Dead-beat control method of permanent magnet synchronous motor based on PSA algorithm parameter identification

Abstract

The invention discloses a dead-beat control method for a permanent magnet synchronous motor based on PSA algorithm parameter identification, and the method comprises the steps: firstly, providing an improved current prediction algorithm based on a mathematical model of the permanent magnet synchronous motor, so as to eliminate one-beat delay in current command calculation; then, current fluctuation and static errors are reduced by combining weight coefficients; and finally, on the basis, parameter identification is carried out by utilizing a PID search algorithm (PSA), and resistance, flux linkage and inductance parameters are accurately identified, so that current static errors caused by parameter mismatching are eliminated. According to the method, the problems that the traditional permanent magnet synchronous motor dead-beat current prediction control depends on the accuracy of motor parameters, current deviation is easy to generate and current fluctuation is large are solved, and the robustness and the control precision of the system are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of permanent magnet synchronous motor speed control, and particularly relates to a deadbeat control method for a permanent magnet synchronous motor based on parameter identification by the PSA algorithm. Background Art

[0002] Permanent Magnet Synchronous Motors (PMSMs) are widely used in industrial automation, electric vehicles, home appliances and other fields due to their advantages such as high efficiency, low noise and high power density. The current control of PMSMs is one of its core technologies, and the fast response of the current loop directly affects the stability and dynamic performance of the motor. Therefore, researching and optimizing current control strategies, especially Deadbeat Predictive Current Control (DPCC), is of great significance for improving the overall performance of PMSMs.

[0003] Currently, traditional current control methods, such as Proportional Integral (PI) control and hysteresis current control, although they can reduce the steady-state error and achieve a relatively fast response to a certain extent, their dynamic performance and robustness are still limited. Especially when the motor parameters change or the external disturbance is large, these methods often cannot guarantee the stability and accuracy of the system. Therefore, how to improve the current control strategy to cope with motor parameter changes and external disturbances has become the focus of research.

[0004] To solve these problems, the DPCC method based on predictive control has gradually become a research hotspot. DPCC achieves more accurate current tracking by optimizing the selection of voltage vectors and has good dynamic performance. However, in practical applications, the control accuracy and robustness of DPCC are often affected by motor parameter mismatches and external disturbances. Therefore, combining high-precision parameter identification techniques (such as the PID Search Algorithm, PSA algorithm) and an improved DPCC control structure to improve the robustness and accuracy of the system is an important direction for improving the performance of PMSM control systems. Summary of the Invention

[0005] The purpose of the present invention is to provide a deadbeat control method for a permanent magnet synchronous motor based on parameter identification by the PSA algorithm, which can eliminate one-beat delay in current command calculation, accurately identify resistance, flux linkage and inductance parameters, eliminate current static error caused by parameter mismatches, and improve the robustness and control accuracy of the system.

[0006] To achieve the above purpose, the technical solution adopted by the present invention is:

[0007] A deadbeat control method for a permanent magnet synchronous motor based on parameter identification of the PSA algorithm, comprising the following steps:

[0008] S1. Establish a mathematical model of the permanent magnet synchronous motor, discretize the mathematical model to obtain the traditional DPCC state equation, construct a current prediction algorithm equation for eliminating one-beat delay, and simultaneously introduce weight coefficients α and β, and then combine them with the traditional DPCC state equation to obtain the DPCC with improved weight coefficients;

[0009] S2. Adopt the PID search algorithm to identify the parameters of the permanent magnet synchronous motor, calculate the difference between the control quantities of the motor at the previous moment and the current moment through the incremental PID controller, and use the difference as the new control quantity to dynamically adjust the motor parameters;

[0010] S3. Substitute the optimal parameters of the motor identified by the PID search algorithm into the DPCC with improved weight coefficients for effective compensation to eliminate the static current error.

[0011] Further, in S1, after transforming the traditional DPCC state equation, obtain the current equation at the k+1 moment, and based on this, establish the state observation coefficient matrix equation: U(k) = AI(k) + LΔI(k) + D Among them,

[0012] In the formula, U(k) is the dq-axis voltage matrix at the k moment, I(k) is the dq-axis current matrix at the k moment, ΔI(k) is the dq-axis deviation current matrix at the k moment, L represents a two-row and two-column matrix, L s and R s are respectively the stator inductance and stator resistance of the motor, ψ f is the magnetic flux of the motor, ω e is the electrical angular velocity of the rotor, T s is the sampling period, i d * (k) and i q * (k) are respectively the given currents of the d-axis and q-axis at the k moment, i d (k) and i q (k) are respectively the d-axis component and q-axis component of the stator current at the k moment.

[0013] Further, predict the current at the k+1 moment according to the state observation coefficient matrix equation, eliminate the one-beat delay of the current command calculation, and obtain the improved current prediction algorithm equation: I Pref (k + 1) = I(k + 2) - A c L -1[LΔI(k)+AI(k)-U(k)+D] Wherein, A C = L -1 *A - E

[0014] In the formula, I Pref (k + 1) is the predicted current matrix at the next beat k + 1 after improvement, I(k + 2) is the dq-axis current matrix at the k + 2 moment, L represents a two-row and two-column matrix, E is the identity matrix, ΔI(k) is the dq-axis deviation current matrix at the k moment, I(k) is the dq-axis current matrix at the k moment, and U(k) is the dq-axis voltage matrix at the kk moment.

[0015] Furthermore, in the improved current prediction algorithm equation, weight coefficients α and β are introduced and then combined with the traditional DPCC state equation to obtain the improved voltage prediction equation at the k moment:

[0016] In the formula, u d (k) and u q (k) are the improved voltages of the d-axis and q-axis at the k moment respectively, L s and R s are the stator inductance and stator resistance of the motor respectively, T s is the sampling period, ψ f is the motor magnetic flux, α and β are weight coefficients, i d (k) and i q (k) are the d-axis component and q-axis component of the stator current at the k moment respectively, i d * (k + 1) and i q * (k + 1) are the given currents of the d-axis and q-axis at the k + 1 moment respectively, i dpref (k + 1) and i qpref (k + 1) are the predicted currents of the d-axis and q-axis at the k + 1 moment after one beat delay, and ω e (k) is the electrical angular velocity of the rotor at the k moment.

[0017] Furthermore, in S1, the dynamic performance of the DPCC with improved weight coefficients is analyzed to obtain the closed-loop discrete transfer function of the actual current i dq and the given current i dq * : Furthermore, the stable interval of the DPCC with improved weight coefficients is obtained as:

[0018] In the formula, idq (z) is the feedback current of the dq axis under the change of z, i dq * (z) is the given current of the dq axis under the change of z, z is the closed-loop pole, α and β are weight coefficients, L s is the stator inductance of the motor, L motor is the actual inductance of the motor.

[0019] Furthermore, in the step S2, combining the physical characteristics of the inductance, resistance, and flux linkage parameters of the motor, a fitness function, a reference model, and an adjustable model are designed to evaluate the error and optimization effect in the parameter identification process.

[0020] Furthermore, in the step S2, when the iteration number is t, the output value Δu(t) of the PID regulation is: Δu(t) = K p ·r 2 ·[e k (t) - e k-1 (t)] + K i ·r 3 ·e k (t) + K d ·r 4 ·[e k (t) - 2e k-1 (t) + e k-2 (t)]

[0021] In the formula, Δu(t) is the control quantity increment at the current moment, r 2 , r 3 and r 4 are vectors of random numbers from 0 to 1 in an n-row 1-column; K p , K i and K d are the adjustment coefficients of proportional, integral, and differential respectively, e k (t), e k-1 (t), e k-2 (t) are the errors at the kth moment, (k - 1)th moment, and (k - 2)th moment respectively.

[0022] Furthermore, in the step S3, the process of obtaining the optimal parameters of the motor is as follows: The PID search algorithm continuously calculates the error between the output values of the motor reference model and the adjustable model through the fitness function. The closer the error value is to 0, the closer the identified parameters are to the actual values.

[0023] Furthermore, the reference model is:

[0024] The adjustable model is:

[0025] Wherein, x represents the system state variable, u represents the system input, p represents the system parameter, y represents the system output, respectively represent the state variable, parameter and output in the adjustable model; f and g are the state equation and output equation of the system, respectively represent the first-order differential of the state variable in the reference model and the first-order differential of the state variable in the adjustable model.

[0026] Furthermore, the fitness function is:

[0027] Wherein, w 1 , w 2 and w 3 are weight factors, u d0 (k) and u q0 (k) are the voltages of the d-axis and q-axis sampled at the current moment respectively, T e * (k) is the electromagnetic torque value of the dq-axis sampled at the current moment, are the voltages of the d-axis and q-axis calculated according to the identification result respectively, T e (k) is the electromagnetic torque value of the dq-axis calculated according to the identification result, and n is the number of iterations.

[0028] Adopting the above technical solutions, the present invention can achieve the following beneficial effects:

[0029] (1) The present invention first proposes an improved current prediction algorithm based on the mathematical model of the permanent magnet synchronous motor to eliminate the one-beat delay in the calculation of the current command; then reduces the current fluctuation and static error by combining the weight coefficients; finally, on this basis, uses the PID search algorithm (PSA) for parameter identification to accurately identify the resistance, flux linkage and inductance parameters to eliminate the current static error caused by parameter mismatch. This method overcomes the problems that the traditional deadbeat current prediction control of the permanent magnet synchronous motor is affected by the accuracy of the motor parameters and is prone to current deviation and large current fluctuation, and improves the robustness and control accuracy of the system.

[0030] (2) The present invention adopts a multi-parameter identification method of PMSM based on PSA, substitutes the accurate parameters identified by PSA into the DPCC with improved weight coefficients for compensation, thereby completely eliminating the current static error. The experimental results show that the DPCC algorithm of the present invention has obvious advantages over the traditional algorithms in terms of electromagnetic torque, stator current control, inductance identification and current fluctuation suppression, especially in terms of current control and system stability. Description of the Drawings

[0031] Figure 1 is Ls = 2L motor Schematic diagram of the traditional DPCC simulation results when

[0032] Figure 2 is L s = 2L motor Schematic diagram of the DPCC simulation results of the improved weight coefficient of the present invention when

[0033] Figure 3 is R s = 4R motor Comparison chart of the id and iq ripple results of the traditional DPCC and the improved weight coefficient DPCC of the present invention when

[0034] Figure 4 is ψ f = 1.5ψ motor Comparison chart of the id and iq ripple results of the traditional DPCC and the improved weight coefficient DPCC of the present invention when

[0035] Figure 5 is L s = 2L motor Comparison chart of the recognition results when

[0036] Figure 6 is R s = 4R motor Comparison chart of the recognition results when

[0037] Figure 7 is ψ f = 1.5ψ motor Comparison chart of the recognition results when

[0038] Figure 8 is L s = 2L motor Id and iq ripple results of the compensated traditional DPCC and the compensated improved weight coefficient DPCC when

[0039] Figure 9 is R s = 4R motor Comparison chart of the id and iq ripple results of the compensated traditional DPCC and the compensated improved weight coefficient DPCC when

[0040] Figure 10 is ψ f = 1.5ψ motor Comparison chart of the id and iq ripple results of the compensated traditional DPCC and the compensated improved weight coefficient DPCC when Detailed implementation manner

[0041] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0042] It should be noted that the following detailed description is illustrative and is intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present application belongs.

[0043] The present invention provides a deadbeat control method for a permanent magnet synchronous motor based on parameter identification of the PSA algorithm, including the following steps:

[0044] S1. Establish a mathematical model of the permanent magnet synchronous motor, discretize the mathematical model to obtain a traditional DPCC state equation, construct a current prediction algorithm equation for eliminating one-beat delay, and simultaneously introduce weight coefficients α and β, and then combine it with the traditional DPCC state equation to obtain a DPCC with improved weight coefficients;

[0045] S2. Adopt a PID search algorithm to identify the parameters of the permanent magnet synchronous motor, calculate the difference between the control quantities of the motor at the previous moment and the current moment through an incremental PID controller, and use the difference as a new control quantity to dynamically adjust the motor parameters;

[0046] S3. Substitute the optimal parameters of the motor identified by the PID search algorithm into the DPCC with improved weight coefficients for effective compensation to eliminate the static current error.

[0047] In order to establish the d-axis and q-axis mathematical models of the PMSM, first assume:

[0048] (1) Ignore the saturation of the iron core;

[0049] (2) Do not consider eddy current and hysteresis losses;

[0050] (3) The stator current is a symmetric three-phase sinusoidal current.

[0051] Step S1. Establish a mathematical model of the permanent magnet synchronous motor in the d-q rotating coordinate system.

[0052] In this embodiment, in the synchronous rotating coordinate system of the surface-mounted permanent magnet synchronous motor, the mathematical model of the PMSM (i.e., the stator voltage equation) is as shown in Equation (1):

[0053] In the formula, i d and i q are respectively the d-axis component and the q-axis component of the motor stator current, u d and u q are respectively the d-axis component and the q-axis component of the motor stator voltage, L s and R s are respectively the stator inductance and the stator resistance of the motor, ψ f is the motor magnetic flux, ωe is the electrical angular velocity of the rotor.

[0054] Take a period T s Within it, discretize Equation (1) to obtain the traditional DPCC state equation, as shown in Equation (2):

[0055] Transform Equation (2) to obtain the current equation at time k + 1, as shown in Equation (3):

[0056] In the formula, u d (k) and u q (k) are the d-axis component and q-axis component of the stator voltage at time k respectively, i d (k) and i q (k) are the d-axis component and q-axis component of the stator current at time k respectively, i d (k + 1) and i q (k + 1) are the d-axis component and q-axis component of the stator current at time k + 1 respectively, ω e (k) is the electrical angular velocity of the rotor at time k, T s is the sampling period, ψ f is the motor magnetic flux, p n is the number of pole pairs of the motor.

[0057] Based on the current equation at time k + 1, establish the state observation coefficient matrix equation, as shown in Equation (4): U(k) = AI(k) + LΔI(k) + D (4)

[0058] Among them,

[0059] In the formula, U(k) is the dq-axis voltage matrix at time k, I(k) is the dq-axis current matrix at time k, ΔI(k) is the dq-axis deviation current matrix at time k, L represents a two-row and two-column matrix, L s and R s are the stator inductance and stator resistance of the motor respectively, ψ f is the motor magnetic flux, ω e is the electrical angular velocity of the rotor, T s is the sampling period, i d * (k) and i q * (k) are the given currents of the d-axis and q-axis at time k respectively, i d (k) and i q (k) are the d-axis component and q-axis component of the stator current at time k respectively.

[0060] Predict the current at the (k + 1)-th moment through Equation (4), thus obtaining the predicted current equation at the (k + 1)-th moment, as shown in Equation (5): I p (k + 1) = I(k) + L -1 [U(k) - AI(k) - D] (5)

[0061] In the formula, I p (k + 1) is the predicted current matrix at the next beat (k + 1)-th moment, I(k) is the dq-axis current matrix at the k-th moment, U(k) is the dq-axis voltage matrix at the k-th moment, and L represents a 2×2 matrix.

[0062] Since the code amount of the voltage command optimization part is small, it basically has no impact on the digital control system. If the control command of the current changes, the system dynamically adjusts the current reference value to prevent the control system from generating incorrect or unstable current due to the command change. At this time, the compensated voltage equation is as shown in Equation (6): ΔU p (k + 1) = LΔI ref (k) (6)

[0063] Among them,

[0064] In the formula, ΔU p (k + 1) is the dq-axis deviation voltage matrix at the (k + 1)-th moment, ΔI ref (k) is the dq-axis deviation current matrix at the k-th moment, L represents a 2×2 matrix, i d * (k + 1) and i q * (k + 1) are the given currents on the d-axis and q-axis at the (k + 1)-th moment respectively, i d * (k) and i q * (k) are the given currents on the d-axis and q-axis at the k-th moment respectively.

[0065] Substitute Equation (6) into Equation (4) to obtain the predicted voltage equation, as shown in Equation (7):

[0066] In the formula, U p (k + 1) is the predicted voltage matrix on the dq-axis at the (k + 1)-th moment, I p (k + 1) is the predicted current matrix on the dq-axis at the (k + 1)-th moment, ΔI p (k + 1) is the dq-axis deviation current matrix at the (k + 1)-th moment, I ref (k + 1) is the given current matrix on the dq-axis at the (k + 1)-th moment.

[0067] Predict the current at the (k + 1)-th moment according to the state observation coefficient matrix equation, eliminate the one-beat delay in the current command calculation, and obtain the improved current prediction algorithm equation.

[0068] Substitute Equation (4) and Equation (5) into Equation (7) to obtain the improved current prediction algorithm equation that eliminates one-beat delay, as shown in Equation (8): I Pref (k + 1) = I(k + 2) - A c L -1 [LΔI(k) + AI(k) - U(k) + D] (8)

[0069] Wherein, A C = L -1 *A - E (9)

[0070] In the formula, I Pref (k + 1) is the predicted current matrix at the next beat (k + 1)-th moment after improvement, I(k + 2) is the dq-axis current matrix at the (k + 2)-th moment, L represents a two-row and two-column matrix, E is the identity matrix, ΔI(k) is the dq-axis deviation current matrix at the k-th moment, I(k) is the dq-axis current matrix at the k-th moment, and U(k) is the dq-axis voltage matrix at the k-th moment.

[0071] When the model inductance parameter is greater than twice the actual inductance parameter, the traditional deadbeat current prediction control algorithm will not converge. To address this problem, a feedforward current and a weight coefficient are introduced to correct the feedback current, a deadbeat control strategy with one-beat delay combined with an improved weight coefficient is designed, and a stability analysis is carried out.

[0072] According to the derived formula (8), two weight coefficients α and β are added, and the relationship formula of α and β is as follows:

[0073] Combine Equation (2) with Equation (8), set the weight of the stator current at the k-th moment to α, and set the weight of the predicted current at the (k + 1)-th moment after one-beat delay to β, to obtain the improved voltage prediction equation at the k-th moment, as shown in Equation (11):

[0074] In the formula, u d (k) and u q (k) are the improved d-axis and q-axis voltages at the k-th moment respectively, L s and R s are the stator inductance and stator resistance of the motor respectively, T s is the sampling period, ψ f is the motor magnetic flux, α and β are weight coefficients, id (k) and i q (k) are the d - axis component and q - axis component of the stator current at time k, respectively, and i d * (k + 1) and i q * (k + 1) are the reference currents of the d - axis and q - axis at time k + 1, respectively, and i dpref (k + 1) and i qpref (k + 1) are the predicted currents of the d - axis and q - axis at time k + 1 after one - step delay, and ω e (k) is the electrical angular velocity of the rotor at time k.

[0075] Substitute Equation (3) into Equation (11) to obtain Equation (12), that is, the relationship between the feedback current and the reference current after adding the weight coefficients is obtained.

[0076] In the formula, i d (k + 1) and i q (k + 1) are the feedback currents of the d - axis and q - axis at time k + 1, respectively, and i d * (k + 1) and i q * (k + 1) are the reference currents of the d - axis and q - axis at time k + 1, respectively, and L motor , Ψ motor , R motor are the actual inductance, flux linkage, and resistance of the motor, respectively, α and β are weight coefficients, and i dpref (k + 1) and i qpref (k + 1) are the predicted currents of the d - axis and q - axis at time k + 1 after one - step delay.

[0077] For the convenience of calculation, let the predicted currents of the d - axis and q - axis at time k + 1 be approximately equal to the reference currents of the d - axis and q - axis at time k, that is, i dqpref (k + 1)≈i dq * (k), and the closed - loop discrete transfer function of i dq and i dq * can be obtained:

[0078] In the formula, i dq (z) is the feedback current of the dq - axis under the change of z, and i dq * (z) is the reference current of the dq - axis under the change of z, z is the closed - loop pole, α and β are weight coefficients, and L s is the stator inductance of the motor.

[0079] The stable interval of the DPCC with the improved weight coefficient is obtained according to formula (13) as follows:

[0080] where L motor is the actual inductance of the motor.

[0081] When the system reaches steady-state operation, the static current difference between the actual feedback value of the current and the given value after introducing the weight coefficient is:

[0082] The analysis of equations (14) and (15) shows that when the deviation of the inductance parameter exceeds 2 times, the traditional control system will lose stability, resulting in current divergence. After introducing robust current control, the decrease of the weight coefficient 1 / α improves the system stability, but sacrifices the dynamic response, resulting in an increase in the steady-state current error. To solve the problem of poor current tracking caused by motor parameter deviation and the introduction of the robust factor, the present invention adopts a PID-based search algorithm (PSA) to identify resistance, inductance, and flux linkage.

[0083] Step S2: Adopt the PID search algorithm to identify the parameters of the permanent magnet synchronous motor. Calculate the difference between the control quantities of the motor at the previous moment and the current moment through the incremental PID controller, and use the difference as the new control quantity to dynamically adjust the motor parameters.

[0084] The calculation formula of the incremental PID controller is as follows: Δu(t) = K p ·[e(t) - e(t - 1)] + K i ·e(t) + K d ·[e(t) - 2e(t - 1) + e(t - 2)] (16)

[0085] where Δu(t) is the control quantity increment at the current moment, K p , K i , K d are the proportional, integral, and differential coefficients respectively, and e(t), e(t - 1), e(t - 2) are the errors at the current moment, the previous moment, and the two previous moments respectively.

[0086] By continuously updating the control quantity through the recursive algorithm, the incremental PID controller can quickly respond to the dynamic changes of the system, adjust the parameters in a timely manner, and make the system output stable near the target value.

[0087] The optimization problem of the PID search algorithm consists of a set of decision variables, constraint conditions, and objective functions. It can be assumed that the number of the decision variable set is d, and the upper and lower limits of the variable are u and l respectively: X ij = (uj -l j )*r 1 +l j , i = 1, 2, ..., n; j = 1, 2, ..., d (17)

[0088] Among them, X ij represents the j-th dimension of the i-th individual; u j and l j are the upper and lower limits of the j-th variable (dimension) respectively; r 1 is a random number between 0 and 1.

[0089] For the minimization problem, the best individual x * (t) at the iteration number t is the individual corresponding to the overall historical minimum. The overall deviation e k (t) for multiple iterations t is:[[]] e k (t) = x * (t - 1) - x(t - 1) (18)

[0090] In the formula, e k (t) is the overall deviation at the t-th iteration, x * (t - 1) represents the best individual at the iteration number t - 1, and x(t - 1) represents the individual at the iteration number t - 1.

[0091] In practical problems, the proportional, integral, and differential factors are adjusted according to different situations and problems. Then, when the iteration number is t, the output value Δu(t) of the PID regulation is:[[]] Δu(t) = K p ·r 2 ·[e k (t) - e k-1 (t)] + K i ·r 3 ·e k (t) + K d ·r 4 ·[e k (t) - 2e k-1 (t) + e k-2 (t)] (19)

[0092] In the formula, Δu(t) is the control quantity increment at the current moment, r 2 , r 3 and r 4 are vectors of random numbers from 0 to 1 in an n-row 1-column; K p , K i and K d are the adjustment coefficients for proportional, integral, and differential respectively, e k (t), ek-1 (t), e k-2 (t) are the errors at time k, time k - 1, and time k - 2 respectively.

[0093] For all individuals, the update is related to Δu(t) and η based on the original Δu(t). The population update formula is defined as: x(t + 1) = x(t) + η·Δu(t) (20)

[0094] In the formula, η is an n - row and 1 - column matrix, expressed as: η = r 5 cos(t / T) (21)

[0095] In the formula, r 5 is a matrix of random numbers from 0 to 1 in an n - row and 1 - column.

[0096] Step S3: Substitute the optimal parameters of the motor identified by the PID search algorithm into the DPCC with the improved weight coefficient for effective compensation to eliminate the static current error.

[0097] The problem of PMSM parameter identification can be transformed into a system optimization problem. Its basic idea is to identify the motor parameters through an identification algorithm and continuously adjust the parameters of the adjustable model according to the difference between the output of the reference model and the adjustable model.

[0098] The dynamic reference model of the motor in the present invention is shown in Equation (22). To identify the unknown parameter p, an adjustable model with the same structure is required, such as Equation (23):

[0099] In the formula, x represents the system state variable, u represents the system input, p represents the system parameter, y represents the system output, represent the state variable, parameter, and output in the adjustable model respectively; f and g are the state equation and output equation of the system, represent the first - order differential of the state variable in the reference model and the first - order differential of the state variable in the adjustable model respectively.

[0100] The PSA algorithm continuously calculates the error between the output values of the motor reference model and the adjustable model through the fitness function (i.e., the fitting function). The closer the fitting value is to 0, the closer the identified parameter is to the actual value, and thus the optimal parameters of the motor can be obtained. The fitness function is:

[0101] In the formula, w 1 , w 2 and w 3 are weight factors, u d0 (k) and u q0(k) are the voltages of the d-axis and q-axis sampled at the current moment, T e * (k) is the electromagnetic torque value of the dq-axis sampled at the current moment, are the voltages of the d-axis and q-axis calculated according to the identification results, T e (k) is the dq-axis electromagnetic torque value calculated according to the identification results, and n is the number of iterations.

[0102] The motor parameters identified by the PSA algorithm are substituted into the DPCC with improved weight coefficients for effective compensation, overcoming the current static error that appears in the traditional deadbeat current predictive control under the condition of motor parameter mismatch.

[0103] In this example, the parameters of the permanent magnet synchronous motor are selected as the rated power P N = 1.2 kW, the rated current I N = 9 A, the rated speed is 3000 r / min, the rated torque is 4 N·m, the number of pole pairs p n = 4, the stator resistance R s = 0.525 Ω, the stator inductance L s = 1.65 mH, the rotor magnetic flux ψ f = 0.0744 Wb, the inertia J = 0.000153 kg·m 2 .

[0104] The number of iterations of PSA is 70; the particle swarm size is 70; Kp = 1.2; Ki = 8; Kd = 0.2; w 1 = 0.3; w 2 = 0.3; w3 = 0.4. L motor , Ψ motor , R motor are the actual inductance, magnetic flux, and resistance of the motor respectively.

[0105] Experiment 1: Simulation comparison between the traditional DPCC and the improved weight coefficient DPCC of the present invention when L s = 2L motor .

[0106] In the experiment based on MATLAB / Simulink, we built a permanent magnet synchronous motor model based on L s = 2L motor , and carried out simulations of the traditional DPCC and the improved weight coefficient DPCC of the present invention respectively. The experimental results are as shown in Figure 1 and Figure 2As shown. Compared with the traditional algorithm, the DPCC based on the improved weight coefficient significantly reduces the fluctuations of the electromagnetic torque Te, d-axis current id, and q-axis current iq. Therefore, the improved weight coefficient algorithm of the present invention shows obvious advantages in electromagnetic torque and stator current control, demonstrating its effectiveness.

[0107] Experiment 2: R s = 4R motor Simulation comparison between traditional DPCC and the improved weight coefficient DPCC of the present invention.

[0108] In another experiment based on MATLAB / Simulink, we built a permanent magnet synchronous motor model with R s = 4R motor and conducted simulations of traditional DPCC and improved weight coefficient DPCC. The results are as shown in Figure 3 (a) and (b), where (a) is traditional DPCC and (b) is improved weight coefficient DPCC. The experimental results show that the DPCC based on the improved weight coefficient can effectively reduce the fluctuations of id and iq, is superior to the traditional algorithm, and demonstrates its advantages in current control.

[0109] Experiment 3: ψ f = 1.5ψ motor Simulation comparison between traditional DPCC and the improved weight coefficient DPCC of the present invention.

[0110] In the permanent magnet synchronous motor experiment based on ψ f = 1.5ψ motor we respectively conducted simulations of id and iq for traditional DPCC and improved weight coefficient DPCC. The experimental results are as shown in Figure 4 (a) and (b), where (a) is traditional DPCC and (b) is improved weight coefficient DPCC. Compared with the traditional algorithm, the DPCC based on the improved weight coefficient can significantly reduce the fluctuations of d-axis and q-axis currents, improving the stability and accuracy of current control.

[0111] Experiment 4: L s = 2L motor Comparison of inductance value identification between PSA and PSO (Particle Swarm Optimization) algorithms when

[0112] In the simulation based on MATLAB / Simulink, we built a permanent magnet synchronous motor model with L s = 2L motor and used the PSA algorithm and PSO algorithm to identify the inductance values of traditional DPCC and improved weight coefficient DPCC. The experimental results are as shown in Figure 5As shown. Experiments show that the PSA algorithm is significantly superior to the PSO algorithm in terms of the accuracy and convergence speed of inductance value identification, with stronger identification ability and faster convergence speed.

[0113] Experiment Five: L s = 2L motor Comparison of the fluctuations of id and iq between the traditional DPCC after compensation and the improved weight coefficient DPCC of the present invention when L = 2L.

[0114] When L s = 2L motor In the simulation of a permanent magnet synchronous motor, we used the PSA algorithm for compensation and compared the fluctuations of id and iq between the traditional DPCC and the improved weight coefficient DPCC. The experimental results are as Figure 8 shown, where (a) is the traditional DPCC after compensation and (b) is the improved weight coefficient DPCC after compensation. Experiments show that after compensation, the fluctuations of id and iq currents of the improved weight coefficient DPCC are significantly smaller than those of the traditional DPCC, demonstrating the significant advantage of the improved algorithm in dynamic performance.

[0115] Experiment Six: R s = 4R motor Comparison of the resistance value identification between the PSA and PSO algorithms when R = 4R.

[0116] When based on R s = 4R motor In the experiment of a permanent magnet synchronous motor, we used the PSA algorithm and the PSO algorithm to identify the resistance values of the traditional DPCC and the improved weight coefficient DPCC. The experimental results are as Figure 6 shown. Through comparison, experiments show that the PSA algorithm is significantly superior to the PSO algorithm in terms of accuracy and convergence speed, proving its advantage in resistance value identification.

[0117] Experiment Seven: R s = 4R motor Comparison of the fluctuations of id and iq between the traditional DPCC after compensation and the improved weight coefficient DPCC of the present invention when R = 4R.

[0118] When based on R s = 4R motor In the simulation of a permanent magnet synchronous motor, we compensated the traditional DPCC and the improved weight coefficient DPCC respectively. The results are as Figure 9 shown, where (a) is the traditional DPCC after compensation and (b) is the improved weight coefficient DPCC after compensation. The experimental results show that after using the PSA algorithm to identify the inductance value, the current fluctuations of the improved weight coefficient DPCC on the id and iq axes are significantly reduced compared with the traditional DPCC after compensation, and the static deviation on the iq axis is eliminated, demonstrating good dynamic performance.

[0119] Experiment VIII: ψ f = 1.5ψ motor Comparison of the PSA and PSO algorithms of the motor for the identification of the flux linkage value.

[0120] At ψ f = 1.5ψ motor In the experiment of the permanent magnet synchronous motor, we used the PSA algorithm and the PSO algorithm to identify the flux linkage value of the traditional DPCC and the improved weight coefficient DPCC respectively. The experimental results are as Figure 7 shown. The experiment shows that the PSA algorithm is significantly superior to the PSO algorithm in terms of the accuracy and convergence speed of the flux linkage value identification, which proves the superiority of the PSA algorithm.

[0121] Experiment IX: ψ f = 1.5ψ motor Comparison of the fluctuations of id and iq of the compensated traditional DPCC and the improved weight coefficient DPCC under the condition.

[0122] At ψ f = 1.5ψ motor In the experiment of the permanent magnet synchronous motor, we compared the performances of the compensated traditional DPCC and the improved weight coefficient DPCC in terms of the fluctuations of id and iq. The experimental results are as Figure 10 shown, where (a) is the compensated traditional DPCC and (b) is the compensated improved weight coefficient DPCC. The results show that after being identified and compensated by PSA, the fluctuations of the id and iq axis currents of the improved weight coefficient DPCC are significantly smaller than those of the traditional DPCC, and the static deviations of id and iq are effectively eliminated, further improving the dynamic performance of the system.

[0123] Through these experiments and combined with the MATLAB / Simulink simulation platform, we verified the significant advantages of the DPCC algorithm based on the improved weight coefficient compared with the traditional algorithm in aspects such as electromagnetic torque, stator current control, inductance identification, and current fluctuation suppression. All experimental results prove the superiority of the improved weight coefficient DPCC algorithm based on PSA in terms of dynamic performance, especially the significant improvement in current control and system stability.

[0124] Finally, it should be noted that the parts not described in detail in the present invention are all prior arts. Those of ordinary skill in the art can understand that the above are only the preferred examples of the invention and are not used to limit the invention. Although the invention has been described in detail with reference to the foregoing examples, for those skilled in the art, they can still modify the technical solutions recorded in the foregoing examples, or perform equivalent replacements for some of the technical features. All modifications, equivalent replacements, etc. made within the spirit and principle of the invention shall be included within the protection scope of the invention.

Claims

1. A deadbeat control method for a permanent magnet synchronous motor based on PSA algorithm parameter identification, characterized in that: The following steps are involved: S1, establish a mathematical model of the permanent magnet synchronous motor, discretize the mathematical model, obtain the traditional DPCC state equation, construct a current prediction algorithm equation that eliminates one-beat delay, and introduce weight coefficients α and β at the same time, and then combine it with the traditional DPCC state equation to obtain a DPCC with improved weight coefficients; S2, using PID search algorithm to identify the parameters of the permanent magnet synchronous motor, using the incremental PID controller to calculate the difference between the motor control quantity at the previous moment and the current moment, and using the difference as the new control quantity to dynamically adjust the motor parameters; S3, the optimal parameters of the motor identified by the PID search algorithm are substituted into the DPCC with improved weight coefficients for effective compensation to eliminate the current static error.

2. The deadbeat control method of a permanent magnet synchronous motor based on PSA algorithm parameter identification according to claim 1 is characterized in that: In S1, the current equation at time k+1 is obtained by modifying the traditional DPCC state equation, based on which the state observation coefficient matrix equation is established: U(k)=AI(k)+LΔI(k)+D in, Where U(k) is the dq axis voltage matrix at time k, I(k) is the dq axis current matrix at time k, ΔI(k) is the dq axis deviation current matrix at time k, L represents a matrix with two rows and two columns, L s and R s are the stator inductance and stator resistance of the motor, ψ f is the motor flux, ω e is the electrical angular velocity of the rotor, T s is the sampling period, i d * (k) and i q * (k) are the given currents of the d-axis and q-axis at time k, i d (k) and i q (k) are the d-axis component and q-axis component of the stator current at time k respectively.

3. The deadbeat control method of a permanent magnet synchronous motor based on PSA algorithm parameter identification according to claim 2 is characterized in that: According to the state observation coefficient matrix equation, the current at k+1 moments is predicted, and the one-shot delay of current command calculation is eliminated to obtain the improved current prediction algorithm equation: I Pref (k+1)=I(k+2)-A c L -1 [LΔI(k)+AI(k)-U(k)+D] in, TO C =L -1 *AE In the formula, I Pref (k+1) is the predicted current matrix at the next beat k+1 after improvement, I(k+2) is the dq axis current matrix at k+2, L represents a matrix of two rows and two columns, E is the unit matrix, ΔI(k) is the dq axis deviation current matrix at k, I(k) is the dq axis current matrix at k, and U(k) is the dq axis voltage matrix at k.

4. The deadbeat control method of a permanent magnet synchronous motor based on PSA algorithm parameter identification according to claim 3 is characterized in that: In the improved current prediction algorithm equation, weight coefficients α and β are introduced, and then combined with the traditional DPCC state equation, the improved voltage prediction equation at time k is obtained: In the formula, u d (k) and u q (k) are the improved voltages of d-axis and q-axis at time k, L s and R s are the stator inductance and stator resistance of the motor, T s is the sampling period, ψ f is the motor flux, α and β are weight coefficients, i d (k) and i q (k) are the d-axis component and q-axis component of the stator current at time k, i d * (k+1) and i q * (k+1) are the given currents of the d-axis and q-axis at time k+1, i dpref (k+1) and i qpref (k+1) is the predicted current of the d-axis and q-axis at time k+1 after a beat delay, ω e (k) is the electrical angular velocity of the rotor at time k.

5. The deadbeat control method of a permanent magnet synchronous motor based on PSA algorithm parameter identification according to claim 4 is characterized in that: In S1, the dynamic performance of the DPCC with improved weight coefficient is analyzed to obtain the actual current i dq and given current i dq * The closed-loop discrete transfer function of is: Then the stable interval of DPCC with improved weight coefficient is obtained as follows: In the formula, i dq (z) is the feedback current of the dq axis under the change of z, i dq * (z) is the given current of the dq axis under the change of z, z is the closed-loop pole, α and β are weight coefficients, L s is the stator inductance of the motor, L motor is the actual inductance of the motor.

6. The deadbeat control method of a permanent magnet synchronous motor based on PSA algorithm parameter identification according to claim 1, characterized in that: In S2, the fitness function, reference model and adjustable model are designed in combination with the physical characteristics of the inductance, resistance and flux parameters of the motor to evaluate the error and optimization effect in the parameter identification process.

7. The deadbeat control method of a permanent magnet synchronous motor based on PSA algorithm parameter identification according to claim 6 is characterized in that: In S2, when the number of iterations is t, the output value Δu(t) of the PID adjustment is: Δu(t)=K p ·r2·[e k (t)-e k-1 (t)]+K i ·r3·e k (t)+K d ·r4·[e k (t)-2e k-1 (t)+e k-2 (t)] Where Δu(t) is the control increment at the current moment, r2, r3 and r4 are vectors of random numbers from 0 to 1 in n rows and 1 column; K p , K i and K d are the adjustment coefficients of proportional, integral and differential respectively, e k (t), e k-1 (t), e k-2 (t) are the errors at time k, time k-1, and time k-2 respectively.

8. The deadbeat control method of a permanent magnet synchronous motor based on PSA algorithm parameter identification according to claim 7, characterized in that: In S3, the process of obtaining the optimal parameters of the motor is as follows: the PID search algorithm continuously calculates the error between the motor reference model and the adjustable model output value through the fitness function. The closer the error value is to 0, the closer the identified parameter is to the actual value.

9. The deadbeat control method of a permanent magnet synchronous motor based on PSA algorithm parameter identification according to claim 8, characterized in that: The reference model is: The adjustable model is: In the formula, x represents the system state variable, u represents the system input, p represents the system parameter, and y represents the system output. They represent the state variables, parameters and outputs in the adjustable model respectively; f and g are the state equation and output equation of the system, They represent the first-order differentials of the state variables of the reference model and the first-order differentials of the state variables in the adjustable model, respectively.

10. The deadbeat control method of a permanent magnet synchronous motor based on PSA algorithm parameter identification according to claim 9, characterized in that: The fitness function is: Where w1, w2 and w3 are weight factors, u d0 (k) and u q0 (k) are the voltages of the d-axis and q-axis sampled at the current moment, T e * (k) is the electromagnetic torque value of the dq axis sampled at the current moment, are the voltages of the d-axis and q-axis calculated according to the identification results, T e (k) is the dq-axis electromagnetic torque value calculated according to the identification result, and n is the number of iterations.

Images