A T-type three-level virtual synchronous machine power point coordination control method and system

Abstract

The application discloses a T-type three-level virtual synchronous generator power midpoint cooperative control method and system, comprising a data acquisition and coordinate transformation module, a power calculation module, a VSG power outer loop control and power feedforward reference generation module, a multi-step model prediction module, a power error prediction module, a midpoint potential prediction and balance module, a unified cost function and optimal control amount selection module and a three-level PWM driving module; data acquisition and coordinate transformation, output power calculation and VSG power outer loop adjustment, obtaining a next control period power feedforward current reference value, multi-step model prediction based on a candidate control amount, predicting active power and reactive power errors, midpoint potential prediction and balance control, unified cost function evaluation and optimal control amount selection, three-level PWM driving and cyclic updating. The application solves the problem that in the T-type three-level virtual synchronous generator control, the traditional model prediction control is difficult to make the active and reactive power references quickly enter the optimization target, thereby leading to the dynamic response lag.

Description

Technical Field

[0001] This invention relates to a control method for a T-type three-level virtual synchronous machine, specifically to a power midpoint collaborative control method and system for a T-type three-level virtual synchronous machine. Background Technology

[0002] Virtual synchronous generator (VSG) control, by simulating the characteristics of a synchronous generator, can effectively improve the converter's ability to support grid frequency and power disturbances. Existing VSG control systems mostly employ multi-loop control structures, but when the active or reactive power reference values ​​experience a step change, traditional structures struggle to quickly transmit power commands to the switching state optimization level, limiting the system's dynamic response. Furthermore, while T-type three-level converters offer advantages such as high efficiency and low harmonics, their DC-side upper and lower capacitor voltages are prone to imbalance. Long-term deviations in the midpoint potential can severely impact power quality and system reliability. Model predictive control (MPC) features fast dynamic response and suitability for multi-objective optimization, but existing methods often focus on minimizing current errors or solely suppressing the midpoint voltage, neglecting to consider how the VSG power outer-loop reference enters the predictive control optimization process. Simply reducing current harmonics does not necessarily guarantee good power command tracking capability; it can easily lead to lag in actual output power when the active power reference experiences a step change. Therefore, how to propose a control method that can unify power feedforward, multi-step prediction, and midpoint potential balance, while simultaneously achieving rapid power tracking and stable control of the midpoint potential, is a current research challenge. Summary of the Invention

[0003] To address the problems existing in the prior art, this invention provides a power midpoint collaborative control method and system for a T-type three-level virtual synchronous machine. This solves the problem that traditional model predictive control in T-type three-level virtual synchronous machine control is unable to quickly bring the active and reactive power references into the optimization target, thus leading to dynamic response lag, and the contradiction that existing control strategies cannot simultaneously take into account the DC side midpoint potential balance and grid-connected current quality.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a power midpoint collaborative control method for a T-type three-level virtual synchronous machine, comprising the following steps: S1. Data Acquisition and Coordinate Transformation: Acquire AC side voltage and current and DC side capacitor voltage information of the T-type three-level converter, and obtain the actual voltage and current in the synchronous rotating coordinate system through coordinate transformation; S2. Output power calculation and VSG power outer loop adjustment: Calculate the actual active power and reactive power based on the collected AC side voltage and current, and generate reference voltage and synchronization angle information through the VSG power outer loop; S3. Obtain the power feedforward current reference value for the next control cycle: Based on the power command and the reference voltage direction of the VSG output, convert the active and reactive power targets into d-axis current reference values ​​and q-axis current reference values ​​for model prediction and evaluation in step S4. S4. Multi-step model prediction based on candidate control quantities: Using the power feedforward current reference value and the current current state as input, predict the current change in future sampling periods under the action of candidate control quantities. S5. Predicted active and reactive power error: Calculate the active and reactive power at future times based on the predicted current and form the corresponding power error evaluation item. S6. Midpoint potential prediction and balance control: Evaluate and suppress the midpoint potential deviation of the T-type three-level converter based on the DC side upper and lower capacitor voltage deviation and AC side current direction. S7. Unified cost function evaluation and optimal control quantity selection: Taking into account current tracking, power prediction error, midpoint potential deviation and control smoothing constraints, the optimal control quantity for the current sampling period is selected. S8, Three-level PWM drive and cyclic update: Convert the optimal control voltage command into a T-type three-level bridge arm switching signal, and repeat the above control process in the next sampling cycle.

[0005] Furthermore, the specific method for step S3 is as follows: First, calculate the magnitude of the reference voltage vector: ; And obtain the direction of the reference voltage: ; In the formula, For the reference voltage vector magnitude, , These are the directional coefficients of the reference voltage vector on the d-axis and q-axis, respectively; Then calculate the measured voltage amplitude: ; Based on the mapping relationship between power and current, the reference value of the power feedforward current is obtained: ; In the formula, and These represent the reference values ​​for the d-axis and q-axis currents, respectively.

[0006] Furthermore, the specific method for step S4 is as follows: First, based on the d-axis current reference value and q-axis current reference value obtained in step S3, and the d-axis actual current and q-axis actual current obtained in step S1, calculate the d-axis current tracking error and q-axis current tracking error respectively: ; In the formula, , The base control voltage in the dq coordinate system consists of grid-connected voltage compensation term, filter resistor voltage drop compensation term, current error feedback term and dq axis coupling compensation term. The equivalent resistance of the AC side filter branch; For AC side filter inductance; This is the current error adjustment coefficient; These are the coupling feedforward coefficients; Centered on the base control voltage, a finite number of discrete offsets are superimposed along the d-axis and q-axis according to a preset voltage step size, forming a finite set of candidate control quantities. For each set of candidate control voltages... , The current in the next sampling period is predicted based on the AC side filtering model: ; In the formula, The sampling period is , The predicted currents for the d-axis and q-axis of the next sampling period are respectively; if the prediction step size is n, then the following is recursively derived based on the above one-step prediction: .

[0007] Furthermore, the specific method of step S5 is as follows: Based on the predicted current , And combined with the current or predicted grid-connected voltage components , Calculate the predicted active power and predicted reactive power: ; In the formula, Let the predicted active power be at the (k+n)th time in the future. This is the predicted reactive power at time k+n; the predictive controller can also calculate the internal power reference value based on the current reference value generated from the acquired power feedforward current reference value. ; In the formula, and These are the active and reactive power reference values ​​used internally by the model predictive controller; Based on the deviation between the predicted power and the reference power, active power error evaluation terms and reactive power error evaluation terms are constructed: ; In the formula, This is the normalized reference value for active power. This is the normalized reference value for reactive power.

[0008] Furthermore, the specific method of step S6 is as follows: Calculate the DC side midpoint potential deviation: ; In the formula, Indicates the degree of voltage imbalance between the upper and lower split capacitors; when When, it means the voltage of the upper capacitor is higher than the voltage of the lower capacitor; when When, it indicates that the voltage of the lower capacitor is higher than the voltage of the upper capacitor; when When the voltage of the upper and lower capacitors is balanced, it indicates that the voltages of the upper and lower capacitors are in equilibrium. The midpoint potential deviation is normalized: ; In the formula, is the normalized threshold for midpoint potential deviation, and sat() is the amplitude limiting function; Based on the normalized midpoint potential deviation and the direction of the three-phase current, construct the three-phase midpoint potential coordination compensation: ; In the formula, Here, sgn() is the midpoint potential compensation coefficient, and sgn() is the sign function. , , These are the midpoint potential coordination compensation amounts for phases A, B, and C, respectively. Simultaneously, a midpoint potential evaluation term is set in the unified cost function: ; In the formula, This serves as the benchmark value for evaluating midpoint potential deviation.

[0009] Furthermore, the specific method for step S7 is as follows: The current tracking evaluation item can be expressed as: ; The control smoothing evaluation term can be expressed as: ; In the formula, Used to evaluate the deviation between the predicted current and the reference current. Used to limit sudden changes in control quantities and improve the quality of output voltage and current; Combining the above evaluation terms, the unified cost function is expressed as: ; In the formula, , , , , These are the weighting coefficients for current tracking, active power tracking, reactive power tracking, neutral point potential balance, and control smoothing constraints, respectively. The controller iterates through the candidate control quantities, selects the control quantity that minimizes J as the optimal control quantity for the current sampling period, and outputs the three-phase control voltage command. , , .

[0010] Furthermore, the specific method of step S8 is as follows: The three-phase control voltage command obtained in step S7 is input into the three-level PWM modulation stage to generate drive signals for the A, B, and C three-phase bridge arm switching devices. Based on the drive signals, the switch devices of each bridge arm are controlled to be turned on or off, so that the T-type three-level converter outputs the corresponding three-phase voltage, and outputs electrical energy to the grid or load through the AC side filtering stage. In the next sampling period, the system re-acquires AC side voltage, current, DC side capacitor voltage, and power reference values, repeating steps S1 to S8.

[0011] A T-type three-level virtual synchronous machine power midpoint cooperative control system is used to implement the above control method. It includes a data acquisition and coordinate transformation module, a power calculation module, a VSG power outer loop control and power feedforward reference generation module, a multi-step model prediction module, a power error prediction module, a midpoint potential prediction and balancing module, a unified cost function and optimal control quantity selection module, and a three-level PWM drive module. The data acquisition and coordinate transformation module is connected to the power calculation module. The power calculation module is connected to the VSG power outer loop control and power feedforward reference generation module. The VSG power outer loop control and power feedforward reference generation module is connected to the multi-step model prediction module. The multi-step model prediction module is connected to the power error prediction module. The power error prediction module, midpoint potential prediction and balancing module are both connected to the unified cost function and optimal control quantity selection module. The unified cost function and optimal control quantity selection module is connected to the three-level PWM drive module.

[0012] Compared with the prior art, the present invention has the following advantages: 1. Power command tracking capability: Unlike traditional predictive control, this invention uses a power feedforward generation module to generate the power command tracking capability of the VSG outer loop. and By directly mapping it to an internal reference quantity of MPC and incorporating it into the cost function, the problem of power reference being difficult to track effectively is solved.

[0013] 2. Dynamic Power Response and Midpoint Potential Balance: This invention incorporates the DC-side voltage difference (du) into a unified cost function. Even under extreme dynamic conditions such as active power step fluctuations or reactive power disturbances, the midpoint potential voltage difference can still be controlled within a small range, maintaining midpoint potential stability.

[0014] 3. Grid-connected current quality and power ripple: The current error term and control smoothing term are integrated into the unified cost function. The three-phase current distortion rate under rated operating conditions is kept within the standard range, and the standard deviation of active power is kept at a low level. The steady-state performance and power quality of the system have achieved a qualitative leap.

[0015] 4. Reactive power disturbance adaptability: Under severe operating conditions with sudden increase in inductive load, the reactive power tracking error is small and the active power channel is stable without obvious distortion, which fully demonstrates the strong robustness of the control system to load disturbances and parameter changes. Attached Figure Description

[0016] Figure 1 This is a flowchart of the control method of the present invention; Figure 2 This is a schematic diagram of the control system structure of the present invention; Figure 3 This is a waveform diagram of the three-phase AC side current under steady-state operation in an embodiment of the present invention; Figure 4 The above are waveforms of active and reactive power response under steady-state operating conditions in this embodiment of the invention. Figure 5 This is a waveform diagram of the DC side midpoint voltage deviation under steady-state operating conditions in an embodiment of the present invention; Figure 6 This is a THD spectrum analysis diagram of the AC side current under steady-state operating conditions in an embodiment of the present invention. Figure 7 The waveforms of active and reactive power response under the active power step condition in this embodiment of the invention are shown. Figure 8 This is a waveform diagram of the DC side midpoint voltage deviation under active power step conditions in an embodiment of the present invention. Figure 9 The waveforms of active and reactive power response under reactive load disturbance conditions in this embodiment of the invention are shown. Figure 10 This is a waveform diagram showing the recovery of the midpoint voltage deviation under the initial imbalance of the DC side midpoint voltage in an embodiment of the present invention. Detailed Implementation

[0017] The invention will now be further described with reference to the accompanying drawings.

[0018] 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.

[0019] like Figure 2 As shown, this invention provides a T-type three-level virtual synchronous machine power midpoint collaborative control system, including a data acquisition and coordinate transformation module, a power calculation module, a VSG power outer loop control and power feedforward reference generation module, a multi-step model prediction module, a power error prediction module, a midpoint potential prediction and balancing module, a unified cost function and optimal control quantity selection module, and a three-level PWM drive module. The VSG power outer loop control and power feedforward reference generation module includes a VSG power outer loop control module and a power feedforward current reference generation module, which are connected. All modules work collaboratively to control the main circuit of the T-type three-level energy storage converter, achieving unified control of active power, reactive power, AC side current, and DC side midpoint potential.

[0020] The main circuit of the T-type three-level energy storage converter, as the controlled object of this invention, includes the DC-side upper capacitor. Lower capacitor The system consists of a midpoint O, a three-phase T-type three-level bridge arm, an AC-side filter inductor L, an equivalent resistance R, a filter capacitor C, a grid-connected switch S, and a load branch. The DC-side split capacitor provides the three-level base voltages (positive, zero, and negative) for the three-level bridge arm. The three-phase bridge arm generates the three-phase output voltage through the switching of the switching devices. The AC-side filter stage suppresses switching harmonics and smoothly delivers the converter's output power to the grid or load.

[0021] The input terminals of the data acquisition and coordinate transformation module are connected to the AC side voltage sampling terminal, AC side current sampling terminal, and DC side upper and lower capacitor voltage sampling terminals of the main circuit of the T-type three-level energy storage converter, respectively, to acquire the grid-connected voltage, AC side current, and DC side capacitor voltage. This module converts the voltage and current in the three-phase stationary coordinate system into the synchronous rotating coordinate system. , , , Its output terminals are connected to the power calculation module, the power feedforward current reference generation module, and the multi-step model prediction module, respectively, providing the above modules with the operating status quantities of the current sampling period.

[0022] The power calculation module calculates the actual output active power P and reactive power Q of the converter based on the voltage and current components obtained after coordinate transformation. The P and Q outputs from this module are fed back as feedback values ​​to the VSG power outer loop control module, and compared with the given... , Together, they participate in power closed-loop regulation. The role of the power calculation module is to convert the AC side voltage and current states into power states, enabling subsequent VSG control to adjust around the power deviation.

[0023] The VSG power outer loop control module is used to implement the virtual synchronous machine power regulation function. This module receives active power reference values. Reactive power reference value Rated angular frequency Rated voltage and the actual power fed back by the power calculation module , Synchronization phase angle is generated through virtual synchronizer control relationship. It outputs the reference voltage in the dq coordinate system. , This module enables the energy storage converter to simulate the power regulation characteristics of a synchronous generator when operating in grid-connected mode, providing a reference voltage direction for subsequent power feedforward and predictive control.

[0024] The power feedforward current reference generation module and the VSG power outer loop control module together constitute the upper-level power regulation section of this invention. This module receives... , , and measured voltage , The current reference value required by the predictive controller is generated based on the power command and voltage vector direction. , Its function is to pre-map the active and reactive power targets as reference quantities that can be directly tracked by the inner current loop, so that the power command can directly participate in the predictive control process, thereby reducing the response lag caused by the gradual adjustment of the traditional outer power loop.

[0025] The multi-step model prediction module is the core prediction component of the control system of this invention. This module receives power feedforward generated by the multi-step model prediction module. , and actual current , Synchronous phase angle Based on the AC side filter parameters of the T-type three-level converter and candidate control quantities, the current change within the future control cycle is predicted using constant state variables. , The module is used to determine the impact of different candidate control variables on the future current state, providing a predictive basis for the unified cost function evaluation.

[0026] The power error prediction module is used to evaluate the impact of candidate control variables on power point tracking performance. This module receives predicted current and predicted power information from the multi-step model prediction module and combines it with a power reference value. , Calculate the predicted active power error and the predicted reactive power error to form an active power evaluation item. and reactive power evaluation items By setting up this module, the control system not only focuses on current tracking error, but can also directly predict future power response during the prediction phase, thereby improving dynamic performance under active and reactive power step conditions.

[0027] The input terminals of the midpoint potential prediction and balancing module are connected to the DC-side upper and lower capacitor voltage sampling terminals and the AC-side three-phase current sampling output terminals, respectively. Its output terminal is connected to the unified cost function and optimal control quantity selection module. This module receives the DC bus voltage. Midpoint potential deviation and AC side current Based on the direction of midpoint potential shift and phase current, the influence of candidate control quantities on the charging and discharging states of the upper and lower capacitors is evaluated, and a midpoint potential balance evaluation item is formed. This module enables the controller to simultaneously consider the midpoint potential constraint when selecting the optimal control quantity, thus avoiding a continuous shift in the DC side midpoint potential during dynamic power adjustment.

[0028] The unified cost function and optimal control quantity selection module is used to comprehensively evaluate various control objectives and output the optimal control command. This module receives current tracking evaluation items. Active power evaluation item Reactive power evaluation item Midpoint potential evaluation item and control smoothing evaluation items Candidate control quantities are compared using a unified cost function, and the control quantity with the smallest cost function is selected as the optimal output for the current sampling period. This module ultimately generates three-phase control voltage commands. , , It is then fed into the three-level PWM drive module.

[0029] The input of the three-level PWM drive module is connected to the output of the unified cost function and optimal control quantity selection module, and its output is connected to the drive terminals of the switching devices of each bridge arm of the T-type three-level main circuit. After receiving the three-phase control voltage command, the module generates PWM drive pulses for the corresponding bridge arm switching devices, controlling the on and off of each power switch in the three-phase bridge arm, so that the main circuit outputs the three-phase voltage corresponding to the optimal control quantity. Thus, the control system completes the prediction, optimization, and drive process within one sampling cycle, and re-acquires the state quantities in the next sampling cycle, cyclically executing the above control process.

[0030] like Figure 1 As shown, this invention provides a power midpoint coordinated control method for a T-type three-level virtual synchronous machine, applicable to energy storage converter systems containing DC split capacitors, T-type three-level bridge arms, AC-side filter inductors, and grid-connected interfaces. Based on the virtual synchronous machine power outer loop, this method integrates power feedforward current reference generation, model predictive current control, active / reactive power prediction and evaluation, and DC-side midpoint potential balance compensation into the same control flow, achieving coordinated control of power dynamic response and midpoint potential suppression. The specific steps are as follows.

[0031] S1. Data Acquisition and Coordinate Transformation: The AC side operating status and DC side neutral point potential status of the T-type three-level energy storage converter are collected, and the voltage and current signals in the three-phase stationary coordinate system are converted into control variables in the dq synchronous rotating coordinate system.

[0032] During the current sampling period k, the three-phase current on the AC side is collected. , , Phase voltage obtained by converting the three-phase voltage or line voltage on the grid-connected side. , , Simultaneously, the voltage of the capacitor on the DC side is collected. Lower capacitor voltage and DC bus voltage Calculate the midpoint potential deviation based on the voltages of the upper and lower capacitors: ; In the formula, This indicates the degree of potential imbalance at the DC side midpoint. and These represent the voltages of the upper and lower split capacitors on the DC side, respectively.

[0033] Synchronization angle signal output from the VSG power outer loop Transform the three-phase voltage and current into the dq coordinate system , , , .in, , These are the d-axis and q-axis components of the grid-connected voltage. , These are the d-axis and q-axis components of the AC side current. These variables are input to the power calculation module, the power feedforward current reference generation module, the multi-step model prediction module, and the midpoint potential prediction and balancing module, respectively.

[0034] S2. Output power calculation and VSG power outer loop adjustment: The converter output power is calculated based on the voltage and current in the dq coordinate system, and the reference voltage direction and synchronization angle information are generated through the VSG power outer loop.

[0035] Power calculation module according to , , , Calculate the actual active power P and reactive power Q: ; ; In the formula, P represents the active power output by the converter to the AC side, and Q represents the reactive power output by the converter to the AC side. , These are the d-axis and q-axis components of the grid-connected voltage in the synchronous rotating coordinate system, respectively. , These are the d-axis and q-axis components of the AC side current in the synchronous rotating coordinate system, respectively.

[0036] The VSG power outer loop control module receives active power reference values. Reactive power reference value Rated angular frequency Rated voltage and feedback power , The active power channel is used to adjust the angular frequency and phase angle of the virtual synchronizer, while the reactive power channel is used to adjust the output voltage amplitude or voltage reference component. Its power deviation can be expressed as: ; ; The active power outer loop can be described using the motion equations of a virtual synchronous machine rotor: ; In the formula, J is the virtual moment of inertia, and D is the virtual damping coefficient. Output angular frequency for the virtual synchronizer. The rated angular frequency, This is the synchronous phase angle. This relationship is used to simulate the inertial and damped response of a synchronous generator under changes in active power.

[0037] The reactive power outer loop can generate voltage regulation based on reactive power deviation: ; in, This is the reactive power voltage regulation coefficient.

[0038] The VSG power outer loop ultimately outputs a synchronization angle signal. and the reference voltage component in the dq coordinate system , It is used to characterize the voltage reference direction of the VSG outer loop output, and also serves as the input of the subsequent power feedforward current reference generation module, so that the power outer loop, current reference and predictive control loop maintain the same synchronous coordinate reference.

[0039] S3. Obtain the reference value of the power feedforward current: Based on the active and reactive power commands and the reference voltage direction of the VSG output, the d-axis and q-axis current reference values ​​that the model predictive controller can directly track are generated.

[0040] Power feedforward current reference generation module receives , , , and measured voltage , First, calculate the magnitude of the reference voltage vector: ; And obtain the direction of the reference voltage: ; In the formula, For the reference voltage vector magnitude, , These are the directional coefficients of the reference voltage vector on the d-axis and q-axis, respectively.

[0041] Then calculate the measured voltage amplitude: ; Based on the mapping relationship between power and current, the reference value of the power feedforward current is obtained: ; In the formula, and These represent the d-axis and q-axis current reference values, respectively. In actual control, to avoid control instability caused by abnormal voltage sampling, sudden changes in power commands, or excessive current commands, the following can be used: , Power deviation and current reference value settings include limiting and protection mechanisms.

[0042] S4. Multi-step model prediction based on candidate control variables: Based on the current current and voltage status and candidate control voltage, predict the trend of AC side current change during future control cycles and provide a predictive quantity for unified cost function evaluation.

[0043] Multi-step model prediction module received power feedforward current reference value , Actual current , Grid-connected voltage , and synchronization angle signal First, construct the base control voltage based on the current tracking error: ; In the formula, , The basic control voltage in the dq coordinate system; The equivalent resistance of the AC side filter branch; For AC side filter inductance; This is the current error adjustment coefficient; For coupling feedforward coefficients. The aforementioned basic control voltage consists of grid-connected voltage compensation term, filter resistor voltage drop compensation term, current error feedback term, and dq-axis coupling compensation term.

[0044] Construct a finite set of candidate control variables near the base control voltage. For each set of candidate control voltages... , The current in the next sampling period is predicted based on the AC side filtering model: ; In the formula, The sampling period is , These are the predicted currents along the d-axis and q-axis for the next sampling period, respectively. If the prediction step size is n, then the following is recursively derived based on the above one-step prediction: ; S5. Predicted active and reactive power errors: The active and reactive power at future times are calculated based on the predicted current and grid voltage, and a power error term is formed for evaluating candidate control quantities.

[0045] Specifically, the power error prediction module receives the predicted current obtained from multi-step model prediction based on candidate control quantities. , And combined with the current or predicted grid-connected voltage components , Calculate the predicted active power and predicted reactive power: ; In the formula, Let the predicted active power be at the (k+n)th time in the future. This represents the predicted reactive power at time k+n. To ensure the power prediction target aligns with the power feedforward current reference, the predictive controller can also calculate the internal power reference value based on the current reference value generated from the acquired power feedforward current reference value. ; In the formula, and These are the active and reactive power reference values ​​used internally by the model predictive controller. They can also be obtained from external power commands. , It can be given directly, or it can be obtained from the power feedforward current reference value. , This is calculated based on the grid connection voltage. In practical applications, it can be used for... , Set a limit to prevent abnormal power commands from causing excessive control input.

[0046] Based on the deviation between the predicted power and the reference power, active power error evaluation terms and reactive power error evaluation terms are constructed: ; In the formula, This is the normalized reference value for active power. This serves as the baseline value for reactive power normalization. Normalization avoids the impact of different dimensions and numerical scales on the unified cost function.

[0047] Candidate control variables are evaluated not only based on current error but also on their impact on future active and reactive power. Therefore, when active or reactive power surges or load disturbances occur, the predictive controller can prioritize control variables that facilitate rapid power convergence, thereby improving power tracking performance.

[0048] S6. Midpoint Potential Prediction and Balance Evaluation: Based on the voltage deviation of the upper and lower capacitors on the DC side and the direction of the current on the AC side, the influence of the candidate control quantity on the midpoint potential balance is evaluated, and a coordinated compensation quantity that is beneficial to suppressing the midpoint offset is generated.

[0049] The midpoint potential prediction and balancing module receives the DC bus voltage. upper capacitor voltage Lower capacitor voltage and AC three-phase current , , First, calculate the DC side midpoint potential deviation: ; In the formula, This indicates the degree of voltage imbalance between the upper and lower split capacitors. When... When, it means the voltage of the upper capacitor is higher than the voltage of the lower capacitor; when When, it indicates that the voltage of the lower capacitor is higher than the voltage of the upper capacitor; when When the voltage of the upper and lower capacitors is in equilibrium, it indicates that the voltages of the upper and lower capacitors are balanced.

[0050] To facilitate unified evaluation within the controller, the midpoint potential deviation can be normalized: ; In the formula, Let be the normalized threshold for the midpoint potential deviation, and be the limiting function. Based on the normalized midpoint potential deviation and the three-phase current directions, construct the three-phase midpoint potential coordination compensation: ; In the formula, Here, sgn() is the midpoint potential compensation coefficient, and sgn() is the sign function. , , These are the midpoint potential coordination compensation values ​​for phases A, B, and C, respectively. They are used to correct the final three-phase control voltage command, enabling the controller to adjust the charging and discharging trends of the upper and lower capacitors according to the direction of the midpoint potential deviation and the direction of the phase current.

[0051] Simultaneously, a midpoint potential evaluation term is set in the unified cost function: ; In the formula, This serves as the benchmark value for evaluating the midpoint potential deviation. This evaluation term penalizes voltage imbalance between the upper and lower capacitors, ensuring that the selection of candidate control quantities no longer only considers current and power tracking, but also simultaneously considers the DC-side midpoint potential balance. The midpoint potential deviation participates in the candidate control quantity optimization as a cost function evaluation term, and also acts on the output voltage command through three-phase coordinated compensation, thereby suppressing continuous voltage deviation of the upper and lower capacitors on the DC side during power dynamic regulation and improving the midpoint potential stability of the T-type three-level converter.

[0052] S7. Evaluation of Unified Cost Function and Selection of Optimal Control Quantity: Considering current tracking error, power prediction error, midpoint potential deviation, and control smoothing constraints, the optimal control quantity for the current sampling period is selected. The unified cost function and optimal control quantity selection module receives evaluation quantities output from the multi-step model prediction module, power error prediction module, and midpoint potential prediction and balancing module. The current tracking evaluation term can be expressed as: ; The control smoothing evaluation term can be expressed as: ; In the formula, Used to evaluate the deviation between the predicted current and the reference current. Used to limit sudden changes in control quantities and improve the quality of output voltage and current.

[0053] Combining the above evaluation terms, the unified cost function is expressed as: ; In the formula, , , , , These are the weighting coefficients for current tracking, active power tracking, reactive power tracking, neutral point potential balance, and control smoothing constraints, respectively. The controller iterates through the candidate control quantities, selects the control quantity that minimizes J as the optimal control quantity for the current sampling period, and outputs the three-phase control voltage command. , , .

[0054] S8, Three-Level PWM Drive and Loop Update: The optimal control quantity is converted into a PWM drive signal for the T-type three-level bridge arm, and the above control process is repeated in the next sampling cycle.

[0055] The three-level PWM drive module receives the three-phase control voltage command output by the unified cost function and the optimal control quantity selection module. , , This generates drive signals for the A, B, and C phase bridge arm switching devices, which control them respectively. to , to , to The T-type three-level converter outputs three-phase voltage under PWM drive. , , It then outputs electrical energy to the power grid or load through the AC side filtering stage.

[0056] In the next sampling period, the system re-acquires AC side voltage, current, DC side capacitor voltage, and power reference values, repeating steps S1 to S8.

[0057] Therefore, this invention achieves the synergistic effect of VSG power outer loop regulation, power feedforward current reference generation, multi-step model prediction, power error evaluation and midpoint potential balance control within the same control cycle, thereby enabling the T-type three-level energy storage converter to maintain good power tracking capability, output current quality and DC side midpoint potential balance under power step, reactive power disturbance and load change conditions.

[0058] To verify the feasibility of the control method of the present invention, a T-type three-level virtual synchronous machine control system was constructed according to Table 1. The system parameters are only used to illustrate this embodiment, and the described embodiments are only some embodiments of the present invention, not all embodiments.

[0059] Within each sampling period, the three-phase current on the AC side, the three-phase voltage on the grid-connected side, and the voltages of the upper and lower capacitors are first collected, and then obtained through coordinate transformation. , , , The actual active and reactive power are calculated; then, the reference voltage component and synchronization angle are generated by the VSG power outer loop, and obtained through power feedforward mapping. , .

[0060] Table 1

[0061] Candidate offsets are superimposed along the d-axis and q-axis, centered on the base control voltage. In this embodiment, the offset coefficients are all set to -2, -1, 0, 1, and 2, resulting in 25 sets of candidate control voltages. For each set of candidate control voltages, the predicted current, predicted active power, and predicted reactive power for the next sampling period are calculated sequentially, along with current tracking, power error, control smoothing, and midpoint potential evaluation terms. The candidate control voltage with the minimum unified cost function is selected as the optimal control quantity for the current period.

[0062] The optimal dq-axis control voltage is converted into a three-phase base voltage. Based on the voltage difference between the upper and lower capacitors and the neutral point potential compensation amount superimposed on the three-phase current direction, a three-phase control voltage command is obtained. This three-phase control voltage command is then modulated by a three-level PWM to generate the bridge arm drive signal. The state variables are re-acquired in the next sampling cycle, and the above process is repeated.

[0063] Under steady-state conditions, with the active power reference value set to 100 kW and the reactive power reference value set to 0 kvar, the steady-state operating conditions are as follows: Figure 3 The T-type three-level energy storage converter shown outputs a basically symmetrical three-phase AC current waveform, with a phase difference of approximately 120° between the three phases. The current amplitude is stable, and no obvious distortion or phase imbalance is observed. Figure 4The active power of the converter shown can gradually increase from the start-up stage and stabilize near the given value, while the reactive power remains basically near zero; Figure 5 As shown, the DC side midpoint voltage deviation was limited to a small range of fluctuation after startup, without any sustained offset or divergence. Figure 6 The spectrum analysis of the AC side current under steady-state operation is shown. The fundamental frequency is 50Hz and the total harmonic distortion (THD) of the current is 1.34%, which verifies that this embodiment can maintain a low current harmonic content while achieving power control and neutral point potential balance.

[0064] Under the active power step condition, the system initially operates at 100 kW, and at 0.25 s, the active power reference value is adjusted to 130 kW, while the reactive power reference value remains at 0 kvar. The resulting power response is as follows: Figure 7 As shown, the active power rises rapidly from the initial setpoint to near the new power reference value and remains stable after a brief adjustment, while the reactive power remains essentially near zero during the active power step change; Figure 8 The diagram demonstrates reactive power tracking when reactive loads are applied or reactive power demand changes, while maintaining the stability of active power output. It verifies that the power feedforward current reference generation module can improve the dynamic response capability when active power commands change and reduce the coupling effect between active and reactive power channels.

[0065] Under reactive load conditions, the system operates with 100 kW of active power. An inductive load is introduced in 0.25 s, and the reactive power reference value is adjusted from 0 kvar to 20 kvar. In this embodiment, the resistance and inductance of the inductive load are 0.1 Ω and 23 mH, respectively. The resulting power response is as follows: Figure 9 As shown, under reactive load disturbance conditions, reactive power rises rapidly from near zero and gradually stabilizes near the new reactive power demand, while active power only experiences a short-term disturbance before returning to stability. This indicates that this embodiment can achieve reactive power tracking when reactive load is applied or reactive power demand changes, while maintaining the stability of active power output.

[0066] Under the condition of initial voltage imbalance at the midpoint, the initial voltage difference between the upper and lower capacitors is set to 30 V, while the rest of the control flow remains unchanged. The resulting midpoint voltage deviation recovery process is as follows: Figure 10 As shown, the midpoint voltage deviation decreases rapidly from a large initial offset and eventually stabilizes within a small fluctuation range near zero.

[0067] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0068] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any minor modifications, equivalent substitutions, and improvements made to the above embodiments based on the technical essence of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for power midpoint coordinated control of a T-type three-level virtual synchronous machine, characterized in that, Includes the following steps: S1. Data Acquisition and Coordinate Transformation: Acquire AC side voltage and current and DC side capacitor voltage information of T-type three-level converter, and obtain the actual voltage and current in synchronous rotating coordinate system through coordinate transformation; S2. Output power calculation and VSG power outer loop adjustment: Calculate the actual active power and reactive power based on the collected AC side voltage and current, and generate reference voltage and synchronization angle information through the VSG power outer loop; S3. Obtain the power feedforward current reference value for the next control cycle: Based on the power command and the reference voltage direction of the VSG output, convert the active and reactive power targets into d-axis current reference values ​​and q-axis current reference values ​​for model prediction and evaluation in step S4. S4. Multi-step model prediction based on candidate control quantities: Using the power feedforward current reference value and the current current state as input, predict the current change in future sampling periods under the action of candidate control quantities. S5. Predicted active and reactive power error: Calculate the active and reactive power at future times based on the predicted current and form the corresponding power error evaluation item. S6. Midpoint potential prediction and balance control: Evaluate and suppress the midpoint potential deviation of the T-type three-level converter based on the DC side upper and lower capacitor voltage deviation and AC side current direction. S7. Unified cost function evaluation and optimal control quantity selection: Taking into account current tracking, power prediction error, midpoint potential deviation and control smoothing constraints, the optimal control quantity for the current sampling period is selected. S8, Three-level PWM drive and cyclic update: Convert the optimal control voltage command into a T-type three-level bridge arm switching signal, and repeat the above control process in the next sampling cycle.

2. The power midpoint collaborative control method for a T-type three-level virtual synchronous machine according to claim 1, characterized in that, The specific method for step S2 is as follows: according to , , , Calculate the actual active power P and reactive power Q: ； ； In the formula, P represents the active power output by the converter to the AC side, and Q represents the reactive power output by the converter to the AC side. , These are the d-axis and q-axis components of the grid-connected voltage in the synchronous rotating coordinate system, respectively. , These are the d-axis and q-axis components of the AC side current in the synchronous rotating coordinate system, respectively. The active power channel is used to adjust the angular frequency and phase angle of the virtual synchronous machine, while the reactive power channel is used to adjust the output voltage amplitude or voltage reference component. Their power deviation can be expressed as: ； ； In the formula, Active power reference value This is a reference value for reactive power. The active power outer loop is described by the motion equations of a virtual synchronous machine rotor: ； In the formula, J is the virtual moment of inertia, and D is the virtual damping coefficient. Output angular frequency for the virtual synchronizer. The rated angular frequency, Synchronization phase angle; The reactive power outer loop generates voltage regulation based on reactive power deviation: ； in, This is the reactive power voltage regulation coefficient.

3. The power midpoint collaborative control method for a T-type three-level virtual synchronous machine according to claim 1, characterized in that, The specific method for step S3 is as follows: First, calculate the magnitude of the reference voltage vector: ； And obtain the direction of the reference voltage: ； In the formula, For the reference voltage vector magnitude, , These are the directional coefficients of the reference voltage vector on the d-axis and q-axis, respectively; Then calculate the measured voltage amplitude: ； Based on the mapping relationship between power and current, the reference value of the power feedforward current is obtained: ； In the formula, and These represent the reference values ​​for the d-axis and q-axis currents, respectively.

4. The power midpoint collaborative control method for a T-type three-level virtual synchronous machine according to claim 1, characterized in that, The specific method for step S4 is as follows: First, based on the d-axis current reference value and q-axis current reference value obtained in step S3, and the d-axis actual current and q-axis actual current obtained in step S1, calculate the d-axis current tracking error and q-axis current tracking error respectively: ； In the formula, , The base control voltage in the dq coordinate system consists of grid-connected voltage compensation term, filter resistor voltage drop compensation term, current error feedback term and dq axis coupling compensation term. The equivalent resistance of the AC side filter branch; For AC side filter inductance; This is the current error adjustment coefficient; These are the coupling feedforward coefficients; Centered on the base control voltage, a finite number of discrete offsets are superimposed along the d-axis and q-axis according to a preset voltage step size, forming a finite set of candidate control quantities. For each set of candidate control voltages... , The current in the next sampling period is predicted based on the AC side filtering model: ； In the formula, The sampling period is , The predicted currents for the d-axis and q-axis of the next sampling period are respectively; if the prediction step size is n, then the following is recursively obtained based on the above one-step prediction: 。 5. The power midpoint collaborative control method for a T-type three-level virtual synchronous machine according to claim 1, characterized in that, The specific method for step S5 is as follows: Based on the predicted current , And combined with the current or predicted grid-connected voltage components , Calculate the predicted active power and predicted reactive power: ； In the formula, The predicted active power at time k+n is... This is the predicted reactive power at time k+n; the predictive controller can also calculate the internal power reference value based on the current reference value generated from the acquired power feedforward current reference value. ； In the formula, and These are the active and reactive power reference values ​​used internally by the model predictive controller; Based on the deviation between the predicted power and the reference power, active power error evaluation terms and reactive power error evaluation terms are constructed: ； In the formula, This is the normalized reference value for active power. This is the normalized reference value for reactive power.

6. The power midpoint collaborative control method for a T-type three-level virtual synchronous machine according to claim 1, characterized in that, The specific method for step S6 is as follows: Calculate the DC side midpoint potential deviation: ； In the formula, Indicates the degree of voltage imbalance between the upper and lower split capacitors; when When, it means the voltage of the upper capacitor is higher than the voltage of the lower capacitor; when When, it indicates that the voltage of the lower capacitor is higher than the voltage of the upper capacitor; when When the voltage of the upper and lower capacitors is balanced, it indicates that the voltages of the upper and lower capacitors are in equilibrium. The midpoint potential deviation is normalized. ； In the formula, is the normalized threshold for midpoint potential deviation, and sat() is the amplitude limiting function; Based on the normalized midpoint potential deviation and the direction of the three-phase current, construct the three-phase midpoint potential coordination compensation: ； In the formula, Here, sgn() is the midpoint potential compensation coefficient, and sgn() is the sign function. , , These are the midpoint potential coordination compensation amounts for phases A, B, and C, respectively. Simultaneously, a midpoint potential evaluation term is set in the unified cost function: ； In the formula, This serves as the benchmark value for evaluating midpoint potential deviation.

7. The power midpoint collaborative control method for a T-type three-level virtual synchronous machine according to claim 1, characterized in that, The specific method for step S7 is as follows: The current tracking evaluation item can be expressed as: ； The control smoothing evaluation term can be expressed as: ； In the formula, Used to evaluate the deviation between the predicted current and the reference current. Used to limit sudden changes in control quantities and improve the quality of output voltage and current; Combining the above evaluation terms, the unified cost function is expressed as: ； In the formula, , , , , These are the weighting coefficients for current tracking, active power tracking, reactive power tracking, neutral point potential balance, and control smoothing constraints, respectively. The controller iterates through the candidate control quantities, selects the control quantity that minimizes J as the optimal control quantity for the current sampling period, and outputs the three-phase control voltage command. , , .

8. The power midpoint collaborative control method for a T-type three-level virtual synchronous machine according to claim 1, characterized in that, The specific method for step S8 is as follows: The three-phase control voltage command obtained in step S7 is input into the three-level PWM modulation stage to generate drive signals for the A, B, and C three-phase bridge arm switching devices. Based on the drive signals, the switch devices of each bridge arm are controlled to turn on or off, so that the T-type three-level converter outputs the corresponding three-phase voltage, and outputs electrical energy to the grid or load through the AC side filtering stage. In the next sampling period, the system re-acquires AC side voltage, current, DC side capacitor voltage, and power reference values, repeating steps S1 to S8.

9. A T-type three-level virtual synchronous machine power midpoint cooperative control system, used to implement the control method according to any one of claims 1-7, characterized in that, It includes a data acquisition and coordinate transformation module, a power calculation module, a VSG power outer loop control and power feedforward reference generation module, a multi-step model prediction module, a power error prediction module, a midpoint potential prediction and balancing module, a unified cost function and optimal control quantity selection module, and a three-level PWM drive module. The data acquisition and coordinate transformation module is connected to the power calculation module. The power calculation module is connected to the VSG power outer loop control and power feedforward reference generation module. The VSG power outer loop control and power feedforward reference generation module is connected to the multi-step model prediction module. The multi-step model prediction module is connected to the power error prediction module. The power error prediction module, midpoint potential prediction and balancing module are both connected to the unified cost function and optimal control quantity selection module. The unified cost function and optimal control quantity selection module is connected to the three-level PWM drive module.

Images