Inertia continuous assessment method for power system considering uncertain system identification

Abstract

The power system inertia continuous evaluation method considering uncertainty system identification, based on the real-time acquisition of frequency and power data in the system by the phasor acquisition unit PMU, the frequency and power data are pretreated, and the evaluation data set is established; according to the frequency dynamic characteristic equation, the traditional output error model is adopted, and the system inertia identification model is established; for the inertia parameter problem of the established system inertia identification model, considering the uncertainty of input and output data, the robust model optimization is adopted to solve, and the system inertia identification under the condition of input and output data uncertainty is realized. The power system inertia continuous evaluation method considering uncertainty system identification, so as to better handle the uncertainty of identification data, the real-time updating model result is closer to the actual value, and the accuracy of identification is guaranteed.

Description

Technical Field

[0001] This invention relates to the field of power system inertia assessment technology, and specifically to a method for continuous assessment of power system inertia that considers uncertainty system identification. Background Technology

[0002] Traditional power systems are primarily powered by primary energy sources, mainly fossil fuels. However, with rapid societal development, the depletion of traditional fossil fuels and increasingly prominent environmental problems have led to the rapid development of new energy sources, causing power systems to increasingly rely on them. In these new power systems, different types of new energy sources exhibit varying inertia response characteristics, resulting in complex nonlinear time-varying inertia that further deteriorates the frequency stability of the power system. The continuous decrease in system inertia weakens the system's inertia support and frequency regulation capabilities under active power surges, posing a severe challenge to the stable operation of the power system. Therefore, conducting research on continuous inertia assessment of new power systems to achieve continuous monitoring and sensing of system inertia is of great significance for early warning of frequency instability risks and ensuring the safe operation of the power grid.

[0003] Inertia assessment has become a hot research topic, and numerous domestic and international research institutions have made relevant progress. However, power grid planners and dispatchers still lack an accurate understanding of the system's inertia characteristics and their changing trends. While research has made valuable attempts at continuous real-time inertia assessment, the insufficient support from the inertia of new energy sources leads to relatively conservative assessment results.

[0004] Furthermore, the strong uncertainty in the output of new energy sources exacerbates the power imbalance, and the system inertia level exhibits strong fluctuations and uncertainties within a day, posing a challenge to the accuracy of continuous real-time assessment of system inertia. Summary of the Invention

[0005] To address the uncertainties in the identification data and ensure identification accuracy, as well as the economic efficiency and safety of system operation, this invention proposes a continuous evaluation method for power system inertia that considers uncertainties in system identification. This method better handles the uncertainties in the identification data, updates the model results in real time to more closely approximate the actual values, and ensures identification accuracy.

[0006] The technical solution adopted in this invention is as follows:

[0007] A continuous evaluation method for power system inertia considering uncertainty system identification includes the following steps:

[0008] Step 1: Based on the phasor acquisition unit (PMU), the frequency and power data of the system are acquired in real time, the frequency and power data are preprocessed, and an evaluation dataset is established.

[0009] Step 2: Based on the frequency dynamic characteristic equation, a system inertia identification model is established using the traditional output error model;

[0010] Step 3: For the inertia parameter problem of the system inertia identification model established in Step 2, considering the uncertainty of input and output data, a robust model is used for optimization to achieve system inertia identification under the condition of uncertain input and output data.

[0011] In step 1, the required dynamic characteristics of the power grid system are obtained from the PMU measurement data center, including: the unbalanced power output ΔP(t) of the generator sets and the node frequency change rate Δf. Through data preprocessing, the inertia characteristic parameters H of each generator set are obtained. i and the center frequency f of the power grid inertia COI .

[0012] In step 2, based on the traditional output error model OE, it is analyzed that the traditional output error model OE has many parameters to be estimated. The traditional output error model is as follows:

[0013]

[0014] In the formula, u(t) is the input of the OE model, i.e., the power change rate; y(t) is the output of the OE model, i.e., the inertia center frequency deviation; x(t) is the unobservable true output within the system; and v(t) is the system noise. n a ,n b Let A(q) and B(q) be the orders of A(q) and B(q).

[0015] To reduce the number of undetermined parameters while ensuring the good fit of the identification model, the inertial response characteristics of the actual system are first analyzed. The transfer function of the system inertial identification model is:

[0016]

[0017] In the formula, △f COI H represents the frequency deviation of the inertia center, ΔP represents the change in system power difference, and H represents the frequency deviation of the inertia center. sys It is the system inertia, D sys For system damping, s is the Laplace operator. ΔP(s) represents the representation of ΔP in the complex field (Laplace).

[0018] Since equation (3) gives the unit impulse response of the system's inertia response as:

[0019]

[0020] In equation (4), h(t) is obtained through the transfer function in equation (3). When the input is a unit pulse, the corresponding output (time domain t) is obtained by mathematically transforming the transfer function through the inverse Laplace transform. This indicates that h(t) decays exponentially; t represents time; when t = 0, the above equation is the initial response of the inertia under disturbance, which is numerically the reciprocal of the system inertia. According to equations (3) and (4), the high-order identification model of the system inertia response can be reduced to the form of a first-order transfer function.

[0021] The traditional OE model can be described in least squares format using the least squares method:

[0022]

[0023] Define data vector and parameter vector θ = [a, b], where:

[0024] Φ(t)=[-x(t-1),…,-x(tn a )], ψ(t)=[u(t),u(t-1),…,u(tn b )];

[0025] In step 3, based on the established system inertia identification model, and considering that the input and output data of the system inertia identification model have uncertainties and may not necessarily follow a probability distribution, the uncertainty of the actual system inertia identification data is described as follows:

[0026]

[0027] H(δ) is the error of the system inertia, H t For the system input, δ Hi H is the input disturbance noise. i Let δ be the system inertia parameter of the i-th unit, m be the total number of units, and i be the unit number; δ is the error.

[0028]

[0029] In the formula, e i Let y(δ) be the i-th column of the m-dimensional identity matrix, and y(δ) be the output error of the system identification model. t For the system input, δ yi This refers to the output disturbance noise;

[0030] By using the Robust Least Squares (RLS) method, which takes into account the uncertainty of the data structure, the identification problem in equation (5) is transformed into an optimization problem, so that the identification results of the unknown parameters can achieve the minimization of residuals.

[0031]

[0032] H(δ) is the error of the system inertia, R m Let m be a vector space, ρ be used to define the range of the residuals, and θ be the parameter vector θ = [a, b].

[0033] Where, δ=[δ y ,δ H ],||δ|| ∞ =max|δ i |;Φ(t)=[-x(t-1),…,-x(tn a )],

[0034] ψ(t)=[u(t),u(t-1),…,u(tn b )];

[0035] In the formula: u(t) is the input of the OE model, i.e., the power change rate; y(t) is the output of the OE model, i.e., the inertia center frequency deviation; x(t) is the unobservable true output within the system; v(t) is the system noise.

[0036] δ=[δ y ,δ H ], defining a simplified calculation, δ y =[δ y1 ,…,δ ym ] T and δ H =[δ H1 ,…,δ Hm ] T Both belong to R m δ i Represents the residual of the i-th unit, ||δ|| ∞ This represents the maximum value of the residual modulus.

[0037] If the worst residual exists, then the minimum θ is the RLS solution of equation (11). Therefore, the existence of the worst residual should be checked before solving.

[0038] Define M(θ) = [H1θ-y1,…,H m θ-y m ], F = M T (θ)M(θ), g=M T (θ)(H t θ-y t ), To simplify the expression.

[0039] H1 represents the inertia of the first unit, H1~H mLet y represent the inertia of each of the m generator units. t It is a matrix representing the output of the OE model. y1~y m Matrix y t The elements of M T (θ) is the transpose of M(θ), H t The first column is h t The lower triangular Toeplitz matrix is ​​represented as:

[0040]

[0041] Since the optimization problems of residuals and residual squares are equivalent, let ρ = 1, and equation (11) is equivalently expressed as:

[0042]

[0043] δ T This represents the transpose of the residual, where T represents the transpose.

[0044] At this point, the squared residual is a quadratic function of the uncertain data δ. The real-time sampled data belongs to a continuous and bounded set, and it can be seen that the maximum and minimum values ​​of this quadratic function exist, thus enabling the optimization solution of the residual.

[0045] Let the auxiliary variable λ≥0, and optimize the quadratic form function of the residual in equation (12) using the S-lemma, we get:

[0046]

[0047] g T Indicates the transpose of g;

[0048] Equation (13) can be rewritten as a function of two variables, λ and τ, if and only if τ ≥ 0:

[0049]

[0050] Let τ be the time constant and I be the identity matrix. Clearly, τ ≥ 0 in the constraint τI ≥ F. The worst residual can be obtained by solving a semidefinite program (SDP) problem involving the functions λ and τ.

[0051] To simplify the problem and obtain the corresponding worst-case perturbation, the constraints of equation (14) are transformed using Schur complement, and the robust identification problem can be transformed into a convex optimization problem of minimizing λ:

[0052] minλ

[0053]

[0054] Accordingly, while preserving the polynomial time complexity, the RLS method transforms the original problem into a deterministic semi-definite programming problem through robust equivalence approximation, and solves equation (15) using a general SDP solver, thus enabling the identification of system inertia under uncertain conditions.

[0055] This invention provides a method for continuous evaluation of power system inertia considering uncertainty system identification, with the following technical advantages:

[0056] 1) The method of the present invention takes into account the impact of uncertainty, improves the identification accuracy by using uncertain data information for online continuous identification, has the ability to monitor the system inertia online, and can be used as an indicator for early warning of frequency instability risk.

[0057] 2) The method of the present invention is highly efficient, has high recognition accuracy, accurate prediction, and complete data structure, and can update the recognition model parameters in real time.

[0058] 3) New energy generating units have inertial response when connected to the grid through virtual inertia control strategies. The efficient virtual inertia in the method of this invention can provide a certain amount of inertia support for the power system.

[0059] 4) The robust least squares (RLS) method of this invention, which considers the uncertainty of data structure, has a strong suppression effect on external noise and exhibits strong robustness in handling uncertainty, making it applicable to engineering. Attached Figure Description

[0060] Figure 1 Flowchart for continuous evaluation of power system inertia.

[0061] Figure 2 It is an improved four-machine two-area system model diagram.

[0062] Figure 3(a) shows the wind speed measurement parameters under 100% operating conditions of the wind farm;

[0063] Figure 3(b) shows the wind power measurement parameters of the wind farm under 100% operating conditions;

[0064] Figure 3(c) shows the frequency measurement parameters of the wind farm under 100% operating conditions.

[0065] Figure 4(a) shows a comparison of the continuous evaluation results of system inertia. Figure 1 ;

[0066] Figure 4(b) shows a comparison of the continuous evaluation results of system inertia. Figure 2 ;

[0067] Figure 4(c) is a comparison of the continuous evaluation results of the system inertia in Figure 3;

[0068] Figure 4(d) is a comparison of the continuous evaluation results of system inertia. Detailed Implementation

[0069] A continuous inertia assessment method for power systems considering uncertainty aims to fully account for the uncertainties in input and output data. It establishes a system identification model based on the input data, transforms the system identification problem into a semidefinite programming problem (SDP) using the Relative Slow Linear Algorithm (RLS) that considers data structure uncertainties, and solves the problem using a general SDP solver. This achieves system inertia identification under uncertain input and output data conditions, improving the accuracy of inertia assessment. Figure 1 The steps shown are as follows:

[0070] Step 1: Collect PMU measured data, preprocess the data, and establish an evaluation dataset.

[0071] In step 1, the required power grid dynamic characteristics data are obtained from the PMU measurement data center, including the unbalanced power output ΔP(t) of the generator units and the node frequency change rate Δf. Through data preprocessing, the inertia characteristic parameters H of each unit are obtained. i and the center frequency f of the power grid inertia COI .

[0072] Step 2: Based on the frequency dynamic response equation, a system inertia identification model is established using the traditional output error model.

[0073] The supporting role of the power system's inertial response is manifested in the dynamic changes of active power and frequency. The frequency dynamic characteristic equation of the overall power system, characterized by the center of inertia, is as follows:

[0074]

[0075] In the formula, H sys and D sys These represent the system inertia and damping, respectively; ΔP is the change in the system power difference; Δf COI This refers to the frequency deviation of the inertia center. The inertia responds constantly to maintain frequency stability under disturbances, and the system's active-frequency dynamic process contains the characteristics of the system's inertia.

[0076] Inertia assessment falls under the category of parameter identification problems. Constructing a system identification model can identify inertia characteristic parameters. In constructing the inertia identification model described in this invention, a traditional output error model is used as a foundation. The traditional output error model is as follows:

[0077]

[0078] In the formula, u(t) is the input of the OE model, i.e., the power change rate; y(t) is the output of the OE model, i.e., the inertia center frequency deviation; x(t) is the unobservable true output within the system; and v(t) is the system noise. n a,n b Let A(q) and B(q) be the orders of A(q) and B(q).

[0079] Based on the traditional Output Error (OE) model, it is found that there are many parameters to be estimated in the OE model. In order to reduce the number of undetermined parameters while ensuring the good fit of the identification model, the inertial response characteristics of the actual system are analyzed first. According to Equation (1), the transfer function from the dynamic change of the system's active power to the frequency to the system's inertial identification model is:

[0080]

[0081] In the formula, s is the Laplace operator.

[0082] The unit impulse response of the system inertia identification model is obtained as follows:

[0083]

[0084] At t=0, the initial response of the inertia under disturbance is numerically the reciprocal of the system inertia. According to equations (3)-(4), the higher-order identification model of the system inertia response can be reduced to a first-order transfer function form.

[0085] Step 3: For the inertia parameter problem of the system inertia identification model established in Step 2, considering the uncertainty of input and output data, a robust model is used for optimization and solution.

[0086] After determining the system inertia identification model, the model structure parameters can be determined according to the identification algorithm. The Least Squares Method (LS) is widely used in the field of system identification, and the parameter identification problem of equation (2) can be described in the least squares format:

[0087]

[0088] Define data vector And the parameter vector θ = [a, b], where Φ(t) = [-x(t-1), ..., -x(tn)]. a )], ψ(t)=[u(t),u(t-1),…,u(tn b )];

[0089] Using the operating status data of the PMU real-time measurement system, with a sampling length of m, the sampled input and output data for t = 1, 2…m are as follows:

[0090]

[0091] Among them, H t The first column is h tThe lower triangular Toeplitz matrix is ​​represented as:

[0092]

[0093] Based on this, we obtain m equations, and equation (5) can be written in matrix form as follows:

[0094] y t =H t θ+v t (8);

[0095] When input H t and output y t While a perfectly linear relationship between inertia values ​​can be achieved using the standard least squares (LS) method for unbiased estimation, this is difficult to implement in engineering applications of system inertia identification. On one hand, the real-time data measurement from the PMU is subject to errors due to the performance of the remote control unit and the data transmission process; on the other hand, the inertia center frequency deviation Δf obtained in step 1... COI The inertia and power of the synchronous machine are determined by the day-ahead start-up and shutdown schedule of the generating units, but the active power output of new energy sources is uncertain. Due to the uncertainty of input and output data, and their non-consistency with probability distributions, the uncertainty of the actual system inertia identification data is described as follows:

[0096]

[0097]

[0098] In the formula, δ Hi δ yi The input and output represent their respective disturbance noises, y(t) is the output of the OE model, i.e., the inertia center frequency deviation, H i and e i Let e ​​be the direction of the disturbance. i H is the i-th column of an m-dimensional identity matrix. i The first column equals e i The lower triangular Toeplitz matrix, where δ is the error.

[0099] To minimize the residuals of the identification results for unknown parameters, the Robust Least Squares (RLS) method, which considers the uncertainty of the data structure, transforms the identification problem in equation (5) into an optimization problem:

[0100]

[0101] Where, δ=[δ y ,δ H ],||δ|| ∞ =max|δ iIf the worst-case residual exists, then the minimum θ is the RLS solution of equation (11). Therefore, the existence of the worst-case residual should be checked before solving.

[0102] Define M(θ) = [H1θ-y1,…,H m θ-y m ], F = M T (θ)M(θ), g=M T (θ)(H t θ-y t ), To simplify the expression, since the optimization problems of residuals and residual squares are equivalent, it is assumed that ρ = 1. Equation (11) is equivalently expressed as:

[0103]

[0104] At this point, the squared residual is a quadratic function of the uncertain data δ. Since the real-time sampled data belongs to a continuous and bounded set, it is easy to see that the maximum and minimum values ​​of this quadratic function exist, and thus the residual can be optimized and solved.

[0105] Let the auxiliary variable λ≥0, and optimize the quadratic form function of the residual in equation (12) using the S-lemma, we get:

[0106]

[0107] Equation (13) can be rewritten as a function of two variables, λ and τ, if and only if τ ≥ 0:

[0108]

[0109] I is the identity matrix, and obviously τ≥0 in the constraint τI≥F. The worst residual can be obtained by solving a semidefinite program (SDP) problem of functions λ and τ. To simplify the problem and obtain the corresponding worst perturbation, the constraint of equation (14) is transformed using Schur complement, and the robust identification problem is transformed into a convex optimization problem of minimizing λ:

[0110] minλ

[0111]

[0112] Therefore, the RLS method, while retaining the polynomial time complexity, transforms the original problem into a deterministic semidefinite programming problem through robust equivalence approximation. Solving equation (15) using the general-purpose SDP solver enables the identification of system inertia under uncertain conditions. The SDP solver is a toolkit in MATLAB used to solve semidefinite programming problems to improve the solution speed; it can be directly called within MATLAB during the solution process.

[0113] Example:

[0114] To verify the effectiveness of the method described in this invention, it was applied to an improved typical four-machine two-zone simulation system, using Matlab / Simulink as the simulation software. Figure 2 As shown, a 500MW wind farm is connected to the system via bus 6. To verify the power system inertia continuous evaluation method based on robust optimization system identification, a double-fed induction generator (DFIG) is connected to the grid using a virtual inertia control strategy. The virtual inertia time constant of the DFIG is set to 5s, and the simulation parameters in the system are shown in Table 1.

[0115] Table 1 Parameters of each generator in the system

[0116]

[0117] Under full-operation conditions at the wind farm, the wind speed, wind farm output power, and wind farm outlet frequency at the same time period are as follows: Figures 3(a) to 3(c) As shown, wind speed fluctuates frequently and uncertainly over a short period, with wind farm output fluctuating within a range of 0.04 Hz around 40% of maximum output. A virtual inertia control strategy is used to smooth out the impact of wind power uncertainty and fluctuations in response to frequency changes, with frequency deviations within 0.05 Hz. After grid connection, the system operates safely and stably, and the frequency fluctuation trend is consistent with wind power fluctuations. This active power output response to dynamic frequency changes demonstrates that virtual inertia support for wind farms is equally applicable in the system's power-frequency dynamic process.

[0118] The simulation results show that the virtual inertia of the new energy unit has a supporting effect on the power system inertia. The present invention can accurately evaluate the system inertia for different operating conditions of the new energy unit, as shown in Table 2.

[0119] Table 2 System Inertia under Different Operating Conditions

[0120]

[0121] Under full-operation conditions of the wind farm, a 30-second data window is used to evaluate the system inertia of the measurement data, and the identification model parameters are updated every 15 seconds to achieve continuous evaluation. The disturbance level δ of the method in this invention is taken as 1. The identification problem considering the influence of uncertainty is transformed into a solution formula (15), and the identification results are compared and analyzed with those of the least squares method (LS) without considering the influence of uncertainty. The identification results of the two methods are shown in Table 3.

[0122] Table 3 Comparison of System Inertia Assessment Results

[0123]

[0124] According to the results in Table 3, the method proposed in this invention can obtain a more accurate system inertia. Unlike the approximately 5% error in system inertia identified by the LS method, the accuracy of the RLS method is improved by an average of 2.37%. The evaluation accuracy is significantly affected by the uncertainty of the new power system. Based on the real-time updated inertia evaluation results, the frequency of the subsequent 1-minute time period is verified according to Equation (1), and the results are as follows. Figures 4(a)-4(d) As shown.

[0125] Figures 4(a)-4(d) The system frequency deviation corresponding to the system inertia is updated sequentially at 15-second intervals. During the selected test time, the system frequency fluctuates continuously over time, with variations within ±0.05Hz, indicating that the system did not experience significant disturbances. In the real-time updated evaluation model, the frequency deviation tested by the RLS method proposed in this invention is closer to the actual value than that tested by the LS method. A comparative analysis is conducted using 35 seconds as an example: at this moment... Figures 4(a)-4(c) The results in the window show that the accuracy of the evaluation results updated by the RLS method is 20.02% and 8.18% higher than that of the LS method, respectively. The accuracy of the RLS method after updating Figure 4(b) is 6.85% higher than that in Figure 4(a). The evaluation model with real-time updates performs better.

[0126] In summary, the method described in this invention can utilize uncertainty information for identification, unlike traditional methods where perturbations are limited to probability distributions, thus better handling the uncertainty of the identification data. The real-time updated model results are closer to the actual values, ensuring the accuracy of the identification. Starting from the worst-case scenario of the residuals in the uncertain data set, identification is achieved by minimizing the residuals, which has a strong suppression effect on external noise. This invention's method considers the uncertainty of the output of new energy units and evaluates the system inertia by monitoring the fluctuation signal of the system's center frequency, exhibiting stronger robustness.

Claims

1. A method for continuous evaluation of power system inertia considering uncertainty system identification, characterized in that... Includes the following steps: Step 1: Based on the phasor acquisition unit (PMU), the frequency and power data of the system are acquired in real time, the frequency and power data are preprocessed, and an evaluation dataset is established. Step 2: Based on the frequency dynamic characteristic equation, and using the output error model, establish the system inertia identification model; Step 3: For the inertia parameter problem of the system inertia identification model established in Step 2, considering the uncertainty of input and output data, the identification problem is transformed into an optimization problem based on the robust least squares (RLS) method to achieve system inertia identification under the condition of uncertain input and output data; In step 2, the transfer function of the system inertia identification model is analyzed based on the inertia response characteristics of the actual system: （3）； In the formula, For the frequency deviation of the center of inertia, It is the change in system power difference. It is the system inertia. For system damping, s It is the Laplace operator; express The representation over the complex field; In step 2, the unit impulse response of the system inertia response is: （4）； In equation (4), h(t) is obtained from the transfer function in equation (3). When the input is a unit impulse, the corresponding output is obtained by mathematically transforming the transfer function through the inverse Laplace transform. This indicates that h(t) decays exponentially; t represents time. When, equation (4) is the initial response of the inertia under disturbance, and numerically it is the reciprocal of the system inertia; The higher-order identification model of the system's inertial response is reduced to a first-order transfer function form. The output error model is described in least squares format using the least squares method: ； Define data vector and parameter vector ,in: ， ； ， 。 2. The method for continuous evaluation of power system inertia considering uncertainty system identification according to claim 1, characterized in that: In step 1, the required data information on the dynamic characteristics of the power grid system is obtained from the PMU measurement data center, including: the unbalanced power output of the generator sets. Node frequency change rate By preprocessing the data, the inertia characteristic parameters of each unit were obtained. and the center frequency of power grid inertia .

3. The method for continuous evaluation of power system inertia considering uncertainty system identification according to claim 1, characterized in that: In step 3, based on the established system inertia identification model, the uncertainty of the actual system inertia identification data is described as follows: （9）； The error is due to the system inertia. For system input, The input disturbance noise, Let m be the system inertia parameter of the i-th unit, m be the total number of units, and i represent the unit number. For error; （10）； In the formula, for m The first dimension of the identity matrix i List, To identify the model output error in the system, For system input, This refers to the output disturbance noise; The identification problem is transformed into an optimization problem by using the robust least squares (RLS) method, so that the identification results of unknown parameters can minimize the residuals. （11）； The error is due to the system inertia. Let m be the vector space. Used to define the range of residuals For parameter vectors, , in, , ; in , ; ， ； In the formula: The input to the output error model is the power change rate; The output of the output error model is the inertia center frequency deviation. This refers to the actual, unobservable output within the system. This is system noise; , define to simplify calculation, and All belong to , Represents the residual of the i-th unit. This represents the maximum value of the residual modulus. If the worst residual exists, then the minimum For the RLS solution of equation (11), we need to check the existence of the residual in the worst case before solving it. definition , , , To simplify the expression; This represents the inertia of the first unit. ~ Let m represent the inertia of each of the m generator units. It is a matrix representing the output of the output error model. ~ Separate matrices The various elements, for transpose, The first column is The lower triangular Toeplitz matrix is ​​represented as: Since the optimization problems of residuals and the squared residuals are equivalent, we set... Equation (11) is equivalent to: （12）； in, This represents the transpose of the residual. Represents transposition; At this point, the squared residual is about the uncertain data. Given a quadratic function, and real-time sampled data belonging to a continuous and bounded set, it can be seen that the maximum and minimum values ​​of this quadratic function exist, thus enabling the optimization solution of the residuals.

4. The method for continuous evaluation of power system inertia considering uncertainty system identification according to claim 3, characterized in that: Let auxiliary variables By applying the S-lemma to optimize the quadratic form function of the residual in equation (12), we obtain: （13）； Indicates the transpose of g; If and only if When, equation (13) is rewritten as about Functional form for two variables: （14）； It is a time constant. It is the identity matrix, obviously Under constraints The worst residual can be solved by... The semidefinite programming problem of the function is obtained by SDP; By using the Schul complement to constrain elementary transformations in equation (14), the robust identification problem can be transformed into Solving the minimization of the convex optimization problem: （15）； Equation (15) can be solved using a general SDP solver, which enables the identification of system inertia under uncertain conditions.

Images