Three-dimensional model prediction control method for open winding motor

Abstract

The invention provides a three-dimensional model prediction control method for an open winding motor, and relates to the technical field of motor control. The invention provides a vector selection method based on subspace selection and sector judgment, which selects an optimal vector output by an inverter through a value function, and comprises the following steps of: classifying a subspace consisting of four adjacent vectors based on three-dimensional space vector distribution and a geometric symmetry relationship of an open winding motor; a reference vector is obtained according to the combination of the deadbeat principle and voltage prediction; according to a sector corresponding to a reference vector in an alpha-beta plane, determining a subspace possibly containing an optimal vector, and reducing the number of candidate vectors from 27 to 7; traversing the seven candidate vectors according to a voltage value function to select four vectors with minimum spatial distances from the reference vector; traversing the four candidate vectors according to a current value function to select an optimal vector; and the inverter outputs the optimal vector. By using a three-dimensional model predictive control method, low switching frequency and high current control accuracy can be realized. In addition, reference values of the zero-sequence current and the q-axis current which enable the stator rotation flux linkage to be unchanged under the open-phase fault are deduced under the condition that the third harmonic flux linkage is considered. By means of the provided three-dimensional model prediction control method, reference values of zero-sequence current and q-axis current are changed, so that fault-tolerant control over the open-phase fault in the open-winding permanent magnet synchronous motor is achieved, and torque ripple is restrained at the same time.

Description

Technical Field

[0001] The present invention relates to the technical field of motor control, and particularly to a three-dimensional model predictive control method for an open-winding motor. Background Art

[0002] Permanent magnet synchronous motors have received increasing attention due to their advantages such as high efficiency, high power density, high reliability, and simple structure. The open-winding permanent magnet synchronous motor powered by a dual-inverter provides an effective solution for achieving higher output power and stronger fault tolerance. The drive topologies of open-winding permanent magnet synchronous motors can be divided into three categories, namely, isolated DC bus topologies, common DC bus topologies, and hybrid topologies with floating capacitors. Among these three topologies, the open-winding permanent magnet synchronous motor topology with a common DC bus is the most popular due to its advantages of low cost and high modulation range. However, the low zero-sequence reactance in the common DC bus topology will cause zero-sequence current, which reduces the operating efficiency. Therefore, measures should be taken to eliminate the zero-sequence current.

[0003] Model predictive control has the advantages of fast dynamic response and flexible multi-objective control, and has broad application prospects in high-power drives. In order to suppress the zero-sequence current of the common DC bus open-winding permanent magnet synchronous motor topology using model predictive control, some research work has been carried out. Although the method of only traversing the vectors that do not generate zero-sequence voltage can achieve good suppression effect, this method will sacrifice the current control accuracy in the torque plane. There is a method to adjust the duration of the zero vectors to achieve controllable zero-sequence voltage. However, using 4 vectors within one switching period will lead to a significant increase in the switching frequency. In order to achieve the goals of low switching frequency and high current accuracy, a literature has made vector selection based on the vector distribution in the αβ0 coordinate system. However, the increase in candidate vectors will lead to a heavy processor operation load. In order to avoid traversing 27 three-dimensional space vectors, this literature selects candidate vectors according to the zero-sequence component and α-β component of the vectors. Although the optimal vector can be finally obtained using the value function, the traversal process is still very complex. There is a literature that obtains candidate vectors closest to the reference vector by performing subspace division in the three-phase coordinate system, and reduces the operation load by simplifying the geometric judgment equation. However, due to the difference in dq-axis inductance parameters, the vector closest to the reference vector may not be the optimal vector. Based on the above analysis, it is difficult for existing research to balance high current accuracy, low complexity, and low switching frequency.

[0004] Therefore, the present invention proposes a three-dimensional model predictive control method for an open-winding motor to solve the above problems. Summary of the Invention

[0005] Object of the Invention. To meet the requirements of low switching frequency and high current control accuracy in high-power applications of open-winding permanent magnet synchronous motor drives, this paper proposes a three-dimensional model predictive control method for an open-winding motor.

[0006] Technical solution: Establish a three-dimensional space vector diagram for the split-winding motor, and divide the three-dimensional space vector diagram into subspaces according to the geometric rotation characteristics between the space vectors. Calculate the reference voltage according to the deadbeat control. Reduce the candidate vectors to 7 according to the sector judgment and subspace judgment, and traverse the 7 candidate vectors according to the voltage value function to obtain 4 vectors closest to the reference vector. Obtain the optimal vector according to the current value function and finally output it by the inverter, so as to achieve the purpose of low switching frequency and high current control accuracy.

[0007] As an optimization, the candidate vectors are reduced from 27 to 7 according to the sector judgment and subspace selection, and 4 vectors closest to the reference vector are further selected as candidate vectors through the voltage value function. Finally, the optimal vector is obtained through the current value function, realizing the low complexity of the algorithm and high current control accuracy.

[0008] As an optimization, a single optimal vector is output by the inverter to achieve a low switching frequency.

[0009] As an optimization, considering the third harmonic flux linkage, the reference expression of the zero-sequence current that can maintain a circular rotating magnetic field under the open-phase fault is deduced, and the tracking of the zero-sequence current reference value is realized through three-dimensional model predictive control, so as to achieve the fault-tolerant control of the open-phase fault of the open-winding permanent magnet synchronous motor and further suppress the torque ripple.

[0010] Compared with the existing technologies, the beneficial effects of the present invention are:

[0011] (1) Compared with the traditional method, it takes into account both low switching frequency and high current control accuracy.

[0012] (2) For the open-phase fault of the permanent magnet synchronous motor, the zero-sequence current and the reference value of the q-axis current are modified on the basis of the proposed control method framework to realize the suppression of torque ripple considering the third harmonic flux linkage. Description of the drawings

[0013] Figure 1 It is the structural diagram of the split-winding motor driven by a common DC bus;

[0014] Figure 2 It is the space vector diagram: (a) Inverter I; (b) Inverter II;

[0015] Figure 3 It is the three-dimensional space vector distribution;

[0016] Figure 4 It is the schematic diagram of the three-dimensional modulation area: (a) Stereoscopic view; (b) Top view;

[0017] Figure 5 It is the subspace division of the upper modulation area;

[0018] Figure 6 It is a block diagram of three-dimensional model predictive control;

[0019] Figure 7 It is a top view of subspaces (I-V); (a) Subspace I; (b) Subspace II; (c) Subspace III; (d) Subspace IV;

[0020] (e) Subspace V; Specific implementation manners

[0021] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings, but the protection scope of the present invention is not limited to the following description.

[0022] In order to make the purpose, technical solution and advantages of the present invention clearer, the present invention will be further described in detail in combination with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention, that is, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Usually, the components of the embodiments of the present invention shown and described in the accompanying drawings herein can be arranged and designed in various different configurations.

[0023] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed present invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts belong to the protection scope of the present invention. It should be noted that relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations.

[0024] Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the phrase "including a..." does not exclude the existence of additional identical elements in the process, method, article or device including the element.

[0025] The features and performance of the present invention will be further described in detail below in combination with embodiments:

[0026] A three-dimensional model predictive control method for an open-winding motor provided by the present invention divides the three-dimensional space modulation region into subspaces, and obtains 4 vectors adjacent to the reference vector for vector synthesis; among them, the subspace division of the three-dimensional space modulation region includes the following process:

[0027] The phase voltage expression of an open-winding permanent magnet synchronous motor in the stationary coordinate system is as follows:

[0028]

[0029] Where u A / B / C , i A / B / C , ψ fA / B / C are the stator voltage, current and three-phase magnetic flux respectively. R s , L s and M s are the stator resistance, self-inductance and mutual inductance.

[0030] Considering the third harmonic in the magnetic flux, the rotor magnetic flux can be expressed as:

[0031]

[0032] Where θ e is the rotor electrical angle, ψ 1f and ψ 3f are the amplitudes of the fundamental magnetic flux and the third harmonic magnetic flux respectively.

[0033] Combining Equation (1) and Equation (2) and performing Park transformation, the phase voltage and electromagnetic torque in the dqz coordinate system can be obtained:

[0034]

[0035]

[0036] Where L d , L q , L0 are the d-axis, q-axis and zero-sequence inductances respectively. n p is the number of pole pairs of the motor, ω r is the rotor angular velocity, θ e is the electrical angle, ψ 1f and ψ 3f are the amplitudes of the fundamental and third harmonic fluxes.

[0037] For the dual-inverter structure, the synthetic voltage vector expression is:

[0038]

[0039] Where u s1 and u s2 represent the space voltage vectors generated by inverter I and inverter II respectively, uK1o and u K2o are the phase voltages of Inverter I and Inverter II (K = A, B, C).

[0040] Figure 2 is the vector diagram of the dual inverter. Each three-phase inverter can generate 8 switching states, i.e., 8 voltage vectors. The voltage vectors generated by Inverter I and Inverter II are named 1 - 8 and 1' - 8' respectively.

[0041] The zero-sequence voltages of the 64 vector combinations can be obtained by the following formula.

[0042]

[0043] S A1 , S B1 , S C1 , S A2 , S B2 , and S C2 are used to represent the switching states of the corresponding inverter legs, where "1" and "0" represent high level and low level respectively.

[0044] Substitute the 64 switching combinations into Equation (5) to obtain the zero-sequence voltage corresponding to each switching combination, as shown in Table I. To more intuitively present the relationship between the voltage positions and the vector combinations, in Table I, the vector combinations with the same position in the three-dimensional space are put together with parentheses.

[0045] Table I Zero-sequence voltages corresponding to the switching combinations

[0046]

[0047] According to Table I, there are 7 types of zero-sequence voltages for the 64 vector combinations. Through the α-β and zero-sequence components of each vector combination, the three-dimensional space vector distribution as shown in Figure 4 can be obtained. Among them, the voltage vector combinations with different voltage amplitudes in the α-β plane are marked with different colors.

[0048] The 27 different vectors are symmetrically distributed in 7 zero-sequence voltage layers. However,[[]] Figure 3 the geometric relationship of the voltage distribution shown in Figure 3 is too complex for designing a geometric criterion to select the optimal vector. Since dc , 1 / 3u dc , -1 / 3u

[0049] In traditional two-dimensional model predictive strategy control, candidate vectors are usually selected by dividing sectors in the α-β plane. After obtaining the sector corresponding to the reference vector, adjacent vectors in the α-β plane are brought into the traversal process. However, in three-dimensional model predictive control, the increase in the zero-sequence dimension complicates the geometric relationship of space vectors. In addition, it is difficult to obtain the optimal vector by only considering the α-β components of the reference vector. In a three-dimensional vector space, four adjacent voltage vectors can construct a basic subspace, and the four vertices of the subspace containing the reference vector can cover all directions in the three-dimensional space. Therefore, the optimal vector must be one of the adjacent vectors constructing the subspace containing the reference vector. In other words, the selection of candidate vectors can be completed by finding the subspace surrounding the reference vector.

[0050] Based on the above analysis, it is necessary to divide the three-dimensional space to reduce candidate vectors. Since the symmetry relationship between the vector with zero-sequence voltage of -1 / 3u dc and the vector with zero-sequence voltage of +1 / 3u dc , first, taking the α-β plane as the boundary, the three-dimensional space is divided into upper and lower parts. Then, the upper space is divided into 16 subspaces, and each subspace is composed of four adjacent vectors. For the convenience of display, the 16 subspaces are further divided into 5 groups, named subspaces (I-V) according to the geometric appearance, as Figure 5 shown. Different colors are used to distinguish each group of subspaces. The subspace distribution of the lower space follows the same rule.

[0051] The three-dimensional model predictive control block diagram for an open-winding permanent magnet synchronous motor is as Figure 6 shown. First, the dead-time control principle is combined with voltage prediction to obtain the reference vector, and a closed-loop current control for the torque plane and the zero-sequence dimension is established. As Figure 6 can be seen, there is a correlation between the voltage reference value, the current reference value, and the current at the (k + 1)th instant. Then, the candidate vectors are reduced to 7 (u 1-7 ) through sector judgment and subspace judgment. Then, in the voltage value function, u 1-7 is traversed to obtain the four vectors (u’ 1-4 ) closest to the reference voltage vector. Then, the optimal vector (u opt ) is obtained through the current value function. Finally, each inverter generates one vector in each control cycle. Low switching frequency and low current harmonics are guaranteed. Based on the proposed three-dimensional model predictive control, considering the third harmonic of the flux, a new fault-tolerant control scheme is proposed to completely eliminate the torque ripple of the open-winding permanent magnet synchronous motor drive under the open-phase fault. The proposed fault-tolerant control can be achieved by only updating the current reference value.

[0052] Reference voltage prediction

[0053] According to the forward Euler method, the dqz-axis voltages in Equation (3) can be expressed as:

[0054]

[0055] Where T s represents the switching period, and the subscripts k and k + n represent the voltages and currents at the corresponding times k and k + n.

[0056] According to Equation (6), the dqz currents at time k + 1 can be expressed as:

[0057]

[0058] The dqz voltages at time k + 1 can be expressed as:

[0059]

[0060] According to the deadbeat principle, the dqz reference voltages can be obtained through Equation (9):

[0061]

[0062] The αβz reference voltages can be obtained through coordinate transformation from Equation (10):

[0063]

[0064] Optimal vector selection

[0065] In the first step, the candidate vectors are reduced from 27 to 7 through sector judgment and subspace judgment. As Figure 7 shown, the α-β plane is divided into 12 30° sectors (I - XII). According to the vector angle on the α-β plane, the sector of the reference vector can be determined. Figure 7 (a)- Figure 7 (e) show the top views of the sector divisions of five subspaces (I - V). Taking sector - I in the upper half modulation part as an example, Figure 7 the subspace - OACE in (a) can cover sector - I, and the vectors OA, OC, OE, and the zero vector are selected as candidate vectors. Similarly, other candidate vectors OI, OS, and OH are obtained from the subspaces - ACIO, ASHO, and AHIO in Figure 7 (b)- Figure 7 (e). Therefore, the 7 adjacent vectors that construct the subspace are selected as candidate vectors. Due to the rotational symmetry relationship of the vectors, regardless of which sector the reference vector is located in, 4 subspaces corresponding to 7 vectors can be obtained. In this way, the number of candidate vectors can be reduced to 7.

[0066] In the second step, the voltage value function of Equation (12) is used to traverse the 7 candidate vectors, and 4 candidate vectors closest to the reference vector are selected.

[0067] F u = |u α * - u αi | 2 + |u α * - u βi | 2 + |u z * - u zi | 2 (12)

[0068] where u α-i , u β-i , u z-i are the αβz voltage components of the candidate vectors.

[0069] These 4 vectors are finally traversed by the current value function of Equation (13) to select the optimal voltage vector:

[0070]

[0071] II. Fault Tolerant Strategy for Open-Winding Motor Phase Loss Fault

[0072] A fault tolerant control considering the third harmonic flux is proposed for the phase loss fault. To ensure that the rotating magnetic field formed by the stator winding remains unchanged, the q-axis current is compensated to further suppress the torque ripple. The following is the derivation of the q-axis current reference value taking the phase A loss fault as an example;

[0073] According to the virtual healthy concept, the zero-sequence current expression that satisfies the generation of the rotating magnetomotive force can be obtained:

[0074] i z * = - i α * = i q * sin(θ e ) (14)

[0075] To compensate for the torque ripple caused by the zero-sequence current and the third harmonic in the flux link, a compensation current i q0 can be injected into the q-axis current, and the torque can be rewritten as:

[0076]

[0077] To completely cancel out the torque ripple caused by the third harmonic flux linkage and the zero-sequence current in Equation (15), the q-axis compensation reference current i q0* It should satisfy the following formula:

[0078] ψ 1f i q0 * = 6ψ 3f sin(3θ e )i z * (16)

[0079] Considering the q-axis compensation reference current i q0 * After that, the zero-sequence reference current can be updated to:

[0080] i z * = (i q * + i q0 * )sin(θ e ) (17)

[0081] Substituting Equation (17) into Equation (16), the q-axis injection current reference value can be expressed as:

[0082]

[0083] By injecting the q-axis compensation current reference value i q0 * in Equation (18) into Figure 6 the q-axis reference current i q * , and updating the zero-sequence reference current i Figure 6 in z * with the zero-sequence current in Equation (17), the open-phase fault tolerance is achieved.

[0084] The above are only the embodiments of the present invention, and do not limit the scope of the present invention. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be equally included in the scope of the invention protection of the present invention.

Claims

1. A three-dimensional model predictive control method for an open-winding motor, characterized in that A three-dimensional space vector diagram of the split-winding motor is established, and the three-dimensional space vector diagram is divided into sub-spaces according to the geometric rotation characteristics between the space vectors; the reference voltage is calculated according to the deadbeat control; the candidate vectors are reduced to 7 according to the sector judgment and sub-space judgment, and 4 vectors closest to the reference vector are obtained by traversing the 7 candidate vectors according to the voltage value function; the optimal vector is obtained according to the current value function and finally output by the inverter, so as to achieve the purpose of low switching frequency and high current control accuracy.

2. The three-dimensional model predictive control method according to claim 1, characterized in that The adjacent vectors of the three-dimensional space vector diagram are connected, and a plane is established for every three adjacent vectors to construct the three-dimensional subspace boundary. The smallest three-dimensional space with an interval in the boundary is used as the three-dimensional subspace. According to the geometric symmetry relationship between the space vectors, the sub-spaces with exactly the same geometric shape are classified.

3. The three-dimensional model predictive control method according to claim 2, characterized in that, Generate the reference voltage vector according to the deadbeat principle and voltage prediction.

4. The three-dimensional model predictive control method according to claim 3, characterized in that, Determine 7 candidate vectors by combining sector judgment and sub-space judgment.

5. The three-dimensional model predictive control method according to claim 4, wherein, Traverse the 7 candidate vectors according to the voltage value function to obtain 4 vectors closest to the reference vector.

6. The three-dimensional model predictive control method according to claim 5, wherein, Traverse the 4 candidate vectors according to the current value function to obtain the optimal vector, and the inverter outputs the optimal vector.

7. The three-dimensional model predictive control method according to claim 6, wherein Considering the third harmonic flux linkage, the reference expressions of the zero-sequence current and q-axis current that can maintain a circular rotating magnetic field under the open-phase fault are derived.

8. The three-dimensional model predictive control method according to claim 7, wherein The tracking of the reference values of the zero-sequence current and q-axis current is realized through three-dimensional model predictive control, so as to realize the fault-tolerant control of the open-phase fault of the open-winding permanent magnet synchronous motor.

Images