A model predictive control method for AC permanent magnet synchronous motor without weight coefficient

Abstract

The present invention discloses a model predictive control method for an AC permanent magnet synchronous motor without weight coefficient, comprising the following steps: collecting the motor rotor operating parameters at time k, obtaining the electromagnetic torque reference value and the stator flux reference value, and obtaining the electromagnetic torque T' at time k. e (k) and stator flux ψ' s (k) and torque angle δ; obtain three sets of voltage vectors from the 2-D lookup table; recalculate the electromagnetic torque T at time k e (k) and stator flux ψ s (k); Load the improved ESO observer to obtain the electromagnetic torque T at time k+2 e (k+2), stator flux ψ s (k+2), predicting the torque angle δ(k+2) at time k+2; and reconstructing the optimal cost function for predictive control of AC permanent magnet synchronous motors. This technical solution effectively reduces torque, flux pulsation, and harmonic components of current. The prediction process does not involve adjusting weight coefficients, significantly simplifying the system's computational complexity.

Description

Technical Field

[0001] The present invention relates to the field of motor control, and in particular to a model predictive control method for an AC permanent magnet synchronous motor without a weight coefficient. Background Art

[0002] Permanent magnet synchronous motors (PMSMs) offer advantages such as low energy consumption and stable operation, and are considered a green alternative to gasoline engines. Motor control is a core control technology for electric drive systems, making the research of high-performance motor control algorithms crucial. With the rapid advancement of industrial technology, finite control set model predictive control (FCS-MPC), which easily implements multi-objective control and multiple constraints, has gradually entered the motor control field. There are two mainstream control methods for AC permanent magnet synchronous motors: field-of-control (FOC) and direct torque control (DTC). FOC decouples flux from torque through a series of coordinate transformations, then uses a PI controller to independently control the speed and current components to achieve the desired control objectives. However, its model is relatively simple, with few adjustable parameters, making it difficult to control and adjust. DTC is a high-performance AC variable frequency control technology that does not rely on complex coordinate transformations or pulse width modulation. Instead, it directly controls flux and torque through a hysteresis controller, selecting a voltage vector based on the error value to control motor operation. This method is simple, practical and easy to implement, but it is limited by the hysteresis band width and sampling frequency, and has problems such as large torque pulsation and excessive switching frequency, which restricts its development in high-performance scenarios.

[0003] In recent years, driven by the rapid development of digital technology and hardware, model predictive control (MPC) has gained considerable attention in the motor drive field due to its advantages, including high dynamic performance, simple and easy-to-implement algorithms, and flexible cost function design for multivariable tracking control. Combining the advantages of MPC and DTC, model predictive torque control (MPTC) directly tracks torque and flux, omitting the hysteresis controller and replacing the switching table with rolling optimization. This method offers excellent dynamic response, smaller torque ripple compared to DTC, and more precise control. However, traditional single-vector model predictive torque control (MPC-DTC) suffers from the following drawbacks: First, traditional MPTC requires the cost function to perform a traversal optimization of all eight feasible voltage vectors (six base vectors and two zero vectors) within each sampling period. This heavy computational burden limits its application in AC permanent magnet synchronous motors. Second, due to its multi-objective control advantages, traditional MPTC includes both electromagnetic torque and stator flux error terms in its constructed cost function. Due to the different dimensions of the two controlled variables, weighting coefficients are required. Typically, selecting appropriate weighting coefficients requires extensive simulation and experimental exploration. The weight coefficient also has a specific range of values; exceeding or falling below this range will result in poor control performance. Therefore, selecting the optimal weight coefficient to achieve optimal motor control performance requires considerable effort. While some improvements exist to address the high computational complexity and difficulty in adjusting the weight coefficient in MPTC, these approaches generally focus on reducing the computational load, which in turn reduces the number of selectable voltage vectors and impacts system control accuracy. Alternatively, they utilize intelligent algorithms to eliminate the weight factor, improving voltage prediction accuracy while also increasing the system's computational burden. There is a lack of torque ripple suppression methods that can simultaneously address both computational reduction and weight factor adjustment. Summary of the Invention

[0004] To solve the above problems, the present application provides a model predictive control method for an AC permanent magnet synchronous motor without weight coefficients, comprising the following steps:

[0005] Collect the motor rotor operating parameters at time k to obtain the electromagnetic torque reference value and stator flux reference value Operating parameters include: electrical angular velocity ω, rotor position θ, three-phase motor current i a (k), i b (k), i c (k), DC bus voltage u dc (k), d, q axis components i d (k), i q (k), α, β axis components i α (k), i β (k);

[0006] Get the electromagnetic torque T' at time k e (k) and stator flux ψ' s (k) and torque angle δ;

[0007] According to the electromagnetic torque T' e (k) and stator flux ψ' s (k) The error values ​​of three sets of voltage vectors are obtained from the 2-D lookup table;

[0008] According to the three sets of voltage vectors, three sets of corresponding operating parameters are obtained and the electromagnetic torque T at time k is recalculated. e (k) and stator flux ψ s (k);

[0009] Load the improved ESO observer to obtain the electromagnetic torque T at time k+2 e (k+2), stator flux ψ s (k+2), predict the torque angle δ(k+2) at time k+2;

[0010] Reconstruct the optimal cost function, which is used to predict the torque angle δ(k+2) at time k+2 based on the torque angle δ at time k;

[0011] Optimal cost function for predictive control of AC permanent magnet synchronous motors.

[0012] Among them, the electromagnetic torque T' at time k e (k) and stator flux ψ' s (k) through the d and q axis components i d (k), i q (k) is obtained and the calculation method is:

[0013]

[0014] Among them, i α 、i β , ψ α , ψ β are the current and flux of the α and β axes of the two-phase stationary coordinate system, ψ s is the stator flux, ψ f is the rotor flux;

[0015] and:

[0016]

[0017] Where θ is the rotor angular position, u α ,u β are the voltages of the α and β axes of the two-phase stationary coordinate system respectively.

[0018] Furthermore, the calculation method of the torque angle is:

[0019] The contents of the 2-D lookup table include the stator flux ψ' at time k s (k) Error with reference stator flux, electromagnetic torque T' e (k) Correspondence with the reference electromagnetic torque error and space voltage vector Un.

[0020] The contents of the 2-D lookup table are determined by dividing the space voltage vector sector table; the space voltage vector sector table includes 12 sectors, which are labeled S1 to S12, Sn is one of the sectors, and 0<n<13; the Sn sector corresponds to the space voltage vector Un; when the stator flux is in the Sn sector, the effective voltage vectors Un, Um and the zero voltage vector are synthesized to achieve stator flux and electromagnetic torque error adjustment, where m=n+1, and m=1 when n=12.

[0021] Recalculate the electromagnetic torque T at time k e (k) and stator flux ψ s (k), using the torque formula, according to the d and q axis components i d (k), i q (k) Calculation implementation: the calculation method is:

[0022]

[0023]

[0024]

[0025] Among them, ψ f is the permanent magnet rotor flux value, L d 、L q is the d-axis inductance and q-axis inductance, L in the surface-mounted permanent magnet synchronous motor d =L q .

[0026] Furthermore, the improved ESO observer is used to realize current error observation and two-step delay compensation; before loading the improved ESO observer, the voltage relationship of the improved ESO observer is defined as:

[0027]

[0028] and:

[0029]

[0030] The method for predicting the torque angle δ(k+2) at time k+2 is:

[0031]

[0032] Furthermore, the optimal cost function does not include the weight coefficient and is expressed as:

[0033] g=[δ ref -δ(k+2)] 2 +I m ,

[0034] Among them, I m is the current limit value, and: Among them, I max is the maximum peak current, i s is the stator current.

[0035] In this invention, the voltage vector sector diagram is divided into 12 sectors, alternative voltage vectors are added, and a novel 2-D lookup table (LUT) is designed. Based solely on the torque error and stator flux error, two valid voltage vectors and a zero vector can be directly selected within a single sampling cycle using the novel switching table. The two valid voltage vectors are not limited to adjacent voltage vectors. Furthermore, the invention proposes a cost function that eliminates weight coefficients. The cost function only contains the torque angle, eliminating the need for weight coefficients and reducing the complexity of the cost function calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a step diagram of a model predictive control method without weight coefficients provided in an embodiment of the present invention;

[0037] Figure 2 2. It is a block diagram of a non-weighted MPTC control according to an embodiment of the present invention;

[0038] Figure 3 is a spatial voltage vector sector division diagram provided according to an embodiment of the present invention;

[0039] Figure 4 This is a schematic diagram of the simulation results of traditional MPTC implementation;

[0040] Figure 5 2 is a schematic diagram of the effect of the model predictive control simulation results provided according to an embodiment of the present invention. DETAILED DESCRIPTION

[0041] The present invention provides an MPTC strategy for AC permanent magnet synchronous motors without weight coefficients. To quickly screen voltage vectors, reduce computational complexity, and increase the concept of candidate voltage vectors, the voltage vector sector diagram is divided into 12 sectors. Based on this, a novel 2-D lookup table (LUT) is designed that can be used to directly select two valid voltage vectors and a zero vector to output the corresponding motor operating state based on torque error and stator flux error within a single cycle. The two valid voltage vectors are not limited to adjacent voltage vectors. The present invention also proposes a prediction model that eliminates weight coefficients, in which the value function only contains the torque angle, thereby eliminating the need for weight coefficients in the value function and reducing the complexity of the value function calculation. The specific implementation of the present invention is described in detail below in conjunction with the accompanying drawings.

[0042] Figure 1 A weight-free model predictive control method for an AC permanent magnet synchronous motor is provided, comprising:

[0043] Step S100: Collect the motor rotor operating parameters at time k and obtain the electromagnetic torque reference value and stator flux reference value The operating parameters include: electrical angular velocity ω, rotor position θ, three-phase motor current i a (k), i b (k), i c (k), DC bus voltage u dc (k), d, q axis components i d (k), i q (k), α, β axis components i α (k), i β (k);

[0044] For obtaining the operating parameters, Figure 2 As shown, the motor rotor electrical angular velocity ω, rotor position θ, and three-phase motor current i at time k are collected through the encoder. a (k),i b (k),i c (k) and DC bus voltage u dc (k), and use Clark transformation to solve the d and q axis components i d (k), i q (k), according to the d and q axis components i d (k), i q (k), using Park transform to get α and β axis components i α (k), i β (k).

[0045] On the other hand, the speed error Δω is adjusted by the PI regulator. e Calculate the electromagnetic torque reference value Obtaining stator flux through MTPA

[0046] Step S110: Obtaining the electromagnetic torque T' at time k e (k) and stator flux ψ' s (k) and torque angle δ;

[0047] The electromagnetic torque T' at time k e (k) and stator flux ψ' s (k) through the d and q axis components i d (k), i q (k) is obtained and the calculation method is:

[0048]

[0049]

[0050] Among them, i α 、i β , ψ α , ψ β are the current and flux of the α and β axes of the two-phase stationary coordinate system, ψ s is the stator flux, ψ f is the rotor flux;

[0051] and:

[0052]

[0053]

[0054]

[0055] Where θ is the rotor angular position, u α ,u β are the voltages of the α and β axes of the two-phase stationary coordinate system respectively.

[0056] The calculation method of the torque angle is:

[0057] Step S120: According to the electromagnetic torque T' e (k) and stator flux ψ' s (k) The error values ​​of three sets of voltage vectors are obtained from the 2-D lookup table;

[0058] The contents of the 2-D lookup table in this step include the stator flux ψ' at time k s (k) Error dψ from the reference stator flux s , electromagnetic torque T' e (k) and the reference electromagnetic torque error dT e, the correspondence between the space voltage vector Un.

[0059] In the present invention, a new spatial voltage vector sector division diagram is established based on the state variables of the control system in the current sampling period, namely the stator flux, electromagnetic torque, torque angle, and rotor position. Then, the directions in which the stator flux and electromagnetic torque need to change are obtained based on the electromagnetic torque and stator flux error values, and a 2-D lookup table (LUT) is determined based on this, as shown in the following table:

[0060]

[0061] In the table, U1 to U6 are non-zero voltage vectors.

[0062] The 2-D lookup table (LUT) provided by this invention allows two valid voltage vectors and one zero vector to be directly selected from the table within a single sampling cycle, based solely on the torque error and stator flux error. These two valid voltage vectors are not limited to adjacent voltage vectors. This allows only three voltage vectors (two base vectors and one zero vector) to be substituted into the cost function for traversal optimization, significantly reducing the subsequent computational effort. Furthermore, these three vectors are determined by the torque and flux errors and can therefore be used for subsequent analysis of torque and flux pulsation levels.

[0063] The space voltage vector sector division diagram established in the present invention is as follows: Figure 3 As shown in the figure, in order to quickly screen the voltage vector and reduce the amount of calculation, alternative voltage vectors are added, and the voltage vector sector diagram is divided into 12 sectors, which are marked as S1 to S12, Sn is one of the sectors, 0<n<13; the Sn sector corresponds to the space voltage vector Un; when the stator flux is in the Sn sector, the effective voltage vectors Un, Um (where m=n+1, m=1 when n=12) are synthesized with the zero voltage vector to achieve stator flux and electromagnetic torque error adjustment.

[0064] Assume that the stator flux ψ s In such Figure 3 The estimated stator flux and electromagnetic torque are compared with the reference stator flux and electromagnetic torque obtained by MPTA and PI controller, and two error values ​​are obtained, which are defined as dψ s and dT eWhen the error is greater than 0, it is defined as (+), and vice versa. For the stator flux in sector S1, if torque and flux need to be increased, two effective voltage vectors (U2 and U3) and a zero voltage vector (U0|U7) are typically selected for synthesis. However, due to the limited number of voltage vectors available in each sector, this leads to significant electromagnetic torque and flux pulsation. Therefore, a new spatial voltage vector sector division diagram needs to be redesigned, dividing it into 12 sectors. Assuming the stator flux is in sector S1, if torque and flux need to be increased, the two effective voltage vectors (U1 and U2) can be applied to the lower half of sector S1 (i.e., optimized S1), and the two effective voltage vectors (U2 and U3) can be applied to the upper half of sector S1 (i.e., optimized S2). This increases the number of alternative voltage vectors for each sector. Multiple voltage vector synthesis methods can be added, provided the inverter switching frequency is not included in the cost function. Furthermore, to reduce the computational burden, only the zero vector (U0) is selected for synthesis.

[0065] Step S121: According to the three sets of voltage vectors obtained in step S120, three sets of corresponding operating parameters are obtained, and the electromagnetic torque T at time k is recalculated using the three sets of operating parameters. e (k) and stator flux ψ s (k);

[0066] Recalculate the electromagnetic torque T at time k e (k) and stator flux ψ s (k) According to the d and q axis components i d (k), i q (k) is calculated using the torque formula, and the calculation method is:

[0067]

[0068]

[0069]

[0070] Among them, ψ f is the permanent magnet rotor flux value, L d , L q is the d-axis inductance and q-axis inductance, L in the surface-mounted permanent magnet synchronous motor d =L q .

[0071] Step S131: Load the improved ESO observer to obtain the electromagnetic torque T at time k+2 e (k+2), stator flux ψ s (k+2);

[0072] When parameter mismatch is considered, an improved ESO observer is designed to observe and compensate for current errors caused by external parameter disturbances in real time. At the same time, a two-step delay compensation is performed considering the existence of computer instruction cycles.

[0073] Before loading the improved ESO observer, the voltage relationship of the improved ESO observer is defined as:

[0074]

[0075] and:

[0076]

[0077] On this basis, the prediction of the torque angle δ(k+2) at time k+2 is realized, that is:

[0078]

[0079] Step S140: reconstructing an optimal cost function, wherein the cost function is used to predict the torque angle δ at time k+2 based on the torque angle δ at time k;

[0080] The newly constructed value function in this step only contains the torque angle and does not include the weight coefficient, which eliminates the complicated tuning process of the weight coefficient and is expressed as:

[0081] g=[δ ref -δ(k+2)] 2 +I m (14),

[0082] Among them, I m is the current limit value, and:

[0083] Among them, I max is the maximum peak current, i s is the stator current.

[0084] Compared with the traditional single-vector MPTC, the strategy of eliminating weight coefficients proposed in this step expands the range of selectable voltage vectors and reduces current harmonics as well as torque and flux pulsations. The optimal value function is finally reconstructed for predictive control of AC permanent magnet synchronous motors. The overall control process is as follows: Figure 2 shown.

[0085] The predictive control method provided by the present invention can effectively reduce torque and flux ripple, as well as the harmonic components of the current. The present invention provides the following simulation experiment: the motor operates stably under certain operating conditions, and steady-state operation is analyzed. The total harmonic distortion (THD) is used to measure the degree of current distortion. In addition, the ripple root mean square error (RMSE) is used to measure the ripple level of its torque and flux ripple. The evaluation index is calculated as follows:

[0086]

[0087]

[0088]

[0089] Where n is the number of samples in each sampling cycle.

[0090] The simulation results using traditional MPTC are as follows: Figure 4 As shown, the simulation results of the control method provided by the present invention are as follows Figure 5 The comparison of evaluation indicators is shown in the following table:

[0091] Traditional MPTC Improved MPTC <![CDATA[THD eq ]]> 21.67％ 12.84％ <![CDATA[T RMSE ]]> 110.03％ 10.75％ <![CDATA[ψ RMSE ]]> 7.97％ 4.24％

[0092] Experimental results show that the model predictive control method without weighting coefficients proposed in this invention achieves significantly lower THD and RMSE values ​​than traditional MPTC, demonstrating that the proposed control method effectively reduces torque and flux pulsation, as well as current harmonics. Furthermore, since the prediction process only requires three voltage vectors, it does not require the adjustment of weighting coefficients, significantly simplifying the system's computational complexity.

[0093] The above disclosures are only a few specific embodiments of the present invention. However, the present invention is not limited thereto. Any changes that can be conceived by those skilled in the art should fall within the scope of protection of the present invention.

Claims

1. A model predictive control method for an AC permanent magnet synchronous motor without weight coefficients, characterized in that: The following steps are involved: Collect the motor rotor operating parameters at time k to obtain the electromagnetic torque reference value and stator flux reference value The operating parameters include: electrical angular velocity ω, rotor position θ, three-phase motor current , DC bus voltage , d, q axis components 、 , 、 Axis component 、 ; Get the electromagnetic torque at time k and stator flux and torque angle ; According to the electromagnetic torque and stator flux The error values ​​of the three voltage vectors are obtained from the 2-D lookup table; According to the three sets of voltage vectors, three sets of corresponding operating parameters are obtained, and the electromagnetic torque at time k is recalculated by the three sets of operating parameters. and stator flux ; Load the improved ESO observer to obtain the electromagnetic torque at time k+2 , stator flux , predict the torque angle at time k+2 (k+2); Reconstruct the optimal cost function, which is used to calculate the torque angle at time k Predict the torque angle at time k+2 (k+2); The optimal cost function is used for predictive control of AC permanent magnet synchronous motor; Among them, the electromagnetic torque at time k is and stator flux Through the d and q axis components 、 Get, the calculation method is: in, 、 、 Two-phase stationary coordinate system Shaft current, magnetic flux, is the stator flux, is the rotor flux; and: , in, is the rotor angular position, Two-phase stationary coordinate system Shaft voltage; The calculation method of the torque angle is: ; The recalculated electromagnetic torque at time k and stator flux When the torque formula is used, according to the d and q axis components 、 Calculation implementation, the calculation method is: in, is the permanent magnet rotor flux value, is the d-axis inductance value and the q-axis inductance value, in the surface-mounted permanent magnet synchronous motor ; The improved ESO observer is used to realize current error observation and two-step delay compensation; Before loading the improved ESO observer, the voltage relationship of the improved ESO observer is defined as: , and: , The predicted torque angle at time k+2 The (k+2) method is: The optimal cost function does not include weight coefficients and is expressed as: in, is the current limit value, and: in, is the maximum peak current, is the stator current.

2. The model predictive control method according to claim 1, characterized in that: The contents of the 2-D lookup table include the stator flux at time k Error from reference stator flux, electromagnetic torque The corresponding relationship with the reference electromagnetic torque error and space voltage vector Un.

3. The model predictive control method according to claim 2, characterized in that: The content of the 2-D lookup table is determined by partitioning the space voltage vector sector table; The space voltage vector sector table includes 12 sectors, which are respectively marked as S1 to S12, Sn is one of the sectors, 0<n<13; the Sn sector corresponds to the space voltage vector Un; when the stator flux is in the Sn sector, the stator flux and electromagnetic torque error are adjusted by synthesizing the effective voltage vectors Un, Um and the zero voltage vector, where m=n+1, and m=1 when n=12.

Images