Construction method of high-speed motorized spindle control system supported by hexapole active magnetic bearing

Abstract

The application discloses a construction method of a high-speed motorized spindle control system supported by a six-pole active magnetic bearing, a transfer learning model is constructed by an edge distribution distance of a source domain data set and a target domain data set, a predicted disturbance function is obtained by training, the predicted disturbance function is built into the transfer learning module, actual displacement is taken as input of the transfer learning module to obtain a predicted disturbance, an optimized active disturbance rejection controller is composed of a tracking differentiator, a nonlinear state error feedback control law, a first compensation factor, a second compensation factor, the transfer learning module and an extended state observer; three optimized active disturbance rejection controllers are connected in series in front of a three-degree-of-freedom composite controlled object, and two optimized active disturbance rejection controllers are connected in series in front of a two-degree-of-freedom composite controlled object to jointly constitute the control system; the active disturbance rejection control based on the transfer learning optimization is used to predict the disturbance, the optimized extended state observer is used to estimate and compensate the total disturbance of the system, and the control performance of the system is improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of electrical and mechanical transmission equipment, and particularly relates to a self-disturbance decoupling control system of a high-speed motorized spindle supported by six-pole active magnetic bearings. BACKGROUND

[0002] The high-speed motorized 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 is a new type of high-performance bearing that suspends the rotor in space by using magnetic field force, realizes no mechanical contact between the rotor and the stator, and solves the problems of thermal deformation, short service life and poor shock resistance of the traditional bearing. The radial magnetic bearing adopts a six-pole symmetrical stator structure, the stator of the six-pole magnetic bearing has a symmetrical structure, the magnetic circuit coupling between the two radial degrees of freedom is small, the bearing capacity is large, and the performance is excellent. The radial control coil is wound on the magnetic pole of the radial magnetic bearing, and is driven by a three-phase AC inverter. After electrification, a radial control magnetic flux is formed to generate suspension force and realize stable suspension in the radial direction. Since the radial magnetic bearings are driven by inverters, the radial magnetic flux inevitably causes coupling, which affects the precise control of the motorized spindle, so a high-efficiency decoupling controller is needed to realize the stable operation of the magnetic bearing at high speed and high precision.

[0003] At present, the methods for solving the decoupling control of the magnetic suspension motorized spindle mainly include the approximate linearization active disturbance rejection decoupling control method, the matrix converter control method, the inverse system decoupling control method and the comprehensive application of other intelligent algorithms. The approximate linearization active disturbance rejection decoupling control is a new decoupling control idea, which regards the mutual coupling between the radial magnetic circuits of the radial magnetic bearings in the five-degree-of-freedom magnetic suspension motorized spindle as internal disturbance of the system, uses an extended state observer to estimate and compensate the internal disturbance, realizes accurate decoupling control, and makes each degree of freedom of the magnetic bearing linear. However, the performance of the traditional approximate linearization active disturbance rejection controller depends on the selection of internal parameters. When the working condition of the high-speed motorized spindle supported by six-pole active magnetic bearings changes, the active disturbance rejection controller cannot accurately compensate the disturbance.

[0004] The construction method provided in the document with the Chinese patent publication number CN103076740B and the title "Construction method of AC magnetic suspension electric spindle controller" selects static experimental data offline to train the inverse system of the control object by using a fuzzy neural network, but adopts fuzzy control, and the control precision is low, which cannot meet the high precision requirement of the electric spindle. The control system provided in the document with the Chinese patent publication number CN112532134A and the title "Five-degree-of-freedom magnetic suspension electric spindle least squares support vector machine optimization control system" selects a tracking signal and a differential signal of a displacement as training samples, predicts and compensates the disturbance of the electric spindle, but the selection of parameters in the least squares support vector machine will lead to slow solving speed, and a large amount of training data is required for each operation, which affects the system control precision. SUMMARY

[0005] The purpose of the present application is to solve the problems of dependence on internal parameter selection, inaccurate disturbance compensation, the need for a large amount of training data for operation, slow solving speed, and poor control precision in the control of the existing five-degree-of-freedom six-pole active magnetic bearing supported high-speed electric spindle, and to provide a construction method of a six-pole active magnetic bearing supported high-speed electric spindle control system that can solve the above problems.

[0006] To achieve the above purpose, the construction method of the six-pole active magnetic bearing supported high-speed electric spindle control system adopts the technical scheme of:

[0007] The high-speed electric spindle has a first six-pole radial active magnetic bearing and an axial magnetic bearing, characterized by comprising the following steps:

[0008] Step 1): constructing a three-degree-of-freedom composite controlled object containing a first six-pole radial active magnetic bearing and an axial magnetic bearing, and a two-degree-of-freedom composite controlled object containing a second six-pole radial active magnetic bearing, the inputs of the three-degree-of-freedom composite controlled object and the two-degree-of-freedom composite controlled object are corresponding four radial control current expected values and one axial control current expected value;

[0009] Step 2): the high-speed electric spindle runs under rated load, collects multiple groups of displacement signals and corresponding disturbance signals as source domain data set, changes the size of the load within the rated load range and collects multiple groups of displacement signals of the high-speed electric spindle as target domain data; calculate the edge distribution distance of the source domain data set and the target domain data set, build a transfer learning model based on the edge distribution distance and train the transfer learning model to obtain a predicted disturbance function, and embed the predicted disturbance function in the transfer learning module (504);

[0010] Step 3): taking the actual radial or axial displacement as the input of the migration learning module (504), obtaining a predicted disturbance according to the predicted disturbance function, and adding the predicted disturbance to the radial or axial control current expected value after the radial or axial control current expected value is compensated by a second compensation factor (506) to obtain a second control quantity;

[0011] Taking the second control quantity and the actual radial or axial displacement as the input of the extended state observer (502), the extended state observer (502) outputs a tracking signal, a first-order differential signal of the tracking signal, and an observed disturbance;

[0012] Adding the predicted disturbance and the observed disturbance, and then compensating by a first compensation factor (505) to obtain a first control quantity;

[0013] A given radial or axial displacement outputs a tracking signal and a differential signal through a tracking differentiator (501), the tracking signal output by the tracking differentiator (501) is subtracted from the tracking signal output by the extended state observer (502), and the differential signal output by the tracking differentiator (501) is subtracted from the first-order differential signal output by the extended state observer (502), the two errors obtained are taken as the input of a nonlinear state error feedback control law (503), and the nonlinear state error feedback control law (503) outputs a feedback control quantity;

[0014] The feedback control quantity is subtracted from the first control quantity to obtain a compensated control quantity as the radial or axial control current expected value;

[0015] The tracking differentiator (501), the nonlinear state error feedback control law (503), the first compensation factor (505), the second compensation factor (506), the migration learning module (504), and the extended state observer (502) jointly constitute an optimal active disturbance rejection controller;

[0016] Step 4): three optimal active disturbance rejection controllers are connected in series in front of the three-degree-of-freedom composite controlled object, and two optimal active disturbance rejection controllers are connected in series in front of the two-degree-of-freedom composite controlled object, to jointly constitute a high-speed motorized spindle control system.

[0017] The technical scheme of the present application has the following advantages:

[0018] (1) This invention employs active disturbance rejection control based on transfer learning optimization, optimizes the performance of the extended state observer, predicts disturbances in the electric spindle, and uses the optimized extended state observer to estimate and compensate for the total system disturbance, thereby achieving 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 a strongly coupled, nonlinear system with multiple inputs and multiple outputs under non-ideal conditions into an uncoupled linear system with multiple inputs and multiple outputs, improving the system control performance.

[0019] (2) This invention employs a transfer learning algorithm to obtain predicted disturbances. It uses displacement and disturbance signals collected from the disturbance rejection controller as the source domain dataset, reducing data requirements. After load replacement, displacement signals and corresponding disturbance signals are collected as the target domain dataset. A transfer learning model is constructed based on the edge distance distribution, and the source domain data is used to train and test the transfer learning model. Since transfer learning has already acquired the characteristics of the source domain dataset, it avoids selecting too many internal parameters, reducing the need for a large amount of training data. Therefore, it improves the computational speed of the control system, giving the system better performance.

[0020] (3) In the process of predicting the output of disturbances, 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.

[0021] (4) In a high-speed electric spindle supported by a six-pole active magnetic bearing, two radial magnetic bearings and one axial magnetic bearing are used to achieve stable levitation of the five degrees of freedom of the spindle shaft. The radial and axial magnetic circuits are independent of each other, reducing coupling. A three-phase inverter drives and controls the radial magnetic bearings. Five active disturbance rejection controllers are used to decouple the high-speed electric spindle, eliminating the need for a precise mathematical model. This approach features small overshoot, fast response, and strong anti-interference capability, significantly improving the control accuracy of the electric spindle control system. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the structure of a high-speed electric spindle supported by a six-pole active magnetic bearing;

[0023] Figure 2 yes Figure 1 A schematic diagram of the radial structure and magnetic circuit of the first six-pole radial active magnetic bearing in China;

[0024] Figure 3 yes Figure 1 A schematic diagram of the axial structure and magnetic circuit of the first six-pole radial active magnetic bearing in the image;

[0025] Figure 4 Based on Figure 1 The equivalent structural block diagram of the three-degree-of-freedom composite controlled object constructed by the first six-pole active radial magnetic bearing and axial magnetic bearing;

[0026] Figure 5 is the equivalent structure block diagram of the second six-pole active radial magnetic bearing in Figure 1 ;

[0027] Figure 6 is the structure block diagram of the high-speed motorized spindle control system supported by the constructed six-pole active magnetic bearing;

[0028] Figure 7 is the control structure block diagram of the first optimized active disturbance rejection controller in Figure 6 ;

[0029] Figure 8 is the control structure block diagram of the traditional active disturbance rejection controller shown in Figure 7 ;

[0030] Figure 9 is the construction method flow chart of the transfer learning module in Figure 7 ;

[0031] Figure: 1. First six-pole radial active magnetic bearing; 2. Axial magnetic bearing; 3. Second six-pole radial active magnetic bearing; 4. High-speed spindle motor; 5. Shaft; 6, 7, 8, 9. Radial displacement sensor; 10. Axial displacement sensor; 11. Auxiliary bearing; 12, 13. End cover; 14. Steel cylinder outer sleeve; 15. Steel cylinder inner sleeve; 16. High-speed spindle motor stator; 17. High-speed spindle motor rotor; 20. First optimized active disturbance rejection controller; 21. Second optimized active disturbance rejection controller; 22. Third optimized active disturbance rejection controller; 23. Fourth optimized active disturbance rejection controller; 24. Fifth optimized active disturbance rejection controller; 25. First displacement sensor; 26. Second displacement sensor; 27. Third displacement sensor; 28. Fourth displacement sensor; 29. Fifth displacement sensor; 31. First Clark inverse transformation; 32 Second Clark inverse transformation; 33. First current tracking type inverter; 34. Second current tracking type inverter; 35. Axial power amplifier;

[0032] 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;

[0033] 201. 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;

[0034] 301. Second radial rotor; 302. Second radial rotor; 303. Second radial control coil;

[0035] 501. 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. DETAILED DESCRIPTION

[0036] Referring to Figure 1 The basic structure of the six-pole magnetic bearing supported electric spindle is shown in the figure, which 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. Among them, the high-speed spindle motor 4 includes a high-speed spindle motor stator 16 and a high-speed spindle motor rotor 17, and the high-speed spindle motor rotor 17 is coaxially fixedly sleeved outside the rotating shaft 5 to drive the rotating shaft 5 to rotate. The left side of the high-speed spindle motor 4 is the second six-pole radial active magnetic bearing 3, and the right side is the axial magnetic bearing 2 and the first six-pole radial active magnetic bearing 1, and the first six-pole radial active magnetic bearing 1 and the second six-pole radial active magnetic bearing 3 are completely the same in structure. 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 are collectively wrapped by a steel cylinder, which 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. The left and right end faces of the steel cylinder outer sleeve 14 and the steel cylinder inner sleeve 15 are respectively fixedly connected with one end cover 12, 13 respectively, and the two end covers 12, 13 are supported outside the rotating shaft 5 through an auxiliary bearing 11 respectively, and the auxiliary bearing 11 is used to support the rotating shaft 5 in the magnetic bearing shutdown or fault state. The rotating shaft 5 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 coaxially.

[0037] 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 respectively. The third radial displacement sensor 8 and the fourth radial displacement sensor 9 are installed between the left end cover 13 and the second six-pole radial active magnetic bearing 3, and the third radial displacement sensor 8 and the fourth radial displacement sensor 9 are installed symmetrically face to face in the radial direction to detect the radial displacement of the rotating shaft 5 in two degrees of freedom.

[0038] 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. In the axial direction, two annular limiting sleeves are respectively arranged 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, for fixing 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. The first radial displacement sensor 6 and the second radial displacement sensor 7 are arranged between the right end cover 12 and the first six-pole radial active magnetic bearing 1, and the first radial displacement sensor 6 and the second radial displacement sensor 7 are symmetrically installed face to face in the radial direction, for detecting the displacement of the two radial degrees of freedom of the right side of the rotating shaft 5. The axial displacement sensor 10 is installed on the right end cover 12, for detecting the displacement of the axial degree of freedom of the rotating shaft 5.

[0039] The probes of the first, second, third, fourth radial displacement sensors 6, 7, 8, 9 and the axial displacement sensor 10 are all eddy current sensors.

[0040] Referring to FIGS. 1 and 2, Figure 2 and Figure 3 The first six-pole radial active magnetic bearing 1 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 and integrated 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 coaxially sleeves outside the first radial rotor 101, and a radial air gap 107 exists between the first radial stator 102 and the first radial rotor 101. The first radial stator 102 is composed of a circular ring stator yoke and six first radial stator magnetic poles 104 in convex shape uniformly arranged along the inner wall circumference of the stator yoke. The side surface of the first radial stator magnetic pole 104 is a cylindrical surface. The first radial control coil 103 with the same direction is wound on each first radial stator magnetic pole 104. The first radial control coils 103 on the opposite poles are connected in series to form a phase, and three-phase coils are formed. The three-phase coils are connected in star connection mode. A three-phase inverter is used to control the first radial control coil 103, so as to control the size and direction of the suspension force.

[0041] The first radial control coil 103 provides bias magnetic flux 105 and control magnetic flux 106 for the first six-pole radial active magnetic bearing 1. The loop of the bias magnetic flux 105 is: from a first radial stator magnetic pole 104, through a radial air gap 107 into the first radial rotor 101, through the opposite radial air gap 107 into the opposite first radial stator magnetic pole 104, and finally through the stator yoke back to the first radial stator magnetic pole 104. The main function of the loop of the bias magnetic flux 105 is to balance the weight of the rotor by the suspension force generated by the bias magnetic motive force. The loop of the control magnetic flux 106 is: from a first radial stator magnetic pole 104, through a radial air gap 107 into the first radial rotor 101, and through the opposite radial air gap 107 into the opposite first radial stator magnetic pole 104. The bias magnetic flux 105 and the control magnetic flux 106 interact to generate radial two-degree-of-freedom suspension force on the first radial rotor 101 of the first six-pole radial active magnetic bearing 1.

[0042] Referring to Figure 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. The structure and assembly method of the second six-pole radial active magnetic bearing 3 are completely 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. Similarly, the bias magnetic flux and the control magnetic flux of the second six-pole radial active magnetic bearing interact to generate radial two-degree-of-freedom suspension force on the second radial rotor 301 of the second six-pole radial active magnetic bearing 3.

[0043] 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 fixed on the shaft 5 and integrated with the shaft 5. The axial control coil 203 is wound around the left and right magnetic poles of the axial stator 202 respectively. The axial control coil 203 is energized by a bipolar DC power amplifier to generate axial control magnetic flux and axial bias magnetic flux. The axial control magnetic flux and the axial bias magnetic flux interact to generate axial suspension force on the axial rotor 202 of the axial magnetic bearing 2.

[0044] 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 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 copper wires with insulating paint with a diameter of 0.67 mm. The auxiliary bearing 11 is a self-aligning ball bearing.

[0045] As Figure 4As shown, based on the first six-pole radial active magnetic bearing 1 and the axial magnetic bearing 2, a three-degree-of-freedom composite controlled object 210 containing the first six-pole radial active magnetic bearing 1 and the axial magnetic bearing 2 is configured, which is composed of the first Clark inverse transformation 31, the first current tracking type inverter 33, the first six-pole radial active magnetic bearing 1 connected in series, and the axial power amplifier 35, the axial magnetic bearing 2 connected in series, and the three-degree-of-freedom composite controlled object 210, the input of which is 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 output of the three-degree-of-freedom composite controlled object 210 is the actual radial displacement x a , y a and the actual axial displacement z a . Wherein, the first radial control current expected value i ax * and the second radial control current expected value i ay * are obtained through the first Clark inverse transformation 31 to obtain the current i au * , i av * and i aw * , the current i au * , i av * and i aw * is output through the first current tracking type inverter 33 to output the radial control current i au , i av and i aw . The radial control current i au , i av , i aw controls the first radial control coil 103 in the first six-pole radial active magnetic bearing 1, and the first six-pole radial active magnetic bearing 1 generates the actual radial displacement x a , y a . The first and second radial displacement sensors 6, 7 respectively detect the actual radial displacement x a , y a . The axial control current expected value i az * is output through the axial power amplifier 35 to output the axial control current i az , and the axial control current i az controls the axial control coil 203 in the axial magnetic bearing 2, and the axial magnetic bearing 2 generates the actual axial displacement za The axial displacement sensor 10 detected the actual axial displacement z. a The controlled object 210 is constructed as a three-degree-of-freedom composite controlled object, which includes a first six-pole radial active magnetic bearing 1 and an axial magnetic bearing 2.

[0046] like Figure 5 As shown, a two-degree-of-freedom composite controlled object 211 is constructed based on the second six-pole radial active magnetic bearing 3. The two-degree-of-freedom composite controlled object 211 is composed of the second Clark inverse transformer 32, the second current-tracking inverter 34, and the second six-pole radial active magnetic bearing 3 connected in series. The input to the two-degree-of-freedom composite controlled object 211 is the desired values ​​i of the third and fourth radial control currents. bx * i by * The output of the two-degree-of-freedom composite controlled object 211 is the actual radial displacement x. b y b Among them, the expected values ​​of the third and fourth radial control currents, i bx * i by * The common input is the second Clark inverse transformer 32, and the output current i of the second Clark inverse transformer 32 is... bu * i bv * and i bw * Current i bu * i bv * and i bw * The radial control current i is output through the second current tracking inverter 34. bu i bv i bw The second six-pole radial active magnetic bearing 3 controls the second radial control coil 302 within it, and the second six-pole radial active magnetic bearing 3 outputs the actual radial displacement x. b y b This constructs a two-degree-of-freedom composite controlled object 211, which includes a second six-pole radial active magnetic bearing 3. The third and fourth radial displacement sensors 8 and 9 detect the actual radial displacement x. b y b .

[0047] like Figure 6 As shown, due to the need for Figure 1The shown six-pole active magnetic bearing support controls five degrees of freedom of the high-speed motorized spindle, and therefore five optimal active disturbance rejection controllers are used to constitute the control system, and the five optimal active disturbance rejection controllers control the five degrees of freedom respectively. The five optimal active disturbance rejection controllers are first, second, third, fourth and fifth optimal active disturbance rejection controllers 20, 21, 22, 23 and 24 respectively. The five optimal active disturbance rejection controllers are only different in input and output signals, and the internal structures are completely the same.

[0048] The first and second radial displacement sensors 6 and 7 collect actual radial displacements x a , y a of the three-degree-of-freedom composite controlled object 210, and the actual radial displacements x a , y a are input into the corresponding first and second optimal active disturbance rejection controllers 20 and 21 respectively. The axial displacement sensor 10 collects an actual axial displacement z a of the three-degree-of-freedom composite controlled object 210, and the actual axial displacement z a is input into the third optimal active disturbance rejection controller 22. At the same time, corresponding given radial displacements x a * are input into the first optimal active disturbance rejection controller 20, and the first optimal active disturbance rejection controller 20 processes the input actual radial displacements x a and the given radial displacements x a * , and outputs first radial control current expected values i ax * to the three-degree-of-freedom composite controlled object 210. Similarly, corresponding given radial displacements y a * are input into the second optimal active disturbance rejection controller 21, and the second optimal active disturbance rejection controller 21 outputs second radial control current expected values i ay * to the three-degree-of-freedom composite controlled object 210; given axial displacements z a * are input into the third optimal active disturbance rejection controller 22, and the third optimal active disturbance rejection controller 22 outputs axial control current expected values i az * to the three-degree-of-freedom composite controlled object 210. Thus, the first radial control current expected values i ax * , the second radial control current expected values i ay * and the axial control current expected values i az * are collectively used as inputs of the three-degree-of-freedom composite controlled object 210.

[0049] The third and fourth radial displacement sensors 8 and 9 collect the actual radial displacement x of the two-degree-of-freedom composite controlled object 211. b y b Actual radial displacement x b y b The corresponding values ​​are input one-to-one into the corresponding fourth optimized active disturbance rejection controller 23 and fifth optimized active disturbance rejection controller 24, respectively, and the corresponding given radial displacement x is applied. b * The input is fed to the fourth optimized active disturbance rejection controller 23, and the fourth optimized active disturbance rejection controller 23 responds to the actual radial displacement x input. b and a given radial displacement x b * Process and output the desired value i of the third radial control current. bx * To the two-degree-of-freedom composite controlled object 211. The corresponding given radial displacement y... b * The input is fed into the fifth optimized active disturbance rejection controller 24, and the fifth optimized active disturbance rejection controller 24 outputs the fourth radial control current expected value i. by * In the two-degree-of-freedom composite controlled object 211, the desired value of the third radial control current i is... bx * and the expected value of the fourth radial control current i by * Together they serve as the input to the two-degree-of-freedom composite controlled object 211.

[0050] like Figure 7 As shown, the construction process of the optimized active disturbance rejection controller is described in detail below, taking the first optimized active disturbance rejection controller 20 as an example:

[0051] The first optimized active disturbance rejection controller 20 consists 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.

[0052] Given radial displacement x a * The given radial displacement x is output by the tracking differentiator 501. a * Tracking signal v a1x and a given radial displacement x a * The differential signal v a2x The mathematical model for the tracking differentiator 501 is as follows:

[0053]

[0054] where sat is a nonlinear function, and the expression is where T is a fast tracking factor, and its value range is [0, 1], the value of T determines the tracking performance of the system; ε is the integral step; h is the system sampling period; is the given displacement is the value at time k; v a1x (k) and v a1x (k+1) are the given radial displacement tracking signals v a1x is the value at time k and k+1; v a2x (k) and v a2x (k+1) are the given radial displacement differential signals v a2x is the value at time k and k+1.

[0055] is the actual radial displacement x a is the tracking signal z a1x output by the extended state observer 502, the first order differential signal z a1x of the tracking signal z a2x output by the extended state observer 502, and the observation disturbance z a3x of the system.

[0056] The tracking signal v a1x output by the tracking differentiator 501 is subtracted from the tracking signal z a1x output by the extended state observer 502 to obtain the first system state error e a1x =v a1x -z a1x . The differential signal v a2x output by the tracking differentiator 501 is subtracted from the first order differential signal z a2x output by the extended state observer 502 to obtain the second system state error e a2x =v a2x -z a2x . The two system state errors e a1x and e a2x are jointly used as the input of the nonlinear state error feedback control law 503, and the nonlinear state error feedback control law 503 outputs the feedback control amount u a0x . The model of the nonlinear state error feedback control law is:

[0057] u a0x =k1fal(e a1x ,α4,δ2)+k2fal(e a2x ,α5,δ2),

[0058] In the formula: k1, k2, α4, α5, and δ2 are adjustable parameters of the nonlinear feedback control law. Generally, α4 is taken as 0.5; α5 is taken as 0.25; the ranges of k1 and k2 are [0, 100], and the range of δ2 is [0, 1]. The nonlinear function... Where e is a natural constant, α is a constant between 0 and 1, and δ is a custom constant that affects the filtering effect.

[0059] Feedback control quantity u a0x After compensation by the first compensation factor of 505, the compensated control quantity u is obtained. ax :

[0060]

[0061] In the formula: 1 / d is the first compensation factor; f ax For the predicted perturbation output by the transfer learning module 504, considering comprehensive factors, the value range of d is [10, 200].

[0062] Compensated control quantity u ax It is also used as input to the second compensation factor 506, which is d. After compensation by the second compensation factor 506, the compensated control quantity du is obtained. ax Then control quantity du ax With predicted disturbance f ax Adding them together, we get the second control quantity (f) ax +du ax The second control quantity (f) after summing ax +du ax This is used as the second input to the extended state observer 502. The discrete mathematical model of the extended state observer 502 is as follows:

[0063]

[0064] In the formula: fal is a nonlinear function, and its expression is: Where e is a natural constant, α is a constant between 0 and 1, and δ is a user-defined constant that affects the filtering effect; α1, a2, α3, β1, β2, β3, and δ1 are 502 adjustable parameters of the extended state observer; z a1x (k+1), z a2x (k+1) and z a3x (k+1) represent z a1x z a2x and z a3x The value at time k+1; h is the sampling period; e a1x e is the first system state error; a2x The second system state error; x a (k) represents the actual radial (axial) displacement at time k, uax (k) represents the compensated control quantity u at time k. ax ;f ax (k) represents the predicted perturbation f output by the transfer learning module (504) at time k. ax α1 is usually taken as 0.75, α2 as 0.5, a3 as 0.25, δ1 is taken as 5 to 10 times the sampling period, and β1, β2, and β3 need to be continuously adjusted according to the tracking effect required by the system.

[0065] The actual radial displacement x collected by the first radial displacement sensor 6 a As the first input to the extended state observer 502, the second control quantity (f) ax +du ax As the second input to the extended state observer 502, the extended state observer 502 outputs the actual radial displacement x. a Tracking signal z a1x Tracking signal z a1x The first-order differential signal z a2x and the system's observed perturbation z a3x .

[0066] The actual radial displacement x a Source domain dataset D s and the target domain dataset D t As input to the transfer learning module 504, the transfer learning module 504 outputs the predicted perturbation f of the system. ax The predicted disturbance f ax The observed perturbation z output by the extended state observer 502 a3x Add them together to get the sum of the perturbations (z). a3x +f ax As the input to the first compensation factor 505, the first control quantity is obtained after the first compensation factor 505 is applied. The feedback control quantity u output by the nonlinear state error feedback control law 503 a0x With the first control quantity By subtracting the values, we obtain the compensated control quantity u. ax The compensated control quantity u ax That is, the expected value of the radial control current i ax * u ax =i ax * As the output of the first optimized active disturbance rejection controller 20.

[0067] The construction methods of the second, third, fourth, and fifth optimized active disturbance rejection controllers 20, 21, 22, 23, and 24 are exactly the same as those of the first optimized active disturbance rejection controller 20. They output the corresponding expected value of control current for different given radial or axial displacements. Therefore, the other four optimized active disturbance rejection controllers will not be described in detail.

[0068] See Figure 7 , Figure 8 and Figure 9 The method for constructing the transfer learning module 504 is as follows:

[0069] Step S1: Data Collection

[0070] Will Figure 1 The high-speed electric spindle supported by the six-pole active magnetic bearing shown is connected to a conventional active disturbance rejection controller. This conventional active disturbance rejection controller is a controller capable of detecting disturbance signals from the electric spindle, for example... Figure 9 The active disturbance rejection controller (ADRC) optimized by the support vector machine (using the ADRC with patent publication number CN112532134A), or other controllers that detect electrical spindle disturbance signals, includes 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. Figure 7 Compared to the first optimized active disturbance rejection controller 20 of the present invention, the only difference is the disturbance prediction output module; the functions and roles of the rest are the same.

[0071] exist Figure 9 The traditional support vector machine optimized active disturbance rejection controller shown enables... Figure 1 The high-speed electric spindle supported by the six-pole active magnetic bearing shown is operating under rated load, and multiple sets of displacement signals x are collected from its output. i and its corresponding disturbance signal y i For displacement signal x i and disturbance signal y i Preprocessing is performed to form the source domain dataset D. s ={(x1,y1),...,(x n ,y n The present invention collects n sets of displacement signals x. i The disturbance signal, n, is the number of source domain data samples, 1≤i≤n. Within the rated load range, the load size is changed, and multiple sets of displacement signals z from the high-speed electric spindle are then acquired. j and the displacement signal z j Preprocessing is performed to form the target domain data D. t ={z1,...,z m}, 1≤j≤m, where m is the acquired displacement signal zj The number is also the number of data samples in the target domain.

[0072] Step S2: Calculation of edge distribution distance

[0073] The key to transfer learning lies in quantifying the data differences between the source and target domains to ensure its effectiveness. This is achieved by mapping the source and target domain datasets D using the mapping function θ(x). s With target domain dataset D t Mapped to the Hilbert space of the reproducing kernel, the maximum mean discrepancy (MMD) is used to characterize the marginal distribution distance D between the source and target domains. H Source domain dataset D s With target domain dataset D t Edge distribution distance D H The calculation is performed using the following formula:

[0074]

[0075] In the formula: H represents the regenerating kernel Hilbert space, x i The displacement signal in the source domain dataset; z j θ(x) represents 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; θ(x) represents the displacement signal in the target domain dataset. i ), θ(z) j Let θ(x) be the mapping function 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 θ(x) is defined as the softmax function, and the perturbation signal y in the source domain dataset is... i This mapping function can be used to transform the original data in the reproducing kernel Hilbert space, thus transforming the perturbation signal y in the source domain dataset. i for:

[0076]

[0077] e is a natural constant.

[0078] Step S3: Based on edge distribution distance D H Building a transfer learning model:

[0079]

[0080] In the formula: f(y) i The expression ) 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, and here L is chosen as the cross-entropy loss function. λ is a correction coefficient, which is selected in the range of [0, 1], and the optimal value in this invention is 0.5.

[0081] Step S4: Train the transfer learning model f(y) i )

[0082] Step S4.1: To train the transfer learning model, the Deep Learning Toolbox is invoked within the MATLAB software environment. From the source domain dataset D... s We randomly select ω groups from the sample, where ω ≤ n, and use half of the ω groups as the training sample set. The other half serves as the test sample set. Considering training accuracy and training time, ω ranges from [100, 600]. The displacement signals x from the ω / 2 training sample set... i As a transfer learning model f(y) i The input of the disturbance signal y is... i As a transfer learning model f(y) i The output of ).

[0083] Step S4.2: Use the training module to train the ω / 2 sets of training samples D train Training is performed. The ω / 2 training sample sets are used as the transfer learning model f(y). i The input and output signals are fitted by software to output ω / 2 sets of mapping functions. The displacement signals in the ω / 2 training sample set are mapped to the reproducing kernel Hilbert space, and the ω / 2 mapping functions are respectively... Substitute edge distribution distance The calculation formula yields the edge distribution distance of group ω / 2.

[0084] Step S4.3: Distribute the edge distribution distance of ω / 2 groups respectively. Substitute into the transfer learning model f(y) i ), thus obtaining ω / 2 sets of transfer learning models f(y) i The corresponding value of ) is, Select the mapping function θ corresponding to the minimum value among the corresponding values ​​of the ω / 2 group of transfer learning models. min (x), the mapping function θ min (x) is the prediction disturbance function f ax (x), training complete.

[0085] Step S5: Test the predicted perturbation function f ax (x)

[0086] Test sample set D test Displacement signal [x] test1 ,x test2 ,…,x testω / 2 Input to the prediction perturbation function f after training is complete ax(x), i.e. the predicted disturbance output set D out = [y out1 , y out2 ,..., y outω / 2 ] is obtained.

[0087] For the test sample set , all the disturbance signals are taken out to form a disturbance signal sample set [y test1 , y test2 ,..., y testω / 2 ], and the mean square error of the disturbance signal sample set [y test1 , y test2 ,..., y testω / 2 ] and the predicted disturbance output set D out = [y out1 , y out2 ,..., y outω / 2 ] is calculated.

[0088]

[0089] wherein y outp , y testp represent the pth disturbance signal in the predicted output set D out and the test sample set D test , respectively, and 0 < p < ω / 2.

[0090] According to the actual accuracy requirement, the mean square error threshold γ is set, and the range of γ set by the present application is [10 -5 , 10 -3 ]. The mean square error Δ is compared with the set error threshold γ, and if the mean square error Δ is less than the error threshold γ, it is considered that the predicted disturbance function f ax (x) obtained in step S4.3 meets the requirement; otherwise, if the mean square error Δ is greater than or equal to the error threshold γ, the correction coefficient λ needs to be adjusted again, and then the transfer learning model f(y i ) is reconstructed again according to the adjusted new correction coefficient λ, and steps S3-S5 are repeatedly executed, i.e. the transfer learning model f(y i ) is reconstructed again based on the adjusted new correction coefficient λ, and the newly constructed transfer learning model f(y i ) is trained and tested until the mean square error is less than γ, at which time the predicted disturbance function f ax (x) meeting the requirement is obtained.

[0091] Step S6: constructing an optimal active disturbance rejection controller

[0092] Referring to Figure 7 , the predicted disturbance function f ax(x) Built into the transfer learning module 504, the transfer learning module 504 is connected to the first optimized active disturbance rejection controller 20. The transfer learning module 504 reduces the burden on the extended state observer 502 and improves the stability of the overall control system.

[0093] Real-time displacement signal x of a high-speed electric spindle supported by a six-pole active magnetic bearing a As the input signal to the transfer learning module 504, the transfer learning module 504 outputs the predicted perturbation f. ax =f ax (x). Predict the disturbance f ax du output with the second compensation factor 506 ax Adding them together, we get the second control quantity (f) ax +du ax This serves as the second input to the extended state observer 502. The predicted perturbation f is then used as... ax The observed perturbation z output by the extended state observer 502 a3x Adding them together, we get the disturbance (f) ax +z a3x ), disturbance quantity (f ax +z a3x The first control quantity is obtained after compensation by the first compensation factor 505. The feedback control quantity u output by the nonlinear state error feedback control law 503 a0x With the first control quantity By taking the difference, we obtain the compensated control quantity. As the desired value of the radial current control, this constitutes an optimized active disturbance rejection controller.

[0094] The second, third, fourth, and fifth optimized active disturbance rejection controllers 21, 22, 23, and 24 are also each connected to a transfer learning module to predict disturbances, using a method similar to the first optimized active disturbance rejection controller 20, and therefore will not be described in detail. This results in five identical optimized active disturbance rejection controllers 20, 21, 22, 23, and 24. Three of these controllers, 20, 21, and 22, are connected in series at the front end of the three-degree-of-freedom composite controlled object 210, while the other two, 23 and 24, are connected in series at the front end of the two-degree-of-freedom composite controlled object 211, together forming a high-speed electric spindle control system.

[0095] According to the above, the high-speed motorized spindle supported by the sextupole active magnetic bearing can be stably suspended in five degrees of freedom of the rotating shaft by driving and controlling the first sextupole radial active magnetic bearing 1 and the second sextupole radial active magnetic bearing 3 with a three-phase inverter and driving the axial magnetic bearing 2 with a bipolar power amplifier. 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 motorized spindle rotating shaft. The five active disturbance rejection controllers are used to decouple the control of the high-speed motorized spindle supported by the sextupole active magnetic bearing, without the need for accurate mathematical models, and have the characteristics of small overshoot, fast response, strong anti-interference ability, etc., which can greatly improve the control accuracy of the motorized spindle system. The transfer learning algorithm is used to obtain the predicted disturbance, the displacement signal and the disturbance signal of the motorized spindle rotor under the rated load are collected as the source domain data set, the displacement signal and the disturbance signal are collected after the load is replaced 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 characteristics of the source data set, the demand for a large amount of data is reduced, the calculation speed of the control system is improved, and the system has better rapidity. The trained transfer learning module is connected to the active disturbance rejection controller, the source domain data set D s , the target domain data set D t and the real-time displacement signal of the high-speed motorized spindle supported by the sextupole active magnetic bearing are used as the input signal of the trained transfer learning model, and the predicted disturbance is used as the output of the transfer learning module, so that the control system can run.

Claims

1. A method of constructing a high speed motorized spindle control system with a six-pole active magnetic bearing support, said high speed motorized spindle having a first and a second six-pole radial active magnetic bearing and an axial magnetic bearing, characterized in that Comprising the following steps: Step 1): Construct a three-degree-of-freedom composite controlled object comprising a first six-pole radial active magnetic bearing and an axial magnetic bearing, and a two-degree-of-freedom composite controlled object comprising a second six-pole radial active magnetic bearing, the inputs of the three-degree-of-freedom composite controlled object and the two-degree-of-freedom composite controlled object are corresponding four radial control current expected values and one axial control current expected value; Step 2): The high-speed motorized spindle runs under rated load, collects multiple sets of displacement signals and corresponding disturbance signals to form a source domain data set, changes the load size within the rated load range and collects multiple sets of high-speed motorized spindle displacement signals to form a target domain data set; calculate the edge distribution distance of the source domain data set and the target domain data set, build a transfer learning model based on the edge distribution distance and train the transfer learning model to obtain a predicted disturbance function, and embed the predicted disturbance function in the transfer learning module (504); Step 3): Take the actual radial or axial displacement as the input of the transfer learning module (504), obtain the predicted disturbance according to the predicted disturbance function, and then add the predicted disturbance to the radial or axial control current expected value after the second compensation factor (506) compensation to obtain the second control quantity; Take the second control quantity and the actual radial or axial displacement as the input of the extended state observer (502), and the extended state observer (502) outputs the tracking signal, the first derivative signal of the tracking signal and the observed disturbance; Add the predicted disturbance to the observed disturbance, and then compensate it through the first compensation factor (505) to obtain the first control quantity; The given radial or axial displacement outputs its tracking signal and differential signal through the tracking differentiator (501), the tracking signal output by the tracking differentiator (501) is subtracted from the tracking signal output by the extended state observer (502), and the differential signal output by the tracking differentiator (501) is subtracted from the first derivative signal output by the extended state observer (502), the two errors obtained are used as the input of the nonlinear state error feedback control law (503), and the nonlinear state error feedback control law (503) outputs the feedback control quantity; The feedback control quantity is subtracted from the first control quantity to obtain the compensated control quantity 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 optimal active disturbance rejection controller; Step 4): Three optimal active disturbance rejection controllers are connected in series in front of the three-degree-of-freedom composite controlled object, and two optimal active disturbance rejection controllers are connected in series in front of the two-degree-of-freedom composite controlled object, which together constitute a high-speed motorized spindle control system.

2. The configuration method of a high-speed motorized spindle control system with a six-pole active magnetic bearing support according to claim 1, characterized in that: In step 2), the edge distribution distance H represents a reproducing kernel Hilbert space, x i , z j are displacement signals in the source domain data set and the target domain data set, respectively; n and m are the sample numbers of the source domain data set and the target domain data set, respectively; 1≤i≤n, 1≤j≤m; θ(x i ), θ(z j ) are mapping functions of the displacement signals in the source domain data set and the target domain data set in the reproducing kernel Hilbert space, respectively, and the initial function of the mapping function θ(x) is a softmax function, then the perturbation signal in the source domain data set is 3. The configuration method of a high-speed motorized spindle control system with a six-pole active magnetic bearing support according to claim 2, characterized in that: The migration learning model argmin represents taking the minimum value of the objective function; λ is a correction coefficient, and is [0, 1].

4. The configuration method of a high-speed motorized spindle control system with a six-pole active magnetic bearing support according to claim 3, characterized in that: First, sample ω groups from the source domain dataset, ω ≤ n, take half of the ω groups as the training sample set and the other half as the test sample set; then take ω / 2 groups of the training sample set as the input and output signals of the transfer learning model f(y i ), output ω / 2 groups of mapping functions, map the displacement signals in the ω / 2 groups of the training sample set to the reproducing kernel Hilbert space, and respectively substitute the ω / 2 groups of the mapping functions into the edge distribution distance D H , calculate to obtain ω / 2 groups of the edge distribution distance D Hi ; finally, respectively substitute the ω / 2 groups of the edge distribution distance into the transfer learning model f(y i ), obtain the corresponding values of the ω / 2 groups of the transfer learning model f(y i ), select the mapping function corresponding to the minimum value in the corresponding values, and the mapping function is the trained prediction disturbance function.

5. The method of constructing a high speed motorized spindle control system with a six-pole active magnetic bearing support according to claim 4, characterized in that: The displacement signal in the test sample set is input into the trained prediction disturbance function to obtain a predicted disturbance output set; all disturbance signals in the test sample set are taken out to form a disturbance signal sample set, and a mean square error of the disturbance signal sample set and the predicted disturbance output set is calculated; the mean square error is compared with a set error threshold value; if the mean square error is less than the error threshold value, the trained prediction disturbance function meets the requirements; otherwise, the correction coefficient λ in the transfer learning model f(y i ) is adjusted, and the transfer learning model f(y i ) is reconstructed again.

6. The method of constructing a six-pole active magnetic bearing supported high speed motorized spindle control system according to claim 1, characterized by: In step 3), the mathematical model of the tracking differentiator (501) is: Non-linear function T is a fast tracking factor, ε is an integration step; h is a sampling period; is a given displacement is a value at k time; v a1x (k) and v a1x (k+1) are given displacement tracking signals v a1x are values at k and k+1 time; v a2x (k) and v a2x (k+1) are given displacement differential signals v a2x are values at k and k+1 time.

7. The method of constructing a six-pole active magnetic bearing supported high speed motorized spindle control system according to claim 6, characterized by: In step 3), the feedback control amount u a0x = klfal(e a1x , a4, d2) + k2fal(e a2x , a5, d2), k1, k2, a4, a5 and d2 are adjustable parameters of the nonlinear feedback control law, a4 is 0.5; a5 is 0.25; the range of k1 and k2 is [0, 100], and the range of d2 is [0, 1], e is a natural constant, a is a constant between 0 and 1, and d is a constant affecting the filtering effect and defined by the user; e a1x = v a1x - z a1x ; e a2x = v a2x - z a2x ; z a1x , z a2x are the tracking signal and the first-order differential signal output by the extended state observer (502), respectively.

8. The method of constructing a six-pole active magnetic bearing supported high speed motorized spindle control system according to claim 7, characterized by: In step 3), the discrete mathematical model of the extended state observer (502) is: a1, a2, a3, b1, b2, b3, and d1 are tunable parameters; z a1x (k + 1), z a2x (k + 1), and z a3x (k + 1) are respectively z a1x , z a2x and the observed disturbance z a3x at time k + 1; x a (k) is the actual displacement at time k, u ax (k) is the compensated control at time k, u ax ; f ax (k) is the predicted disturbance f ax output by the transfer learning module (504) at time k.

9. The method of constructing a six-pole active magnetic bearing supported high speed motorized spindle control system according to claim 1, characterized by: In step 3), the compensated control quantity u a0x is a feedback control quantity, 1 / d is a first compensation factor, and takes a value of [10, 200]; z a3x is an observed disturbance output by the extended state observer (502); f ax is a predicted disturbance.

10. The method of constructing a six-pole active magnetic bearing supported high speed motorized spindle control system according to claim 1, characterized by: In step 1), the three-degree-of-freedom composite controlled object is composed of a first Clark inverse transformation, a first current tracking type inverter, the first six-pole radial active magnetic bearing, and an axial power amplifier and the axial magnetic bearing connected in series; and the two-degree-of-freedom composite controlled object is composed of a second Clark inverse transformation, a second current tracking type inverter, and the second six-pole radial active magnetic bearing connected in series.

Images