A parameter adaptive MPC-VSG island microgrid frequency stability control method

Abstract

The application discloses a parameter self-adaptive MPC-VSG island micro-grid frequency stability control method and belongs to the technical field of power system operation control. The method takes virtual rotational inertia J and virtual damping coefficient D as state variables into the optimization framework of MPC (model predictive control), and constructs a variable parameter rotational inertia dynamic model. By solving a multi-objective cost function in each control cycle, J, D and output power P are integrated and optimized online, and the action cost of the energy storage system is included in the optimization target. The application solves the problem of fixed J and D parameters or rigid adjustment rule in the existing VSG (virtual synchronous generator) control, realizes the collaborative optimization of core parameters and power output, significantly improves the frequency stability and dynamic response performance of the island micro-grid, and takes into account the service life of the energy storage.

Description

Technical Field

[0001] This invention belongs to the field of power system operation and control technology, specifically relating to a virtual synchronous generator frequency stability control method based on parameter adaptive model predictive control, which is particularly suitable for islanded microgrid scenarios with a high proportion of renewable energy access. Background Technology

[0002] With the increasing penetration of intermittent and fluctuating renewable energy sources such as wind and solar power in microgrids, the generation-side structure of power systems is undergoing fundamental changes. Traditional synchronous generators are gradually being replaced by renewable energy generation units based on power electronic converters. However, most power electronic converters adopt a grid-following control strategy based on phase-locked loops (PLLs), and their output power is determined by upstream energy sources (such as solar power and wind speed), lacking autonomous response capabilities to system frequency changes. This means that the converter inherently lacks the rotational inertia and damping characteristics provided by the rotating mass of a synchronous generator, resulting in a sharp decline in the equivalent inertia and damping level of microgrids, especially those operating in islanded mode. When the system encounters disturbances such as sudden load changes or severe fluctuations in wind and solar power output, the system's frequency stability domain decreases, and the rate of frequency change increases, easily triggering severe frequency oscillations or even instability, seriously threatening the safe and reliable operation of the microgrid.

[0003] To address these challenges, Virtual Synchronous Generator (VSG) technology has emerged and is considered by academia and industry as a key technological path to improve the grid-friendly integration capability of renewable energy sources. The core idea of ​​VSG technology is to simulate the mathematical model and external operating characteristics of a synchronous generator within a converter using control algorithms. Its control typically embeds the rotor motion equations of the synchronous generator, where J is the virtual moment of inertia and D is the virtual damping coefficient. Through these equations, the VSG can dynamically adjust its active power output according to grid frequency deviations, thereby providing the system with an inertial response and damping support similar to that of a synchronous generator.

[0004] However, traditional VSG control schemes have revealed significant limitations in practical applications. The vast majority of schemes employ fixed parameters for J and D. While this design can operate stably under specific steady-state conditions, it struggles to adapt to the complex and dynamic conditions of microgrids. Specifically, when the system experiences power deficit, excessive inertia, while effectively reducing the initial frequency change rate, hinders rapid frequency recovery, leading to a prolonged frequency sag. Conversely, insufficient damping fails to effectively suppress oscillations during frequency recovery and may even trigger sustained power and frequency oscillations, threatening system stability. In other words, fixed parameters inherently contradict different disturbances, failing to achieve a dynamic optimal balance between suppressing frequency overshoot and accelerating frequency recovery.

[0005] To overcome the limitations of fixed parameters, researchers have proposed various adaptive VSG control strategies. For example, the values ​​of J and D are adjusted piecewise based on the magnitude of the frequency deviation or the sign of the rate of change of frequency. While these methods improve dynamic performance to some extent, they still rely on pre-defined, experience-based adjustment rules. The design of these rules heavily depends on expert experience, and their coverage is limited, making it difficult to accurately characterize the optimal parameter mapping relationship of the nonlinear system under all possible operating conditions. Furthermore, parameter switching at rule thresholds often leads to abrupt changes, which can themselves become internal disturbances, potentially triggering secondary oscillations in the system, contradicting the initial control objective. Therefore, these methods have inherent limitations in achieving global, smooth, and multi-objective collaborative optimization.

[0006] On the other hand, Model Predictive Control (MPC), as an advanced optimization control algorithm, has attracted widespread attention in the field of power electronics due to its powerful ability to handle multivariable and constrained optimization problems. Existing research has attempted to combine MPC with VSG (MPC-VSG). However, existing schemes often have limitations: they typically treat MPC as a higher-level optimizer, using only the active / reactive power output of the VSG as the control variable, while J and D are still considered fixed or externally given parameters that need to be determined by other lower-level rules. This architecture fails to incorporate J and D, the two core parameters determining the dynamic response of the VSG, into the optimization framework of MPC, resulting in limited optimization freedom for MPC. This prevents the fundamental realization of synergistic optimization of VSG inertia, damping characteristics, and power output, thus failing to fully realize the technical potential of VSG.

[0007] Furthermore, from a system configuration perspective, without an energy storage system, the VSG's power support in response to power disturbances relies entirely on passive changes in the DC-side voltage or primary energy sources at the front end, lacking active power regulation capabilities and having limited effectiveness in suppressing frequency fluctuations. While configuring an energy storage system (such as a battery or supercapacitor) can provide rapid power support, improper control strategies (such as overly aggressive frequency regulation) can lead to frequent charging and discharging of the energy storage units. This not only accelerates their aging and shortens their lifespan but may also cause them to shut down due to exceeding the state of charge (SOC) limit, thereby weakening system stability.

[0008] In summary, a significant gap exists in existing technologies: the lack of an advanced control strategy capable of integrating, online, and rolling optimization of the virtual moment of inertia J, virtual damping coefficient D, and output power P of a VSG. Particularly in islanded microgrid scenarios incorporating energy storage, there is an urgent need for a frequency stability control method that can collaboratively optimize the internal parameters (J, D) of the VSG with the external power command (P), while also considering the lifespan of energy storage, to address the severe challenges posed by the integration of high proportions of renewable energy. This invention aims to fill this technological gap. Summary of the Invention

[0009] The technical problem this invention aims to solve is to overcome the shortcomings of existing VSG control methods, which use fixed parameters or piecewise adjustments based on empirical rules for virtual rotational inertia J and virtual damping coefficient D, resulting in the inability to achieve dynamic optimal coordination of inertia, damping, and output power under complex dynamic conditions of microgrids. Furthermore, existing MPC-VSG schemes do not incorporate J and D into the optimization framework, limiting the degree of optimization freedom. This invention provides a parameter-adaptive MPC-VSG islanded microgrid frequency stability control method that integrates online rolling optimization of J, D, and output power P.

[0010] To achieve the above objectives, this invention provides a frequency stability control method for VSG islanded microgrids based on parameter adaptive model predictive control. The core idea of ​​this method is to construct a high-level control framework with multi-parameter collaborative optimization to achieve a comprehensive improvement in the system's dynamic performance. The technical solution adopted in this invention is as follows:

[0011] Step 1: Collect the grid angular frequency and three-phase voltage of the AC bus, obtain the voltage and current components in the rotating coordinate system through coordinate transformation, and calculate the electromagnetic power P output by the converter in real time. e and reactive power Q e ;

[0012] Step 2: Input the reactive power into the droop control loop, and generate a mechanical power reference value P through active-frequency droop control and reactive-voltage droop control. m and voltage reference value E;

[0013] Step 3: Incorporate the virtual moment of inertia J and the virtual damping coefficient D as state variables into the state-space equations to construct a dynamic model of variable-parameter moment of inertia that includes dynamic adjustment of J and D, so that J and D become dynamically adjustable optimization variables in the model predictive control (MPC) optimization framework.

[0014] Step 4: Within each control cycle, predict the system state at multiple future moments based on the variable parameter moment of inertia dynamic model. Quantify and weigh each predicted state using a multi-objective cost function and solve for the optimal control sequence. The multi-objective cost function comprehensively considers frequency deviation, power deviation, change in moment of inertia, and change in damping coefficient. Simultaneously, frequency fluctuation safety constraints and upper and lower bound constraints of J and D are introduced into the optimization solution, forming a constrained MPC optimization problem.

[0015] Step 5: Use the virtual moment of inertia obtained in Step 4 for optimization. Virtual damping coefficient And the power generated or absorbed by the energy storage system in the predicted time domain. (i.e., the difference between the mechanical power of the virtual synchronous generator in the predicted time domain and the reference value) The angular frequency at the current moment is calculated using the rotor inertia equation, and then the voltage control command is generated by integrating and combining the sine function with the voltage reference value.

[0016] Step Six: The voltage control command is corrected through a virtual impedance circuit, and then the voltage and current dual closed-loop control system outputs a PWM (pulse width modulation) drive signal to complete the converter control.

[0017] Preferably, the variable parameter moment of inertia dynamic model described in step three is as follows:

[0018]

[0019] in, This is the adjustment coefficient for the virtual moment of inertia J. Let J be the coefficient of inertia for the virtual moment of inertia. This is the adjustment coefficient for the virtual damping coefficient D. The inertia coefficient is the virtual damping coefficient D. Let J be the initial value of the virtual moment of inertia. The initial value of the virtual damping coefficient D; These are the rates of change of angular frequency, virtual moment of inertia, and virtual damping coefficient over time, respectively.

[0020] Preferably, the multi-objective cost function in step four is:

[0021]

[0022] The weighting coefficients for the corresponding parameters are Δω and ΔP. m ΔJ and ΔD are the differences between the predicted angular frequency, mechanical power, moment of inertia, and damping coefficient and their respective reference values ​​in the predicted time domain.

[0023] Preferably, the constraints of the constrained MPC optimization problem in step four include frequency fluctuation constraints, upper and lower limits of virtual moment of inertia J, and upper and lower limits of virtual damping coefficient D.

[0024] Preferably, the rotor inertia equation in step five is:

[0025] ;

[0026] This is a reference value for rated active power. This is the reference value for angular frequency.

[0027] Preferably, the coordinate transformation in step one is the Park transformation, which converts the three-phase voltage and current into the d-axis and q-axis components in a two-phase rotating coordinate system.

[0028] Preferably, the virtual impedance link in step six is ​​constructed by software definition to build a voltage correction mechanism based on real-time output current feedback, and the preset virtual resistance and inductance parameters are superimposed on the voltage control command to simulate the resistance-inductance voltage drop effect in the control loop.

[0029] Preferably, the energy storage system described in step five operates in conjunction with the virtual synchronous generator (VSG) to provide rapid power support during power disturbances by generating or absorbing power. The operational cost of energy storage is incorporated into the multi-objective cost function to avoid overuse and frequent charging and discharging of energy storage.

[0030] Compared with the prior art, the present invention has the following advantages and technical effects:

[0031] 1. Enhanced Dynamic Adaptability: By incorporating virtual moment of inertia J and virtual damping coefficient D as dynamic optimization variables into the MPC framework, online and real-time collaborative optimization of J, D, and P is achieved. The controller can adaptively match optimal parameters for different dynamic operating conditions (such as load step changes and wind and solar power fluctuations), avoiding control failure of fixed parameter models under complex operating conditions.

[0032] 2. Superior performance of multi-objective control: By establishing a multi-objective cost function, multiple control objectives such as frequency stability, power smoothness, and parameter stability are comprehensively considered, and the global optimal balance under different control requirements is achieved through flexible adjustment of weight coefficients, avoiding suboptimal problems caused by single-objective optimization.

[0033] 3. Balancing real-time performance and stability: The rolling optimization mechanism ensures that control commands can respond quickly to system changes, while the prediction process avoids potential fluctuation risks in advance, thus improving the reliability and adaptability of the grid-connected operation of the new energy power generation system.

[0034] 4. Synergistic optimization of energy storage lifespan: The operational costs of energy storage are incorporated into the optimization objectives. While providing rapid power support, the excessive use of energy storage and frequent charging and discharging are effectively avoided, thereby achieving synergistic optimization of frequency oscillation suppression and energy storage lifespan. Attached Figure Description

[0035] Figure 1 A simplified model diagram of the connection between the synchronous generator and the power grid;

[0036] Figure 2 This is the overall control block diagram;

[0037] Figure 3 A comparison curve of the output power of energy storage systems;

[0038] Figure 4 A comparison graph of the system frequency response;

[0039] Figure 5 A comparison curve of the changes in the rotational inertia parameter J;

[0040] Figure 6 This is a comparison curve of the changes in the damping coefficient D parameter. Detailed Implementation

[0041] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0042] like Figure 1 and Figure 2 As shown, the parameter-adaptive MPC-VSG islanded microgrid frequency stability control method of the present invention is applied to the grid-connected control of new energy power generation systems.

[0043] Step 1: Signal Acquisition and Processing

[0044] In actual operation, the phase-locked loop (PLL 5) first acquires the grid angular frequency of the AC bus. The three-phase voltages are transformed using Park transformation to obtain the voltage and current components in a two-phase rotating coordinate system. The actual electromagnetic power output by the converter is then obtained in real time by the power calculation module. and reactive power The power calculation is shown in formula (1):

[0045] (1)

[0046] In equation (1), These are the three-phase voltages after Park transformation, representing either the grid voltage or the stator voltage. Shaft (excitation component) and Projected components on the axis (torque component), These are the three-phase voltages after Park transformation and the grid-connected current at... shaft and Projected components on the axis.

[0047] Step Two: Sagging Control

[0048] Subsequently, reactive power The data is transmitted to droop control stage 8; droop control includes active-frequency droop control (P-droop control) and reactive-voltage droop control (Q-droop control), generating a mechanical power reference value. and voltage reference value As shown in formula (2):

[0049] (2)

[0050] in, This is a reference value for rated active power. This is the reference value for rated reactive power. This is the reference value for the rated voltage. ω is the angular frequency reference value, and all the above reference values ​​are set values; m is the active power-frequency droop coefficient, and n is the reactive power-voltage droop coefficient.

[0051] Step 3: Construction of the extended state-space model

[0052] MPC control module 7 receives data from the power calculation module. and from the sagging control link As input variables. The core is to construct a dynamic model of variable parameter moment of inertia based on the rotor inertia equation, that is, to incorporate the virtual moment of inertia J and the virtual damping coefficient D as state variables into the state space equation. The state space equation of the model predictive control (MPC) design of this invention is shown in formula (3):

[0053] (3)

[0054] in, This is the adjustment coefficient for the virtual moment of inertia J. Let J be the coefficient of inertia for the virtual moment of inertia. This is the adjustment coefficient for the virtual damping coefficient D. The inertia coefficient is the virtual damping coefficient D. Let J be the initial value of the virtual moment of inertia. Let D be the initial value of the virtual damping coefficient. These are the rates of change of angular frequency, virtual moment of inertia, and virtual damping coefficient over time, respectively.

[0055] Step 4: Rolling Optimization and Multi-Objective Cost Function

[0056] After completing the dynamic update of parameters and states, the method enters the rolling optimization phase. Within each control cycle, it predicts the system state at multiple future moments, including changes in angular frequency, mechanical power, moment of inertia, and damping coefficient, and quantifies and weighs these predicted states using preset multi-objective evaluation rules.

[0057] To quantitatively evaluate the control effect of the system, the following multi-objective cost function was designed, as shown in Equation (4):

[0058] (4)

[0059] In equation (4), These are the weighting coefficients corresponding to the respective parameters. , , , They represent the parameters at the i-th time starting from time k. , , , The difference between the value and its respective reference value; by reasonably setting the weighting coefficient, the frequency deviation, power deviation, and changes in rotational inertia and damping coefficient can be flexibly adjusted according to actual control needs.

[0060] In actual microgrid operating environments, frequency fluctuations are limited to a certain safe range due to constraints such as the capacity of power electronic equipment and the limitations of system protection devices. Based on this reality, the constrained MPC optimization problem can be described as the optimization solution problem of formula (5):

[0061] (5)

[0062] Rolling-optimized virtual rotational inertia With virtual damping coefficient The difference between the set reference value and the optimized value The difference between the mechanical power of the virtual synchronous generator in the predicted time domain and the reference value is obtained by summation calculation. This difference is generated or absorbed by the energy storage system of the isolated microgrid, that is, the power generated or absorbed by the energy storage system. .

[0063] Step 5: Rotor Inertia Equation

[0064] Virtual rotational inertia obtained from MPC after rolling optimization With virtual damping coefficient And the power generated or absorbed by the energy storage system in the predicted time domain. The angular frequency at the current moment is calculated using the rotor inertia equation, and the calculation formula is as follows;

[0065] (6);

[0066] The resulting voltage control command is then calculated using integration and a sine function with the voltage reference value E.

[0067] Step Six: Virtual Impedance and Voltage / Current Dual Closed-Loop Control

[0068] The virtual impedance loop is an equivalent output impedance reshaping strategy based on a control algorithm, which actively regulates the external characteristics of a power electronic converter through software definition. Its core mechanism lies in constructing a voltage correction mechanism based on real-time output current feedback, superimposing preset virtual resistance and inductance parameters onto the original reference voltage (voltage control command), thereby artificially simulating a specific resistance-inductance voltage drop effect within the control loop.

[0069] Then, a dual-loop voltage and current control system achieves decoupling and coordinated control of the system's macroscopic state variables and microscopic execution variables. In the outer loop control, the system uses the output voltage as the controlled object. A proportional-integral (PI) regulator is introduced to perform closed-loop correction of the steady-state error between the target reference voltage and the actual sampled voltage, thereby mapping the voltage deviation to the dynamic current command required by the inner loop. The inner loop serves as the system's fast follow-up subsystem, and its core task is to accurately track the inductor current or armature current at high frequency. Because the current inner loop has a control bandwidth much higher than the voltage outer loop, it can quickly suppress internal disturbances caused by grid fluctuations, load changes, etc. Finally, the dual-loop voltage and current control system outputs the PWM drive signal to complete the converter control.

[0070] Simulation verification experiment:

[0071] To fully verify the effectiveness and robustness of the proposed method, this study constructed a complete and high-fidelity microgrid simulation model in the MATLAB / Simulink simulation platform. The specific simulation experiment settings are as follows: after the system has been running stably for 0.5 seconds, the same sudden load power disturbance is applied to the microgrid to simulate common load change scenarios in actual operation. The experiment compares the control performance of the traditional MPC-VSG strategy and the A-MPC-VSG strategy proposed in this invention.

[0072] Simulation parameter settings: The new energy power input of both sets of experiments is kept constant at 50000W. When the simulation runs for 0.5s, the load-side power of both sets of experiments changes from 50000W to 60000W simultaneously.

[0073] Depend on Figure 3It can be seen that after a load disturbance occurs, the output power of the energy storage system under the traditional MPC-VSG strategy changes slowly; while the output power of the energy storage system using the A-MPC-VSG strategy performs better, and it tracks the load power changes and maintains frequency stability faster.

[0074] Depend on Figure 4 It can be seen that after a load disturbance occurs, the system frequency under the traditional MPC-VSG strategy deviates significantly, and the recovery process has obvious overshoot and a long adjustment time; while the system frequency fluctuation amplitude using the A-MPC-VSG strategy is significantly reduced.

[0075] Depend on Figure 5 and Figure 6 It can be seen that after load disturbance occurs, the system performance under the traditional MPC-VSG strategy still has a high room for improvement, while the A-MPC-VSG strategy can improve the system performance to a certain extent, with a more stable dynamic response and faster recovery to the rated value, showing stronger anti-interference ability and frequency support characteristics.

[0076] The simulation results above fully demonstrate that the A-MPC-VSG method proposed in this invention is significantly superior to the traditional MPC-VSG control method in improving the stability of microgrid frequency dynamic response and enhancing the system's frequency maintenance capability under load disturbances, and has superior control performance and engineering application potential.

Claims

1. A parameter-adaptive MPC-VSG islanded microgrid frequency stability control method, characterized in that, Includes the following steps: Step 1: Collect the angular frequency and three-phase current of the AC bus grid, obtain the voltage and current components in the rotating coordinate system through coordinate transformation, and calculate the electromagnetic power P output by the converter. e and reactive power Q e ; Step 2: Convert the reactive power Q e The input droop control loop generates a mechanical power reference value P through active-frequency and reactive-voltage droop control. m and voltage reference value E; Step 3: Incorporate the virtual moment of inertia J and the virtual damping coefficient D as state variables into the state-space equations to construct a dynamic model of variable parameter moment of inertia, so that J and D become dynamically adjustable optimization variables in the model predictive control (MPC) optimization framework. Step 4: Within each control cycle, predict the future system state based on the dynamic model, quantify the trade-offs by comprehensively considering the changes in frequency deviation, power deviation, moment of inertia and damping coefficient through a multi-objective cost function, and introduce frequency fluctuation and J and D upper and lower limit constraints to solve for the optimal control sequence. Step 5: Using the virtual rotational inertia from the optimal control sequence in Step 4 Virtual damping coefficient and the mechanical power difference of the virtual synchronous generator. That is, the power generated or absorbed by the energy storage system. The angular frequency at the current moment is calculated using the rotor inertia equation, and then the voltage control command is generated by integrating and combining the sine function with the voltage reference value E. Step 6: Correct the voltage control command through a virtual impedance loop, and output a pulse width modulation drive signal through voltage and current dual closed-loop control to complete the converter control.

2. The parameter-adaptive MPC-VSG islanded microgrid frequency stability control method according to claim 1, characterized in that, The variable parameter moment of inertia dynamic model described in step three is as follows: in, This is the adjustment coefficient for the virtual moment of inertia J. Let J be the coefficient of inertia for the virtual moment of inertia. This is the adjustment coefficient for the virtual damping coefficient D. The inertia coefficient is the virtual damping coefficient D. Let J be the initial value of the virtual moment of inertia. The initial value of the virtual damping coefficient D; These represent the rates of change of angular frequency, virtual moment of inertia, and virtual damping coefficient over time, respectively.

3. The parameter-adaptive MPC-VSG islanded microgrid frequency stability control method according to claim 2, characterized in that, The multi-objective cost function mentioned in step four is: The weighting coefficients for the corresponding parameters are Δω and ΔP. m ΔJ and ΔD are the differences between the predicted angular frequency, mechanical power, moment of inertia, and damping coefficient and their respective reference values ​​in the predicted time domain.

4. The parameter-adaptive MPC-VSG islanded microgrid frequency stability control method according to claim 3, characterized in that, The constraints of the constrained MPC optimization problem described in step four include frequency fluctuation constraints, upper and lower bound constraints on the virtual moment of inertia J, and upper and lower bound constraints on the virtual damping coefficient D.

5. The parameter-adaptive MPC-VSG islanded microgrid frequency stability control method according to claim 4, characterized in that, The rotor inertia equation in step five is: This is a reference value for rated active power. This is the reference value for angular frequency.

6. The parameter-adaptive MPC-VSG islanded microgrid frequency stability control method according to claim 5, characterized in that, The coordinate transformation described in step one is the Park transformation, which converts the three-phase voltage and current into the d-axis and q-axis components in a two-phase rotating coordinate system.

7. The parameter-adaptive MPC-VSG islanded microgrid frequency stability control method according to claim 6, characterized in that, The virtual impedance link described in step six is ​​constructed using a software-defined method to create a voltage correction mechanism based on real-time output current feedback. The preset virtual resistance and inductance parameters are superimposed on the voltage control command to simulate the resistance-inductance voltage drop effect within the control loop.

8. The parameter-adaptive MPC-VSG islanded microgrid frequency stability control method according to claim 7, characterized in that, The energy storage system described in step five operates in conjunction with the virtual synchronous generator (VSG). During power disturbances, it provides rapid power support by generating or absorbing power. The operational cost of energy storage is incorporated into the multi-objective cost function to avoid overuse and frequent charging and discharging of energy storage.

Images