Five-phase permanent magnet synchronous motor single-phase open circuit model prediction fault-tolerant control method adopting variable virtual voltage vector

Abstract

The invention discloses a five-phase permanent magnet synchronous motor single-phase open-circuit model prediction fault-tolerant control method adopting a variable virtual voltage vector, and belongs to the technical field of multi-phase motor fault-tolerant control. The method comprises the following steps: detecting a feedback rotating speed omega m of a motor, obtaining iqref through an error between the feedback rotating speed omega m and a given rotating speed omega *, and setting idref as 0; determining the izref under the principle that the copper loss is equal and the copper loss is minimum; sampling each phase current iA, iB, iC, iD and iE, and combining the transformation matrix to obtain feedback d-q-z axis current id (k), iq (k) and iz (k); a z-axis reference voltage Uzref is calculated by using a dead-beat algorithm; constructing an alternative variable virtual voltage vector and determining dynamic vector action time; substituting the alternative voltage vector into a value function to obtain an optimal voltage vector; the switching state corresponding to the optimal voltage vector is selected and input into a PWM module, and an obtained switching signal is input into an inverter to control the motor, so that single-phase open-circuit fault-tolerant control of the five-phase permanent magnet synchronous motor is realized.

Description

Technical Field

[0001] This invention relates to the field of fault-tolerant control technology for multiphase motors, and specifically to a single-phase open-circuit model prediction fault-tolerant control method for a five-phase permanent magnet synchronous motor using a variable virtual voltage vector. Background Technology

[0002] Permanent magnet synchronous motors (PMSMs) are increasingly widely used in electric vehicles, aerospace, and maritime navigation systems due to their high torque density, high efficiency, and high reliability. Furthermore, stable and reliable motor drive systems are particularly important for applications with high reliability requirements, such as aircraft and electric vehicles. Three-phase PMSMs suffer from insufficient reliability in certain special applications. Multiphase PMSMs, with their low torque ripple and high fault tolerance, are a hot research topic in the field of motors. Taking a five-phase motor as an example, open-circuit faults in the motor windings are a common type of fault. When a single-phase open-circuit fault occurs in the motor windings, adjusting the phase and amplitude of the inverter output current to ensure that the magnetomotive force is consistent before and after the fault can guarantee that the motor operates smoothly as under normal conditions. Currently, scholars both domestically and internationally have conducted numerous studies on fault-tolerant control of five-phase motors.

[0003] In recent years, scholars both domestically and internationally have conducted in-depth research on model predictive control (MPC) and achieved fruitful results. MPC features fast dynamic response, simple and flexible control, and ease of handling nonlinear constraints. Compared to direct torque control (DTC), MPC is more accurate and effective in vector selection. It selects the optimal voltage vector by predicting the motor state and substituting the predicted values ​​into a value evaluation function. Compared to vector control, MPC does not require current loop parameter tuning and can achieve better dynamic response characteristics. However, single-vector MPC has a large error, resulting in more harmonic currents. Furthermore, multi-vector MPC mainly focuses on the normal operating conditions of the motor, with limited research on fault-tolerant conditions. Research on fault-tolerant MPC control for five-phase motors mainly focuses on vector selection and vector synthesis under single-phase open-circuit faults, which still has significant limitations. Traditional virtual voltage vectors synthesize vectors by eliminating the cubic space voltage components, which cannot accurately control the cubic space current. Therefore, further optimization of vector synthesis under fault-tolerant conditions for five-phase motors has become a research hotspot. Summary of the Invention

[0004] To address the limitations of traditional virtual voltage vectors in accurately controlling the three-dimensional space current and enabling fault-tolerant operation of the motor under single-phase open-circuit faults, this invention proposes a single-phase open-circuit model prediction fault-tolerant control method for a five-phase permanent magnet synchronous motor using variable virtual voltage vectors.

[0005] To achieve the technical objectives, the present invention adopts the following technical solution:

[0006] A method for predictive fault-tolerant control of a five-phase permanent magnet synchronous motor using a single-phase open-circuit model with a variable virtual voltage vector includes the following steps:

[0007] Step 1, the detected feedback speed ω of the five-phase permanent magnet synchronous motor m By comparing the given rotational speed ω* with the actual rotational speed, the motor's speed error is obtained. A PI controller is then used to calculate the q-axis reference current i of the five-phase permanent magnet synchronous motor based on this speed error. qref i d For the d-axis current, due to the use of i d =0 control, d-axis reference current i of the five-phase permanent magnet synchronous motor dref Set to 0;

[0008] Step 2: When a single-phase open-circuit fault occurs in the five-phase permanent magnet synchronous motor, the corresponding z-axis reference current i can be determined according to the principles of minimum copper loss and equal copper loss, respectively. zref This enables control of the z-subspace of a five-phase permanent magnet synchronous motor after a single-phase open-circuit fault.

[0009] Step 3: Sample the A, B, C, D, and E phase currents i of the five-phase permanent magnet synchronous motor using a current sensor. A i B i C i D i E Based on the sampled current, the reduced-order matrix corresponding to the phase loss during an open-circuit fault is determined. Using the selected reduced-order matrix, a matrix transformation is performed on the sampled phase currents of the five-phase permanent magnet motor to obtain the dqz-axis current i fed back by the five-phase permanent magnet synchronous motor at the current time k during the fault. d (k), i q (k), i z (k);

[0010] Step 4: Based on the z-axis current i at time k fed back by the five-phase permanent magnet synchronous motor. z (k) and reference current i zref The z-axis reference voltage U can be predicted using a deadbeat current prediction algorithm. zref ;

[0011] Step 5, based on the synthesized vector z-axis voltage U synz Tracking the z-axis reference voltage U zref Based on the principle of [the above], alternative variable virtual voltage vectors are constructed and the dynamically changing vector action time is determined;

[0012] Step 6: Combining the five-phase motor discrete prediction model and the transformation matrix from the stationary coordinate system to the rotating coordinate system, substitute the candidate variable virtual voltage vector into the value function based on the dqz axis current error, and perform rolling optimization to find the optimal voltage vector and corresponding action time allocation.

[0013] Step 7: Select the optimal voltage vector and input its corresponding switching state to the PWM module to obtain the switching signals of each phase. Input the obtained switching signals to the inverter to control the motor and realize the single-phase open-circuit fault-tolerant control of the five-phase permanent magnet synchronous motor.

[0014] Furthermore, the z-axis reference current i under the principles of minimum and equal copper loss described in step 2 is... zref as follows:

[0015] After a single-phase open-circuit fault occurs in a five-phase motor, the copper loss generated by the motor can be expressed as the sum of the copper losses generated by the d, q, and z axis currents:

[0016] R s (i2 d+i2 q+i2 z); where R s i is the phase resistance of the motor; x (x=d, q, z) represents the dqz axis current of the motor;

[0017] When the principle of minimizing copper loss is adopted (LJL), the z-axis reference current i zref It should be set as: ;

[0018] The five-phase current under the principle of minimizing copper loss is: ;

[0019] When the principle of equal copper loss (EJL) is adopted, it means that the magnitudes of the remaining phase currents are equal, i.e., the remaining phase currents satisfy: i B =-i D and i C =-i E z-axis reference current i zref It should be set as: ;

[0020] The five-phase current under the principle of equal copper loss is: ;

[0021] Where, iy xref (x=d, q, z, y=LJL, EJL) are the dqz axis reference currents under the principles of minimum copper loss and equal copper loss; iy x (x= A, B, C, D, E, y=LJL, EJL) are the phase currents under the principles of minimum copper loss and equal copper loss; I m θ represents the current amplitude of the motor under normal operating conditions. e The electric angle is the motor angle.

[0022] Furthermore, the calculation of the feedback dqz-axis current mentioned in step 3 is as follows:

[0023] Since the prediction of the z-axis reference voltage and the rolling selection of the value function in model predictive control require feedback of the dq z-axis current, the current i collected by the current sensor is used. A i B i C i D i E The current i at time k is obtained by transforming the data using the Clarke matrix to the α-β axis, and then by transforming it using the Park matrix to the dqz axis. d (k), i q (k), i z (k):

[0024] ;

[0025] in, The missing Park transformation matrix for phase A. Here is the Clarke transformation matrix for phase A missing.

[0026] ,

[0027] ;

[0028] Where α = 2 / 5π, is the electrical angle difference between adjacent windings of the five-phase motor.

[0029] Furthermore, in step 4, the z-axis reference voltage u zref The calculation is as follows:

[0030] Step 4.1: Discretizing the motor's z-axis voltage equation yields the corresponding z-axis current prediction equation:

[0031] ;

[0032] Among them, L s For motor leakage inductance; T s i is the sampling time; z (k) represents the current at time k; i z (k+1) represents the current at time k+1; u z (k) represents the voltage at time k;

[0033] Step 4.2, based on the deadbeat current prediction method, the reference current i zref The current i at time k+1 z The reference voltage U can be predicted by (k+1). zrefTherefore, by substituting the corresponding z-axis reference currents under the different principles of minimum and equal copper loss in step 2, we can obtain the z-axis reference voltages under the two principles: ;

[0034] Where uy zref (y=LJL, EJL) is the z-axis reference voltage under the principles of minimum copper loss and equal copper loss.

[0035] Furthermore, based on the synthesized vector z-axis voltage U synz Tracking the z-axis reference voltage U zref Based on the principle of synthesizing variable virtual voltage vectors, determining the action time of two vectors in each vector combination, and performing amplitude limiting and normalization correction on the action time, the α-β-z axis voltage components of the synthesized vector are then calculated to construct alternative variable virtual voltage vectors.

[0036] Step 5.1: Based on the Clarke transformation matrix and switch states after the single-phase loss in Step 3, the vector distribution after single-phase open-circuit fault tolerance can be obtained, including the α-β subspace and the z subspace containing U1~U 14 There are a total of 14 valid vectors and U0, U 15 Two zero vectors are divided into [U0], [U9], and [U8] based on their distribution in the α-β subspace. 13 ]、[U 12 U 10 ]、[U 14 ,U4],[U6],[U7,U2],[U3,U5],[U1,U 11 ]、[U 15 The combination of ] and the combination of two voltage vectors, whose vector composition relationships in the α-β subspace and z subspace are as follows: ;

[0037] Among them, T opt1 T opt2 U represents the duration of action of vectors 1 and 2 in the combination; opt1z U opt2z U synz U represents the voltage components of vector 1, vector 2, and the composite vector in the z-subspace. opt1αβ U opt2αβ U synαβ Let U be the voltage components of vector 1, vector 2, and the composite vector in the α-β subspace, satisfying U opt1αβ =U opt1α +jU opt1β U opt2αβ =U opt2α +jU opt2β U synαβ =U synα +jUsynβ ;

[0038] Step 5.2, based on the synthesized vector z-axis voltage U synz Tracking the z-axis reference voltage U zref Based on the principle of synthesizing variable virtual voltage vectors and determining the vector action time, considering the voltage vector distribution and spatial symmetry, individual vectors [U0], [U9], [U6], and [U... 15 The remaining combinations containing two voltage vectors are divided into two types;

[0039] Type 1 includes [U8, U 13 ]、[U 14 [U4], [U7, U2], [U1, U 11 ], with [U8, U 13 For example, voltage vectors U8 and U 13 The components in the α-β subspace are 0.4413U. dc e j60 / 180π 0.3245U dc e j46 / 180π The z-subspace components are -0.2351U. dc 0.3804U dc U dc For the DC bus voltage, according to the principle of vector composition, voltage vectors U8 and U... 13 The range of the vertex of the composite vector in the α-β subspace is the third side of the vector triangle formed by the two vectors. The range of the vertex of the composite vector in the z-subspace is the line segment formed by the vertices of the two vectors. This is to satisfy the z-axis voltage U of the composite vector. synz Tracking the z-axis reference voltage U zref We can obtain: ;

[0040] Among them, T * opt1(U8) T * opt2(U13) For voltage vectors U8 and U 13 Duration of action;

[0041] Considering that the vector action time must be within the interval of 0 to 1, when it exceeds this interval, the action time is restricted to the boundaries of the interval, 0 and 1. Furthermore, if T... * opt1(U8) +T * opt2(U13) If the total duration of action exceeds 1, the following correction will be made: ;

[0042] Among them, T opt1(U8) T opt2(U13)For voltage vectors U8 and U 13 The revised duration of action;

[0043] After obtaining the corrected vector action time, alternative variable virtual voltage vectors can be established: ;

[0044] Among them, U synz(U8- U13) For voltage vectors U8 and U 13 The voltage component of the composite vector in the z-subspace; U synzαβ(U8- U13) For voltage vectors U8 and U 13 The voltage component of the composite vector in the α-β subspace;

[0045] Type 2 includes [U 12 U 10 ]、[U3、U5]、with[U8、U 13 For example, voltage vector U 12 U 10 The components in the α-β subspace are 0.6155U. dc e j90 / 180π 0.1453U dc e j90 / 180π The z-subspace components are 0.1453U. dc -0.6155U dc According to the principle of vector composition, the voltage vector U 12 U 10 The range of the vertex of the composite vector in the α-β subspace and z subspace is the line segment formed by the two vector vertices. This is to satisfy the z-axis voltage U of the composite vector. synz Tracking the z-axis reference voltage U zref We can obtain: ;

[0046] Among them, T * opt1(U12) T * opt2(U10) Voltage vector U 12 U 10 Duration of action;

[0047] Considering that the vector action time must be within the interval of 0 to 1, when it exceeds this interval, the action time is restricted to the boundaries of the interval, 0 and 1. Furthermore, if T... * opt1(U12) +T * opt2(U10) If the total duration of action exceeds 1, the following correction will be made:

[0048] ;

[0049] Among them, T opt1(U12)T opt2(U10) Voltage vector U 12 U 10 The revised duration of action;

[0050] After obtaining the corrected vector action time, alternative variable virtual voltage vectors can be established:

[0051] ;

[0052] Among them, U synz(U12- U10) Voltage vector U 12 U 10 The voltage component of the composite vector in the z-subspace; U synzαβ(U12- U10) Voltage vector U 12 U 10 The voltage component of the composite vector in the α-β subspace;

[0053] Step 5.3, similarly, can solve for the virtual voltage vector formed by the combination of the remaining voltage vectors, and add a single effective voltage vector and a zero vector to form a set of alternative variable virtual voltage vectors, denoted as VV0~VV9.

[0054] Furthermore, the specific steps of voltage vector rolling optimization in step 6 are as follows:

[0055] Step 6.1: Discretizing the dqz axis voltage equations of the five-phase motor yields the corresponding dqz axis current prediction equations:

[0056] ;

[0057] Among them, L d L q T is the inductance of the dq axis of the motor; s i is the sampling time; d (k), i q (k) represents the current at time k; i d (k+1),i q (k+1) represents the current at time k+1; u d (k), u q (k) represents the voltage at time k;

[0058] Step 6.2, substitute the α-β-z-axis components of the variable virtual voltage vector generated in step 5 into the values ​​from step 3. From the matrix, the corresponding dqz axis components can be obtained: ;

[0059] Step 6.3, convert the dqz axis components U of the variable virtual voltage vector synd U synq Usynz Substitute the voltage u at time k into the prediction equation in step 6.1 d (k), u q (k), u z (k) gives the current i at time k+1 of the candidate variable virtual voltage vector. d (k+1),i q (k+1),i z (k+1);

[0060] Step 6.4, the current i at time k+1 of the candidate variable virtual voltage vector. d (k+1),i q (k+1),i z Substituting (k+1) into the following value function based on the dqz axis current error, the optimal voltage vector and corresponding action time allocation can be obtained through rolling optimization: .

[0061] This invention enhances the control of the z-axis current by constructing a variable virtual voltage vector, thereby satisfying different constraints such as minimum and equal copper losses, ultimately achieving fault-tolerant control of a five-phase permanent magnet motor under single-phase open-circuit faults. It is suitable for applications with high reliability requirements, such as aerospace, electric vehicles, and ship propulsion systems.

[0062] This invention has the following beneficial effects: Based on the traditional virtual voltage vector constructed from a single-phase open-circuit fault vector, this invention utilizes a deadbeat-free algorithm to transform the tracking of z-axis current into the tracking of z-axis voltage, and based on the synthesized vector z-axis voltage U... synz Tracking the z-axis reference voltage U zref Based on the principle of [missing information], alternative variable virtual voltage vectors are constructed, and the dynamically changing vector action time is established. Finally, these vectors are substituted into the model's predictive value function for rolling optimization to obtain the optimal voltage vector combination, thereby achieving precise control of harmonic space current. This approach can satisfy different principles, such as minimum copper loss and equal copper loss, and is computationally simple and easy to implement. It can significantly improve the current waveform of a five-phase PMSM under fault conditions while ensuring motor torque performance. Experimental results show that this fault-tolerant control strategy can achieve high-quality operation of a five-phase motor drive system under single-phase open-circuit faults. Attached Figure Description

[0063] Figure 1 The vector distribution diagram for a single-phase open circuit is shown in Figure 1; (a) fundamental frequency space; (b) harmonic frequency space.

[0064] Figure 2 U8 and U 13 Vector synthesis; (a) fundamental wave space; (b) harmonic wave space;

[0065] Figure 3:U 12 and U 10 Vector synthesis; (a) fundamental wave space; (b) harmonic wave space;

[0066] Figure 4 : Variable virtual voltage vector distribution diagram;

[0067] Figure 5 : Block diagram of single-phase open-circuit model prediction fault-tolerant control for five-phase permanent magnet synchronous motor based on variable virtual voltage vector;

[0068] Figure 6 The current and torque waveforms of a five-phase permanent magnet synchronous motor from normal to fault to fault-tolerant; (a) minimum copper loss; (b) equal copper loss;

[0069] Figure 7 Current and torque waveforms of a five-phase permanent magnet synchronous motor with single-phase open circuit faults (speed 100 r / min); (a) Method 1; (b) Method 2; (c) Method 3;

[0070] Figure 8 Current and torque waveforms of a five-phase permanent magnet synchronous motor with single-phase open circuit faults (speed 300 r / min); (a) Method 1; (b) Method 2; (c) Method 3. Detailed Implementation

[0071] The specific implementation methods and effects of this embodiment will be described in detail below with reference to the accompanying drawings.

[0072] Step 1, the detected feedback speed ω of the five-phase permanent magnet synchronous motor m By comparing the given rotational speed ω* with the actual rotational speed, the motor's speed error is obtained. A PI controller is then used to calculate the q-axis reference current i of the five-phase permanent magnet synchronous motor based on this speed error. qref i d For the d-axis current, due to the use of i d =0 control, d-axis reference current i of the five-phase permanent magnet synchronous motor dref Set to 0;

[0073] Step 2: When a single-phase open-circuit fault occurs in the five-phase permanent magnet synchronous motor, the corresponding z-axis reference current i can be determined according to the principles of minimum copper loss and equal copper loss, respectively. zref This enables control of the z-subspace of a five-phase permanent magnet synchronous motor after a single-phase open-circuit fault.

[0074] After a single-phase open-circuit fault occurs in a five-phase motor, the copper loss generated by the motor can be expressed as the sum of the copper losses generated by the dqz axis currents: R s (i2 d+i2 q+i2 z); where R s i is the phase resistance of the motor; x(x=d, q, z) represents the dqz axis current of the motor;

[0075] When the principle of minimizing copper loss is adopted (LJL), the z-axis reference current i zref It should be set as: ;

[0076] The five-phase current under the principle of minimizing copper loss is: ;

[0077] When the principle of equal copper loss (EJL) is adopted, it means that the magnitudes of the remaining phase currents are equal, i.e., the remaining phase currents satisfy: i B =-i D and i C =-i E z-axis reference current i zref It should be set as: ;

[0078] The five-phase current under the principle of equal copper loss is: ;

[0079] Where, iy xref (x=d, q, z, y=LJL, EJL) are the dqz axis reference currents under the principles of minimum copper loss and equal copper loss; iy x (x= A, B, C, D, E, y=LJL, EJL) are the phase currents under the principles of minimum copper loss and equal copper loss; I m θ represents the current amplitude of the motor under normal operating conditions. e The electric angle is the motor angle.

[0080] Step 3: Sample the A, B, C, D, and E phase currents i of the five-phase permanent magnet synchronous motor using a current sensor. A i B i C i D i E Based on the sampled current, the reduced-order matrix corresponding to the phase loss during an open-circuit fault is determined. Using the selected reduced-order matrix, a matrix transformation is performed on the sampled phase currents of the five-phase permanent magnet motor to obtain the dqz-axis current i fed back by the five-phase permanent magnet synchronous motor at the current time k during the fault. d (k), i q (k), i z (k);

[0081] Since the prediction of the z-axis reference voltage and the rolling selection of the value function in model predictive control require feedback of the dq z-axis current, the current i collected by the current sensor is used. A i B i C i D i EThe current i at time k is obtained by transforming the data using the Clarke matrix to the α-β axis, and then by transforming it using the Park matrix to the dqz axis. d (k), i q (k), i z (k):

[0082] ;

[0083] in, The missing Park transformation matrix for phase A. Here is the Clarke transformation matrix for phase A missing.

[0084] ,

[0085] ;

[0086] Where α = 2 / 5π, is the electrical angle difference between adjacent windings of the five-phase motor.

[0087] Step 4: Based on the z-axis current i at time k fed back by the five-phase permanent magnet synchronous motor. z (k) and reference current i zref The z-axis reference voltage U can be predicted using a deadbeat current prediction algorithm. zref ;

[0088] Step 4.1: Discretizing the motor's z-axis voltage equation yields the corresponding z-axis current prediction equation:

[0089] ;

[0090] Among them, L s For motor leakage inductance; T s i is the sampling time; z (k) represents the current at time k; i z (k+1) represents the current at time k+1; u z (k) represents the voltage at time k;

[0091] Step 4.2, based on the deadbeat current prediction method, the reference current i zref The current i at time k+1 z The reference voltage U can be predicted by (k+1). zref Therefore, by substituting the corresponding z-axis reference currents under the different principles of minimum and equal copper loss in step 2, we can obtain the z-axis reference voltages under the two principles: ;

[0092] Where uy zref (y=LJL, EJL) is the z-axis reference voltage under the principles of minimum copper loss and equal copper loss.

[0093] Step 5, based on the synthesized vector z-axis voltage U synz Tracking the z-axis reference voltage U zref Based on the principle of [the above], alternative variable virtual voltage vectors are constructed and the dynamically changing vector action time is determined;

[0094] Step 5.1: Based on the Clarke transformation matrix and switch states after the single-phase loss in Step 3, the vector distribution after single-phase open-circuit fault tolerance can be obtained, as shown in the attached figure. Figure 1 As shown, the vector distribution diagram contains the α-β subspace and the z subspace containing U1~U 14 There are a total of 14 valid vectors and U0, U 15 The two zero vectors, their magnitudes, and their switching states are shown in Table 1:

[0095] Table 1 shows the vector magnitude and switching status:

[0096] ;

[0097] Based on the distribution of vectors in the α-β subspace, the vectors are divided into [U0], [U9], [U8], and [U9]. 13 ]、[U 12 U 10 ]、[U 14 ,U4],[U6],[U7,U2],[U3,U5],[U1,U 11 ]、[U 15 The combination of ] and the combination of two voltage vectors, whose vector composition relationships in the α-β subspace and z subspace are as follows: ;

[0098] Among them, T opt1 T opt2 U represents the duration of action of vectors 1 and 2 in the combination; opt1z U opt2z U synz U represents the voltage components of vector 1, vector 2, and the composite vector in the z-subspace. opt1αβ U opt2αβ U synαβ Let U be the voltage components of vector 1, vector 2, and the composite vector in the α-β subspace, satisfying U opt1αβ =U opt1α +jU opt1β U opt2αβ =U opt2α +jU opt2β U synαβ =U synα +jU synβ ;

[0099] Step 5.2, based on the synthesized vector z-axis voltage U synz Tracking the z-axis reference voltage U zref Based on the principle of synthesizing variable virtual voltage vectors and determining the vector action time, considering the voltage vector distribution and spatial symmetry, individual vectors [U0], [U9], [U6], and [U... 15 The remaining combinations containing two voltage vectors are divided into two types;

[0100] Type 1 includes [U8, U 13 ]、[U 14 [U4], [U7, U2], [U1, U 11 ], with [U8, U 13 For example, voltage vectors U8 and U 13 The components in the α-β subspace are 0.4413U. dc e j60 / 180π 0.3245U dc e j46 / 180π The z-subspace components are -0.2351U. dc 0.3804U dc U dc This is the DC bus voltage, as shown in the attached diagram. Figure 2 As shown, according to the principle of vector composition, voltage vectors U8 and U... 13 The range of the vertex of the composite vector in the α-β subspace is the third side of the vector triangle formed by the two vectors. The range of the vertex of the composite vector in the z-subspace is the line segment formed by the vertices of the two vectors. This is to satisfy the z-axis voltage U of the composite vector. synz Tracking the z-axis reference voltage U zref We can obtain:

[0101] ;

[0102] Among them, T * opt1(U8) T * opt2(U13) For voltage vectors U8 and U 13 Duration of action;

[0103] Considering that the vector action time must be within the interval of 0 to 1, when it exceeds this interval, the action time is restricted to the boundaries of the interval, 0 and 1. Furthermore, if T... * opt1(U8) +T * opt2(U13) If the total duration of action exceeds 1, the following correction will be made:

[0104] ;

[0105] Among them, Topt1(U8) T opt2(U13) For voltage vectors U8 and U 13 The revised duration of action;

[0106] After obtaining the corrected vector action time, alternative variable virtual voltage vectors can be established:

[0107] ;

[0108] Among them, U synz(U8- U13) For voltage vectors U8 and U 13 The voltage component of the composite vector in the z-subspace; U synzαβ(U8- U13) For voltage vectors U8 and U 13 The voltage component of the composite vector in the α-β subspace;

[0109] Type 2 includes [U 12 U 10 ]、[U3、U5]、with[U8、U 13 For example, voltage vector U 12 U 10 The components in the α-β subspace are 0.6155U. dc e j90 / 180π 0.1453U dc e j90 / 180π The z-subspace components are 0.1453U. dc -0.6155U dc , roots as attached Figure 3 As shown, according to the principle of vector composition, the voltage vector U 12 U 10 The range of the vertex of the composite vector in the α-β subspace and z subspace is the line segment formed by the two vector vertices. This is to satisfy the z-axis voltage U of the composite vector. synz Tracking the z-axis reference voltage U zref We can obtain:

[0110] ;

[0111] Among them, T * opt1(U12) T * opt2(U10) Voltage vector U 12 U 10 Duration of action;

[0112] Considering that the vector action time must be within the interval of 0 to 1, when it exceeds this interval, the action time is restricted to the boundaries of the interval, 0 and 1. Furthermore, if T... * opt1(U12) +T * opt2(U10)If the total duration of action exceeds 1, the following correction will be made:

[0113] ;

[0114] Among them, T opt1(U12) T opt2(U10) Voltage vector U 12 U 10 The revised duration of action;

[0115] After obtaining the corrected vector action time, alternative variable virtual voltage vectors can be established:

[0116] ;

[0117] Among them, U synz(U12- U10) Voltage vector U 12 U 10 The voltage component of the composite vector in the z-subspace; U synzαβ(U12- U10) Voltage vector U 12 U 10 The voltage component of the composite vector in the α-β subspace;

[0118] Step 5.3, similarly, can be used to solve for the virtual voltage vector formed by combining the remaining voltage vectors. Adding this to the single effective voltage vector and the zero vector will constitute a set of alternative variable virtual voltage vectors, the vector distribution of which is shown in the attached figure. Figure 4 Let them be denoted as VV0~VV9, and their corresponding vector combinations are shown in Table 2:

[0119] Table 2 is the vector combination table:

[0120] ;

[0121] Step 6: Combining the five-phase motor discrete prediction model and the transformation matrix from the stationary coordinate system to the rotating coordinate system, substitute the candidate variable virtual voltage vector into the value function based on the dqz axis current error, and perform rolling optimization to find the optimal voltage vector and corresponding action time allocation.

[0122] Step 6.1: Discretizing the dqz axis voltage equations of the five-phase motor yields the corresponding dqz axis current prediction equations:

[0123] ;

[0124] Among them, L d L q T is the inductance of the dq axis of the motor; s i is the sampling time; d (k), i q (k) represents the current at time k; id (k+1),i q (k+1) represents the current at time k+1; u d (k), u q (k) represents the voltage at time k;

[0125] Step 6.2, substitute the α-β-z-axis components of the variable virtual voltage vector generated in step 5 into the values ​​from step 3. From the matrix, the corresponding dqz axis components can be obtained:

[0126] ;

[0127] Step 6.3, convert the dqz axis components U of the variable virtual voltage vector synd U synq U synz Substitute the voltage u at time k into the prediction equation in step 6.1 d (k), u q (k), u z (k) gives the current i at time k+1 of the candidate variable virtual voltage vector. d (k+1),i q (k+1),i z (k+1);

[0128] Step 6.4, the current i at time k+1 of the candidate variable virtual voltage vector. d (k+1),i q (k+1),i z Substituting (k+1) into the following value function based on the dqz axis current error, the optimal voltage vector and corresponding action time allocation can be obtained through rolling optimization: .

[0129] Step 7: Select the optimal voltage vector and its corresponding switching state, inputting it to the PWM module to obtain the switching signals for each phase. These switching signals are then input to the inverter to control the motor, achieving single-phase open-circuit fault-tolerant control of the five-phase permanent magnet synchronous motor. A complete control block diagram of this fault-tolerant control strategy is attached. Figure 5 As shown.

[0130] Appendix Figure 6 This invention relates to the switching of a five-phase permanent magnet synchronous motor from normal operation to single-phase open-circuit fault to fault-tolerant control. Regardless of whether the principle of minimizing copper losses or equalizing copper losses is met, the open-circuit fault-tolerant control proposed in this invention can reduce torque ripple caused by faults, and the torque performance is close to that under normal operating conditions. At the same time, it restores the sinusoidal nature of the healthy phase current, exhibiting excellent fault-tolerant performance. Figure 7 and attached Figure 8The torque-current waveforms of three methods—traditional virtual voltage vector, variable virtual voltage vector (under the principle of minimum copper loss), and variable virtual voltage vector (under the principle of equal copper loss)—were compared at 100 r / min and 300 r / min. In comparison, the variable virtual voltage vector method further enhances the suppression of z-axis harmonic current while ensuring the same torque performance, resulting in better sinusoidal phase current and achieving the goals of minimum copper loss and equal copper loss, respectively.

[0131] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0132] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for predictive fault-tolerant control of a five-phase permanent magnet synchronous motor using a single-phase open-circuit model with a variable virtual voltage vector, characterized in that, Includes the following steps: Step 1, the detected feedback speed ω of the five-phase permanent magnet synchronous motor m By comparing the given rotational speed ω* with the actual rotational speed, the motor's speed error is obtained. A PI controller is then used to calculate the q-axis reference current i of the five-phase permanent magnet synchronous motor based on this speed error. qref i d For the d-axis current, due to the use of i d =0 control, d-axis reference current i of the five-phase permanent magnet synchronous motor dref Set to 0; Step 2: When a single-phase open-circuit fault occurs in the five-phase permanent magnet synchronous motor, the corresponding z-axis reference current i can be determined according to the principles of minimum copper loss and equal copper loss, respectively. zref This enables control of the z-subspace of a five-phase permanent magnet synchronous motor after a single-phase open-circuit fault. Step 3: Sample the A, B, C, D, and E phase currents i of the five-phase permanent magnet synchronous motor using a current sensor. A i B i C i D i E Based on the sampled current, the reduced-order matrix corresponding to the phase loss during an open-circuit fault is determined. Using the selected reduced-order matrix, a matrix transformation is performed on the sampled phase currents of the five-phase permanent magnet motor to obtain the dqz-axis current i fed back by the five-phase permanent magnet synchronous motor at the current time k during the fault. d (k), i q (k), i z (k); Step 4: Based on the z-axis current i at time k fed back by the five-phase permanent magnet synchronous motor. z (k) and reference current i zref The z-axis reference voltage U can be predicted using a deadbeat current prediction algorithm. zref ; Step 5, based on the synthesized vector z-axis voltage U synz Tracking the z-axis reference voltage U zref Based on the principle of [the above], alternative variable virtual voltage vectors are constructed and the dynamically changing vector action time is determined; Step 6: Combining the five-phase motor discrete prediction model and the transformation matrix from the stationary coordinate system to the rotating coordinate system, substitute the candidate variable virtual voltage vector into the value function based on the dqz axis current error, and perform rolling optimization to find the optimal voltage vector and corresponding action time allocation. Step 7: Select the optimal voltage vector and input its corresponding switching state to the PWM module to obtain the switching signals of each phase. Input the obtained switching signals to the inverter to control the motor and realize the single-phase open-circuit fault-tolerant control of the five-phase permanent magnet synchronous motor.

2. The method for predictive fault-tolerant control of a single-phase open-circuit model of a five-phase permanent magnet synchronous motor using a variable virtual voltage vector, as described in claim 1, is characterized in that... The z-axis reference current i under the principle of minimum and equal copper loss described in step 2 zref as follows: After a single-phase open-circuit fault occurs in a five-phase motor, the copper loss generated by the motor can be expressed as the sum of the copper losses generated by the dqz axis currents: R s (i2 d+i2 q+i2 z); where R s i is the phase resistance of the motor; x (x=d, q, z) represents the dqz axis current of the motor; When the principle of minimizing copper loss is adopted (LJL), the z-axis reference current i zref It should be set as: ; The five-phase current under the principle of minimizing copper loss is: ; When the principle of equal copper loss (EJL) is adopted, it means that the magnitudes of the remaining phase currents are equal, i.e., the remaining phase currents satisfy: i B =-i D and i C =-i E z-axis reference current i zref It should be set as: ; The five-phase current under the principle of equal copper loss is: ; Where iy xref (x=d, q, z; y=LJL, EJL) are the dqz axis reference currents under the principles of minimum copper loss and equal copper loss; iy x (x= A, B, C, D, E; y=LJL, EJL) are the phase currents under the principles of minimum copper loss and equal copper loss; I m θ represents the current amplitude of the motor under normal operating conditions. e The electric angle is the motor angle.

3. The method for predictive fault-tolerant control of a five-phase permanent magnet synchronous motor using a variable virtual voltage vector single-phase open-circuit model according to claim 1, characterized in that, The calculation of the feedback dqz axis current mentioned in step 3 is as follows: Since the prediction of the z-axis reference voltage and the rolling selection of the value function in model predictive control require feedback of the dq z-axis current, the current i collected by the current sensor is used. A i B i C i D i E The current i at time k is obtained by transforming the data using the Clarke matrix to the α-β axis, and then by transforming it using the Park matrix to the dqz axis. d (k), i q (k), i z (k): ； in, The missing Park transformation matrix for phase A. Here is the Clarke transformation matrix with phase A missing. ； ； Where α = 2 / 5π, is the electrical angle difference between adjacent windings of the five-phase motor.

4. The method for predictive fault-tolerant control of a five-phase permanent magnet synchronous motor using a variable virtual voltage vector single-phase open-circuit model as described in claim 1, characterized in that, In step 4, the z-axis reference voltage u zref The calculation is as follows: Step 4.1: Discretizing the motor's z-axis voltage equation yields the corresponding z-axis current prediction equation: ； Among them, L s For motor leakage inductance; T s i is the sampling time; z (k) represents the current at time k; i z (k+1) represents the current at time k+1; u z (k) represents the voltage at time k; Step 4.2, based on the deadbeat current prediction method, the reference current i zref The current i at time k+1 z The reference voltage U can be predicted by (k+1). zref Therefore, by substituting the corresponding z-axis reference currents under the different principles of minimum and equal copper loss in step 2, we can obtain the z-axis reference voltages under the two principles: ; Where uy zref (y=LJL, EJL) is the z-axis reference voltage under the principles of minimum copper loss and equal copper loss.

5. The method for predictive fault-tolerant control of a single-phase open-circuit model of a five-phase permanent magnet synchronous motor using a variable virtual voltage vector, as described in claim 1, is characterized in that... The construction of the variable virtual voltage vector in step 5 is shown below: Step 5.1: Based on the Clarke transformation matrix and switch states after the single-phase loss in Step 3, the vector distribution after single-phase open-circuit fault tolerance can be obtained, including the α-β subspace and the z subspace containing U1~U 14 There are a total of 14 valid vectors and U0, U 15 Two zero vectors are divided into multiple combinations based on their distribution in the α-β subspace; Step 5.2, based on the synthesized vector z-axis voltage U synz Tracking the z-axis reference voltage U zref Based on the principle of synthesizing variable virtual voltage vectors, determining the action time of two vectors in each vector combination, and performing amplitude limiting and normalization correction on the action time, the α-β-z axis voltage components of the synthesized vector are calculated to construct alternative variable virtual voltage vectors. Step 5.3: Solve for the virtual voltage vector formed by the combination of the remaining voltage vectors. Add the single effective voltage vector and the zero vector to form a set of alternative variable virtual voltage vectors, denoted as VV0~VV9.

6. The method for predictive fault-tolerant control of a five-phase permanent magnet synchronous motor using a variable virtual voltage vector single-phase open-circuit model according to claim 1, characterized in that, The specific steps of voltage vector rolling optimization in step 6 are as follows: Step 6.1: Discretizing the dqz axis voltage equations of the five-phase motor yields the corresponding dqz axis current prediction equations: ； Among them, L d L q T is the inductance of the dq axis of the motor; s i is the sampling time; d (k), i q (k) represents the current at time k; i d (k+1),i q (k+1) represents the current at time k+1; u d (k), u q (k) represents the voltage at time k; Step 6.2, substitute the α-β-z-axis components of the variable virtual voltage vector generated in step 5 into the values ​​from step 3. From the matrix, the corresponding dqz axis components can be obtained: ; Step 6.3, convert the dqz axis components U of the variable virtual voltage vector. synd U synq U synz Substitute the voltage u at time k into the prediction equation in step 6.1 d (k), u q (k), u z (k) gives the current i at time k+1 of the candidate variable virtual voltage vector. d (k+1),i q (k+1),i z (k+1); Step 6.4, the current i at time k+1 of the candidate variable virtual voltage vector. d (k+1),i q (k+1),i z Substituting (k+1) into the following value function based on the dqz axis current error, the optimal voltage vector and corresponding action time allocation can be obtained through rolling optimization: .

Images