Characterization method of power grid frequency modulation AGC signal

Abstract

This invention discloses a method for characterizing a power grid frequency regulation AGC signal, comprising the following steps: S1. Setting a statistical time interval for the power grid frequency regulation AGC signal and obtaining historical data of the amplitude of the power grid frequency regulation AGC signal under a fixed statistical time interval; S2. Preprocessing the historical data of the amplitude of the power grid frequency regulation AGC signal and converting the absolute value of the amplitude of the power grid frequency regulation AGC signal into a change in the AGC signal reflecting AGC demand; S3. Performing probability distribution statistics on the change in the AGC signal, considering the impact of renewable energy and load on AGC demand, and establishing an autoregressive model reflecting the distribution of AGC demand based on the statistical results of the probability distribution of the change in the AGC signal; S4. Determining the parameters of the autoregressive model reflecting the distribution of AGC demand, and finally obtaining a power grid frequency regulation AGC signal model considering the impact of renewable energy and load on AGC demand; S5. Using the power grid frequency regulation AGC signal model considering the impact of renewable energy and load on AGC demand, characterizing the uncertainty of the power grid frequency regulation AGC signal.

Description

Technical Field

[0001] This invention belongs to the field of new power technology, specifically, it relates to a method for characterizing power grid frequency regulation AGC signals. Background Technology

[0002] As is well known, power system frequency deviation reflects the power imbalance between generation and load in the system. Power systems have certain peak-shaving and frequency regulation needs, which are distinguished based on their time scale and objectives. Peak-shaving has a longer time scale and its main purpose is economic dispatch of the power system; frequency regulation has a shorter time scale and its main purpose is the safe and stable operation of the power system. Power system peak-shaving refers to the process of adjusting various links in the power system, such as generation, transmission, storage, and consumption, to match supply and demand in order to address the mismatch between power supply and load. Taking the generation side as an example, peak-shaving includes start-stop peak-shaving and deep peak-shaving. Start-stop peak-shaving (upward peak-shaving) refers to the additional commissioning of generator units beyond normal operation during peak load periods to increase power output. Deep peak-shaving (downward peak-shaving) refers to power plants reducing output to meet off-peak loads or the consumption of new energy sources.

[0003] Power system frequency regulation mainly includes primary frequency regulation and secondary frequency regulation. When power fluctuations occur in the power system, generator units respond quickly to second-level fluctuations through primary frequency regulation. Secondary frequency regulation addresses power deviations that primary frequency regulation cannot balance by using AGC (Automatic Generation Control) units for minute-level power balancing. AGC is an important function in the Energy Management System (EMS). Generally, the AGC unit receives frequency regulation signals from the power grid control center and adjusts accordingly, performing real-time output control with specific response speeds, adjustment rates, and adjustment precision to achieve real-time power balance. This meets the constantly changing power demands of users, conforms to control performance standards, maintains system frequency stability, and ensures the system operates in an economical state. In a standalone power system, AGC controls the frequency of that system's frequency through the frequency regulation unit. In a combined power system, AGC operates on a regional system basis, with each region controlling the output of its generators.

[0004] The global energy transition is accelerating, and my country has proposed building a new power system based on new energy sources. Distributed power generation systems, represented by photovoltaic and wind power, have been widely adopted. However, the intermittency and volatility of wind and solar power pose a significant challenge to grid frequency stability.

[0005] Regarding frequency regulation requirements, the minute-level frequency regulation response requirements mainly consist of two parts: a deterministic part and an uncertain part. The deterministic part primarily addresses insufficient power source ramp-up rates caused by the "duck curve," while the uncertain part mainly includes minute-level uncertainties in load, wind power output, and solar power output. Currently, there is still no mature calculation method for the minute-level frequency regulation requirements of the power system after the integration of renewable energy sources such as wind and solar power.

[0006] For power systems with low early renewable energy penetration, the uncertainty of frequency regulation AGC signals is mainly affected by load fluctuations, which have a large time scale and are highly predictable, resulting in low uncertainty. However, for power systems with high renewable energy penetration, the frequency regulation AGC signals are affected by both load fluctuations and renewable energy (wind power, photovoltaic) fluctuations. Furthermore, the time scales of renewable energy and load fluctuations are different, and they do not have the characteristic of mutually weakening or canceling each other out. The superposition of these two factors causes significant uncertainty, making it impossible to predict accurately using traditional experience-based methods.

[0007] Therefore, there is an urgent need for a probability distribution-based method to mathematically describe frequency-modulated AGC signals with significant uncertainties, so as to build a model and use it for frequency modulation-related research and applications. Summary of the Invention

[0008] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for characterizing the frequency regulation AGC signal of the power grid. This invention can fully consider the significant uncertainty of the AGC signal in the power grid with high penetration of renewable energy, establish a model of the frequency regulation AGC signal of the power grid, and the relevant signal characteristics can provide a reference for operators to schedule AGC units to participate in frequency regulation services.

[0009] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0010] A method for characterizing power grid frequency modulation AGC signals includes the following steps:

[0011] S1. Set the statistical time interval for the power grid frequency regulation AGC signal and obtain historical data of the power grid frequency regulation AGC signal amplitude under a fixed statistical time interval;

[0012] S2. Preprocess the historical data of the amplitude of the power grid frequency regulation AGC signal, and convert the absolute value of the amplitude of the power grid frequency regulation AGC signal into the change of the AGC signal reflecting the AGC demand;

[0013] S3. Perform probability distribution statistics on the changes in AGC signals, consider the impact of renewable energy and load on AGC demand, and establish an autoregressive model reflecting the distribution of AGC demand based on the probability distribution statistics of the changes in AGC signals.

[0014] S4. Determine the parameters of the autoregressive model that reflects the distribution of AGC demand, and finally obtain the grid frequency regulation AGC signal model that considers the impact of renewable energy and load on AGC demand;

[0015] S5. Using a grid frequency regulation AGC signal model that considers the impact of renewable energy and load on AGC demand, the uncertainty of the grid frequency regulation AGC signal is characterized.

[0016] Furthermore, in step S1, both the amplitude and time interval of the valid AGC signal are random. By fixing the statistical time interval of the AGC signal, only the amplitude of the AGC signal is counted. At this time, the amplitude of the AGC signal is allowed to be 0, thereby transforming the dual randomness of amplitude and time interval into single randomness.

[0017] Furthermore, in step S2, the historical data of the power grid frequency regulation AGC signal amplitude is collected, sorted, and preprocessed by removing bad data. The change in the AGC signal reflecting the AGC demand is the difference in the absolute value of the AGC signal amplitude between two adjacent time intervals.

[0018] Furthermore, in step S3, based on the statistical analysis of the probability distribution of the AGC signal change, and by comparing the AGC signal change data in different time periods, it is determined that the AGC signal change in different time periods within a day all follow a normal distribution in probability.

[0019] Furthermore, renewable energy and load exhibit varying uncertainties at different times. The variation ξ of the AGC signal follows a normal distribution within a certain time period Δt. A time period Δt = 1 hour is selected, and an autoregressive model is chosen to model the AGC signal for a typical day. Specifically:

[0020] For a typical day at 24:00, the hourly AGC amplitude data forms a normal distribution. The mean of this normal distribution is set to 0. The variance of the AGC signal distribution varies across different time periods, resulting in the following variation in the AGC signal:

[0021] ξ t ~(μ,σ t 2 t = 1, 2, ..., 24

[0022] Where, ξ t Let σ be the change in the AGC signal at time t. t 2Let be the variance of the AGC signal distribution at time t, and μ be the mean of the normal distribution, μ = 0;

[0023] For the uncertainty of typical intraday AGC demand, an autoregressive model is established for the time series variance of AGC signal distribution at different time periods:

[0024]

[0025] Right now

[0026]

[0027] Where p is the total number of terms in the regression model. Let e ​​be the autoregressive coefficient of the i-th regression model. t Let t be the white noise excitation at time t.

[0028] Furthermore, in step S4, the parameters of the autoregressive model are obtained through the observed values ​​and the YW equation system, which is as follows:

[0029]

[0030] Where k is the number of equations in the YW equation system, γ k Let γ be the autocorrelation function of the k-th equation. k-i Let δ be the autocorrelation function of the ki-th equation. e Let η be the standard deviation of white noise. k,0 The cross-correlation function of white noise;

[0031] The YW equation system with k > 0 is transformed into matrix form as follows:

[0032]

[0033] Right now

[0034] RΦ=B

[0035] In the formula, R is the autocorrelation matrix, Φ is the autoregression coefficient matrix, and B is the autocorrelation function matrix;

[0036] The autocorrelation matrix R is obtained from the observed values, and the autocorrelation function matrix B is calculated. Based on the autocorrelation matrix R and the autocorrelation function matrix B, the autoregressive coefficients are calculated. The autoregressive coefficient matrix Φ is obtained;

[0037] When k = 0, the YW equation system is:

[0038]

[0039] Based on autoregression coefficients And from the YW equations at k=0, the standard deviation δ of the white noise can be obtained.e The final grid frequency regulation AGC signal model that takes into account the impact of renewable energy and load on AGC demand is obtained.

[0040] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0041] 1. The method for characterizing the frequency regulation AGC signal of the power grid in this invention, compared with the traditional empirical method, can fully consider the significant uncertainty of the AGC signal in the power grid with high penetration of renewable energy, making the frequency regulation demand characterization model closer to the current power system reality.

[0042] 2. The method for characterizing the frequency regulation AGC signal of the power grid in this invention fixes the statistical time interval of the AGC signal and only counts the amplitude of the AGC signal. At this time, the amplitude of the AGC signal is allowed to be 0. This transforms the dual randomness of amplitude and time interval into single randomness, reduces the number of random variables, and makes the characterization process more efficient.

[0043] 3. The method for characterizing the power grid frequency modulation AGC signal of the present invention has a finer time granularity, dividing a typical day into 24 points, which can fully consider the characteristics of the frequency modulation signal in different hours and characterize it more accurately. Attached Figure Description

[0044] Figure 1 This is a flowchart illustrating the method for characterizing the power grid frequency modulation AGC signal according to the present invention.

[0045] Figure 2 This is a schematic diagram of AGC signal conversion in the power grid frequency regulation AGC signal characterization method of the present invention.

[0046] Figure 3 This is a schematic diagram of the statistical characteristics of AGC signals within an hour in a certain AGC unit in southern China. Detailed Implementation

[0047] The method for characterizing the power grid frequency modulation AGC signal of the present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0048] Please see Figure 1 This invention discloses a method for characterizing power grid frequency modulation AGC signals, comprising the following steps:

[0049] S1. Set the statistical time interval for the power grid frequency regulation AGC signal and obtain historical data of the power grid frequency regulation AGC signal amplitude under a fixed statistical time interval.

[0050] S2. Preprocess the historical data of the amplitude of the power grid frequency regulation AGC signal, and convert the absolute value of the amplitude of the power grid frequency regulation AGC signal into the change of the AGC signal reflecting the AGC demand.

[0051] S3. Perform probability distribution statistics on the changes in AGC signals, consider the impact of renewable energy and load on AGC demand, and establish an autoregressive model reflecting the distribution of AGC demand based on the probability distribution statistics of the changes in AGC signals.

[0052] S4. Determine the parameters of the autoregressive model that reflects the distribution of AGC demand, and finally obtain the grid frequency regulation AGC signal model that considers the impact of renewable energy and load on AGC demand.

[0053] S5. Using a grid frequency regulation AGC signal model that considers the impact of renewable energy and load on AGC demand, the uncertainty of the grid frequency regulation AGC signal is characterized.

[0054] Specifically, in step S1, both the amplitude and time interval of the valid AGC signal (amplitude not equal to 0) are random. This invention fixes the statistical time interval of the AGC signal and only counts the amplitude of the AGC signal, allowing the amplitude of the AGC signal to be 0. This transforms the dual randomness of amplitude and time interval into a single randomness. Mature data is obtained through data governance, and the AGC signal is transformed from an absolute quantity to a variable quantity. Probability distribution statistics are performed using a large number of samples. The distribution type is identified using the probability distribution graph, and a corresponding mathematical model is selected for modeling. After modeling, the parameters of the mathematical model can be identified using both continuous and discrete methods. Finally, a power grid frequency regulation AGC signal model under uncertain factors is established, forming a mathematical characterization method for AGC signals.

[0055] In step S2, a large amount of historical AGC frequency modulation signal data from a certain AGC unit is collected and processed, bad data is removed, and mature data is obtained after the above data processing. The absolute value of the AGC signal is converted into a change quantity ξ, such as... Figure 2 As shown, the change ξ of the AGC signal is the difference between the absolute values ​​of the AGC signal within a certain time period Δt. For example, the change ξ1 of the AGC signal is the difference between the absolute value A1 of the AGC signal within the time period T1-T0 and the absolute value A2 of the AGC signal within the time period T2-T1. The change ξ of the AGC signal can be directly used in actual modeling and calculation.

[0056] In step S3, the present invention fixes the statistical time interval of the AGC signal, sets upward power adjustment as the positive direction and positive value, and downward power adjustment as the opposite direction and negative value. Statistical analysis is performed on the changes ξ of a large number of historical AGC signals received by an AGC unit in southern China during a certain hour; the changes approximately follow a normal distribution, such as... Figure 3 As shown.

[0057] Meanwhile, by comparing the changes in AGC signals ξ over different time periods, the same distribution characteristics can be obtained. Therefore, it can be considered that the changes in AGC signals ξ over different time periods within a day all follow a normal distribution in probability, that is, the changes in AGC signals ξ ~ N(μ, σ 2 ).

[0058] The probability density function of the normal distribution is:

[0059]

[0060] In the formula, x is a random variable, μ is the mean of the normal distribution, and σ is the standard deviation of the normal distribution. 2 The variance is the variance of the normal distribution.

[0061] Influenced by natural resources and users' production and lifestyles, renewable energy and load exhibit varying degrees of uncertainty across different time periods. Since the variation ξ of the AGC signal follows a normal distribution within a certain time period Δt, a time period of Δt = 1 hour is chosen for more precise modeling compared to the peak, valley, and flat time periods. Furthermore, an autoregressive model is selected to model the AGC signal for a typical day, thus fully reflecting the randomness of the signal under time series conditions.

[0062] Specifically, for the 24:00 mark of a typical day, the hourly AGC amplitude data are statistically analyzed to form a normal distribution. The mean of the normal distribution is set to 0, and the variance of the distribution varies in different time periods, as shown below.

[0063] ξ t ~(μ,σ t 2 ) t=1,2,...,24 (1)

[0064] Where, ξ t Let σ be the change in the AGC signal at time t. t 2 Let be the variance of the AGC signal distribution at time t, and μ be the mean of the normal distribution, μ = 0.

[0065] As can be seen from formula (1), the distribution of AGC demand in each time period of a typical day is determined by the variance σ of the AGC signal distribution in different time periods. 2 Determined. For the uncertainty of typical intraday AGC demand, the variance σ of the AGC signal distribution at different time periods is determined. 2 Establish an autoregressive model for the time series data. The autoregressive model is as follows:

[0066]

[0067] Right now

[0068]

[0069] Where p is the total number of terms in the regression model. Let e ​​be the autoregressive coefficient of the i-th regression model. t Let t be the white noise excitation at time t.

[0070] In step S4, the parameter estimates in the autoregressive model are obtained through certain observations and the Yule-Walker equations, abbreviated as the YW equations. The YW equations are as follows:

[0071]

[0072] Where k is the number of equations in the YW equation system, γ k Let γ be the autocorrelation function of the k-th equation. k-i Let δ be the autocorrelation function of the ki-th equation. e Let η be the standard deviation of white noise. k,0 is the cross-correlation function of white noise.

[0073] As can be seen from formula (4), the second part of the right side of the YW equation system is non-zero only when k = 0. Therefore, formula (4) can be solved by the equation system or matrix equation when k > 0, so as to obtain the autoregressive model coefficients.

[0074] The YW equation system of formula (4) can be transformed into matrix form as follows:

[0075]

[0076] Right now

[0077] RΦ=B (6)

[0078] In the formula, R is the autocorrelation matrix, Φ is the autoregression coefficient matrix, and B is the autocorrelation function matrix;

[0079] Since the variances at different time periods are stationary time series, the autocorrelation matrix R is a constant value, which can be obtained from the observed values. The variance σ of the AGC signal distribution can be calculated according to formula (5). 2 Autoregression coefficient

[0080] When k = 0, the YW equation system is:

[0081]

[0082] Based on autoregression coefficients The standard deviation δ of white noise is obtained by using formula (7). e Thus, a typical daily frequency modulation signal model with randomness is obtained.

[0083] The uncertainty of the grid frequency regulation AGC signal is characterized by using a grid frequency regulation AGC signal model that considers the impact of renewable energy and load on AGC demand.

[0084] This invention characterizes the frequency regulation demand of the power grid under high renewable energy penetration by establishing a frequency regulation demand characterization model that considers significant uncertainties. The invention proposes a research method that uses a fixed statistical time interval for AGC signals and focuses solely on the amplitude of the AGC signals. For characterizing typical daily frequency regulation demand, based on the probability distribution characteristics of frequency regulation demand in different hourly segments, a frequency regulation demand characterization method based on an autoregressive model is proposed, and the model parameters are determined using the YW equation.

[0085] In summary, the present invention has the following advantages and beneficial effects:

[0086] 1. The method for characterizing the frequency regulation AGC signal of the power grid in this invention, compared with the traditional empirical method, can fully consider the significant uncertainty of the AGC signal in the power grid with high penetration of renewable energy, making the frequency regulation demand characterization model closer to the current power system reality.

[0087] 2. The method for characterizing the frequency regulation AGC signal of the power grid in this invention fixes the statistical time interval of the AGC signal and only counts the amplitude of the AGC signal. At this time, the amplitude of the AGC signal is allowed to be 0. This transforms the dual randomness of amplitude and time interval into single randomness, reduces the number of random variables, and makes the characterization process more efficient.

[0088] 3. The method for characterizing the power grid frequency modulation AGC signal of the present invention has a finer time granularity, dividing a typical day into 24 points, which can fully consider the characteristics of the frequency modulation signal in different hours and characterize it more accurately.

[0089] The above description is a detailed description of the preferred embodiments of the present invention. However, the embodiments are not intended to limit the scope of the patent application of the present invention. All equivalent changes or modifications made under the technical spirit disclosed in the present invention should fall within the patent scope covered by the present invention.

Claims

1. A method for characterizing power grid frequency modulation AGC signals, characterized in that, Includes the following steps: S1. Set the statistical time interval for the power grid frequency regulation AGC signal and obtain historical data of the amplitude of the power grid frequency regulation AGC signal under a fixed statistical time interval; S2. Preprocess the historical data of the amplitude of the power grid frequency regulation AGC signal, and convert the absolute value of the amplitude of the power grid frequency regulation AGC signal into the change of the AGC signal reflecting the AGC demand; S3. Perform probability distribution statistics on the changes in AGC signals, consider the impact of renewable energy and load on AGC demand, and establish an autoregressive model reflecting the distribution of AGC demand based on the probability distribution statistics of the changes in AGC signals. S4. Determine the parameters of the autoregressive model that reflects the distribution of AGC demand, and finally obtain the grid frequency regulation AGC signal model that considers the impact of renewable energy and load on AGC demand; S5. Using a grid frequency regulation AGC signal model that considers the impact of renewable energy and load on AGC demand, the uncertainty of the grid frequency regulation AGC signal is characterized; In step S1, the amplitude and time interval of the valid AGC signal are both random. By fixing the statistical time interval of the AGC signal, only the amplitude of the AGC signal is counted. At this time, the amplitude of the AGC signal is allowed to be 0, thereby transforming the dual randomness of amplitude and time interval into single randomness. In step S2, the historical data of the power grid frequency regulation AGC signal amplitude is collected, sorted, and bad data is removed in the preprocessing. The change in the AGC signal reflecting the AGC demand is the difference in the absolute value of the AGC signal amplitude between two adjacent time intervals. In step S3, the statistical time interval of the AGC signal is fixed, and the upward adjustment of power is set as the positive direction with a positive value, and the downward adjustment of power is set as the opposite direction with a negative value. Based on the statistical analysis of the probability distribution of the change in the AGC signal, and by comparing the change data of the AGC signal in different time periods, it is determined that the change in the AGC signal in different time periods of the day follows a normal distribution in terms of probability. Renewable energy and load exhibit varying uncertainties at different times. The variation ξ of the AGC signal follows a normal distribution within a certain time period Δt. A time period Δt = 1 hour is selected, and an autoregressive model is chosen to model the AGC signal for a typical day. Specifically: For a typical day at 24:00, the hourly AGC amplitude data forms a normal distribution. The mean of this normal distribution is set to 0. The variance of the AGC signal distribution varies across different time periods, resulting in the following variation in the AGC signal: x t ~(m,s t 2 )t=1,2,…,24 Where, ξ t Let σ be the change in the AGC signal at time t. t 2 Let be the variance of the AGC signal distribution at time t, and μ be the mean of the normal distribution, μ = 0; For the uncertainty of typical intraday AGC demand, an autoregressive model is established for the time series variance of AGC signal distribution at different time periods: Right now Where p is the total number of terms in the regression model. Let e ​​be the autoregressive coefficient of the i-th regression model. t The white noise excitation is at time t; In step S4, the parameters of the autoregressive model are obtained through the observed values ​​and the YW equation system, which is as follows: Where k is the number of equations in the YW equation system, γ k Let γ be the autocorrelation function of the k-th equation. k-i Let δ be the autocorrelation function of the ki-th equation. e Let η be the standard deviation of white noise. k,0 The cross-correlation function of white noise; The YW equation system with k > 0 is transformed into matrix form as follows: Right now RΦ=B In the formula, R is the autocorrelation matrix, Φ is the autoregression coefficient matrix, and B is the autocorrelation function matrix; The autocorrelation matrix R is obtained from the observed values, and the autocorrelation function matrix B is calculated. Based on the autocorrelation matrix R and the autocorrelation function matrix B, the autoregressive coefficients are calculated. The autoregressive coefficient matrix Φ is obtained; When k = 0, the YW equation system is: Based on autoregression coefficients And from the YW equations at k=0, the standard deviation δ of the white noise can be obtained. e The final grid frequency regulation AGC signal model that takes into account the impact of renewable energy and load on AGC demand is obtained.

Images