Dead-beat current prediction control method based on parameter adaptive online identification

Abstract

The invention discloses a dead-beat current prediction control method based on parameter adaptive online identification. The method comprises the steps of S1, mathematical modeling of a three-phase permanent magnet synchronous motor; s2, designing and simulating a vector control system based on PI control; s3, designing a dead-beat predictive current control algorithm; s4, vector control system simulation based on a dead-beat predictive current control algorithm; s5, analyzing the performance of the dead-beat predictive current control algorithm under the parameter mismatch condition; s6, a parameter identification algorithm based on model reference self-adaption; s7, performing model reference adaptive parameter identification algorithm simulation; and S8, performing parameter identification-based permanent magnet motor dead-beat predictive current control system simulation. According to the method, permanent magnet synchronous motor parameter identification based on a model reference adaptive algorithm is combined with dead-beat predictive current control, so that the stability and the interference resistance of the system are improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of permanent magnet synchronous motor control technology in industrial automation, involving motor mathematical modeling, parameter adaptive online identification and current predictive control, and particularly to a deadbeat current predictive control method based on parameter adaptive online identification. BACKGROUND

[0002] With the development of industrial automation and electrification, permanent magnet synchronous motors are widely used in high-precision driving fields such as new energy vehicles and industrial robots. The deadbeat predictive current control algorithm has attracted widespread attention due to its high response speed and zero steady-state error tracking capability. However, this algorithm has strong dependence on motor parameters, and parameter mismatch can easily lead to a decline in control performance. Specifically:

[0003] When the actual motor parameters do not match the preset parameters of the controller, the deadbeat predictive current control performance will be significantly affected, leading to increased current harmonics, torque fluctuations, and intensified electromagnetic vibrations. In severe cases, the system may even lose stability, which is inconsistent with the design purpose. Meanwhile, during operation, the electromagnetic parameters of permanent magnet motors may change due to factors such as temperature rise and magnetic saturation, leading to a mismatch between the control system and the theoretical design expectations, and further affecting the performance of the deadbeat predictive current control system.

[0004] Therefore, the main purpose of the present application is to combine permanent magnet synchronous motor model reference adaptive parameter identification with deadbeat predictive current control, allowing the system to automatically correct the predictive model parameters when the motor parameters are mismatched, maintaining the effectiveness of the control strategy. This method has engineering significance in providing a control system with stronger disturbance rejection and better environmental adaptability for permanent magnet motors, ensuring the stability and reliability of motor control under complex working conditions.

[0005] Comparison with prior art technology as follows:

[0006] Comparison with patent CN119051500A "PIR type permanent magnet synchronous motor deadbeat predictive control method with disturbance suppression"

[0007] I. Patent CN119051500A designs a PIR type deadbeat current predictive controller based on the electromagnetic equation of a permanent magnet synchronous motor, which improves the anti-steady-state error and alternating current disturbance capability of the deadbeat controller by adding an integral element and a quasi-resonant element. Our design is based on a permanent magnet motor vector control system using deadbeat predictive current control, which does not use the PIR structure. The core is to combine permanent magnet synchronous motor model reference adaptive (MRAS) parameter identification with deadbeat predictive current control, specifically addressing the problems of increased current harmonics, torque fluctuations, and other issues caused by motor parameter mismatch, rather than focusing on anti-steady-state error and alternating current disturbance.

[0008] Secondly, patent CN119051500A uses MRAS with nonlinear ADRC adaptive law for parameter identification and disturbance observation to improve the overshoot problem of linear ADRC and retain the ability to resist high-frequency disturbances. We use a parameter adaptive online identification mechanism without relying on nonlinear ADRC adaptive law. The core goal is to automatically correct the prediction model parameters when the electromagnetic parameters change due to temperature rise and magnetic circuit saturation, maintain the effectiveness of the control strategy, provide a control system with stronger disturbance resistance and better environmental adaptability for permanent magnet motors, and ensure stability and reliability under complex working conditions.

[0009] Comparison with the technology of patent CN111969914A "Permanent Magnet Synchronous Motor Deadbeat Current Prediction Control Method and Device, Storage Medium"

[0010] Firstly, patent CN111969914A aims to improve the dynamic response of permanent magnet synchronous motors by determining whether the motor enters a dynamic state, determining the dynamic prediction reference voltage, and selecting the basic voltage vector with the smallest error to improve the motor's adaptive speed to dynamics. We focus on the parameter mismatch problem of the motor, not on dynamic response optimization. The core is to combine permanent magnet synchronous motor MRAS parameter identification with deadbeat predictive current control to solve the problem of performance degradation or instability of the control system caused by parameter mismatch, rather than improving dynamic response capability.

[0011] Secondly, the control strategy of patent CN111969914A focuses on dynamic working conditions. It only outputs dynamic control pulse signals that adapt to the dynamic prediction reference voltage to achieve fast adaptation to motor dynamics. We have developed a permanent magnet synchronous motor deadbeat predictive current control system based on parameter identification. It can automatically correct model parameters when there is a parameter mismatch, and it can also support high-precision control of the motor under complex working conditions to ensure the stability and reliability of the control, rather than being limited to adaptive control under dynamic working conditions. SUMMARY

[0012] The purpose of the present application is to provide a deadbeat current prediction control method based on parameter adaptive online identification. By combining permanent magnet synchronous motor parameter identification based on model reference adaptive algorithm with deadbeat predictive current control, the stability and disturbance resistance of the system are improved.

[0013] To achieve the above purpose, the technical solution adopted by the present application is as follows:

[0014] A deadbeat current prediction control method based on parameter adaptive online identification, comprising the following steps:

[0015] S1, mathematical modeling of a three-phase permanent magnet synchronous motor;

[0016] S2, PI control based vector control system design simulation;

[0017] S3, deadbeat predictive current control algorithm design;

[0018] S4, vector control system simulation based on deadbeat predictive current control algorithm;

[0019] S5, performance analysis of deadbeat predictive current control algorithm under parameter mismatch condition;

[0020] S6, parameter identification algorithm based on model reference adaptive;

[0021] S7, simulation of model reference adaptive parameter identification algorithm;

[0022] S8, simulation of deadbeat predictive current control system of permanent magnet motor based on parameter identification.

[0023] As a preferred technical scheme of the application: step S1 is specifically as follows:

[0024] S11, mathematical model of three-phase permanent magnet synchronous motor in three-phase stationary coordinate system, specifically as follows:

[0025] Among them, the voltage equation is:

[0026]

[0027] The flux equation is:

[0028] Ψ 3s =L 3s i 3s +ψ f ·F 3S (θ e ) (2);

[0029] Among them: u 3s , R, i 3s , ψ 3s Respectively, the phase voltage, resistance, current and flux; L 3s For three-phase winding inductance; ψ f For permanent magnet flux;

[0030] F 3S (θ e ) is the unit flux and satisfies:

[0031]

[0032] Among them: L m3 For stator mutual inductance; L sσ For stator leakage inductance; i A i B i Cis the three-phase current; R is the resistance; ψ A ψ B ψ C is the three-phase flux linkage; u A u B u C is the three-phase voltage; θ e is the electrical angle representing the electrical angle of the armature rotation; and there is also an electromagnetic torque equation:

[0033]

[0034] where p n is the number of magnetic pole pairs of the three-phase PMSM; T e is the electromagnetic torque; θ m is the mechanical angle;

[0035] In addition, the mechanical motion equation of the motor is as follows:

[0036]

[0037] where J is the moment of inertia of the motor; T L is the load torque; B represents the viscous damping coefficient; ω m represents the motor mechanical angular velocity;

[0038] S12, to realize vector control, coordinate transformation is performed on the above mathematical model to obtain the mathematical model in the synchronous rotating coordinate system, which is as follows:

[0039] S121, coordinate transformation:

[0040] The quantities in the three-phase stationary coordinate system ABC are transformed to the two-phase stationary coordinate system α-β by Clark transformation, and the coordinate transformation formula is as follows:

[0041]

[0042] In the formula, the coordinate transformation matrix is:

[0043]

[0044] The two-phase stationary coordinate system α-β is transformed to the synchronous rotating coordinate system d-q by Park transformation, and the coordinate transformation formula is as follows:

[0045]

[0046] In the formula, the coordinate transformation matrix is:

[0047]

[0048] where f α f βf o represent the components of a, b, o in the a-β coordinate system;f A f B f C represent the components of A, B, C in the three-phase coordinate system; is a transformation matrix for conversion from the three-phase 3s coordinate system to the two-phase stationary 2s coordinate system;f d f q components in the d-q coordinate system; represents a transformation matrix from the a-β coordinate system to the d-q coordinate system, where the subscripts 2s and 2r respectively represent the two-phase stationary coordinate system and the two-phase rotating coordinate system;

[0049] S122, mathematical model of permanent magnet synchronous motor in synchronous rotating coordinate system:

[0050] Using Clark transformation and Park transformation, the permanent magnet motor can be converted from the three-phase stationary coordinate system to the two-phase rotating coordinate system, that is, the stator voltage equation can be obtained:

[0051]

[0052] Stator flux linkage equation:

[0053]

[0054] It can be obtained:

[0055]

[0056] Where, u d , u q are the d, q axis components of the stator voltage; ψ d , ψ q , ψ f are the d, q axis components of the stator flux linkage and the permanent magnet flux linkage; i d , i q are the d, q axis components of the stator current; R s is the stator resistance; L d , L q are the d, q axis components of the inductance; ω e is the electrical angular velocity,

[0057] The electromagnetic torque equation is:

[0058]

[0059] The motion equation is:

[0060]

[0061] Adopting surface-mounted three-phase permanent magnet synchronous motor, its L d = L q , the torque formula shown in formula (13) can be seen that the electromagnetic torque is determined by the cross-axis current i q , that is, to ensure i d = 0, the maximum torque under the same current can be obtained.

[0062] As a preferred technical solution of the application: step S2 is as follows:

[0063] S21, build a vector control system based on PI control

[0064] Under the Simulink environment, a permanent magnet synchronous motor vector control system based on PI control of speed-current double closed loop structure is built, which is i d = 0 three-phase permanent magnet synchronous motor vector control system, the speed loop of the system is the difference between the given speed and the actual speed, and the q-axis reference current is output by the speed loop regulator, and the current loop is also the difference between the respective current given signal and the current feedback signal, and the reference voltage is output by the current loop regulator, wherein the speed loop and the current loop regulator are designed based on PI control algorithm;

[0065] S22, system control strategy design

[0066] The closed loop control strategy of the above-mentioned permanent magnet synchronous motor vector control system with speed-current double closed loop structure adopts PID control strategy, and its mathematical model is as follows:

[0067] u(t)=K p e(t)+K i ∫e(t)dt (15);

[0068] Wherein, K p is the proportional coefficient, K i is the integral coefficient, and K d is the differential coefficient; u(t) represents the output of the controller, that is, the control amount; e(t) represents the error signal, and ∫e(t)dt represents the integral of the error signal e(t) over time;

[0069] S23, system simulation

[0070] The simulation time is set to 0.4s, the initial given speed is 750r / min, and the no-load start is started; the speed is unchanged at 0.2s, and a step load of 5Nm is suddenly added; the load is unchanged at 0.3s, and the given speed is changed to 1000r / min,

[0071] Analysis of motor speed of permanent magnet synchronous motor vector control system simulation based on PI control:

[0072] When the motor is started under no-load, the motor reaches the given speed 750r / min for the first time at 0.015s, overshoots to 779r / min, the overshoot is about 3.87%, and enters the given speed ±1% range for stable operation at 0.019s, the motor speed fluctuates in the range of 752r / min-754r / min, when a 5Nm step load is suddenly added at 0.2s, the motor speed drops to 745r / min at the lowest, and recovers to stable state within 0.04s, so that the motor speed fluctuates in the range of 746r / min-748r / min, the given speed rises to 1000r / min at 0.3s, the motor reaches the given speed 1000r / min for the first time at 0.014s, overshoots to 1012r / min, the overshoot is about 1.2%, and enters the given speed ±1% range for stable operation at 0.016s, the motor speed fluctuates in the range of 997r / min-999r / min;

[0073] Motor torque analysis based on PI control of permanent magnet synchronous motor vector control system simulation:

[0074] When the motor is started under no-load, the motor first outputs high torque, the highest can reach 54Nm, and as the motor enters the steady state, the torque drops to fluctuate in the range of -1Nm-2Nm, and only overcomes the motor rotating damping, the torque fluctuation rate is:

[0075]

[0076] Wherein, T max is the maximum torque, T min is the minimum torque,

[0077] When a 5Nm step load is suddenly added at 0.2s, the motor output torque increases rapidly and fluctuates in the range of 3.5Nm-7.5Nm, at this time the torque fluctuation rate is:

[0078]

[0079] When the given speed rises to 1000r / min at 0.3s, the motor outputs high torque again for acceleration, and after the motor enters the steady state, it falls to fluctuate in the range of 4.5Nm-6.5Nm, at this time the torque fluctuation rate is:

[0080]

[0081] As a preferred technical scheme of the application, step S3 is specifically as follows:

[0082] The voltage equation of the three-phase permanent magnet synchronous motor in the dq synchronous rotating coordinate system is:

[0083]

[0084] wherein: u d , u q are d, q-axis components of stator voltage; ψ f is permanent magnet flux linkage; i d , i q are d, q-axis components of stator current; R S is stator resistance; L d , L q are d, q-axis components of inductance; ω e is electrical angular velocity,

[0085] Discretization is performed on the voltage equation, when sampling time T s reaches a preset threshold, it is considered that the system input quantity remains constant in a sampling period, and since the mechanical time constant of the motor is greater than the electrical time constant, the speed and back EMF change much slower than the current, so the back EMF term can be considered constant in a sampling period, then equation (19) can be rewritten as:

[0086]

[0087] k is a discrete time point, on this basis, the given current is taken as the predicted current value in the next period, that is, the motor can reach the given current value in the next period, then:

[0088]

[0089] Therefore, for the q-axis current loop, the predicted current value in the next period is the speed loop output, and for the d-axis current loop, the predicted current value in the next period is the current loop output.

[0090] As a preferred technical solution of the application, step S4 is specifically as follows:

[0091] S41, a vector control system of the dead-beat predictive current control algorithm is built

[0092] A permanent magnet synchronous motor vector control system based on the dead-beat predictive current control algorithm is built in the Simulink environment, wherein the current loop PI controller is replaced by the dead-beat predictive current control algorithm, the d, q-axis voltage calculation is completely consistent with equation (20), and the speed loop output is equivalent to i q (k+1) as the q-axis given current, at the same time, since the system uses i d =0 control, i d (k+1)=0.

[0093] S42, system simulation

[0094] The simulation duration is set to 0.4s, the initial given speed is 750r / min, the load is started, the speed is unchanged at 0.2s, and a step load of 5Nm is suddenly added at 0.3s, the given speed is changed to 1000r / min,

[0095] Motor speed analysis of permanent magnet synchronous motor vector control system simulation based on dead-beat predictive current control:

[0096] Under dead-beat predictive current control, the motor first reaches the given speed of 750r / min at 0.015s when starting under no load, overshoots to 764r / min, with an overshoot of about 1.87%, and quickly enters steady state, keeping the motor speed fluctuating in the range of 752.5r / min~752.7r / min, compared with PI control overshoot of 3.87% and speed fluctuation of 2r / min,

[0097] When a step load of 5Nm is suddenly added at 0.2s, the motor speed under dead-beat predictive current control drops to 745.3r / min at the lowest, and recovers to stable state within 0.03s, keeping the motor speed fluctuating in the range of 747r / min~747.2r / min, compared with PI control recovery time of 0.04s and speed fluctuation of 2r / min,

[0098] At 0.3s, the given speed rises to 1000r / min, the motor first reaches the given speed of 1000r / min within 0.014s, with an overshoot of less than 0.1%, and the motor speed fluctuates in the range of 998.3r / min~998.5r / min,

[0099] Motor torque analysis of permanent magnet synchronous motor vector control system simulation based on dead-beat predictive current control algorithm:

[0100] Under dead-beat predictive current control, the motor first outputs high torque when starting under no load, with a maximum of 54Nm, and as the motor enters steady state, the torque drops to fluctuate in the range of 0.2Nm~1.1Nm, with a torque fluctuation rate of:

[0101]

[0102] Under the same conditions, the torque fluctuation rate of PI control is 300%,

[0103] When a step load of 5Nm is suddenly added at 0.2s, the motor output torque increases rapidly and fluctuates in the range of 5.1Nm~6.1Nm, at this time the torque fluctuation rate is:

[0104]

[0105] Under the same conditions, the torque fluctuation rate of PI control is 36.36%,

[0106] 0.3s when the given speed rises to 1000r / min, the motor output high torque for acceleration again, and after the motor into the steady state back to 5.4Nm-6.2Nm fluctuation range, the torque fluctuation rate is:

[0107]

[0108] Under the same conditions, the PI control torque fluctuation rate is 18.18%.

[0109] As a preferred technical solution of the application: step S5 is specifically as follows:

[0110] In the permanent magnet synchronous motor vector control system based on the deadbeat predictive current control algorithm, the motor inductance and resistance are respectively:

[0111]

[0112] Where, R S is the stator resistance; L d , L q are the d, q axis components of inductance,

[0113] Keep the motor parameters unchanged, and reduce the resistance and inductance based on the deadbeat predictive current control algorithm to 0.2 times of the original, that is:

[0114]

[0115] Where, L d0 and L q0 represent the initial inductance in the direct axis and quadrature axis directions; L0 represents the initial inductance; R0 represents the initial resistance,

[0116] At this time, the permanent magnet synchronous motor vector control system based on the deadbeat predictive current control algorithm is simulated,

[0117] Motor speed analysis of the deadbeat predictive current control algorithm under the parameter mismatch condition:

[0118] Under the parameter mismatch condition, when the motor is started at no load, the motor reaches the given speed 750r / min for the first time at 0.016s, overshoots to 765r / min, the overshoot is about 2.00%, and quickly enters the steady state, so that the motor speed is maintained in the range of 751.2r / min-752.2r / min fluctuation, compared with the case without parameter mismatch, the motor overshoot is 1.87%, the speed fluctuation is 0.2r / min, compared with the PI control case, the motor overshoot is 3.87%, the speed fluctuation is 2r / min,

[0119] 0.2s sudden 5Nm step load, motor in the deadbeat predictive current control under the speed of the lowest drop to 743.9r / min, and in 0.03s recovery stable, make motor speed maintain in 745.5r / min~746.6r / min range fluctuation,

[0120] 0.3s when the given speed rises to 1000r / min, motor in 0.015s first reach the given speed 1000r / min, overshoot to 1001.7r / min, motor speed maintain in 999r / min~999.3r / min range fluctuation,

[0121] Parameter mismatch condition under the deadbeat predictive current control algorithm simulation motor torque analysis:

[0122] Motor in parameter mismatch condition, no-load start, motor first output high torque, highest 45Nm, with motor into steady state, torque drop to-0.65Nm~1.73Nm range fluctuation, torque fluctuation rate is:

[0123]

[0124] Under the same condition, PI control torque fluctuation rate is 300%, no parameter mismatch deadbeat predictive current control torque fluctuation rate is 69.23%,

[0125] 0.2s sudden 5Nm step load, motor output torque increases rapidly, and maintain in 4.3Nm~6.8Nm range fluctuation, at this time torque fluctuation rate is:

[0126]

[0127] Under the same condition, PI control torque fluctuation rate is 36.36%, no parameter mismatch deadbeat predictive current control torque fluctuation rate is 8.93%,

[0128] 0.3s when the given speed rises to 1000r / min, motor again output high torque for acceleration, and after motor into steady state, back to 5.1Nm~6.3Nm range fluctuation, at this time torque fluctuation rate is:

[0129]

[0130] Under the same condition, PI control torque fluctuation rate is 18.18%, no parameter mismatch deadbeat predictive current control torque fluctuation rate is 6.90%.

[0131] As a preferred technical scheme of the application: step S6 is specifically as follows:

[0132] S61, the determination of adjustable model

[0133] The three-phase permanent magnet synchronous motor is regarded as a reference model, and the mathematical model of the three-phase permanent magnet synchronous motor in the d-q synchronous rotating coordinate system is regarded as an adjustable model. Since the expected identified inductance and resistance parameters are included in the voltage equation, the adjustable model can use the voltage equation shown in equation (12),

[0134] Rewrite equation (12) into the form of a state equation, and then have:

[0135]

[0136] where u d and u q are the d and q axis components of the stator voltage respectively; ψ f is the permanent magnet flux linkage; i d and i q are the d and q axis components of the stator current respectively; L d and L q are the d and q axis components of the inductance respectively; ω e is the electrical angular velocity; and α is the estimated value of the ratio of the resistance to the inductance of the motor, L is the inductance, and R is the resistance. In the equation, all quantities with hat represent the output values or identified values of the adjustable model;

[0137] S62, reference adaptive law design

[0138] Lyapunov stability theory is selected to design the adaptive law of the three-phase permanent magnet synchronous motor parameter identification system. The large range asymptotic stability judgment theorem in Lyapunov stability theory is as follows:

[0139] To ensure large range uniform asymptotic stability in the equilibrium state, for a continuous time nonlinear time-varying self-regulating system, if a scalar function V(x, t) can be constructed, V(0, t) = 0, and for all non-zero state points x in the state space, the following conditions are met:

[0140] V(x, t) is positive definite and bounded, and when ||x||→∞, V(x, t)→∞;

[0141] The partial derivative with respect to time is negative and bounded,

[0142] then the zero equilibrium state x = 0 of the system is a large range uniform asymptotic stable equilibrium point,

[0143] According to the above Lyapunov stability theory, the adaptive law is designed,

[0144] First, simplify the adjustable model shown in equation (30), and let:

[0145]

[0146] Wherein, e is error vector, e1, e2 is vector element; I is actual current vector, U is voltage vector, W is angular velocity vector, A is system matrix, B is input matrix, C is output matrix, L is motor inductance,

[0147] Simultaneous formula (31) in all terms, and partial derivative with respect to time, then have:

[0148]

[0149] Let:

[0150]

[0151] Wherein, The transpose of a, b, c three difference value; s is column vector,

[0152] Substitute formula (33) into formula (31), then have:

[0153]

[0154] The constructor V (x, t):

[0155]

[0156] Wherein, The input vector of function V is represented by e T The transpose of error vector; The state vector,

[0157] Then formula (35) satisfies the condition 1 of Lyapunov stability theory, that is, V ([e, φ], t) is positive definite and bounded, and when || [e, φ] ||→∞, V ([e, φ], t)→∞,

[0158] At this time, only need to derive when Negative, write Expression and simplify, have:

[0159]

[0160] According to formula (31), have

[0161]

[0162] Wherein A T The transpose of system matrix A,

[0163] That is, the first term of the formula is obviously negative, so it is also necessary to ensure that the second term is zero, that is, to ensure Negative, therefore, let:

[0164]

[0165] Then we have:

[0166]

[0167] Solving the equations, we get the adaptive law as:

[0168]

[0169] where K L and K R are gain coefficients related to inductance and resistance estimation, L0 represents the initial inductance, R0 represents the initial resistance,

[0170] The adaptive law designed according to Lyapunov stability theory is shown as formula (40), and a model reference adaptive parameter identification system can be constructed in combination with the adjustable model.

[0171] As a preferred technical scheme of the present application, step S7 is specifically as follows:

[0172] S71, simulation verification is performed in the MATLAB\Simulink environment

[0173] S711, feedback currents i d , i q , electrical angular velocity and current loop output voltage u d , u q are obtained from the vector control system based on the deadbeat predictive current control, which are taken as inputs of the model reference adaptive parameter identification algorithm and used for calculating estimated currents;

[0174] S712, actual currents and estimated currents are subtracted to obtain current errors, and the current errors are input into the adaptive law, and resistance and inductance parameters are output from the adaptive law;

[0175] S713, the identified parameters are fed back to the adjustable model, and the parameters in the adjustable model are adjusted, so that the estimated currents output by the adjustable model are more and more close to the actual currents, and the errors of the two are smaller, that is, the adjustable model is more accurate, and the identified parameters are more accurate;

[0176] S72, system simulation

[0177] The simulation time is set to 0.4s, the given speed is 750r / min, and the load is 5Nm. Since the surface-mounted permanent magnet synchronous motor is selected, the d-axis and q-axis inductances are equal, so only one of them needs to be identified. The identified inductance starts from 0 and gradually approaches the actual inductance. At 0.1s, it reaches 10.9mH, at 0.15s, it reaches 11.5mH, and it is maintained at 11.5±0.5mH. Compared with the actual inductance of 12mH, the identification error is about 4.17%,

[0178] The identified resistance gradually approaches the actual resistance from 0, reaches 5.41Ω at 0.15s, and remains at 5.35±0.5Ω, compared with the actual inductance of 4.875Ω, the identification error is about 9.74%,

[0179] Based on the identification results of the above simulation, the error of the identification results is always kept within 10%.

[0180] As a preferred technical solution of the application, step S8 is specifically as follows:

[0181] The identified parameters are fed back to the dead-beat predictive current control algorithm to form a permanent magnet motor dead-beat predictive current control system simulation based on parameter identification, which is specifically as follows:

[0182] The simulation duration is set to 0.4s, and the initial given speed is kept at 750r / min, while the motor is started under no load,

[0183] Motor speed analysis using parameter identification under no parameter mismatch condition, parameter mismatch condition and parameter mismatch condition:

[0184] The motor no-load starting performance is basically the same as the parameter mismatch condition, and as the motor enters the steady state, at 0.05s, the identified parameters are added, the motor speed fluctuation starts to decrease, and is maintained in the range of 752.6r / min~753.1r / min, compared with the motor overshoot of 1.87% and the speed fluctuation of 0.2r / min under the condition of no parameter mismatch, the speed fluctuation is obviously lower than 1r / min without using parameter identification under the condition of parameter mismatch,

[0185] When a 5Nm step load is suddenly added at 0.2s, the motor speed is lowest and decreases to 745.5r / min, and recovers to stable within 0.03s, so that the motor speed is maintained in the range of 747r / min~747.3r / min, compared with the case that the motor speed is lowest and decreases to 743.9r / min without using parameter identification under the condition of parameter mismatch, and the motor speed is maintained in the range of 745.5r / min~746.6r / min after 0.03s, which is obviously optimized,

[0186] The given speed is increased to 1000r / min at 0.3s, the motor reaches the given speed of 1000r / min for the first time within 0.015s, and the overshoot is 1001r / min, and the motor speed is maintained in the range of 998.5r / min~998.8r / min,

[0187] Motor torque analysis using parameter identification under no parameter mismatch condition, parameter mismatch condition and parameter mismatch condition:

[0188] The motor's no-load starting performance is consistent with that under parameter mismatch conditions. As the motor enters steady state, at 0.05s, the identification parameters are applied, and the torque stabilizes within the range of 0.1Nm to 1.1Nm, with a torque fluctuation rate of:

[0189]

[0190] When a 5 Nm step load is suddenly applied in 0.2 seconds, the motor output torque increases rapidly and fluctuates within the range of 5.1 Nm to 6.1 Nm. The torque fluctuation rate at this time is:

[0191]

[0192] Under the same operating conditions, the torque ripple rate of PI control is 36.36%;

[0193] The torque ripple rate of the deadbeat predictive current control without parameter mismatch is 8.93%.

[0194] The torque ripple rate of the deadbeat predictive current control with parameter mismatch is 22.52%.

[0195] At 0.3s, the given speed increases to 1000 r / min, and the motor outputs high torque again for acceleration. After the motor enters steady state, the torque drops back to fluctuate within the range of 5.4 Nm to 6.4 Nm. At this time, the torque fluctuation rate is:

[0196]

[0197] Under the same operating conditions, the torque ripple rate of PI control is 18.18%;

[0198] The torque ripple rate of the deadbeat predictive current control without parameter mismatch is 6.90%.

[0199] The torque ripple rate of the deadbeat predictive current control with parameter mismatch is 10.53%.

[0200] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0201] Since the deadbeat predictive current control algorithm is highly dependent on the accuracy of the motor model parameters, this invention optimizes the parameter identification algorithm based on model reference adaptation and combines the permanent magnet synchronous motor parameter identification based on the model reference adaptation algorithm with deadbeat predictive current control to improve the system performance under parameter mismatch conditions. Moreover, this invention develops a deadbeat predictive current control system for permanent magnet synchronous motors based on parameter identification, which provides strong support for high-precision control of permanent magnet synchronous motors under complex operating conditions. Attached Figure Description

[0202] Figure 1 A simulation model for a vector control system of a permanent magnet synchronous motor;

[0203] Figure 2 PID control system block diagram

[0204] Figure 3 Motor speed under PI control based vector control system

[0205] Figure 4 Motor torque under PI control based vector control system

[0206] Figure 5 Permanent magnet synchronous motor vector control system simulation model based on deadbeat predictive current control algorithm

[0207] Figure 6 D-axis voltage calculation module

[0208] Figure 7 Q-axis voltage calculation module

[0209] Figure 8 Motor speed comparison diagram of PI control based vector control system and vector control system based on deadbeat predictive current control algorithm

[0210] Figure 9 Motor torque comparison diagram of PI control based vector control system and vector control system based on deadbeat predictive current control algorithm

[0211] Figure 10 Motor speed comparison diagram under parameter mismatch condition and no parameter mismatch condition

[0212] Figure 11 Motor torque comparison diagram under parameter mismatch condition and no parameter mismatch condition

[0213] Figure 12 Model reference adaptive parameter identification algorithm simulation model

[0214] Figure 13 Model reference adaptive parameter identification algorithm

[0215] Figure 14 Adjustable model diagram of model reference adaptive parameter identification algorithm

[0216] Figure 15 Adaptive law circuit diagram

[0217] Figure 16 Inductance identification result curve diagram

[0218] Figure 17 Resistance identification result curve diagram

[0219] Figure 18A comparison chart of motor speed using parameter identification for permanent magnet synchronous motor under no parameter mismatch condition, parameter mismatch condition and parameter mismatch condition;

[0220] Figure 19 A comparison chart of motor torque using parameter identification for permanent magnet synchronous motor under no parameter mismatch condition, parameter mismatch condition and parameter mismatch condition. DETAILED DESCRIPTION

[0221] The application will be described in further detail below with reference to the drawings and specific embodiments:

[0222] The application discloses a parameter self-adaptive online identification-based deadbeat current prediction control method, which comprises the following steps:

[0223] Step S1, mathematical modeling of a three-phase permanent magnet synchronous motor

[0224] The motor mathematical model is the basis for developing motor control system research, therefore, first, the mathematical model of a permanent magnet synchronous motor in a three-phase stationary coordinate system is introduced, which is as follows:

[0225] The voltage equation of the three-phase permanent magnet synchronous motor in the mathematical model of the three-phase stationary coordinate system is:

[0226]

[0227] The flux linkage equation is:

[0228] Ψ 3s =L 3s i 3s +ψ f ·F 3S (θ e ) (2);

[0229] Wherein, u 3s , R, i 3s , ψ 3s are phase voltage, resistance, current and flux linkage respectively; L 3s is the inductance of the three-phase winding; ψ f is the permanent magnet flux linkage;

[0230] F 3S (θ e ) is a unit flux linkage and satisfies:

[0231]

[0232] Wherein, L m3 is the stator mutual inductance; L sσ is the stator leakage inductance; i A i B i C are three-phase currents; R is resistance; ψA ψ B ψ C For three-phase flux linkage; u A u B u C For three-phase voltage; θ e To represent an electrical angle, it is the electrical angle of rotation of the armature in an electric motor.

[0233] There is also the electromagnetic torque equation:

[0234]

[0235] Where, p n T represents the number of pole pairs in a three-phase PMSM. e For electromagnetic torque; θ m For mechanical angle,

[0236] In addition, the mechanical motion equations of the electric motor are as follows:

[0237]

[0238] Where J represents the moment of inertia of the motor; T L To represent the load torque; B represents the viscous damping coefficient; ω m Indicates the mechanical angular velocity of the motor.

[0239] Equations (1) to (5) are the mathematical models of the PMSM in the three-phase stationary coordinate system. Obviously, the mathematical model of the motor is quite complex at this time, and there are complex coupling relationships between the parameters. In order to realize vector control, it is also necessary to perform coordinate transformation on the model to obtain its mathematical model in the synchronous rotating coordinate system, as follows:

[0240] Coordinate transformation:

[0241] The Clark transformation is a method for transforming quantities in a three-phase stationary coordinate system ABC to a two-phase stationary coordinate system α-β, and has the following coordinate transformation formula:

[0242]

[0243] In the formula, the coordinate transformation matrix is:

[0244]

[0245] The Park transformation is a method for transforming a two-phase stationary coordinate system α-β to a synchronously rotating coordinate system dq, and it has the following coordinate transformation formula:

[0246]

[0247] In the formula, the coordinate transformation matrix is:

[0248]

[0249] where f α f β f o represent the components of a, b, o in the a- b coordinate system; f A f B f C represent the components of A, B, C in the three-phase coordinate system; is a transformation matrix for conversion from the three-phase 3s coordinate system to the two-phase stationary 2s coordinate system; f d f q components in the d-q coordinate system; represents a transformation matrix from the a- b coordinate system to the d-q coordinate system, where the subscripts 2s and 2r represent the two-phase stationary coordinate system and the two-phase rotating coordinate system, respectively.

[0250] Mathematical model of the permanent magnet synchronous motor in the synchronous rotating coordinate system:

[0251] Using the Clark transformation and the Park transformation, the permanent magnet motor can be converted from the three-phase stationary coordinate system to the two-phase rotating coordinate system, i.e. the stator voltage equation can be obtained:

[0252]

[0253] Stator flux linkage equation:

[0254]

[0255] It can be obtained that:

[0256]

[0257] where f d , u q are the d, q axis components of the stator voltage; f d , f q , f f are the d, q axis components of the stator flux linkage and the permanent magnet flux linkage; f d , i q are the d, q axis components of the stator current; f s is the stator resistance; f d , f q are the d, q axis components of the inductance; f e is the electrical angular velocity,

[0258] The electromagnetic torque equation is:

[0259]

[0260] The motion equation is:

[0261]

[0262] where: p n is the number of pole pairs of the three-phase PMSM; i d , i q are the d, q-axis components of the stator current; L d , L q are the d, q-axis components of the inductance; ψ f is the permanent magnet flux linkage, J represents the moment of inertia of the motor; T e is the electromagnetic torque; T L represents the load torque; B represents the viscous damping coefficient; ω m represents the mechanical angular velocity of the motor.

[0263] The present application adopts a surface-mounted three-phase permanent magnet synchronous motor, and L d = L q As can be seen from the torque formula shown in equation (13), the electromagnetic torque at this time is completely determined by the cross-axis current i q , that is, as long as i d = 0, the maximum torque under the same current can be obtained.

[0264] Step S2, simulation of a vector control system based on PI control

[0265] A permanent magnet synchronous motor vector control system based on a PI control speed-current double closed-loop structure is built in a Simulink environment, and the simulation model is as shown in Figure 1 .

[0266] The control system is an i d = 0 permanent magnet synchronous motor vector control system. The speed loop outputs the q-axis reference current by taking the difference between the given speed and the actual speed and adjusting the speed loop. The current loop also outputs the reference voltage by taking the difference between the respective current given signal and the current feedback signal and adjusting the current loop regulator.

[0267] Among them, the speed loop and current loop regulators of the system are designed based on the PI control algorithm. In various closed-loop control strategies, the PID control strategy is the most widely used control method at present due to its simple structure, low cost, and not over-reliance on system mathematical model and parameters. In fact, the PID control strategy is divided into proportional (Proportional), integral (Integral), and differential (Differential) three parts, and its block diagram is as shown in Figure 2 .

[0268] Its mathematical model can be written in the following form:

[0269]

[0270] The proportional adjustment compensates for the system deviation according to a certain proportion, with the proportional coefficient K. p Increasing the proportional coefficient K can speed up the system's adjustment speed and reduce the steady-state error, but the proportional coefficient K... p If the integral coefficient is too large, it will cause the system to oscillate continuously or even become unstable within a certain range; integral control integrates the system deviation until there is no deviation in the system, therefore the integral coefficient K... i Increasing the value of K can reduce the steady-state error of the system, but it will also reduce the dynamic response speed and stability of the system; differential adjustment K d It can improve the dynamic performance of the system and reduce overshoot, but an excessively large differential coefficient can easily lead to system instability.

[0271] Because motor control systems are nonlinear systems with fast response times and significant system signal noise, and because derivative control easily amplifies noise interference and makes the system unstable, motor control typically only uses proportional and integral control, forming a PI controller. The mathematical model of the PI controller is as follows:

[0272] u(t) = K pe (t)+K i ∫e(t)dt (15);

[0273] Among them, K p K is the proportionality coefficient. i K is the integral coefficient. d Here, is the differential coefficient; u(t) represents the controller output, i.e., the control quantity; e(t) represents the error signal, and ∫e(t)dt represents the integral of the error signal e(t) over time;

[0274] The simulation duration was set to 0.4 s, with an initial set speed of 750 r / min and no-load start-up. At 0.2 s, the speed remained constant, and a sudden step load of 5 Nm was applied. At 0.3 s, the load remained constant, and the set speed changed to 1000 r / min. Motor parameters are shown in Table 1. The simulation results of this system are as follows: Figure 2 , Figure 3 As shown.

[0275]

[0276] Table 1 Parameters of Permanent Magnet Synchronous Motor

[0277] like Figure 3The motor speed waveform is shown, and it can be seen that when the motor is started under no load, the given speed 750 r / min is reached for the first time at 0.015 s, overshoots to 779 r / min, the overshoot is about 3.87%, and at 0.019 s, it enters the given speed ±1% range and runs stably, and the motor speed fluctuates in the range of 752 r / min-754 r / min. When a 5 Nm step load is suddenly added at 0.2 s, the motor speed drops to 745 r / min at the lowest, and recovers to stability within 0.04 s, so that the motor speed fluctuates in the range of 746 r / min-748 r / min. At 0.3 s, the given speed rises to 1000 r / min, and the motor reaches the given speed 1000 r / min for the first time within 0.014 s, overshoots to 1012 r / min, the overshoot is about 1.2%, and at 0.016 s, it enters the given speed ±1% range and runs stably, and the motor speed fluctuates in the range of 997 r / min-999 r / min.

[0278] As shown in Figure 4 The motor torque waveform is shown, and it can be seen that when the motor is started under no load, in order to obtain a larger acceleration of the motor, the motor first outputs a higher torque, and from the figure it can be seen that the highest can reach 54 Nm, and as the motor enters the steady state, the torque drops to fluctuate in the range of -1 Nm-2 Nm, only overcoming the motor rotational damping. According to the torque fluctuation rate in the national standard "GB / T30549-2014 AC General Technical Conditions", the torque fluctuation rate is:

[0279]

[0280] Among them, T max is the maximum torque, T min is the minimum torque,

[0281] When a 5 Nm step load is suddenly added at 0.2 s, the motor output torque increases rapidly and fluctuates in the range of 3.5 Nm-7.5 Nm, and at this time the torque fluctuation rate is:

[0282]

[0283] At 0.3 s, the given speed rises to 1000 r / min, and the motor outputs a higher torque again for acceleration, and after the motor enters the steady state, it falls back to fluctuate in the range of 4.5 Nm-6.5 Nm, and at this time the torque fluctuation rate is

[0284]

[0285] In summary, the permanent magnet synchronous motor vector control system based on PI control can make the motor maintain stable operation, and the dynamic and steady-state performance is good, but there is still certain speed and torque fluctuation. To further improve the performance of the motor control system, the application will design a deadbeat predictive current algorithm for permanent magnet synchronous motor.

[0286] Step S3: Design of deadbeat predictive current control algorithm

[0287] Deadbeat predictive current control of permanent magnet motor is a control strategy based on model design. According to the voltage equation combined with feedback current, the required voltage of the motor in the next period is calculated for control, realizing fast and static error free tracking of motor current, ensuring that the motor can achieve the purpose of efficient and stable operation.

[0288] The voltage equation of permanent magnet synchronous motor in dq synchronous rotating coordinate system is:

[0289]

[0290] Where: u d , u q are the d, q axis components of stator voltage; ψ f is the permanent magnet flux linkage; i d , i q are the d, q axis components of stator current; R S is the stator resistance; L d , L q are the d, q axis components of inductance; ω e is the electrical angular velocity,

[0291] Discretization processing is performed on the voltage equation. When the sampling time T s is very small, it can be considered that the system input quantity remains constant within one sampling period, and since the mechanical time constant of the motor is much larger than the electrical time constant, the speed and back EMF change much slower than the current, so it can be considered that the back EMF term is constant within one sampling period. Therefore, equation (19) can be rewritten as:

[0292]

[0293] k is the discrete time point, on this basis, the given current is taken as the predicted current value in the next period, that is, the motor can reach the given current value in the next period, then

[0294]

[0295] Therefore, for the q-axis current loop, the predicted current value in the next period is the speed loop output; for the d-axis current loop, the predicted current value in the next period is the current loop output.

[0296] Step S4, vector control system simulation based on dead-beat predictive current control algorithm

[0297] A permanent magnet synchronous motor vector control system based on dead-beat predictive current control algorithm is built in Simulink environment, and the simulation model is shown in Figure 5

[0298] Compared with the permanent magnet synchronous motor vector control system based on dead-beat predictive current control algorithm shown in Figure 1 Figure 5 In the permanent magnet synchronous motor vector control system based on dead-beat predictive current control algorithm shown in Figure 6 Figure 7 It can be seen from the figure that the d-axis and q-axis voltage calculation is completely consistent with formula (20), and the speed loop output is equivalent to i q (k+1) as the q-axis given current. At the same time, since the system uses i d =0 control, i d (k+1) =0.

[0299] The simulation duration is set to 0.4s, the initial given speed is 750r / min, and the no-load start; the speed is unchanged at 0.2s, and a step load of 5Nm is suddenly added; the load is unchanged at 0.3s, and the given speed is changed to 1000r / min. The motor parameters are the same as in Table 1. The simulation results of the system and PI control are shown in Figure 8 Figure 9

[0300] As can be seen from Figure 8 , under dead-beat predictive current control, the motor starts with no load, and reaches the given speed of 750r / min for the first time at 0.015s, with an overshoot of 764r / min, an overshoot of about 1.87%, and quickly enters the steady state, so that the motor speed fluctuates in the range of 752.5r / min~752.7r / min. Compared with the PI control overshoot of 3.87% and the speed fluctuation of 2r / min, the dead-beat predictive current control significantly reduces the overshoot of the motor, and significantly improves the steady-state performance.

[0301] When a step load of 5Nm is suddenly added at 0.2s, the motor speed under dead-beat predictive current control drops to 745.3r / min, and recovers to stability within 0.03s, so that the motor speed fluctuates in the range of 747r / min~747.2r / min. Compared with the PI control recovery time of 0.04s and the speed fluctuation of 2r / min, the dead-beat predictive current control significantly improves the dynamic response speed of the motor, and maintains superior steady-state performance under different loads.

[0302] ​​​​​0.3s, the motor reaches the given speed of 1000r / min in the first time within 0.014s, almost without overshoot, and the motor speed fluctuates in the range of 998.3r / min~998.5r / min.

[0303] The simulation results prove that the dead-beat predictive current algorithm significantly improves the dynamic and steady-state performance of the motor control system. Compared with the PI control algorithm, which needs to continuously adjust parameters to obtain ideal control effect, this method does not need parameter tuning to achieve the best effect, greatly reducing the difficulty of algorithm use.

[0304] From Figure 9 It can be seen that under the dead-beat predictive current control, the motor outputs a high torque at no-load start to obtain a larger acceleration. As shown in the figure, the highest torque can reach 54Nm, which is similar to PI control. As the motor enters steady state, the torque drops to fluctuate in the range of 0.2Nm~1.1Nm, and the torque fluctuation rate is

[0305]

[0306] As known from the foregoing, under the same conditions, the PI control torque fluctuation rate is 300%, and it can be seen that the dead-beat predictive current control significantly improves the steady-state performance of the torque.

[0307] When a 5Nm step load is suddenly added at 0.2s, the motor output torque increases rapidly and fluctuates in the range of 5.1Nm~6.1Nm, and the torque fluctuation rate is

[0308]

[0309] As known from the foregoing, under the same conditions, the PI control torque fluctuation rate is 36.36%.

[0310] At 0.3s, the given speed rises to 1000r / min, and the motor outputs a higher torque for acceleration again, and after the motor enters steady state, it falls to fluctuate in the range of 5.4Nm~6.2Nm, and the torque fluctuation rate is

[0311]

[0312] As known from the foregoing, under the same conditions, the PI control torque fluctuation rate is 18.18%.

[0313] In summary, the permanent magnet synchronous motor vector control system based on deadbeat predictive current control can maintain stable operation of the motor, and compared with PI control, its dynamic and steady-state performance is significantly improved. However, it can be seen from equation (20) that deadbeat predictive current control uses a large number of motor parameters in the process of calculating the given voltage, resulting in a high degree of dependence of the algorithm on the accuracy of the model. To verify the dependence of deadbeat predictive current control on motor parameters, the performance of the deadbeat predictive current control algorithm under parameter mismatch conditions will be simulated and analyzed first.

[0314] Step S5, performance analysis of the deadbeat predictive current control algorithm under parameter mismatch conditions:

[0315] Parameter mismatch refers to the mismatch between the actual parameters of the motor and the parameters input into the control system. In the permanent magnet synchronous motor control system, various factors such as environmental impact on the motor, winding insulation aging, measurement error, etc. can cause parameter changes, leading to parameter mismatch. Parameter mismatch is a very common situation.

[0316] Deadbeat predictive current control has the disadvantage of high dependence on motor parameters, so parameter mismatch can cause the system performance to decline. To verify the impact of the deadbeat predictive current control algorithm under parameter mismatch conditions, this section will simulate this special condition.

[0317] As shown in equation (20), since the inductance term has a derivative, the deadbeat predictive current control algorithm is most affected by changes in inductance. At the same time, the resistance is also easily affected by factors such as motor temperature rise and can change significantly, while the permanent magnet flux linkage usually changes little. Therefore, this application mainly focuses on the impact of motor inductance and resistance mismatch on the performance of the control system.

[0318] In the simulation above, the motor inductance and resistance are:

[0319]

[0320] where R S is the stator resistance; L d , L q are the d and q axis components of inductance,

[0321] The motor parameters remain unchanged, and the resistance and inductance of the input control algorithm are reduced to 0.2 times the original, i.e.

[0322]

[0323] where L d0 and L q0 represent the initial inductance in the direct axis and quadrature axis directions, respectively; L0 represents the initial inductance; R0 represents the initial resistance,

[0324] At this time, still with the control system shown in Figure 5 The simulation results are shown in Figure 10 , Figure 11

[0325] As shown in Figure 10 It can be seen that:

[0326] Under the parameter mismatch condition, when starting with no load, the motor reaches the given speed 750 r / min for the first time at 0.016 s, overshoots to 765 r / min, the overshoot is about 2.00%, and quickly enters the steady state, so that the motor speed is maintained in the range of 751.2 r / min-752.2 r / min fluctuation. Compared with the motor overshoot of 1.87% and the speed fluctuation of 0.2 r / min without parameter mismatch, the performance of the motor system under the parameter mismatch condition is obviously decreased, but compared with the PI control overshoot of 3.87% and the speed fluctuation of 2 r / min, it still has certain advantages, and only the dynamic performance of the motor is decreased.

[0327] When a 5 Nm step load is suddenly added at 0.2 s, the motor speed under deadbeat predictive current control is the lowest and drops to 743.9 r / min, and is recovered to stable within 0.03 s, so that the motor speed is maintained in the range of 745.5 r / min-746.6 r / min fluctuation.

[0328] At 0.3 s, the given speed rises to 1000 r / min, the motor reaches the given speed 1000 r / min for the first time within 0.015 s, overshoots to 1001.7 r / min, and the motor speed is maintained in the range of 999 r / min-999.3 r / min fluctuation.

[0329] The above simulation results prove that the performance of the deadbeat predictive current algorithm is decreased under the parameter mismatch condition, but it still has certain advantages compared with the PI control.

[0330] As shown in Figure 11 It can be seen that:

[0331] Under the parameter mismatch condition, when starting with no load, in order to make the motor obtain a larger acceleration, the motor first outputs a higher torque, as can be seen from the figure, which is as high as 45 Nm, which is decreased more compared with the PI control and the condition without parameter mismatch, which leads to the motor acceleration will be smaller and the dynamic performance is decreased.

[0332] With the motor entering the steady state, the torque drops to-0.65 Nm-1.73 Nm fluctuation, and the torque fluctuation rate is

[0333]

[0334] ​As mentioned earlier, under the same operating conditions, the torque fluctuation rate of PI control is 300%, while the torque fluctuation rate of deadbeat predictive current control without parameter mismatch is 69.23%. This demonstrates that parameter mismatch has a significant impact on torque fluctuation.

[0335] When a 5 Nm step load is suddenly applied in 0.2 seconds, the motor output torque increases rapidly and fluctuates within the range of 4.3 Nm to 6.8 Nm. At this time, the torque fluctuation rate is...

[0336]

[0337] As mentioned above, under the same operating conditions, the torque fluctuation rate of PI control is 36.36%, while the torque fluctuation rate of deadbeat predictive current control without parameter mismatch is 8.93%.

[0338] At 0.3s, the given speed increases to 1000 r / min, and the motor outputs higher torque again for acceleration. After the motor enters steady state, the torque drops back to fluctuate within the range of 5.1 Nm to 6.3 Nm, at which point the torque fluctuation rate is...

[0339]

[0340] As mentioned above, under the same operating conditions, the torque fluctuation rate of PI control is 18.18%; the torque fluctuation rate of deadbeat predictive current control without parameter mismatch is 6.90%.

[0341] In summary, when the motor is in a parameter mismatch condition, the performance of the control system will be significantly affected, but it still has certain advantages compared to PI control.

[0342] To minimize the impact of parameter mismatch on the deadbeat predictive current control algorithm, a parameter identification algorithm needs to be designed to achieve online identification of permanent magnet motor parameters and feed these parameters back to the deadbeat predictive current control algorithm in real time. Below, this application will utilize model reference adaptive theory to design the motor's parameter identification algorithm.

[0343] Step S6: Parameter identification algorithm based on model reference adaptation

[0344] Determining the adjustable model:

[0345] In the model reference adaptive identification system used in this application, the permanent magnet synchronous motor is regarded as the reference model, and the mathematical model of the permanent magnet synchronous motor in the dq synchronous rotating coordinate system is regarded as the adjustable model. Since the inductance and resistance parameters that this application expects to identify are all included in the voltage equation, the adjustable model can use the voltage equation shown in equation (12).

[0346] Rewriting equation (12) in the form of a state equation, we have:

[0347]

[0348] where u d , u q are the d, q-axis components of stator voltage; ψ f is the permanent magnet flux linkage; i d , i q are the d, q-axis components of stator current; L d , L q are the d, q-axis components of inductance; ω e is the electrical angular velocity; α is the estimated value of the ratio of resistance to inductance of the motor, L is inductance, and R is resistance,

[0349] In the formula, all quantities with hat represent the output values or identified values of the adjustable model;

[0350] Reference adaptive law design:

[0351] In the process of establishing the model reference adaptive identification system, the most critical link of the whole system is to design a reasonable and suitable adaptive law. Generally, the adaptive law can be designed based on Lyapunov stability theory. In this application, Lyapunov stability theory is selected to design the adaptive law of the permanent magnet synchronous motor parameter identification system.

[0352] Discrimination theorem of large range asymptotic stability in Lyapunov stability theory:

[0353] Lyapunov stability theory can guarantee large range uniform asymptotic stability in equilibrium state. For a continuous time nonlinear time-varying self-regulating system, if a scalar function V(x, t) can be constructed, V(0, t) = 0, and for all non-zero state points x in the state space, the following conditions are met:

[0354] V(x, t) is positive definite and bounded, and when ||x||→∞, V(x, t)→∞;

[0355] The partial derivative with respect to time is negative and bounded.

[0356] Then the zero equilibrium state x = 0 of the system is a large range uniform asymptotic stable equilibrium point.

[0357] According to the above Lyapunov stability theory, the adaptive law is designed.

[0358] First, simplify the adjustable model shown in formula (30), let:

[0359]

[0360] where e is the error vector, e1, e2 are the vector elements; I is the actual current vector, U is the voltage vector, W is the angular velocity vector, A is the system matrix, B is the input matrix, C is the output matrix, L is the motor inductance,

[0361] Differentiate all terms in equation (31) with respect to time, we have:

[0362]

[0363] Let:

[0364]

[0365] where, is the transpose of the three difference values a, b, c; s is a column vector,

[0366] Substitute equation (33) into equation (32), we have:

[0367]

[0368] The constructor V(x, t):

[0369]

[0370] It is not difficult to see that equation (35) satisfies the condition 1 of Lyapunov stability theory, that is, V([e, φ], t) is positive definite and bounded, and when ||[e, φ]||→∞, V[e, φ], t)→∞.

[0371] At this time, we only need to derive when to meet negative. Write expression and simplify, we have:

[0372]

[0373] According to equation (31), we have:

[0374]

[0375] where A T is the transpose of the system matrix A,

[0376] That is, the first term of the formula is obviously negative, so we also need to ensure that the second term is zero, that is, we can guarantee negative. Therefore, let:

[0377]

[0378] Then we have:

[0379]

[0380] Solving the system of equations yields the adaptive law as follows:

[0381]

[0382] Among them, K L and K R It is a gain coefficient related to inductance and resistance estimation, where L0 represents the initial inductance and R0 represents the initial resistance.

[0383] The adaptive law designed based on Lyapunov stability theory is shown in equation (40). Combined with the adjustable model, a model reference adaptive parameter identification system can be constructed.

[0384] Step S7: Simulation of Model Reference Adaptive Parameter Identification Algorithm

[0385] To verify the effectiveness of the model reference adaptive parameter identification algorithm, simulation verification is performed in the MATLAB / Simulink environment. The simulation model is as follows: Figure 12 As shown in the figure, the control system is the same as the vector control system based on deadbeat predictive current control, and the feedback current i is obtained from the control system. d i q Electric angular velocity and current loop output voltage u d u q It serves as the input to the model reference adaptive parameter identification algorithm.

[0386] The model reference adaptive parameter identification algorithm part is as follows: Figures 13-15 As shown. Figure 13 The diagram shows the overall structure of the model reference adaptive parameter identification algorithm. The current loop output voltage u... d u q and electric angular velocity ω e As input to the adjustable model, it is used to calculate the estimated current; the difference between the actual current and the estimated current is used to obtain the current error, and the current error is input into the adaptive law, which outputs the resistance and inductance parameters; the identified parameters are then fed back to the adjustable model. By adjusting the parameters in the adjustable model, the estimated current output by the adjustable model is made to get closer and closer to the actual current. The smaller the error between the two, the more accurate the adjustable model is considered to be, and the more accurate the identified parameters are.

[0387] Figure 14 The diagram shows the adjustable model part of the model reference adaptive parameter identification algorithm. This part is divided into d-axis and q-axis adjustable models, corresponding to the d-axis and q-axis voltage equations in equation (30). The function of the adjustable model is to utilize the current loop output voltage u d u q and electric angular velocity ω eThe estimated current is calculated according to the motor parameters output by the adaptive law, and the identification parameters are gradually close to the actual parameters by making the difference between the estimated current and the actual current smaller and smaller.

[0388] Figure 15 For the adaptive law implementation part, it is designed according to formula (40). The purpose of the adaptive law is to adjust the parameters according to the error between the estimated current output by the adjustable model and the actual current, and feedback to the adjustable model, so that it gradually approaches the actual model.

[0389] The simulation is carried out using the above model, and the simulation time is still set to 0.4s, the given speed is 750r / min, and the load is 5Nm. The parameter identification result is shown in Figure 16 、 Figure 17

[0390] Since the surface-mounted permanent magnet synchronous motor selected in the application has equal d and q axis inductances, only one of them needs to be identified. As shown in the figure, the inductance identification starts from 0 and gradually approaches the actual inductance, reaches 10.9mH at 0.1s, reaches 11.5mH at 0.15s, and remains around 11.5mH. Compared with the actual inductance of 12mH, the identification error is about 4.17%.

[0391] As shown in Figure 17 , the resistance identification starts from 0 and gradually approaches the actual resistance, reaches 5.41Ω at 0.15s, and remains around 5.35Ω. Compared with the actual inductance of 4.875Ω, the identification error is about 9.74%.

[0392] From the identification result, the model reference adaptive parameter identification algorithm has fast response speed and high identification accuracy. Since the non-linear factors such as inverter tube voltage drop are not considered in the adjustable model, there is inevitably some error in the identification result, but the error can be kept within 10%.

[0393] Step S8, simulation of permanent magnet motor deadbeat predictive current control system based on parameter identification:

[0394] After completing the parameter identification algorithm, the identified parameters are fed back to the deadbeat predictive current control algorithm to improve its anti-disturbance ability to parameter mismatch.

[0395] The simulation model is consistent with Figure 15 According to the simulation results in the foregoing, the parameter identification algorithm needs a certain convergence time to make its identification value close to the actual value. Therefore, considering the motor response speed, the application selects to input the identified parameters to the deadbeat predictive current algorithm at 0.05s.

[0396] ​For the convenience of comparing the simulation results, the motor selected parameters are consistent with Table 1, and all simulation basic settings are consistent with the foregoing. That is, the simulation duration is still set to 0.4s, the initial given speed is kept at 750r / min, and the motor is started under no load; the speed is kept unchanged at 0.2s, and a step load of 5Nm is suddenly added; the load is kept unchanged at 0.3s, and the given speed is raised to 1000r / min. The initial value of the parameter in the deadbeat predictive current algorithm is consistent with that in Section 3.3, which is 0.2 times the actual value, and the identified value is input into the algorithm at the 0.05s moment.

[0397] As shown in FIG. 6, the motor speed comparison diagram of the permanent magnet synchronous motor under the conditions of no parameter mismatch, parameter mismatch, and parameter mismatch using parameter identification is shown. It can be seen that, since the identified parameter is added to the system after 0.05s, the no-load starting performance is basically consistent with that under the parameter mismatch condition, and the motor speed still reaches the given speed of 750r / min for the first time at 0.016s, overshoots to 765.5r / min, and the overshoot amount is about 2.07%, and after entering the steady state, the motor speed fluctuates in the range of 752r / min-752.9r / min. Figure 18 After the identified parameter is added to the algorithm at 0.05s, the motor speed fluctuation is reduced and maintained in the range of 752.6r / min-753.1r / min. Compared with the motor overshoot of 1.87% and the speed fluctuation of 0.2r / min under the condition of no parameter mismatch, the performance is slightly reduced, but the speed fluctuation is significantly lower than the 1r / min under the condition of parameter mismatch without using parameter identification.

[0398] When a step load of 5Nm is suddenly added at 0.2s, the motor speed is reduced to a minimum of 745.5r / min and recovers to stability within 0.03s, so that the motor speed fluctuates in the range of 747r / min-747.3r / min. Compared with the case where the motor speed is reduced to a minimum of 743.9r / min under the condition of parameter mismatch without using parameter identification, and the motor speed fluctuates in the range of 745.5r / min-746.6r / min after 0.03s, there is a certain optimization.

[0399] At 0.3s, the given speed is raised to 1000r / min, and the motor reaches the given speed of 1000r / min for the first time within 0.015s, overshoots to 1001r / min, and the motor speed fluctuates in the range of 998.5r / min-998.8r / min.

[0400] The above simulation results prove that the parameter identification algorithm can effectively improve the performance of the deadbeat predictive current algorithm under the condition of parameter mismatch.

[0401] As shown in FIG. 7, the motor speed comparison diagram of the permanent magnet synchronous motor under the conditions of no parameter mismatch, parameter mismatch, and parameter mismatch using parameter identification is shown. It can be seen that, since the identified parameter is added to the system after 0.05s, the no-load starting performance is basically consistent with that under the parameter mismatch condition, and the motor speed still reaches the given speed of 750r / min for the first time at 0.016s, overshoots to 765.5r / min, and the overshoot amount is about 2.07%, and after entering the steady state, the motor speed fluctuates in the range of 752r / min-752.9r / min.

[0402] Figure 19 ​As shown, the same because the identification parameters are input into the deadbeat predictive current algorithm at 0.05s, the motor starting performance is basically consistent with the parameter mismatch condition.

[0403] With the motor entering the steady state, the identification parameters are added to the deadbeat current predictive control algorithm after 0.05s, the torque is stabilized to fluctuate in the range of 0.1Nm-1.1Nm, and the torque fluctuation rate is:

[0404]

[0405] From the foregoing, under the same condition, the torque fluctuation rate of PI control is 300%; the torque fluctuation rate of the deadbeat predictive current control without parameter mismatch is 69.23%; and the torque fluctuation rate of the deadbeat predictive current control with parameter mismatch is 220.37%. It can be seen that the parameter identification algorithm can effectively improve the performance of the deadbeat predictive current algorithm under the condition of parameter mismatch.

[0406] When a 5Nm step load is suddenly added at 0.2s, the motor output torque rapidly increases and fluctuates in the range of 5.1Nm-6.1Nm, and the torque fluctuation rate is:

[0407]

[0408] From the foregoing, under the same condition, the torque fluctuation rate of PI control is 36.36%; the torque fluctuation rate of the deadbeat predictive current control without parameter mismatch is 8.93%; and the torque fluctuation rate of the deadbeat predictive current control with parameter mismatch is 22.52%. It can be seen that under the action of the parameter identification algorithm, the motor torque steady state performance is basically consistent with that without parameter mismatch.

[0409] At 0.3s, the given speed rises to 1000r / min, the motor outputs higher torque again for acceleration, and after the motor enters the steady state, it falls back to fluctuate in the range of 5.4Nm-6.4Nm, and the torque fluctuation rate is

[0410]

[0411] From the foregoing, under the same condition, the torque fluctuation rate of PI control is 18.18%; the torque fluctuation rate of the deadbeat predictive current control without parameter mismatch is 6.90%; and the torque fluctuation rate of the deadbeat predictive current control with parameter mismatch is 10.53%.

[0412] In summary, under the action of the parameter identification algorithm, the performance of the control system is significantly improved under the condition of parameter mismatch.

[0413] The above merely describes the preferred embodiments of the present application, but does not constitute any other form of limitation to the present application, and any modification or equivalent change made according to the technical essence of the present application still falls within the scope of the present application.

Claims

1. A dead-beat current predictive control method based on parameter adaptive online identification, characterized in that, Comprising the following steps: S1, mathematical modeling of three-phase permanent magnet synchronous motor; S2, simulation of vector control system based on PI control; S3, design of deadbeat predictive current control algorithm; S4, simulation of vector control system based on deadbeat predictive current control algorithm; S5, performance analysis of deadbeat predictive current control algorithm under parameter mismatch condition; S6, parameter identification algorithm based on model reference self-adaptation; S7, simulation of model reference self-adaptation parameter identification algorithm; S8, simulation of permanent magnet motor deadbeat predictive current control system based on parameter identification.

2. The method of claim 1, wherein, Step S1 is as follows: S11, mathematical model of three-phase permanent magnet synchronous motor in three-phase stationary coordinate system, specifically as follows: Among them, the voltage equation is: The flux linkage equation is: Ψ 3s = L 3s i 3s + ψ f · F 3s (θ e ) (2) where: u 3s , R, i 3s , ψ 3s are the phase voltage, resistance, current and flux linkage respectively; L 3s is the inductance of the three-phase winding; ψ f is the permanent magnet flux linkage; F 3S (θ e ) is the unit flux linkage and satisfies: where: L m3 is the stator mutual inductance; L sσ is the stator leakage inductance; i A i B i C is the three-phase current; R is the resistance; ψ A ψ B ψ C is the three-phase flux linkage; u A u B u C is the three-phase voltage; θ e is the electrical angle representing the electrical angle of the armature rotation in the motor; In addition, the mechanical motion equation of the motor is as follows: where p n is the number of pole pairs of the three-phase PMSM; T e is the electromagnetic torque; θ m is the mechanical angle; S12, to realize vector control, coordinate transformation is performed on the above mathematical model to obtain the mathematical model in the synchronous rotating coordinate system, specifically as follows: where J is the moment of inertia of the motor; T L is the load torque; B is the viscous damping coefficient; ω m is the motor mechanical angular velocity; S121, coordinate transformation: The quantities in the three-phase stationary coordinate system ABC are transformed to the two-phase stationary coordinate system α-β by Clark transformation, and the coordinate transformation formula is as follows: In the formula, the coordinate transformation matrix is: In the formula, the coordinate transformation matrix is: By Park transformation, the two-phase stationary coordinate system α-β is transformed into the synchronous rotating coordinate system d-q, and the coordinate transformation formula is as follows: S122, mathematical model of permanent magnet synchronous motor in synchronous rotating coordinate system: wherein f α f β f o represent the components of a, b, o in the a- b coordinate system; f A f B f C represent the components of A, B, C in the three-phase coordinate system; is a transformation matrix for conversion from the three-phase 3s coordinate system to the two-phase stationary 2s coordinate system; f d f q components in the d-q coordinate system; denotes a transformation matrix from the a- b coordinate system to the d-q coordinate system, wherein the indices 2s and 2r denote the two-phase stationary coordinate system and the two-phase rotating coordinate system, respectively; Using Clark transformation and Park transformation, the permanent magnet motor can be converted from three-phase stationary coordinate system to two-phase rotating coordinate system, that is, the stator voltage equation can be obtained: The stator flux linkage equation is: The electromagnetic torque equation is: The motion equation is: where u d , u q are the d, q-axis components of the stator voltage; ψ d , ψ q , ψ f are the d, q-axis components of the stator flux and the permanent magnet flux; i d , i q are the d, q-axis components of the stator current; R s is the stator resistance; L d , L q are the d, q-axis components of the inductance; ω e is the electrical angular velocity, Step S2 is as follows: S21, build vector control system based on PI control Adopting surface-mounted three-phase permanent magnet synchronous motor, its L d = L q , the torque formula shown in formula (13) can be seen that the electromagnetic torque is completely determined by the cross-axis current i q , that is, to ensure i d = 0, the maximum torque under the same current can be obtained.

3. The method of claim 1, wherein the parameters are updated according to the following equation: ###0001### where K is a constant, and ###0002### is the difference between the current value of the parameter and the previous value of the parameter. S22, system control strategy design The closed-loop control strategy of the above speed-current double closed-loop structure permanent magnet synchronous motor vector control system adopts PID control strategy, and its mathematical model is as follows: A permanent magnet synchronous motor vector control system based on PI control of speed-current double closed loop structure is built in Simulink environment, which is i d A three-phase permanent magnet synchronous motor vector control system is built. The speed loop of the system is the difference between the given speed and the actual speed, and the q-axis reference current is outputted by the speed loop regulator. The current loop is also the difference between the current given signal and the current feedback signal, and the reference voltage is outputted by the current loop regulator. The speed loop and the current loop regulator are both designed based on PI control algorithm. S23, system simulation The simulation time is set to 0.4s, the initial given speed is 750r / min, and the motor starts under no-load condition; At 0.2s, the speed remains unchanged, and a step load of 5Nm is suddenly added; At 0.3s, the load remains unchanged, and the given speed changes to 1000r / min, u(t) = K p e(t) + K i ∫e(t)dt (15); Among them, K p K is the proportionality coefficient. i K is the integral coefficient. d Here, is the differential coefficient; u(t) represents the controller output, i.e., the control quantity; e(t) represents the error signal, and ∫e(t)dt represents the integral of the error signal e(t) over time; Analysis of motor speed in simulation of permanent magnet synchronous motor vector control system based on PI control: ​ ​ The motor reaches the given speed 750 r / min for the first time at 0.015 s, overshoots to 779 r / min, the overshoot is about 3.87%, and enters the given speed ±1% range for stable operation at 0.019 s, the motor speed fluctuates in the range of 752 r / min-754 r / min, when a 5 Nm step load is suddenly added at 0.2 s, the motor speed drops to 745 r / min at the lowest, and recovers to stable state within 0.04 s, so that the motor speed fluctuates in the range of 746 r / min-748 r / min, the given speed rises to 1000 r / min at 0.3 s, the motor reaches the given speed 1000 r / min for the first time within 0.014 s, overshoots to 1012 r / min, the overshoot is about 1.2%, and enters the given speed ±1% range for stable operation at 0.016 s, the motor speed fluctuates in the range of 997 r / min-999 r / min; Motor torque analysis of the permanent magnet synchronous motor vector control system simulation based on PI control: When the motor starts under no load, the motor first outputs high torque, which can reach 54 Nm at the highest, and as the motor enters the steady state, the torque drops to fluctuate in the range of -1 Nm-2 Nm, and only overcomes the motor rotation damping, the torque fluctuation rate is: where T max is the maximum torque value, T min is the minimum torque value, When a 5 Nm step load is suddenly added at 0.2 s, the motor output torque increases rapidly and fluctuates in the range of 3.5 Nm-7.5 Nm, at this time the torque fluctuation rate is: When the given speed rises to 1000 r / min at 0.3 s, the motor outputs high torque again for acceleration, and after the motor enters the steady state, it falls to fluctuate in the range of 4.5 Nm-6.5 Nm, at this time the torque fluctuation rate is:

4. The method of claim 1, wherein, Step S3 is specifically as follows: The voltage equation of the three-phase permanent magnet synchronous motor in the dq synchronous rotating coordinate system is: where: u d , u q are the d, q-axis components of the stator voltage; ψ f is the permanent magnet flux linkage; i d , i q are the d, q-axis components of the stator current; R S is the stator resistance; L d , L q are the d, q-axis components of the inductance; ω e is the electrical angular velocity, Discretization of the voltage equation is performed when the sampling time T s When the preset threshold is reached, it is considered that the system input quantity remains constant within a sampling period, and since the mechanical time constant of the motor is greater than the electrical time constant, the speed and back EMF change much slower than the current. It can be considered that the back EMF term is constant within a sampling period, and equation (19) can be rewritten as: k is a discrete time point, on this basis, the given current is taken as the predicted current value of the next period, that is, it is considered that the motor can reach the given current value in the next period, so that: Therefore, for the q-axis current loop, the predicted current value of the next period is the output of the speed loop, and for the d-axis current loop, the predicted current value of the next period is the output of the current loop.

5. The method of claim 1, wherein, Step S4 is specifically as follows: S41, build a vector control system of the dead-beat predictive current control algorithm The permanent magnet synchronous motor vector control system based on the deadbeat predictive current control algorithm is built in the Simulink environment, wherein the PI controller of the current loop is replaced by the deadbeat predictive current control algorithm, the d and q axis voltage calculation is completely consistent with formula (20), and the speed loop output as the q axis given current is equivalent to i q (k+1) is obtained, and at the same time, since the system uses i d (k+1) = 0 control, i d (k+1) = 0. S42, system simulation The simulation duration is set to 0.4 s, the initial given speed is 750 r / min, the motor starts under no load, the speed does not change at 0.2 s, and a step load of 5 Nm is suddenly added; the load does not change at 0.3 s, and the given speed becomes 1000 r / min, Motor speed analysis of the permanent magnet synchronous motor vector control system simulation based on dead-beat predictive current control: Under the dead-beat predictive current control, when the motor starts under no load, it reaches the given speed 750 r / min for the first time at 0.015 s, overshoots to 764 r / min, the overshoot is about 1.87%, and quickly enters the steady state, so that the motor speed fluctuates in the range of 752.5 r / min-752.7 r / min, compared with the PI control overshoot of 3.87% and the speed fluctuation of 2 r / min, 0.2s sudden 5Nm step load, the motor in the deadbeat predictive current control under the speed of the lowest drop to 745.3r / min, and in 0.03s recovery stable, make the motor speed maintain in 747r / min~747.2r / min fluctuation, compared with PI control recovery time 0.04s, speed fluctuation 2r / min, 0.3s given speed up to 1000r / min, the motor in 0.014s first reaches the given speed 1000r / min, overshoot less than 0.1%, the motor speed maintain in 998.3r / min~998.5r / min fluctuation, The motor torque analysis of permanent magnet synchronous motor vector control system simulation based on deadbeat predictive current control algorithm: Motor in the deadbeat predictive current control, no load start, the motor first output high torque, the highest can reach 54Nm, with the motor into steady state, torque drop to 0.2Nm~1.1Nm fluctuation, torque fluctuation rate: Under the same conditions, PI control torque fluctuation rate is 300%, 0.2s sudden 5Nm step load, the motor output torque increases rapidly, and maintain in 5.1Nm~6.1Nm fluctuation, at this time the torque fluctuation rate: Under the same conditions, PI control torque fluctuation rate is 36.36%, 0.3s given speed up to 1000r / min, the motor again output high torque for acceleration, and after the motor into steady state back to 5.4Nm~6.2Nm fluctuation, at this time the torque fluctuation rate: Under the same conditions, PI control torque fluctuation rate is 18.18%.

6. The method of claim 1, wherein, Step S5 is specifically as follows: In the permanent magnet synchronous motor vector control system based on deadbeat predictive current control algorithm, the motor inductance, resistance is respectively: wherein R S is the stator resistance; L d , L q are the d, q-axis components of the inductance, respectively, Keep the motor parameters unchanged, while the input based on deadbeat predictive current control algorithm resistance, inductance is reduced to the original 0.2 times, namely: wherein L d0 and L q0 respectively represent initial inductance in the direct axis and quadrature axis directions; L0 represents initial inductance; R0 represents initial resistance, At this time, the permanent magnet synchronous motor vector control system based on deadbeat predictive current control algorithm is simulated, The motor speed analysis of parameter mismatch condition under deadbeat predictive current control algorithm simulation: Motor in the parameter mismatch condition, no load start, the motor in 0.016s first reaches the given speed 750r / min, overshoot to 765r / min, overshoot about 2.00%, and quickly into steady state, make the motor speed maintain in 751.2r / min~752.2r / min fluctuation, compared with no parameter mismatch condition, motor overshoot 1.87%, speed fluctuation 0.2r / min, compared with PI control, motor overshoot 3.87%, speed fluctuation 2r / min, 0.2s sudden 5Nm step load, the motor in the deadbeat predictive current control under the speed of the lowest drop to 743.9r / min, and in 0.03s recovery stable, make the motor speed maintain in 745.5r / min~746.6r / min fluctuation, 0.3s, the given speed rises to 1000r / min, the motor reaches the given speed 1000r / min for the first time in 0.015s, overshoots to 1001.7r / min, and the motor speed fluctuates in the range of 999r / min-999.3r / min, Motor torque analysis of the simulation of the deadbeat predictive current control algorithm under the condition of parameter mismatch: When the motor is started under no-load condition, the motor first outputs high torque, up to 45Nm, and as the motor enters steady state, the torque drops to fluctuate in the range of-0.65Nm-1.73Nm, with torque fluctuation rate of: Under the same condition, the torque fluctuation rate of PI control is 300%, and the torque fluctuation rate of deadbeat predictive current control without parameter mismatch is 69.23%, When a 5Nm step load is suddenly added at 0.2s, the motor output torque increases rapidly and fluctuates in the range of 4.3Nm-6.8Nm, with torque fluctuation rate of: Under the same condition, the torque fluctuation rate of PI control is 36.36%, and the torque fluctuation rate of deadbeat predictive current control without parameter mismatch is 8.93%, When the given speed rises to 1000r / min at 0.3s, the motor outputs high torque again for acceleration, and after the motor enters steady state, it falls to fluctuate in the range of 5.1Nm-6.3Nm, with torque fluctuation rate of: Under the same condition, the torque fluctuation rate of PI control is 18.18%, and the torque fluctuation rate of deadbeat predictive current control without parameter mismatch is 6.90%.

7. The method of claim 1, wherein the parameters are updated according to the following equation: ###00003### where K is a constant, and ###00004### is the difference between the current value of the parameter and the previous value of the parameter. Step S6 is specifically as follows: S61, determination of adjustable model The three-phase permanent magnet synchronous motor is regarded as a reference model, and the mathematical model of the three-phase permanent magnet synchronous motor in the d-q synchronous rotating coordinate system is regarded as an adjustable model. Since the expected identified inductance and resistance parameters are all included in the voltage equation, the adjustable model can use the voltage equation shown in formula (12), Rewrite formula (12) into the form of state equation, then where u d , u q are the d, q-axis components of the stator voltage; ψ f is the permanent magnet flux linkage; i d , i q are the d, q-axis components of the stator current; L d , L q are the d, q-axis components of the inductance; ω e is the electrical angular velocity; and α is an estimate of the ratio of the resistance to the inductance of the machine, L is the inductance, and R is the resistance. In the formula, all quantities with hat represent output values or identified values of the adjustable model; S62, reference adaptive law design Lyapunov stability theory is used to design the adaptive law of the three-phase permanent magnet synchronous motor parameter identification system. The large-range asymptotic stability judgment theorem in Lyapunov stability theory is as follows: To ensure large-range consistent asymptotic stability in the equilibrium state, for a continuous-time nonlinear time-varying self-regulating system, if a scalar function V(x, t) can be constructed, V(0, t)=0, and for all non-zero state points x in the state space, the following conditions are met: V(x, t) is positive definite and bounded, and when ||x||→∞, V(x, t)→∞; The partial derivative with respect to time is negative, definite, and bounded. Then the zero equilibrium state x=0 of the system is a large-range consistent asymptotic stable equilibrium point, According to the above Lyapunov stability theory, the adaptive law is designed, First, simplify the adjustable model shown in formula (30), let Where e is the error vector, e1 and e2 are vector elements; I is the actual current vector, U is the voltage vector, W is the angular velocity vector, A is the system matrix, B is the input matrix, C is the output matrix, and L is the motor inductance; Solve all terms in formula (31) simultaneously and take the partial derivative with respect to time, then Let wherein is the transpose of the three differences a, b, c; s is a column vector, Substitute equation (33) into equation (31), then we have: The constructor V(x, t): wherein, represents an input vector to the function V; e T is the transpose of the error vector; is the state vector, Equation (35) satisfies the condition 1 of Lyapunov stability theory, that is, V([e, φ], t) is positive definite and bounded, and when ||[e, φ]||→∞, V([e, φ], t)→∞, At this time, we only need to derive when to meet what conditions negative, write expression and simplify, have: According to equation (31), we have where A T is the transpose of the system matrix A, That is, the first term of the equation is obviously negative, so it is necessary to ensure that the second term is zero, that is, to ensure negative, so let Then we have: Solving the equation set, we can get the adaptive law as: where K L and K R are gain coefficients related to inductance and resistance estimation, L0 represents initial inductance; R0 represents initial resistance, the adaptive law designed according to Lyapunov stability theory is shown as formula (40), and the model reference adaptive parameter identification system can be constructed in combination with the adjustable model.

8. The method of claim 1, wherein, Step S7 is specifically as follows: S71, simulation verification is carried out in the MATLAB\Simulink environment S711, obtaining a feedback current i from the vector control system based on the dead-beat predictive current control d 、 q , an electrical angular velocity and a current loop output voltage u d 、 q as an input of a model reference adaptive parameter identification algorithm for calculating an estimated current; S712, the actual current and the estimated current are subtracted to obtain the current error, and the current error is input into the adaptive law, and the resistance and inductance parameters are output from the adaptive law; S713, the identified parameters are fed back to the adjustable model, the parameters in the adjustable model are adjusted, the estimated current output by the adjustable model is closer and closer to the actual current, the error between the two is smaller and smaller, and it is considered that the adjustable model is more accurate, that is, the identified parameters are more accurate; S72, system simulation The simulation time is set to 0.4s, the given speed is 750r / min, and the load is 5Nm. Since the surface-mounted permanent magnet synchronous motor is selected, the d-axis and q-axis inductances are equal, so only one of them needs to be identified. The identified inductance starts from 0 and gradually approaches the actual inductance. At 0.1s, it reaches 10.9mH, at 0.15s, it reaches 11.5mH, and it is maintained at 11.5±0.5mH. Compared with the actual inductance of 12mH, the identification error is about 4.17%, The identified resistance starts from 0 and gradually approaches the actual resistance. At 0.15s, it reaches 5.41Ω, and is maintained at 5.35±0.5Ω. Compared with the actual inductance of 4.875Ω, the identification error is about 9.74%, Based on the identification results of the above simulation, the error of the identification results is always maintained within 10%.

9. The method of claim 1, wherein, Step S8 is specifically as follows: The identified parameters are fed back to the deadbeat predictive current control algorithm to form a permanent magnet motor deadbeat predictive current control system simulation based on parameter identification, which is specifically as follows: The simulation time is set to 0.4s, and the initial given speed is maintained at 750r / min. At the same time, the motor is started under no-load, Motor speed analysis using parameter identification under no parameter mismatch condition, parameter mismatch condition and parameter mismatch condition: The motor no-load starting performance is basically the same as that under the parameter mismatch condition. With the motor entering the steady state, at 0.05s, the identified parameters are added, the motor speed fluctuation starts to decrease and is maintained in the range of 752.6r / min~753.1r / min. Compared with the motor overshoot of 1.87% and the speed fluctuation of 0.2r / min under the condition of no parameter mismatch, the speed fluctuation is significantly lower than 1r / min under the condition of no parameter mismatch without using parameter identification, 0.2s sudden 5Nm step load, the motor speed drops to 745.5r / min, and recovers to stable in 0.03s, the motor speed maintains in 747r / min~747.3r / min range fluctuation, compared with the parameter mismatch without using parameter identification when the motor speed drops to 743.9r / min, and the motor speed maintains in 745.5r / min~746.6r / min range fluctuation after 0.03s, it has obvious optimization, 0.3s given speed rises to 1000r / min, the motor reaches the given speed 1000r / min in 0.015s for the first time, overshoots to 1001r / min, and the motor speed maintains in 998.5r / min~998.8r / min range fluctuation, Motor torque analysis under the condition of no parameter mismatch, parameter mismatch and parameter mismatch using parameter identification: The motor no-load starting performance is consistent with the parameter mismatch condition, and with the motor entering steady state, at 0.05s, the identification parameters are added, the torque is stabilized to 0.1Nm~1.1Nm range fluctuation, the torque fluctuation rate is: 0.2s sudden 5Nm step load, the motor output torque increases rapidly, and maintains in 5.1Nm~6.1Nm range fluctuation, at this time the torque fluctuation rate is: Under the same condition, the PI control torque fluctuation rate is 36.36%; The torque fluctuation rate of deadbeat predictive current control without parameter mismatch is 8.93%; The torque fluctuation rate of deadbeat predictive current control with parameter mismatch is 22.52%; 0.3s given speed rises to 1000r / min, the motor outputs high torque again for acceleration, and falls to 5.4Nm~6.4Nm range fluctuation after the motor enters steady state, at this time the torque fluctuation rate is: Under the same condition, the PI control torque fluctuation rate is 18.18%; The torque fluctuation rate of deadbeat predictive current control without parameter mismatch is 6.90%; The torque fluctuation rate of deadbeat predictive current control with parameter mismatch is 10.53%.

Images