Fast modulation type model prediction control method applied to three-level inverter

Abstract

The invention provides a fast modulation type model prediction control method applied to a three-level inverter, and relates to the technical field of inverter optimization control, and the method comprises the steps: carrying out the delay compensation of a converted reference current based on a delay compensation strategy, and combining a converted output current with an optimal voltage vector of a previous control period, thereby achieving the optimal control of the three-level inverter. Therefore, the output current at the next moment is accurately estimated. On the basis, an optimal large sector and an optimal small sector are quickly positioned by introducing a hierarchical query mechanism of a mapping table, an initial optimal voltage vector is screened by utilizing an upper and lower capacitor voltage relationship and combining a cost function minimum principle, then the other two voltage vectors in the small sector are complemented, duty ratio calculation and vector synthesis are completed, and the optimal voltage vector is obtained. Operation efficiency is greatly improved, accuracy and high efficiency of sector positioning and vector selection are ensured through a layered query mechanism of the mapping table, and control precision equivalent to that of a traditional method is maintained while calculation burden is effectively relieved.

Description

Technical Field

[0001] This invention relates to the field of inverter optimization control technology, and in particular to a fast modulation model predictive control method applied to three-level inverters. Background Technology

[0002] Modulated Model Predictive Control (MMPC), as an advanced control strategy, has received widespread attention in the field of power electronic converters, especially in inverter control, in recent years. It cleverly combines the dynamic optimization capabilities of model predictive control with the fixed switching frequency advantage of modulation techniques, thereby achieving excellent control performance. However, when applying the traditional MMPC strategy to three-level inverters, its computational burden becomes particularly prominent.

[0003] Compared to two-level inverters, three-level inverters have a significantly increased number of voltage vector states. In traditional three-level MMPC three-vector strategies, determining the optimal voltage vector combination requires a comprehensive evaluation of all possible voltage vectors or sectors. For example, one existing method evaluates all 24 sub-sectors to select the optimal voltage vector combination. While this method offers high control accuracy, it involves enormous computational demands, placing extremely high demands on the controller's processing power and limiting its application in situations requiring high control frequencies or low-cost controllers.

[0004] To reduce computational burden, researchers have proposed several voltage vector pre-selection methods. For example, a two-stage pre-selection method has been proposed, first evaluating six large sectors, and then evaluating the intermediate vectors of four smaller sectors to quickly locate the optimal smaller sector. This method reduces computational complexity to some extent by decreasing the number of vectors requiring detailed evaluation. However, while simplifying the computational process, such pre-selection methods may also introduce the risk of decreased control performance, as the pre-selection process may exclude the actually optimal voltage vector, resulting in a non-globally optimal control effect, i.e., the so-called "suboptimal" problem. In addition, some researchers have proposed the concept of "zero suboptimal," evaluating performance loss by comparing the difference in cost functions between traditional and simplified methods. This reflects the industry's focus on maintaining consistent control performance while reducing computational burden.

[0005] Therefore, in the field of modulation model predictive control research, how to significantly reduce the computational complexity of the MMPC strategy for three-level inverters while ensuring that its control performance is not inferior to traditional, computationally intensive methods has become a key problem that urgently needs to be solved. Summary of the Invention

[0006] To address the aforementioned problems in the background technology, this invention provides a fast modulation model predictive control method for three-level inverters, which reduces the computational burden of the traditional three-vector method in the vector preselection stage and the entire control algorithm for three-level inverters, while ensuring control performance consistent with the traditional three-vector method. That is, it effectively reduces the computational burden while maintaining control accuracy comparable to the traditional method.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A fast modulation model predictive control method for a three-level inverter includes: S1. During the current control cycle, acquiring the three-phase reference current and the three-phase output current of the three-level inverter, and performing coordinate transformations on them respectively, while simultaneously acquiring the upper and lower capacitor voltages on the DC side of the three-level inverter; S2. Based on a delay compensation strategy, performing delay compensation on the transformed reference current, and simultaneously predicting the output current at the next moment by combining the transformed output current with the optimal voltage vector obtained in the previous control cycle; S3. Based on the predicted output current at the next moment, pre-selecting candidate voltage vectors and calculating the corresponding cost functions, and gradually determining the output current by querying a mapping table based on the relationship between the various cost functions. S4. Using the optimal large sector and optimal small sector, select an initial optimal voltage vector within the optimal small sector based on the principle of minimizing the cost function; S5. Using the optimal large sector and optimal small sector to look up the mapping table, complete the remaining two initial optimal voltage vectors within the optimal small sector, and calculate the corresponding duty cycle based on the cost functions of the three initial optimal voltage vectors; S6. Based on the voltage relationship between the upper and lower capacitors, look up the mapping table, and combine the three initial optimal voltage vectors and the duty cycle to obtain the ultimate optimal voltage vector for the current control cycle, which is used for output current prediction in the next control cycle. At the same time, control pulses are generated to control the three-level inverter; where each cost function is obtained through modulation model prediction calculation.

[0008] This invention provides a preferred embodiment, wherein S3 specifically includes: S31. Based on the predicted output current at the next moment, three candidate voltage vectors are pre-selected from all voltage vectors of the three-level inverter, and the predicted output current and corresponding cost function applied to the two forward moments are calculated respectively; S32. According to the magnitude relationship of the three cost functions, a predefined first mapping table is consulted to determine the optimal large sector number, the minimum cost function of the first stage, and the other two candidate voltage vectors to be evaluated within the optimal large sector and their cost functions; S33. The cost functions of the other two candidate voltage vectors to be evaluated are compared with the minimum cost function of the first stage to determine the minimum cost function of the second stage, and the predefined second mapping table is consulted to determine the sixth candidate voltage vector to be evaluated and its cost function; S34. The cost function of the sixth candidate voltage vector to be evaluated is compared with the minimum cost function of the second stage, and the second mapping table is consulted according to the comparison result to determine the optimal small sector number and one of the initial optimal voltage vectors within the optimal small sector and its cost function. Section S3 describes a fast sector localization method that evaluates only six voltage vectors to obtain the optimal small sector. Compared to the inefficient search of traditional methods that require traversing all vectors, this method significantly improves computational efficiency and reduces algorithm complexity, making it particularly suitable for real-time control requirements in high-frequency PWM modulation scenarios. This strategy effectively narrows the search range through a staged filtering mechanism, avoiding redundant calculations caused by full vector enumeration, and achieving fast response while ensuring control accuracy.

[0009] This invention provides a preferred solution, with step S4 specifically including: S41. Based on the optimal large sector number and the optimal small sector number, query a predefined third mapping table (Table 3) to obtain the other two initial optimal voltage vectors and their cost functions within the optimal small sector; wherein, the third mapping table contains the mapping relationship between the optimal large sector number and the optimal small sector number, and the three initial optimal voltage vectors within the optimal small sector; S42. Based on the cost functions of the three voltage vectors in the optimal small sector, calculate the corresponding duty cycles respectively. Step S4 mainly describes the method for calculating the optimal voltage vector and its duty cycle. This method requires evaluating the cost functions of the other two voltage vectors, i.e., the other two initial optimal voltage vectors within the optimal small sector. Therefore, the total calculation process only requires evaluating 8 voltage vectors to obtain the optimal voltage vector. Compared with the traditional model predictive control scheme that requires traversing all 7 basic voltage vectors, this significantly reduces the amount of computation and improves the real-time performance of the control.

[0010] This invention provides a preferred solution, with S5 specifically including: S51. Based on the magnitude relationship of the upper and lower capacitor voltages, a predefined fourth mapping table is consulted to select three optimal voltage vectors from the voltage vector pairs with redundancy characteristics, forming an optimal voltage vector combination; S52. The three-phase duty cycle and the ultimate optimal voltage vector of the current control cycle are calculated according to the duty cycle corresponding to the selected three optimal voltage vector combinations for use in predicting the output current of the next control cycle; S53. Control pulses are generated according to the three-phase duty cycle to control the three-level inverter. This solution transforms the complex cost function evaluation into efficient logical judgment through a four-stage mapping table lookup mechanism, significantly reducing the computational load of the algorithm. By combining the magnitude relationship of the upper and lower capacitor voltages and fully considering the capacitor voltage balance factor, the optimal voltage vector is further screened, thereby reducing capacitor voltage fluctuations and ensuring the balance and stability of the output voltage. The entire control process does not require real-time solving of multivariable equations, making it suitable for motor drive systems with high dynamic performance requirements, significantly improving response speed and system efficiency while ensuring control accuracy. By employing a hierarchical discrimination mechanism based on a pre-defined mapping table, the system only needs to perform a finite number of logical lookups and comparisons within each control cycle to quickly locate the optimal voltage vector and accurately calculate the duty cycle. This method simplifies the complex global cost function optimization process in traditional model predictive control into staged table lookups and conditional judgments, significantly reducing the computational burden on the controller.

[0011] This invention provides a preferred embodiment where, in step S52, the ultimate optimal voltage vector is obtained by multiplying each of the three optimal voltage vector combinations by its corresponding duty cycle and then summing the results. This calculation method achieves accurate synthesis of the output voltage through linear combination, ensuring a smooth and continuous vector switching process and effectively reducing the current harmonic distortion rate.

[0012] Compared with the prior art, the present invention has the following beneficial technical effects: S1. During the current control cycle, acquire the three-phase reference current and the three-phase output current of the three-level inverter, and perform coordinate transformations on each. Simultaneously, acquire the upper and lower capacitor voltages on the DC side of the three-level inverter. S2. Based on a delay compensation strategy, perform delay compensation on the transformed reference current. Simultaneously, combine the transformed output current with the optimal voltage vector obtained in the previous control cycle to predict the output current at the next moment. S3. Based on the predicted output current at the next moment, pre-select candidate voltage vectors and calculate the corresponding cost functions. Based on the relationships between the various cost functions, gradually determine the optimal large sector and optimal small sector by querying a mapping table. S4. Using the optimal large sector and optimal small sector to look up the mapping table, complete the other two initial optimal voltage vectors in the optimal small sector, and calculate the corresponding duty cycle by combining the cost functions of the three initial optimal voltage vectors; S5. Based on the voltage relationship between the upper and lower capacitors, look up the mapping table, and combine the three initial optimal voltage vectors and the duty cycle to obtain the ultimate optimal voltage vector of the current control cycle for use in predicting the output current of the next control cycle, and generate control pulses to control the three-level inverter; wherein, each cost function is obtained by prediction calculation through a modulation model.

[0013] This invention first uses a delay compensation strategy to compensate for the delay in the transformed reference current, and then combines the transformed output current with the optimal voltage vector from the previous control cycle to accurately estimate the output current at the next moment. Based on this, a hierarchical lookup mechanism using a mapping table is introduced to quickly locate the optimal large and small sectors. The initial optimal voltage vector is then selected using the relationship between the upper and lower capacitor voltages and the principle of minimizing the cost function. The remaining two voltage vectors within the small sector are then completed, fulfilling the duty cycle calculation and vector synthesis. This significantly reduces computational complexity while improving the system's dynamic response accuracy. This method effectively avoids the high computational burden of traversing all voltage vectors in traditional model predictive control, greatly improving computational efficiency and making it suitable for three-level inverter control scenarios with high switching frequencies. Simultaneously, the hierarchical lookup mechanism of the mapping table ensures the accuracy of sector location and vector selection, enhancing system stability under load changes or non-ideal operating conditions. Therefore, this method can reduce the computational burden of the traditional three-vector method for three-level inverters in the vector pre-selection stage and the entire control algorithm, while maintaining control performance consistent with the traditional three-vector method, effectively reducing computational burden while maintaining control accuracy comparable to the traditional method.

[0014] This invention enables rapid sector localization through step S3, obtaining the optimal small sector by evaluating only 6 voltage vectors; step S4 calculates the optimal voltage vector and its duty cycle, and evaluates the cost function of the other 2 initial voltage vectors. Therefore, the total calculation process only requires evaluating 8 voltage vectors to obtain the optimal voltage vector. Compared with the traditional method, which requires evaluating all 24 small sectors to select the optimal voltage vector, this invention significantly reduces computational complexity. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0016] Figure 1 A flowchart illustrating the steps of a fast modulation model predictive control method for a three-level inverter provided in a specific embodiment of the present invention; Figure 2 A flowchart of a fast modulation model predictive control method for a three-level inverter, provided as a specific embodiment of the present invention; Figure 3 The overall control block diagram of a fast modulation model predictive control method for a three-level inverter provided in a specific embodiment of the present invention is shown. Figure 4 The diagram showing the large and small sector numbering in a fast modulation model predictive control method for three-level inverters provided in a specific embodiment of the present invention; Figure 5 The three-phase current waveforms and current harmonic values ​​of the fast modulation model predictive control method (proposed method) for a three-level inverter provided by a specific embodiment of the present invention are compared in the SIMULINK simulation environment when the reference current increases from 10 A to 15 A, using both traditional methods and this invention. Figure 6 The DC bus capacitor current waveforms are compared between the conventional method and the fast modulation model predictive control method (proposed method) for three-level inverters provided by a specific embodiment of the present invention in the SIMULINK simulation environment when the reference current increases from 10 A to 15 A. Figure 7 The small sector numbering values ​​are compared between the conventional method and the fast modulation model predictive control method (proposed method) for three-level inverters provided by a specific embodiment of the present invention in the SIMULINK simulation environment when the reference current increases from 10 A to 15 A. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Please refer to Figure 1 In one optional implementation, a fast modulation model predictive control method for a three-level inverter is provided, which is mainly implemented through the following steps: S1. During the current control cycle, the three-phase reference current and the three-phase output current of the three-level inverter are collected and coordinate transformations are performed respectively. At the same time, the voltages of the upper and lower capacitors on the DC side of the three-level inverter are collected. S2. Based on the delay compensation strategy, the transformed reference current is delayed and compensated. At the same time, the transformed output current is combined with the optimal voltage vector obtained in the previous control cycle to predict the output current at the next moment. S3. Based on the predicted output current at the next moment, pre-select candidate voltage vectors and calculate the corresponding cost functions. Based on the relationship between each cost function, determine the optimal large sector and the optimal small sector step by step by querying the mapping table, and select an initial optimal voltage vector in the optimal small sector according to the principle of minimizing the cost function. S4. Use the optimal large sector and optimal small sector to look up the mapping table, complete the other two initial optimal voltage vectors in the optimal small sector, and calculate the corresponding duty cycle by combining the cost function of the three initial optimal voltage vectors. S5. Based on the voltage relationship between the upper and lower capacitors, look up the mapping table, combine the three initial optimal voltage vectors and the duty cycle, obtain the ultimate optimal voltage vector of the current control cycle for the next control cycle to predict the output current, and generate control pulses to control the three-level inverter. Each cost function is obtained through prediction and calculation using a modulation model.

[0019] Please refer to Figure 2 and Figure 3 Based on the above implementation method, a more preferred and detailed implementation method is given as follows: S1. Within the current control cycle (i.e., the current sampling moment; one control cycle corresponds to one sampling moment), the three-phase reference current and the three-phase output current of the three-level inverter are acquired, and coordinate transformations are performed on them respectively. Simultaneously, the upper and lower capacitor voltages on the DC side of the three-level inverter are acquired. In a preferred embodiment, the coordinate transformation specifically transforms the reference current and output current into a two-phase rotating coordinate system (α-β coordinate system). More specifically, based on the given three-phase reference current...i abc * ( k The reference current in the α-β coordinate system is obtained by coordinate transformation. i αβ * ( k Sampling output current i abc ( k The output current in the α-β coordinate system is obtained by coordinate transformation. i αβ ( k ).

[0020] S2. Based on a delay compensation strategy, the transformed reference current is delayed and compensated. Simultaneously, the transformed output current is combined with the optimal voltage vector obtained in the previous control cycle to predict the output current at the next moment. In a preferred embodiment, the transformed reference current is delayed and compensated in two forward steps to obtain the reference current at the two forward moments. In another preferred embodiment, the output current at the next moment is also calculated based on the equivalent filter inductance and equivalent resistance on the output side of the three-level inverter, as well as the control cycle prediction. More specifically, based on the reference current obtained in step S1... i αβ * ( k The delay compensation method is calculated using a two-step forward prediction approach. k Reference current at +2 time i αβ * ( k +2): (1) in T s Indicates the control cycle. ω =2π f , f It is the frequency of the reference current.

[0021] Then, based on the output current obtained in step S1 i αβ ( k ),predict k Output current at time +1 i αβ ( k +1): (2) in V oldThese represent the optimal voltage vector (i.e., the ultimate optimal voltage vector) output by the inverter in the previous control cycle. L , R These represent the equivalent filter inductance and equivalent resistance on the inverter output side, respectively.

[0022] S3. Based on the predicted output current at the next moment, pre-select candidate voltage vectors and calculate the corresponding cost functions. Based on the relationship between the various cost functions, determine the optimal large sector and the optimal small sector step by step by querying the mapping table, and select an initial optimal voltage vector within the optimal small sector according to the principle of minimizing the cost function. In a preferred embodiment, S3 is specifically implemented through the following steps: S31. Based on the predicted output current at the next moment, three candidate voltage vectors are pre-selected from all voltage vectors of the three-level inverter, and the predicted output current and corresponding cost function applied to each of the two forward moments are calculated respectively. Specifically, in this sub-step, based on the output current obtained in step S2... i αβ ( k +1), calculate Figure 4 Predicted output currents of voltage vectors 18, 22, and 26 (candidate voltage vectors) i αβx ( k +2) and cost function G x { x ∈1,2,3}: (3) (4) in V i ( i (∈18,22,26) represents the candidate voltage vector.

[0023] S32. Based on the magnitude relationship of the three cost functions, query the predefined first mapping table (Table 1: Large Sector Selection Table) to determine the optimal large sector number, the minimum cost function for the first stage, and the other two candidate voltage vectors to be evaluated within the optimal large sector and their cost functions. Specifically, in this sub-step, based on the cost function from step S31... G 1. G 2. G 3 and Table 1 yielded Figure 4 Optimal large sector numbering S big And define the minimum cost function for the first stage. G min_3 for G 1. G 2 and GThe minimum value in 3. Calculate the cost function value of the voltage vector under different sectors in Table 1 according to equations (3) and (4). G 4 and G 5. That is, the other two candidate voltage vectors to be evaluated within the optimal large sector and their cost functions.

[0024] Table 1: Large Sector Selection Table

[0025] S33. Compare the cost functions of the other two candidate voltage vectors to be evaluated with the minimum cost function of the first stage to determine the minimum cost function of the second stage. Then, query the predefined second mapping table (Table 2: Small Sector Selection Table) to determine the sixth candidate voltage vector to be evaluated and its cost function. Specifically, this sub-step involves further comparison... G min_3 , G 4. G 5. Obtain the minimum cost function value for the second stage. G min_5 = G 4. Obtain the 6th voltage vector according to Table 2. V 19 And calculate its cost function value. G 6.

[0026] Table 2: Small Sector Selection Table

[0027] S34. Compare the cost function of the sixth candidate voltage vector to be evaluated with the minimum cost function of the second stage. Based on the comparison result, query the second mapping table (Table 2: Small Sector Selection Table) to determine the optimal small sector number and one of the initial optimal voltage vectors and its cost function within that optimal small sector. Specifically, this sub-step further compares... G min_5 ( G min_5 = G 4) and G 6. Through S big =2、 G min_5 > G 6. Query Table 2 to obtain the optimal small sector number. S small =3. Voltage Vector V m3 = V 19 and its cost function value G best = G6. That is, in this embodiment, the minimum cost function in the second stage G best for G 6. V m3 This is one of the initial optimal voltage vectors within the optimal small sector.

[0028] In a preferred embodiment, the first mapping table (Table 1: Large Sector Selection Table) contains the mapping relationship between the magnitude of the cost functions of the three pre-selected candidate voltage vectors and the first-stage minimum cost function, the optimal large sector number, and the two candidate voltage vectors to be evaluated within the corresponding sector. The second mapping table (Table 2: Small Sector Selection Table) contains the mapping relationship between the magnitude of the first-stage minimum cost function and the cost functions of the two candidate voltage vectors to be evaluated and the second-stage minimum cost function, the sixth candidate voltage vector number to be evaluated, the optimal large sector number, the magnitude of the cost function of the sixth candidate voltage vector to be evaluated and the second-stage minimum cost function, the optimal small sector number, and the cost function of the initial optimal voltage vector.

[0029] S4. Using the optimal large sector and optimal small sector lookup mapping table, complete the remaining two initial optimal voltage vectors within the optimal small sector, and calculate the corresponding duty cycle based on the cost function of the three initial optimal voltage vectors. In a preferred embodiment, S4 is specifically implemented through the following steps: S41. Based on the optimal large sector number and the optimal small sector number, query the predefined third mapping table (Table 3: Voltage Vector Table Corresponding to Small Sectors) to obtain the other two initial optimal voltage vectors and their cost functions within the optimal small sector; wherein, the third mapping table contains the mapping relationship between the optimal large sector number and the optimal small sector number, and the three initial optimal voltage vectors within the optimal small sector. Specifically, in this sub-step, according to steps S32 and S34... S big =2 and S small =3, obtain the three voltage vector numbers within the optimal small sector from Table 3, and calculate the remaining two initial optimal voltage vectors. V m1 = V 7 and V m2 = V Cost function value of 9 G 7 and G 8.

[0030] Table 3: Voltage Vector Table for Small Sectors

[0031] S42. Calculate the corresponding duty cycle based on the cost function of the three voltage vectors of the optimal small sector. In this sub-step, the cost function values ​​from steps S34 and S41 are used... G 7. G 8 and G best calculate V m1 , V m2 , V m3 duty cycle t m1 , t m2 , t m3 : (5) S5. Based on the voltage relationship between the upper and lower capacitors, a mapping table is consulted. Combining the three initial optimal voltage vectors and the duty cycle, the ultimate optimal voltage vector for the current control cycle is obtained for predicting the output current in the next control cycle. Simultaneously, control pulses are generated to control the three-level inverter. In a preferred embodiment, S5 is implemented through the following steps: S51. Based on the magnitude relationship of the upper and lower capacitor voltages, query the predefined fourth mapping table (Table 4: Redundancy Vector Selection Table) to select three optimal voltage vectors for the three initial optimal voltage vectors, forming an optimal voltage vector combination. In a preferred embodiment, the fourth mapping table includes the magnitude relationship between the upper and lower capacitor voltages and the optimal large sector number, as well as the mapping relationship between redundant voltage vector pairs. Specifically, in this sub-step, the upper and lower capacitor voltages on the DC side of the three-level inverter at the current moment are obtained from the sampling in step S1. v c1 and v c2 Select the optimal voltage vector according to Table 4. V m11 , V m22 , V m33 This refers to the three optimal voltage vector combinations. Step S4 obtains the voltage vectors and control cycle without considering capacitor voltage balance factors. Step S5 involves resolving the voltage vectors obtained in step S4. V 4 to V The voltage vector between 15 is selected according to Table 4, with the goal of reducing capacitor voltage fluctuations. In Table 3... V m1 , V m2 , V m3 Possibly V 4 to VThe values ​​are between 1 and 5, therefore, the corresponding values ​​are obtained after selecting from Table 4. Vm11 , Vm22 , Vm33 For example, voltage vectors 5, 7, and 2 are selected in Table 3, and... v c1 ≥ v c2 The final output will be voltage vectors 4, 6, and 2. At this point, the capacitor voltage difference decreases.

[0032] Table 4: Redundancy Vector Selection Table

[0033] S52. Calculate the three-phase duty cycle and the ultimate optimal voltage vector for the current control cycle based on the duty cycles corresponding to the selected three optimal voltage vector combinations, for use in predicting the output current for the next control cycle. In a preferred embodiment, the ultimate optimal voltage vector is obtained by multiplying each of the three optimal voltage vector combinations by its corresponding duty cycle and then summing the results. Specifically, in this sub-step, the three optimal voltage vector combinations obtained in step S51 are used... V m11 , V m22 , V m33 and the duty cycle obtained in step S42 t m1 , t m2 , t m3 The three-phase duty cycle of the inverter output is calculated using equation (6). S a , S b and S c And calculate the ultimate optimal voltage vector for the current control cycle according to equation (7). V opt As V old Apply to the next control cycle.

[0034] (6) (7) In the formula: S a | V m11 , S b | V m11 , S c | Vm11 、S a | V m22 , S b | V m22 , S c | V m22 、S a | V m33 , S b | V m33 , S c | V m33 They represent V m11 , V m22 and V m33 The corresponding three-phase duty cycle.

[0035] S53. Generate control pulses based on the three-phase duty cycle to control the three-level inverter.

[0036] Through the preferred embodiments described above, this invention presents a fast sector localization method in step S3, which only requires evaluating 6 voltage vectors to obtain the optimal small sector. Step S4 mainly describes the optimal voltage vector and its duty cycle calculation method. This method requires evaluating the cost function of another 2 voltage vectors, i.e., the other two initial optimal voltage vectors within the optimal small sector. Therefore, the total calculation process only requires evaluating 8 voltage vectors to obtain the optimal voltage vector. Compared with the traditional method, which requires evaluating all 24 small sectors to select the optimal voltage vector, this significantly reduces the computational complexity. In addition, this method pre-stores sector division rules, voltage vector distribution, and upper and lower capacitor voltage balance relationships in a mapping table, further improving query efficiency and real-time decision-making. Under dynamic operating conditions, the current prediction mechanism combined with delay compensation effectively suppresses the error accumulation caused by model mismatch, ensuring that the system has good steady-state accuracy and anti-disturbance capability.

[0037] Additionally, it should be noted that for control system parameters, such as the equivalent filter inductance on the inverter output side... L and equivalent resistance R Control cycle T s DC side upper and lower capacitors C 1. C 2. DC side voltage V dcFor details, please refer to Table 5: Control System Parameter Table.

[0038] Table 5: Control System Parameter Table

[0039] In summary, this invention proposes a fast modulation model predictive control method for three-level inverters, utilizing... V 18 , V 22 and V 26 The relationship between the cost function values ​​and Table 1 determine the large sector S big and G min_3 And obtain the cost function values ​​of the other two voltage vectors in the large sector. G 4 and G 5. Based on G min_3 , G 4 and G The relationship between 5 and Table 2 is obtained G min_5 The 6th vector and its cost function value G 6. Comparison G min_5 and G 6. Obtain the optimal small sector S small and G best .according to S big and S small The small sector number is determined using Table 3, and the cost function values ​​of the remaining two voltage vectors are calculated. G 7 and G 8. Finally, according to G best , G 7 and G 8. Calculate the duty cycle of the corresponding voltage vector, and obtain the optimal vector sequence according to Table 4 and the capacitor voltage relationship. Then, generate control pulses and the optimal voltage vector based on the optimal vector sequence and duty cycle. The main contribution of this invention is that it reduces the computational burden of the voltage preselection stage and the entire control algorithm by 40% and 38.5% respectively, while achieving control performance consistent with traditional methods.

[0040] This invention uses MATLAB / SIMULINK for simulation. Figure 5 The current waveforms of the traditional modulated model predictive control method and the proposed method are presented when the reference current increases from 10 A to 15 A, while the DC-side capacitor voltage is as follows: Figure 6 As shown. Figure 7 The small sectors selected by two methods are presented. The results show that the present invention has the same control performance as the traditional method. In the voltage vector preselection and overall control strategy, the number of voltage vector evaluations is reduced by 40% and 38.5% respectively, resulting in a lower computational burden.

[0041] In summary, the present invention, through the above embodiments, can achieve the following beneficial technical effects: (1) Compared with the existing voltage vector preselection method, the method proposed in this invention reduces the number of preselected voltage vectors from 10 to 6, reducing the computational burden by about 40%.

[0042] (2) Compared with the existing three-vector strategy applied to three-level inverters, the method proposed in this invention reduces the number of voltage vectors to be evaluated from 13 to 8, which is a reduction of about 38.5%.

[0043] (3) Compared with other existing voltage vector preselection methods, the method proposed in this invention exhibits control performance consistent with cost function-based preselection methods.

[0044] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0045] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification. Furthermore, the above embodiments only illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. For those skilled in the art, several modifications and improvements can be made without departing from the concept of the present invention, and these all fall within the protection scope of the present invention.

Claims

1. A fast modulation model predictive control method applied to a three-level inverter, characterized in that, include: S1. During the current control cycle, the three-phase reference current and the three-phase output current of the three-level inverter are collected and coordinate transformations are performed respectively. At the same time, the voltages of the upper and lower capacitors on the DC side of the three-level inverter are collected. S2. Based on the delay compensation strategy, the transformed reference current is delayed and compensated. At the same time, the transformed output current is combined with the optimal voltage vector obtained in the previous control cycle to predict the output current at the next moment. S3. Based on the predicted output current at the next moment, pre-select candidate voltage vectors and calculate the corresponding cost functions. Based on the relationship between each cost function, determine the optimal large sector and the optimal small sector step by step by querying the mapping table, and select an initial optimal voltage vector in the optimal small sector according to the principle of minimizing the cost function. S4. Use the optimal large sector and optimal small sector to look up the mapping table, complete the other two initial optimal voltage vectors in the optimal small sector, and calculate the corresponding duty cycle by combining the cost function of the three initial optimal voltage vectors. S5. Based on the voltage relationship between the upper and lower capacitors, look up the mapping table, combine the three initial optimal voltage vectors and the duty cycle, obtain the ultimate optimal voltage vector of the current control cycle for the next control cycle to predict the output current, and generate control pulses to control the three-level inverter. Each cost function is obtained through prediction and calculation using a modulation model.

2. The fast modulation model predictive control method for a three-level inverter according to claim 1, characterized in that, In S1, the coordinate transformation specifically transforms into the reference current and output current in a two-phase rotating coordinate system.

3. The fast modulation model predictive control method for a three-level inverter according to claim 1, characterized in that, In S2, the step of performing delay compensation on the transformed reference current specifically involves performing two-step forward delay compensation on the transformed reference current to obtain the reference current at the two forward moments.

4. The fast modulation model predictive control method for a three-level inverter according to claim 1, characterized in that, S3 specifically includes: S31. Based on the predicted output current at the next moment, three candidate voltage vectors are pre-selected from all voltage vectors of the three-level inverter, and their predicted output currents and corresponding cost functions at the two forward moments are calculated respectively. S32. Based on the relationship between the three cost functions, query the predefined first mapping table to determine the optimal large sector number, the minimum cost function of the first stage, and the other two candidate voltage vectors to be evaluated and their cost functions within the optimal large sector; S33. Compare the cost functions of the other two candidate voltage vectors to be evaluated with the minimum cost function of the first stage to determine the minimum cost function of the second stage, and query the predefined second mapping table to determine the sixth candidate voltage vector to be evaluated and its cost function; S34. Compare the cost function of the sixth candidate voltage vector to be evaluated with the minimum cost function of the second stage, and query the second mapping table according to the comparison result to determine the optimal small sector number and one of the initial optimal voltage vectors and its cost function within the optimal small sector.

5. The fast modulation model predictive control method for a three-level inverter according to claim 4, characterized in that, The first mapping table contains the relationship between the cost functions of the three pre-selected candidate voltage vectors and the mapping relationship between the minimum cost function of the first stage, the optimal large sector number, and the two candidate voltage vectors to be evaluated in the corresponding sector. The second mapping table contains the mapping relationship between the first-stage minimum cost function and the cost functions of the two candidate voltage vectors to be evaluated, and the mapping relationship between the second-stage minimum cost function, the number of the sixth candidate voltage vector to be evaluated, the optimal large sector number, the cost function of the sixth candidate voltage vector to be evaluated and the second-stage minimum cost function, the optimal small sector number, and the cost function of the initial optimal voltage vector.

6. The fast modulation model predictive control method for a three-level inverter according to claim 4, characterized in that, S4 specifically includes: S41. Based on the optimal large sector number and the optimal small sector number, query the predefined third mapping table to obtain the other two initial optimal voltage vectors and their cost functions within the optimal small sector; wherein, the third mapping table contains the mapping relationship between the optimal large sector number and the optimal small sector number, and the three initial optimal voltage vectors within the optimal small sector. S42. Based on the cost function of the three voltage vectors of the optimal small sector, calculate the corresponding duty cycle respectively.

7. The fast modulation model predictive control method for a three-level inverter according to claim 1, characterized in that, S5 specifically includes: S51. Based on the relationship between the upper and lower capacitor voltages, query the predefined fourth mapping table, select three optimal voltage vectors from the voltage vector pairs with redundancy characteristics for the three initial optimal voltage vectors respectively, and form the optimal voltage vector combination; S52. Calculate the three-phase duty cycle and the ultimate optimal voltage vector of the current control cycle based on the duty cycle corresponding to the three selected optimal voltage vector combinations for use in predicting the output current of the next control cycle; S53. Generate control pulses based on the three-phase duty cycle to control the three-level inverter.

8. The fast modulation model predictive control method for a three-level inverter according to claim 4, characterized in that, The fourth mapping table contains the size relationship between the upper and lower voltage capacitors and the optimal large sector number, as well as the mapping relationship between redundant voltage vector pairs.

9. The fast modulation model predictive control method for a three-level inverter according to claim 1, characterized in that, In S2, the output current at the next moment is also obtained based on the equivalent filter inductance and equivalent resistance on the output side of the three-level inverter, as well as the control cycle prediction calculation.

10. The fast modulation model predictive control method for a three-level inverter according to claim 1, characterized in that, In S52, the ultimate optimal voltage vector is obtained by multiplying each of the three optimal voltage vector combinations by its corresponding duty cycle and then adding them together.

Images