024-05-10 珠海格力电器股份有限公司 一种磁轴承控制方法、装置、介质、电子设备及系统
CN118167735A (zh) * 2024-05-14 2024-06-11 山东志伟环保科技有限公司 一种电磁悬浮轴承结构、风机、真空泵及空分系统

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20170300605A1 (en) * 2016-04-19 2017-10-19 General Electric Company Creating predictive damage models by transductive transfer learning
CN110131312A (zh) * 2019-04-03 2019-08-16 江苏大学 五自由度交流主动磁轴承自抗扰解耦控制器及构造方法
CN112532134A (zh) * 2020-11-26 2021-03-19 江苏大学 五自由度磁悬浮电主轴最小二乘支持向量机优化控制系统
CN114928074A (zh) * 2022-05-12 2022-08-19 西安交通大学 基于迁移学习和自抗扰控制的次同步振荡抑制方法及系统
CN115470903A (zh) * 2022-10-17 2022-12-13 中国科学院计算技术研究所 一种基于迁移学习的血压预测模型训练方法及预测方法

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP6828762B2 (ja) * 2019-03-15 2021-02-10 ダイキン工業株式会社 機械学習装置、及び、磁気軸受装置
CN111829782B (zh) * 2020-07-16 2021-12-07 苏州大学 一种基于自适应流形嵌入动态分布对齐的故障诊断方法
CN115289137A (zh) * 2022-07-29 2022-11-04 常州大学 六极混合磁轴承的线性自抗扰控制器设计消除扰动方法
CN115903497A (zh) * 2022-11-17 2023-04-04 江苏大学 五自由度主动磁悬浮电主轴遗传粒子群优化解耦控制系统

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20170300605A1 (en) * 2016-04-19 2017-10-19 General Electric Company Creating predictive damage models by transductive transfer learning
CN110131312A (zh) * 2019-04-03 2019-08-16 江苏大学 五自由度交流主动磁轴承自抗扰解耦控制器及构造方法
CN112532134A (zh) * 2020-11-26 2021-03-19 江苏大学 五自由度磁悬浮电主轴最小二乘支持向量机优化控制系统
CN114928074A (zh) * 2022-05-12 2022-08-19 西安交通大学 基于迁移学习和自抗扰控制的次同步振荡抑制方法及系统
CN115470903A (zh) * 2022-10-17 2022-12-13 中国科学院计算技术研究所 一种基于迁移学习的血压预测模型训练方法及预测方法

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN119717547A (zh) * 2025-02-27 2025-03-28 深圳市高川自动化技术有限公司 一种主从轴耦合控制方法、系统、终端设备及介质
CN120447350A (zh) * 2025-05-07 2025-08-08 北京品德技术有限公司 一种线性自抗扰控制器及其设计方法和参数整定方法
CN120217899A (zh) * 2025-05-21 2025-06-27 南昌航空大学 基于ddpg算法的电磁轴承转子系统不平衡控制方法
CN120507992A (zh) * 2025-07-18 2025-08-19 中北大学 一种基于efdt-eso的显模型跟踪飞行控制方法
CN120729123A (zh) * 2025-08-28 2025-09-30 华能国际电力股份有限公司上海石洞口第二电厂 无人值守螺旋卸船机的电流控制方法
CN120928776A (zh) * 2025-10-16 2025-11-11 江苏美弛智造科技有限公司 一种非标自动化检具的检测精度动态校准控制方法
CN121356340A (zh) * 2025-12-16 2026-01-16 江西水利电力大学 基于线性扩张状态观测器和mpc的降压变换器控制方法

Also Published As

Publication number Publication date
CN116557421B (zh) 2025-11-07
CN116557421A (zh) 2023-08-08

Similar Documents

Publication Publication Date Title
WO2024259754A1 (zh) 六极主动磁轴承支承的高速电主轴控制系统的构造方法
CN109245449B (zh) 一种轴向分相磁悬浮开关磁阻飞轮电机的优化设计方法
CN113472226B (zh) 基于滑模观测器的并网逆变器双矢量无模型预测控制方法
CN111692210B (zh) 一种基于查表法的主动磁轴承控制器的构造方法
CN112422007B (zh) 混合磁轴承最小二乘支持向量机优化控制系统的构造方法
CN110131312B (zh) 五自由度交流主动磁轴承自抗扰解耦控制器及构造方法
CN115903497A (zh) 五自由度主动磁悬浮电主轴遗传粒子群优化解耦控制系统
CN112532134A (zh) 五自由度磁悬浮电主轴最小二乘支持向量机优化控制系统
CN110705174A (zh) 一种五自由度磁悬浮电主轴转子位移自检测系统及方法
CN106612089B (zh) 五自由度无轴承永磁同步电机强化学习控制器及构造方法
CN110048657B (zh) 无轴承异步电机神经网络逆系统解耦控制方法
Wang et al. Particle swarm optimization of fuzzy PI control for PMSMs
CN113054881B (zh) 外转子定子无铁心无轴承永磁同步电动机控制器构造方法
CN214707585U (zh) 一种混合磁轴承最小二乘支持向量机优化解耦控制器
CN102510253B (zh) 一种无轴承同步磁阻电机无传感器控制器及其控制方法
CN102097986A (zh) 无轴承同步磁阻电机神经网络广义逆解耦控制器构造方法
CN116164039A (zh) 径向六极混合磁轴承无位移传感器控制系统的构造方法
JP7267642B2 (ja) 参照テーブル法に基づくアクティブ磁気軸受コントローラーの構築方法
CN109039190B (zh) 一种四自由度磁悬浮电机悬浮系统解耦控制方法
Wang et al. Robust tracking control for permanent magnet linear servo system using intelligent fractional-order backstepping control
CN110425220A (zh) 一种径向磁轴承改进抗扰动控制器的构造方法
CN109672380B (zh) 五自由度无轴承永磁同步电机悬浮力子系统解耦控制器
CN109600083A (zh) 二自由度无轴承永磁同步电机悬浮力子系统解耦控制器
Schroder et al. Multi-objective optimisation of distributed active magnetic bearing controllers
Ganeshaperumal et al. Optimal fractional controller design methodology for electric drive train

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23942022

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE