阅读说明：本技术 变步长变角度搜索遗传算法优化转子振动补偿系统及方法 (System and method for optimizing rotor vibration compensation through variable step length and variable angle search genetic algorithm ) 是由 朱熀秋 周锴 于 2021-02-04 设计创作，主要内容包括：本发明公开无轴承永磁同步电机的变步长变角度搜索遗传算法优化转子振动补偿系统及方法,转速作为改进的陷波器的输入,改进的陷波器输出的是转子x,y方向上与转子同频的振动信号的正弦系数和余弦系数,正弦和余弦系数输入到不平衡补偿模块中,不平衡补偿模块将振动总幅值与振动目标值作比较,若振动总幅值大于振动目标值,则对转子振动采用变步长变角度搜索方法进行补偿控制,当后一步振动比前一步振动小时,增加步长和角度,当后一步振动比前一步振动大时,角度变90°且步长变为初始搜索步长,以此不断改变搜索角度和方向得到辨识系数,计算出最优补偿电流；加快了转子振动补偿的反应速度,能在电机全转速运行情况下进行转子振动补偿。(The invention discloses a variable step length and variable angle search genetic algorithm optimized rotor vibration compensation system and method of a bearingless permanent magnet synchronous motor, wherein the rotating speed is used as the input of an improved wave trap, the improved wave trap outputs sine coefficients and cosine coefficients of vibration signals with the same frequency as the rotor in the x and y directions, the sine coefficients and the cosine coefficients are input into an unbalance compensation module, the unbalance compensation module compares the total vibration amplitude with a vibration target value, if the total vibration amplitude is larger than the vibration target value, then the rotor vibration is compensated and controlled by adopting a variable step length and variable angle searching method, when the vibration of the next step is smaller than that of the previous step, the step length and the angle are increased, when the vibration of the next step is larger than that of the previous step, the angle is changed to 90 DEG and the step size is changed to the initial search step size, the search angle and the search direction are continuously changed to obtain an identification coefficient, and the optimal compensation current is calculated; the reaction speed of the rotor vibration compensation is accelerated, and the rotor vibration compensation can be carried out under the condition that the motor runs at the full rotating speed.)

1. A variable step length and variable angle search genetic algorithm optimized rotor vibration compensation system comprises a torque control part and a suspension force control part, and is characterized in that a photoelectric encoder (7) is adopted to detect the rotating speed omega of a bearingless permanent magnet synchronous motor: the rotating speed omega is used as the input of an improved wave trap (8), the improved wave trap (8) outputs the sine coefficient a of the vibration signal with the same frequency as the rotor in the x and y directions of the rotor_AX，a_AYAnd cosine coefficient b_AX，b_AXThe output end of the improved wave trap (8) is connected with the input end of the unbalance compensation module (9) and the sine coefficient a_AX，a_AYAnd cosine coefficient b_AX，b_AXThe current is input to an unbalance compensation module (9), and the unbalance compensation module (9) outputs optimal compensation current i in the x direction and the y direction_c-x-BAnd i_c-y-BThe optimal solution compensates for the current i_c-x-BAnd i_c-y-BInput into the levitation force control portion.

2. The variable step size and variable angle search genetic algorithm optimized rotor vibration compensation system of claim 1, wherein: the improved wave trap (8) comprises a first sine multiplication module (101), a first sine integration module (111) and a second sine multiplication module (121) which are sequentially connected in series, and a first cosine multiplication module (102), a first cosine integration module (112) and a second cosine multiplication module (122) which are sequentially connected in series, wherein the output ends of the second sine multiplication module (121) and the second cosine multiplication module (122) are both connected with the input end of the unbalance compensation module (9), and the second sine multiplication module (121) outputs sine coefficient a_AX，a_AYThe cosine coefficient b is output by the cosine second multiplication module (122)_AX，b_AY。

3. The variable step size and variable angle search genetic algorithm optimized rotor vibration compensation system of claim 2, wherein: the sine coefficient a_AX，a_AYAnd cosine coefficient b_AX，b_AYThe method comprises the following steps:

4. the variable step size and variable angle search genetic algorithm optimized rotor vibration compensation system of claim 1, wherein: the suspension control module (5) outputs three-phase suspension control current i_Ba，i_Bb，i_BcControlling the suspension winding of a bearingless permanent-magnet synchronous motor, three-phase suspension control current i_Ba，i_Bb，i_BcConverted into a suspension force feedback current i by a CLARK/PARK conversion module (62)_Bd，i_BqThe actual displacement x and y of the bearingless permanent magnet synchronous motor are detected by using a displacement sensor, and the reference suspension force F in the x and y directions is obtained by passing a displacement difference obtained by subtracting the actual displacement x and y and a set reference displacement x, y through PID modules (21 and 22)_x ^*，F_y ^*Reference levitation force F_x ^*，F_y ^*The suspension force reference current i in the x and y directions is obtained by the force/current conversion module (3) after being input into the force/current conversion module (3)_Bd ^*，i_Bq ^*According to formula (I)Obtaining the input current i of the levitation force_c-Bd，i_c-BqInputting a levitation force into a current i_c-Bd，i_c-BqAs input to the levitation control module (5).

5. A method for compensating rotor vibration in a system for optimizing rotor vibration compensation using a variable step size and variable angle search genetic algorithm as claimed in claim 1, comprising the steps of:

step 1): the unbalance compensation module (9) is based on the sine coefficient a_AX，a_AYAnd cosine coefficient b_AX，b_AYCalculating the total vibration amplitude A (k) of the rotor;

step 2): the total vibration amplitude A (k) and the vibration target value A are compared_objComparing, if the value of the total vibration amplitude A (k) is larger than the vibration target value A_objThen, a variable step length and variable angle searching method is adopted for the rotor vibration to carry out compensation control, and the variable step length and variable angle searching method is adopted: when the vibration of the next step is smaller than that of the previous step, the step length and the angle are increased, when the vibration of the next step is larger than that of the previous step, the angle is changed to 90 degrees, the step length is changed to be the set initial search step length, the search angle and the search direction are continuously changed, and the identification coefficients of the x direction and the y direction are respectively alpha_AX，β_AXAnd alpha_AY，β_AY；

Step 3): using genetic algorithm to identify the coefficient alpha_AX，β_AX，α_AY，β_AYOptimizing to obtain the optimal solution alpha of the identification coefficient_AX-B，β_A，α_AY-B，β_AY-B；

Step 4): optimal solution alpha according to identification coefficient_AX-B，β_AX-B，α_AY-B，β_AY-BCalculating the optimal compensation current i in the x direction and the y direction_c-x-BAnd i_c-y-BAnd output.

6. A method of compensating for vibration of a rotor according to claim 5, wherein: in step 1), the unbalance compensation module (9) is according to the formulaCalculating the total amplitude A (k) of the vibration of the rotor,

7. a method of compensating for vibration of a rotor according to claim 5, wherein: in step 2), in the variable-step-length and variable-angle searching method, if the total vibration amplitude A (k) of the k-th step is smaller than the total vibration amplitude A (k-1) of the k-1-th step, the searching step is increased by a step difference delta R_kSearch angle increase Δ θ_kIf the total vibration amplitude A (k) in the k step is greater than or equal to the total vibration amplitude A (k-1), the search angle is changed by 90 degrees clockwise, and the search step length is R, the step length is shown in the specificationSaidR_k-1Step size at time k-1, θ_k-1Is the angle at time k-1.

8. A method of compensating for vibration of a rotor according to claim 7, wherein: in step 2), sequentially increasing the step number k, and repeatedly executing the step length and angle variable searching steps until the searched identification coefficient alpha_AX，β_AXAnd alpha_AY，β_AYConvergence stops the search in the square ABDC convergence domain.

9. A method of compensating for vibration of a rotor according to claim 5, wherein: in step 4), the optimal compensation current i in the x direction and the y direction_c-x-BAnd i_c-y-BThe method comprises the following steps:

Technical Field

The invention relates to a control technology of a bearingless permanent magnet synchronous motor, in particular to rotor offset and vibration control of the bearingless permanent magnet synchronous motor, which improves the traditional fixed-step fixed-angle search algorithm and performs compensation control on rotor unbalance.

Background

The bearingless permanent magnet synchronous motor is a motor which utilizes a suspension winding to provide suspension force and a torque winding to provide torque to enable a rotor to suspend and rotate, so the bearingless permanent magnet synchronous motor has the advantages of no friction, no abrasion, no need of lubrication, high rotating speed, high precision and the like, and has important application value in the high and new technical fields of centrifuges, sealing pumps, turbomolecular pumps, flywheel energy storage, semiconductor industry, life science, aerospace and the like. For the rotor of the bearingless permanent magnet synchronous motor, mass unbalance exists to a certain extent, namely the mass center and the centroid cannot be completely overlapped, so that an excitation force with the same frequency as the rotor can be generated when the rotor rotates, the rotor loses balance by the force, the original position is deviated, and the radial vibration of the rotor is caused. When the rotating speed reaches a certain degree, the amplitude of unbalanced vibration exceeds an air gap, and the rotor and the stator rub to ensure that the system cannot work normally, so that the vibration compensation of the suspended rotor in the bearingless permanent magnet synchronous motor is very important.

The following two cases can be roughly classified for the rotor vibration compensation control: the first is the force minimization criterion and the second is the vibration displacement minimization criterion. The first kind of minimum acting force criterion is to minimize the pass-frequency exciting force in the closed-loop control system and make the rotor rotate around the inertial shaft, that is to say before the feedback signal enters the controller, the minimum acting force criterion is realized by eliminating the vibration signal in the feedback signal, the minimum acting force criterion utilizes the notch filter technology, the principle is to add a stable controller in the forward channel of the closed-loop control, the controller filters out the signal with the same frequency as the rotating speed, and other signals basically do not pass through the attenuation, thus the damping and the rigidity of the suspension rotor are very small in the very narrow frequency band range set by the wave trap, and the vibration of the rotor is also reduced. The second type of displacement minimization principle compensation technique is to add a compensation signal to the output signal of the closed-loop system controller to force the rotor to rotate around its geometric center, which is equivalent to adding a counter mass to the rotor to counteract the mass imbalance of the rotor, thereby effectively counteracting the centrifugal force caused by the mass imbalance of the rotor. The first kind of compensation control can not meet the requirement of high rotor rotation precision, and the second kind of compensation control is not fast enough in response and not sensitive enough in response. Therefore, a compensation control strategy is needed which can adapt to multiple situations and can optimize the compensation effect of the compensation control strategy, and the vibration compensation speed is accelerated under the second type of compensation control.

The traditional fixed-step search starts with a certain step length, the search angle is fixed, if the vibration of the next step is smaller than that of the previous step, the search direction and the step length are kept unchanged, and if the vibration of the next step is larger than that of the previous step, the search angle is changed by 90 degrees, and the search step length is unchanged. In the traditional fixed-step search algorithm, the search step size is constant all the time, so that the compensation speed is reduced, and if the step size and the angle are changed continuously in the search process, the compensation speed can be increased.

Disclosure of Invention

The invention aims to overcome the defects of vibration, eccentricity and the like of a rotor when a bearingless permanent magnet synchronous motor works, particularly the problems that the rotor cannot be positioned at high precision due to deviation and vibration when the motor works at different rotating speeds, and the like.

The technical scheme adopted by the variable-step variable-angle search genetic algorithm optimized rotor vibration compensation system provided by the invention is as follows: the device comprises a torque control part and a suspension force control part, a photoelectric encoder is adopted to detect the rotating speed omega of a bearingless permanent magnet synchronous motor, the rotating speed omega is used as the input of an improved wave trap, and the sine coefficient a of a vibration signal with the same frequency as a rotor in the x and y directions of the rotor is output by the improved wave trap_AX，a_AYAnd cosine coefficient b_AX，b_AXThe output end of the improved wave trap is connected with the input end of the unbalance compensation module, and the sine coefficient a_AX，a_AYAnd cosine coefficient b_AX，b_AXThe optimal compensation current i in the x direction and the y direction is output by the unbalance compensation module_c-x-BAnd i_c-y-BThe optimal solution compensates for the current i_c-x-BAnd i_c-y-BInput into the levitation force control portion.

The improved wave trap comprises a first sine multiplication module and a first sine product which are sequentially connected in seriesThe output ends of the sine second multiplication module and the cosine second multiplication module are connected with the input end of the unbalance compensation module, and the output end of the sine second multiplication module is the sine coefficient a_AX，a_AYThe cosine coefficient b is output by the cosine second multiplication module (122)_AX，b_AY。

The sine coefficient a_AX，a_AYAnd cosine coefficient b_AX，b_AYThe method comprises the following steps:

the technical scheme adopted by the method for optimizing the rotor vibration compensation by the variable-step variable-angle search genetic algorithm comprises the following steps:

step 1): the unbalance compensation module is based on sine coefficient a_AX，a_AYAnd cosine coefficient b_AX，b_AYCalculating the total vibration amplitude A (k) of the rotor;

step 2): the total vibration amplitude A (k) and the vibration target value A are compared_objComparing, if the value of the total vibration amplitude A (k) is larger than the vibration target value A_objThen, a variable step length and variable angle searching method is adopted for the rotor vibration to carry out compensation control, and the variable step length and variable angle searching method is adopted: when the vibration of the next step is smaller than that of the previous step, the step length and the angle are increased, when the vibration of the next step is larger than that of the previous step, the angle is changed to 90 degrees, the step length is changed to be the set initial search step length, the search angle and the search direction are continuously changed, and the identification coefficients of the x direction and the y direction are respectively alpha_AX，β_AXAnd alpha_AY，β_AY；

Step 3): using genetic algorithm to identify the coefficient alpha_AX，β_AX，α_AY，β_AYOptimizing to obtain the optimal solution alpha of the identification coefficient_AX-B，β_A，α_AY-B，β_AY-B；

Step 4): optimal solution alpha according to identification coefficient_AX-B，β_AX-B，α_AY-B，β_AY-BCalculating the optimal compensation current i in the x direction and the y direction_c-x-BAnd i_c-y-BAnd output.

In the variable-step-length and variable-angle searching method, if the total vibration amplitude A (k) of the k step is smaller than the total vibration amplitude A (k-1) of the k-1 step, the searching step is increased by a step difference delta R_kSearch angle increase Δ θ_kIf the total vibration amplitude A (k) in the k step is greater than or equal to the total vibration amplitude A (k-1), the search angle is changed by 90 degrees clockwise, and the search step length is R, the step length is shown in the specificationSaidR_k-1Step size at time k-1, θ_k-1Is the angle at time k-1.

The invention has the advantages that after the technical scheme is adopted:

1. the invention adopts the step-variable and angle-variable iterative search to accelerate the reaction speed of the rotor vibration compensation, can carry out the rotor vibration compensation under the condition of the full-rotating-speed operation of the motor, improves the reliability of the system and ensures the normal operation of the motor.

2. And the search result is optimized by adopting a genetic algorithm, so that the search precision is higher, the obtained compensation current is more accurate, the vibration and the offset of the rotor are inhibited to the greatest extent, and the high-precision performance of the bearingless permanent magnet synchronous motor is ensured.

3. The vibration compensation of the rotor can be realized in a large rotating speed range, and the vibration compensation device can be applied to various occasions.

4. Because the vibration compensation of the invention utilizes Maxwell force generated by the current of the suspension winding to balance the centrifugal force, no additional equipment is needed to balance the rotor, and the vibration compensation is realized only by a software control algorithm, so that the debugging is more convenient.

5. The invention can ensure the reliability of stable operation of other parts of the system on the basis of inhibiting the rotor vibration to the maximum extent, and has higher compensation speed and small interference to the system.

Drawings

FIG. 1 is a structural control block diagram of a variable step size and angle search genetic algorithm optimized rotor vibration compensation system provided by the present invention;

figure 2 is an improved structure and schematic block diagram of the improved trap of figure 1:

FIG. 3 is a flow chart of a compensation method of the imbalance compensation module 9 of FIG. 1;

FIG. 4 is a flow chart of a method of varying step size and angle search identification coefficients in FIG. 3;

FIG. 5 is a schematic diagram illustrating a conventional fixed step search and a variable step and angle search process shown in FIG. 4;

FIG. 6 is a flow chart of a method for selecting an optimal solution for the genetic algorithm of FIG. 3;

in the figure: 3. a force/current conversion module; 4. a torque control module; 5. a suspension control module; 7. a photoelectric encoder; 8. an improved wave trap; 9. an imbalance compensation module; 21. a PID module; a PI module; 61. a CLARK/PARK transform module; 101. a sinusoidal first multiplication module; 102. a cosine first multiplication module; 111. a sinusoidal first integration module; 112. a cosine first integration module; 121. a sinusoidal second multiplication module; 122. a cosine second multiplication module; 131. a sinusoidal third multiplication module; 132. and a cosine third multiplication module.

Detailed Description

Referring to fig. 1, the variable-step and variable-angle search genetic algorithm optimized rotor vibration compensation system (hereinafter referred to as vibration compensation system) according to the present invention includes a torque control portion, a levitation force control portion, an improved wave trap 8, and an imbalance compensation module 9. The torque control part and the suspension force control part are both conventional control parts of the bearingless permanent magnet synchronous motor control system. The torque control part consists of a torque control module 4, a CLARK/PARK conversion module 61 and a PI module 23, and the levitation force control part consists of a levitation force control module 5, a CLARK/PARK conversion module 62, a force/current conversion module 3 and two PID modules 21 and 22.

For the torque control part, the output is from the torque control module 4Three-phase torque control current i_Ma，i_Mb，i_McControlling torque winding of bearingless permanent magnet synchronous motor, three-phase torque control current i_Ma，i_Mb，i_McThe torque feedback current i is converted into a dq axis coordinate system through a CLARK/PARK conversion module 61_Mq，i_Md. The rotation speed omega of the bearingless permanent magnet synchronous motor is detected by adopting the photoelectric encoder 7, the rotation speed omega and the reference rotation speed omega are differenced to obtain a rotation speed difference, and the rotation speed difference is input into the PI module 23 to obtain a reference torque current i_Mq ^*Setting a reference torque current i_Md ^*Is 0. The reference torque current i under a q-axis coordinate system_Mq ^*，i_Md ^*Torque feedback current i in dq axis coordinate system_Mq，i_MdCorrespondingly making difference to obtain current difference, using the current difference as input of torque control module 4, and outputting the three-phase-change control current i by torque control module 4_Ma，i_Mb，i_McThe bearing-free permanent magnet synchronous motor is provided.

The rotating speed omega of the bearingless permanent magnet synchronous motor detected by the photoelectric encoder 7 is input into the improved wave trap 8 to be used as the input of the improved wave trap 8, and the sine coefficient a of the vibration signal with the same frequency as the rotor is obtained after the processing of the improved wave trap 8_AX，a_AYAnd cosine coefficient b_AX，b_AX. The output end of the improved wave trap 8 is connected with the input end of the unbalance compensation module 9, and the sine coefficient a is converted into the sine coefficient a_AX，a_AYAnd cosine coefficient b_AX，b_AXInputting the current to an unbalance compensation module 9, and obtaining the optimal compensation current i in the x direction and the y direction after the processing of the unbalance compensation module 9_c-x-BAnd i_c-y-BThe unbalance compensation module 9 compensates the current i with the optimal solution_c-x-BAnd i_c-y-BInput into the levitation force control portion.

For the levitation force control part, a three-phase levitation force control current i is output by the levitation force control module 5_Ba，i_Bb，i_BcControlling a levitation force winding of a bearingless permanent magnet synchronous machine, the three-phase levitation force controlling a current i_Ba，i_Bb，i_BcBy CLAThe RK/PARK conversion module 62 converts the suspension force feedback current i into dq axis coordinate system_Bd，i_Bq. Meanwhile, the actual displacement x and y of the bearingless permanent magnet synchronous motor are detected by using a displacement sensor, the actual displacement x and y and a set reference displacement x and y are subjected to difference (the reference displacement x and y are generally set to be 0), and the displacement difference obtained by difference is processed by PID modules 21 and 22 to obtain a reference suspension force F in the x and y directions_x ^*，F_y ^*Reference levitation force F_x ^*，F_y ^*Inputting the current into a force/current conversion module 3, and obtaining a suspension force reference current i in the x and y directions through the conversion of the force/current conversion module 3_Bd ^*，i_Bq ^*And the current is in a dq axis coordinate system. Referencing the levitation force with a current i_Bd ^*，i_Bq ^*And the optimum compensation current i in the x-direction and the y-direction input by the unbalance compensation module 9_c-x-B，i_c-y-BAdded and then added with the suspension feedback current i output by the CLARK/PARK conversion module 62_Bd，i_BqMaking difference to obtain suspension force input current i_c-Bd，i_c-BqComprises the following steps:

inputting a levitation force into a current i_c-Bd，i_c-BqAs an input signal of the levitation control module 5, the levitation control module 5 outputs the three-phase levitation control current i_Ba，i_Bb，i_BcThe bearing-free permanent magnet synchronous motor is provided.

Referring to fig. 2, the improved wave trap 8 is composed of a sine multiplication module, a sine integration module, and a cosine multiplication module and a cosine integration module. The sinusoidal first multiplication module 101, the sinusoidal first integration module 111 and the sinusoidal second multiplication module 121 are sequentially connected in series; the cosine first multiplication module 102, the cosine first integration module 112 and the cosine second multiplication module 122 are connected in series in sequence. The output end of the sinusoidal second multiplication module 121 is directly connected with the input end of the unbalance compensation module 9, and the output of the sinusoidal second multiplication module 121 is in the x directionAnd sine coefficient a in the y direction_AX，a_AYThe sine coefficients a in the x and y directions_AX，a_AYInput to the imbalance compensation module 9. Similarly, the output end of the cosine second multiplying module 122 is directly connected to the input end of the unbalance compensating module 9, and the cosine coefficients b in the x direction and the y direction are output by the cosine second multiplying module 122_AX，b_AYThe cosine coefficient b_AX，b_AYInput to the imbalance compensation module 9. In the conventional trap filter, a sine third multiplication module 131 is connected in series after the sine second multiplication module 121, and a cosine third multiplication module 132 is connected in series after the cosine second multiplication module 122. The sine third multiplying module 131 and the cosine third multiplying module 132 output the same-frequency unbalanced vibration signals. Therefore, the invention directly outputs the sine coefficient a after the input signal passes through the first multiplication module, the first integration module and the second multiplication module on the basis of the traditional wave trap_AX，a_AYAnd cosine coefficient b_AX，b_AY. It can be seen that the improved trap 8 of the present invention eliminates the third multiplication module 131 for sine and the third multiplication module 132 for cosine in the conventional trap, and the sine coefficient a is obtained_AX，a_AYAnd cosine coefficient b_AX，b_AYDirectly to the unbalance compensation module 9.

When the motor rotates, when the rotor vibrates, the rotation speed ω measured by the photoelectric encoder 7 is sent to the improved wave trap 8, and is used as the input of the improved wave trap 8, the improved wave trap 8 generates a sine signal sin ω t and a cosine signal cos ω t which have the same frequency as the rotor, the sine signal sin ω t is input into the corresponding sine first multiplication module 101 and the corresponding sine second multiplication module 121, and the cosine signal cos ω t is input into the corresponding cosine first multiplication module 102 and the corresponding cosine second multiplication module 122. Thereby obtaining a sine coefficient a_AX，a_AYAnd cosine coefficient b_AX，b_AYThe following were used:

in the formula: a is_AXThe sine coefficient of the vibration signal with the same frequency as the rotor in the x direction of the rotor; a is_AYThe sine coefficient of the vibration signal with the same frequency as the rotor in the y direction of the rotor; b_AXThe cosine coefficient of the vibration signal with the same frequency as the rotor in the x direction of the rotor; b_AYThe cosine coefficient of the vibration signal with the same frequency as the rotor in the y direction of the rotor; ω is the rotational speed of the rotor.

As shown in FIG. 3, the imbalance compensation module 9 depends on the sine coefficient a_AX，a_AYAnd cosine coefficient b_AX，b_AYCalculating the vibration amplitudes in the x direction and the y direction as follows:

in the formula: a. the_AX(k)，A_AY(k) The vibration amplitudes of the rotor in x and y directions at the k-th time instant, respectively.

And then calculating the total vibration amplitude of the rotor as follows:

in the formula: a (k) is the total amplitude of the vibration of the rotor at the k-th moment, and is also the total amplitude of the vibration at the k-th step.

Referring to fig. 3 and 4, after the unbalance compensation module 9 obtains the total vibration amplitude a (k), the vibration total amplitude a (k) and the vibration target value a are calculated_objMaking a comparison based on the vibration target value A_objAnd then carrying out variable step length and variable angle search compensation on the comparison result with the actual vibration total amplitude A (k). If the value of the vibration total amplitude A (k) is larger than the vibration target value A_objAnd performing compensation control on the rotor vibration, and otherwise, not working. Theoretical vibration target value A_objIt is possible to take 0, but considering that the rotor vibration is not necessarily 0 in the actual process, the vibration target value a_objA very small value can be taken. When the unbalance compensation module 9 performs compensation control on the rotor vibration, based on the vibration total amplitude A (k), the variable step length and the variable angle are adoptedThe identification coefficients of two x directions and two y directions obtained by the degree search method are respectively alpha_AX，β_AXAnd alpha_AY，β_AY。

The step length and angle variable search path is as shown in fig. 5, when the vibration of the next step is smaller than that of the previous step, the step length and the angle are increased, when the vibration of the next step is larger than that of the previous step, the angle is changed to 90 degrees, the step length is changed to be the set fixed and unchangeable initial search step length, so that the search angle and the search direction are continuously changed, and the required identification coefficient is obtained. To search the identification coefficient alpha in the x direction_AX，β_AXFor example, first, the identification coefficient α in the x direction is determined_AX，β_AXInitializing and setting an initial value (alpha)_AX,β_AX)₀(0,0), initial search angle θ₀Is 0 deg., and the initial search step R is 0. If the vibration of the rotor in the k step is smaller than that in the k-1 step, namely the total vibration amplitude A (k) in the k step is smaller than that in the k-1 step (k-1), the search step length is increased by a step difference delta R_kSearch angle increase Δ θ_kIn the present invention, the kth step is equivalent to the rotor being at the kth time. If the rotor vibration in the k step is larger than or equal to the vibration in the k-1 step, namely the total vibration amplitude A (k) is larger than or equal to the total vibration amplitude A (k-1), the search angle is changed by 90 degrees clockwise, and the search step is R. Wherein the step difference Δ R_kAnd the angular difference Δ θ_kRespectively as follows:

in the formula: Δ R_kThe step length difference between the kth step and the k-1 step is obtained; a (k-1) is the total amplitude of the rotor vibration at the moment of k-1; r_kAnd R_k-1Step lengths at the time of k and k-1 respectively; delta theta_kThe angle difference between the kth step and the k-1 step is obtained; theta_kAnd theta_k-1Respectively, the angles at times k and k-1.

Searching for an identification coefficient alpha in the x-direction_AX，β_AXThe specific process is as follows:

when k is 1, the vibration total amplitude A of the step 1 is compared_AX(1) And a vibration target value A_objComparing, if the vibration total amplitude A after the first compensation_AX(1) Is less than or equal to A_objIf yes, stopping and exiting; if after the first compensation A_AX(1) Greater than A_objThen, executing variable step length and variable angle search: if A is satisfied_AX(1)＜A_AX(0) I.e. total amplitude of vibration A after first compensation_AX(1) Less than the total amplitude A of the initial vibration_AX(0) Then with (α)_AX,β_AX)₀As a starting point, the search step size is increased by Δ R₁Search angle increase Δ θ₁Obtaining the identification coefficient (alpha)_AX,β_AX)₁. On the contrary, if A_AX(1)≥A_AX(0) I.e. total amplitude of vibration A after first compensation_AX(1) Greater than or equal to the total amplitude A of the initial vibration_AX(0) Then with (α)_AX,β_AX)₀For the starting point, the search step size is R, and the search angle is changed by 90 ° clockwise. Wherein (alpha)_AX,β_AX)₁The identification coefficient obtained for the first search; Δ R₁Step size for the first change; delta theta₁Is the first angle change.

When k is 2, the vibration total amplitude A of the step 2 is compared_AX(2) And a vibration target value A_objMaking a comparison, if A_AX(2) Is less than or equal to A_objThen the search is terminated and exited, if A is after the second compensation_AX(2) Greater than A_objThen, executing variable step length and variable angle search: if A is satisfied_AX(2)＜A_AX(1) I.e. the total amplitude of the vibration after the second compensation is smaller than the total amplitude of the vibration after the first compensation, is calculated by (alpha)_AX,β_AX)₁Increment by Δ R for the starting point search step₂Search angle increase Δ θ₂Obtaining the identification coefficient (alpha)_AX,β_AX)₂. If A_AX(2)≥A_AX(1) I.e. the total amplitude of the vibration after the second compensation is greater than or equal to the total amplitude of the vibration after the first compensation, is calculated by (alpha)_AX,β_AX)₁For the starting point, the search step size is R, and the search angle is changed by 90 ° clockwise. Wherein (alpha)_AX,β_AX)₂The identification coefficient obtained by the second search; Δ R₂Step size for the second change; delta theta₂Is the angle of the second change.

When k is 3, the vibration total amplitude A of the step 3 is compared_AX(3) And A_objMaking a comparison, if A_AX(3) Is less than or equal to A_objIf the search is stopped, the search is stopped and quit, if A_AX(3) Greater than A_objThen, executing variable step length and variable angle search: if A is satisfied_AX(3)＜A_AX(2) I.e. the total amplitude of the vibration after the third compensation is smaller than the total amplitude of the vibration after the second compensation, is calculated by (alpha)_AX,β_AX)₂Increment by Δ R for the starting point search step₃Search angle increase Δ θ₃Obtaining the identification coefficient (alpha)_AX,β_AX)₃. If A_AX(3)≥A_AX(2) That is, the total amplitude of the vibration after the third compensation is greater than or equal to the total amplitude of the vibration after the second compensation, is (alpha)_AX,β_AX)₂For the starting point, the search step size is R, and the search angle is changed by 90 ° clockwise. Wherein (alpha)_AX,β_AX)₃The identification coefficient obtained by the third search; Δ R₃Step size for the third change; delta theta₃The angle for the third change.

Sequentially increasing the step number k in this way, and repeatedly executing the step length and angle variable searching steps until the searched identification coefficient alpha_AX，β_AXStopping searching when the signal is converged in the square ABDC convergence domain, and obtaining the identification coefficient alpha in the x direction in the convergence domain_AX，β_AX。

Similarly, the same variable step length and variable angle searching method is adopted to obtain the identification coefficient alpha in the y direction in the convergence domain_AY，β_AY。

Compared with the fixed-step and fixed-angle search, it can be seen from the search path shown in fig. 5 that the identification coefficient obtained by the variable-step and variable-angle iterative search in the invention can increase the search speed, and the search speed is greatly improved.

Coefficient of identification alpha in x-direction in convergence domain_AX，β_AXAnd the identification coefficient alpha in the y direction_AY，β_AYThe precision is not high enough, and then the genetic algorithm is adopted to further optimize the precision. The genetic algorithm evaluates the quality of an individual through a fitness function. The finite number in the converged domain is selected in the square ABDC for binary encoding as the population to be solved. As shown in fig. 5 and 6, the selection corresponds to a selection operator in the genetic algorithm. Crossover is the crossover operator in the corresponding algorithm, i.e., genetic recombination produces a new set of solutions. Mutation corresponds to the process by which a mutation operator in the algorithm, i.e., a component of the code, changes. Firstly, binary coding is carried out to generate an initial population, referring to fig. 5, the abscissa range is from the abscissa of point C to the abscissa of point B, the ordinate range is from the ordinate of point C to the ordinate of point a, the abscissa of the segment in this range is divided into eight equal parts, 64 identification coefficients are taken, the range of the abscissa coding is 000-111, the range of the ordinate coding is 000-111, the size of the population is 4, the population is composed of 16 individuals, and each individual can be generated by a random method. From these 64 identification coefficients as the initial population, the next step is to count gen from 0 and use the objective function value as the fitness of the individual. If the fitness precision is achieved, outputting a result, if the fitness precision is not achieved, executing selection operation, summing the four groups of calculation results, dividing the result of each group by the sum of the results of all samples to obtain a probability percentage, wherein the higher the probability percentage is, the better the fitness is, and the individual can be copied to the next generation of population. After the selection operation is finished, a cross operation is performed, a single-point cross method is adopted, a group is randomly paired, for example, two binary codes 110110 and 101011 are crossed, 1 and 4 are crossed, namely 110, 101 and 011 are crossed, and after recombination, the two groups become new two individuals 110011 and 101110. And performing mutation operation after the cross operation is finished. Randomly determining the position of a variation point, negating the original code value to obtain a new generation group, sequentially circulating to reach the specified algebraic precision, and finally outputting the optimal solution alpha of the identification coefficients in the x direction and the y direction meeting the conditions_AX-B，β_A，α_AY-B，β_AY-B。

The imbalance compensation module 9 optimally solves alpha based on the identification coefficient_AX-B，β_AX-B，α_AY-B，β_AY-BCalculating the optimal compensation current to obtain the optimal compensation current i in the x direction and the y direction_c-x-B，i_c-y-BThe calculation process is as follows:

to compensate for rotor vibration, a levitation force equal to and opposite to the imbalance force is required, and the levitation force is generated by a levitation current. The imbalance force calculation formula is as follows:

in the formula: m is the mass of the rotor; omega is the rotating speed; rho is the eccentricity of the rotor inertia shaft relative to the geometric central shaft; beta is a₀Is an initial phase angle; f_x1Is the unbalanced force in the x direction; f_y1Is an unbalanced force in the y-direction.

The suspension force calculation formula is as follows:

in the formula: mu.s_METhe initial phase of the equivalent magnetic field magnetomotive force is synthesized by the torque winding and the permanent magnet magnetic field; mu.s_BThe initial phase of the corresponding suspension magnetic field magnetomotive force of the suspension winding is shown; i.e. i_xAnd i_yFor controlling the levitation force current in x and y directions, m is a motor-related constant (obtained by multiplying the current amplitude of the torque winding, the number of turns of the torque winding and the levitation winding, and the radius and length of the rotor by 4.5), and F_xIs the levitation force in the x-direction, F_yIs the levitating force in the y-direction.

To achieve the purpose of compensating the rotor vibration, the imbalance force is counteracted by controlling the levitation force, and the following levitation force currents in the x and y directions can be obtained according to the calculation formulas of the levitation force and the imbalance force of the formulas (7) and (8):

if the vibration compensation of the rotor is required to be realized, the levitation force generated by the compensation current is equal to the unbalanced force, and the calculation of the compensation current is consistent with that of the levitation current. The rotation speed omega is taken as a variable, the product of the rest unknowns is set as an identification coefficient, the relation between the identification coefficient and the current can be obtained, and the optimal compensation current i in the x direction and the y direction is obtained according to the optimal solution of the identification coefficient_c-x-BAnd i_c-y-BAnd identifying the optimal solution alpha of the coefficient_AX-B，β_AX-B，α_AY-B，β_AY-BThe relational calculation formula is as follows:

the unbalance compensation module 9 obtains the optimal compensation current i in the x direction and the y direction_c-x-BAnd i_c-y-BThe levitation force control part in FIG. 1 is inputted with a levitation force reference current i_Bd ^*，i_Bq ^*And levitation force feedback current i_Bd，i_BqThe combined action obtains the suspension force input current i_c-Bd，i_c-BqThe suspension force control module 5 controls the suspension force to achieve the purpose of eliminating unbalanced force, and vibration compensation of the rotor can be achieved.