Permanent magnet synchronous motor torque harmonic suppression system

Abstract

The invention relates to a permanent magnet synchronous motor torque harmonic suppression system, which comprises a PMSM model, an iterative learning control module, a two-degree-of-freedom control module and a genetic algorithm parameter optimization module, and is characterized in that the two-degree-of-freedom control module is connected in series with the iterative learning control module, and the expected current output by the two-degree-of-freedom control module is equal to the expected current output by the genetic algorithm parameter optimization module. The current is added with compensation current output by the iterative learning control module to serve as expected current, and the expected current serves as input of the PMSM model; the iterative learning control module adopts a closed-loop PI type iterative learning control algorithm; and the genetic algorithm parameter optimization module is adopted to carry out optimization setting on the parameters of the iterative learning control module. Compared with a traditional iterative learning controller, the harmonic suppression system adopted by the invention not only can suppress periodic torque harmonics, but also can suppress non-periodic torque harmonics; and the genetic algorithm is adopted to automatically perform parameter optimization on the parameters of the iterative learning control module, so that a complex parameter adjustment process is omitted.

Description

Technical Field

[0001] The present invention relates to the technical field of motor control, and in particular to a permanent magnet synchronous motor torque harmonic suppression system. Background Art

[0002] Permanent magnet synchronous motors have many excellent characteristics, such as high efficiency, high torque-to-current ratio, low noise and robustness, and are widely used in industrial servo applications. However, permanent magnet synchronous motors have obvious torque harmonics during operation, which will directly lead to harmonics in the motor output speed. At higher operating speeds, these torque harmonics are naturally filtered out by the rotor and load inertia and will not be reflected in the motor speed; but at low speeds, the torque harmonics cannot be filtered, and the speed oscillation caused by them seriously limits the performance of the servo system.

[0003] The current method for suppressing torque harmonics is usually the iterative learning control method (ILC), but this method has many control parameters, which makes parameter adjustment complicated. At the same time, the above method only considers periodic torque harmonics. In fact, there are also non-periodic torque harmonics caused by changes in motor parameters and sudden changes in load torque in the control system, and this method cannot effectively suppress non-periodic torque harmonics. Summary of the invention

[0004] The present invention provides a permanent magnet synchronous motor torque harmonic suppression system to solve the above technical problems.

[0005] In order to solve the above technical problems, the present invention provides a permanent magnet synchronous motor torque harmonic suppression system, including a PMSM model, an iterative learning control module, a two-degree-of-freedom control module and a genetic algorithm parameter optimization module.

[0006] The two-degree-of-freedom control module is connected in series with the iterative learning control module, the desired current output by the two-degree-of-freedom control module is added to the compensation current output by the iterative learning control module as the desired current, and the desired current is used as the input of the PMSM model;

[0007] The iterative learning control module adopts a closed-loop PI-type iterative learning control algorithm; and the genetic algorithm parameter optimization module is used to optimize and adjust the parameters of the iterative learning control module.

[0008] Preferably, the iterative learning control module includes a PI controller and an iterative learning controller, and its working process includes:

[0009] The reference current generated by the PI controller is converted into a reference torque The error between the reference torque and the actual torque is used as the input of the iterative learning controller;

[0010] Obtain the iteration period through the torque harmonic period;

[0011] The compensation current is generated by iterating the learning rate, and the output current of the PI controller is compensated by the compensation current.

[0012] Preferably, the iterative process of the iterative learning controller is divided into the following four stages: a feedforward control stage, an error preservation stage, a feedback correction stage and an iterative control stage.

[0013] Preferably, the control rate of the iterative learning controller includes a forgetting factor.

[0014] Preferably, the two-degree-of-freedom control module includes a two-degree-of-freedom controller, and the two-degree-of-freedom controller parameter tuning step includes: firstly adjusting some parameters using a conventional PID parameter tuning method, and then selecting a two-degree-of-freedom coefficient.

[0015] Preferably, the two-degree-of-freedom controller adopts a set value filter type two-degree-of-freedom control structure.

[0016] Preferably, the implementation steps of the genetic algorithm parameter optimization module include:

[0017] Initialize the population: set the value range of the parameter and generate the initial population by a random function;

[0018] Individual evaluation: Set the fitness function and calculate the fitness function value of each individual in the current initial population;

[0019] Selection operation: Determine whether to keep the fitness function value based on its relative size;

[0020] Crossover operation: According to the set crossover probability, the selected parent individuals undergo gene crossover to produce new offspring individuals;

[0021] Mutation operation: Gene mutation is performed on a portion of the offspring individuals to introduce new genetic information;

[0022] Update population: replace the generated offspring individuals with the original population to form a new population;

[0023] Termination judgment: If the number of iterations reaches the maximum number of iterations, the set of parameters with the largest fitness function value is selected and output.

[0024] Preferably, the fitness function is taken as the inverse of the minimum objective function, and the minimum objective function is the squared error value.

[0025] Preferably, the fitness function is:

[0026]

[0027] Among them, J(x) is the minimum objective function and σ is a custom minimum value.

[0028] Preferably, the minimum value σ is 10 -10 .

[0029] Compared with the prior art, the permanent magnet synchronous motor torque harmonic suppression system provided by the present invention has the following advantages:

[0030] 1. Compared with the traditional iterative learning controller, the present invention adopts an iterative learning control module and a two-degree-of-freedom control module, which can not only suppress periodic torque harmonics, but also suppress non-periodic torque harmonics;

[0031] 2. The traditional iterative learning controller has complex parameter adjustment and cannot achieve the optimal suppression effect. The present invention uses a genetic algorithm to automatically optimize parameters, eliminating the complex parameter adjustment process. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a structural block diagram of an iterative learning control module in a specific implementation manner of the present invention;

[0033] Figure 2 It is the structural block diagram of the two-degree-of-freedom control module based on the internal model controller;

[0034] Figure 3 It is the structural block diagram of the two-degree-of-freedom control of the set value filter type;

[0035] Figure 4 This is the block diagram of permanent magnet synchronous motor vector control based on 2DOFC;

[0036] Figure 5 It is a structural block diagram of a permanent magnet synchronous motor torque harmonic suppression system in a specific implementation manner of the present invention;

[0037] Figure 6 It is a fitness function convergence curve diagram in a specific implementation mode of the present invention;

[0038] Figure 7 is the FFT spectrum of the speed under different control strategies;

[0039] Figure 8 The waveform diagram of the motor speed under different control strategies;

[0040] Fig. 9 The dynamic performance diagram of speed under different control strategies. DETAILED DESCRIPTION

[0041] In order to describe the technical solution of the above invention in more detail, specific embodiments are listed below to demonstrate the technical effects; it should be emphasized that these embodiments are used to illustrate the present invention but not to limit the scope of the present invention.

[0042] The permanent magnet synchronous motor torque harmonic suppression system provided by the present invention is as follows: Figure 5 As shown, it includes a PMSM model, an iterative learning control module, a two-degree-of-freedom control module and a genetic algorithm parameter optimization module.

[0043] In the speed loop, the two-degree-of-freedom control module is connected in series with the iterative learning control module, and the desired current output by the two-degree-of-freedom control module is added to the compensation current output by the iterative learning control module as the desired current, and the desired current is used as the input of the PMSM model, that is, the compensation current of the current loop; the iterative learning control module adopts a closed-loop PI iterative learning control algorithm. Different from the traditional PI controller that uses the current and adjacent moment deviation information to calculate the control instruction in the feedback control framework, the iterative learning control jumps out of the feedback control framework and can compensate for certain regular interference or uncertainty in advance, thereby effectively overcoming the shortcomings of the PI controller in harmonic suppression; and the use of a two-degree-of-freedom control module can simultaneously meet the purposes of tracking and anti-disturbance.

[0044] The genetic algorithm parameter optimization module is used to optimize and adjust the parameters of the iterative learning control module, thereby realizing automatic parameter optimization and eliminating the complicated parameter adjustment process.

[0045] In some embodiments, a surface-mounted permanent magnet synchronous motor is used as a research object, and the voltage equation established in the dq axis rotating coordinate system is:

[0046]

[0047] where u d 、u q Respectively represent the stator voltage of dq axis; i d 、i q Respectively represent the stator current of dq axis; ψ f represents the rotor flux; ω e Indicates the rotor electrical angular velocity; L s Represents the stator equivalent inductance; R s is the phase resistance of the stator winding.

[0048] The electromagnetic torque equation and motion equation of the permanent magnet synchronous motor can be expressed as:

[0049]

[0050] Where T e Represents electromagnetic torque; T Lrepresents load torque; J represents moment of inertia; B represents viscous friction coefficient; ω r It represents the mechanical angular velocity of the rotor; P represents the number of pole pairs of the motor.

[0051] In some embodiments, Figure 1 As shown, the iterative learning control module includes a PI controller and an iterative learning controller. Figure 1 middle, The expected current output by the two-degree-of-freedom control module is added to the compensation current output by the iterative learning controller as the final expected current. That is the input current of the current loop.

[0052] In some embodiments, the iterative process of the iterative learning controller is divided into the following four stages: a feedforward control stage, an error preservation stage, a feedback correction stage, and an iterative control stage.

[0053] Specifically, the working process of the iterative learning control module includes:

[0054] The reference current generated by the PI controller is converted into a reference torque The error between the reference torque and the actual torque is used as the input of the iterative learning controller;

[0055] Obtain the iteration period through the torque harmonic period;

[0056] The compensation current is generated by iterative learning rate, and the output current of the PI controller is compensated by the compensation current, thereby achieving early compensation for interference.

[0057] In some embodiments, the control rate of the iterative learning controller includes a forgetting factor. Since there is high-frequency uncertainty interference when the motor is running, in order to avoid the differential term in the control rate as much as possible, a closed-loop PI-type iterative learning control algorithm with a forgetting factor is used to suppress periodic torque harmonics. The specific form of the iterative learning control rate is as follows:

[0058]

[0059] in, is the q-axis current compensation value of the current cycle; is the q-axis current compensation value of the previous cycle; α is the forgetting factor; k p and k i is the PI gain of the controller; e k (t) is the torque error of the current cycle,

[0060] In some embodiments, please refer to Figure 2The two-degree-of-freedom control module includes a two-degree-of-freedom controller (2DOFC). The present invention designs the two-degree-of-freedom controller based on the internal model control principle and replaces the PI controller with the two-degree-of-freedom controller. Figure 2 In the above equation, Q1(s) and Q2(s) constitute a two-DOF internal model controller; G P (s) is the actual object of the charge; G m (s) is the mathematical model of the controlled object, i.e., the internal model. The present invention adopts a two-degree-of-freedom control structure of a set value filter type, such as Figure 3 shown. Figure 2 By equivalent transformation Figure 3 , they have the same input r, output y, and interference d. Therefore, the following relationship can be obtained:

[0061]

[0062] From equations 4 and 5, we can see that C(s) can be obtained by designing Q(s). In order to suppress the influence of model error on the system, enhance the robustness of the system, and ensure the practical feasibility of the controller, according to the design principle of the internal model controller, Q1(s) and Q2(s) are taken as follows:

[0063]

[0064] Among them, F1(s) and F2(s) are low-pass filters, and their design forms vary depending on the controlled objects. In order to make the servo system obtain better steady-state and dynamic characteristics, the following low-pass filters are used:

[0065]

[0066] Combining equation 5, equation 6 and equation 7, we can get:

[0067]

[0068] Among them, λ1 and λ2 are adjustable parameters of the controller, and the performance of controllers C1(s) and C2(s) can be changed by optimizing the parameters.

[0069] The block diagram of permanent magnet synchronous motor vector control based on two degrees of freedom is as follows Figure 4 As shown, Figure 4 In, K t is the conversion factor of torque. Figure 4 And the two-degree-of-freedom controller expression of equation 8, the closed-loop expression of the system is obtained:

[0070]

[0071] It can be seen from equation 9 that parameter λ1 affects the system's following response performance, while parameter λ2 can affect both the system's following response performance and anti-interference performance. When performing parameter tuning of a two-degree-of-freedom controller, λ2 can be adjusted first, that is, some parameters can be adjusted using conventional PID parameter tuning methods to ensure the system's anti-interference requirements, and then λ1 can be adjusted, that is, the two-degree-of-freedom coefficient can be selected to enable the system to obtain better following response performance.

[0072] From the above analysis, it can be seen that the parallel connection of the iterative learning controller and the output end of the two-degree-of-freedom controller has a strong nonlinear characteristic, so it is considered to use a genetic algorithm to achieve parameter tuning. Specifically, the implementation steps of the genetic algorithm parameter optimization module include:

[0073] Initialize the population: set the range of parameter values, each set of parameters constitutes an individual, and all individuals constitute the population. Use binary coding, generate the initial population by a random function, set the maximum number of iterations G and the population size.

[0074] Individual evaluation: Set the fitness function and calculate the fitness function value of each individual in the current initial population.

[0075] Selection operation: Determine whether it is retained according to the relative size of the fitness function value. In this embodiment, a roulette selection method can be used for selection. The higher the fitness value, the higher the probability of being selected and retained.

[0076] Crossover operation: According to the set crossover probability P c , the selected parent individuals undergo gene crossover to generate new offspring individuals. The crossover operation generates new solutions by exchanging and combining the genetic information of the parent individuals.

[0077] Mutation operation: Gene mutation is performed on a portion of the offspring individuals to introduce new genetic information. Mutation operation can increase the diversity of the population and help avoid local optimal solutions.

[0078] Update population: Replace the generated offspring individuals with the original population to form a new population.

[0079] Termination judgment: If the number of iterations is less than the maximum number of iterations, the individual evaluation is performed again to implement the loop operation; if the number of iterations is equal to the maximum number of iterations, the set of parameters with the largest fitness function value is selected for output.

[0080] Selecting a suitable objective function plays a vital role in achieving parameter optimization. This application uses the squared error as the minimum objective function J(x) for parameter selection. The individual fitness function F(x) is taken as the inverse of the minimum objective function. Since the expected value of the minimum objective function is 0, this will cause the denominator of the fitness function to be 0, so it is necessary to add a minimum value to the fitness function to ensure the normal operation of the program. The final individual fitness function is as follows:

[0081]

[0082] Where σ is a self-defined minimum value, which is 10 in this embodiment. -10 .

[0083] The control effect of the present invention is described below by a specific embodiment. A model of a permanent magnet synchronous motor vector control system based on two-degree-of-freedom iterative learning control is built in Matlab / Simulink simulation software. For a specific system block diagram, please refer to Figure 5 .

[0084] The specific simulation parameters of the motor are shown in Table 1.

[0085] Table 1 Permanent magnet synchronous motor simulation parameters

[0086]

[0087] Since the simulation system model is an idealized model, it is impossible to simulate the actual torque harmonics of the permanent magnet synchronous motor. Therefore, an equivalent periodic harmonic torque is injected into the input end of the motor. At the same time, a custom white noise is added to simulate the unknown non-periodic harmonic torque.

[0088] In order to optimize the parameters of the iterative learning controller, a genetic algorithm program suitable for this system was written in M ​​language. The specific parameters of the genetic algorithm are set as follows: p ∈[0,100], k i ∈[0,100], α∈[0,1], size=10, G=100, P c =0.95. After iteration, the convergence curve of the fitness function is obtained, as shown in Figure 6 As shown. Figure 6 It can be seen that the genetic algorithm determines the optimal value of the fitness function at the 48th iteration, that is, the minimum value of the objective function. The specific parameter values ​​are as follows: p =98.09, k i =12.53, α=0.09, and substituting the obtained parameters into the iterative learning controller can obtain a better harmonic suppression effect.

[0089] Figure 7 shows the FFT spectra of the speed under the three control strategies: traditional PI control, PI-ILC control, and 2DOFC-ILC. Figure 7a It can be seen that under the traditional PI control, there are periodic harmonic components in the speed, mainly 1st, 2nd, 6th and 12th harmonics, among which the 1st harmonic component is 0.006%, the 2nd harmonic component is 0.012%, the 6th harmonic component is 0.052%, and the 12th harmonic component is 0.045%. Figure 7b and Figure 7c It can be seen that no matter PI-ILC control or 2DOF-ILC control, the 1st, 2nd, 6th, and 12th harmonic components in the speed are reduced to 10 -4 %, both control strategies effectively suppress periodic harmonics.

[0090] Figure 8 The following are simulation waveforms of the motor speed under the PI-ILC control and 2DOF-ILC control strategies. When simulating 0-3s, the 2DOF-ILC control strategy is adopted; when simulating 3-6s, the PI-ILC control strategy is adopted. It can be seen that the speed fluctuation amplitude under the PI-ILC control strategy is about 20 -3 r / min. The speed fluctuation amplitude under the 2DOF-ILC control strategy is about 10 -3 r / min, which is 50% lower than that of the PI-ILC control strategy. In combination with the similar periodic harmonic suppression capabilities of the two control strategies in Figure 7, it is concluded that the 2DOF-ILC control strategy has a better suppression effect on non-periodic harmonics.

[0091] Fig. 9 In the figure, when the motor runs for 1s, a load disturbance of 10N·m is added to observe the anti-disturbance performance of the two strategies. The dotted line in the figure is the speed dynamic response of the PI-ILC strategy. The PI-ILC strategy has an overshoot of about 21.6r / min and reaches stability after 0.25s; the solid line is the speed dynamic response of the 2DOF-ILC strategy. The 2DOF-ILC strategy has an overshoot of about 2.2r / min and reaches stability after 0.005s. Therefore, the 2DOF-ILC control strategy has better robustness and anti-disturbance capabilities when facing a sudden increase in load disturbances.

[0092] In summary, the permanent magnet synchronous motor torque harmonic suppression system provided by the present invention includes a PMSM model, an iterative learning control module, a two-degree-of-freedom control module and a genetic algorithm parameter optimization module. The two-degree-of-freedom control module is connected in series with the iterative learning control module. The desired current output by the two-degree-of-freedom control module is added to the compensation current output by the iterative learning control module as the desired current, and the desired current is used as the input of the PMSM model, that is, the compensation current of the current loop; the iterative learning control module adopts a closed-loop PI iterative learning control algorithm, which effectively overcomes the shortcomings of the PI controller in harmonic suppression; and the use of a two-degree-of-freedom control module can simultaneously meet the purposes of tracking and anti-disturbance. The present invention uses the genetic algorithm parameter optimization module to optimize and adjust the parameters of the iterative learning control module, thereby realizing automatic parameter optimization and eliminating the complex parameter adjustment process.

[0093] Obviously, those skilled in the art can make various changes and modifications to the invention without departing from the spirit and scope of the invention. Thus, if these modifications and variations of the invention fall within the scope of the claims of the invention and their equivalents, the invention is also intended to include these modifications and variations.

Claims

1. A permanent magnet synchronous motor torque harmonic suppression system, characterized in that: Including PMSM model, iterative learning control module, two-degree-of-freedom control module and genetic algorithm parameter optimization module. The two-degree-of-freedom control module is connected in series with the iterative learning control module, the desired current output by the two-degree-of-freedom control module is added to the compensation current output by the iterative learning control module as the desired current, and the desired current is used as the input of the PMSM model; The iterative learning control module adopts a closed-loop PI-type iterative learning control algorithm; and the genetic algorithm parameter optimization module is used to optimize and adjust the parameters of the iterative learning control module.

2. The permanent magnet synchronous motor torque harmonic suppression system according to claim 1, characterized in that: The iterative learning control module includes a PI controller and an iterative learning controller, and its working process includes: The reference current generated by the PI controller is converted into a reference torque The error between the reference torque and the actual torque is used as the input of the iterative learning controller; Obtain the iteration period through the torque harmonic period; The compensation current is generated by iterating the learning rate, and the output current of the PI controller is compensated by the compensation current.

3. The permanent magnet synchronous motor torque harmonic suppression system according to claim 2, characterized in that: The iterative process of the iterative learning controller is divided into the following four stages: a feedforward control stage, an error preservation stage, a feedback correction stage and an iterative control stage.

4. The permanent magnet synchronous motor torque harmonic suppression system according to claim 2, characterized in that: The control rate of the iterative learning controller includes a forgetting factor.

5. The permanent magnet synchronous motor torque harmonic suppression system according to claim 1, characterized in that: The two-degree-of-freedom control module includes a two-degree-of-freedom controller, and the two-degree-of-freedom controller parameter tuning step includes: firstly adjusting some parameters using a conventional PID parameter tuning method, and then selecting a two-degree-of-freedom coefficient.

6. The permanent magnet synchronous motor torque harmonic suppression system according to claim 5, characterized in that: The two-degree-of-freedom controller adopts a set value filter type two-degree-of-freedom control structure.

7. The permanent magnet synchronous motor torque harmonic suppression system according to claim 1, characterized in that: The implementation steps of the genetic algorithm parameter optimization module include: Initialize the population: set the value range of the parameter and generate the initial population by a random function; Individual evaluation: Set the fitness function and calculate the fitness function value of each individual in the current initial population; Selection operation: Determine whether to keep the fitness function value based on its relative size; Crossover operation: According to the set crossover probability, the selected parent individuals undergo gene crossover to produce new offspring individuals; Mutation operation: Gene mutation is performed on a portion of the offspring individuals to introduce new genetic information; Update population: replace the generated offspring individuals with the original population to form a new population; Termination judgment: If the number of iterations reaches the maximum number of iterations, the set of parameters with the largest fitness function value is selected and output.

8. The permanent magnet synchronous motor torque harmonic suppression system according to claim 7, characterized in that: The fitness function is taken as the inverse of the minimum objective function, and the minimum objective function is the square error value.

9. The permanent magnet synchronous motor torque harmonic suppression system according to claim 8, characterized in that: The fitness function is: Among them, J(x) is the minimum objective function and σ is a custom minimum value.

10. The permanent magnet synchronous motor torque harmonic suppression system according to claim 9, characterized in that: The minimum value σ is 10 -10 .

Images