Low-complexity dual three-phase PMSM model prediction control method
By synthesizing virtual voltage vectors and designing switch sequences in dual three-phase PMSM motor control, avoiding inverse trigonometric function calculations, the problems of complexity and computational burden in traditional control methods are solved, and more efficient steady-state performance and simplified Design process.
Patent Information
- Application Number
- CN202410982426.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-22
- Publication Date
- 2024-11-08
AI Technical Summary
Traditional multiphase motor vector control has difficulty in setting current loop saturation and parameter, direct torque control has poor steady state performance, model prediction control calculation is complex and has a high computational burden, especially in the calculation of inverse trigonometric function on DSP28335 chip, it is necessary to check the table. Increased design difficulty.
The prediction control method of the low-complexity dual three-phase PMSM model is adopted. By synthesizing 24 virtual voltage vectors, the reference voltage vector is calculated using the principle of no-beat and prediction equations, the duty cycle calculation method is designed to avoid inverse trigonometric function calculation, and logic is used. The calculation determines the position of the reference voltage vector, and designs a switch sequence of 24 virtual voltage vectors combined with zero vectors, and uses triangle waves to generate a switch sequence.
The algorithm is simplified, the computational complexity is reduced, the modulation range is increased, the steady-state performance is improved, and the software is written in a low-cost microprocessor.
Smart Images

Figure CN118920937A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of multi-phase motor control, and in particular to a low-complexity dual three-phase PMSM model predictive control method. Background Art
[0002] Compared with three-phase motors, multi-phase motors have higher efficiency, better control performance and higher power density. They are suitable for applications with high performance and precision requirements, such as industrial machinery, aerospace and other fields. More control freedom increases the fault tolerance of multi-phase drives.
[0003] As one of the most mature control methods, vector control has been widely used in industry. However, vector control involves multiple current loops, which inevitably causes problems of current loop saturation and parameter setting. This is particularly evident in the vector control of multi-phase motors, because it usually involves multiple planes and requires the introduction of multiple PI controllers, which further increases the difficulty of parameter setting. Therefore, scholars have studied direct torque control with a simple structure. However, direct torque control has poor steady-state performance and requires the construction of a flux observer and the addition of a switch table, which undoubtedly increases the complexity of system design. Therefore, model prediction technology has gradually entered the field of vision of scholars.
[0004] Model predictive control is considered to be one of the most effective methods in the field of power electronic control because of its simple structure, good dynamic performance, multi-objective optimization and other advantages. It has been widely studied by scholars in the past decade. For the model predictive control technology of motor drive system, the current loop PI controller can be omitted, which simplifies the design difficulty. However, there are still some limitations for traditional model predictive control. The limited control of the centralized voltage vector will limit the improvement of control performance such as harmonic current and torque pulsation; and the model predictive control method using the deadbeat principle determines the position of the reference voltage vector by introducing inverse trigonometric function calculation. For the DSP28335 chip, both traditional software programming and MBD automatic code generation technology need to avoid inverse trigonometric function calculation, because the calculation of inverse trigonometric function usually needs to be realized in disguise by table lookup, which greatly increases the calculation burden and workload. In addition, the switching frequency of the traditional model predictive control method involving multiple vectors is usually not fixed, which also increases the workload of the underlying design. Therefore, how to improve the steady-state performance by adding extension sets, simplify the algorithm, reduce the computational burden, and design a switching sequence that is easier to implement has become an important step in the dual three-phase PMSM model predictive control method. Summary of the invention
[0005] In view of the shortcomings of the prior art, the present invention provides a low-complexity dual three-phase PMSM model predictive control method; first, 24 virtual voltage vectors are synthesized to increase the modulation range. Second, the reference voltage vector is obtained by combining the deadbeat principle with the prediction equation. Third, a duty cycle calculation method is designed to calculate the action time of the zero vector to change the amplitude of the optimal voltage vector, thereby improving the steady-state performance. Fourth, a U constant for the reference voltage is designed. α and U β The method of determining the position of the reference voltage vector by simple addition, subtraction and logical operations avoids the use of inverse trigonometric functions and reduces the computational burden and complexity. Finally, 24 virtual voltage vectors are designed to combine the switching sequence of the zero vector. 00 and V 77 The method of the present invention is simple in structure, small in calculation amount, increases the modulation range, and is easier to implement on low-cost microprocessors such as DSP28335.
[0006] A low-complexity dual three-phase PMSM model predictive control method comprises the following steps:
[0007] Step 1: With the goal of limiting the harmonics in the xy plane to zero, 24 virtual voltage vectors are synthesized to form an extended control set; which is divided into group G1, including virtual voltage vectors VV1 ~ VV 12 Virtual voltage vector, and G2 group, including VV E1 Virtual voltage vector~VV E12 Virtual voltage vector;
[0008] The specific synthesis method is as follows:
[0009] VV E1 The virtual voltage vector is V 44 , V 54 , V 65 and V 45 For synthesis, the action time ratios were designed to be: 0.366, 0134, 0.134 and 0.366;
[0010] VV1 virtual voltage vector is composed of V 44 and V 65 The synthesis and action time ratios are designed to be 0.732 and 0.268 respectively.
[0011] VV E2 The virtual voltage vector is V 44 , V 46 , V 65 and V 64For synthesis, the action time ratios are designed to be 0.366, 0134, 0.134 and 0.366 respectively.
[0012] The VV2 virtual voltage vector is composed of V 44 , V 64 and V 66 The synthesis and action time ratios are designed to be 0.268, 0.464 and 0.268 respectively.
[0013] VV E3 The virtual voltage vector is V 24 , V 64 , V 66 and V 46 For synthesis, the action time ratios are designed to be 0.366, 0134, 0.134 and 0.366 respectively.
[0014] The VV3 virtual voltage vector is composed of V 24 and V 66 The synthesis and action time ratios are designed to be 0.732 and 0.268 respectively.
[0015] VV E4 The virtual voltage vector is V 24 , V 26 , V 66 and V 62 For synthesis, the action time ratios are designed to be 0.366, 0134, 0.134 and 0.366 respectively.
[0016] The VV4 virtual voltage vector is composed of V 22 , V 26 and V 66 The synthesis and action time ratios are designed to be 0.268, 0.464 and 0.268 respectively.
[0017] VV E5 The virtual voltage vector is V 22 , V 62 , V 36 and V 26 For synthesis, the action time ratios are designed to be 0.366, 0134, 0.134 and 0.366 respectively.
[0018] The VV5 virtual voltage vector is composed of V 22 and V 36 The synthesis and action time ratios are designed to be 0.732 and 0.268 respectively.
[0019] VV E6 The virtual voltage vector is V 22 , V 23 , V 36 and V 32For synthesis, the action time ratios are designed to be 0.366, 0134, 0.134 and 0.366 respectively.
[0020] The VV6 virtual voltage vector is composed of V 22 , V 32 and V 33 The synthesis and action time ratios are designed to be 0.268, 0.464 and 0.268 respectively.
[0021] VV E7 The virtual voltage vector is V 12 , V 32 , V 33 and V 23 For synthesis, the action time ratios are designed to be 0.366, 0134, 0.134 and 0.366 respectively.
[0022] VV7 virtual voltage vector is composed of V 22 and V 33 The synthesis and action time ratios are designed to be 0.732 and 0.268 respectively.
[0023] VV E8 The virtual voltage vector is V 12 , V 13 , V 33 and V 31 For synthesis, the action time ratios are designed to be 0.366, 0134, 0.134 and 0.366 respectively.
[0024] The VV8 virtual voltage vector is composed of V 11 , V 13 and V 33 The synthesis and action time ratios are designed to be 0.268, 0.464 and 0.268 respectively.
[0025] VV E9 The virtual voltage vector is V 11 , V 31 , V 53 and V 13 For synthesis, the action time ratios are designed to be 0.366, 0134, 0.134 and 0.366 respectively.
[0026] The VV9 virtual voltage vector is composed of V 11 and V 53 The synthesis and action time ratios are designed to be 0.732 and 0.268 respectively.
[0027] VV E10 The virtual voltage vector is V 11 , V 15 , V 53 and V 51For synthesis, the action time ratios are designed to be 0.366, 0134, 0.134 and 0.366 respectively.
[0028] VV 10 The virtual voltage vector is V 11 , V 51 and V 55 The synthesis and action time ratios are designed to be 0.268, 0.464 and 0.268 respectively.
[0029] VV E11 The virtual voltage vector is V 15 , V 55 , V 51 and V 41 For synthesis, the action time ratios are designed to be 0.366, 0134, 0.134 and 0.366 respectively.
[0030] VV 11 The virtual voltage vector is V 41 and V 55 The synthesis and action time ratios are designed to be 0.732 and 0.268 respectively.
[0031] VV E12 The virtual voltage vector is V 41 , V 45 , V 55 and V 54 For synthesis, the action time ratios are designed to be 0.366, 0134, 0.134 and 0.366 respectively.
[0032] VV 12 The virtual voltage vector is V 44 , V 45 and V 55 The synthesis and action time ratios are designed to be 0.268, 0.464 and 0.268 respectively.
[0033] Step 2: Calculate the reference voltage vector position;
[0034] The forward Euler method is used to discretize the mathematical model of the dual three-phase permanent magnet synchronous motor, which is expressed as:
[0035]
[0036] Where x is a variable, k represents the kth sampling period, T s Represents the sampling period. Since the average value of the virtual voltage vector on the xy plane is zero, the mathematical model of the motor on the xy plane is not considered, and only the mathematical model on the dq axis is considered. The discrete expression of the dual three-phase PMSM considering the motor parameters is expressed as:
[0037]
[0038] In the formula, ψ f , R s and ω represent the permanent magnet flux, stator resistance and electrical angular velocity respectively, L d and L q are the d-axis and q-axis inductances, u d (k) and u d (k) are the d-axis and q-axis voltages applied at time k, respectively, i d (k) and i q (k) d-axis and q-axis currents at time k respectively;
[0039] The current prediction value at time k+1 is obtained by rearranging equations (1) and (2). A two-step prediction is used to compensate for the calculation delay. The current prediction at instant k+2 is:
[0040]
[0041] According to the deadbeat current prediction control principle, the following equation (4) should be satisfied:
[0042]
[0043] In the formula, and are the reference currents for the d-axis and q-axis respectively.
[0044] Substituting (4) into (3), the d-axis component of the reference voltage vector and the q-axis component It is expressed as the following formula:
[0045]
[0046] The reference voltage vector in (5) is transformed to the α–β plane by inverse Park transform:
[0047]
[0048] in is the component of the reference voltage vector on the α axis, is the component of the reference voltage vector on the β axis, the reference voltage vector is calculated, and θ is the rotor position angle.
[0049] Step 3: Determine the sector position of the reference voltage vector by designing the judgment coefficient and the pre-judgment function to avoid inverse trigonometric function calculation;
[0050] The judgment coefficient is given by u α * and u β *The value of the judgment coefficient obtained by adding and subtracting is compared with zero and provided as input for the sign function. In fact, the judgment coefficient is converted into a logical judgment. Specifically, the reference voltage vector u is calculated by formula (6): d * and u q * , the reference voltage vector is converted into the α-β plane in the dq plane through the inverse park transformation, that is, the u of the reference voltage vector is calculated α * and u β * , 12 judgment coefficients are introduced;
[0051] The pre-judgment function F is: the position of the reference voltage vector is determined by calculating the value of the pre-judgment function F and combining it with the designed truth table, as follows:
[0052] Introduce 12 judgment coefficients B0, B1, B2, B3, B4, B5, B6, B7, B8, B9, B 10 ,B 11 and B 12 ,make:
[0053]
[0054] Using B1~B 12 These 12 judgment coefficients are used to determine the 24 sectors and design the pre-judgment function F:
[0055]
[0056] Among them, sign(x) is the sign function, where x represents B1~B 12 , when x>0, sign(x)=1, when x<0, sign(x)=0, use the pre-judgment function value to determine the sector value;
[0057] Step 4: Design a truth table to determine the optimal voltage vector;
[0058] The position of the optimal voltage vector is determined by calculating F in step 3; each value of F corresponds to a unique sector value and a unique optimal voltage vector; a truth table for sector determination is designed:
[0059]
[0060]
[0061] The sector position of the reference voltage vector is determined by a truth table to determine the optimal voltage vector;
[0062] Step 5: Duty cycle calculation;
[0063] By introducing 24 virtual voltage vectors as the control set, the voltage on the xy axis of the virtual voltage vector can be regarded as zero, and the design cost function G is:
[0064]
[0065] Where, d opt is the duty cycle, i.e., the effective virtual voltage vector action time, VV αi 、VV βi are the components of the virtual voltage vector on the α and β axes respectively; G is taken as the dependent variable, d opt As the independent variable, the inverse processing is performed, and the specific calculation of the duty cycle is as follows:
[0066]
[0067] Step 6: Design the switching sequence, insert the zero vector V on both sides and in the middle of the voltage vector action sequence 00 and V 77 , ensuring that the number of switching actions in each control cycle is the same;
[0068] A switching sequence of 24 virtual voltage vectors combined with a zero vector is designed, as shown in the following table:
[0069]
[0070]
[0071] Step 7: Generate a switching sequence using a triangular carrier.
[0072] The virtual voltage vector after combining with the zero vector, the vector with completely symmetric first half cycle and second half cycle of the switching sequence is the virtual voltage vector of group G1, and the vector with incomplete symmetry first half cycle and second half cycle of the switching sequence is the virtual voltage vector of group G2. For the existing symmetrical and asymmetrical switching sequences, a triangular carrier is uniformly used to generate them, that is, in the increase and decrease mode of DSP, for the voltage vector of group G1, a reference value is required to be compared with the triangular carrier; for the voltage vector of group G2, two reference values are required to be compared with the triangular carrier. The specific implementation method is: in half of the control cycle, the values of registers CMPA and CMPB are updated;
[0073] The beneficial effects of adopting the above technical solution are:
[0074] The present invention provides a low-complexity dual three-phase PMSM model predictive control method. Compared with the traditional model predictive control method, the modulation range is increased by expanding the control set, and a duty cycle calculation method is designed to calculate the action time of the zero vector to change the amplitude of the optimal voltage vector, thereby achieving the purpose of improving the steady-state performance; in addition, u α * and u β * The method of determining the position of the reference voltage vector by simple addition, subtraction and logical operations avoids the use of complex numbers and inverse trigonometric functions in traditional model predictive control methods, reducing the computational burden and complexity; 24 virtual voltage vectors are designed in combination with the switching sequence of the zero vector. 00 and V 77 The method is arranged in the middle and on both sides of the virtual voltage vector, and the switch sequence of the proposed method is generated by using a triangular wave, so that the difficulty of software writing is reduced. The method of the present invention is more easily implemented on a low-cost microprocessor such as DSP28335. BRIEF DESCRIPTION OF THE DRAWINGS
[0075] Figure 1 A flow chart of a predictive control method provided by an embodiment of the present invention;
[0076] Figure 2 A voltage vector distribution diagram of the α-β subspace provided by an embodiment of the present invention;
[0077] Figure 3 A voltage vector distribution diagram of an xy subspace provided by an embodiment of the present invention;
[0078] Figure 4 A distribution diagram of virtual voltage vectors of an extended control set in an α-β subspace provided by an embodiment of the present invention;
[0079] Figure 5 VV provided by the embodiment of the present invention E1 ~VV E12 Switching sequences and generating schematics;
[0080] Figure 6 VV1, VV3, VV5, VV7, VV9 and VV provided in the embodiments of the present invention 11 Switch sequence diagram;
[0081] Figure 7 VV2, VV4, VV6, VV8, VV provided in the embodiment of the present invention 10 and VV 12 Switch sequence diagram; DETAILED DESCRIPTION
[0082] The specific implementation of the present invention is further described in detail below in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.
[0083] A low-complexity dual three-phase PMSM model predictive control method, such as Figure 1 As shown, the following steps are included:
[0084] Step 1: With the goal of limiting the harmonics in the xy plane to zero, 24 virtual voltage vectors are synthesized to form an extended control set; which is divided into group G1, including virtual voltage vectors VV1 ~ VV 12 Virtual voltage vector, and G2 group, including VV E1 Virtual voltage vector~VV E12 Virtual voltage vector; such as Figure 2 and Figure 3 As shown, several voltage vectors are found that are opposite in the xy plane and basically similar in the α-β plane direction. Here, multiple vectors are found to satisfy the volt-second balance principle to synthesize a virtual voltage vector, and the average value of the multiple vectors in the xy plane is zero.
[0085] The specific synthesis method is as follows:
[0086] VV E1 The virtual voltage vector is V 44 , V 54 , V 65 and V 45 For synthesis, the action time ratios were designed to be: 0.366, 0134, 0.134 and 0.366;
[0087] VV1 virtual voltage vector is composed of V 44 and V 65 The synthesis and action time ratios are designed to be 0.732 and 0.268 respectively.
[0088] VV E2 The virtual voltage vector is V 44 , V 46 , V 65 and V 64 For synthesis, the action time ratios are designed to be 0.366, 0134, 0.134 and 0.366 respectively.
[0089] The VV2 virtual voltage vector is composed of V 44 , V 64 and V 66 The synthesis and action time ratios are designed to be 0.268, 0.464 and 0.268 respectively.
[0090] VV E3 The virtual voltage vector is V 24 , V64 , V 66 and V 46 For synthesis, the action time ratios are designed to be 0.366, 0134, 0.134 and 0.366 respectively.
[0091] The VV3 virtual voltage vector is composed of V 24 and V 66 The synthesis and action time ratios are designed to be 0.732 and 0.268 respectively.
[0092] VV E4 The virtual voltage vector is composed of V 24 , V 26 , V 66 and V 62 For synthesis, the action time ratios are designed to be 0.366, 0134, 0.134 and 0.366 respectively.
[0093] The VV4 virtual voltage vector is composed of V 22 , V 26 and V 66 The synthesis and action time ratios are designed to be 0.268, 0.464 and 0.268 respectively.
[0094] VV E5 The virtual voltage vector is V 22 , V 62 , V 36 and V 26 For synthesis, the action time ratios are designed to be 0.366, 0134, 0.134 and 0.366 respectively.
[0095] The VV5 virtual voltage vector is composed of V 22 and V 36 The synthesis and action time ratios are designed to be 0.732 and 0.268 respectively.
[0096] VV E6 The virtual voltage vector is composed of V 22 , V 23 , V 36 and V 32 For synthesis, the action time ratios are designed to be 0.366, 0134, 0.134 and 0.366 respectively.
[0097] The VV6 virtual voltage vector is composed of V 22 , V 32 and V 33 The synthesis and action time ratios are designed to be 0.268, 0.464 and 0.268 respectively.
[0098] VV E7 The virtual voltage vector is V 12 , V 32, V 33 and V 23 For synthesis, the action time ratios are designed to be 0.366, 0134, 0.134 and 0.366 respectively.
[0099] VV7 virtual voltage vector is composed of V 22 and V 33 The synthesis and action time ratios are designed to be 0.732 and 0.268 respectively.
[0100] VV E8 The virtual voltage vector is V 12 , V 13 , V 33 and V 31 For synthesis, the action time ratios are designed to be 0.366, 0134, 0.134 and 0.366 respectively.
[0101] The VV8 virtual voltage vector is composed of V 11 , V 13 and V 33 The synthesis and action time ratios are designed to be 0.268, 0.464 and 0.268 respectively.
[0102] VV E9 The virtual voltage vector is V 11 , V 31 , V 53 and V 13 For synthesis, the action time ratios are designed to be 0.366, 0134, 0.134 and 0.366 respectively.
[0103] The VV9 virtual voltage vector is composed of V 11 and V 53 The synthesis and action time ratios are designed to be 0.732 and 0.268 respectively.
[0104] VV E10 The virtual voltage vector is V 11 , V 15 , V 53 and V 51 For synthesis, the action time ratios are designed to be 0.366, 0134, 0.134 and 0.366 respectively.
[0105] VV 10 The virtual voltage vector is V 11 , V 51 and V 55 The synthesis and action time ratios are designed to be 0.268, 0.464 and 0.268 respectively.
[0106] VV E11 The virtual voltage vector is V 15 , V 55, V 51 and V 41 For synthesis, the action time ratios are designed to be 0.366, 0134, 0.134 and 0.366 respectively.
[0107] VV 11 The virtual voltage vector is V 41 and V 55 The synthesis and action time ratios are designed to be 0.732 and 0.268 respectively.
[0108] VV E12 The virtual voltage vector is V 41 , V 45 , V 55 and V 54 For synthesis, the action time ratios are designed to be 0.366, 0134, 0.134 and 0.366 respectively.
[0109] VV 12 The virtual voltage vector is V 44 , V 45 and V 55 The synthesis and action time ratios are designed to be 0.268, 0.464 and 0.268 respectively.
[0110] There are a total of 24 virtual voltage vectors in the extended control set, which increases the modulation range and improves the control performance, while limiting the xy plane harmonics to zero and does not introduce more harmonic currents.
[0111] Step 2: Calculate the reference voltage vector position;
[0112] The forward Euler method is used to discretize the mathematical model of the dual three-phase permanent magnet synchronous motor, which is expressed as:
[0113]
[0114] Where x is a variable, which can represent current, voltage, etc. k represents the kth sampling period, T s Represents the sampling period. Since the average value of the virtual voltage vector on the xy plane is zero, the mathematical model of the motor on the xy plane is not considered, and only the mathematical model on the dq axis is considered. The discrete expression of the dual three-phase PMSM considering the motor parameters is expressed as:
[0115]
[0116] In the formula, ψ f , R s and ω represent the permanent magnet flux, stator resistance and electrical angular velocity respectively, L d and L q are the d-axis and q-axis inductances, ud (k) and u d (k) are the d-axis and q-axis voltages applied at time k, respectively, i d (k) and i q (k) d-axis and q-axis currents at time k respectively;
[0117] The current prediction value at time k+1 is obtained by rearranging equations (1) and (2). However, considering that the one-step delay between the expected vector and the actual vector introduced by digital processing will deteriorate the performance of the model predictive current control, in order to solve this problem, a two-step prediction is used to compensate for the calculation delay. The current prediction at instant k+2 is:
[0118]
[0119] According to the deadbeat current prediction control principle, the following equation (4) should be satisfied:
[0120]
[0121] In the formula, and are the reference currents for the d-axis and q-axis respectively.
[0122] Substituting (4) into (3), the d-axis component of the reference voltage vector and the q-axis component It is expressed as the following formula:
[0123]
[0124] The reference voltage vector in (5) is transformed to the α–β plane by inverse Park transform:
[0125]
[0126] in is the component of the reference voltage vector on the α axis, is the component of the reference voltage vector on the β axis, the reference voltage vector is calculated, and θ is the rotor position angle.
[0127] Step 3: Determine the sector position of the reference voltage vector by designing the judgment coefficient and the pre-judgment function to avoid inverse trigonometric function calculation;
[0128] like Figure 4 As shown, the entire α-β is divided into 24 sectors, and the angle of each sector is 15°. The judgment coefficient is obtained by u α * and u β *The value of the judgment coefficient obtained by adding and subtracting is compared with zero and provided as input for the sign function. In fact, the judgment coefficient is converted into a logical judgment. Specifically, the reference voltage vector u is calculated by formula (6): d * and u q * , the reference voltage vector is converted into the α-β plane in the dq plane through the inverse park transformation, that is, the u of the reference voltage vector is calculated α * and u β * , 12 judgment coefficients are introduced;
[0129] The pre-judgment function F is: by calculating the value of the pre-judgment function F and combining it with the designed truth table to judge the position of the reference voltage vector, this method does not need to use complex numbers and inverse trigonometric functions to perform sector judgment, as follows:
[0130] Introduce 12 judgment coefficients B0, B1, B2, B3, B4, B5, B6, B7, B8, B9, B 10 ,B 11 and B 12 ,make:
[0131]
[0132] Using B1~B 12 These 12 judgment coefficients are used to determine the 24 sectors and design the pre-judgment function F:
[0133]
[0134] Among them, sign(x) is the sign function, where x represents B1~B 12 , when x>0, sign(x)=1, when x<0, sign(x)=0, use the pre-judgment function value to determine the sector value;
[0135] Step 4: Design a truth table to determine the optimal voltage vector;
[0136] The position of the optimal voltage vector is determined by calculating F in step 3; each value of F corresponds to a unique sector value and a unique optimal voltage vector; a truth table for sector determination is designed:
[0137]
[0138]
[0139] The sector position of the reference voltage vector is determined by the truth table to determine the optimal voltage vector; the judgment of the 24 sectors is very accurate. In order to avoid the heavy computational burden caused by too many traversal prediction calculations of the voltage vector in the control set, the traditional model predictive control method uses inverse trigonometric function calculations for the judgment of the reference voltage vector position, which is not desirable in the actual software coding or MBD automatic code generation technology process.
[0140] Step 5: Duty cycle calculation;
[0141] In the model predictive control method, the role of the cost function is to adjust the control behavior of the system by comparing the predicted value and the reference value. By introducing 24 virtual voltage vectors as the control set, the voltage on the xy axis of the virtual voltage vector can be regarded as zero, so only the voltage in the α-β subspace needs to be considered in the cost function, and the voltage problem in the xy subspace does not need to be considered. Therefore, the harmonic weighting factor is eliminated. The design cost function G is:
[0142]
[0143] Where, d opt is the duty cycle, i.e., the effective virtual voltage vector action time, VV αi 、VV βi are the components of the virtual voltage vector on the α and β axes respectively; G is taken as the dependent variable, d opt As the independent variable, the inverse processing is performed, and the specific calculation of the duty cycle is as follows:
[0144]
[0145] Step 6: Design a simpler switching sequence, inserting zero vector V on both sides and in the middle of the voltage vector action sequence 00 and V 77 , ensuring that the number of switching actions in each control cycle is the same;
[0146] A switching sequence of 24 virtual voltage vectors combined with a zero vector is designed, as shown in the following table:
[0147]
[0148]
[0149] Step 7: Generate a switching sequence using a triangular carrier.
[0150] The virtual voltage vector after combining with the zero vector, the vector with completely symmetrical first half cycle and second half cycle of the switching sequence is the virtual voltage vector of group G1, and the vector with incomplete symmetrical first half cycle and second half cycle of the switching sequence is the virtual voltage vector of group G2. For the existing symmetrical and asymmetrical switching sequences, a triangular carrier is uniformly used to generate them, that is, in the increase and decrease mode of DSP, for the voltage vector of group G1, a reference value is required to be compared with the triangular carrier; for the voltage vector of group G2, two reference values are required to be compared with the triangular carrier. The specific implementation method is: in half of the control cycle, the values of registers CMPA and CMPB are updated; because V 00 The two sides of the fixed switch sequence, that is, the initial level is low, when the reference value is equal to the triangular wave for the first time, it is converted from low level to high level, and when the reference value is equal to the triangular carrier for the second time, it is converted from high level to low level. The switch sequence designed by the present invention can be generated by using the triangular carrier, which is easy to implement on a low-cost microprocessor such as DSP28335.
[0151] Two examples are given in this embodiment:
[0152] like Figure 5 As shown, the virtual voltage vector VV E1 As an example, the six-phase PWM wave generation principle is listed here, VV E1 ~VV E12 The switching sequence of can be obtained in the same way.
[0153] First, the calculation of the reference value modulation wave is given:
[0154]
[0155] Among them, m a 、m b 、m c and m d are the values of the four reference values, calculated as the ratio of the high level time of each phase in the six phases to the switching cycle; T 44 , T 54 , T 65 and T 45 V 44 、V 54 、V 65 and V 45 action time.
[0156] The proposed carrier wave is a triangular wave. Since the switching sequence of phase A is completely symmetrical, the reference value of phase A is consistent before and after two half switching cycles, which is (T 44 +T 54 +T 65 +T 45 ) / T s; For the asymmetric switching sequence of phase B, the reference value in the first half cycle is (-T 44 -T 54 -T 65 -T 45 ) / T s , in the second half of the cycle is 2*(T 65 -T 45 ) / T s ; For the asymmetric switching sequence of phase C, the reference value is also at T s / 2, the first half cycle is 2*(T 54 -T 44 ) / T s , the second half cycle is (-T 44 -T 54 -T 65 -T 45 ) / T s ; U phase is a completely symmetrical switching sequence, which is (T 44 +T 54 +T 65 +T 45 ) / T s , the V phase reference value is also constant in one cycle, which is (-T 44 -T 54 -T 65 -T 45 ) / T s , the reference value of phase W in the first half cycle is (-T 44 -T 54 -T 65 -T 45 ) / T s , the second half cycle is (T 44 +T 54 +T 65 +T 45 ) / T s ;
[0157] For T0, T 44 , T 54 , T 65 and T 45 The calculation of is as follows:
[0158]
[0159] like Figure 6 As shown, taking the virtual voltage vector VV1 as an example, the six-phase PWM wave generation principle is listed here. Here, for T0, T 44 and T 65 The calculation of is as follows:
[0160]
[0161] VV3, VV5, VV7, VV9 and VV 11 The switching sequence of can be obtained in the same way.
[0162] like Figure 7 As shown, taking the virtual voltage vector VV2 as an example, the six-phase PWM wave generation principle is listed here. 44 , T 64 and T 66 The calculation of is as follows:
[0163]
[0164] VV4, VV6, VV8, VV 10 and VV 12 The switching sequence of can be obtained in the same way.
[0165] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present disclosure is not limited to the technical solutions formed by a specific combination of the above-mentioned technical features, but should also cover other technical solutions formed by any combination of the above-mentioned technical features or their equivalent features without departing from the above-mentioned inventive concept. For example, the above-mentioned features are replaced with the technical features with similar functions disclosed in the embodiments of the present disclosure (but not limited to) to form a technical solution.
Claims
1. A low-complexity dual three-phase PMSM model predictive control method, characterized in that: The following steps are involved: Step 1: With the goal of limiting the harmonics in the xy plane to zero, 24 virtual voltage vectors are synthesized to form an extended control set; which is divided into group G1, including virtual voltage vectors VV1 ~ VV 12 Virtual voltage vector, and G2 group, including VV E1 Virtual voltage vector~VV E12 Virtual voltage vector; Step 2: Calculate the reference voltage vector position; Step 3: Determine the sector position of the reference voltage vector by designing the judgment coefficient and the pre-judgment function to avoid inverse trigonometric function calculation; Step 4: Design a truth table to determine the optimal voltage vector; Step 5: Duty cycle calculation; Step 6: Design the switching sequence, insert the zero vector V on both sides and in the middle of the voltage vector action sequence 00 and V 77 , ensuring that the number of switching actions in each control cycle is the same; Step 7: Generate the switching sequence by triangular carrier.
2. A low-complexity dual three-phase PMSM model predictive control method according to claim 1, characterized in that: The specific synthesis method in step 1 is as follows: VV E1 The virtual voltage vector is composed of V 44 , V 54 , V 65 and V 45 For synthesis, the action time ratios were designed to be: 0.366, 0134, 0.134 and 0.366; VV1 virtual voltage vector is composed of V 44 and V 65 For synthesis, the action time ratios were designed to be: 0.732, 0.268; VV E2 The virtual voltage vector is composed of V 44 , V 46 , V 65 and V 64 For synthesis, the action time ratios were designed to be: 0.366, 0134, 0.134 and 0.366; The VV2 virtual voltage vector is composed of V 44 , V 64 and V 66 For synthesis, the action time ratios were designed to be: 0.268, 0.464, and 0.268; VV E3 The virtual voltage vector is composed of V 24 , V 64 , V 66 and V 46 For synthesis, the action time ratios were designed to be: 0.366, 0134, 0.134 and 0.366; The VV3 virtual voltage vector is composed of V 24 and V 66 For synthesis, the action time ratios were designed to be: 0.732, 0.268; VV E4 The virtual voltage vector is composed of V 24 , V 26 , V 66 and V 62 For synthesis, the action time ratios were designed to be: 0.366, 0134, 0.134 and 0.366; The VV4 virtual voltage vector is composed of V 22 , V 26 and V 66 For synthesis, the action time ratios were designed to be: 0.268, 0.464, and 0.268; VV E5 The virtual voltage vector is composed of V 22 , V 62 , V 36 and V 26 For synthesis, the action time ratios were designed to be: 0.366, 0134, 0.134 and 0.366; The VV5 virtual voltage vector is composed of V 22 and V 36 For synthesis, the action time ratios were designed to be: 0.732, 0.268; VV E6 The virtual voltage vector is composed of V 22 , V 23 , V 36 and V 32 For synthesis, the action time ratios were designed to be: 0.366, 0134, 0.134 and 0.366; The VV6 virtual voltage vector is composed of V 22 , V 32 and V 33 For synthesis, the action time ratios were designed to be: 0.268, 0.464, and 0.268; VV E7 The virtual voltage vector is composed of V 12 , V 32 , V 33 and V 23 For synthesis, the action time ratios were designed to be: 0.366, 0134, 0.134 and 0.366; VV7 virtual voltage vector is composed of V 22 and V 33 For synthesis, the action time ratios were designed to be: 0.732, 0.268; VV E8 The virtual voltage vector is composed of V 12 , V 13 , V 33 and V 31 For synthesis, the action time ratios were designed to be: 0.366, 0134, 0.134 and 0.366; The VV8 virtual voltage vector is composed of V 11 , V 13 and V 33 For synthesis, the action time ratios were designed to be: 0.268, 0.464, and 0.268; VV E9 The virtual voltage vector is composed of V 11 , V 31 , V 53 and V 13 For synthesis, the action time ratios were designed to be: 0.366, 0134, 0.134 and 0.366; The VV9 virtual voltage vector is composed of V 11 and V 53 For synthesis, the action time ratios were designed to be: 0.732, 0.268; VV E10 The virtual voltage vector is composed of V 11 , V 15 , V 53 and V 51 For synthesis, the action time ratios were designed to be: 0.366, 0134, 0.134 and 0.366; VV 10 The virtual voltage vector is composed of V 11 , V 51 and V 55 For synthesis, the action time ratios were designed to be: 0.268, 0.464, and 0.268; VV E11 The virtual voltage vector is composed of V 15 , V 55 , V 51 and V 41 For synthesis, the action time ratios were designed to be: 0.366, 0134, 0.134 and 0.366; VV 11 The virtual voltage vector is composed of V 41 and V 55 For synthesis, the action time ratios were designed to be: 0.732, 0.268; VV E12 The virtual voltage vector is composed of V 41 , V 45 , V 55 and V 54 For synthesis, the action time ratios were designed to be: 0.366, 0134, 0.134 and 0.366; VV 12 The virtual voltage vector is composed of V 44 , V 45 and V 55 The synthesis and action time ratios are designed to be 0.268, 0.464 and 0.268 respectively.
3. A low-complexity dual three-phase PMSM model predictive control method according to claim 1, characterized in that: The step 2 is specifically: discretizing the mathematical model of the dual three-phase permanent magnet synchronous motor using the forward Euler method, which is expressed as: Where x is a variable, k represents the kth sampling period, T s Represents the sampling period. Since the average value of the virtual voltage vector on the xy plane is zero, the mathematical model of the motor on the xy plane is not considered, and only the mathematical model on the dq axis is considered. The discrete expression of the dual three-phase PMSM considering the motor parameters is expressed as: In the formula, ψ f , R s and ω represent the permanent magnet flux, stator resistance and electrical angular velocity respectively, L d and L q are the d-axis and q-axis inductances, u d (k) and u d (k) are the d-axis and q-axis voltages applied at time k, respectively, i d (k) and i q (k) d-axis and q-axis currents at time k respectively; The current prediction value at time k+1 is obtained by rearranging equations (1) and (2). A two-step prediction is used to compensate for the calculation delay. The current prediction at instant k+2 is: According to the deadbeat current prediction control principle, the following equation (4) should be satisfied: In the formula, and are the reference currents of the d-axis and q-axis respectively; Substituting (4) into (3), the d-axis component of the reference voltage vector and the q-axis component It is expressed as the following formula: The reference voltage vector in (5) is transformed to the α–β plane by inverse Park transform: in is the component of the reference voltage vector on the α axis, is the component of the reference voltage vector on the β axis, and is calculated to be The reference voltage vector, θ is the rotor position angle.
4. A low-complexity dual three-phase PMSM model predictive control method according to claim 1, characterized in that: The judgment coefficient in step 3 is u α * and u β * The value of the judgment coefficient obtained by adding and subtracting is compared with zero and provided as input for the sign function. In fact, the judgment coefficient is converted into a logical judgment. Specifically, the reference voltage vector u is calculated by formula (6): d * and u q * , the reference voltage vector is converted into the α-β plane in the dq plane through the inverse park transformation, that is, the u of the reference voltage vector is calculated α * and u β * , 12 judgment coefficients are introduced; The pre-judgment function F is: the position of the reference voltage vector is determined by calculating the value of the pre-judgment function F and combining it with the designed truth table, as follows: Introduce 12 judgment coefficients B0, B1, B2, B3, B4, B5, B6, B7, B8, B9, B 10 ,B 11 and B 12 ,make: Using B1~B 12 These 12 judgment coefficients are used to determine the 24 sectors and design the pre-judgment function F: Among them, sign(x) is the sign function, where x represents B1~B 12 When x>0, sign(x)=1, when x<0, sign(x)=0, and the sector value is determined by using the pre-judgment function value.
5. A low-complexity dual three-phase PMSM model predictive control method according to claim 4, characterized in that: The step 4 is specifically as follows: the position of the optimal voltage vector is determined by calculating F in step 3; each value of F corresponds to a unique sector value and a unique optimal voltage vector; a truth table for sector determination is designed: The sector position of the reference voltage vector is determined through a truth table to determine the optimal voltage vector.
6. A low-complexity dual three-phase PMSM model predictive control method according to claim 1, characterized in that: The step 5 is specifically as follows: by introducing 24 virtual voltage vectors as the control set, the voltage on the xy axis of the virtual voltage vector can be regarded as zero, and the design cost function G is: Where, d opt is the duty cycle, i.e., the effective virtual voltage vector action time, VV αi 、VV βi are the components of the virtual voltage vector on the α and β axes respectively; G is taken as the dependent variable, d opt As the independent variable, the inverse processing is performed, and the specific calculation of the duty cycle is as follows:
7. A low-complexity dual three-phase PMSM model predictive control method according to claim 1, characterized in that: The step 6 is specifically as follows: designing a switching sequence of 24 virtual voltage vectors combined with a zero vector, as shown in the following table:
8. A low-complexity dual three-phase PMSM model predictive control method according to claim 1, characterized in that: The step 7 is specifically as follows: after combining with the zero vector, the virtual voltage vector is a G1 group virtual voltage vector whose first half cycle and second half cycle of the switching sequence are completely symmetrical, and the vector whose first half cycle and second half cycle of the switching sequence are not completely symmetrical is a G2 group virtual voltage vector; for the existing symmetrical and asymmetrical switching sequences, a triangular carrier is uniformly used to generate them, that is, in the increase and decrease mode of the DSP, for the G1 group voltage vector, a reference value is required to be compared with the triangular carrier; for the G2 group voltage vector, two reference values are required to be compared with the triangular carrier, and the specific implementation method is: at half of the control cycle, the values of registers CMPA and CMPB are updated.
