An inertia level estimation method for a high-proportion new energy new-type power system

Abstract

This invention discloses a method for estimating the inertia of a high-proportion renewable energy power system. The steps are as follows: First, acquire frequency deviation and power disturbance data collected when power disturbances occur in the power system, and preprocess the acquired data using a first-order inertial element. Second, construct a parameterized controlled autoregressive moving average (CARMA) model containing a stacked white noise vector to describe the dynamic relationship between power disturbance and frequency deviation, further transforming the system inertia assessment problem into a model parameter identification problem. Third, solve for the unknown parameters in the CARMA model using a least squares iterative algorithm. Fourth, use impulse invariant transformation to transform the transfer function from the discrete domain to the continuous domain, making the discrete transfer function continuous. Finally, use the continuous transfer function to solve for the initial value of the impulse response to obtain the system inertia. This invention effectively improves the robustness of the power grid inertia parameter estimation method, while reducing the impact of the measurement accuracy of data at a single moment, thus improving the parameter estimation accuracy and fitting degree.

Description

Technical Field

[0001] This invention relates to a method for estimating the inertia level of a power system, and in particular to a method for estimating the inertia level of a new power system with a high proportion of new energy sources. Background Technology

[0002] With the rapid development of new energy sources and the advancement of clean energy substitution, the number and capacity of asynchronous power sources such as wind power and photovoltaics connected to the system are constantly increasing. In actual operation, these asynchronous power sources cannot actively provide inertia support to the system, and the equivalent inertia of the power grid decreases accordingly, which increases the risk of system frequency instability. When the power of the power grid fluctuates significantly, it will cause the system frequency to drop rapidly, which may lead to major power outages and huge economic losses.

[0003] When power imbalance exists in high-proportion renewable energy grid lines, considering the power system's inertial response and primary frequency regulation process, the system releases rotor kinetic energy into electromagnetic power while simultaneously adjusting the system's active power input, responding to power changes in real time and dynamically compensating for unbalanced power to suppress frequency fluctuations. Solving for the system inertia level within the rotor motion equations of synchronous generators, which include system inertial response and primary frequency regulation, is essentially a system parameter identification problem. Traditional methods suffer from insufficient evaluation accuracy and require improved fitting speed. Summary of the Invention

[0004] Purpose of the invention: The purpose of this invention is to provide a method for estimating the inertia level of a new type of power system with a high proportion of new energy sources, thereby enabling a rapid and accurate estimation of the power system's inertia.

[0005] Technical solution: The present invention provides a method for estimating the inertia level of a new type of power system with a high proportion of new energy sources, comprising the following steps:

[0006] (1) Acquire frequency deviation and power disturbance data collected when power disturbance occurs in the power system, and preprocess the acquired data through a first-order inertial element to effectively avoid overfitting during system identification and improve system identification accuracy.

[0007] (1.1) Construct a power system model and set power disturbances of different scales in the system.

[0008] (1.2) Collect power disturbance time series data ΔP and frequency deviation time series data Δf under power disturbance conditions.

[0009] (1.3) The collected data is filtered by a first-order inertial circuit to remove high-frequency noise interference and improve the identification accuracy.

[0010] (2) Construct a CARMA model to describe the dynamic relationship between power disturbance and frequency deviation, and further transform the system inertia assessment problem into the model parameter identification problem.

[0011] (2.1) The power disturbance data is used as the input sequence u(t), the frequency deviation data is used as the output sequence y(t), and a white noise sequence v(t) is introduced to simulate the random disturbances to the system.

[0012] (2.2) Construct a standard CARMA model to describe the dynamic relationship between power disturbance and frequency deviation.

[0013] (2.3) Extract the specific unknown parameters to be identified in the CARMA model.

[0014] (3) Based on the input and output data obtained in step (1), the unknown parameters in the CARMA model are solved using the least squares iterative algorithm; a stacked output vector and a stacked information matrix are constructed from the power disturbance and frequency deviation data, the criterion function constructed by the least squares principle is used as the objective function, and the unmeasurable noise is processed by the hierarchical idea to continuously update the least squares iterative estimate. The iterative results identified the unknown parameters in the CARMA model, and the discrete transfer function G was further obtained. HP (z).

[0015] (3.1) Construct the stacked output vector Y(t) and the stacked information matrix φ(t) using the collected system node frequency deviation and power disturbance data.

[0016] (3.2) Construct the least squares identification model of CARMA model based on the stacked output vector Y(t) and the stacked information matrix φ(t).

[0017] (3.3) Construct a criterion function using the least squares principle. When the criterion function reaches its minimum value, the estimated values ​​of the model parameters can be obtained.

[0018] (3.4) The noise term is processed by the hierarchical approach and the least squares iteration result of the parameter estimate is continuously updated to identify the k-th iteration result of the unknown parameter in the CARMA model.

[0019] (3.5) Substitute the identified parameter results into the CARMA model expression to calculate the discrete transfer function G of the system. HP (z).

[0020] (4) The discrete transfer function G is obtained in step (3). HP (z) Then, the impulse invariant transform is used to transform the transfer function from the discrete domain to the continuous domain, thus making the discrete transfer function continuous as G. HP(s), and then the initial value of the impulse response can be solved using the continuous transfer function to obtain the magnitude of the system inertia.

[0021] (4.1) Using the impulse-invariant Zs transform, the discrete transfer function G of the system is obtained. HP (z) is transformed into a continuous transfer function G in the s-domain. HP (s).

[0022] (4.2) By continuous transfer function G HP (s) The system impulse response expression g is obtained by inverse Laplace transform. HP (t).

[0023] (4.3) The system inertia H is derived from the rotor motion equation of the synchronous generator set. sys With g HP The relationship between (t) is used to finally determine the system inertia value H. sys .

[0024] (4.3.1) Perform a Laplace transform on both sides of the rotor motion equation of the synchronous generator set.

[0025] (4.3.2) Derive the transfer function expressions for the two cases where the frequency change is less than the frequency modulation dead zone threshold and the frequency change is greater than the frequency modulation dead zone threshold, respectively.

[0026] (4.3.3) The system inertia H is derived from the initial value theorem and the transfer function expression. sys With g HP A unified relational expression between (t).

[0027] (4.3.4) Based on the derived H sys With g HP The relationship between (t) and the system inertia H can be obtained by solving the time-domain impulse response function obtained in (4.2). sys .

[0028] A computer storage medium storing a computer program that, when executed by a processor, implements the aforementioned method for estimating the inertia level of a new power system with a high proportion of new energy sources.

[0029] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-described method for estimating the inertia level of a new type of power system with a high proportion of new energy sources.

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

[0031] 1. This invention uses a parametric CARMA model to characterize the system, while considering the random factors that affect the system. It introduces a stacked white noise vector into the model and uses a hierarchical identification principle to process the noise term, which can effectively improve the robustness of the power grid inertia parameter estimation method, while reducing the impact of the measurement accuracy of data at a single moment, and improving the parameter estimation accuracy and fitting degree.

[0032] 2. For the CARMA model, this invention adopts the least squares identification method, using the mean square error as the criterion function. The calculation is relatively simple and convenient, the convergence speed is fast, and the prediction error can be effectively reduced.

[0033] 3. The method proposed in this invention has high identification accuracy under different types of disturbances. It can be applied to offline estimation of the inertia of high-proportion new energy power systems after large disturbance events, and is also suitable for online monitoring of small disturbances in system inertia. Attached Figure Description

[0034] Figure 1 This is a schematic diagram of the method described in this invention;

[0035] Figure 2 A schematic diagram of the system structure described by the CARMA model;

[0036] Figure 3 This is a schematic diagram of the IEEE 39-node system model simulated in the embodiment.

[0037] Figure 4 A comparison chart of system inertia assessment results under different disturbances. Detailed Implementation

[0038] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0039] like Figure 1 As shown, a method for estimating the inertia level of a new power system with a high proportion of new energy sources includes the following steps:

[0040] (1) Acquire frequency deviation and power disturbance data collected when power disturbance occurs in the power system, and preprocess the acquired data through a first-order inertial element to effectively avoid overfitting during system identification and improve system identification accuracy.

[0041] (1.1) Construct a power system model and set power disturbances of different scales in the system.

[0042] A simulation model of the actual power system requiring inertia estimation is constructed and debugged to normal operating conditions. Subsequently, random power disturbances of different scales are set at the system nodes.

[0043] (1.2) Collect power disturbance time series data ΔP and frequency deviation time series data Δf under power disturbance conditions.

[0044] Find the time point where the disturbance occurs in the model's output data, and collect the power disturbance time series data ΔP and frequency deviation time series data Δf for a period of time after the disturbance occurs.

[0045] (1.3) The collected data is filtered by a first-order inertial circuit to remove high-frequency noise interference and improve the identification accuracy.

[0046] The acquired data needs to be preprocessed using a first-order inertial element. Low-pass filtering is performed to remove high-frequency noise interference. Note that the appropriate inertial time constant T needs to be adjusted for different systems. F When the inertial time constant T of the introduced first-order inertial element F When the time constant T is too small, the low-pass filtering effect is poor, and there is still a lot of high-frequency interference; while when the time constant T is too small, the low-pass filtering effect is poor, and there is still a lot of high-frequency interference. F When the value is too large, the system response becomes severely delayed, which will have a significant impact on the original waveform data and affect the final identification results.

[0047] (2) Construct a CARMA model to describe the dynamic relationship between power disturbance and frequency deviation, and further transform the system inertia assessment problem into the model parameter identification problem.

[0048] (2.1) The power disturbance data is used as the input sequence u(t), the frequency deviation data is used as the output sequence y(t), and a white noise sequence v(t) is introduced to simulate the random disturbances to the system.

[0049] The power disturbance time series data ΔP, when the system experiences unbalanced power disturbances, is used as the model's input sequence u(t), and the frequency disturbance time series data Δf is used as the model's observed output sequence y(t). A mean of 0 and a variance of σ are introduced. 2 A white noise sequence v(t) is used to simulate the random disturbances experienced by the system.

[0050] (2.2) Construct a standard CARMA model to describe the dynamic relationship between power disturbance and frequency deviation. For example... Figure 2 As shown, the generator rotor motion equations are modeled as a CARMA model, as follows:

[0051] A(z)y(t)=B(z)u(t)+D(z)v(t)

[0052] In the formula, u(t) and y(t) are the input sequence and observation output sequence of the CARMA model, respectively; v(t) is the white noise sequence; and A(z), B(z) and D(z) are the model parameters.

[0053] (2.3) Extract the specific unknown parameters to be identified in the CARMA model

[0054] A(z), B(z), and D(z) are model parameters, which are the unit shift operator z. -1 A time-invariant polynomial with constant coefficients can be expanded as follows:

[0055]

[0056] In the formula, z -n (n = 1, 2, 3...) is the shift operator, a n b n d n (n = 1, 2, 3...) are the unknown parameters in the specific expansion of the model parameters.

[0057] Therefore, it can be seen that in the standard CARMA model, (n) need to be identified. a +n b +n d () unknown parameters.

[0058] (3) Based on the input and output data obtained in step (1), the unknown parameters in the CARMA model are solved using the least squares iterative algorithm; a stacked output vector and a stacked information matrix are constructed from the power disturbance and frequency deviation data, the criterion function constructed by the least squares principle is used as the objective function, and the unmeasurable noise is processed by the hierarchical idea to continuously update the least squares iterative estimate. The iterative results identified the unknown parameters in the CARMA model, and the discrete transfer function G was further obtained. HP (z).

[0059] (3.1) Construct the stacked output vector Y(t) and the stacked information matrix φ(t) using the collected system node frequency deviation and power disturbance data.

[0060] To identify unknown parameters using the least squares iterative algorithm, we first need to construct the stacked output vector Y(t) and the stacked information matrix φ(t) using the collected system node frequency deviation and power disturbance data.

[0061] (3.2) Construct the least squares identification model of CARMA model based on the stacked output vector Y(t) and the stacked information matrix φ(t).

[0062] The least-squares identification model of this CARMA model is expressed as:

[0063] Y(t)=φ(t)θ+V(t)

[0064]

[0065] In the formula, Y(t) is the stacked output matrix, φ(t) is the stacked information matrix, V(t) is the stacked noise matrix, θ is the vector of parameters to be identified, and y(t) is the output vector (i.e., frequency deviation data). This is an information vector.

[0066] Among them, information vector This can be further expanded to:

[0067]

[0068] In the formula, y(t) represents the frequency deviation data, u(t) represents the power disturbance data, and v(t) represents the stacked white noise vector.

[0069] (3.3) Construct a criterion function using the least squares principle. When the criterion function reaches its minimum value, the estimated values ​​of the model parameters can be obtained.

[0070] To minimize the prediction error, a criterion function is constructed using the least squares principle:

[0071] J(θ) = ||Y(t) - φ(t)θ|| 2

[0072] In the formula, Y(t) is the stacked output matrix, φ(t) is the stacked information matrix, and θ is the parameter vector to be identified.

[0073] When the criterion function J(θ) reaches its minimum value, the estimated values ​​of the model parameters can be obtained, that is, the least squares iterative estimates of the parameter vector θ. Satisfying the equation:

[0074]

[0075] In the formula, Let Y(t) be the estimated vector of the parameters to be identified, Y(t) be the stacked output matrix, and φ(t) be the stacked information matrix.

[0076] (3.4) The noise term is processed by the hierarchical approach and the least squares iteration result of the parameter estimate is continuously updated to identify the k-th iteration result of the unknown parameter in the CARMA model.

[0077] Because an unmeasurable noise term v(ti) is introduced during system identification, where i = 0, 1, ..., p-1, a hierarchical identification principle is adopted to quantitatively describe the impact of the noise term on system identification. First, each noise term v(ti) is assigned a random initial value, and the information vector is... The unknown quantity v(ti) that appears in the process is estimated by its (k-1)th iteration value v. k-1(ti) is used as a substitute to obtain The expression for the k-th iteration is as follows:

[0078]

[0079] Finally, the least squares iterative estimate is obtained. The result of the kth iteration:

[0080]

[0081] In the formula, Y(t) is the estimated vector of the parameters to be identified after k iterations, and Y(t) is the stacked output matrix. k (t) is the stacked information matrix after k iterations.

[0082] (3.5) Substitute the identified parameter results into the CARMA model expression to calculate the discrete transfer function G of the system. HP (z).

[0083] Substituting the parameters obtained after least squares identification back into the CARMA model expression yields the discrete transfer function G of the system. HP (z).

[0084] (4) The discrete transfer function G is obtained in step (3). HP (z) Then, the impulse invariant transform is used to transform the transfer function from the discrete domain to the continuous domain, thus making the discrete transfer function continuous as G. HP (s), and then the initial value of the impulse response can be solved using the continuous transfer function to obtain the magnitude of the system inertia.

[0085] (4.1) Using the impulse-invariant Zs transform, the discrete transfer function G of the system is obtained. HP (z) is transformed into a continuous transfer function G in the s-domain. HP (s).

[0086] The system's discrete transfer function G is obtained by using the impulse-invariant Zys transform. HP (z) is converted to the s-domain continuous transfer function G. HP (s), the conversion formula is as follows:

[0087]

[0088] In the formula, T is the sampling period, and the integration range is the area surrounding G. HP (z) Curves of all poles.

[0089] (4.2) By continuous transfer function G HP (s) The system impulse response expression g is obtained by inverse Laplace transform. HP(t).

[0090] G HP (s) The system's time-domain impulse response expression g can be obtained through the inverse Laplace transform. HP (t).

[0091] (4.3) The system inertia H is derived from the rotor motion equation of the synchronous generator set. sys With g HP The relationship between (t) is used to finally determine the system inertia value H. sys .

[0092] (4.3.1) Perform a Laplace transform on both sides of the rotor motion equation of the synchronous generator set.

[0093] Considering only the inertial response and primary frequency regulation of the synchronous generator set, the rotor motion equation of the synchronous generator set is:

[0094]

[0095] In the formula, H sys The system inertia level is represented by Δf(t); the frequency change is represented by ΔP. m ΔP L These are the active power adjustment amount of the first frequency regulation and the difference in active power after the disturbance, respectively. The sum of the two can be written as ΔP; D is the damping coefficient.

[0096] The input to this equation is ΔP, and the output is Δf(t), which means that when the system experiences an unbalanced power disturbance ΔP, the frequency deviation generated under the action of the rotor motion equation is Δf.

[0097] By performing a Laplace transform on both sides of the rotor motion equation, we can easily obtain:

[0098] 2H sys sΔf(s)=ΔP(s)-DΔf(s)

[0099] (4.3.2) Derive the transfer function expressions for the two cases where the frequency change is less than the frequency modulation dead zone threshold and the frequency change is greater than the frequency modulation dead zone threshold, respectively.

[0100] When the frequency change Δf is less than the dead zone threshold for initiating frequency modulation, the frequency response can be considered to exclude primary frequency modulation. Therefore, the s-domain transfer function G can be obtained at this time. HP (s) is:

[0101]

[0102] In the formula, G HP H(s) is the transfer function of the system in the s-domain. sys Let D be the system inertia level, and D be the damping coefficient.

[0103] When the frequency change Δf is greater than the frequency modulation dead zone threshold, the first-order inertial element and the primary frequency modulation element are considered simultaneously. Combining the primary frequency modulation closed-loop feedback, the s-domain transfer function G can be obtained at this time. HP (s):

[0104]

[0105] In the formula, G HP H(s) is the transfer function of the system in the s-domain. sys Let D be the system inertia level, and D be the damping coefficient.

[0106] (4.3.3) The system inertia H is derived from the initial value theorem and the transfer function expression. sys With g HP A unified relational expression between (t).

[0107] The same time-domain results can be obtained from equations (1) and (2) using the initial value theorem:

[0108]

[0109] In the formula, g HP (t) is the time-domain impulse response function of the system, G HP H(s) is the transfer function of the system in the s-domain. sys This represents the system inertia level.

[0110] (4.3.4) Based on the derived H sys With g HP The relationship between (t) and the system inertia H can be obtained by solving the time-domain impulse response function obtained in (4.2). sys .

[0111] From equation (3), it can be seen that when the system is disturbed, if the initial value g of the impulse response of the transfer function in the time domain is known... HP (0) allows us to determine the equivalent inertia H of the system. sys Solve the problem. The time-domain impulse response function g of the system model has already been obtained in (4.2). HP (t), therefore the system's equivalent inertia H sys It can be obtained from the following formula:

[0112]

[0113] In the formula, H sys For the system inertia level, g HP (0) is the value of the system’s time-domain impulse response function at t=0.

[0114] Simulation verification:

[0115] To verify the reliability and accuracy of the system inertia estimation method based on the controlled autoregressive moving average model proposed in this invention under different perturbations, an IEEE 39-node system model was first built on the PSCAD / EMTDC simulation platform (as shown in the attached figure). Figure 3 As shown in the figure, two different fault disturbance states, large disturbance and random small disturbance, were set to obtain disturbance data for system inertia assessment. The acquired data first needs to be noise processed. Simulation tests show that when the inertial time constant T of the introduced first-order inertial element... F When the value is less than 5, the low-pass filtering effect is poor, while when the time constant T is less than 5, the low-pass filtering effect is poor. F When the value is greater than 10, the system response exhibits severe hysteresis, which significantly impacts the original waveform data. Therefore, in this embodiment, the time constant T of the first-order inertial element is selected. F =5s to filter the frequency signal, which improves the anti-interference ability of the model without affecting the accuracy of the data, thereby improving the accuracy of system inertia identification.

[0116] In this embodiment, the large disturbance fault acts on the system in the form of a large load disconnection from the grid. In the constructed IEEE 39-node system model, the load (600MW) under node 8 is disconnected at t=58s, and the system recovers to a stable state after 6s. The operating information of 58-62s, during which the system power and frequency changes significantly under this large disturbance fault, is selected as the sampling data. 1000 sets of frequency and power data of each generator are collected with a sampling step size of 0.004s. After processing the collected data, the unbalanced power disturbance ΔP(t) and frequency deviation value Δf(t) time series are obtained as the input and output data of the identification model, respectively. The least squares iterative algorithm is used to estimate the inertia of each generator set during this time period, and the optimal fitting result is identified. After obtaining the identification result of the system transfer function through the above steps, the inertia value of each generator set in the model can be calculated according to equation (3). The results are shown in Table 1.

[0117] Table 1. Estimation results of generator unit inertia under large disturbances.

[0118]

[0119] By comparing the inertia estimate obtained by this algorithm under large disturbance with the actual inertia value of the generator set, it can be seen that under the condition of large disturbance in the system, the inertia estimate of each unit in the whole network can closely match the actual value. The maximum error of inertia estimation is 7.21%, and the minimum error is 2.88%, which verifies that the algorithm has high accuracy in inertia identification under large disturbance scenarios.

[0120] In this embodiment, a small disturbance fault is set during the stable operation of the system from t=80 to 88s, with a total random small load fluctuation of 108MW set at system buses 7, 10, 19, 25, and 29. The set random small disturbance acts on the system with a frequency variation of 1Hz, and at the same time, the frequency and power data of each generator within 80 to 88s are collected with a sampling step size of 0.004s as the source of system identification data. A total of 1905 sets of frequency and power data of each generator set are collected during the time period of the random small disturbance. The inertia of each generator set during this time period is estimated by the least squares iterative algorithm, and the optimal fitting result is identified. The inertia value of each generator set in the model can be calculated according to equation (3), and the results are shown in Table 2.

[0121] Table 2. Estimation results of generator unit inertia under random small disturbances.

[0122]

[0123] Based on the results in Table 2, a comparison between the estimated inertia values ​​and the actual inertia parameters of each unit shows that the inertia estimation error for all units in the network is controlled within a small range, indicating high identification accuracy. The maximum error in unit inertia estimation is 7.10%, and the minimum error is only 1.16%.

[0124] Appendix Figure 4 A comparison chart of system inertia evaluation results under different disturbances is presented. It can be seen that the identification method proposed in this invention has high identification accuracy and good adaptability for both large and small disturbance data.

Claims

1. A method for estimating the inertia level of a new power system with a high proportion of new energy sources, characterized in that, Includes the following steps: (1) Acquire frequency deviation and power disturbance data collected when power disturbance occurs in the power system, and preprocess the acquired data through a first-order inertial link to effectively avoid overfitting during system identification and improve system identification accuracy. (2) Construct a CARMA model to describe the dynamic relationship between power disturbance and frequency deviation, and further transform the system inertia assessment problem into a model parameter identification problem. (3) Based on the input and output data obtained in step (1), the unknown parameters in the CARMA model are solved using the least squares iterative algorithm; a stacked output vector and a stacked information matrix are constructed from the power disturbance and frequency deviation data, the criterion function constructed by the least squares principle is used as the objective function, and the unmeasurable noise is processed by the hierarchical idea to continuously update the least squares iterative estimate. The iterative results identified the unknown parameters in the CARMA model, and the discrete transfer function G was further obtained. HP (z); (4) The discrete transfer function G is obtained in step (3). HP (z) Then, the impulse invariant transform is used to transform the transfer function from the discrete domain to the continuous domain, thus making the discrete transfer function continuous as G. HP (s), and then the initial value of the impulse response can be solved using the continuous transfer function to obtain the magnitude of the system inertia.

2. The method for estimating the inertia level of a new power system with a high proportion of new energy sources according to claim 1, characterized in that, The specific steps (1) are as follows: (1.1) Construct a power system model and set power disturbances of different scales in the system; (1.2) Collect power disturbance time series data ΔP and frequency deviation time series data Δf under power disturbance conditions; (1.3) The collected data is filtered by a first-order inertial circuit to remove high-frequency noise interference and improve the identification accuracy.

3. The method for estimating the inertia level of a new power system with a high proportion of new energy sources according to claim 1, characterized in that, Step (2) specifically involves: (2.1) The power disturbance data is used as the input sequence u(t), the frequency deviation data is used as the output sequence y(t), and a white noise sequence v(t) is introduced to simulate the random disturbances to the system. (2.2) Construct a standard CARMA model to describe the dynamic relationship between power disturbance and frequency deviation; (2.3) Extract the specific unknown parameters to be identified in the CARMA model.

4. The method for estimating the inertia level of a new power system with a high proportion of new energy sources according to claim 1, characterized in that, Step (3) specifically involves: (3.1) Construct the stacked output vector Y(t) and the stacked information matrix φ(t) using the collected system node frequency deviation and power disturbance data; (3.2) Construct the least squares identification model of the CARMA model based on the stacked output vector Y(t) and the stacked information matrix φ(t); (3.3) Construct a criterion function using the least squares principle. When the criterion function reaches its minimum value, the estimated values ​​of the model parameters can be obtained. ; (3.4) The noise term is processed by the hierarchical approach and the least squares iteration result of the parameter estimate is continuously updated to identify the k-th iteration result of the unknown parameter in the CARMA model; (3.5) Substitute the identified parameter results into the CARMA model expression to calculate the discrete transfer function G of the system. HP (z).

5. The method for estimating the inertia level of a new type of power system with a high proportion of new energy sources according to claim 1, characterized in that, Step (4) specifically involves: (4.1) Using the impulse-invariant Zs transform, the discrete transfer function G of the system is obtained. HP (z) is transformed into a continuous transfer function G in the s-domain. HP (s); (4.2) By continuous transfer function G HP (s) The system impulse response expression g is obtained by inverse Laplace transform. HP (t); (4.3) The system inertia H is derived from the rotor motion equation of the synchronous generator set. sys With g HP The relationship between (t) is used to finally determine the system inertia value H. sys； (4.3.1) Perform a Laplace transform on both sides of the rotor motion equation of the synchronous generator set; (4.3.2) Derive the transfer function expressions for the two cases: when the frequency change is less than the frequency modulation dead zone threshold and when the frequency change is greater than the frequency modulation dead zone threshold. (4.3.3) The system inertia H is derived from the initial value theorem and the transfer function expression. sys With g HP A unified relational expression between (t); (4.3.4) Based on the derived H sys With g HP The relationship between (t) and the system inertia H can be solved from the time-domain impulse response function obtained in (4.2). sys .

6. A computer storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements a method for estimating the inertia level of a new type of power system with a high proportion of new energy sources as described in any one of claims 1-5.

7. 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 computer program, it implements a method for estimating the inertia level of a new type of power system with a high proportion of new energy sources as described in any one of claims 1-5.

Images