New energy power grid virtual inertia estimation method and system and readable storage medium

Abstract

The invention discloses a new energy power grid virtual inertia estimation method and system and a readable storage medium, and belongs to the technical field of power grid control. The method comprises the following steps: acquiring input and output data of a power grid in real time through a PMU; constructing a transfer function model based on the input and output data by using a system identification method so as to determine orders of a pole and a zero point of the transfer function model; setting the order of the AR part as the pole order of the transfer function model, and setting the order of the X part as the zero order of the transfer function model, so as to construct an ARMAX model based on the input and output data; and calculating the virtual inertia of the power grid by using the ARMAX model. According to the invention, the PMU data is collected in real time, so that the system can accurately capture the frequency fluctuation of the power grid. Besides, a transfer function pole and a zero point identified by the system are mapped to the ARMAX model, so that a virtual inertia response strategy can be adjusted in real time, and the frequency stability of the power grid is ensured to be kept under frequency disturbance.

Description

Technical Field

[0001] The present invention belongs to the technical field of power grid control, and in particular relates to a new energy power grid virtual inertia estimation method, system and readable storage medium. Background Art

[0002] With the widespread application of new energy (such as wind power and photovoltaic) power generation systems, the proportion of traditional synchronous generators in power systems has gradually decreased. Synchronous generators can provide natural support for grid frequency fluctuations by releasing or absorbing rotational kinetic energy due to their mechanical rotational inertia. However, new energy power generation systems mostly rely on power electronic equipment and lack mechanical inertia when connected to the grid, resulting in weak buffering ability for frequency fluctuations.

[0003] In order to enhance the frequency stability of renewable energy systems, virtual inertia control technology has been proposed, which enables these power generation systems to simulate the inertia response of traditional synchronous generators. However, most existing virtual inertia control technologies assume that the system inertia is a fixed value and cannot be dynamically adjusted according to the actual operating conditions of the power grid. This method responds slowly to complex grid frequency disturbances and is difficult to adapt to the rapidly changing loads and frequency fluctuations in modern power grids.

[0004] In the modeling of power system frequency response, the ARMAX model, as a classic time series analysis tool, has been widely used in the dynamic modeling of the system. The ARMAX model can fit the autoregressive characteristics of the system, the impact of external input on output, and the interference of system noise through input and output data. The main steps of estimating the inertia of the power grid system using the ARMAX model include:

[0005] (1) Data collection: Obtain relevant data such as grid frequency, power flow, control signals, etc. These data will be used to build and verify the model.

[0006] (2) Model construction: Determine the structure of the ARMAX model. The general ARMAX model can be expressed as:

[0007] y(t)=A(q) -1 [B(q)u(t)+C(q)e(t)]

[0008] Among them, y(t) represents output, such as frequency deviation, u(t) represents input, such as control signal; e(t) represents random error, A(q), B(q), C(q) are polynomials, and q is a time-delay operator.

[0009] (3) Select the model order: Use information criteria such as AIC, BIC or cross-validation to determine the order of AR (autoregression), MA (moving average), and X (input) parts.

[0010] (4) Parameter estimation: Use least squares, maximum likelihood estimation or other optimization methods to fit the model based on the input and output data and estimate the model parameters.

[0011] (5) Use the model to exploit the relationship between parameters and grid dynamics and identify parameters that represent inertia characteristics.

[0012] However, the limitation of the traditional ARMAX model is that the selection of its order usually needs to be set manually, which may result in the model failing to fully capture the true dynamic characteristics of the frequency response in complex systems. Especially in renewable energy power generation systems, the frequency response is affected by a variety of nonlinear factors and complex disturbances, and a fixed model order is difficult to accurately describe its dynamic behavior, resulting in limited accuracy of virtual inertia evaluation.

[0013] In order to improve the frequency response capability of renewable energy power generation systems, nonlinear control technologies such as fuzzy control and neural networks have been introduced into the design of virtual inertia control in recent years. This type of technology dynamically adjusts the output power of renewable energy power generation units through self-learning algorithms, which can better deal with nonlinear problems in the system. However, although nonlinear control technology improves the responsiveness of the system, its computational complexity is high, especially when applied in large-scale power grids, it faces challenges in real-time performance and computational efficiency. In addition, this type of method focuses on the optimization of control strategies, but fails to accurately model the frequency response characteristics of the system, so it still has limitations under some complex working conditions.

[0014] The state-space model is a modeling method commonly used to describe the dynamic behavior of power systems. It can predict the frequency response by capturing the state changes of the system. However, in renewable energy power generation systems, the frequency response usually exhibits randomness and nonlinearity. The accuracy of the state-space model depends on the precise setting of system parameters, which are difficult to obtain accurately in a complex power grid environment. In addition, the state-space model has limited response to frequent disturbances and noise, and cannot fully adapt to the high dynamic changes of renewable energy power generation systems.

[0015] In general, existing technologies face many challenges in virtual inertia control and frequency response modeling. Traditional control and modeling methods are difficult to flexibly respond to the complex dynamic characteristics of renewable energy power generation systems, especially in power grids with a high proportion of wind power and photovoltaic access. Fixed inertia and frequency response models cannot meet the dynamic adjustment needs of the system. Summary of the invention

[0016] The purpose of the present invention is to provide a new energy grid virtual inertia estimation method, system and readable storage medium to accurately estimate the virtual inertia of each area of ​​the new energy grid.

[0017] According to a first aspect of the present invention, a method for estimating virtual inertia of a new energy power grid is provided, comprising:

[0018] Collect input and output data of the power grid in real time through PMU;

[0019] constructing a transfer function model based on the input and output data using a system identification method to determine the order of poles and zeros of the transfer function model;

[0020] Setting the order of the AR part to the pole order of the transfer function model, and setting the order of the X part to the zero order of the transfer function model, so as to construct an ARMAX model based on the input and output data;

[0021] The ARMAX model is used to calculate the virtual inertia of the power grid.

[0022] In other embodiments of the above method, the use of a system identification method to construct a transfer function model based on the input and output data includes: preprocessing the input and output data of the power grid collected in real time by the PMU, and the preprocessing includes:

[0023] Remove the mean of the data;

[0024] Filter the data after de-averaging;

[0025] The active power signal of the filtered data is subjected to feature extraction using the differential Teager-Kaiser energy operator method.

[0026] In other embodiments of the above method, the use of a system identification method to construct a transfer function model based on the input and output data includes: evaluating the current transfer function model according to the fitting results, and gradually adding poles and / or zeros according to the evaluation results to update the transfer function model.

[0027] In other embodiments of the above method, the calculating the virtual inertia of the power grid by using the ARMAX model includes:

[0028] Convert the discrete-time ARMAX model to a continuous-time model of high-order polynomials;

[0029] Simplifying the converted high-order polynomial continuous-time model into a low-order transfer function model;

[0030] The low-order transfer function model is used to analyze the initial moment of the power grid impulse response to estimate the virtual inertia of each area.

[0031] According to a second aspect of the present invention, a new energy grid virtual inertia estimation system is provided, comprising:

[0032] PMU unit, used to collect input and output data of the power grid in real time;

[0033] A fitting unit is configured to: construct a transfer function model based on the input and output data using a system identification method to determine the order of poles and zeros of the transfer function model;

[0034] A modeling unit is configured to: set the order of the AR part to the pole order of the transfer function model, and set the order of the X part to the zero order of the transfer function model, so as to construct an ARMAX model based on the input and output data;

[0035] A calculation unit is used to calculate the virtual inertia of the power grid using the ARMAX model.

[0036] In other embodiments of the above system, the fitting unit includes a preprocessing module for preprocessing the grid input and output data collected in real time by the PMU unit, and the preprocessing includes:

[0037] Remove the mean of the data;

[0038] Filter the data after de-averaging;

[0039] The active power signal of the filtered data is subjected to feature extraction using the differential Teager-Kaiser energy operator method.

[0040] In other embodiments of the above system, the fitting unit constructs a transfer function model based on the input and output data using a system identification method, including: evaluating the current transfer function model according to the fitting results, and gradually adding poles and / or zeros according to the evaluation results to update the transfer function model.

[0041] In other embodiments of the above system, the calculation unit calculates the virtual inertia of the power grid using the ARMAX model, including:

[0042] Convert the discrete-time ARMAX model to a continuous-time model of high-order polynomials;

[0043] Simplifying the converted high-order polynomial continuous-time model into a low-order transfer function model;

[0044] The low-order transfer function model is used to analyze the initial moment of the power grid impulse response to estimate the virtual inertia of each area.

[0045] According to a third aspect of the present invention, a readable storage medium is provided, on which a computer program is stored; when the computer program is executed, the new energy power grid virtual inertia estimation method is implemented.

[0046] The present invention collects PMU data in real time to ensure that the system accurately captures grid frequency fluctuations. The high resolution and synchronization of PMU data provide accurate input and output information for virtual inertia evaluation, ensuring that the model can respond to dynamic changes in the grid in a timely manner.

[0047] In addition, the present invention adopts a system identification method to automatically construct a transfer function model of the power grid and dynamically adjust the poles and zeros of the model, which solves the limitation of the traditional ARMAX model that relies on manually set orders, ensures that the model structure can more accurately reflect the frequency response characteristics of the power grid, and improves the adaptability and accuracy of the model.

[0048] In addition, the present invention improves the construction method of the ARMAX model, maps the poles and zeros of the transfer function model to the autoregressive and input parts, and combines online optimization algorithms such as the recursive least squares method to ensure that the model parameters can be adjusted in real time, and realizes the dynamic modeling and real-time optimization of the power grid frequency response characteristics. Finally, the virtual inertia can be evaluated in real time through pulse response analysis, and the virtual inertia response strategy can be dynamically adjusted according to the power grid frequency fluctuations to ensure that the new energy power generation unit can provide effective support under frequency disturbances. Compared with the traditional fixed inertia assumption, the dynamic adjustment strategy significantly improves the robustness and frequency stability of the power grid.

[0049] Further features and advantages of the present invention will become apparent from the following detailed description of exemplary embodiments of the present invention with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0051] Figure 1 A schematic diagram of a flow chart of a method for estimating virtual inertia of a new energy power grid according to an embodiment of the present invention;

[0052] Figure 2 Schematic diagram of the optimization process with gradually increasing poles and zeros;

[0053] Figure 3 Schematic diagram of the composition of a new energy grid virtual inertia estimation system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0054] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0055] Terminology explanation:

[0056] ARMAX model: AutoRegressive Moving Average with eXogenous inputs, which is a mathematical model used for time series analysis and control system modeling. It combines autoregressive (AR), moving average (MA) and external input (X).

[0057] PMU: Phasor Measurement Unit, phasor measurement unit.

[0058] The time series analysis method based on the ARMAX model has been applied to the frequency response modeling of the power system. However, in its application process, for example, when estimating the virtual inertia, the order of the ARMAX model usually needs to be set manually, which cannot accurately reflect the complex dynamic characteristics of the system, resulting in its accuracy in the frequency response characteristic modeling being limited. Therefore, how to ensure that the virtual inertia response of the new energy system can accurately reflect its dynamic characteristics through reasonable selection of the model order is a technical problem that needs to be solved urgently.

[0059] To solve the above problems, the present invention proposes a virtual moment of inertia observation method for key nodes of a new energy power grid, such as Figure 1 As shown, the following steps are included:

[0060] Step 101, collecting input and output data in the power grid in real time through a phasor measurement unit;

[0061] In order to overcome the shortcomings of the ARMAX model in terms of order selection, the present invention introduces a system identification method based on PMU (phasor measurement unit) data, aiming to more accurately capture the complex dynamic response characteristics of the new energy system by reasonably selecting the order of the transfer function and dynamically adjusting the model structure.

[0062] PMU has the advantages of high time resolution and synchronous acquisition. It can collect key input and output data in the power grid in real time and with high resolution, including frequency deviation, power output and voltage phase angle, etc., so that the system can accurately capture the dynamic response of the power grid when frequency disturbance occurs. These data provide an accurate data basis for virtual inertia evaluation, ensuring that the response information during power grid frequency disturbance is comprehensive and accurate.

[0063] Step 102, constructing a transfer function model of the power grid based on input and output data using a system identification method to determine the order of the poles and zeros of the transfer function model;

[0064] To ensure the quality and validity of the data, the data collected by the PMU needs to be preprocessed before system identification. In one example, the preprocessing includes:

[0065] (1) Remove the mean of the data;

[0066] The influence of DC component is eliminated by de-averaging.

[0067] (2) filtering the data after de-averaging;

[0068] The data after de-averaging is processed by filtering technology to remove high-frequency noise and ensure the smoothness of the signal.

[0069] (3) extracting features from the active power signal of the filtered data;

[0070] According to an embodiment of the present invention, a differential Teager-Kaiser energy operator (dTKEO) method is used to extract features from the active power signal of the filtered data to capture the disturbance characteristics in the signal, thereby accurately detecting the time and duration of the disturbance. In one example, the extraction process includes the following steps:

[0071] (i) Apply the differential Teager-Kaiser energy operator to the active power signal P(t) of the filtered data and calculate the dTKEO value of each energy point:

[0072] ψ(P[n])=P[n] 2 -P[n-1]·P[n+1]

[0073] (ii) Based on the calculated dTKEO values, find abnormally high energy points, which may correspond to rapid changes or anomalies in the power signal. Perform statistical analysis on the calculated dTKEO values, such as calculating the mean value, standard deviation, etc.

[0074] The identified high energy points or transients can be used to determine whether they are related to power grid events (such as faults, sudden load changes, etc.).

[0075] System identification is a method to construct a mathematical model of a dynamic system by measuring the input and output signals of the system. Based on the key input and output data collected by the phasor measurement unit, the system identification technology is used to dynamically analyze the pre-processed input and output data to automatically determine the transfer function model of the system.

[0076] Similar to polynomial curve fitting, the accuracy of system identification depends on the choice of transfer function order. Low-order models have low computational complexity, but may not fully capture the complex dynamic characteristics of new energy systems, resulting in insufficient identification accuracy. Although high-order models improve accuracy, they also significantly increase computational complexity.

[0077] In order to achieve a balance between identification accuracy and computational complexity, during the identification process, according to an embodiment of the present invention, an optimization strategy of gradually increasing the number of poles and zeros is adopted, such as Figure 2 As shown, the optimization process includes:

[0078] (1) Constructing the initial model;

[0079] The initial transfer function model uses low-order poles and zeros, for example, a first-order model with only a single pole and no zeros, that is, the number of poles p is 1 and the number of zeros z is 0.

[0080] (2) Use the current model to fit the preprocessed input and output data;

[0081] For example, the least squares method, maximum likelihood estimation or other suitable parameter estimation methods may be used.

[0082] (3) Evaluate the current model based on the fitting results, and add poles or zeros based on the evaluation results to update the model;

[0083] After fitting, the prediction error, residual analysis, information criteria (such as AIC or BIC) or coefficient of determination (R 2 ) and other methods to evaluate the performance of the current model. If the current model does not adequately capture the dynamics of the data, consider adding poles. Adding poles usually increases the logarithm of the model, and adding zeros can help the model better capture the complex interactions between input and output.

[0084] In one example, after fitting, it is determined whether the recognition precision (recognition accuracy) of the fitting result exceeds a set value, such as 90%.

[0085] If yes, that is, it exceeds the set value, then multiple times (as an example, Figure 2 The number of poles and zeros is increased (3 times in the example), and the system is identified after each increase of the number of poles and zeros. The difference in identification accuracy of the fitting results of multiple system identifications is calculated. If the difference is less than a set value, such as 1%, it means that there is almost no improvement in the identification accuracy of the fitting results after increasing the number of poles and zeros. The number of poles and zeros before the increase is output. Otherwise, if the difference is greater than (or equal to) the set value, the order is continued to be increased, and the process returns to step (2) until the identification accuracy requirement is met.

[0086] If not, that is, it does not exceed the above set value, then continue to increase the order and return to step (2) until the identification accuracy requirement is met.

[0087] In the above process, when increasing the order, first determine the size relationship between the number of zeros z and the number of poles p. If the number of zeros z is less than the number of poles p, continue to increase the number of zeros; if the number of zeros z is equal to the number of poles p, set the number of zeros z to 0 and continue to increase the number of poles p.

[0088] Based on the preliminary setting of the transfer function model structure, the present invention gradually optimizes the model order by evaluating and adjusting the number of poles and zeros to ensure that the fitting effect of system identification reaches the expected result. This process avoids the uncertainty of manually setting the order in the traditional ARMAX model through feedback iterative optimization, so that the final model can fully characterize the dynamic behavior of the system.

[0089] As an example, the transfer function model obtained through system identification can be expressed as follows:

[0090]

[0091] Among them, n a is the order of the denominator, n b is the order of the molecule.

[0092] n determined by identification method a and n b Will be used to set the order of the ARMAX model.

[0093] The present invention is based on the key input and output data in the power grid collected by PMU, and adopts system identification technology to automatically construct a transfer function model of the power grid. By analyzing the input and output data, the poles and zeros of the transfer function are dynamically adjusted to ensure that the constructed model can accurately reflect the complex frequency response characteristics of the power grid. It no longer relies on the traditional manually set model order, and improves the model's ability to adapt to different working conditions.

[0094] In addition, the transfer function model obtained by the optimization strategy of gradually increasing the number of poles and zeros provides a reliable model basis for the identification of the ARMAX model and achieves a balance between identification accuracy and computational complexity.

[0095] Step 103, constructing an ARMAX model based on the order of the poles and zeros of the transfer function model;

[0096] The ARMAX model is a mathematical model used for time series analysis and control system modeling, which combines autoregression (AR), moving average (MA) and external input (X). Specifically, the output variable depends not only on its own past value (autoregression part), but also on the lagged value of the external input variable (external input part) and the past error term (moving average part). This structure enables the ARMAX model to handle complex systems containing noise interference and external input well. Its structure can be described as:

[0097]

[0098] Where y(t) is the current output value; is the autoregressive part, where a i is the autoregressive coefficient, y(ti) is the lagged output value, and n is the order of the autoregressive term; is the external input part, where b j is the external input coefficient, u(tj) is the lagged external input value, and m is the order of the exogenous input term; is the sliding average part, where c k is the sliding average coefficient, e(tk) is the lagged error term, and p is the order of the sliding average term; e(t) is the error term, which represents random disturbance or noise, and represents the unexplained part of the model.

[0099] After the order of the transfer function model is determined in the aforementioned system identification process, the improvement of the ARMAX model can be effectively implemented. The order of the autoregressive part (AR), the input polynomial part (X), and the moving average (MA) part in the ARMAX model can be determined by the number of poles and zeros of the transfer function model. Specifically, the order of the autoregressive part (AR) in the ARMAX model is equal to the order of the denominator of the transfer function n a (the order of the poles), which represents the dynamic characteristics of the system; the order of the external input (X) is equal to the order n of the numerator of the transfer function b (the order of the zero point) plus 1, that is, n b +1, indicating the influence of the signal on the output; the order of the moving average part (MA) can be set according to the noise characteristics and can be determined by further model adjustment. Here, n c Therefore, the ARMAX model can be expressed as follows:

[0100]

[0101] For the virtual inertia assessment of new energy systems, the accuracy and robustness of the model are crucial, because the dynamic characteristics of new energy power generation units often fluctuate with changes in external conditions (such as wind speed, light, etc.) and grid status. The above operations provide a reliable theoretical basis for the structure of the ARMAX model, making the model more accurate and robust in the virtual inertia assessment of new energy systems. Among them, in the aforementioned system identification process, the order selection of the traditional ARMAX model is optimized by dynamically adjusting the model order, so that the model can more accurately describe the frequency response characteristics of the new energy system.

[0102] In addition, the present invention constructs an improved ARMAX model based on the identification results of the transfer function model, which integrates autoregression, input and noise processing into a dynamically optimized model framework through the mapping of the transfer function poles and zeros identified by the system. Through the impulse response analysis described below, the virtual inertia response strategy can be adjusted in real time to ensure that the new energy power generation unit can provide effective inertia support under frequency disturbances and maintain the frequency stability of the power grid.

[0103] The improved ARMAX model adopted in the present invention can not only better characterize the dynamic behavior of the system through reasonable order selection, but also perform online parameter estimation through recursive least squares method, and dynamically adjust model parameters during operation to adapt to the time-varying characteristics of the system.

[0104] Step 104, calculating virtual inertia using the ARMAX model;

[0105] By constructing the ARMAX model, an approximate model describing the dynamic relationship between load changes and frequency changes can be established. Although the effective inertia component of the system is included in the ARMAX model, the inertia is not expressed in a clear numerical form, so further analysis of the determined model is required to extract and determine the virtual inertia response of the system. Therefore, the ARMAX model constructed using the above steps can be used to evaluate the virtual inertia through pulse response analysis by converting the ARMAX model into a transfer function form.

[0106] First, the discrete-time ARMAX model is converted into a continuous-time model. This step is crucial because in many system controls, the continuous-time model can more intuitively reflect the physical characteristics of the system. However, the discrete-time model cannot always be successfully converted into a stable continuous-time model, so the stability of the converted system model needs to be checked. Usually, the stability of the model is judged by checking whether the poles of the system are located in the left half of the complex plane (that is, the real part of the poles is less than zero).

[0107] If the model conversion is successful, the next step is to simplify the high-order polynomial model into a low-order transfer function model. To this end, the model reduction technique in the system identification method can be used. First, the system is converted into a state space model, and then methods such as balanced truncation or singular value decomposition are used to identify and remove states that have little impact on the system dynamics, thereby simplifying the system model. Finally, an appropriate algorithm is used to convert the simplified state space model into a transfer function form to obtain a low-order transfer function with fewer poles and zeros.

[0108] The above process can also be implemented through the built-in functions of commercial simulation software.

[0109] The expected structure of the transfer function is of the general form:

[0110]

[0111] Ideally, we expect the transfer function to have a simple structure that is easy to analyze and calculate. However, in the process of model simplification, some marginal non-zero terms may appear, which have little effect on the overall dynamics of the system but increase the complexity of the transfer function. In order to facilitate the estimation of the effective inertia value, these marginal terms can be properly processed or ignored, so as to obtain a more concise and expected transfer function structure.

[0112] In order to better understand the process of determining the virtual inertia response, a simplified example can be used. Assuming a single-machine system without a speed regulator or frequency control, the swing equation of the system can be modeled as a first-order transfer function:

[0113]

[0114] Where M is the effective inertia and D is the frequency dependence of the load. The unit impulse response of the system is

[0115]

[0116] At the moment after the disturbance, i.e., at t = 0, the initial value of the impulse response is Therefore, the initial response of the system to a load disturbance can be considered the inverse of the effective inertia. That is, the greater the inertia of the system, the smaller the initial response. Since the input signal is a load change, the proportional term is negative, so

[0117] Through the above methods, the complex system transfer function obtained by system identification contains more dynamic components and control mechanisms, but the analysis of inertial response still follows similar principles. Regardless of whether the system has a control link, the inertial response is usually the fastest part of the system, thus determining the initial response of the system to disturbances. By analyzing the initial moment of the impulse response of the complex system, the virtual inertia of each area can be estimated.

[0118] In more complex multi-machine systems, the transfer function obtained through system identification contains multiple poles and zeros, which characterize the dynamic characteristics of each unit. Although the overall dynamics of the system are relatively complex, the initial inertial response is still dominated by the total inertia of the system. Therefore, the virtual inertia of the system can be determined by the value of the impulse response at the initial moment. On this basis, the improved ARMAX model shows significant advantages in the virtual inertia evaluation of new energy systems. By optimizing the structure of the model and the parameter estimation method, the present invention can dynamically adjust the calculation of the virtual inertia according to the actual operation of the new energy power generation unit, thereby providing more accurate basic data for grid frequency control.

[0119] Through the above improvements, the ARMAX model not only improves the system identification accuracy, but also shows higher adaptability and robustness in the renewable energy power generation system. The improved ARMAX model can optimize the structure and parameter estimation of the model through the system identification method, so that it can show high prediction accuracy under various operating conditions, significantly improving the effectiveness of the virtual inertia evaluation of the renewable energy power generation system.

[0120] According to another embodiment of the present invention, a new energy grid virtual inertia estimation system is also provided, such as Figure 3 As shown, the system comprises:

[0121] PMU unit 301, used for collecting input and output data of the power grid in real time;

[0122] The fitting unit 302 is configured to: construct a transfer function model based on the input and output data using a system identification method to determine the order of the poles and zeros of the transfer function model;

[0123] The modeling unit 303 is configured to: set the order of the AR part to the pole order of the transfer function model, and set the order of the X part to the zero order of the transfer function model, so as to construct an ARMAX model based on the input and output data;

[0124] The calculation unit 304 is used to calculate the virtual inertia of the power grid using the ARMAX model.

[0125] In the above system, the fitting unit includes a preprocessing module for preprocessing the grid input and output data collected in real time by the PMU unit, and the preprocessing includes:

[0126] Remove the mean of the data;

[0127] Filter the data after de-averaging;

[0128] The active power signal of the filtered data is subjected to feature extraction using the differential Teager-Kaiser energy operator method.

[0129] In the above system, the fitting unit constructs a transfer function model based on the input and output data using a system identification method, including: evaluating the current transfer function model according to the fitting results, and gradually adding poles and / or zeros according to the evaluation results to update the transfer function model.

[0130] In the above system, the calculation unit calculates the virtual inertia of the power grid using the ARMAX model, including:

[0131] Convert the discrete-time ARMAX model to a continuous-time model of high-order polynomials;

[0132] Simplifying the converted high-order polynomial continuous-time model into a low-order transfer function model;

[0133] The low-order transfer function model is used to analyze the initial moment of the power grid impulse response to estimate the virtual inertia of each area.

[0134] According to a third aspect of the present invention, a readable storage medium is provided, on which a computer program is stored; when the computer program is executed, the new energy power grid virtual inertia estimation method is implemented.

[0135] The present invention is applicable to multi-unit renewable energy power grids, and can independently evaluate the virtual inertia of each unit and coordinately control its frequency response. This multi-unit collaborative control method not only improves the overall frequency stability of the power grid, but also enhances the flexibility and reliability of the system, and is particularly suitable for complex power grid scenarios with a high proportion of renewable energy access.

[0136] The above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or some technical features can be replaced by equivalents without departing from the spirit of the technical solution of the present invention, which should be included in the scope of the technical solution for protection of the present invention.

Claims

1. A new energy grid virtual inertia estimation method, characterized in that: include: Collect input and output data of the power grid in real time through PMU; constructing a transfer function model based on the input and output data using a system identification method to determine the order of poles and zeros of the transfer function model; Setting the order of the AR part to the pole order of the transfer function model, and setting the order of the X part to the zero order of the transfer function model, so as to construct an ARMAX model based on the input and output data; The ARMAX model is used to calculate the virtual inertia of the power grid.

2. The virtual inertia estimation method according to claim 1, characterized in that: The method of using the system identification method to construct a transfer function model based on the input and output data includes: preprocessing the input and output data of the power grid collected in real time by the PMU, and the preprocessing includes: Remove the mean of the data; Filter the data after de-averaging; The active power signal of the filtered data is subjected to feature extraction using the differential Teager-Kaiser energy operator method.

3. The virtual inertia estimation method according to claim 1, characterized in that: The method of using the system identification method to construct a transfer function model based on the input and output data includes: evaluating the current transfer function model according to the fitting result, and gradually adding poles and / or zeros according to the evaluation result to update the transfer function model.

4. The virtual inertia estimation method according to claim 1, characterized in that: The method of calculating the virtual inertia of the power grid by using the ARMAX model includes: Convert the discrete-time ARMAX model to a continuous-time model of high-order polynomials; Simplifying the converted high-order polynomial continuous-time model into a low-order transfer function model; The low-order transfer function model is used to analyze the initial moment of the power grid impulse response to estimate the virtual inertia of each area.

5. A new energy grid virtual inertia estimation system, characterized in that: include: PMU unit, used to collect input and output data of the power grid in real time; A fitting unit is configured to: construct a transfer function model based on the input and output data using a system identification method to determine the order of poles and zeros of the transfer function model; A modeling unit is configured to: set the order of the AR part to the pole order of the transfer function model, and set the order of the X part to the zero order of the transfer function model, so as to construct an ARMAX model based on the input and output data; A calculation unit is used to calculate the virtual inertia of the power grid using the ARMAX model.

6. The virtual inertia estimation system according to claim 5, characterized in that: The fitting unit includes a preprocessing module, which is used to preprocess the power grid input and output data collected in real time by the PMU unit, and the preprocessing includes: Remove the mean of the data; Filter the data after de-averaging; The active power signal of the filtered data is subjected to feature extraction using the differential Teager-Kaiser energy operator method.

7. The virtual inertia estimation system according to claim 5, characterized in that: The fitting unit constructs a transfer function model based on the input and output data using a system identification method, including: evaluating the current transfer function model according to the fitting result, and gradually adding poles and / or zeros according to the evaluation result to update the transfer function model.

8. The virtual inertia estimation system according to claim 5, characterized in that: The calculation unit calculates the virtual inertia of the power grid using the ARMAX model, including: Convert the discrete-time ARMAX model to a continuous-time model of high-order polynomials; Simplifying the converted high-order polynomial continuous-time model into a low-order transfer function model; The low-order transfer function model is used to analyze the initial moment of the power grid impulse response to estimate the virtual inertia of each area.

9. A readable storage medium, characterized in that: The readable storage medium stores a computer program; when the computer program is executed, the new energy power grid virtual inertia estimation method as described in any one of claims 1 to 4 is implemented.

Images