A T-type three-level L-filter grid-connected inverter control system and control method

Abstract

The application provides a T-type three-level L-filter grid-connected inverter control system and control method, and belongs to the technical field of three-level inverter control. It solves the problems of existing linear control relying on linear approximation, poor working condition adaptability, and inability to handle multiple nonlinear constraints. The T-type three-level L-filter grid-connected inverter control system and control method adopts a 3L-TNPC topology, the bridge arm structure is more simplified, the driving circuit complexity is lower, the hardware cost and wiring difficulty are effectively reduced, the system on-state loss is reduced, the whole machine operation efficiency is improved, the single L filter is used in the filtering link instead of the LCL filter, the number of passive devices is greatly reduced, the equipment volume cost is reduced, the inherent resonance nonlinear problem of the LCL filter is completely eliminated, and the system stability is improved; meanwhile, the first-order simple mathematical model corresponding to the single L filter can further simplify the state calculation process of the nonlinear predictive control, reduce the controller computing power load, and improve the real-time performance and stability of the nonlinear control algorithm.

Description

Technical Field

[0001] This invention belongs to the field of three-level inverter control technology, and in particular to a T-type three-level L-filter grid-connected inverter control system and control method. Background Technology

[0002] As the core interface device between the new energy power generation system and the AC power grid, the three-phase grid-connected inverter's operating performance directly determines the power quality, energy conversion efficiency, operational reliability, and economy of the entire power generation system. The three-level inverter synthesizes a stepped output voltage waveform by connecting multiple DC voltage sources in series, offering significant advantages such as low output voltage harmonic content, low voltage stress on switching devices, low electromagnetic interference, and high efficiency. The control technology of the three-level inverter directly determines its operating performance.

[0003] Currently, the mainstream PI control, PR control, and repetitive control in the industry all belong to traditional linear control strategies. These control methods generally rely on the small-signal linearization approximation of the system and the decoupling assumption of the dq coordinate system, making it difficult to maintain optimal control performance under wide-range and large-dynamic operating conditions. Among them, PI control based on a synchronous rotating coordinate system achieves steady-state error-free tracking by converting AC quantities into DC quantities through coordinate transformation, but it has obvious technical bottlenecks: the dq axis electrical quantities have a strong coupling relationship, requiring additional complex decoupling circuits; the control parameters are tuned based on a linearized model of a specific operating point, and when the inverter output base frequency fluctuates or the load conditions change over a wide range, the system's nonlinear characteristics become prominent, and the adaptability of PI parameters drops sharply, causing problems such as current waveform distortion and dynamic response lag. At the same time, the linear control framework cannot directly accommodate the multi-constraint and multi-objective control requirements of inverter switching state constraints, midpoint voltage fluctuations, and device loss nonlinearity, making it difficult to adapt to the complex nonlinear operating characteristics of three-level inverters. Summary of the Invention

[0004] The purpose of this invention is to address the aforementioned problems in existing technologies by proposing a T-type three-level L-filter grid-connected inverter control system and control method. The technical problem to be solved by this invention is: how to solve the problems of existing linear control relying on linear approximation, poor adaptability to operating conditions, and inability to handle multiple nonlinear constraints.

[0005] The objective of this invention can be achieved through the following technical solutions: A T-type three-level L-filter grid-connected inverter control system and control method, characterized by comprising the following steps: Step 1: The topology of the T-type three-level inverter L-filter includes a DC bus unit, a T-type three-level inverter, a single L-filter, and a three-phase power grid. The three-phase power grid voltages are respectively... , , DC link voltage Assumed to be a constant value, and They represent capacitors respectively. and The voltage of the upper and lower capacitors, and The voltage at both ends is The current at the neutral point is denoted as , Indicates the filter inductance; Step 2: Based on Kirchhoff's voltage law, establish a mathematical model in a three-phase stationary coordinate system, focusing on the filter inductor. Introducing its equivalent series internal resistance R, we can obtain the dynamic voltage equation for the three-phase circuit:

[0006] , , For the inverter bridge arm side current, , , The inverter bridge arm outputs three-phase voltage; Step 3: In the αβ stationary coordinate system, according to Kirchhoff's voltage law, the voltage across the filter inductor is equal to the difference between the inverter output voltage and the grid voltage. Therefore, the continuous-time state equation of the system can be expressed as:

[0007] and These represent the components of the grid-connected current along the α-axis and β-axis, respectively. and This represents the inverter output voltage component; and Represents the grid voltage component; By discretizing the continuous model and performing the forward Euler method, the current prediction model for the next sampling period can be obtained:

[0008] , This represents the predicted current value at time k+1; Step 4: Based on the principle of space vector modulation, the region where the reference voltage vector is located is divided into several sectors. For the current sector, only two adjacent active vectors and one zero vector are selected to participate in the control calculation. Let the candidate vectors be... , , ,in It is a zero vector. and Given the adjacent active power vectors within the current sector, based on the system mathematical model, the current changes corresponding to each candidate voltage vector can be further obtained, resulting in:

[0009] Where n = 0, 1, 2, corresponding to the zero vector and the two active vectors, respectively; To achieve a fixed switching frequency and reduce the harmonic content of the output current, a symmetrical seven-segment switching sequence is used:

[0010] Let the action times of the three candidate vectors be respectively , , ,satisfy:

[0011] Based on the aforementioned current change rate and the time allocation of the seven-segment switching sequence, a complete single-phase current trajectory prediction model is established to predict the grid-connected current value at time k+1:

[0012] The model uses vector action time , , , as the independent variable, can quickly calculate the future grid-connected current trajectory under different time allocations; Step 5: According to the active power command set by the grid connection system. With reactive power command Combined with the real-time data acquisition and Clarke transformation of the grid voltage components in the αβ coordinate system and The reference current expression at time k can be derived, and the instantaneous active and reactive power in the three-phase stationary coordinate system satisfy the following:

[0013] Substituting the active and reactive power command values ​​into the above formula and performing inverse derivation on the current variable, we can obtain the formula for calculating the fundamental reference current in the αβ coordinate system:

[0014] In the formula, , Let α be the β-axis component of the grid voltage sampled at time k. , These are the active and reactive power control commands given by the system, respectively. Step 6: Use the Lagrange extrapolation method to extrapolate the reference current over multiple timeframes to obtain the predicted reference current values ​​for the next two timeframes:

[0015] According to (10), we can further obtain:

[0016] Then, a two-step prediction cost function is established:

[0017] In the formula Indicates by The square operation of the current vector composed of the components is equivalent to the sum of squares of the errors of the two-axis components; Substituting the current trajectory prediction expression from equation (7) into equation (12), the cost function... It can be converted into vector action time , , Given a quadratic convex function of the independent variable, and combining the time and constraints of equation (6), the extreme values ​​can be solved using the gradient descent iterative method to obtain a set of optimal action times that minimize the cost function. , , ; Step 7: Achieve midpoint voltage balance control by dynamically adjusting the duration of the redundant small vector, and define the midpoint voltage deviation:

[0018] Define a midpoint balance coefficient to quantify the relative degree of voltage deviation:

[0019] Let the total duration of the small vectors within the current optimal switching sequence be... The compensation adjustment time is calculated based on the balance coefficient:

[0020] Midpoint voltage deviation When the voltage of the upper capacitor is greater than that of the lower capacitor, the duration of the negative small vector is increased. This corresponds to reducing the duration of action of small positive vectors. This causes the midpoint current to flow towards the undervoltage lower capacitor side; when the midpoint voltage deviation... <0, meaning when the voltage of the lower capacitor is higher than that of the upper capacitor, the duration of the positive small vector is increased. This corresponds to reducing the duration of action of the negative small vector. This causes the midpoint current to flow towards the undervoltage upper capacitor side.

[0021] This invention proposes a CR-OSS-MPC nonlinear control implementation scheme adapted to T-type three-level inverters. The inverter in this scheme adopts a 3L-TNPC topology, resulting in a simpler bridge arm structure and lower drive circuit complexity, effectively reducing hardware costs and wiring difficulty, while also reducing system conduction losses and improving overall operating efficiency. The filtering stage uses a single-L filter instead of an LCL filter, significantly reducing the number of passive components, decreasing equipment size and cost, and completely eliminating the inherent resonant nonlinearity problem of the LCL filter, thus improving system stability. Simultaneously, the simple first-order mathematical model corresponding to the single-L filter further simplifies the state calculation process of nonlinear predictive control, reduces the computing load on the controller, and improves the real-time performance and stability of the nonlinear control algorithm. While retaining the core advantages of CR-OSS-MPC nonlinear multi-objective optimal control, this scheme greatly optimizes the system hardware architecture and algorithm operating efficiency, providing a feasible solution for the industrial application of high-performance, low-cost nonlinear control strategies in the field of three-level grid-connected inverters.

[0022] Compared with existing technologies, the T-type three-level L-filter grid-connected inverter control system and control method of the present invention have the following advantages: The present invention uses the advanced nonlinear control algorithm CR-OSS-MPC (Reduced Complexity Optimal Switching Sequence Model Predictive Control) to fully retain its excellent nonlinear multi-objective optimization performance, such as low output harmonics, fast dynamic response, and low computational complexity; on this basis, through the simplified adaptation of the main circuit topology and filtering scheme, the overall system control performance and hardware economy are significantly improved simultaneously. At the topology level, a T-type three-level topology is adopted, which reduces the number of power devices and significantly reduces the difficulty of hardware wiring. At the same time, the conduction path in the zero-level operating range is shorter, significantly reducing power conduction losses and improving the overall operating efficiency. The heat dissipation structure design and the convenience of later maintenance are optimized simultaneously. In the filtering stage, a single L filter is used to replace the LCL filter, eliminating many passive components such as filter capacitors, grid-side inductors and resonant damping components, effectively reducing the size of the equipment and reducing material costs. At the same time, the inherent resonant nonlinear disturbance of the LCL filter is eliminated from the source, eliminating the need for additional damping suppression strategies, and the system is naturally more stable. Furthermore, the single L-filter corresponds to a simple first-order mathematical model with fewer state variables and lower predictive model complexity. This can further reduce the computational load of the nonlinear predictive control algorithm in online rolling optimization and lower the computing power requirements of the main control chip. The overall solution fully leverages all the advantages of CR-OSS-MPC nonlinear constraint optimal control while ensuring that the grid-connected power quality and dynamic control effect remain unchanged. It significantly reduces the overall cost of software and hardware and is more suitable for industrial mass promotion and application in medium and low voltage high-power grid-connected power generation scenarios such as distributed photovoltaics and industrial and commercial energy storage. Attached Figure Description

[0024] Figure 1 This is a topology diagram of the L-filter of the grid-connected T-type three-level inverter of the present invention.

[0025] Figure 2 This is a control block diagram of the method proposed in this invention.

[0026] Figure 3 This is the present invention. Three-phase grid-connected current waveform at 3kW Detailed Implementation

[0027] The following are specific embodiments of the present invention, which are described in conjunction with the accompanying drawings. However, the present invention is not limited to these embodiments.

[0028] A T-type three-level L-filter grid-connected inverter control system and control method, characterized by comprising the following steps: Step 1: The topology of the T-type three-level inverter L-filter includes a DC bus unit, a T-type three-level inverter, a single L-filter, and a three-phase power grid. The three-phase power grid voltages are respectively... , , DC link voltage Assumed to be a constant value, its function is to provide a stable DC input voltage for the inverter and buffer power fluctuations on the inverter input side. and They represent capacitors respectively. and The voltage of the upper and lower capacitors, and The voltage at both ends is The current at the neutral point is denoted as , Indicates the filter inductance; The specific wiring method for each phase arm of the T-type three-level inverter is as follows: Figure 1 upper bridge arm switch tube The collector is connected to the positive terminal of the DC bus, and the emitter is connected to the lower bridge arm switch. The collector of the lower bridge arm is the connection point, which is the output terminal of each phase arm; the lower bridge arm switching transistor... The emitter is connected to the negative terminal of the DC bus; the midpoint clamping switch transistor The collector is connected to the midpoint O of the DC bus, and the emitter is connected to the midpoint clamping switch. Emitter; Midpoint clamping switch transistor The collector of the transistor is connected to the midpoint O of the DC bus. By controlling the on / off combinations of these four transistors, the bridge arm can output three different voltage levels, thus achieving a three-level inverter. Compared with the traditional two-level topology, it has the advantages of lower output harmonic content and lower switching stress.

[0029] Each phase arm of a T-type three-level inverter has three legal switching states, corresponding to three output voltage levels. The on state is defined as "1" and the off state as "0". The output phase voltage is the potential of the bridge arm output terminal relative to the DC bus midpoint O. The correspondence between the switching states and output voltages of the T-type three-level inverter is shown in the table below: Table 1. Three switching state combinations for a T-type three-level inverter

[0030] For a T-type three-phase three-level inverter, each bridge arm can output three different switching states, resulting in a total of 27 possible output states. Based on the angle of each voltage vector, it can be divided into six large sectors, and based on the magnitude of each voltage vector, it can be divided into four types: zero vector, small vector, medium vector, and large vector.

[0031] Step 2: A single-L filter is connected between the AC output terminal of the T-type three-level inverter and the power grid, consisting of only three filter inductors connected in series in the three-phase output circuits. The inverter output is directly connected to the three-phase grid voltage via a filter inductor. , , Compared to LCL filters, single-L filters eliminate the need for three filter capacitors and corresponding damping circuits, eliminating the need for complex damping control strategies and significantly reducing parameter design complexity. Furthermore, single-L filters are smaller, lighter, and less expensive, making them more suitable for installation and maintenance in industrial settings.

[0032] Based on Kirchhoff's voltage law, a mathematical model is established in a three-phase stationary coordinate system, focusing on the filter inductor. Introducing its equivalent series internal resistance R, we can obtain the dynamic voltage equation for the three-phase circuit:

[0033] , , For the inverter bridge arm side current, , , This provides the three-phase output voltage for the inverter bridge arm. The first-order differential equations fully describe the dynamic characteristics of the single-L-filter grid-connected system, providing a theoretical basis for subsequent control strategy design based on OSS-MPC. Compared to the traditional second-order model of LCL filtering, this first-order model reduces capacitor-related state variables, significantly lowering computational complexity and improving the real-time performance of the control algorithm.

[0034] Based on the core control logic of CR-OSS-MPC, this invention makes necessary adaptation adjustments to suit the physical characteristics of the T-type three-level topology and single-L filter, ensuring that all the excellent features of the original algorithm are fully preserved in the simplified topology. This invention uses the core framework of main sector judgment, vector mapping, generation of three candidate switch sequences, symmetrical seven-segment switch sequence, current trajectory prediction, and analytical solution of duty cycle, while adapting and adjusting the switch state mapping, prediction model, and voltage equalization mechanism directly related to the hardware topology, maximizing the reuse of the technical accumulation and engineering experience of the original algorithm.

[0035] Step 3: For a T-type three-level grid-connected inverter using an L-filter, in the αβ stationary coordinate system, according to Kirchhoff's voltage law, the voltage across the filter inductor is equal to the difference between the inverter output voltage and the grid voltage. Therefore, the continuous-time state equation of the system can be expressed as:

[0036] and These represent the components of the grid-connected current along the α-axis and β-axis, respectively. and This represents the inverter output voltage component; and Represents the grid voltage component; Since the controller is implemented using a digital processor, the continuous model needs to be discretized. After discretization using the forward Euler method, the current prediction model for the next sampling period can be obtained.

[0037] , This represents the predicted current value at time k+1; The above model describes the influence of the inverter output voltage vector on the trend of grid-connected current variation, providing a mathematical basis for subsequent current trajectory prediction and optimal switching sequence selection.

[0038] Step 4: To reduce the computational burden of traditional finite control set model predictive control requiring traversal of all switching states, optimal switching sequence model predictive control selects only candidate voltage vectors within the neighborhood of the current reference voltage vector for optimization calculations in each control cycle. Based on the principle of space vector modulation, the region containing the reference voltage vector is divided into several sectors. For the current sector, only two adjacent active vectors and one zero vector are selected for control calculations. Let the candidate vectors be... , , ,in It is a zero vector. and For the adjacent active power vectors within the current sector, limiting the search range of candidate vectors can significantly reduce the number of cost function calculations, improve the operating efficiency of the control system, and maintain high current tracking performance. Based on the system's mathematical model, the current changes corresponding to each candidate voltage vector can be further obtained, resulting in:

[0039] Where n = 0, 1, 2, corresponding to the zero vector and the two active vectors, respectively; To achieve a fixed switching frequency and reduce the harmonic content of the output current, a symmetrical seven-segment switching sequence is used:

[0040] Let the action times of the three candidate vectors be respectively , , ,satisfy:

[0041] Based on the aforementioned current change rate and the time allocation of the seven-segment switching sequence, a complete single-phase current trajectory prediction model is established to predict the grid-connected current value at time k+1:

[0042] The model uses vector action time , , , as the independent variable, can quickly calculate the future grid-connected current trajectory under different time allocations; For model predictive control systems, the optimization objective of the cost function is to make the predicted grid-connected current accurately track the reference current. Therefore, it is necessary to first determine the reference current benchmark value in the αβ stationary coordinate system to provide a tracking target for subsequent optimization calculations.

[0043] Step 5: This invention targets a three-phase grid-connected inverter system, employing instantaneous reactive power theory to calculate the reference current, based on the active power command set by the grid-connected system. With reactive power command Combined with the real-time data acquisition and Clarke transformation of the grid voltage components in the αβ coordinate system and The reference current expression at time k can be derived, and the instantaneous active and reactive power in the three-phase stationary coordinate system satisfy the following:

[0044] Substituting the active and reactive power command values ​​into the above formula and performing inverse derivation on the current variable, we can obtain the formula for calculating the fundamental reference current in the αβ coordinate system:

[0045] In the formula, , Let α be the β-axis component of the grid voltage sampled at time k. , These are the active and reactive power control commands given by the system; under normal grid-connected operating conditions, the reactive power command can be... =0, at this point the formula can be further simplified, the amount of calculation is lower, and it is suitable for the single L-type filter grid topology used in this invention. Since the system only contains a first-order filter inductor, there is no need to introduce additional current correction terms such as resonance suppression and damping compensation. The reference current calculation link is simple, which can further reduce the overall computational burden of the control algorithm.

[0046] Step 6: The above formula only yields the reference current value at the current time k. However, in actual digital control platforms, sampling, coordinate transformation, optimization calculation, and pulse update all have inherent delays, and the overall control delay can typically reach one control cycle. If only the reference current at the current time is relied upon, along with a single-step prediction model to construct the cost function, the control output will lag behind the system state, ultimately leading to decreased grid-connected current tracking accuracy, increased steady-state harmonic content, and slower dynamic response speed. To offset the negative impact of digital control delay on the CR-OSS-MPC control performance, this paper adopts a two-step prediction strategy, using the Lagrange extrapolation method to extrapolate the reference current over multiple timeframes to obtain the predicted reference current values ​​for the next two timeframes:

[0047] According to (10), we can further obtain:

[0048] Then, a two-step prediction cost function is established:

[0049] In the formula Indicates by The square operation of the current vector composed of the components is equivalent to the sum of squares of the errors of the two-axis components; Substituting the current trajectory prediction expression from equation (7) into equation (12), the cost function... It can be converted into vector action time , , Given a quadratic convex function of the independent variable, and combining the time and constraints of equation (6), the extreme values ​​can be solved using the gradient descent iterative method to obtain a set of optimal action times that minimize the cost function. , , ; Furthermore, the DC side of the T-type three-level inverter consists of two series-connected voltage-dividing capacitors. If the charging and discharging states of these two capacitors are unbalanced during system operation, it will cause a midpoint potential shift, resulting in output voltage waveform distortion and increased grid-connected current harmonics, which can severely affect system stability. Traditional model predictive control typically incorporates the midpoint voltage balance objective into the cost function for joint optimization; however, this method requires additional weight coefficient design, and the weight parameters are difficult to adapt to the entire operating range, making parameter tuning cumbersome and increasing the computational load of the algorithm.

[0050] Each set of small vectors in the T-type three-level space vector corresponds to two sets of redundant switch states, defined as a positive small vector and a negative small vector: when the positive small vector is active, the midpoint current flows into the midpoint of the DC bus, raising the midpoint potential; when the negative small vector is active, the midpoint current flows out of the midpoint of the DC bus, lowering the midpoint potential. By dynamically adjusting the time allocation ratio of the two sets of redundant small vectors, the direction and amplitude of the midpoint current can be precisely controlled, achieving voltage balance between the upper and lower capacitors on the DC side. This voltage equalization mechanism is independent of the current tracking cost function, requires no weighting coefficients, and does not increase the optimization calculation dimension.

[0051] Step 7: Achieve midpoint voltage balance control by dynamically adjusting the duration of the redundant small vector, and define the midpoint voltage deviation:

[0052] Define a midpoint balance coefficient to quantify the relative degree of voltage deviation:

[0053] Let the total duration of the small vectors within the current optimal switching sequence be... The compensation adjustment time is calculated based on the balance coefficient:

[0054] Midpoint voltage deviation When the voltage of the upper capacitor is greater than that of the lower capacitor, the duration of the negative small vector is increased. This corresponds to reducing the duration of action of small positive vectors. This causes the midpoint current to flow towards the undervoltage lower capacitor side; when the midpoint voltage deviation... <0, meaning when the voltage of the lower capacitor is higher than that of the upper capacitor, the duration of the positive small vector is increased. This corresponds to reducing the duration of action of the negative small vector. This causes the midpoint current to flow towards the undervoltage upper capacitor side. During the adjustment process, the total duration of the small vector action is maintained. The optimal vector action time allocation result obtained by the CR-OSS-MPC algorithm remains constant and unchanged. The duration allocation of redundant switching states is only completed in the pulse orchestration stage, which has no impact on the current tracking control performance.

[0055] This method achieves automatic voltage balancing of the upper and lower capacitors by dynamically adjusting the duty cycle of redundant small vectors. It does not require additional weighting factor design and does not increase the complexity of solving the cost function. It has the advantages of simple implementation, low computational cost, and stable voltage balancing effect.

[0056] The overall control strategy block diagram corresponding to this invention is as follows: Figure 2 As shown, the specific execution flow of the control method of the present invention is as follows: Signal sampling: Real-time acquisition of three-phase grid-connected current, three-phase grid voltage, and DC-side upper and lower capacitor voltages; Coordinate transformation and sub-interval location: The sampled three-phase voltage and current are transformed to the αβ stationary coordinate system using Clarke transformation, and the current spatial sub-interval is located based on the grid voltage vector; Candidate vector reduction and filtering: Based on the current sub-interval position, a reduced set of candidate voltage vector combinations is obtained, thus completing the candidate sub-interval reduction; Reference current generation: Based on the grid-connected active and reactive power control commands, generate the future two-phase reference current in the αβ stationary coordinate system; Current change rate calculation: Based on the voltage loop equation of the L-type filter, calculate the grid-connected current change rate corresponding to each candidate vector; Current prediction modeling: Based on the discrete current trajectory equation and combined with the current change rate, a prediction expression for the grid-connected current in the next two phases with respect to the vector action time is established; Gradient iteration to find the optimal duration: Substitute the prediction expression and the reference current into the cost function, take the initial duration as the starting point of the iteration, and iteratively update the vector duration along the cost descent direction. After the iteration converges, the optimal duration and minimum cost value corresponding to each candidate vector are obtained. Global optimization: Compare the minimum substitution value of all candidate vector groups, and select the group with the minimum substitution value as the optimal switching sequence for the current cycle; Midpoint potential compensation: Based on the voltage deviation of the upper and lower capacitors on the DC side, the duty cycle allocation ratio of redundant small vectors in the optimal sequence is corrected to achieve DC bus voltage balance; Pulse arrangement: Generate a seven-segment PWM drive pulse according to the corrected optimal time allocation; Pulse output: Output drive signal to control the operation of T-type three-level inverter.

[0057] To verify the effectiveness and control performance of the proposed nonlinear control system, a complete simulation model of a three-phase T-type three-level grid-connected inverter was built on the Matlab / Simulink simulation platform. The main circuit of the model adopts a T-type neutral-point clamped three-level topology. The DC side consists of two series-connected voltage-dividing capacitors, and the AC side is connected to the three-phase power grid via a single-L filter. The control loop fully implements the CR-OSS-MPC control strategy, including all aspects such as signal sampling, coordinate transformation, sector positioning and candidate vector screening, current trajectory prediction, cost function construction, gradient iteration to solve for the optimal action time, neutral-point potential compensation, and seven-segment PWM generation.

[0058] The main parameters for the simulation are set as shown in the table below: Table 2 Simulation Parameter Design

[0059] Figure 3 The waveform shows the steady-state current of the three-phase grid-connected system under the condition of rated active power of 3kW and reactive power command of 0. As can be seen from the waveform, the three-phase grid-connected current... , , The three-phase symmetry is good, with phases differing by 120° sequentially. The waveform has high sinusoidal strength and no obvious distortion, spikes, or DC bias. The peak current is approximately 12.6A, which is basically consistent with the theoretical calculation value, indicating that the reference current generation logic is accurate and the current tracking control has no steady-state error.

[0060] Meanwhile, the high-frequency ripple amplitude of the grid-connected current is small and the overall waveform is smooth, which verifies the effectiveness of the T-type three-level topology combined with the L-type filter and the CR-OSS-MPC nonlinear control strategy adopted in this invention: by eliminating the filter capacitor and damping branch of the LCL filter and simplifying the system hardware structure, it is still possible to achieve low harmonic and high sinusoidal grid-connected current output, with excellent steady-state grid-connected performance, meeting the grid-connected power quality requirements.

[0061] The specific embodiments described herein are merely illustrative of the spirit of the invention. Those skilled in the art to which this invention pertains may make various modifications or additions to the described specific embodiments or use similar methods to substitute them, without departing from the spirit of the invention or exceeding the scope defined by the appended claims.

Claims

1. A control system and control method for a T-type three-level L-filter grid-connected inverter, characterized in that, Includes the following steps: Step 1: The topology of the T-type three-level inverter L-filter includes a DC bus unit, a T-type three-level inverter, a single L-filter, and a three-phase power grid. The three-phase power grid voltages are respectively... , , DC link voltage Assumed to be a constant value, and They represent capacitors respectively. and The voltage of the upper and lower capacitors, and The voltage at both ends is The current at the neutral point is denoted as , Indicates the filter inductance; Step 2: Based on Kirchhoff's voltage law, establish a mathematical model in a three-phase stationary coordinate system, focusing on the filter inductor. Introducing its equivalent series internal resistance R, we can obtain the dynamic voltage equation for the three-phase circuit: , , For the inverter bridge arm side current, , , The inverter bridge arm outputs three-phase voltage; Step 3: In the αβ stationary coordinate system, according to Kirchhoff's voltage law, the voltage across the filter inductor is equal to the difference between the inverter output voltage and the grid voltage. Therefore, the continuous-time state equation of the system can be expressed as: and These represent the components of the grid-connected current along the α-axis and β-axis, respectively. and This represents the inverter output voltage component; and Represents the grid voltage component; By discretizing the continuous model and performing the forward Euler method, the current prediction model for the next sampling period can be obtained: , This represents the predicted current value at time k+1; Step 4: Based on the principle of space vector modulation, the region where the reference voltage vector is located is divided into several sectors. For the current sector, only two adjacent active vectors and one zero vector are selected to participate in the control calculation. Let the candidate vectors be... , , ,in It is a zero vector. and Given the adjacent active power vectors within the current sector, based on the system mathematical model, the current changes corresponding to each candidate voltage vector can be further obtained, resulting in: Where n = 0, 1, 2, corresponding to the zero vector and the two active vectors, respectively; To achieve a fixed switching frequency and reduce the harmonic content of the output current, a symmetrical seven-segment switching sequence is used: Let the action times of the three candidate vectors be respectively , , ,satisfy: Based on the aforementioned current change rate and the time allocation of the seven-segment switching sequence, a complete single-phase current trajectory prediction model is established to predict the grid-connected current value at time k+1: The model uses vector action time , , , as the independent variable, can quickly calculate the future grid-connected current trajectory under different time allocations; Step 5: According to the active power command set by the grid connection system. With reactive power command Combined with the real-time data acquisition and Clarke transformation of the grid voltage components in the αβ coordinate system and The reference current expression at time k can be derived, and the instantaneous active and reactive power in the three-phase stationary coordinate system satisfy the following: Substituting the active and reactive power command values ​​into the above formula and performing inverse derivation on the current variable, we can obtain the formula for calculating the fundamental reference current in the αβ coordinate system: In the formula, , Let α be the β-axis component of the grid voltage sampled at time k. , These are the active and reactive power control commands given by the system, respectively. Step 6: Use the Lagrange extrapolation method to extrapolate the reference current over multiple timeframes to obtain the predicted reference current values ​​for the next two timeframes: According to (10), we can further obtain: Then, a two-step prediction cost function is established: In the formula Indicates by The square operation of the current vector composed of the components is equivalent to the sum of squares of the errors of the two-axis components; Substituting the current trajectory prediction expression from equation (7) into equation (12), the cost function... It can be converted into vector action time , , Given a quadratic convex function of the independent variable, and combining the time and constraints of equation (6), the extreme values ​​can be solved using the gradient descent iterative method to obtain a set of optimal action times that minimize the cost function. , , ; Step 7: Achieve midpoint voltage balance control by dynamically adjusting the duration of the redundant small vector, and define the midpoint voltage deviation: Define a midpoint balance coefficient to quantify the relative degree of voltage deviation: Let the total duration of the small vectors within the current optimal switching sequence be... The compensation adjustment time is calculated based on the balance coefficient: Midpoint voltage deviation When the voltage of the upper capacitor is greater than that of the lower capacitor, the duration of the negative small vector is increased. This corresponds to reducing the duration of action of small positive vectors. This causes the midpoint current to flow towards the under-voltage lower capacitor side; Midpoint voltage deviation <0, meaning when the voltage of the lower capacitor is higher than that of the upper capacitor, the duration of the positive small vector is increased. This corresponds to reducing the duration of action of the negative small vector. This causes the midpoint current to flow towards the undervoltage upper capacitor side.

Images