A fuzzy algorithm-based adaptive control method for permanent magnet synchronous motor

Abstract

The application discloses a permanent magnet synchronous motor control method based on a fuzzy algorithm, which comprises the following steps: firstly, the motor speed is calculated and taken as the speed loop input of a controller; then, three-phase currents of the motor are collected; subsequently, the currents are filtered; a two-section current filtering method is adopted to eliminate current spikes and high-order harmonics, and the filtering result is taken as the current loop input; finally, according to the reference speed input by the system, a disturbance rejection controller is adopted, a nonlinear function in the controller is improved, and parameters in a nonlinear state error feedback control law are corrected on line through a fuzzy algorithm, so that parameter self-adaptation is realized, and then stable and accurate control of the motor is realized.

Description

Technical Field

[0001] This invention belongs to the field of motor control, specifically an adaptive control method for permanent magnet synchronous motors based on fuzzy algorithms. Background Technology

[0002] With the development of power electronics, microelectronics, microcomputers, sensor technology, rare-earth permanent magnet materials, and motor control theory, AC permanent magnet servo control technology has made remarkable progress. Currently, further research on AC permanent magnet servo systems has become a global hot topic, attracting numerous researchers and developers. Its application areas have been gradually expanded, especially in aviation, aerospace, CNC machine tools, machining centers, robotics, and national defense weaponry.

[0003] Permanent magnet synchronous motors (PMSMs) are widely used in high-performance servo system motor drive control due to their high torque-to-current ratio, high power density, low losses, and ease of maintenance. A PMSM is a typical nonlinear, strongly coupled, and time-varying system, difficult to describe with an accurate mathematical model. Common control methods are based on the motor's mathematical model combined with typical PID control, which has proven effective in practice. However, PID controller design relies excessively on the controlled object model, exhibiting poor parameter robustness and failing to meet high-precision control requirements. When a wide speed range is required, it cannot simultaneously meet the requirements of fast response and high steady-state accuracy; the system may even fail to operate normally at low speeds. Modern control theories such as adaptive neural network control and sliding membrane variable structure control can effectively improve the operating performance of PMSMs, but these involve significant computational loads and high control system costs. Summary of the Invention

[0004] The purpose of this invention is to provide an adaptive control method for permanent magnet synchronous motors based on fuzzy algorithms, which can automatically correct the parameters of the controller and has good parameter adaptability, anti-interference, robustness and high control accuracy.

[0005] The technical solution to achieve the purpose of this invention is: an adaptive control method for permanent magnet synchronous motors based on fuzzy algorithms, comprising the following steps:

[0006] Step 1: Use the encoder to obtain the current angle value of the permanent magnet synchronous motor, and with the help of the timer interrupt, calculate the mechanical angle increment value. Then calculate the motor speed at a certain frequency and use it as the speed feedback signal input to the speed loop, and proceed to Step 2.

[0007] Step 2, the current of the permanent magnet synchronous motor is collected by using the ADC, a two-stage current filtering method is adopted, the first stage filtering removes part of the current spikes, and the second stage filtering removes the remaining current spikes and high-order harmonics by dynamically adjusting the filtering coefficient, the calculated filtering result is taken as the current feedback signal input to the current loop, and step 3 is entered.

[0008] Step 3, the reference speed of the permanent magnet synchronous motor is set, the signal error and the differential of the error calculated by using the output of the tracking differentiator and the extended state observer module in the active disturbance rejection controller are taken as the input of the fuzzy algorithm, the fuzzy algorithm outputs three coefficient increments Δβ1, Δβ2 and Δβ3, and the increment threshold is set, the three gain coefficients β0, β1 and β2 in the nonlinear error feedback control law in the active disturbance rejection controller are corrected, the parameter self-adaptation of the active disturbance rejection controller is realized, the nonlinear function of the active disturbance rejection controller is improved, the disturbance compensation of the active disturbance rejection controller is combined, the dynamic correction control quantity is realized, and finally the permanent magnet synchronous motor is accurately controlled.

[0009] Further, in step 1, the certain frequency calculation formula is as follows:

[0010]

[0011] In the formula, f s represents the clock frequency of the timer when calculating the motor speed, T clk represents the input clock frequency of the timer when calculating the motor speed, ARR represents the automatic reload value of the timer, and PSC represents the pre-frequency division coefficient.

[0012] Compared with the prior art, the present application has the following advantages:

[0013] (1) The present application is based on the active disturbance rejection controller, the controller unifies the internal and external disturbances into a lumped disturbance and performs real-time compensation, improves the nonlinear function in the active disturbance rejection controller, and corrects the core parameters in the active disturbance rejection controller online by using the fuzzy algorithm. When the disturbance on the motor is time-varying disturbance, the fuzzy algorithm can flexibly adjust the parameters in the active disturbance rejection controller, so that the active disturbance rejection controller has good self-adaptation ability.

[0014] (2), in the motor current sampling, the sliding recursive median value average filtering algorithm is taken as the first stage current filtering algorithm to filter most of the current spikes and high-order harmonics, in the second stage filtering algorithm, by setting a flag and a counter, when the sampling data changes, the contribution value of the last frame sampling value to the current frame sampling value is dynamically adjusted, and the filtering effect can be timely followed up; when the sampling data tends to be stable, the filtering result gradually approaches and finally equals to the sampling data.

[0015] The application will be further described in detail in combination with the drawings and specific embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 The permanent magnet synchronous motor adaptive control method based on fuzzy algorithm is shown in the principle block diagram.

[0017] Figure 2 The motor speed calculation flow chart is shown in the principle block diagram.

[0018] Figure 3 The current filtering algorithm flow chart is shown in the principle block diagram.

[0019] Figure 4 The permanent magnet synchronous motor control method using fuzzy control algorithm and active disturbance rejection controller is shown in the principle block diagram. DETAILED DESCRIPTION

[0020] In combination Figure 1 The permanent magnet synchronous motor adaptive control method based on fuzzy algorithm includes the following steps:

[0021] Step 1, calculate the motor speed: as shown in the figure, use the encoder to obtain the current angle value of the permanent magnet synchronous motor, cooperate with the timer interrupt, calculate the mechanical angle increment value, and then calculate the photoelectric pod speed at a certain frequency. Figure 2

[0022] The calculation formula at a certain frequency is as follows:

[0023]

[0024] In the formula, f s represents the clock frequency of the timer when calculating the motor speed, T clk represents the input clock frequency of the timer when calculating the motor speed, ARR represents the automatic reload value of the timer, and PSC represents the pre-frequency division coefficient.

[0025] A 14-bit absolute magnetic induction encoder (the calculation frequency of the angle increment is 42MHz, and the calculation frequency of the angular velocity is 2MHz) of AS5048A is used as the motor angle position feedback device, and the current rotor position can be obtained by sending instruction 0xffff to the encoder through SPI protocol, and the size range is 0-16383. Since the magnetic encoder can only feedback the angle position, it needs to cooperate with the timer peripheral to calculate the rotor angular velocity. The main calculation of the timer interrupt is the increment of the electric angle, that is, the difference between the previous time electric angle value and the current time electric angle, and then the electric angle difference is converted into the mechanical angle difference Mec_Angle, the period of the timer is represented as T, and the speed Speed is calculated and input as the speed feedback signal into the speed loop:

[0026] ​

[0027] Go to Step 2.

[0028] Step 2, use ADC to collect the current of permanent magnet synchronous motor, as Figure 3 shown, using two-stage current filtering method, the first stage filtering removes part of the current spikes, introduces data temporary area L, the second stage filtering calculates the filtering result according to the current size and filtering coefficient, and then removes the remaining current spikes and high harmonics, and the filtering result is input as the current feedback signal into the current loop, as follows:

[0029] Step 2-1, use ADC to collect the current of permanent magnet synchronous motor, use two-stage current filtering method to filter the current, go to Step 2-2.

[0030] Step 2-2, in the first stage filtering, the median filtering method is combined with the sliding recursive average filtering method, and is improved to a sliding recursive median average filtering algorithm, that is, a sampling current array is maintained, each time new current sampling data is obtained, the data at the head of the array is cleared, and the new data is inserted at the tail, the average value of the elements between the maximum value and the minimum value in the sampling current array is calculated, the first stage filtering result is obtained, and the first stage filtering result is recorded as data_1, after obtaining the result of the first stage filtering, the size L of the temporary area is dynamically adjusted, so as to improve the sensitivity and smoothness, go to Step 2-3.

[0031] When the difference between the original sampling value and the first stage filtering result value is greater than 0.5 times the original sampling value, decrease L, when the difference between the first stage filtering result value and the original sampling value is greater than 0.5 times the original sampling value, increase L, and L does not change in other cases.

[0032] Step 2-3, the second stage filtering is based on the traditional low-pass filtering algorithm, the difference between the result of the current frame after the first stage filtering algorithm and the result of the second stage filtering algorithm of the last frame is calculated, the result of the second stage filtering of the current frame is recorded as data_2_new, the filtering result of the previous frame is recorded as data_2_last, the size of data_1 and the filtering result of the previous frame data_2_last (the initial value of data_2_last calculated for the first time is 0) is compared, if the difference is not greater than the set difference threshold, go to Step 2-4; if the difference is greater than the difference threshold, go to Step 2-5.

[0033] Step 2-4, when the difference generated by the second stage filtering is not greater than the threshold, judge whether the difference is positive or negative:

[0034] Set the direction flag of the data change as Flag_new, the direction flag of the last frame data change as Flag_last, when the difference data_1-data_2_last is positive, Flag_new is set to positive; when the difference data_1-data_2_last is negative, Flag_new is set to negative; then judge whether the value of the current frame flag Flag_new and the value of the last frame flag are the same:

[0035] When the value of the current frame flag Flag_new and the value of the last frame flag Flag_last are different, the counter is cleared, and the filter coefficient k is restored to the default value, and step 2-7 is entered; when the value of the current frame flag Flag_new and the value of the last frame flag Flag_last are the same, the value of the counter is increased by N to record the change speed of the current, and step 2-6 is entered.

[0036] Step 2-5, when the difference data_1-data_2_last is greater than the set threshold, the value of the counter is increased by M (M>N) to record the change speed of the current, and step 2-6 is entered;

[0037] Step 2-6, judge the size of the current counter value and the set counter threshold: when the value of the counter is greater than the set threshold, it means that the growth speed of the current has reached a certain value, the filter coefficient k is increased, and the contribution value of the last frame filter result data_2_last to the current frame filter result data_2_new is increased, and step 2-7 is entered; when the value of the counter is not greater than the set threshold, step 2-7 is directly entered.

[0038] Step 2-7, according to the current current size and the filter coefficient k, the filter result is calculated, and the filter result is used as the current loop input.

[0039] This filtering method solves the problem that the controller cannot accurately control the motor due to the large deviation of the current sampling value caused by the oscillation and noise in the current sampling process.

[0040] Step 3, set the reference signal of the permanent magnet synchronous motor, as shown in Figure 4 The error and the differential of the error output by the TD module in the active disturbance rejection controller are used as the input of the fuzzy algorithm (the calculation frequency is 42MHz), and three coefficient increments Δβ1, Δβ2, Δβ3 are output by the fuzzy algorithm, and the increment threshold is set. The three gain coefficients β0, β1, β2 in the NLSEF of the active disturbance rejection controller are modified, the parameter adaptation of the active disturbance rejection controller is realized, the disturbance compensation of the active disturbance rejection controller is combined, the non-linear function in the active disturbance rejection controller is improved to have better smoothness, the dynamic correction control amount is realized, and finally the permanent magnet synchronous motor is accurately controlled. The specific steps are as follows:

[0041] Step 3-1, set the reference speed and position of the permanent magnet synchronous motor as the input of the active disturbance rejection controller, and go to step 3-2.

[0042] The active disturbance rejection controller (ADRC) is composed of three parts: tracking differentiator (TD), nonlinear state error feedback control law (NLSEF) and extended state observer (ESO). TD realizes the smooth processing of the input signal through the transition process, reduces the overshoot of the control process output, and gives the transition amount and the differential signal of the transition amount; ESO estimates the uncertain state of the system and the external disturbance according to the output y of the permanent magnet synchronous motor, and outputs it to NLSEF; NLSEF realizes the synchronous compensation of the control amount with the output signals of TD and ESO, and finally outputs the control amount.

[0043] Step 3-2, use the TD module in the active disturbance rejection controller to calculate the error and the differential of the error according to the reference signal input in step 3-1 and the state observation output by the extended state observer, and take them as the input of the fuzzy algorithm, and go to step 3-3.

[0044] The input of TD is the reference signal of the control system, that is, the reference position signal or the reference speed signal, and the nonlinear method is introduced into the differentiator. According to the reference signal, the transition process is arranged and the differential value of this process is provided, which can effectively reduce the initial error of the system and solve the contradiction between the response speed and the overshoot of the system. The transition signal v1(t) and the differential v2(t) of the transition signal output by TD will be used to calculate the input signals e0, e1, e2 of NLSEF, where e1 is the difference between the tracking signal output by the TD module and the estimated signal output by the ESO module, e2 is the difference between the differential of the tracking signal output by the TD module and the differential of the estimated signal output by the ESO module, and e0 is the integral of e1.

[0045] Step 3-3, use the fuzzy algorithm to set the fuzzy domain and fuzzy reasoning method, calculate and output three gain coefficients Δβ0, Δβ1, Δβ2 according to the input error and the differential of the error, and limit the amplitude of the gain coefficients. When the increment is greater than the threshold value, the increment size is the threshold value, and go to step 3-4; when the increment is not greater than the threshold value, the increment size is the actual calculation size, and go to step 3-4.

[0046] Fuzzy algorithm design process:

[0047] A) Determination and fuzzification of input and output

[0048] The input of the fuzzy algorithm is e1, e2, and the output is β0, β1, β2 of the NLSEF part of the ADRC. The fuzzy domain of e1 and e2 is selected as [-1, 1] and [-10, 10] respectively, and the membership function is selected as gaussmf; the fuzzy domain of Δβ0, Δβ1, and Δβ2 is selected as [-0.2, 0.2], [-0.2, 0.2], and [-0.06, 0.06] respectively.

[0049] Considering the precise stability required by the control of the permanent magnet synchronous motor, 7 different fuzzy subsets are defined for each fuzzy domain, denoted as positive big (PB), positive medium (PM), positive small (PS), zero (ZO), negative small (NS), negative medium (NM), and negative big (NB).

[0050] B) Step, making fuzzy control rules

[0051] The fuzzy control rules need to determine the value of the output according to the distribution of the input e1 and e2, so as to realize the adjustment of the parameter value in the ADRC. According to e1 and e2, combined with actual experience and prior knowledge, the adjustment rules of β0, β1, and β2 are designed as shown in Table 1:

[0052] Table 1 Fuzzy control rule table of Δβ0, Δβ1, and Δβ2

[0053]

[0054]

[0055] For example, when e1 and e2 are both NB, Δβ0, Δβ1, and Δβ2 are NB, PB, and PS respectively, i.e. when e1 and e2 are both negative big, according to prior knowledge and actual experience, Δβ0, Δβ1, and Δβ2 are taken as negative big, positive big, and positive small respectively. Since there are two input quantities and three output quantities in this design, each input quantity has seven fuzzy domain subsets, so a total of 49 fuzzy control rules need to be specified.

[0056] C) Step, selecting defuzzification rules

[0057] The fuzzy control quantity obtained after A) step and B) step cannot be directly used as the adjustment quantity by the ADRC, and needs to be defuzzified to an accurate quantity, i.e. defuzzification. The widely used centroid method is used for defuzzification, which has a relatively smooth reasoning output.

[0058] After A), B), and C) steps, the fuzzy controller can output accurate values Δβ0, Δβ1, and Δβ2, which are used to update β0, β1, and β2.

[0059] Step 3-4, using Δβ0, Δβ1, Δβ2 as adjustment amount, and adding processing with three gain coefficients β0, β1, β2, to adjust the size of gain coefficients β0, β1, β2, and go to step 3-5.

[0060] The fuzzy controller outputs three output quantities, Δβ0, Δβ1, Δβ2, and limits the amplitude of Δβ0, Δβ1, Δβ2, as the adjustment value of β0, β1, β2, and the rule of parameter adjustment follows the formula:

[0061]

[0062] And limit the amplitude, when the updated value of β0, β1, β2 is greater than 1.2 times of the initial value, the value is set to 1.2 times of the initial value, and when it is not greater than 1.2 times of the initial value, it is set to the current value of β0, β1, β2.

[0063] Step 3-5, using the updated gain coefficients β0, β1, β2, combining the state observation z1(t), z2(t), z3(t) output by the ESO state observer, to compensate the disturbance and improve the nonlinear function, and finally output the current control quantity to realize the accurate and stable control of the permanent magnet synchronous motor.

[0064] Using ESO to observe the output of permanent magnet synchronous motor, the output signal observation z1(t) and the observation of the differential of the output signal z2(t) are obtained, as well as the disturbance observation z3(t). The input of ESO is the output y of permanent magnet synchronous motor and the product of control quantity u and control quantity gain coefficient b0 output by NLSEF, and the output is z1(t), z2(t), z3(t). z1(t), z2(t) are the output quantity and the differential of the output quantity observed by ESO, and z3(t) is the disturbance, which is compensated in the NLSEF link to simplify the system to a standard integral series system.

[0065] The nonlinear function fal(e, α, δ) determines the nonlinear control performance of the active disturbance rejection controller. In this design, fal(e, α, δ) is improved as follows. This function has better continuity, smoothness and derivability around the origin, and enhances the nonlinear control ability of the active disturbance rejection controller.

[0066]

[0067] In the above formula, e represents the signal error quantity, α is a nonlinear factor, different values affect the nonlinear degree of fal(e, α, β) function, and this parameter is related to the working effect of ESO; δ represents the length of the linear interval of fal(e, α, β) function, the smaller the value, the stronger the anti-interference ability of ESO.

[0068] The output z1(t), z2(t), z3(t) of the ESO will be the input of the NLSEF, which is used to calculate e1, e2 with the output of the TD, and provide the disturbance compensation for the NLSEF.

[0069] After the ESO obtains the output observation and the disturbance of the permanent magnet synchronous motor, the NLSEF is used as the final calculation and output of the system control quantity u0, and the disturbance z3(t) / b0 observed by the ESO and u0 are weighted to obtain the final control quantity u, and b0 represents the control quantity gain coefficient. The NLSEF combines the nonlinear combination of various states of the error together, and through the correction of the fuzzy algorithm, the parameters β0, β1, β2 are always in the best state, and through the flexible combination of proportion, integral and differential, the expansion state observer and the tracking differentiator, the compensation of the disturbance is realized. The input of the NLSEF is the integral of e1, e1 and e2, and the output is u0. The parameters β0, β1, β2 in the NLSEF represent the error, integral and differential gain coefficient, which directly affects the output of the active disturbance rejection controller.

[0070] The NLSEF combines the nonlinear function fal(e, alpha, beta), and finally outputs the current control quantity driver, and then drives the permanent magnet synchronous motor, controls the stable operation of the permanent magnet synchronous motor, realizes the control effect of "small error with large gain, and large error with small gain", and has good robustness.

[0071] The effect of the application is to propose a permanent magnet synchronous motor control method based on fuzzy algorithm, which adopts two-stage parameter adaptive filtering method in current detection, filters out the noise interference in the current signal, adopts fuzzy algorithm in the control method to automatically correct the active disturbance rejection controller online, makes the active disturbance rejection controller have parameter self-correction ability, improves the nonlinear function in the active disturbance rejection controller, and makes the control method proposed by the application have strong adaptability, anti-interference, robustness and high precision. When the permanent magnet synchronous motor encounters various nonlinear disturbances during operation, the motor can run stably.

Claims

1. A permanent magnet synchronous motor adaptive control method based on fuzzy algorithm, characterized in that, The method comprises the following steps: Step 1, the current angle value of the permanent magnet synchronous motor is obtained by using an encoder, a timer interrupt is cooperated, the mechanical angle increment value is calculated, and then the motor speed is calculated at a certain frequency as the speed loop input, and step 2 is entered; Step 2, the current of the permanent magnet synchronous motor is collected by using an ADC, a two-stage current filtering method is adopted, the first stage filtering removes part of the current spikes, the second stage filtering removes the remaining current spikes and high-order harmonics by dynamically adjusting the filtering coefficient, and the calculated filtering result is taken as the current loop input, and step 3 is entered; Step 3, the reference speed and position of the permanent magnet synchronous motor are set, the error and the differential of the error output by the tracking differentiator module and the extended state observer module in the active disturbance rejection controller are taken as the input of the fuzzy algorithm, three coefficient increments Δβ1, Δβ2 and Δβ3 are output by the fuzzy algorithm, and the increment threshold is set, the three gain coefficients β0, β1 and β2 in the nonlinear error feedback control law of the active disturbance rejection controller are modified, the parameter adaptation of the active disturbance rejection controller is realized, the disturbance compensation of the active disturbance rejection controller is combined, the dynamic correction control quantity is realized, and finally the permanent magnet synchronous motor is accurately controlled; In step 1, the certain frequency calculation formula is as follows: where f s represents the clock frequency of the timer when calculating the motor speed, T clk represents the input clock frequency of the timer when calculating the motor speed, ARR represents the timer auto-reload value, and PSC represents the pre-scaling factor; In step 2, the following is specifically provided: Step 2-1, the current of the permanent magnet synchronous motor is collected by using an ADC, a two-stage current filtering method is adopted to filter the current, and step 2-2 is entered; Step 2-2, the first stage filtering adopts a sliding recursive median average filtering method combining a median filtering method and a sliding recursive average filtering method to eliminate part of the current spikes and high-order harmonics, and step 2-3 is entered; Step 2-3, the second stage filtering is based on a traditional low-pass filtering algorithm, the difference between the result of the current frame after the first stage filtering algorithm and the result of the second stage filtering algorithm of the last frame is calculated, if the difference is not greater than the set threshold, step 2-4 is entered; if the difference is greater than the threshold, step 2-5 is entered; Step 2-4, when the difference is not greater than the threshold, it is judged whether the difference is positive or negative: When the difference is positive, the flag is positive; When the difference is negative, the flag is negative; Then it is judged whether the values of the current frame flag and the last frame flag are the same: When the values of the current frame flag and the last frame flag are different, the counter is cleared, the filtering coefficient is restored to the default value, and step 2-7 is entered; when the values of the current frame flag and the last frame flag are the same, the value of the counter is increased by N to record the change speed of the current, and step 2-5 is entered; Step 2-5, when the difference is greater than the set threshold, the value of the counter is increased by M to record the change speed of the current, M>N, and step 2-6 is entered; Step 2-6, it is judged whether the value of the current counter is greater than the set counter threshold: when the value of the counter is greater than the set threshold, the filtering coefficient is increased, and then the contribution value of the last frame filtering result to the current frame filtering result is increased, and step 2-7 is entered; when the value of the counter is not greater than the set threshold, step 2-7 is directly entered; Step 2-7, the second stage filtering result is calculated according to the current first stage filtering result and the filtering coefficient, and the filtering result is taken as the current loop input. The step 3 is specifically as follows: Step 3-1, setting the reference speed and position of the permanent magnet synchronous motor as the input of the active disturbance rejection controller, and turning to step 3-2; Step 3-2, using the tracking differentiator module in the active disturbance rejection controller, according to the reference signal input in step 3-1 and the state observation output by the extended state observer, calculating the error and error differential of the signal, and taking them as the input of the fuzzy algorithm, and turning to step 3-3; Step 3-3, using the fuzzy algorithm, setting the fuzzy domain and fuzzy reasoning method, according to the input error and error differential, calculating and outputting three gain coefficient increments Δβ1, Δβ2 and Δβ3, and limiting the amplitude of the increments, when the increment is greater than the threshold value, the increment size is the threshold value, and turning to step 3-4; when the increment is not greater than the threshold value, the increment size is the actual calculation size, and turning to step 3-4; Step 3-4, using Δβ1, Δβ2 and Δβ3 as the adjustment amount, and adding β0, β1 and β2 to adjust the size of the gain coefficients β0, β1 and β2, and turning to step 3-5; Step 3-5, using the corrected three gain coefficients β0, β1 and β2 to weight the error and error differential, combining the disturbance compensation function of the active disturbance rejection controller, to realize the accurate and stable control of the permanent magnet synchronous motor.

2. The adaptive control method for permanent magnet synchronous motor based on fuzzy algorithm according to claim 1, characterized in that: Using the fuzzy algorithm as the parameter correction algorithm of the active disturbance rejection controller, the input of the fuzzy algorithm is the output of the tracking differentiator in the active disturbance rejection controller, and the three gain coefficient increments output by the fuzzy algorithm are limited in amplitude, and the output of the fuzzy algorithm is taken as the increment value of the integral, error and differential gain coefficients in the active disturbance rejection controller.

Images