Power system inertia control method and device based on model-free adaptive control strategy

Abstract

The invention relates to the technical field of smart power grids, in particular to a power system inertia control method and device based on a model-free adaptive control strategy. A dynamic model of the virtual synchronous generator is constructed, changes of the rotating speed and the rotor angle of the virtual synchronous generator are described, and dynamic adjustment of the frequency and the torque is achieved by controlling input power and voltage. According to the method, a generic model of a power system is established, modeling is carried out on the VSG by using a state space method, and prediction of future output is realized through multi-step prediction. The prediction capability enables the system to keep stable operation under the condition of changing load and power generation, and the online optimization algorithm is adopted to dynamically adjust and control the input increment and weighting parameters so as to optimize the system performance. According to the invention, through dynamic modeling and real-time optimization control, the stability of the power system is enhanced, and the fluctuation of frequency and voltage can be effectively handled. The multi-step prediction function enables the system to pre-judge the power demand in advance, thereby optimizing the control strategy and improving the power quality.

Description

Technical Field

[0001] The present invention relates to the field of smart grid technology, and specifically to a method and device for controlling inertia of a power system based on a model-free adaptive control strategy. Background Art

[0002] In the power system, inertia refers to the system's resistance to frequency changes, which is traditionally provided by large rotating equipment such as generators. However, with the rapid development of renewable energy such as wind and solar energy, the inertia level of the power system has generally decreased, which has weakened the stability of the system to a certain extent and led to increased frequency fluctuations. Therefore, how to effectively provide sufficient inertia support has become one of the hot topics in current power system research.

[0003] The existing technical inertia support mainly focuses on the following aspects:

[0004] (1) Traditional inertia compensation methods: Many researchers have proposed inertia compensation strategies based on traditional generator sets. These methods rely on modeling and analyzing detailed models of the power system to determine the scheduling and operation of the generator sets. Although these methods are effective to a certain extent, they require a large amount of real-time data and complex models, and are difficult to quickly adapt to system dynamic changes.

[0005] (2) Virtual inertia technology: As the proportion of renewable energy in power systems increases, virtual inertia technology has gradually become a research focus. This technology simulates the inertia characteristics of traditional generators by controlling the power output of renewable energy power generation equipment (such as wind power and photovoltaics). However, existing virtual inertia strategies often rely on mathematical models of the system, lack flexibility and adaptability, and are difficult to cope with complex power system dynamics.

[0006] (3) Adaptive control strategies: Some studies have explored the application of adaptive control strategies in inertia support. These methods use machine learning and control theory to try to adjust control parameters in real time without relying on detailed models. However, these strategies often require a large amount of training data and show instability in real power systems.

[0007] (4) Distributed control schemes: In recent years, distributed control schemes have gradually attracted attention, aiming to achieve system inertia compensation through multi-point coordinated control. These schemes usually rely on network communication and information sharing, but in practical applications, system delays and communication failures may lead to poor control effects.

[0008] In summary, although the existing technology has made some progress in providing inertia support, it still has problems such as strong model dependence, insufficient adaptability, and poor real-time performance. Therefore, it is an urgent need for current research to develop an inertia control method based on model-free adaptive control that can effectively improve the stability of low-inertia power systems. Summary of the invention

[0009] The purpose of the present invention is to provide a method and device for controlling the inertia of a power system based on a model-free adaptive control strategy, aiming to solve the stability challenges faced by the current power system under low inertia conditions, especially in the context of the gradual increase in the proportion of renewable energy such as wind energy and solar energy, and to propose new solutions for the inertia support of the power system. Its innovation focuses on the application of the model-free adaptive control strategy, aiming to improve the stability and adaptability of the low inertia power system.

[0010] The technical solution of the present invention is as follows:

[0011] A method for controlling inertia of a power system based on a model-free adaptive control strategy, characterized by comprising the following contents:

[0012] S1: Construct a virtual synchronous generator model:

[0013] The virtual synchronous generator speed dynamic equation can be expressed by the rotor inertia J and electrical power Pe as follows:

[0014]

[0015] Where D is the damping coefficient, ω and ω0 are the actual frequency and the rated frequency; Pm is the mechanical input power and;

[0016] The rate of change of the virtual synchronous generator rotor angle θ is related to the speed:

[0017] dt / dθ=ω (2), the stator voltage can be expressed as:

[0018] Vs=RI+Ldt / dI+Lωθ (3), where: Vs is the stator terminal voltage, R is the stator resistance, L is the stator inductance, and I is the current;

[0019] The electrical power Pe can be expressed as:

[0020] Pe=VsIcos(φ) (4), where φ is the phase difference between current and voltage;

[0021] S2: Determine the virtual synchronous generator control strategy:

[0022] The control strategy includes one or more of the following: frequency control, torque control, voltage control, active and reactive power control, and PWM control of the inverter;

[0023] S21: Build frequency control model

[0024] Tracking grid frequency changes by adjusting the mechanical input power Pm:

[0025] Pm=Pref+Kf(ω grid -ω) (5)

[0026] Where Kf is the frequency control gain.

[0027] S22: Establishing torque control model

[0028] To ensure system stability, the virtual synchronous generator can adjust the output torque according to the speed and rotor angle:

[0029] T = Tref - Kt (ω grid -ω) (6)

[0030] S23: Establishing voltage control model

[0031] The voltage regulation of VSG is achieved by simulating the excitation system of synchronous generator, usually using automatic voltage regulator (AVR). Its mathematical model can be expressed as:

[0032]

[0033] Where Vset is the voltage setting value, V is the output voltage, Kv is the voltage feedback gain, Ka is the excitation gain, and Vf is the excitation voltage

[0034] S24: Establish active and reactive power control model

[0035] Active power control is achieved by adjusting the output frequency of the inverter. The frequency regulation equation is:

[0036] ΔP=Kp(ωref-ωr)+Ki∫(ωref-ωr)dt (8) Wherein: ΔP is the active power compensation amount, Kp is the proportional gain, Ki is the integral gain, and ωref is the reference frequency.

[0037] Reactive power control is achieved by adjusting the output voltage amplitude of the inverter. The adjustment equation is:

[0038] ΔQ=Kq(Vref-V)+Ki∫(Vref-V)dt (8) Where: ΔQ is the reactive power compensation, Kq is the proportional gain, Ki is the integral gain, and Vref is the reference voltage

[0039] S25: Establish the PWM control model of the inverter

[0040] The pulse width modulation (PWM) control of the inverter is the key to realize the VSG function. The PWM control signal is determined by the active and reactive power regulation calculated by the above control equation.

[0041] S3: Establishing data-driven pan-functional estimation of power systems:

[0042] Based on S1 and S2, the input and output of the virtual synchronous generator are clarified, and the model-free generalized predictive control is implemented on the virtual synchronous generator. A general model is established to describe the dynamic behavior of the power system. The expression of the general model is:

[0043] y(t+j / t)=y(t+j-1 / t)+ψ(t)Δu(t) (10),

[0044] Where y(t)=[Pe(t),ω(t),V s (t),T(t)] T , y(t+j / t) is the j-step predicted value of the system output under the condition of time t; ψ(t) is the estimated value of the power system model parameter, which needs to be obtained through fitting; Δu(t) is the change in the control input, including five control quantities: frequency control, torque control, voltage control, active and reactive power control, and PWM control of the inverter;

[0045] S4: Multi-step Forecast

[0046] According to the model-free adaptive control theory, multi-step prediction is performed based on the power system universal model, and the prediction step number is set j = d, d+1, ..., d+p-1, we can get:

[0047] y(t+d / t)=y(t)+ψ(t)Δu(t)+ψ(t)[u(t-1)-u(td)]+dv(t)

[0048] (11)

[0049] Where Np is the time domain length of the control input;

[0050] S5: Online Optimization:

[0051] Setting the objective function and adding the control input increment constraint, the objective function can be expressed as:

[0052]

[0053] Where J is the total cost function; L(yj,yref,j) is the error loss function between the state output and the reference trajectory, y ref,j is the desired reference output; R(Δu k ) is the penalty term of the control input increment, which is used to limit the change of the control input; Δuk is the increment of the control input; Nu is the number of prediction steps of the control input, d is the index of the current time step, indicating the moment from which prediction and optimization start, and p is the number of future prediction steps, indicating the number of future time steps that need to be considered in the objective function. By minimizing the objective function, the performance of the current control strategy is evaluated.

[0054] In the control decision, adjust the weighted parameters to optimize the control parameters:

[0055] By calculating gradients, iterative optimization, and real-time updates, control parameters are adjusted to improve system performance; this adaptive optimization method can cope with dynamic changes in system status and ensure system stability under different load and power generation conditions, including:

[0056] (1) Calculate the gradient: By calculating the partial derivative of the system output y(t) with respect to the control parameter ψ, we can obtain the pseudo-derivative The calculation method is shown in formula (13):

[0057]

[0058] (2) Iterative optimization: After calculating the gradient of the control parameter, an iterative optimization algorithm is used to update the weighting coefficient to optimize the control effect. The objective function is shown in formula (14):

[0059]

[0060] (3) Real-time update: Apply the optimized parameters to control decisions in real time to achieve better control effects.

[0061] The power system inertia control method based on model-free adaptive control strategy is characterized in that the frequency control strategy in S2 is:

[0062] Tracking grid frequency changes by adjusting the mechanical input power Pm:

[0063] Pm=Pref+Kf(ω grid -ω) (5),

[0064] Where Kf is the frequency control gain.

[0065] The power system inertia control method based on model-free adaptive control strategy is characterized in that the torque control strategy in S2 is:

[0066] To ensure system stability, the virtual synchronous generator adjusts the output torque according to the speed and rotor angle:

[0067] T = Tref - Kt (ω grid -ω) (6),

[0068] The power system inertia control method based on model-free adaptive control strategy is characterized in that the voltage control strategy in S2 is:

[0069] The voltage regulation of the virtual synchronous generator VSG is achieved by simulating the excitation system of the synchronous generator, using an automatic voltage regulator AVR, whose mathematical model can be expressed as:

[0070]

[0071] Where Vset is the voltage setting value, V is the output voltage, Kv is the voltage feedback gain, Ka is the excitation gain, and Vf is the excitation voltage.

[0072] The power system inertia control method based on model-free adaptive control strategy is characterized in that the active and reactive power control strategies in S2 are:

[0073] Active power control is achieved by adjusting the output frequency of the inverter. The frequency adjustment equation is:

[0074] ΔP=Kp(ωref-ωr)+Ki∫(ωref-ωr)dt (8), where: ΔP is the active power compensation, Kp is the proportional gain, Ki is the integral gain, and ωref is the reference frequency;

[0075] Reactive power control is achieved by adjusting the output voltage amplitude of the inverter. The adjustment equation is:

[0076] ΔQ=Kq(Vref-V)+Ki∫(Vref-V)dt (9), where: ΔQ is the reactive power compensation amount, Kq is the proportional gain, Ki is the integral gain, and Vref is the reference voltage.

[0077] The power system inertia control method based on model-free adaptive control strategy is characterized in that the PWM control strategy of the inverter in S2 is:

[0078] The pulse width modulation (PWM) control of the inverter is the key to realizing the function of the virtual synchronous generator (VSG). The PWM control signal is determined by the active and reactive power regulation values ​​calculated by the control equation.

[0079] The power system inertia control method based on model-free adaptive control strategy is characterized in that the virtual synchronous generator model in S1 is implemented by the following steps:

[0080] Step 1. Initialize parameters: set the moment of inertia J, damping coefficient D, voltage setting value Vset, frequency setting value ωref, and control gains Kp, Ki, Kq;

[0081] Step 2. Calculate the torque: Based on the active power P e and electromagnetic torque T e Calculate the electromagnetic torque.

[0082]

[0083] Step 3. Solve the rotation equation: Use the Runge-Kutta method to solve the rotation equation (1) and obtain the real-time rotor angular velocity;

[0084] Step 4. Voltage regulation: Calculate the excitation voltage Vf based on the output voltage V and the reference voltage Vref;

[0085] Step 5. Power regulation: Calculate the active and reactive power compensation according to the frequency deviation ωref-ωr and the voltage deviation Vref-V;

[0086] Step 6. PWM control: convert the compensated active and reactive power into PWM control signals to adjust the output of the inverter;

[0087] Furthermore, the power system is regarded as a black box and model-free adaptive control is performed.

[0088] A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of a method for controlling inertia of an electric power system based on a model-free adaptive control strategy as described in any one of claims 1 to 8 are implemented.

[0089] The present invention relates to modeling and control technology of virtual synchronous generator (VSG), aiming to improve the stability and responsiveness of power system. Firstly, the dynamic model of virtual synchronous generator is constructed in the present invention to describe the change of its speed and rotor angle, and realize the dynamic regulation of frequency and torque by controlling input power and voltage. This process ensures that VSG can effectively track the change of grid frequency, so as to achieve higher power quality.

[0090] Secondly, the present invention establishes a pan-model of the power system, models the VSG using the state space method, and estimates future output through multi-step prediction. This prediction capability enables the system to maintain stable operation under changing load and power generation conditions.

[0091] In terms of control strategy, the present invention adopts an online optimization algorithm to dynamically adjust the control input increment and weighted parameters to optimize system performance. The objective function design takes into account the error between the system output and the predicted value to ensure the real-time and effectiveness of the control decision. Through the adaptive control optimization method, the patent realizes the continuous optimization of control parameters, thereby enhancing the application potential of virtual synchronous generators in renewable energy access and grid dispatching.

[0092] The present invention enhances the stability of the power system through dynamic modeling and real-time optimization control, especially when renewable energy is connected on a large scale, and can effectively cope with frequency and voltage fluctuations. Its multi-step prediction function enables the system to predict power demand in advance, thereby optimizing the control strategy and improving the quality of power. In addition, the application of adaptive control strategies improves the responsiveness and flexibility of the system, reduces scheduling costs, and provides a technical basis for the development of smart grids. These advantages not only support the integration of renewable energy, but also promote the reliability and efficiency of the power system, and have important practical value and broad application prospects.

[0093] In general, the present invention provides an innovative solution for the control and optimization of virtual synchronous generators, which has important practical application value.

[0094] The advantages of the present invention are:

[0095] (1) Improving power system stability: By providing effective inertia support, the power system's ability to resist frequency fluctuations is enhanced, thereby ensuring the reliability of power supply and system stability against the backdrop of an increasing proportion of renewable energy.

[0096] (2) Simplify the modeling process: Without relying on a detailed power system model, the modeling process of virtual inertia is refined so that the control strategy can respond to system dynamic changes more flexibly and efficiently, reducing dependence on complex models.

[0097] (3) Realize adaptive control: Propose an adaptive and flexible inertia support strategy that can adjust control parameters in real time to adapt to changes in the power system and improve the response speed and accuracy of the control system.

[0098] (4) Enhanced versatility of the control strategy: This enables the inertia control method to be widely applicable to different types of power systems, especially in environments with low inertia and high renewable energy penetration, providing a universal solution.

[0099] (5) Reduce implementation costs: By reducing the need for complex models and large amounts of real-time data, the implementation and maintenance costs of inertia control strategies can be reduced, thereby improving the feasibility of practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0100] Figure 1 Schematic diagram of the virtual synchronous generator model.

[0101] Figure 2 Provides schematic diagram of integrated inertia for VSG.

[0102] Figure 3 Schematic diagram of output prediction for model-free adaptive control.

[0103] Figure 4 It is a model-free adaptive control flow chart.

[0104] The specific steps are as follows:

[0105] See also Figure 4 A method for controlling inertia of a power system based on a model-free adaptive control strategy comprises the following steps:

[0106] (1) Establish a virtual synchronous generator model:

[0107] The mathematical model of virtual synchronous generator is as follows Figure 1 as follows:

[0108] The virtual synchronous generator speed dynamic equation can be expressed by the rotor inertia J and electrical power Pe as follows:

[0109]

[0110] , where D is the damping coefficient and ω0 is the rated frequency.

[0111] The rate of change of the virtual synchronous generator rotor angle θ is related to the speed:

[0112] dt / dθ=ω (2), the stator voltage can be expressed as:

[0113] Vs=RI+Ldt / dI+Lωθ (3),

[0114] Where: Vs is the stator terminal voltage, R is the stator resistance, L is the stator inductance, and I is the current.

[0115] The electrical power Pe can be expressed as:

[0116] Pe=VsIcos(φ) (4),

[0117] Where φ is the phase difference between current and voltage.

[0118] The virtual synchronous generator adopts the following control strategy:

[0119] a. Frequency control

[0120] Tracking grid frequency changes by adjusting the mechanical input power Pm:

[0121] Pm=Pref+Kf(ω grid -ω) (5), where Kf is the frequency control gain.

[0122] b. Torque control

[0123] To ensure system stability, the virtual synchronous generator can adjust the output torque according to the speed and rotor angle:

[0124] T = Tref - Kt (ω grid -ω) (6),

[0125] c. Voltage control

[0126] The voltage regulation of VSG is achieved by simulating the excitation system of synchronous generator, usually using automatic voltage regulator (AVR). Its mathematical model can be expressed as:

[0127]

[0128] Where Vset is the voltage setting value, V is the output voltage, Kv is the voltage feedback gain, Ka is the excitation gain, and Vf is the excitation voltage

[0129] d. Active and reactive power control

[0130] Active power control is achieved by adjusting the output frequency of the inverter. The frequency regulation equation is:

[0131] ΔP=Kp(ωref-ωr)+Ki∫(ωref-ωr)dt (8), where: ΔP is the active power compensation amount, Kp is the proportional gain, Ki is the integral gain, and ωref is the reference frequency.

[0132] Reactive power control is achieved by adjusting the output voltage amplitude of the inverter. The adjustment equation is:

[0133] ΔQ=Kq(Vref-V)+Ki∫(Vref-V)dt (9), where: ΔQ is the reactive power compensation amount, Kq is the proportional gain, Ki is the integral gain, and Vref is the reference voltage

[0134] e. PWM control of inverter

[0135] The pulse width modulation (PWM) control of the inverter is the key to realize the function of virtual synchronous generator (VSG). The PWM control signal is determined by the active and reactive power regulation calculated by the above control equation.

[0136] Comprehensive Model

[0137] Integrate the above parts to form a complete virtual synchronous generator VSG model. The model can be implemented through the following steps:

[0138] Step 1. Initialize parameters: set the moment of inertia J, damping coefficient D, voltage setting value Vset, frequency setting value ωref, and control gains Kp, Ki, Kq, etc.

[0139] Step 2. Calculate the torque: Calculate the electromagnetic torque based on the relationship between active power P and electromagnetic torque.

[0140] Step 3. Solve the rotation equation: Use numerical methods (such as the Euler method or the Runge-Kutta method) to solve the rotation equation and obtain the real-time rotor angular velocity.

[0141] Step 4. Voltage regulation: Calculate the excitation voltage Vf based on the output voltage V and the reference voltage Vref.

[0142] Step 5. Power regulation: Calculate the active and reactive power compensation according to the frequency deviation ωref-ωr and the voltage deviation Vref-V.

[0143] Step 6. PWM control: Convert the compensated active and reactive power into PWM control signals to adjust the output of the inverter.

[0144] Furthermore, the power system is regarded as a black box, and model-free adaptive control can be performed, such as Figure 2 as follows:

[0145] (2) Establishing a universal model of the power system

[0146] For simulation analysis, model-free generalized predictive control can be implemented for virtual synchronous generators. First, a general model is established to describe the dynamic behavior of the system. The expression of the general model is:

[0147] A general model is established to describe the dynamic behavior of the power system. The expression of the general model is:

[0148] y(t+j / t)=y(t+j-1 / t)+ψ(t)Δu(t) (10),

[0149] Where y(t)=[Pe(t),ω(t),V s (t),T(t)] T , y(t+j / t) is the j-step predicted value of the system output under the condition of time t; ψ(t) is the estimated value of the power system model parameter, which needs to be obtained through fitting; Δu(t) is the change in the control input, formula 5-9, including five control quantities including frequency control, torque control, voltage control, active and reactive power control, and PWM control of the inverter;

[0150] (3) Multi-step prediction

[0151] like Figure 3 As shown, according to the general model, multi-step prediction is performed, and the number of prediction steps is set j = d, d+1, ..., d+p-1, we can get:

[0152] y(t+d / t)=y(t)+ψ(t)Δu(t)+ψ(t)[u(t-1)-u(td)]+dv(t)

[0153] (11), where Np is the time domain length of the control input;

[0154] (4) Online Optimization

[0155] Setting the objective function and adding the control input increment constraint, the objective function can be expressed as:

[0156]

[0157] Where J is the total cost function; L(yj,yref,j) is the error loss function between the state output and the reference trajectory, yref,j is the desired reference output; R(Δu k ) is the penalty term of the control input increment, which is used to limit the change of the control input; Δuk is the increment of the control input; Nu is the number of prediction steps of the control input, d is the index of the current time step, indicating the moment from which the prediction and optimization start, and p is the number of future prediction steps, which indicates the number of future time steps that need to be considered in the objective function. The performance of the current control strategy is evaluated by minimizing the objective function.

[0158] (5) Adjust weighting parameters

[0159] In order to optimize the control performance, it is necessary to adjust the weighting coefficient parameters through the online optimization algorithm. In the control decision, adjust the weighting parameters to optimize the control parameters:

[0160] By calculating gradients, iterative optimization, and real-time updates, control parameters are adjusted to improve system performance; this adaptive optimization method can cope with dynamic changes in system status and ensure system stability under different load and power generation conditions, including:

[0161] (1) Calculate the gradient: By calculating the partial derivative of the system output y(t) with respect to the control parameter ψ, we can obtain the pseudo-derivative The calculation method is shown in formula (13):

[0162]

[0163] (2) Iterative optimization: After calculating the gradient of the control parameter, an iterative optimization algorithm is used to update the weighting coefficient to optimize the control effect. The objective function is shown in formula (14):

[0164]

[0165] (3) Real-time update: Apply the optimized parameters to control decisions in real time to achieve better control effects.

[0166] (4) Finally, the above optimization process is integrated into a closed-loop control system by combining the generalized predictive control method. In this system, the output prediction model interacts with the general model, and the control strategy is continuously optimized through the adaptive control algorithm, ultimately achieving the best control effect of the system.

[0167] Through the above steps, the universal model of the power system not only provides an effective description of the dynamic behavior of the system, but also provides a solid foundation for the implementation of model-free generalized predictive control, enabling the system to achieve efficient and stable operation in a complex power environment.

[0168] Application examples:

[0169] For example, in a regional power grid, a virtual synchronous generator (VSG) control system is deployed to optimize power dispatch and improve system stability. First, multiple distributed energy resources (such as wind farms, photovoltaic power plants, and energy storage systems) in the regional power grid are connected to the main grid through smart substations. For these distributed energy resources, a dynamic model is established to describe their power generation capacity, response characteristics, and relationship with the grid frequency.

[0170] Next, a real-time control strategy is designed to use VSG technology to dynamically adjust the output power and voltage of each distributed energy source according to the real-time load demand and frequency changes of the power grid to ensure the balance of supply and demand of the power grid. The system also uses a multi-step prediction algorithm to predict future power demand and renewable energy generation changes based on historical load data and meteorological information. This prediction result is used to optimize the dispatch plan and prepare for the operation of the power grid in advance.

[0171] During the implementation process, the VSG control system uses an online optimization algorithm to monitor the grid status in real time, including frequency, load and power quality, and adjusts control parameters based on these real-time data. The goal is to minimize the error between power generation and demand and improve the reliability and flexibility of the system.

[0172] Finally, by coordinating with the grid dispatch center, the performance of the VSG control system is regularly evaluated and its effects under different operating conditions are analyzed, including indicators such as frequency stability, power fluctuations and power quality. Through this series of measures, the regional power grid can more effectively integrate renewable energy, improve overall operating efficiency, enhance the anti-interference ability of the power grid, and promote the sustainable development of green energy.

Claims

1. A method for controlling inertia of a power system based on a model-free adaptive control strategy, characterized in that: Include the following: S1: Construct a virtual synchronous generator model: The virtual synchronous generator speed dynamic equation can be expressed by the rotor inertia J and electrical power Pe as follows: Where D is the damping coefficient, ω and ω0 are the actual frequency and the rated frequency; Pm is the mechanical input power and; The rate of change of the virtual synchronous generator rotor angle θ is related to the speed: dt / dθ=ω (2), the stator voltage can be expressed as: Vs=RI+Ldt / dI+Lωθ (3), where: Vs is the stator terminal voltage, R is the stator resistance, L is the stator inductance, and I is the current; The electrical power Pe can be expressed as: Pe=VsIcos(φ) (4), where φ is the phase difference between current and voltage; S2: Determine the virtual synchronous generator control strategy: The control strategy includes one or more of the following: frequency control, torque control, voltage control, active and reactive power control, PWM control of inverter; treating the power system as a black box and performing model-free adaptive control; S3: Establish a universal model of the power system: Based on S1 and S2, the input and output of the virtual synchronous generator are clarified, and the model-free generalized predictive control is implemented on the virtual synchronous generator. A general model is established to describe the dynamic behavior of the power system. The expression of the general model is: y(t+j / t)=y(t+j-1 / t)+ψ(t)Δu(t) (10), where y(t)=[Pe(t),ω(t),V s (t),T(t)] T , y(t+j / t) is the j-step predicted value of the system output under the condition of time t; ψ(t) is the estimated value of the power system model parameter, which needs to be obtained through fitting; Δu(t) is the change in the control input, including five control quantities: frequency control, torque control, voltage control, active and reactive power control, and PWM control of the inverter; S4: Multi-step Forecast Based on the model-free generalized prediction theory, multi-step prediction is performed according to the power system pan-model, and the prediction step number is set j = d, d+1, ..., d+p-1, and the following is obtained: y(t+d / t)=y(t)+ψ(t)Δu(t)+ψ(t)[u(t-1)-u(td)]+dv(t) (11) Where Np is the time domain length of the control input; S5: Online Optimization: Set the objective function and add the control input increment constraint. The objective function is expressed as: Where J is the total cost function; L(y j ,y ref,j ) is the error loss function between the state output and the reference trajectory, y ref,j is the desired reference output; R(Δu k ) is the penalty term of the control input increment, which is used to limit the change of the control input; Δu k is the increment of the control input; Nu is the number of prediction steps of the control input, d is the index of the current time step, indicating the moment from which prediction and optimization start, and p is the number of future prediction steps, indicating the number of future time steps that need to be considered in the objective function. By minimizing the objective function, the performance of the current control strategy is evaluated.

2. The power system inertia control method based on model-free adaptive control strategy according to claim 1 is characterized in that: In the control decision, adjust the weighted parameters to optimize the control parameters: By calculating gradients, iterative optimization, and real-time updates, control parameters are adjusted to improve system performance; this adaptive optimization method can cope with dynamic changes in system status and ensure system stability under different load and power generation conditions, including: (1) Calculate the gradient: By calculating the partial derivative of the system output y(t) with respect to the control parameter ψ, we can obtain the pseudo-derivative The calculation method is shown in formula (13): (2) Iterative optimization: After calculating the gradient of the control parameter, an iterative optimization algorithm is used to update the weighting coefficient to optimize the control effect. The objective function is shown in formula (14): (3) Real-time update: Apply the optimized parameters to control decisions in real time to achieve better control effects.

3. The method for controlling inertia of a power system based on a model-free adaptive control strategy according to claim 1 or 2, characterized in that: Combined with the method of generalized predictive control, the above optimization process is integrated into a closed-loop control system. In this system, the output prediction model interacts with the universal model, and the control strategy is continuously optimized through an adaptive control algorithm to ultimately achieve the best control effect of the system.

4. The method for controlling inertia of a power system based on a model-free adaptive control strategy according to claim 1, characterized in that: The frequency control strategy in S2 is: Tracking grid frequency changes by adjusting the mechanical input power Pm: Pm=Pref+Kf(ω grid -ω) (5), Where Kf is the frequency control gain.

5. The method for controlling inertia of a power system based on a model-free adaptive control strategy according to claim 1, characterized in that: The torque control strategy in S2 is: To ensure system stability, the virtual synchronous generator adjusts the output torque according to the speed and rotor angle: T=Tref-Kt(ω grid -ω) (6)。 6. The method for controlling inertia of a power system based on a model-free adaptive control strategy according to claim 1, characterized in that: The voltage control strategy in S2 is: The voltage regulation of the virtual synchronous generator VSG is achieved by simulating the excitation system of the synchronous generator, using an automatic voltage regulator AVR, whose mathematical model can be expressed as: Where Vset is the voltage setting value, V is the output voltage, Kv is the voltage feedback gain, Ka is the excitation gain, and Vf is the excitation voltage.

7. The method for controlling inertia of a power system based on a model-free adaptive control strategy according to claim 1, characterized in that: The active and reactive power control strategies in S2 are: Active power control is achieved by adjusting the output frequency of the inverter. The frequency adjustment equation is: ΔP=Kp(ωref-ωr)+Ki∫(ωref-ωr)dt (8), Where: ΔP is the active power compensation, Kp is the proportional gain, Ki is the integral gain, ωref is the reference frequency; Reactive power control is achieved by adjusting the output voltage amplitude of the inverter. The adjustment equation is: ΔQ=Kq(Vref-V)+Ki∫(Vref-V)dt (9), Among them: ΔQ is the reactive power compensation amount, Kq is the proportional gain, Ki is the integral gain, and Vref is the reference voltage.

8. The method for controlling inertia of a power system based on a model-free adaptive control strategy according to claim 1, characterized in that: The PWM control strategy of the inverter in S2 is: The pulse width modulation (PWM) control of the inverter is the key to realizing the function of the virtual synchronous generator (VSG). The PWM control signal is determined by the active and reactive power regulation values ​​calculated by the control equation.

9. The method for controlling inertia of a power system based on a model-free adaptive control strategy according to claim 1, characterized in that: The virtual synchronous generator model in S1 is implemented by the following steps: Step 1. Initialize parameters: set the moment of inertia J, damping coefficient D, voltage setting value Vset, frequency setting value ωref, and control gains Kp, Ki, Kq; Step 2. Calculate the torque: Based on the active power P e and electromagnetic torque T e Calculate the electromagnetic torque. Step 3. Solve the rotation equation: Use the Runge-Kutta method to solve the rotation equation (1) and obtain the real-time rotor angular velocity; Step 4. Voltage regulation: Calculate the excitation voltage Vf based on the output voltage V and the reference voltage Vref; Step 5. Power regulation: Calculate the active and reactive power compensation according to the frequency deviation ωref-ωr and the voltage deviation Vref-V; Step 6. PWM control: convert the compensated active and reactive power into PWM control signals to adjust the output of the inverter; Furthermore, the power system is regarded as a black box and model-free adaptive control is performed.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, an inertia control method for a power system based on a model-free adaptive control strategy is implemented as described in any one of claims 1-9.

Images