Ship propulsion six-phase permanent magnet motor model predictive control method

Abstract

The application discloses a ship propulsion six-phase permanent magnet motor double-subspace composite virtual vector cooperative robust predictive control method, a vector space decoupling mathematical model of a ship propulsion six-phase permanent magnet synchronous motor is established, a decoupling voltage equation of an alpha-beta basic wave subspace and an x-y harmonic subspace is obtained, a double-subspace extended composite virtual vector control set containing a basic virtual vector and a composite virtual vector is independently constructed, independent closed-loop model predictive current controllers of the alpha-beta subspace and the x-y subspace are respectively designed, respective independent predictive cost functions are constructed, a sector-based vector preselection mechanism is adopted to quickly optimize the extended composite virtual vector control set, and optimal voltage vectors and optimal duty cycles of corresponding subspaces are generated. A model-free sliding mode speed controller is designed to replace a traditional PI speed regulator, is used to generate a torque or current reference instruction, and is combined with the double-subspace model predictive current controller to form a robust predictive control system.

Description

Technical Field

[0001] This invention relates to the field of motor control technology, and more particularly to a bi-space composite virtual vector cooperative robust predictive control method for a six-phase permanent magnet synchronous motor used in a marine electric propulsion system. Background Technology

[0002] As the shipbuilding industry moves towards higher reliability, electrification, intelligence, and green development, ship electric propulsion technology has gradually become a research hotspot. Six-phase permanent magnet synchronous motors (PMSMs) have shown great potential in ship electric propulsion systems due to their high power density, low torque ripple, and strong fault tolerance. Model predictive control (MRC) has received widespread attention in this field due to its fast dynamic response and convenient handling of multiple constraints. However, traditional virtual vector-based MRC methods have significant limitations: First, they typically only perform closed-loop control in the α-β fundamental subspace, while employing open-loop or simple suppression strategies for the xy harmonic subspace, making it difficult to effectively address low-order harmonic currents caused by inverter nonlinearity and dead-zone effects. Second, the fixed amplitude and direction of traditional virtual vectors limit further improvements in current tracking accuracy. Finally, the speed loop generally relies on traditional PI controllers, which are complex to tune and lack robustness under nonlinear, time-varying disturbances such as ship propeller loads. Although existing technologies have optimized harmonic suppression through virtual vector modulation, they have not yet achieved active closed-loop control of the harmonic subspace and have not solved the dynamic robustness problem of the speed loop. Summary of the Invention

[0003] To address the aforementioned problems, this invention proposes a robust predictive control method for a six-phase permanent magnet motor in a ship propulsion system, employing a dual-space composite virtual vector cooperative control mechanism. This method aims to achieve independent closed-loop cooperative control of the α-β and xy dual-space systems, actively suppressing harmonics and improving current tracking accuracy and system dynamic robustness. The technical means employed in this invention are as follows: This invention provides a robust predictive control method for a six-phase permanent magnet motor in a ship propulsion system, employing a dual-space composite virtual vector cooperative control mechanism, comprising:

[0004] S1: Establish a vector space decoupling mathematical model for a six-phase permanent magnet synchronous motor for ship propulsion, and obtain the decoupling voltage equations for the α-β fundamental subspace and the xy harmonic subspace; S2: Independently construct a basic virtual vector control set and a composite virtual vector extended control set in the α-β fundamental subspace and xy harmonic subspace, respectively. The control set controls the effective voltage vector in the corresponding subspace, wherein the mapped voltage component in the other subspace is zero. S3: Based on the vector space decoupling mathematical model, design independent closed-loop model predictive current controllers for α-β subspace and xy harmonic subspace respectively, construct their respective independent prediction cost functions, and use a sector-based vector pre-selection mechanism to optimize the composite virtual vector extended control set to generate the optimal voltage vector and its optimal duty cycle for the corresponding subspace. S4: Coordinate and allocate the application time of the optimal voltage vectors in the α-β subspace and xy harmonic subspace, and synthesize the final application voltage vector within one control cycle; S5: Design a modelless sliding mode speed controller to generate torque or current reference commands, and together with the bispace model predictive current controller, form a robust predictive control system.

[0005] Furthermore, the extended control set of the composite virtual vector is constructed using the following method: S21: Based on the traditional virtual vector synthesis principle, synthesize the fundamental virtual vector set VVs-αβ of the fundamental subspace that generates zero average voltage in the xy harmonic subspace; S22: Based on the traditional principle of virtual vector synthesis, the basic virtual vector set VVs-xy of the harmonic subspace that generates zero average voltage in the α-β fundamental wavelet subspace is synthesized by using the voltage vectors that are distributed in opposite directions in the xy subspace. S23: Using the discrete spatial virtual vector modulation method, based on the fundamental virtual vector set VVs-αβ and the harmonic subspace fundamental virtual vector set VVs-xy respectively, and taking the fractional nodes of adjacent fundamental virtual vectors as mother vectors, a composite virtual vector of medium amplitude is synthesized to form the final extended composite virtual vector control set CVVs-αβ and extended composite virtual vector control set CVVs-xy; the required basic virtual voltage vector synthesis process is as follows:

[0006]

[0007] Where: v53 and v36 are real vectors with the same direction but different amplitudes in the α-β subspace, the ratio of the action time of v53 and v36 in a single cycle is denoted as Tn and Tm, the total execution cycle is Ts, and Vdc is the DC bus voltage.

[0008] Furthermore, when using a sector-based vector pre-selection mechanism to optimize the extended control set of composite virtual vectors: the composite virtual vectors in each subspace are divided into four sectors according to the coordinate axes; the optimization process is carried out in two stages: the first stage selects a representative vector from each sector for preliminary prediction and evaluation to determine the sector where the optimal vector is located; the second stage performs detailed prediction and evaluation on the remaining vectors in the optimal sector.

[0009] Furthermore, the extended composite virtual vector control set implements closed-loop control of harmonic currents in the xy subspace within predictive control. When the composite virtual vector is selected as a candidate control set, the principal plane composite virtual vector regulates the active component of electromagnetic torque in the α-β subspace, while the secondary plane composite virtual vector suppresses the reactive component of harmonic currents in the xy subspace. The virtual voltage vectors in the two subspaces map to zero in each other's spaces, thus allowing independent control of the two subspaces. Combining predictive delay compensation and zero-vector space modulation, the dual-subspace current prediction model and cost function are expressed as follows:

[0010]

[0011] To determine the modulation duty cycles rαβ and rxy of the composite virtual vector, based on the constructed bi-subspace current prediction model, the bi-subspace cost function is used... and Starting from this point, we derive the analytical expression for the duty cycle:

[0012]

[0013]

[0014]

[0015]

[0016] Where: ud, uq, id, iq, Ld, and Lq are the stator voltage, current, and inductance components on the dq axis, respectively; ux, uy, ix, and iy are the stator voltage and current components in the xy subspace, respectively; R is the stator resistance; ωr is the rotor angular velocity; ψf is the permanent magnet flux linkage; P is the number of pole pairs; J is the moment of inertia; Te and TL are the electromagnetic torque and load torque, respectively; B is the damping coefficient; L0 is the leakage inductance in the xy plane; and Ts is the sampling period.

[0017] Furthermore, the sector-based vector pre-selection mechanism is as follows: the composite virtual vectors in each subspace are divided into four sectors according to the coordinate axes, and the optimization process is carried out in two stages: the first stage selects a representative vector from each sector for preliminary prediction and evaluation to determine the sector where the optimal vector is located; the second stage only performs detailed prediction and evaluation on the remaining vectors in the optimal sector.

[0018] Furthermore, a model-free sliding mode speed controller is designed based on a hyperlocal model of the velocity loop. The control law of the model-free sliding mode speed controller includes a sliding mode feedback controller and a sliding mode observer. The sliding mode observer is used to estimate the lumped disturbance of the system in real time. The sliding mode feedback controller uses this estimate and combines it with the speed tracking error to calculate the disturbance-resistant torque reference command, the expression of which is:

[0019] Constructing a velocity loop hyperlocal model based on model-free control theory:

[0020] In the above formula: g The part is unknown. The torque adjustment coefficient is the control objective, which is to design the torque command. This causes the rotor speed to... Accurately track reference values ke is the scaling factor of the sliding surface. The sliding surface is defined by c, which is the convergence coefficient used to adjust the response speed, and k1 and k2 are the switching and feedback gain coefficients.

[0021] This invention focuses on a six-phase permanent magnet synchronous motor in a ship propulsion system, and deeply analyzes the fundamental theory of its traditional virtual vector model predictive current control. A bi-space composite virtual vector cooperative robust predictive control method for a six-phase permanent magnet motor in ship propulsion is developed, which achieves the following control effects: Active Cooperative Harmonic Suppression in Twin Subspaces: By constructing an independent virtual vector set of harmonic subspaces and implementing closed-loop predictive control, active tracking and suppression of harmonic currents are achieved, overcoming the shortcomings of open-loop control of harmonic subspaces in traditional methods, and exhibiting strong robustness to nonlinear disturbances such as dead zones.

[0022] High-precision current tracking: The use of an extended composite virtual vector set increases the diversity and precision of control degrees of freedom, thereby improving the current tracking performance in the fundamental subspace.

[0023] Strong dynamic robustness: The model-free sliding mode speed controller does not rely on precise model parameters and has strong robustness to external load disturbances and changes in internal parameters, which significantly improves the system's response performance under dynamic conditions such as sudden load changes.

[0024] High computational efficiency and practicality: The sector pre-selection mechanism significantly reduces the computational burden caused by the expansion of the vector set; and the system does not require tuning of complex weighting factors and PI parameters, making it easy to implement and apply in engineering. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of the present invention 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 some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 A structural block diagram of a six-phase PMSM propulsion system for ships provided in an embodiment of the present invention; Figure 2 A control block diagram for a robust predictive control method for a six-phase permanent magnet motor in a twin-space composite virtual vector cooperative system for ship propulsion. Figure 3 This is a schematic diagram showing the distribution of the basic virtual vector set (VVs-αβ, VVs-xy) in the bi-subspace constructed in this invention; Figure 4 This is a schematic diagram showing the distribution of the bi-space composite virtual vector set (CVVs-αβ, CVVs-xy) synthesized in this invention; Figure 5 This is a schematic diagram of a sector-based vector preselection mechanism; Figure 6 This is a schematic diagram of the experimental verification platform; Figure 7 Block diagram for modeling propeller load characteristics; Figure 8 The relationship between propulsion motor speed and load torque, (a) motor speed variation curve, (b) load torque; Figure 9 The steady-state experimental results of the method of the present invention and the comparative method under typical working conditions are as follows: (a) conventional VVMPC, (b) existing SVO-MPC method, and (c) CVV-DCRPC method disclosed in this invention. Figure 10 The transient experimental results of the method of the present invention and the comparative method under typical working conditions are shown in (a) conventional VVMPC, (b) existing SVO-MPC method, and (c) CVV-DCRPC method disclosed in the present invention. Detailed Implementation

[0027] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0028] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0029] like Figure 1 As shown, this invention provides a model predictive control method for a six-phase permanent magnet motor for ship propulsion based on virtual voltage vector space modulation, specifically including the following steps: S1: Establish a vector space decoupling mathematical model for a six-phase permanent magnet synchronous motor (PMSM) for ship propulsion. For a Y-connected, 30° phase-shifted six-phase PMSM, a vector space decoupling transformation is used to obtain the decoupling voltage equations in the α-β fundamental subspace and the xy harmonic subspace, and the α-β subspace (transformed to the dq axis by Park transformation).

[0030] S2: Construct a composite virtual vector control set. The principle of traditional virtual vector synthesis is as follows: Figure 3 As shown, VVs-αβ is synthesized with the goal of eliminating the voltage in the xy subspace. Based on this, the present invention further synthesizes a VVs-xy specifically for regulating the xy subspace. To further improve control accuracy, such as... Figure 4 As shown, using DSVVM technology, the fractional vector of the basic virtual vector is used as the mother vector to synthesize the composite virtual vector extended control set CVVs-αβ and CVVs-xy with amplitudes between the zero vector and the basic virtual vector, thus forming an extended control set containing multiple amplitude and direction selections.

[0031] S3: Dual-subspace closed-loop predictive control and fast optimization. Independent cost functions are designed for the α-β subspace and the xy subspace respectively. To reduce the computational cost of traversing and optimizing the extended vector set (e.g., 24 vectors / subspace), a method is adopted... Figure 5 The sector pre-selection mechanism shown reduces the number of optimization attempts per subspace from 24 to 10, effectively ensuring real-time performance.

[0032] S4: Coordinated Action Time. After obtaining the optimal vectors and their optimal duty cycles rαβ and rxy of the two subspaces, the action time is coordinated and allocated according to the modulation rule of prioritizing the principal plane and then the harmonic plane in the two subspaces. The vectors are applied sequentially within one Ts to synthesize the final voltage output.

[0033] S5: Model-less sliding mode speed control: speed loop structure as follows Figure 2 As shown in the MFSC diagram. Based on the hyperlocal model, a sliding mode observer is designed for estimation, and a sliding mode control law is designed to calculate the torque command Te. The controller has good parameter adaptability and does not require frequent tuning as motor parameters change.

[0034] Specifically, Figure 2 The overall control steps shown are as follows: (1) Sample the phase current, rotor position and speed at time k, perform VSD and Park transformation on the phase current to obtain the current id, q and ix, y of each subspace at time k; using the concept of virtual voltage vector of dual subspace, synthesize multiple basic voltage vectors of dual subspace in a single cycle, and then synthesize the composite virtual voltage vector set of dual subspace according to the spatial discrete virtual voltage vector modulation, so as to realize the closed-loop control of harmonic current in the xy subspace in predictive control, and further improve the tracking and suppression capability of harmonic current; (2) Calculate the current values ​​id, q, x, y(k+1) at time k+1 based on the optimal voltage vector compensation at the previous time; obtain the q-axis current reference values ​​id, q-ref through the speed loop; (3) Calculate the predicted current values ​​id, q, x, y(k+2) at time k+2 corresponding to the candidate voltage vector using the current compensation value at time k+1, and obtain the value function; When CVVs are selected as the candidate control set, CVVs-αβ regulates the active component of electromagnetic torque in the α-β subspace, while CVVs-xy suppresses the reactive component of harmonic current in the xy subspace. Since the virtual voltage vectors of the two subspaces map to zero in each other's spaces, independent control of the two subspaces can be achieved. Combining predictive delay compensation and zero-vector space modulation techniques, the reconstructed dual-subspace current prediction model and dual-subspace cost function are as follows:

[0035]

[0036] (4) Insert virtual vector modulation coefficients r into the prediction model and reconstruct the prediction model. This invention derives the function about r from the value function J.

[0037]

[0038]

[0039]

[0040]

[0041]

[0042] Compared with traditional methods, this study focuses on suppressing harmonic currents caused by system nonlinearity and load disturbances, and proposes a solution of implementing model predictive current closed-loop control in the xy harmonic subspace. Since the two subspaces use independent control sets, optimization needs to be performed separately to select their respective optimal voltage vectors: one acts on the α-β subspace, and the other acts on the xy subspace.

[0043] The constructed composite virtual vector has the following important characteristic: the composite virtual vector in any subspace is mapped to a zero voltage vector in another subspace. Based on this characteristic, the zero vector traditionally used for control of one subspace can be equivalently replaced by the effective voltage vector of the other subspace. Therefore, the current control of the two subspaces can be completed collaboratively within the same control cycle, achieving synchronous closed-loop regulation of the current components in both subspaces.

[0044] However, within a single control cycle, the sum of the action times of the two effective vectors may exceed Ts. To prioritize the system's ability to regulate the active component (i.e., the α-β subspace current), this paper imposes the following constraint on the action times of the two optimal subspace vectors:

[0045] (5) Real-time computation is ensured through sector partitioning and vector pre-selection mechanisms. First, the composite virtual vectors in each subspace are divided into four sectors using the α-β and xy coordinate axes. Taking the α-β subspace as an example: First, four representative composite virtual vectors (CVV3, CVV6, CVV9, CVV12) are selected for initial prediction optimization to determine the sector where the optimal vector is located (taking the first sector corresponding to CVV3 as an example, its partitioning is shown in the diagram below). Figure 5 (As shown in Table 1). Subsequently, a second prediction optimization is performed only on the remaining 6 vectors within the sector to finally determine the optimal composite virtual vector in this subspace. Using this strategy, the number of model evaluations required for each subspace is reduced from 24 to 10, thereby reducing the total computational cost of the two subspaces from 48 to 20. The specific optimization order of the voltage vectors is shown in Table 1.

[0046] TABLE Ⅰ Gemini Space Composite Virtual Vector Optimization Order

[0047] (6) Design a modelless sliding mode speed controller to achieve robust speed control under ship operating conditions.

[0048] In traditional methods, the speed loop of a Six-Phase PMSM typically relies on a PI controller to generate a torque or current reference. However, when the system is subjected to disturbances, the output accuracy of the PI controller drops significantly, and its optimal parameters are difficult to tune precisely in scenarios requiring rapid deployment, such as "plug and play," thus limiting the system's dynamic performance and robustness. Therefore, this paper proposes a method based on the input of the speed loop... Based on the output relationship, an independent model-free speed controller was designed to further improve the system's robustness under complex ship operating conditions.

[0049] The dynamic equation in the velocity loop under external disturbance can be expressed as:

[0050] Constructing a velocity loop hyperlocal model based on model-free control theory:

[0051] The model-free sliding mode controller is designed as follows:

[0052] By treating the velocity error as a state variable, we can obtain the velocity tracking error, and introduce an integral error to eliminate the steady-state error:

[0053] To reduce steady-state error, the sliding mode surface can be designed as follows:

[0054] Using the exponential reaching law design, the control law can be obtained as follows:

[0055] The expression for the model-free sliding mode controller (SMC) is as follows:

[0056] To accurately estimate unknown disturbances in the velocity loop, a sliding diaphragm observer is designed as follows: in, k 3>0, k 4>0, 0 <v<1。

[0057] Based on the newly constructed predictive control algorithm for a six-phase permanent magnet motor for ship propulsion, which is based on a dual-space composite virtual vector cooperative robustness, the control algorithm disclosed in this paper is verified by building an experimental platform. The experimental platform demonstrates the following: Figure 6 As shown.

[0058] In the experimental verification, to realistically simulate the actual operating conditions of the ship's propulsion system and verify the effectiveness of the control algorithm, this experiment used the actual ship parameters of the "New Red and Expert" electric propulsion teaching experimental vessel of Dalian Maritime University (see Table 2). Based on the propeller mathematical model (such as... Figure 7 (As shown) and actual ship data, the static characteristic curve of propeller thrust torque as a function of rotational speed was calculated, as follows: Figure 8 As shown. The discrete data of this torque curve, after being appropriately scaled, serves as the torque command for the load motor (simulating a propeller), thereby achieving dynamic loading of the propulsion motor. The DC bus voltage Udc is 380V, and the sampling frequency is 10kHz.

[0059] Table II. Actual Ship Parameters of the Xin Hongzhuan Ship

[0060] To evaluate the control performance of different control methods, a comparative experiment was first conducted under typical ship propulsion conditions. The target speed of the propulsion motor was set at 200-300 rpm, and the load motor was determined according to... Figure 8 (b) shows the load characteristic curve, with a dynamic load applied at the corresponding rotational speed. The control methods compared include: existing Virtual Vector Model Predictive Control (VVMPC), Space Vector Optimization Model Predictive Control (SVO-MPC), and the CVV-DCRPC method disclosed in this invention for a six-phase permanent magnet motor for ship propulsion. Experimental results are as follows: Figure 9 and 10 As shown.

[0061] Figure 9 The comprehensive performance indicators of various control methods under steady-state operation are presented, including speed fluctuation, phase currents (iA and iD), dq-axis currents (id and iq), xy-harmonic subspace currents (ix, iy), and total harmonic distortion (THD) of phase currents. Figure 9 (c) It can be seen that, using the proposed CVV-DCRPC method, the system exhibits excellent steady-state performance under the condition that the q-axis current is given as the rated value of 2.2 A. Specifically, compared with the VVMPC and SVO-MPC methods, CVV-DCRPC limits the speed fluctuation to within ±2 rpm, resulting in higher steady-state accuracy; the pulsation of the dq-axis current is significantly reduced; and most notably, the xy harmonic current is strictly controlled within the amplitude range of 0.35 A, with an optimization of approximately 30%. In terms of harmonic suppression, CVV-DCRPC achieves the most significant improvement: the phase current THD is reduced to 5.74%, which is 51.2% and 47.6% lower than VVMPC and SVO-MPC, respectively, with the most significant suppression effect on low-order harmonics (especially the 5th harmonic).

[0062] The above steady-state experimental results demonstrate that the CVV-DCRPC method for robust predictive control of a six-phase permanent magnet motor for ship propulsion, disclosed in this invention, combines the MFSC speed outer loop with a dual-subspace current predictive control inner loop based on composite virtual vectors, achieving synergistic optimization of the system's steady-state performance. Its core lies in implementing closed-loop predictive control of the xy harmonic subspace currents, significantly suppressing harmonic currents while also improving the control accuracy of the fundamental plane current, thereby enhancing the overall steady-state performance.

[0063] To further evaluate the performance of the control strategy under transient conditions, a transient response experiment was conducted to investigate potential load surges that may occur during shipboard operations. The experiment simulated a typical operating condition: at 0.5 s, the target speed of the main propulsion motor jumped from 200 rpm to 300 rpm. The transient response waveforms under different control methods are shown below. Figure 10 As shown, the dynamic changes of rotational speed, id, iq, ix, iy, and ia, id are compared and demonstrated. Figure 10 (c) The dynamic tracking performance of the proposed CVV-DCRPC method is demonstrated. Experimental results show that the method achieves smooth transition and rapid tracking during dynamic load surges. Specifically, compared with VVMPC and SVO-MPC methods, CVV-DCRPC significantly reduces the system's dynamic response time from approximately 0.9 s to 0.3 s, an optimization of 66.7%. This performance improvement stems mainly from two aspects: firstly, the adopted MFSC overcomes the dynamic lag problem caused by the difficulty in parameter tuning and adaptability to multiple operating conditions of traditional PI controllers; secondly, the predictive control framework itself possesses superior real-time performance and speed. Experimental results verify that MFSC has a faster response speed and stronger system robustness when dealing with dynamic load surges on ships.

[0064] In terms of current control, CVV-DCRPC also performs excellently. Before and after sudden changes in operating conditions, the ripple of the dq-axis current is smaller, the phase current waveform maintains better sinusoidality, and the harmonic content is significantly reduced. Crucially, the amplitudes of ix and iy remain stable throughout the dynamic process, effectively suppressed to within 0.35 A. In contrast, existing methods (VVMPC, SVO-MPC) typically employ open-loop control strategies based on fundamental virtual vectors for ix and iy, resulting in a significant increase in harmonic current amplitude with increasing load. In high-power traction applications such as ship propulsion, the motor operates under heavy loads for extended periods, leading to considerable additional harmonic losses and reduced system energy efficiency. The CVV-DCRPC method disclosed in this invention achieves robust speed tracking and effective suppression of harmonic currents during dynamic processes through the collaborative design of the speed outer loop MFSC and the current inner loop (DS-CVVMPCC). Its core advantage lies in the fact that by implementing closed-loop predictive control of the xy harmonic subspace current, it can effectively suppress harmonic currents even under heavy load or dynamic load conditions, while improving the overall current control accuracy of the fundamental plane and the harmonic plane.

[0065] Therefore, the feasibility and effectiveness of a robust predictive control method for a six-phase permanent magnet motor in a ship propulsion system, based on a bi-space composite virtual vector cooperative, are demonstrated.

[0066] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A robust predictive control method for a six-phase permanent magnet motor for ship propulsion using a dual-space composite virtual vector cooperative system, characterized in that... include: S1: Establish a vector space decoupling mathematical model for a six-phase permanent magnet synchronous motor for ship propulsion, and obtain the decoupling voltage equations for the α-β fundamental subspace and the xy harmonic subspace; S2: Independently construct a basic virtual vector control set and a composite virtual vector extended control set in the α-β fundamental subspace and xy harmonic subspace, respectively. The control set controls the effective voltage vector in the corresponding subspace, wherein the mapped voltage component in the other subspace is zero. S3: Based on the vector space decoupling mathematical model, design independent closed-loop model predictive current controllers for α-β subspace and xy harmonic subspace respectively, construct their respective independent prediction cost functions, and use a sector-based vector pre-selection mechanism to optimize the composite virtual vector extended control set to generate the optimal voltage vector and its optimal duty cycle for the corresponding subspace. S4: Coordinate and allocate the application time of the optimal voltage vectors in the α-β subspace and xy harmonic subspace, and synthesize the final application voltage vector within one control cycle; S5: Design a modelless sliding mode speed controller to generate torque or current reference commands, and together with the bispace model predictive current controller, form a robust predictive control system.

2. The robust predictive control method for a six-phase permanent magnet motor for ship propulsion, as described in claim 1, is characterized in that: The extended control set of the composite virtual vector is constructed using the following method: S21: Based on the traditional virtual vector synthesis principle, synthesize the fundamental virtual vector set VVs-αβ of the fundamental subspace that generates zero average voltage in the xy harmonic subspace; S22: Based on the traditional principle of virtual vector synthesis, the basic virtual vector set VVs-xy of the harmonic subspace that generates zero average voltage in the α-β fundamental wavelet subspace is synthesized by using the voltage vectors that are distributed in opposite directions in the xy subspace. S23: Using the discrete spatial virtual vector modulation method, based on the fundamental virtual vector set VVs-αβ and the harmonic subspace fundamental virtual vector set VVs-xy respectively, and taking the fractional nodes of adjacent fundamental virtual vectors as mother vectors, a composite virtual vector of medium amplitude is synthesized to form the final extended composite virtual vector control set CVVs-αβ and extended composite virtual vector control set CVVs-xy; the required basic virtual voltage vector synthesis process is as follows: Where: v53 and v36 are real vectors with the same direction but different amplitudes in the α-β subspace, the ratio of the action time of v53 and v36 in a single cycle is denoted as Tn and Tm, the total execution cycle is Ts, and Vdc is the DC bus voltage.

3. The robust predictive control method for a six-phase permanent magnet motor for ship propulsion, as described in claim 1, is characterized in that: When using a sector-based vector pre-selection mechanism to optimize the extended control set of composite virtual vectors: the composite virtual vectors in each subspace are divided into four sectors according to the coordinate axes; The optimization process is carried out in two stages: the first stage selects a representative vector from each sector for preliminary prediction and evaluation to determine the sector where the optimal vector is located. The second stage involves a detailed prediction and evaluation of the remaining vectors within the optimal sector.

4. The robust predictive control method for a six-phase permanent magnet motor for ship propulsion, as described in claim 1, is characterized in that: The extended composite virtual vector control set achieves closed-loop control of harmonic currents in the xy subspace within predictive control. When the composite virtual vector is selected as the candidate control set, the primary plane composite virtual vector regulates the active component of electromagnetic torque in the α-β subspace, while the secondary plane composite virtual vector suppresses the reactive component of harmonic currents in the xy subspace. The virtual voltage vectors in the two subspaces are mutually mapped to zero in their respective spaces, thus allowing independent control of the two subspaces. Combining predictive delay compensation and zero-vector space modulation, the dual-subspace current prediction model and cost function are expressed as follows: To determine the modulation duty cycles rαβ and rxy of the composite virtual vector, based on the constructed bi-subspace current prediction model, the bi-subspace cost function is used... and Starting from this point, we derive the analytical expression for the duty cycle: Where: ud, uq, id, iq, Ld, and Lq are the stator voltage, current, and inductance components on the dq axis, respectively; ux, uy, ix, and iy are the stator voltage and current components in the xy subspace, respectively; R is the stator resistance; ωr is the rotor angular velocity; ψf is the permanent magnet flux linkage; P is the number of pole pairs; J is the moment of inertia; Te and TL are the electromagnetic torque and load torque, respectively; B is the damping coefficient; L0 is the leakage inductance in the xy plane; and Ts is the sampling period.

5. The robust predictive control method for a six-phase permanent magnet motor for ship propulsion, as described in claim 1, is characterized in that: The sector-based vector pre-selection mechanism is as follows: the composite virtual vector in each subspace is divided into four sectors according to the coordinate axes. The optimization process is carried out in two stages: the first stage selects a representative vector from each sector for preliminary prediction and evaluation to determine the sector where the optimal vector is located. The second stage only performs detailed prediction and evaluation of the remaining vectors within the optimal sector.

6. The robust predictive control method for a six-phase permanent magnet motor for ship propulsion, as described in claim 1, is characterized in that: A model-free sliding mode speed controller is designed based on a hyperlocal model of the velocity loop. The control law of the model-free sliding mode speed controller includes a sliding mode feedback controller and a sliding mode observer. The sliding mode observer is used to estimate the lumped disturbance of the system in real time. The sliding mode feedback controller uses this estimate and combines it with the speed tracking error to calculate the disturbance-resistant torque reference command, the expression of which is: Constructing a velocity loop hyperlocal model based on model-free control theory: In the above formula: g The part is unknown. The torque adjustment coefficient is the control objective, which is to design the torque command. This causes the rotor speed to... Accurately track reference values ke is the scaling factor of the sliding surface. The sliding surface is defined by c, which is the convergence coefficient used to adjust the response speed, and k1 and k2 are the switching and feedback gain coefficients.

Images