Low-harmonic model predictive control method for five-phase motor based on double space vector synthesis

Abstract

The application discloses a five-phase motor low-harmonic model predictive control method based on double space vector synthesis, belongs to the technical field of multiphase motor control, and aims at solving the problem of serious phase current distortion when the motor has a third harmonic flux in the prior art.The method comprises the following steps: S1, in the fundamental wave space and the third harmonic space, ten virtual voltage vectors with equal amplitude, uniform distribution and mutual decoupling in the two spaces are generated by synthesizing large vectors and medium vectors with the same direction according to a preset ratio, and the control sets of the two spaces are formed; S2, in the two spaces, the optimal vector and the suboptimal vector are selected from the corresponding control sets according to the current reference value and the current sampling value, the action time of the two vectors is calculated through angle adjustment and amplitude adjustment, and the final synthetic voltage vector is obtained; and S3, signal generation: the final synthetic voltage vectors obtained in the two spaces are decomposed into the action time of the basic voltage vectors, the modulation waves of the two spaces are formed and superposed, and the switching signal is generated.

Description

Technical Field

[0001] This invention relates to the field of multiphase motor control technology, and in particular to a low-harmonic model predictive control method for five-phase motors based on dual-space vector synthesis. Background Technology

[0002] Five-phase permanent magnet synchronous motors are widely used in fields with extremely high system reliability requirements, such as aerospace, electric vehicles, and ship propulsion, due to their advantages of high power density, low torque ripple, strong fault tolerance, and multiple degrees of control freedom. Model predictive control, especially finite set model predictive control (FCS-MPC), has become a research hotspot in the field of multiphase motor drive control due to its advantages such as fast dynamic response, no need for PI parameter tuning, and ease of implementing multi-objective constraint control.

[0003] However, in practical engineering applications, due to limitations in rotor structure design and permanent magnet magnetization methods, third harmonic flux linkages inevitably exist in the air gap magnetic field of permanent magnet synchronous motors. When the motor operates at high speed, the rotating third harmonic flux linkages induce significant third harmonic back electromotive force (EMF) in the stator windings. Most existing FCS-MPC strategies, such as those based on virtual voltage vectors, passively "suppress" harmonic currents by synthesizing specific voltage vectors to make the voltage projection of the third harmonic space zero. However, this open-loop suppression method cannot effectively address the third harmonic currents generated by back EMF excitation.

[0004] When a third harmonic flux linkage exists in a motor, an uncontrollable third harmonic current flows through the stator windings, causing severe distortion of the phase current and an increase in total harmonic distortion (THD), which in turn increases motor losses and reduces system operating efficiency. To improve control accuracy, active control of the third harmonic space is required. However, directly increasing the switching frequency to achieve closed-loop control of the high-frequency third harmonic current is limited by controller bandwidth, switching losses, and processor computing power, making it difficult to implement in practical engineering.

[0005] Therefore, there is an urgent need for a control method that can effectively suppress the third harmonic current without significantly increasing the computational burden and switching losses, so as to improve the operating performance of a five-phase permanent magnet synchronous motor under the condition of harmonic flux linkage. Summary of the Invention

[0006] To address the issues of severe phase current distortion and high third harmonic current content in existing model predictive control techniques for five-phase permanent magnet synchronous motors when the motor has a third harmonic flux linkage, this invention provides a low-harmonic model predictive control method for five-phase motors based on dual-space vector synthesis.

[0007] The low-harmonic model predictive control method for a five-phase motor based on dual space vector synthesis described in this invention includes the following steps:

[0008] S1. Dual-space virtual voltage vector synthesis steps: In the fundamental wave space and the third harmonic wave space, the large vector and the medium vector in the same direction are synthesized according to a preset ratio to generate ten virtual voltage vectors with equal amplitude, uniform distribution and mutual decoupling in the two spaces, which constitute the virtual vector control set of the fundamental wave space and the virtual vector control set of the third harmonic wave space respectively.

[0009] S2. Screening and Action Time Calculation Steps: In two spaces, based on the given current reference value and the current sampling value at the current moment, the optimal vector and the second-best vector are screened from the corresponding virtual vector control set, and the action time of the two is calculated by angle adjustment and amplitude adjustment to calculate the final synthesized voltage vector.

[0010] S3. Signal generation steps: Decompose the final synthesized voltage vector obtained from the two spaces into the action time of the inverter's basic space voltage vector. Based on their respective action times, form the fundamental space modulation wave and the third harmonic space modulation wave respectively. Then, superimpose the modulation waves of the two spaces to generate a switching signal.

[0011] Preferably, the specific process of step S1 is as follows: the basic space voltage vector of the five-phase inverter is divided into large vector, medium vector and small vector according to the magnitude, in the fundamental space... In the coordinate system, the large and medium vectors in the same direction are synthesized with a weight ratio of 0.618:0.382 to obtain ten virtual voltage vectors in the fundamental space. These ten vectors have equal amplitudes, uniform phase angles, and adjacent intervals of 72° in the fundamental space, and their projections in the third harmonic space are zero. In the coordinate system, the large vector and the medium vector in the same direction are combined with a weight ratio of 0.618:0.382 to obtain ten virtual voltage vectors in the third harmonic space. These ten vectors have equal amplitudes, uniform phase angles, and adjacent intervals of 72° in the third harmonic space, and their projection in the fundamental space is zero.

[0012] Preferably, the specific process of selecting the optimal and suboptimal vectors in step S2 is as follows: First, from five non-adjacent virtual voltage vectors, a preliminary voltage vector is selected by minimizing the voltage cost function; then, from two virtual voltage vectors adjacent to the preliminary voltage vector, a suboptimal voltage vector is selected by minimizing the same voltage cost function; the vector with the lagging phase angle in the preliminary and suboptimal voltage vectors is named the optimal vector, and the vector with the leading phase angle is named the suboptimal vector; wherein, the selection process in the fundamental space and the third harmonic space is the same.

[0013] Preferably, the voltage cost function in the fundamental space is constructed based on the sum of squares of the differences between the reference voltage vector and the voltage vector at the current moment, wherein the reference voltage vector is calculated from the current reference value, the current current sample value, the motor parameters, and the rotor electrical angle; the voltage cost function in the third harmonic space is constructed using an open-loop given method, wherein the reference voltage vector is calculated from the rotor angular velocity and the amplitude of the third harmonic flux linkage of the permanent magnet, and the direct-axis reference voltage is set to zero.

[0014] Preferably, the angle adjustment and amplitude adjustment in step S2 are specifically as follows: by introducing the angle adjustment duty cycle to adjust the ratio of the action time of the optimal vector and the suboptimal vector, the direction of the final synthesized voltage vector is consistent with the direction of the reference voltage vector; then by introducing the amplitude adjustment duty cycle to adjust the amplitude of the synthesized vector, the amplitude of the final synthesized voltage vector is consistent with the amplitude of the reference voltage vector, and a zero vector is added to participate in the synthesis.

[0015] Preferably, the action time of the decomposition into the basic space voltage vector of the inverter in step S3 is specifically as follows: based on the angle adjustment duty cycle, amplitude adjustment duty cycle, control cycle, and the preset synthesis ratio of the large vector and the medium vector, the action time of the large vector in the same direction as the optimal vector, the medium vector in the same direction as the optimal vector, the large vector in the same direction as the second-best vector, the medium vector in the same direction as the second-best vector, and the two zero vectors are calculated respectively.

[0016] Preferably, the screening and action time calculation steps in the fundamental frequency space and the third harmonic frequency space are executed independently and in parallel.

[0017] The beneficial effects of this invention are:

[0018] 1. The third harmonic flux linkage of a five-phase permanent magnet synchronous motor induces a third harmonic back electromotive force (EMF) in the windings. Traditional methods synthesize a virtual voltage vector only in the fundamental space, which avoids generating additional voltage in the third harmonic space, but cannot cancel the third harmonic current caused by the back EMF itself. This invention constructs an independent control loop in parallel in the third harmonic space. By actively synthesizing a voltage component with the same amplitude but opposite direction to the back EMF, voltage compensation is achieved in the third harmonic space, thereby directly suppressing the third harmonic current generated by flux linkage excitation and reducing the degree of phase current distortion.

[0019] 2. The frequency of the third harmonic current is three times that of the fundamental frequency. If closed-loop tracking control is used, theoretically, the switching frequency needs to be increased several times to meet the sampling and control bandwidth requirements, leading to a significant increase in switching losses. This invention employs an open-loop voltage input strategy in the third harmonic space, directly calculating the required compensation voltage based on the rotor angular velocity and the amplitude of the third harmonic flux linkage, eliminating the need for high-frequency current sampling and feedback adjustment. Therefore, the third harmonic control loop and the fundamental control loop operate at the same low-speed sampling frequency, avoiding the problems of increased switching frequency and increased processor computational burden.

[0020] 3. The presence of third harmonic current generates torque ripple six times the fundamental frequency and increases copper and iron losses. This invention applies a compensation voltage to the third harmonic space, bringing the net voltage of the third harmonic space close to zero, thereby significantly reducing the current component in this subspace and decreasing harmonic losses and torque fluctuations. Simultaneously, the fundamental space and the third harmonic space are decoupled through a virtual voltage vector, preventing mutual interference between the two spaces. Fundamental current tracking is unaffected by harmonic current changes, thus maintaining an independent and rapid fundamental current response during sudden load or speed changes.

[0021] 4. This invention designs the control loop of the third harmonic space as a structure completely parallel to that of the fundamental space. Both spaces employ the same virtual voltage vector synthesis method, two-step screening process, and angle and amplitude adjustment algorithms. This means that based on the existing fundamental space continuous modulation FCS-MPC code, only a few parameters need to be copied and modified (such as replacing the fundamental inductance and flux linkage with the corresponding values ​​of the third harmonic, and adjusting the Park transform angle from θ to 3θ) to achieve active control of the third harmonic space, reducing the complexity of algorithm porting and engineering deployment. Attached Figure Description

[0022] Figure 1 This is a typical topology of a five-phase permanent magnet synchronous motor driven by a five-phase two-level voltage source inverter as described in this invention;

[0023] Figure 2 This is a system block diagram of the five-phase motor low harmonic model predictive control method based on dual space vector synthesis described in this invention;

[0024] Figure 3 This is a distribution diagram of the 32 basic space voltage vectors of a five-phase inverter in the fundamental and third harmonic spaces; among them... Figure 3 (a) represents the fundamental voltage vector in the fundamental space. Distribution map in Figure 3 (b) represents the fundamental space voltage vector in the third harmonic space. Distribution map in;

[0025] Figure 4 This is a schematic diagram showing the distribution of the ten virtual voltage vectors synthesized in this invention in the fundamental frequency space and the third harmonic frequency space; wherein, Figure 4 (a) Ten virtual voltage vectors in the fundamental frequency space Distribution map in Figure 4 (b) Ten virtual voltage vectors in the third harmonic space Distribution map in;

[0026] Figure 5A schematic diagram illustrating the selection from five non-adjacent virtual voltage vectors during the initial screening of the optimal vector;

[0027] Figure 6 A schematic diagram of two virtual voltage vectors adjacent to the initially selected vector when further filtering for suboptimal vectors;

[0028] Figure 7 This is a vector geometric relationship analysis diagram used to calculate the duty cycle of angle adjustment;

[0029] Figure 8 This is a schematic diagram of the dual-space vector synthesis method for generating PWM waveform superposition according to the present invention;

[0030] Figure 9 The diagram shows the phase current waveform and its total harmonic distortion (THD) spectrum when controlled using traditional methods; among them, Figure 9 (a) is a waveform diagram of the phase current. Figure 9 (b) is the phase current spectrum diagram;

[0031] Figure 10 The diagram shows the phase current waveform and its total harmonic distortion (THD) spectrum when controlled by the method described in this invention; wherein, Figure 10 (a) is a waveform diagram of the phase current. Figure 10 (b) is the phase current spectrum diagram. Detailed Implementation

[0032] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Although some embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the present invention. It should be understood that the accompanying drawings and embodiments of the present invention are for illustrative purposes only and are not intended to limit the scope of protection of the present invention.

[0033] It should be understood that the various steps described in the method embodiments of the present invention may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present invention is not limited in this respect.

[0034] The term "comprising" and its variations as used herein are open-ended, meaning "including but not limited to"; the term "based on" means "at least partially based on"; and the term "one embodiment" means "at least one embodiment". Definitions of other terms will be given in the following description. It should be noted that the concepts of "first," "second," etc., mentioned in this invention are used only to distinguish different devices, modules, or units, and are not intended to limit the order of functions performed by these devices, modules, or units or their interdependencies.

[0035] It should be noted that the terms "a" and "a plurality of" used in this invention are illustrative rather than restrictive. Those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0036] The names of the messages or information exchanged between the multiple devices in the embodiments of the present invention are for illustrative purposes only and are not intended to limit the scope of these messages or information.

[0037] In the field of motor control technology, model predictive control strategies for five-phase permanent magnet synchronous motors, especially finite set model predictive control (FCS-MPC), are widely used due to their advantages such as fast dynamic response and no need for PI parameter tuning. However, in actual motors, due to limitations in rotor structure and permanent magnet magnetization methods, third harmonic flux linkages inevitably exist in the air gap magnetic field. When the motor runs at high speed, the rotating third harmonic flux linkages induce significant third harmonic back electromotive force in the stator windings. Existing FCS-MPC strategies, such as those based on virtual voltage vectors, passively suppress harmonic currents by synthesizing specific voltage vectors to make the voltage projection of the third harmonic space zero. However, in-depth analysis reveals that this open-loop suppression method can only avoid introducing additional voltage components into the third harmonic space, but cannot actively control the third harmonic currents generated by back electromotive force excitation. From the perspective of electromagnetic induction, the back electromotive force generated by the rotating third harmonic flux in the winding is proportional to the rotational speed. When the motor runs at high speed, the amplitude of this back electromotive force increases significantly, exciting an uncontrollable third harmonic current in the stator winding. This harmonic current superimposed on the fundamental current leads to severe distortion of the stator phase current, increases total harmonic distortion (THD), and consequently increases the motor's copper and iron losses, generating torque pulsation six times the fundamental frequency and reducing system operating efficiency.

[0038] From a control theory perspective, the frequency of the third harmonic current is three times the fundamental frequency. If a closed-loop control strategy is used to actively suppress it, theoretically, the switching frequency needs to be increased to several times the fundamental frequency to meet the sampling and control bandwidth requirements. However, in practical engineering, the controller switching frequency is limited by power device losses and processor computing power, making it difficult to increase indefinitely. Directly increasing the switching frequency will lead to a sharp increase in switching losses, increased heat dissipation pressure, a heavier computational burden on the processor, a compressed control cycle, and may even cause system malfunction. Therefore, there is an urgent need for a control method that can effectively suppress the third harmonic current without significantly increasing the switching frequency and computational burden.

[0039] To address the aforementioned technical problems, this invention provides a low-harmonic model predictive control method for a five-phase motor based on dual-space vector synthesis.

[0040] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0041] The controlled object in this embodiment of the invention is as follows Figure 1 The system shown is a five-phase permanent magnet synchronous motor driven by a five-phase two-level voltage source inverter. Figure 1 In the middle, the DC bus voltage is Two voltage-dividing capacitors are connected in series between the positive and negative terminals of the DC bus. and The connection point o of the two capacitors is the midpoint of the DC bus; the five-phase inverter consists of five parallel bridge arms, each containing two switching transistors (phase A bridge arm is...). Phase B bridge arm is C-phase bridge arm is The D-phase bridge arm is E-phase bridge arm is These correspond to the five-phase windings A, B, C, D, and E, respectively. One end of each phase winding is connected to the midpoint of the corresponding bridge arm of the inverter, and the other end is connected to the neutral point N. The five-phase windings of the motor are symmetrically distributed in space, and the angle between the axes of adjacent phase windings is θ. (Electrical angle). The execution subject of the method of the present invention can be an embedded controller such as a digital signal processor (DSP) or a microcontroller unit (MCU).

[0042] To facilitate the description of the electrical quantities and control variables of the motor, this invention employs multiple coordinate system transformations. Physical quantities in the natural coordinate system (five-phase coordinate system A, B, C, D, E) are transformed to a stationary two-phase coordinate system using Clark transformation. Under a single-phase open-circuit fault, Clark transformation reduces the five-phase system to two orthogonal stationary subspaces: Fundamental space and Third harmonic space. Among them, The fundamental frequency space corresponds to electromechanical energy conversion, which is the main control object of this invention; The third harmonic space only generates losses and torque ripples; this invention uses virtual voltage vector design to make its projection zero. Furthermore, the Park transform is used to... Transformation of stationary coordinate system components in fundamental space to rotating coordinate system in fundamental space Below, among them The shaft and rotor magnetic flux direction are aligned. Axis ahead axis .exist In a rotating coordinate system, the inductance matrix in the motor equations is a constant matrix, which facilitates controller design. The current reference value and current sampling value in this invention are both in... Given and feedback in a coordinate system.

[0043] A low-harmonic model predictive control method for a five-phase motor based on dual-space vector synthesis is described, which includes the following steps:

[0044] S1. Dual-space virtual voltage vector synthesis steps: In the fundamental wave space and the third harmonic space, the large vector and the medium vector in the same direction are synthesized according to a preset ratio to generate ten virtual voltage vectors with equal amplitude, uniform distribution and mutual decoupling in the two spaces, which constitute the virtual vector control set of the fundamental wave space and the virtual vector control set of the third harmonic space, respectively.

[0045] Specifically, the core of this step lies in constructing two independent and non-interfering virtual voltage vector control sets, laying the foundation for achieving parallel control in both spaces. Analyzing the principle of vector space decoupling, the 32 basic space voltage vectors of a five-phase inverter exhibit different distribution characteristics in the fundamental and third harmonic spaces. The projection amplitude and direction of the same basic vector in the two spaces have a definite mapping relationship. If the basic space voltage vectors are directly used for control, the control in the fundamental space will inevitably generate interference components in the third harmonic space, and vice versa. This invention synthesizes large and medium vectors of the same direction at a specific ratio, utilizing the characteristic that their projection directions in the third harmonic space are opposite, so that the net projection of the synthesized vector in the third harmonic space is zero, thus achieving complete decoupling between the virtual vector in the fundamental space and the third harmonic space. Similarly, the projection of the virtual vector in the third harmonic space synthesized at the same ratio in the fundamental space is also zero. This decoupling mechanism eliminates mutual interference between the two spaces at the vector synthesis level, allowing for independent design of control strategies in the two spaces without mutual interference.

[0046] This step constructs a virtual voltage vector set offline, avoiding the burden of online vector synthesis calculations. The uniform distribution of the ten vectors (with adjacent vectors spaced 72° apart) ensures the control set's coverage uniformity across the entire plane, providing a sufficient set of candidate vectors for subsequent high-precision vector selection. When this control set is used for model predictive control, the selected virtual vectors do not generate additional voltage components in the third harmonic space, thus avoiding the introduction of new harmonic interference by the control itself.

[0047] S2. Screening and Action Time Calculation Steps: In two spaces, based on the given current reference value and the current sampling value at the current moment, the optimal vector and the second-best vector are screened from the corresponding virtual vector control set, and the action time of the two is calculated by angle adjustment and amplitude adjustment to calculate the final synthesized voltage vector.

[0048] Specifically, the core of this step lies in rapidly determining the optimal and second-best vectors through a two-step screening method, and then achieving precise control of the angle and amplitude of the synthesized vector through continuous modulation. From the perspective of balancing computational complexity and accuracy, traditional FCS-MPC requires traversing all candidate vectors to evaluate the cost function, which significantly increases the computational burden when the number of candidate vectors is large. The two-step screening method used in this invention first selects an initial voltage vector from five non-adjacent virtual vectors, and then selects a second-best voltage vector from its two adjacent vectors, reducing the number of evaluations and significantly reducing the computational burden without significantly sacrificing optimality.

[0049] Furthermore, traditional FCS-MPC can only output a finite number of discrete voltage vectors, resulting in significant voltage tracking errors and current ripple in steady state. This invention introduces angle-adjustable duty cycle and amplitude-adjustable duty cycle to perform time-weighted synthesis of the optimal and second-best vectors, allowing the final synthesized voltage vector to continuously vary within the sector spanned by the two candidate vectors. The angle-adjustable duty cycle controls the ratio of the two vectors' action time, ensuring the synthesized vector's direction is precisely aligned with the reference voltage vector's direction; the amplitude-adjustable duty cycle controls the magnitude of the synthesized vector's amplitude, while simultaneously introducing a zero vector to fill the remaining time. Through this dual-degree-of-freedom adjustment mechanism, the final synthesized voltage vector can achieve continuous modulation in both angle and amplitude dimensions, theoretically enabling error-free tracking of any reference voltage vector, significantly reducing steady-state current ripple and torque fluctuation.

[0050] S3. Signal generation steps: Decompose the final synthesized voltage vector obtained from the two spaces into the action time of the inverter's basic space voltage vector. Based on their respective action times, form the fundamental space modulation wave and the third harmonic space modulation wave respectively. Then, superimpose the modulation waves of the two spaces to generate a switching signal.

[0051] Specifically, the core of this step lies in converting the final synthesized voltage vector calculated independently in the two spaces into a switching sequence executable by the inverter. From the perspective of PWM waveform generation, the fundamental space and the third harmonic space each calculate the duration of the final synthesized voltage vector, corresponding to two different sets of PWM modulation waves. This invention superimposes these two sets of modulation waves in the time domain to generate the final switching signal to control the inverter. The physical basis of the superposition principle is that the actual voltage vector output by the inverter is the result of a weighted synthesis of the two sets of basic space voltage vectors according to their duration. The modulation waves in the fundamental space and the third harmonic space reflect the control requirements of different frequency components, and the switching sequence generated after superposition can simultaneously meet the control requirements of both spaces within the same switching cycle. Since the control loops of the two spaces operate in parallel and independently, and use the same control cycle and sampling frequency, the superposition operation does not introduce additional computational burden or delay.

[0052] Further, the specific process of step S1 is as follows: the basic space voltage vector of the five-phase inverter is divided into large vector, medium vector and small vector according to the magnitude, in the fundamental frequency space. In the coordinate system, the large and medium vectors in the same direction are synthesized with a weight ratio of 0.618:0.382 to obtain ten virtual voltage vectors in the fundamental space. These ten vectors have equal amplitudes, uniform phase angles, and adjacent intervals of 72° in the fundamental space, and their projections in the third harmonic space are zero. In the coordinate system, the large vector and the medium vector in the same direction are combined with a weight ratio of 0.618:0.382 to obtain ten virtual voltage vectors in the third harmonic space. These ten vectors have equal amplitudes, uniform phase angles, and adjacent intervals of 72° in the third harmonic space, and their projection in the fundamental space is zero.

[0053] Specifically, the core of this further definition lies in clarifying the specific synthesis method of the virtual voltage vector and the basis for determining its weight ratio. From the perspective of vector synthesis principles, the large and medium vectors of a five-phase inverter have the same direction in the fundamental space and opposite directions in the third harmonic space. By synthesizing them in an appropriate ratio, a larger amplitude can be retained in the fundamental space while the projections in the third harmonic space cancel each other out, achieving complete decoupling between the two spaces. The selection of the ratio 0.618:0.382 ensures that the synthesized vector maximizes its amplitude in its respective target space while satisfying the decoupling condition, which is beneficial for improving voltage utilization.

[0054] In a preferred embodiment, combined with Figure 3 and Figure 4 ,

[0055] 32 basic space voltage vectors of a five-phase two-level inverter It includes two zero vectors. And 30 non-zero fundamental voltage vectors, 30 non-zero fundamental voltage vectors In fundamental space and third harmonic space Distribution in Figure 3 As shown. Based on the fundamental frequency amplitude, these vectors can be divided into 10 large vectors, 10 medium vectors, and 10 small vectors. This invention uses the following synthesis formula to combine a large vector and a medium vector in the same direction in the fundamental frequency space into a single virtual voltage vector:

[0056] i=1,2,…,10

[0057] In the formula, For the fundamental spatial virtual voltage vector, For fundamental space Large vectors in For fundamental space The median vector in the vector.

[0058] Similarly, this invention uses the following synthesis formula to combine a large vector and a medium vector in the same direction in the third harmonic space into a virtual voltage vector:

[0059] i=1,2,…,10

[0060] In the formula, The virtual voltage vector in space for the third harmonic. For the third harmonic space Large vectors in For the third harmonic space The median vector in the vector.

[0061] This composition operation makes the vector The projections in the third harmonic space cancel each other out to zero, making The projections in the fundamental space cancel each other out to zero, thus eliminating the mutual influence between the two spaces and allowing for independent control. Ultimately, the ten virtual voltage vectors synthesized in both spaces exhibit identical distribution across all spaces, with equal amplitude and uniform phase angles (adjacent vectors spaced 2π / 5 apart), as shown below. Figure 4 As shown.

[0062] Furthermore, the specific process of selecting the optimal and suboptimal vectors in step S2 is as follows: First, from five non-adjacent virtual voltage vectors, the initial voltage vector is selected by minimizing the voltage cost function; then, from the two virtual voltage vectors adjacent to the initial voltage vector, the suboptimal voltage vector is selected by minimizing the same voltage cost function; the vector with the lagging phase angle in the initial and suboptimal voltage vectors is named the optimal vector, and the vector with the leading phase angle is named the suboptimal vector; the selection process is the same in the fundamental space and the third harmonic space.

[0063] Specifically, the core of this further definition lies in rapidly determining the two fundamental vectors used to synthesize the final voltage vector through a two-step screening method. From a computational efficiency perspective, evaluating the cost function for each of the ten virtual vectors individually would require ten cost calculations. The two-step screening method employed in this invention utilizes the geometric distribution characteristics of virtual vectors on the complex plane: the ten vectors are evenly distributed, with an angle of 36° between adjacent vectors, while the five non-adjacent vectors (interval of 72°) can essentially cover the entire direction. The first step evaluates only the five non-adjacent vectors to determine the initial voltage vector with the optimal direction; the second step evaluates only the two vectors adjacent to the initial voltage vector to determine the secondary voltage vector. The two steps require a total of seven cost calculations, reducing the computational burden by 30%. Furthermore, due to the small spacing between candidate vectors, the two selected vectors are necessarily the optimal and second-best combination, and the optimal solution will not be lost due to downsampling.

[0064] From the perspective of phase relationship, the one with the phase lag between the initial voltage vector and the secondary voltage vector is named the optimal vector. The vector with the leading phase angle is called the suboptimal vector. This naming established a unified coordinate system for subsequent angle adjustment duty cycle calculations. and The area formed by the extension is a fan-shaped region, and the reference voltage vector must fall within this region or its extension, ensuring the solvability of the subsequent angle adjustment formula.

[0065] In a preferred embodiment, combined with Figure 5 and Figure 6 In the fundamental frequency space, preliminary screening is performed on five non-adjacent virtual voltage vectors (such as...). Figure 5 shown , , , , In this process, each virtual voltage vector is substituted sequentially into the voltage cost function of the fundamental space, and the virtual voltage vector that minimizes the voltage cost function is selected. As the initial voltage vector . Figure 6 In, with Two adjacent vectors are , Substitute these two vectors into the voltage cost function again, and select the vector that minimizes the voltage cost function. Secondary voltage vector .Compare and In the fundamental space virtual vector control set, the phase with phase lag is named the optimal vector. The vector with the leading phase angle is named the suboptimal vector. .

[0066] Furthermore, the voltage cost function in the fundamental space is constructed based on the sum of squares of the difference between the reference voltage vector and the voltage vector at the current moment, wherein the reference voltage vector is calculated from the current reference value, the current current sample value, the motor parameters, and the rotor electrical angle; the voltage cost function in the third harmonic space is constructed using an open-loop given method, wherein the reference voltage vector is calculated from the rotor angular velocity and the amplitude of the third harmonic flux linkage of the permanent magnet, and the direct-axis reference voltage is set to zero.

[0067] Specifically, this further clarifies the calculation methods for the reference voltage vector and the construction of the cost function in the fundamental frequency space and the third harmonic frequency space, respectively. From the control principle analysis of the fundamental frequency space, based on the voltage equation of the permanent magnet synchronous motor, the reference voltage vector can be calculated by proportionally adjusting the deviation between the current reference value and the current sampled value, plus a back electromotive force compensation term. This method is equivalent to predicting the required voltage vector based on the current tracking error in each control cycle, ensuring that the current in the next cycle can track the reference value after applying the voltage. The cost function uses the square of the Euclidean distance between the reference voltage and the candidate voltage, which has a clear physical meaning, is simple to calculate, and requires no weighting coefficient tuning.

[0068] From the control principle of the third harmonic space, the main excitation source of the third harmonic current is the back electromotive force generated by the rotation of the rotor's third harmonic flux linkage. This back electromotive force... In the rotating coordinate system, this is represented by the q-axis component, with an amplitude of The d-axis component is zero. Therefore, this invention employs an open-loop feedforward method to directly calculate the voltage component with the same amplitude but opposite direction to the back electromotive force, serving as the reference voltage vector for the third harmonic space. The core advantage of this open-loop compensation strategy is that it eliminates the need for closed-loop sampling and adjustment of the high-frequency third harmonic current, avoiding a significant increase in switching frequency to improve control bandwidth. Furthermore, the open-loop feedforward calculation only requires motor speed and permanent magnet parameters, resulting in minimal computational load and no increase in processor burden. From the control effect analysis, as long as the amplitude of the permanent magnet's third harmonic flux linkage is accurately identified, this open-loop compensation can effectively counteract the influence of the back electromotive force, bringing the net voltage of the third harmonic space close to zero, thereby significantly suppressing the third harmonic current.

[0069] In a preferred embodiment,

[0070] sampling The stator phase current at time t is obtained by Clark transform and Park transform. Current sampling value in coordinate system , Obtain from the speed loop controller or other upper-level strategies. Moment Shaft current reference value , ; Obtain the current electric angular velocity of the motor ; Use a position sensor to obtain the angle The following equation (fundamental wave space prediction equation) is used to obtain Moment Shaft voltage reference value:

[0071]

[0072] In the formula, , They are respectively Spatial Shaft reference voltage vector value, Stator phase resistance, , They are respectively time Spatial Shaft current sampling value, , The fundamental space in the synchronous rotating coordinate system are respectively Shaft inductor, To control the cycle, , They are respectively Spatial Shaft current reference value, The electric angular velocity of the motor. This represents the fundamental amplitude of the permanent magnet flux linkage.

[0073] The shaft voltage is given through an open loop and obtained using the following formula. Moment Shaft voltage reference value:

[0074]

[0075] In the formula, , They are respectively Spatial Shaft reference voltage vector value, This represents the amplitude of the third harmonic flux linkage of the permanent magnet.

[0076] Obtained through the inverse Park transform spatial The axis reference voltage vector is given by the following formula:

[0077]

[0078] In the formula, , They are respectively Spatial Shaft reference voltage vector value; The rotor electrical angle.

[0079] The fundamental spatial voltage cost function is constructed as follows:

[0080]

[0081] In the formula, The fundamental space voltage cost function, , They are respectively time Spatial Axis voltage vector value.

[0082] Because the fundamental frequency of the third harmonic space current is relatively high, closed-loop control requires a significant increase in the switching frequency. To reduce switching losses and prevent the control cycle from being compressed, a method based on... A third harmonic current suppression strategy is given by the open-loop axis voltage. This strategy suppresses the third harmonic current by directly injecting a voltage component corresponding to the back electromotive force into the third harmonic rotating coordinate system. The calculation process of the cost function in the third harmonic space is as follows:

[0083] The reference voltage vector is calculated from the voltage component corresponding to the back electromotive force using the following formula:

[0084]

[0085] In the formula, , They are respectively Spatial Shaft reference voltage vector value, This represents the amplitude of the third harmonic flux linkage of the permanent magnet.

[0086] Obtained through the inverse Park transform The reference voltage vector in space is given by the following formula:

[0087]

[0088] In the formula, , They are respectively Spatial Axis reference voltage vector value.

[0089] The third harmonic space voltage cost function is constructed as follows:

[0090]

[0091] In the formula, The third harmonic space voltage cost function. , They are respectively time Spatial Axis voltage vector value.

[0092] Further, the angle adjustment and amplitude adjustment in step S2 are specifically as follows: by introducing the angle adjustment duty cycle to adjust the ratio of the action time of the optimal vector and the suboptimal vector, the direction of the final synthesized voltage vector is consistent with the direction of the reference voltage vector; then by introducing the amplitude adjustment duty cycle to adjust the amplitude of the synthesized vector, the amplitude of the final synthesized voltage vector is consistent with the amplitude of the reference voltage vector, and a zero vector is added to participate in the synthesis.

[0093] Specifically, the core of this further limitation lies in achieving continuous adjustment of the angle and amplitude of the synthesized vector through dual duty cycle adjustment. From the geometric principles of vector synthesis, the optimal vector... and suboptimal vector The vector spans a sector-shaped region. Any vector falling within this sector can be represented as... and Convex combination: ,in The duty cycle is used to adjust the angle, with a value range of [0,1]. Adjustment is made by... It can make The direction is and The values ​​change continuously. When the reference voltage vector direction is exactly aligned with... When the directions are consistent, the goal of angle adjustment is achieved.

[0094] After the angle adjustment is completed The direction is aligned with the reference voltage vector, but the amplitude may not be equal to the reference voltage vector. In this case, an amplitude adjustment duty cycle is introduced. ,Will Combined with the zero vector: Since the addition of the zero vector does not change the direction of the composite vector, but only reduces the equivalent amplitude, it can be adjusted... This allows the magnitude of the final synthesized voltage vector to be exactly equal to the magnitude of the reference voltage vector. When When =1, the magnitude of the composite vector is the largest (equal to 1). (amplitude); when When =0, the composite vector is the zero vector. and With dual-degree-of-freedom adjustment, the final synthesized voltage vector can achieve continuous stepless adjustment in both angle and amplitude dimensions, with a theoretical tracking error of zero, which is impossible to achieve with traditional finite set model predictive control.

[0095] In a preferred embodiment, combined with Figure 7 Angle adjustment duty cycle Calculate using the following formula:

[0096]

[0097] In the formula, , Voltage vectors Voltage components along the α and β axes; , Voltage vectors Voltage components along the α and β axes; , Voltage vectors Voltage components along the α and β axes. Voltage vector. This refers to the fundamental frequency space reference voltage vector or the third harmonic frequency space reference voltage vector. In the fundamental space, the two components correspond to: , In the third harmonic space, the two components correspond to: .

[0098] Amplitude adjustment duty cycle Calculate using the following formula:

[0099]

[0100] The final composite voltage vector formula can be obtained as follows:

[0101]

[0102] In the formula, For the final synthesized voltage vector, , Do not use the zero vector for all 0s or all 1s switching states. =0000, =1111.

[0103] At this point, the magnitude and angle of the final synthesized voltage vector are exactly the same as those of the reference voltage vector, enabling continuous modulation.

[0104] Furthermore, the action time of the basic space voltage vector of the inverter described in step S3 is specifically calculated as follows: based on the angle adjustment duty cycle, amplitude adjustment duty cycle, control cycle, and the preset synthesis ratio of the large vector and the medium vector, the action time of the large vector in the same direction as the optimal vector, the medium vector in the same direction as the optimal vector, the large vector in the same direction as the second-optimal vector, the medium vector in the same direction as the second-optimal vector, and the two zero vectors are calculated respectively.

[0105] Specifically, the core of this further limitation lies in decomposing the synthesized final voltage vector into the action time series of the basic spatial voltage vector that the inverter can directly execute. From the hierarchical structure analysis of the vector decomposition, the final synthesized voltage vector... The synthesis path is a three-layer nested structure: the first layer, Depend on According to the zero vector Proportional synthesis; second layer, Depend on and according to Proportional synthesis; third layer, and Each of the two vectors, a large one and a small one, is combined in a ratio of 0.618:0.382. Therefore, [the following is a possible interpretation:] Decomposing it into basic voltage vectors requires calculating the duration of action of each basic vector layer by layer, following the reverse process of the synthesis path described above.

[0106] The decomposition results involve six basic vectors: and Large vectors in the same direction, and Same direction median vector, and Large vectors in the same direction, and The median vectors in the same direction, and two zero vectors ( , The two zero vectors evenly distribute the remaining time, which helps optimize switching losses and common-mode voltage. This decomposition ensures that the equivalent voltage vector of the inverter output is exactly equal to the zero vector within a single switching cycle. Meanwhile, the symmetry of the switching sequence helps to reduce harmonic content.

[0107] In a preferred embodiment, the formula for calculating the duration of action of the basic space voltage vector is:

[0108]

[0109] In the formula, To and The duration of action of large vectors with the same direction; To and The duration of action of the median vectors in the same direction; To and The duration of action of large vectors with the same direction; To and The duration of action of the median vectors in the same direction; , zero vector , The duration of their respective effects.

[0110] Furthermore, the screening and action time calculation steps in the fundamental frequency space and the third harmonic frequency space are executed independently and in parallel.

[0111] Specifically, the core of this further limitation lies in emphasizing that the control loops of the two spaces can operate independently and in parallel without interfering with each other. From the perspective of parallel processing, the fundamental space and the third harmonic space have already achieved complete decoupling in the virtual voltage vector synthesis stage—the virtual vector in the fundamental space projects to zero in the third harmonic space, and vice versa. Therefore, the cost function evaluation, vector selection, and duty cycle calculation of the two spaces can be performed completely independently, without cross-coupling or iterative coordination. This feature makes the algorithm of this invention naturally suitable for parallel implementation on dual-core or multi-core processors, with the two cores handling the control tasks of the fundamental space and the third harmonic space respectively, without waiting for each other, doubling the computational efficiency. Even with a single-core processor, the control calculations of the two spaces can be executed sequentially. Since the two algorithms have identical structures, the code reusability is high, and the programming implementation is simple. Another benefit of independence is that if the control of one space malfunctions (such as sensor failure), the other space can still operate normally, giving the system partial fault tolerance.

[0112] Furthermore, the specific process of generating the switching signal by superimposing the modulation waves in step S3 is as follows: the first set of PWM modulation waves is generated based on the action time of each basic space voltage vector calculated from the fundamental wave space, and the second set of PWM modulation waves is generated based on the action time of each basic space voltage vector calculated from the third harmonic space. The two sets of modulation waves are superimposed in the time domain to obtain the final switching control signal.

[0113] Specifically, the core of this further definition lies in elucidating the method for generating switching signals through the superposition of dual-space modulated waves. Analyzing the PWM generation principle, the duration of each set of basic space voltage vectors can be mapped to the switching duty cycle of each phase arm of a five-phase inverter. The duration calculated by the fundamental space control loop corresponds to a duty cycle sequence, and the duration calculated by the third harmonic space control loop corresponds to another duty cycle sequence. These two duty cycle sequences reflect the requirements of the fundamental and third harmonic components on the inverter's output voltage, respectively. Since the inverter can only output one set of switching states within a single switching cycle, it is necessary to merge the two control requirements into a single switching sequence. This invention employs a time-domain superposition method: the duty cycles of the two sets of PWM modulated waves are algebraically added to obtain the final duty cycle command, which is then used to generate the switching signal via carrier comparison.

[0114] From a physical perspective, superposition operation is essentially a linear combination of the fundamental frequency control requirement and the third harmonic control requirement. Since there is a linear relationship between the inverter output voltage and the switching duty cycle, the output voltage generated by the superimposed duty cycle is equivalent to the vector sum of the fundamental and third harmonic components. This superposition principle is mathematically accurate, ensuring that the goal of independent dual-space control can be simultaneously achieved on a single inverter.

[0115] In a preferred embodiment, combined with Figure 8 Each vector in the fundamental frequency space generates a modulation wave during its interaction time, and each vector in the third harmonic frequency space also generates a modulation wave during its interaction time. The two modulation waves are added together to form the final modulation wave, which is then sent to the five-phase inverter. When superimposing modulation waves, it is important to note that the amplitudes of the fundamental frequency space modulation wave and the third harmonic frequency space modulation wave may exceed the carrier range. Normalization and amplitude limiting are necessary to ensure the inverter operates normally.

[0116] Examples and performance verification:

[0117] To verify the superior performance of the method proposed in this invention, a simulation system for a five-phase high-speed permanent magnet synchronous motor with parameters as shown in Table 1 was built.

[0118] Table 1

[0119]

[0120] The switching frequency is 25kHz. A third harmonic flux linkage, with an amplitude of 3% of the fundamental flux linkage amplitude, is added to the permanent magnet flux linkage. Using traditional control methods without controlling the third harmonic space, the phase current and phase current spectrum of the motor at steady state under rated operating conditions are as follows: Figure 9 As shown, the phase current is significantly distorted, with the third harmonic accounting for 9.7%. Therefore, it is necessary to actively control the third harmonic space and suppress the third harmonic space current.

[0121] With the switching frequency maintained at 25kHz and the motor operating under rated conditions, the phase current waveform and its spectrum are shown below when the third harmonic current reaches steady state after open-loop suppression using the method of this invention. Figure 10 As shown. It can be seen that, with Figure 9 In comparison, after incorporating the third harmonic space open-loop control algorithm, phase current distortion was significantly reduced, with the proportion of third harmonic phase current decreasing from 9.7% to 7.2%, a reduction of 25.8%. This verifies the effectiveness of the method in suppressing third harmonic current. Furthermore, since this invention uses the same control cycle (switching frequency 25kHz) as the fundamental space current for third harmonic control, there is no need to increase the switching frequency, avoiding the problems of increased switching losses and increased processor computational burden. In terms of dynamic performance, the response speed of the fundamental space current loop is unaffected by the third harmonic open-loop control, and the system can still maintain fast current tracking capability and stable torque output during load changes.

[0122] While the invention has been described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the invention. Therefore, it should be understood that many modifications can be made to the exemplary embodiments, and other arrangements can be designed without departing from the spirit and scope of the invention as defined by the appended claims. It should be understood that different dependent claims and features described herein can be combined in ways different from those described in the original claims. It is also understood that features described in conjunction with individual embodiments can be used in other described embodiments.

Claims

1. A low-harmonic model predictive control method for a five-phase motor based on dual-space vector synthesis, characterized in that, The method includes the following steps: S1. Dual-space virtual voltage vector synthesis steps: In the fundamental wave space and the third harmonic wave space, the large vector and the medium vector in the same direction are synthesized according to a preset ratio to generate ten virtual voltage vectors with equal amplitude, uniform distribution and mutual decoupling in the two spaces, which constitute the virtual vector control set of the fundamental wave space and the virtual vector control set of the third harmonic wave space respectively. S2. Screening and Action Time Calculation Steps: In two spaces, based on the given current reference value and the current sampling value at the current moment, the optimal vector and the second-best vector are screened from the corresponding virtual vector control set, and the action time of the two is calculated by angle adjustment and amplitude adjustment to calculate the final synthesized voltage vector. S3. Signal generation steps: Decompose the final synthesized voltage vector obtained from the two spaces into the action time of the inverter's basic space voltage vector. Based on their respective action times, form the fundamental space modulation wave and the third harmonic space modulation wave respectively. Then, superimpose the modulation waves of the two spaces to generate a switching signal.

2. The five-phase motor low-harmonic model predictive control method based on dual-space vector synthesis according to claim 1, characterized in that, The specific process of step S1 is as follows: the basic space voltage vector of the five-phase inverter is divided into large vector, medium vector and small vector according to the magnitude of the vector, in the fundamental space... In the coordinate system, the large and medium vectors in the same direction are synthesized with a weight ratio of 0.618:0.382 to obtain ten virtual voltage vectors in the fundamental space. These ten vectors have equal amplitudes, uniform phase angles, and adjacent intervals of 72° in the fundamental space, and their projections in the third harmonic space are zero. In the coordinate system, the large vector and the medium vector in the same direction are combined with a weight ratio of 0.618:0.382 to obtain ten virtual voltage vectors in the third harmonic space. These ten vectors have equal amplitudes, uniform phase angles, and adjacent intervals of 72° in the third harmonic space, and their projection in the fundamental space is zero.

3. The five-phase motor low-harmonic model predictive control method based on dual space vector synthesis according to claim 1, characterized in that, The specific process of selecting the optimal and suboptimal vectors in step S2 is as follows: First, from five non-adjacent virtual voltage vectors, the initial voltage vector is selected by minimizing the voltage cost function; then, from the two virtual voltage vectors adjacent to the initial voltage vector, the suboptimal voltage vector is selected by minimizing the same voltage cost function; the vector with the lagging phase angle in the initial and suboptimal voltage vectors is named the optimal vector, and the vector with the leading phase angle is named the suboptimal vector; the selection process is the same in the fundamental space and the third harmonic space.

4. The five-phase motor low-harmonic model predictive control method based on dual space vector synthesis according to claim 3, characterized in that, The voltage cost function in the fundamental space is constructed based on the sum of squares of the differences between the reference voltage vector and the voltage vector at the current moment, wherein the reference voltage vector is calculated from the current reference value, the current current sample value, the motor parameters, and the rotor electrical angle; the voltage cost function in the third harmonic space is constructed using an open-loop given method, wherein the reference voltage vector is calculated from the rotor angular velocity and the amplitude of the third harmonic flux linkage of the permanent magnet, and the direct-axis reference voltage is set to zero.

5. The five-phase motor low-harmonic model predictive control method based on dual-space vector synthesis according to claim 3, characterized in that, The angle adjustment and amplitude adjustment mentioned in step S2 are specifically as follows: by introducing an angle adjustment duty cycle, the action time ratio of the optimal vector and the suboptimal vector is adjusted so that the direction of the final synthesized voltage vector is consistent with the direction of the reference voltage vector; Then, by introducing an amplitude adjustment duty cycle, the amplitude of the synthesized vector is adjusted so that the amplitude of the final synthesized voltage vector is consistent with that of the reference voltage vector, and a zero vector is added to participate in the synthesis.

6. The five-phase motor low-harmonic model predictive control method based on dual space vector synthesis according to claim 5, characterized in that, The action time of the basic space voltage vector of the inverter described in step S3 is as follows: based on the angle adjustment duty cycle, amplitude adjustment duty cycle, control cycle, and the preset synthesis ratio of the large vector and the medium vector, the action time of the large vector in the same direction as the optimal vector, the medium vector in the same direction as the optimal vector, the large vector in the same direction as the second-optimal vector, the medium vector in the same direction as the second-optimal vector, and the two zero vectors are calculated respectively.

7. The five-phase motor low-harmonic model predictive control method based on dual-space vector synthesis according to claim 1, characterized in that, The screening and action time calculation steps in the fundamental frequency space and the third harmonic frequency space are executed independently and in parallel.

Images