Three-level inverter-based pmsm three-vector model predictive current control method

Abstract

The application discloses a PMSM three-vector model predictive current control method based on a three-level inverter without weight coefficients, which selects two non-zero voltage vectors through two rounds of prediction in each period, and performs time distribution on the two non-zero voltage vectors and zero vectors, so that the applied vectors are closer to a reference voltage vector, and a virtual vector method is used to process small vectors, the amount of charges flowing in and out from a midpoint is controlled by changing the action time of the small vectors and redundant small vectors of the small vectors, the midpoint potential is changed, and the fluctuation of the midpoint voltage of the three-level inverter is suppressed. Compared with a traditional single-vector model predictive control method, the three-vector model predictive current control method used in the application reduces the torque ripple of the motor and improves the current waveform. The midpoint voltage balance term in the value function of the traditional model predictive control method is eliminated, the weight coefficient is eliminated, the calculation amount is reduced, and the current control precision is improved.

Description

Technical Field

[0001] This invention relates to the field of motor control, and in particular to a PMSM three-vector model predictive current control method based on a three-level inverter without weighted coefficients. Background Technology

[0002] Model predictive control algorithms rely on vector optimization using a predictive model and a value function. By inputting each voltage vector into the predictive model and selecting the optimal voltage vector through the value function, the motor is controlled. Since there is a midpoint current when the load is connected to the midpoint of a three-level inverter, the midpoint potential inevitably shifts during operation, affecting the accuracy of control. Traditional model predictive methods add a midpoint potential balancing term to the value function to suppress the fluctuation of the midpoint potential of the three-level inverter. However, since its weighting coefficient is not easy to tune, and the addition of the midpoint potential balancing term inevitably has a certain impact on the accurate control of the current. Summary of the Invention

[0003] To address the problems existing in the prior art, this invention discloses a PMSM three-vector model predictive current control method based on a three-level inverter without weighted coefficients, specifically including the following steps:

[0004] During motor operation, its rotor position information θ and motor speed n are collected, and the given speed n is... * The difference between the actual speed n fed back by the motor and the speed n is used by a PI controller to obtain the q-axis current of the motor. Combination The control strategy completes the motor speed loop control and collects the three-phase current i of the motor. a i b i c Three-phase voltage u a u b u c And the inverter midpoint voltage, used for current loop three-vector model predictive current control.

[0005] The prediction model for the dq-axis stator current of a permanent magnet synchronous motor is as follows:

[0006]

[0007] Where i d (k),i q (k) represents the stator current at the current moment, i d (k+1),i q (k+1) represents the predicted stator current, u d (k),u q (k) The current time is the dq axis stator voltage, R s ω is the stator winding resistance, L is the stator winding self-inductance, and ω is the stator winding resistance.e Let ψ be the electric angular velocity of the motor rotor. f For permanent magnet flux linkage in motors.

[0008] The first voltage vector is selected using the value function. Then, the 27 voltage vectors are input into the stator current prediction model, and the vectors selected are those that maximize the value function. The smallest voltage vector is taken as the first voltage vector.

[0009] After selecting the first voltage vector, the slope of the current change along the dq axis of the motor under the action of each voltage vector is calculated using the following formula.

[0010]

[0011] Where f d0 f q0 f d_opt1 f q_opt1 f d_i f q_i These are the slopes of the d-axis and q-axis currents generated when the zero vector, the first voltage vector, and the alternative vector of the second vector are applied, respectively.

[0012] Using the deadbeat principle of current, the time allocation of the selected first voltage vector and the 27 candidate voltage vectors and zero vector of the second voltage vector is performed, and the action time t of the three vectors when different voltage vectors are used as the second voltage vector is calculated. opt1 ,t opt2 From t0, a total of 27 composite voltage vectors composed of three vectors can be obtained. The deadbeat formula for current is:

[0013]

[0014] Where i d (k+1) and i q (k+1) represent the predicted current values ​​along the d and q axes, respectively, i d * and i q * These are the q-axis current setpoints.

[0015] Substituting the 27 synthesized voltage vectors obtained above into the current prediction model, the second voltage vector used to obtain the synthesized vector that minimizes the value function is taken as the optimal voltage vector obtained in the second round of prediction.

[0016] The obtained three voltage vectors and their corresponding durations are output to the inverter to control the motor operation.

[0017] If the selected vector is a small vector, since the small vector and its corresponding redundant small vector generate currents at the midpoint that are equal in magnitude but opposite in direction, the following formula can be used to allocate time between it and its corresponding redundant small vector, thereby reducing the potential fluctuation at the midpoint of the three-level inverter. The principle of time allocation is as follows:

[0018] When the first voltage vector is a small vector, the time allocation is as follows:

[0019] From the capacitance formula Q = CU and the charge formula Q = it, we can obtain that within one period

[0020]

[0021] Seeking

[0022]

[0023] U o U is the voltage at the midpoint of the inverter. dc Where C is the DC bus voltage, C is the DC side capacitor value, and t is the DC bus voltage. opt1 To calculate the time that the first vector should act, t1 is the actual time that the small vector acts, and t2 is the actual time that its corresponding redundant small vector acts.

[0024] Similarly, when the second voltage vector is a small vector, the time allocation is as follows:

[0025]

[0026] By adopting the above technical solution, this invention provides a PMSM three-vector model predictive current control method for three-level inverters without weighting coefficients. This method applies three voltage vectors in each cycle to improve the control effect. By using the virtual vector method to balance the midpoint voltage, and by changing the action time of the small vector and its redundant small vector, the amount of charge flowing into and out of the midpoint is controlled, thereby changing the midpoint potential. This eliminates the weighting coefficient, reduces the amount of calculation, and improves the accuracy of current control. Attached Figure Description

[0027] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0028] Fig. 1 This is a voltage vector diagram of the three-level inverter in this invention.

[0029] Fig. 2This is a block diagram of the three-vector model predictive current control system without weighting coefficients in this invention.

[0030] Fig. 3 This is a speed diagram of the three-phase permanent magnet synchronous motor in this invention.

[0031] Fig. 4 Torque diagram of the three-phase permanent magnet synchronous motor in this invention.

[0032] Fig. 5 The diagram shows the three-phase current of the three-phase permanent magnet synchronous motor in this invention.

[0033] Fig. 6 This is a diagram showing the voltage fluctuation at the midpoint of the three-level inverter in this invention. Detailed Implementation

[0034] To make the technical solutions and advantages of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings:

[0035] Model predictive control algorithms rely on vector optimization using a predictive model and a value function. By inputting each voltage vector into the predictive model and selecting the optimal voltage vector through the value function, the motor is controlled. Since there is a midpoint current when the load is connected to the midpoint of a three-level inverter, the midpoint potential inevitably shifts during operation, affecting the accuracy of control. Traditional model predictive methods add a midpoint potential balancing term to the value function to suppress the fluctuation of the midpoint potential of the three-level inverter. However, since its weighting coefficient is not easy to tune, and the addition of the midpoint potential balancing term inevitably has a certain impact on the accurate control of the current.

[0036] like Figs. 1-2 As shown, this invention proposes a PMSM three-vector model predictive current control method for three-level inverters without weighting coefficients. It uses a virtual vector method to balance the midpoint voltage. By changing the action time of the small vector and its redundant small vector, the amount of charge flowing into and out of the midpoint is controlled, and the midpoint potential is changed to maintain balance. This eliminates the weighting coefficients, reduces the amount of calculation, and improves the accuracy of current control.

[0037] Within a control cycle, two current predictions are performed. The first prediction selects the voltage vector that minimizes the value function as the first optimal voltage vector. Using the current deadbeat principle, the selected first and second voltage vectors, along with 27 candidate voltage vectors and a zero vector, are time-allocated. The duration of the three vectors is calculated for different vectors used as the second voltage vector, resulting in 27 different combinations. These 27 composite voltage vectors are then substituted into the value function to select the composite vector that minimizes the value function, and the second voltage vector used by it is selected as the second optimal voltage vector. A virtual vector method is used to process small vectors. By acquiring the three-phase current of the motor, the duration of the small vector and its corresponding redundant small vectors is calculated to suppress fluctuations in the neutral point voltage. The specific steps are as follows:

[0038] S1: During motor operation, collect its rotor position information θ and motor speed n, and give the given speed n. * The difference between the actual speed fed back by the motor and the actual speed is used by a PI controller to obtain the q-axis current of the motor. Combination The control strategy completes the motor speed loop control and collects the three-phase current i of the motor. a i b i c Three-phase voltage u a u b u c And the inverter midpoint voltage, used for current loop three-vector model predictive current control.

[0039] The prediction model for the dq-axis stator current of a permanent magnet synchronous motor is as follows:

[0040]

[0041] Where i d (k),i q (k) represents the stator current at the current moment, i d (k+1),i q (k+1) represents the predicted stator current, u d (k),u q (k) The current time is the dq axis stator voltage, R s ω is the stator winding resistance, L is the stator winding self-inductance, and ω is the stator winding resistance. e Let ψ be the electric angular velocity of the motor rotor. f For permanent magnet flux linkage in motors.

[0042] The first voltage vector is selected using the value function. Then, the 27 voltage vectors are input into the stator current prediction model, and the vectors selected are those that maximize the value function. The smallest voltage vector is taken as the first voltage vector.

[0043] S2: After selecting the first voltage vector, use the following formula to calculate the slope of the current change along the dq axis of the motor under the action of each voltage vector.

[0044]

[0045] Where f d0 f q0 f d_opt1 f q_opt1 f d_i f q_i These are the d-axis and q-axis current slopes generated when the zero vector, the first voltage vector, and the alternative vector of the second vector are applied, respectively.

[0046] Using the deadbeat principle of current, the time allocation of the 27 candidate voltage vectors for the selected first and second voltage vectors and the zero vector is performed, and the action time t of the three vectors when different voltage vectors are used as the second voltage vector is calculated. opt1 ,t opt2 From t0, a total of 27 composite voltage vectors composed of three vectors can be obtained. The deadbeat formula for current is:

[0047]

[0048] Where i d (k+1) and i q (k+1) represent the predicted current values ​​along the d and q axes, respectively, i d * and i q * These are the current setpoints for the dq axes, respectively.

[0049] Substituting the 27 synthesized voltage vectors obtained above into the current prediction model, the second voltage vector used to obtain the synthesized vector that minimizes the value function is taken as the optimal voltage vector obtained in the second round of prediction.

[0050] The obtained three voltage vectors and their corresponding durations are output to the inverter to control the motor operation.

[0051] S3: If the selected vector is a small vector, since the small vector and its corresponding redundant small vector generate currents at the midpoint that are equal in magnitude but opposite in direction, the following formula can be used to allocate time between it and its corresponding redundant small vector, thereby reducing the potential fluctuation at the midpoint of the three-level inverter. The principle of time allocation is as follows:

[0052] When the first voltage vector is a small vector, its corresponding redundant small vector is looked up according to Table 1, and time allocation is performed on it, as follows:

[0053] From the capacitance formula Q = CU and the charge formula Q = it, we can deduce that within one period:

[0054]

[0055] Seeking

[0056]

[0057] U o U is the voltage at the midpoint of the inverter. dc Where C is the DC bus voltage, C is the DC side capacitor value, and t is the DC bus voltage. opt1 To calculate the time that the first vector should act, t1 is the actual time that the small vector acts, and t2 is the actual time that its corresponding redundant small vector acts.

[0058] Similarly, when the second voltage vector is a small vector, the time allocation is as follows:

[0059]

[0060] The redundant small vectors corresponding to each small vector are shown in Table 1:

[0061] Table 1. Small vectors and their corresponding redundant small vectors.

[0062]

[0063] Fig. 3 , Fig. 4 , Fig. 5 and Fig. 6 These correspond to the motor's speed diagram, torque diagram, three-phase current waveform diagram, and three-level inverter midpoint voltage fluctuation diagram during operation. A high-power motor was used for simulation verification. The DC bus voltage was set to 1800V, the initial given speed was 120 r / min, and the given load was 97600 N·m. The speed was increased to 200 r / min in 0.5 seconds, and the load was adjusted to 195200 N·m in 1 second. The simulation results show that the motor operates well under different speeds and loads, and the three-level inverter midpoint voltage fluctuation is less than 2%.

[0064] Simulation results show that the three-vector model predictive current control method without weighted coefficients can enable the three-phase permanent magnet synchronous motor to have good current sinusoidality and small torque ripple, while ensuring that the midpoint voltage deviation of the three-level inverter is less than 2%.

[0065] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A PMSM three-vector model predictive current control method based on a three-level inverter without weighted coefficients, characterized in that... include: Collect rotor position information during motor operation and actual speed Based on a given rotational speed The actual speed fed back by the motor The difference is used to obtain the q-axis current of the motor through a PI controller. , combined =0 control strategy is used to control the motor speed loop, and the three-phase current of the motor is collected. i a , i b , i c and three-phase voltage u a , u b , u c and inverter midpoint voltage U o , used for predictive current control using a three-vector model of the current loop; Two current predictions are performed within a control cycle. The first prediction selects the voltage vector that minimizes the value function as the first optimal voltage vector. The action time of the candidate voltage and the zero vector of the first and second voltage vectors is calculated using the current deadbeat principle to obtain multiple composite vectors. These composite vectors are then sequentially substituted into the current prediction model to select the composite vector that minimizes the value function. The second voltage vector used by this composite vector is then selected as the second optimal voltage vector. If the selected vector is a small vector, the virtual vector method is used to allocate time between the small vector and its corresponding redundant small vector, thereby controlling the midpoint potential balance; The switching states corresponding to the first optimal voltage vector, the second optimal voltage vector, and the zero vector are output to the inverter to control the motor operation; The value function for both voltage vector selections is: in i d ( k +1) i q ( k +1) are the predicted values ​​of the d-axis and q-axis currents, respectively. i q Set the q-axis current value; After the first voltage vector is selected, the time allocation of the selected first voltage vector, the 27 candidate voltage vectors for the second voltage vector, and the zero vector is performed using the current deadbeat principle. The effective time of the three vectors when different vectors are used as the second voltage vector is calculated. The three vectors are then combined to obtain a total of 27 composite voltage vectors. The current deadbeat formula is as follows: in i d ( k +1) and i q ( k +1) represent the predicted current values ​​for the d and q axes, respectively. i d and i q These are the q-axis current setpoints. , , , , , These are the d-axis and q-axis current slopes generated when the zero vector, the first voltage vector, and the alternative vector of the second vector are applied, respectively. Substitute the 27 synthesized voltage vectors obtained above into the current prediction model in the following formula to obtain the synthesized vector that minimizes the value function. The second voltage vector used in this formula is taken as the best voltage vector obtained in the second round of prediction. in i d ( k ), i q ( k ( ) represents the stator current at the current moment. i d ( k +1) i q ( k +1) is the predicted value of the stator current. u d ( k ), u q ( k The dq-axis stator voltages at the current time are respectively. R s For stator winding resistance, L For the stator winding self-inductance The electric angular velocity of the motor rotor. For permanent magnet flux linkage in motors; When the selected vector is a small vector, the following method is used to allocate the time between the small vector and its corresponding redundant small vector to reduce the midpoint potential fluctuation of the three-level inverter. The midpoint potential is balanced by utilizing the characteristic that the small vector and its redundant small vector of a three-level inverter generate currents of equal magnitude but opposite sign at the midpoint. When the first voltage vector is a small vector, the time allocation is as follows: Within one period, according to the capacitance formula and the charge formula get: Seeking in U o This is the voltage at the midpoint of the inverter. U dc Where C is the DC bus voltage, and C is the capacitance value of the DC side capacitor. t opt1 To calculate the time it should take for the first vector to act, t 1 represents the actual duration of action of this small vector. t 2 represents the actual duration of action of the corresponding redundant small vector. i 1 represents the midpoint current generated by this small vector. i 2 represents the current generated at the midpoint of the redundant small vector corresponding to this small vector; Similarly, when the second voltage vector is a small vector, the time allocation is as follows: 。

Images