Agc frequency modulation and collaborative optimization control method and system for thermal power generating unit

Abstract

The present application relates to the technical field of thermal power unit operation control, and discloses a thermal power unit AGC frequency modulation and collaborative optimization control method and system, acquires NOx concentration data sequences and denitration regulating valve opening data sequences of a flue gas purification system under different loads, establishes a flue gas purification system response lag time function through cross-correlation analysis, reads unit rotor inertia constant and obtains historical operation parameters, establishes an inertia-flue gas emission coupling coefficient function in combination with the lag time function, calculates a virtual power compensation amount according to current operation parameters, generates a linkage control instruction set based on the virtual power compensation amount, and issues the linkage control instruction set to a unit control system and a denitration control system; the present application realizes deep collaboration of AGC frequency modulation and denitration regulation, and takes into account unit frequency stability and flue gas emission standard compliance.

Description

Technical Field

[0001] This invention relates to the field of thermal power unit operation control technology, and more specifically, to a method and system for AGC frequency regulation and collaborative optimization control of thermal power units. Background Technology

[0002] In the field of thermal power unit operation control, Automatic Generation Control (AGC) frequency regulation is a core means of ensuring grid frequency stability. It needs to respond quickly to grid load commands to balance power generation and consumption demand. Meanwhile, flue gas NOx emission control is a key link for thermal power units to meet environmental protection requirements. It needs to adjust the denitrification system in real time according to unit load changes to avoid excessive NOx content in flue gas. With the increasing peak-valley difference of grid load, units often need to respond frequently and quickly to AGC frequency regulation commands within a wide load range. At this time, the coordination between the frequency inertia response of AGC frequency regulation and the adjustment of the denitrification system has become a core technical pain point affecting the balance between unit operational stability and environmental compliance.

[0003] In the prior art, there are already optimization schemes for rapid load change and AGC frequency regulation response of thermal power units. For example, Chinese patent application CN120578040A discloses a multivariable collaborative rapid load change control system and method for thermal power units. After receiving AGC commands through the acquisition module, the load change command preprocessing module calculates the load change rate and predicts the energy gap in the next 30 seconds. The dynamic load feedforward compensation module allocates fuel quantity, feedwater quantity and feedforward commands of the turbine control valve according to the energy gap. Combined with the steam-combustion side decoupling control matrix and parameter self-optimization mechanism established by the multivariable coordinated control module, the stability of the unit's rapid load change process is improved, effectively ensuring the execution speed and accuracy of AGC frequency regulation commands.

[0004] However, in actual operating conditions where multiple units are jointly dispatched and the unit load increases or decreases rapidly, existing technologies have limitations and cannot solve the core problem of "AGC frequency regulation inertial response and denitrification regulation lag coupling": Although existing technologies can optimize the multi-variable coordination of the load change process, their control object only focuses on the energy balance of the unit's power generation side and has not established a coupling correlation model between the unit's rotor inertial constant and the response lag of the flue gas purification system. When the unit suddenly reduces the load, the rotor inertial absorption of kinetic energy can easily cause a temporary drop in frequency. At this time, the denitrification system fails to reduce the absorbent supply in time due to the regulation lag, which not only wastes the reagent, but the lag in regulation action will also further increase the unit's load regulation burden, leading to an expansion of the AGC frequency regulation command execution deviation and an increase in the risk of frequency exceeding the limit. More importantly, existing technologies have not quantified the coupling strength between the response lag of the denitrification system and the inertial response of the unit under different load conditions. They cannot reserve lag compensation in advance when AGC commands are issued, so "frequency stability" and "emission compliance" are always in a state of mutual constraint: if priority is given to ensuring the frequency regulation response speed of AGC, emissions will exceed the standard due to the lag of denitrification; if priority is given to controlling NOx emissions, frequency fluctuations will occur due to the interference of denitrification regulation on the unit, making it difficult to meet the dual requirements of the power grid for frequency regulation quality and environmental protection. Summary of the Invention

[0005] To overcome the aforementioned deficiencies in existing technologies, this invention provides a method and system for AGC frequency regulation and collaborative optimization control of thermal power units. By establishing a response lag time function of the flue gas purification system and an inertial-flue gas emission coupling coefficient function, the virtual power compensation amount is calculated and a linkage control instruction set is generated. This effectively solves the problem of coupling between denitrification regulation lag and unit inertial response, achieving synchronization between AGC frequency regulation and denitrification response. It can reduce the number of times the unit frequency exceeds the limit and the probability of NOx emission exceeding the standard, ensuring grid frequency stability while meeting environmental protection requirements, and improving the operational adaptability and comprehensive benefits of thermal power units in multi-unit joint scheduling.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] AGC frequency regulation and collaborative optimization control methods for thermal power units include:

[0008] Data sequences of NOx concentration in flue gas from the flue gas purification system and the opening sequence of the regulating valve in the denitrification system of thermal power units under different load power were collected; cross-correlation analysis was performed on the NOx concentration data sequence and the regulating valve opening sequence to establish the response lag time function of the flue gas purification system.

[0009] The rotor inertia constant of the unit is read from the unit control system, and the historical operating parameters of the unit are obtained. Based on the rotor inertia constant, the historical operating parameters of the unit, and the response lag time function of the flue gas purification system, an inertia-flue gas emission coupling coefficient function is established. The current operating parameters of the unit are obtained, and the virtual power compensation amount to be injected under the current operating condition is calculated according to the rotor inertia constant, the current operating parameters of the unit, and the inertia-flue gas emission coupling coefficient function.

[0010] Based on the virtual power compensation amount that needs to be injected under the current operating conditions, a set of linkage control instructions is generated; the set of linkage control instructions is then sent to the unit control system and denitrification control system of the thermal power unit.

[0011] The NOx concentration data sequence of the flue gas and the opening degree data sequence of the denitrification system control valve are accompanied by the load power identifier of the corresponding time for each collection moment; the NOx concentration data sequence of the flue gas is formed by sorting the NOx concentration values ​​of the flue gas at different data set collection moments in time, and the opening degree data sequence of the denitrification system control valve is formed by sorting the NOx concentration values ​​of the flue gas at different data set collection moments in time.

[0012] The method for cross-correlation analysis of flue gas NOx concentration data sequences and denitrification system regulating valve opening data sequences includes:

[0013] Using the power value corresponding to the load power identifier as an index, the flue gas NOx concentration data sequence and the denitrification system regulating valve opening data sequence under different load powers are divided into n1 flue gas NOx concentration data subsequences and n1 denitrification system regulating valve opening data subsequences.

[0014] The n1 subsequences of flue gas NOx concentration data and the n1 subsequences of denitrification system control valve opening data are combined into n1 subsequence pairs;

[0015] Perform cross-correlation analysis on each of the n1 subsequence pairs and calculate the cross-correlation coefficient for each subsequence pair.

[0016] The method for dividing the flue gas NOx concentration data sequence and the denitrification system regulating valve opening data sequence under different load power into n1 flue gas NOx concentration data subsequences and n1 denitrification system regulating valve opening data subsequences includes:

[0017] Set a load power division interval δP, and group all flue gas NOx concentration values ​​that fall within the same δP load interval into a flue gas NOx concentration data subsequence. Then, group the corresponding valve opening values ​​into a denitrification system valve opening data subsequence.

[0018] The method for establishing the response lag time function of the flue gas purification system includes:

[0019] Based on the cross-correlation coefficient of each subsequence pair, the response lag time of the flue gas purification system for the corresponding load range of each subsequence pair is obtained;

[0020] The response lag time function of the flue gas purification system is obtained by fitting the response lag time of the n1 subsequences to the typical values ​​of the corresponding load ranges.

[0021] The historical operating parameters of the unit include historical load sequence, historical speed sequence, and historical flue gas flow sequence;

[0022] The method for establishing the inertial-flue gas emission coupling coefficient function includes:

[0023] Substitute the historical load sequence into the response lag time function of the flue gas purification system to calculate and generate the historical lag time sequence;

[0024] Multiple regression analysis was performed on the unit rotor inertia constant, historical load sequence, historical speed sequence, historical flue gas flow sequence, and historical lag time sequence to obtain the inertia-flue gas emission coupling coefficient function.

[0025] The current operating parameters of the unit include the current unit speed, the current unit load power, and the current unit flue gas flow rate;

[0026] The method for calculating the amount of virtual power compensation to be injected under the current operating condition includes:

[0027] Extract the target power change from the AGC frequency regulation command, substitute the current load power in the unit's current operating parameters into the flue gas purification system response lag time function, and obtain the flue gas purification system response lag time under the current operating conditions.

[0028] Substitute the unit rotor inertia constant, unit current speed, unit current load power, unit current flue gas flow rate, and flue gas purification system response lag time under current operating conditions into the inertia-flue gas emission coupling coefficient function to calculate the inertia-flue gas emission coupling coefficient under current operating conditions.

[0029] Based on the inertial-flue gas emission coupling coefficient, the target power change, and the response lag time of the flue gas purification system under the current operating conditions, the amount of virtual power compensation that needs to be injected under the current operating conditions is calculated.

[0030] The method for generating the linkage control instruction set includes:

[0031] The target power change is superimposed with the virtual power compensation to generate a power dispatch command that includes the corrected load adjustment range;

[0032] Based on the response lag time function of the flue gas purification system, a response lag prediction curve of the flue gas purification system is generated; the cumulative lag time corresponding to each load point in the response lag prediction curve of the flue gas purification system is analyzed to determine the pre-adjustment command of the denitrification system.

[0033] The power scheduling command and the denitrification system pre-adjustment command are integrated to form a linkage control command set.

[0034] The method for generating the response hysteresis prediction curve of the flue gas purification system includes:

[0035] Obtain the target load power sequence from the AGC frequency modulation command, substitute the target load power sequence into the flue gas purification system response lag time function, and calculate the flue gas purification system response lag time corresponding to each load point.

[0036] The cumulative lag time of the flue gas purification system at each load point is summed to obtain the cumulative lag time of each load point. The cumulative lag time of each load point is then plotted as the flue gas purification system response lag prediction curve.

[0037] A thermal power unit AGC frequency regulation and collaborative optimization control system, used to implement the above-mentioned thermal power unit AGC frequency regulation and collaborative optimization control method, the system comprising:

[0038] Response lag modeling module: used to collect flue gas NOx concentration data sequences of the flue gas purification system and denitrification system regulating valve opening data sequences under different load power; to perform cross-correlation analysis on the flue gas NOx concentration data sequences and denitrification system regulating valve opening data sequences, and to establish the response lag time function of the flue gas purification system;

[0039] Coupled modeling module: used to read the rotor inertia constant of the unit from the unit control system and obtain the historical operating parameters of the unit. Based on the rotor inertia constant, the historical operating parameters of the unit and the response lag time function of the flue gas purification system, the inertia-flue gas emission coupling coefficient function is established.

[0040] Power compensation calculation module: used to obtain the current operating parameters of the unit, and calculate the amount of virtual power compensation to be injected under the current operating conditions based on the unit rotor inertia constant, the current operating parameters of the unit and the inertia-flue gas emission coupling coefficient function;

[0041] Linkage control module: Based on the virtual power compensation amount that needs to be injected under the current operating conditions, it generates a linkage control instruction set; and issues the linkage control instruction set to the unit control system and denitrification control system of the thermal power unit.

[0042] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0043] This invention collects NOx concentration data sequences from the flue gas purification system and denitrification system regulating valve opening data sequences under different load power of thermal power units, and performs cross-correlation analysis to establish a response lag time function for the flue gas purification system. This allows for precise quantification of the denitrification system's regulation lag characteristics, providing reliable lag parameter support for subsequent compensation. Simultaneously, by reading the unit's rotor inertia constant, obtaining historical and current operating parameters, and combining the aforementioned lag time function to construct an inertia-flue gas emission coupling coefficient function, and calculating the virtual power compensation under the current operating conditions, the coupling interference between the unit's inertial response and the flue gas purification system can be effectively offset, avoiding mutual influence when the two change. Finally, by generating a set of linkage control instructions and synchronously sending them to the unit control system and the denitrification control system, the execution of AGC frequency modulation instructions and denitrification regulation actions are coordinated. This effectively solves the problems of slow denitrification system regulation failing to synchronize with AGC instructions, frequency fluctuations caused by inertia and flue gas purification coupling, and emission exceedances. It significantly improves the operational stability and environmental compliance of thermal power units when responding to dispatch instructions, ensuring that the unit can still meet the requirements of frequency control and emission control under scenarios of rapid load increases or decreases. Attached Figure Description

[0044] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0045] Figure 1 This is a flowchart of a method for AGC frequency regulation and collaborative optimization control of thermal power units provided in an embodiment of the present invention;

[0046] Figure 2 A flowchart illustrating a method for establishing an inertial-flue gas emission coupling coefficient function, provided in an embodiment of the present invention;

[0047] Figure 3 A flowchart illustrating the principle of generating a linkage control instruction set and updating the model, provided in an embodiment of the present invention;

[0048] Figure 4 This is a functional block diagram of an AGC frequency regulation and collaborative optimization control system for thermal power units provided in an embodiment of the present invention. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] Example 1

[0051] Please see Figure 1 As shown, this embodiment provides a method for AGC frequency regulation and collaborative optimization control of thermal power units, including:

[0052] Step S10: Collect flue gas NOx concentration data sequence of the flue gas purification system and denitrification system regulating valve opening data sequence under different load power; perform cross-correlation analysis on the flue gas NOx concentration data sequence and denitrification system regulating valve opening data sequence to establish the response lag time function of the flue gas purification system.

[0053] Step S10 aims to establish the response lag time function of the flue gas purification system through data acquisition and analysis, and generate a response lag prediction curve, so as to provide accurate lag characteristic parameters for subsequent inertial-flue gas emission coupling compensation and linkage control, and solve the technical problems that the adjustment speed of the denitrification system is slower than the execution speed of the AGC frequency modulation command and the lag characteristic differences under different load conditions cannot be quantified.

[0054] Further, step S10 includes:

[0055] Step S11: Collect the flue gas NOx concentration data sequence C(t) of the flue gas purification system and the denitrification system regulating valve opening data sequence V(t) under different load power. The values ​​of the flue gas NOx concentration data sequence and the denitrification system regulating valve opening data sequence at each collection time are accompanied by the load power identifier of the corresponding time.

[0056] Specifically, the NOx concentration data sequence C(t) of the flue gas is collected using the Continuous Flue Gas Monitoring System (CEMS) already configured in the thermal power unit, where t represents the data acquisition time and C(t) is the NOx concentration value at time t. The opening sequence V(t) of the denitrification system's regulating valve is collected using the position sensor integrated into the valve, where V(t) is the valve opening value at time t, and the opening range is typically mapped to a standardized value of 0-100%. That is, the NOx concentration data sequence is formed by sorting the NOx concentration values ​​from different datasets at different acquisition times, and the denitrification system regulating valve opening data sequence is also formed by sorting the NOx concentration values ​​from different datasets at different acquisition times. During the data collection process, the unit load power value P at each acquisition time is correlated with the NOx concentration value and regulating valve opening value at that time through the real-time data interaction interface of the unit control system, forming a dataset containing the load power identifier. Step S11 addresses the deficiency of existing data acquisition processes where flue gas parameters are disconnected from load conditions. It ensures that subsequent analysis can accurately match the response characteristics of the flue gas purification system under specific loads, provides operating condition anchors for response lag analysis under different load conditions, avoids deviations in lag time calculations caused by load changes, and provides a clear classification basis for subsequence segmentation in step S12.

[0057] Step S12: Perform cross-correlation analysis on the flue gas NOx concentration data sequence and the denitrification system regulating valve opening data sequence to establish the flue gas purification system response lag time function;

[0058] Further, step S12 includes:

[0059] Step S121: Using the power value corresponding to the load power identifier as an index, divide the flue gas NOx concentration data sequence and the denitrification system regulating valve opening data sequence under different load powers into n1 flue gas NOx concentration data subsequences and n1 denitrification system regulating valve opening data subsequences.

[0060] Step S122: Calculate the cross-correlation coefficient between the NOx concentration data subsequence of flue gas and the opening data subsequence of the denitrification system regulating valve, and determine the response lag time of the flue gas purification system under the load power represented by the load power identifier;

[0061] Step S123: Fit the functional relationship between the response lag time of the flue gas purification system and the load power, and use it as the response lag time function of the flue gas purification system.

[0062] Specifically, using the power value (P value) corresponding to the load power identifier as an index, the collected flue gas NOx concentration data sequence C(t) containing the load power identifier and the denitrification system regulating valve opening data sequence V(t) are classified and segmented. Specifically, a load power division interval δP is set. The determination of δP depends on the unit's commonly used load range and the sensitivity of the flue gas purification system response to load changes. For example, if the unit's commonly used load range is 200MW-600MW, and the flue gas purification system response shows a significant difference when the load changes by 10MW, then δP can be set to 10MW. All flue gas NOx concentration values ​​with P values ​​falling within the same δP load interval are grouped into a single flue gas NOx concentration data subsequence C. i (t), which assigns the valve opening value at the corresponding time to a denitrification system valve opening data subsequence V. i (t), where i is the subsequence number, ranging from 1 to n1, and n1 is the total number of subsequences after segmentation, determined by the ratio of the unit's normal load range to δP. During the segmentation process, timestamp comparison is required to ensure that the flue gas NOx concentration value and the regulating valve opening value within the same subsequence are completely consistent at the time of acquisition, achieving time alignment. That is, for any acquisition time ts, if C(ts) is assigned to C... i (t), then the corresponding V(ts) must be classified into the V corresponding to the same i. i (t), and the ts of both are exactly the same. This segmentation method solves the problem of difficulty in distinguishing the lag characteristics caused by the mixing of data from different load conditions in traditional overall data sequence analysis. It ensures that each subsequence corresponds to the load condition of a single load range, ensuring that the subsequent lag time calculation can accurately reflect the true response characteristics of the load range. At the same time, time alignment ensures the time correlation between the control valve action and the change in NOx concentration, laying the foundation for the accuracy of subsequent cross-correlation analysis.

[0063] For each subsequence pair (C) i (t), V i (t) Perform cross-correlation analysis to calculate the cross-correlation coefficient for each subsequence pair, and calculate the response lag time τ of the flue gas purification system for the corresponding load range based on the cross-correlation coefficient. i Specifically, the cross-correlation function R is used. xy (τ) for C i (t) and V i (t) is used for calculation, and the cross-correlation function is used to calculate the degree of coordinated change between two signals under different time delays τ. Its mathematical expression is as follows: The integration interval is the acquisition time length of the subsequence, where V i (t-τ) is the subsequence of the opening data of the i-th denitrification system control valve after a delay of τ. It is the start time of the acquisition of the i-th subsequence. It is the end time of the acquisition of the i-th subsequence. - This refers to the "sampling time length of the subsequence". Due to the cross-correlation function R... xy The magnitude of cross-correlation function R(τ) is affected by the signal amplitude and cannot directly represent the degree of correlation. Therefore, it is necessary to normalize the cross-correlation function to obtain the cross-correlation coefficient R(τ). The mathematical expression of the normalization is: ,in, , is a subsequence of flue gas NOx concentration data The zero-delay autocorrelation function value characterizes Its own energy; V is the data subsequence of the denitrification system regulating valve opening. i The zero-delay autocorrelation function value of V(t) characterizes V i (t) The energy of the signal itself. After normalization, the cross-correlation coefficient R(τ) ranges from -1 to 1. The closer R(τ) is to 1, the stronger the positive correlation between the two signals under the time delay τ; the closer R(τ) is to -1, the stronger the negative correlation; and the closer R(τ) is to 0, the less correlation there is. The value range of τ is usually set to 0-30s, covering the common lag time range of most thermal power unit denitrification systems. A fixed time step is used when traversing τ. , The determination of the lag time needs to take into account both the accuracy and efficiency of the calculation: An excessively large value will result in insufficient accuracy in the calculation of lag time, and may miss the peak value of the cross-correlation coefficient corresponding to the true lag time. If the value is too small, it will result in too many traversals, increasing the computation time. The typical value range is 0.1s to 1s, and the specific value is determined based on the response characteristics of the flue gas purification system and the data acquisition cycle. For example, if the acquisition cycle of the flue gas NOx concentration data sequence and the denitrification system regulating valve opening data sequence is 1s, then... It can be set to 1 second, consistent with the acquisition cycle, to ensure that the traversal step size matches the data resolution; if the acquisition cycle is 0.5 seconds and higher accuracy is required for the lag time, then... It can be set to 0.5s. The value of τ ranges from 0 to 30s. Taking 1 second as an example, the number of iterations N = (30-0) / 1+1 = 31 times. Each iteration requires calculating the cross-correlation function value once, resulting in a computational complexity of O(N), which is linear. Using the computing power of existing industrial control systems, the calculation time for a single cross-correlation function is typically in the millisecond range. The total computation time for 31 iterations does not exceed 100ms, far less than the typical response time (seconds) for load regulation of thermal power units, thus meeting real-time requirements. With a time limit of 0.1s and the number of iterations increased to 301, the total computation time can still be controlled within 1 second, thus meeting the real-time requirements of engineering applications. By iterating through the possible range of values ​​for τ, the value of τ that maximizes R(τ) is found; this value of τ corresponds to the load interval (subsequence C). i (t), V i The response lag time τ of the flue gas purification system in the P interval corresponding to (t) i The reason for selecting the τ value corresponding to the maximum cross-correlation coefficient as the lag time is that after the opening of the denitrification system's regulating valve is adjusted, the change in flue gas NOx concentration has a transmission delay. When the time delay τ is exactly equal to this transmission delay, the regulating valve opening signal and the NOx concentration response signal are aligned in time, and their degree of synchronization is the highest. The normalized cross-correlation coefficient reaches its maximum value, therefore, the τ corresponding to this maximum value is the true response lag time of the flue gas purification system. This method addresses the deficiency of existing technologies that use a fixed lag time and ignore load differences. Through cross-correlation analysis, it quantifies the time difference between the regulating valve action and the change in NOx concentration under a specific load, accurately capturing the differences in lag characteristics across different load ranges. The obtained τ... i It can accurately reflect the response speed of the flue gas purification system under the corresponding load, providing reliable sample data for the function fitting in step S123.

[0064] For n1 subsequence pairs (C i (t), V i Cross-correlation analysis was performed on (t) to obtain n1 groups (P) i , τ i ) data, where P i This represents the typical value of the load interval corresponding to the i-th subsequence, usually taken as the midpoint of that interval; for group n1 (P i , τ i The response lag time function τ=f(P) of the flue gas purification system is obtained by fitting the data. The fitting process requires scatter plot analysis (P) first. i , τ i To determine the type of functional relationship (linear or nonlinear) between the two, we first use linear regression to analyze the distribution trend of (P). i , τ i The data were initially fitted, and the coefficient of determination R for linear fitting was calculated. 2 If R 2 If the value is ≥0.95, the scatter points are determined to be approximately linearly distributed. A linear regression method is used to fit the data, and the resulting response lag time function of the flue gas purification system is in the form τ=a1×P+b1, where a1 is the slope coefficient and b1 is the intercept coefficient. If R... 2If the value is less than 0.95, the scatter points are determined to exhibit a non-linear distribution, requiring further fitting using a non-linear regression method. For the non-linear distribution case, a quadratic polynomial form τ = a² × P is used. 2 The equations +b2×P+c2 and the exponential form τ=a3×exp(b3×P)+c3 are fitted together, where, Let be an exponential function with base e of the natural logarithm, where a2 is the coefficient of the quadratic term, b2 is the coefficient of the linear term, a3 is the magnitude coefficient, representing the magnitude of the exponential change, b3 is the exponential variation coefficient, representing the rate of change (decline or increase) of the lag time with load power, and c3 is the asymptote coefficient. The coefficient of determination R for calculating the two nonlinear fittings is also shown. 2 Choose R 2 The fitted form with a larger value is used as the final response lag time function of the flue gas purification system.

[0065] The following two typical unit implementation examples illustrate the specific application of the above fitting method:

[0066] Example 1 (Linear Distribution): A 660MW supercritical coal-fired unit uses a honeycomb SCR denitrification reactor with 3 catalyst layers, and the ammonia injection grid employs a vortex mixer design. The unit's load regulation range is from 200MW to 660MW. Subsequences are divided into intervals δP with a load power of 50MW, resulting in n1=10 groups (P...). i , τ i The sample data is shown in Table 1:

[0067] Table 1. Ten groups of linear distribution cases (P) i , τ i Sample data

[0068] Serial number i <![CDATA[Midpoint P of the load range i (MW)]]> <![CDATA[Response lag time τ i (s)]]> 1 215 17.2 2 265 15.9 3 315 14.3 4 365 12.8 5 415 11.5 6 465 10.1 7 515 8.6 8 565 7.4 9 615 6.1 10 650 5.2

[0069] Linear regression was performed on the above data, and the coefficient of determination R was calculated. 2 =0.9973, R 2 ≥0.95 indicates that the scatter points exhibit an approximately linear distribution. A linear regression method was used to fit the response lag time function of the flue gas purification system, yielding τ=-0.0276×P+23.12, where a1=-0.0276s / MW and b1=23.12s. This function shows that the response lag time of the unit's flue gas purification system decreases linearly with increasing load power. For every 100MW increase in load, the lag time decreases by approximately 2.76 seconds. This is because the honeycomb catalyst used in this unit has low flow resistance, and the effect of increased flue gas velocity on the lag time at high loads exhibits a uniform variation characteristic.

[0070] Example 2 (Nonlinear Distribution): A 330MW subcritical coal-fired unit uses a plate-type SCR denitrification reactor with two catalyst layers, and the ammonia injection grid adopts a direct pipe injection design. The unit's load regulation range is from 120MW to 330MW. Subsequences are divided into intervals δP with a load power of 30MW, resulting in a total of n1=8 groups (P... i , τ i The sample data is shown in Table 2:

[0071] Table 2 Groups of Nonlinear Distributions (P) i , τ i Sample data

[0072] Serial number i <![CDATA[Midpoint P of the load range i (MW)]]> <![CDATA[Response lag time τ i (s)]]> 1 130 24.6 2 160 18.3 3 190 13.8 4 220 10.9 5 250 8.7 6 280 7.3 7 310 6.4 8 330 5.9

[0073] Linear regression was performed on the above data, and the coefficient of determination R was calculated. 2 =0.8938, R 2 The value is less than 0.95, indicating a non-linear distribution of the scatter points, requiring further fitting using a non-linear regression method. Fitting is performed using both quadratic polynomial and exponential forms: The quadratic polynomial form is τ = a² × P. 2 By fitting the equation +b²×P+c², we obtain a² = 0.000409 s / MW. 2 b² = -0.2811 s / MW, c² = 54.12 s, coefficient of determination R 2 =0.9939; Using the exponential form τ=a3×exp(b3×P)+c3 for fitting, we calculated a3=116s, b3=-0.0133 / MW, c3=4.5s, and the coefficient of determination R0 is 0.9939; 2 =0.9982. Comparing the coefficients of determination for the two nonlinear fits, the exponential form R0... 2 =0.9982 is greater than R in the form of a quadratic polynomial. 2 =0.9939, therefore, the exponential form is chosen as the final response lag time function of the flue gas purification system: τ=116×exp(-0.0133×P)+4.5. This function shows that the response lag time of the unit's flue gas purification system decreases exponentially with increasing load power. The lag time changes drastically in the low load range, and the lag time tends to approach the asymptotic value of 4.85 seconds in the high load range. This is because the plate catalyst used in this unit has significant non-uniformity in flue gas distribution at low load, making the denitrification reaction kinetic response speed more sensitive to load changes.

[0074] In practical applications of thermal power units, the response characteristics of flue gas purification systems are influenced by a combination of factors, including combustion conditions, flue gas velocity, and denitrification reaction kinetics. Different units and operating conditions (P) i , τ iThe data distribution patterns differ: for units with good combustion stability and relatively uniform load changes affecting flue gas characteristics, the lag time and load power typically exhibit an approximately linear relationship; for units where load changes significantly impact flue gas velocity and temperature field, the lag time and load power typically exhibit a non-linear relationship, with longer lag times in low-load ranges and shorter lag times in high-load ranges, conforming to exponential decay or quadratic curve characteristics. Therefore, this scheme adopts an adaptive fitting strategy of "first linear fitting judgment—then non-linear fitting selection," which can automatically select the optimal fitting form based on the distribution characteristics of the actual collected data, ensuring that the response lag time function of the flue gas purification system is applicable to different types of units. The calculation of the fitting coefficients in the fitting equation needs to be completed through mathematical software or the built-in algorithm of the control system, with the minimum sum of squared residuals as the objective function, ensuring that the deviation between the fitting function and the sample data is minimized. Step S123 solves the problem of not being able to continuously characterize the lag time under different loads. The response lag time function τ=f(P) of the flue gas purification system can calculate the corresponding response lag time of the flue gas purification system in real time according to any given load power, without having to repeat cross-correlation analysis for each load point, which significantly improves the efficiency of obtaining the lag time. At the same time, it provides a continuous lag characteristic model for the generation of the response lag prediction curve in step S13 and the calculation of the virtual power compensation in step S23, realizing the leap from discrete samples to continuous functions of lag time.

[0075] Step S13: Generate the response lag prediction curve of the flue gas purification system based on the response lag time function of the flue gas purification system.

[0076] Further, step S13 includes:

[0077] Step S131: Obtain the target load power sequence in the AGC frequency modulation command, substitute the target load power sequence into the flue gas purification system response lag time function, and calculate the flue gas purification system response lag time corresponding to each load point.

[0078] Step S132: The cumulative lag time of the flue gas purification system corresponding to each load point is summed to obtain the cumulative lag time of each load point, and the cumulative lag time of each load point is plotted as the flue gas purification system response lag prediction curve.

[0079] Specifically, the target load power sequence P in the AGC (Automatic Generation Control) frequency regulation command is obtained through the command interface of the AGC system. target (t), the target load power sequence contains the variation of the load power that the unit needs to achieve over a period of time, that is, each time t' corresponds to a target load P. target (t'). P targetSubstituting (t') into τ=f(P) one by one, the response lag time τ corresponding to each target load is calculated. target (t')=f(P target (t')), thereby obtaining P target (t) Synchronous response lag time series. This process solves the problem that the lag time corresponding to future load changes cannot be predicted during AGC frequency regulation. Its beneficial effect is that it obtains the lag time distribution of different stages in the entire frequency regulation process in advance, providing dynamic lag parameters for subsequent cumulative summation, ensuring that the prediction curve can match the load change trend of AGC.

[0080] The cumulative lag time T at each load point is obtained by summing the response lag time series. acc (t'). The physical meaning of the cumulative lag time is: from the start time t0 of AGC frequency regulation (corresponding to the initial load P0=P target (t0) to any time t' (corresponding to target load P1=P) target (t')), the weighted sum of the response lag times of the flue gas purification system corresponding to each load point. This is due to the target load power sequence P in the AGC frequency modulation command. target (t) is typically a discrete load point sequence, therefore the cumulative lag time is calculated using a discrete summation method. The specific calculation method is as follows: Assume the target load power sequence contains m discrete time points {t1, t2, ..., t...}. m The corresponding target load is {P}. target (t1),P target (t2),...,P target (t m Substituting each target load into the response lag time function τ=f(P) of the flue gas purification system, we obtain the response lag time sequence {τ}. target (t1),τ target (t2),...,τ target (t m If )}, then the cumulative lag time T at time i' is acc (t i' The formula for calculating ) is: Where k is the summation index variable, taking values ​​from 1 to i', τ target (t k Let be the response lag time of the flue gas purification system at time k. This formula shows that the cumulative lag time is equal to the direct sum of the lag times corresponding to all load points from time 1 to time i'.

[0081] Taking the 330MW unit in Example 2 above as an example, the specific calculation process is explained as follows: Assume that the target load power sequence in the AGC frequency regulation command is P target(t) = {150MW, 200MW, 250MW, 300MW}, corresponding to four discrete time points {t1, t2, t3, t4}. Substituting each target load into the response lag time function τ = 116 × exp(-0.0133 × P) + 4.5 of the unit's flue gas purification system, the calculated response lag time series is shown in Table 3.

[0082] Table 3. Calculation results of response lag time series and cumulative lag time.

[0083] To sum the index variable k <![CDATA[Target load P target (t k )(MW)]]> <![CDATA[Response lag time τ target (t k )(s)]]> <![CDATA[Cumulative lag time T acc (t k )(s)]]> 1 150 <![CDATA[τ target (t1)=20.28]]> <![CDATA[T acc (t1)=20.28]]> 2 200 <![CDATA[τ target (t2)=12.62]]> <![CDATA[T acc (t2)=20.28+12.62=32.90]]> 3 250 <![CDATA[τ target (t3)=8.68]]> <![CDATA[T acc (t3)=32.90+8.68=41.58 4 300 <![CDATA[τ target (t4)=6.59]]> <![CDATA[T acc (t4)=41.58+6.59=48.17]]>

[0084] In Table 3, T acc (t1)=τ target (t1) = 20.28, T acc (t2)=τ target (t1)+τ target (t2) = 20.28 + 12.62 = 32.90 s, and so on. Cumulative lag time T acc (t i' The value increases monotonically with the time index i', reflecting the cumulative effect of hysteresis during load changes. For each time t... i' The corresponding T acc (t i' ) and P at that moment target (t i' Associate them with P target (t i' ) represents the horizontal axis, T acc (t i' Plotting a curve on the vertical axis yields the response lag prediction curve of the flue gas purification system. This method addresses the shortcomings of existing technologies that only consider the lag time of a single load point and ignore the cumulative effect of load changes. Its advantage lies in the fact that the prediction curve can intuitively reflect the total lag time that the denitrification system needs to compensate during AGC frequency regulation, providing key time parameters for generating the denitrification system pre-adjustment command in step S313: by querying the cumulative lag time corresponding to a target load in the prediction curve, it can be determined that the denitrification system needs to send the adjustment command in advance of the cumulative lag time, ensuring that the NOx concentration adjustment of the denitrification system is completed synchronously with the power adjustment of the unit. At the same time, the cumulative summation takes into account the dynamic process of load changes, avoiding the problem of insufficient or excessive pre-adjustment time caused by continuous load changes.

[0085] Step S10, through a technical path of "data acquisition - subsequence segmentation - cross-correlation analysis - function fitting - cumulative summation," solves the core problems of difficulty in quantifying the response lag time of the denitrification system, the inability to dynamically adjust with load, and the inability to predict the lag time during AGC frequency regulation, providing accurate lag characteristic parameters for subsequent steps. The response lag time function τ=f(P) of the flue gas purification system can be adapted to different units of the same type. By simply collecting load-lag time sample data of the unit and refitting, a specific lag time function can be quickly established, significantly improving the versatility of the method. The generation of the response lag prediction curve of the flue gas purification system shifts the pre-regulation of the denitrification system from "empirical judgment" to "data-driven." Combined with virtual power compensation in subsequent steps, deep synergy between frequency regulation and denitrification response can be achieved, avoiding the vicious cycle of emission exceedances and frequency fluctuations caused by lag. If step S10 is missing, the inertial-flue gas emission coupling coefficient function in the subsequent step S20 will lack the key lag time parameter, making it impossible to accurately quantify the coupling strength; the linkage control command set in step S30 will lose the basis for calculating the denitrification pre-adjustment time, resulting in the AGC command and denitrification adjustment remaining in an independent state, making it impossible to achieve the dual synchronization target.

[0086] Step S20: Read the unit rotor inertia constant from the unit control system and obtain the unit's historical operating parameters. Based on the unit rotor inertia constant, the unit's historical operating parameters, and the response lag time function of the flue gas purification system, establish the inertia-flue gas emission coupling coefficient function. Obtain the unit's current operating parameters and calculate the virtual power compensation amount to be injected under the current operating condition based on the unit rotor inertia constant, the unit's current operating parameters, and the inertia-flue gas emission coupling coefficient function.

[0087] Step S20 aims to acquire the core characteristic parameters of the unit, historical operating data, and the established response lag time function of the flue gas purification system, construct the inertia-flue gas emission coupling coefficient function, and then calculate the virtual power compensation amount under the current operating condition. This solves the strong coupling problem between the unit rotor inertial response and the flue gas purification system regulation, achieves decoupling control between the two, and provides a quantitative compensation basis for the generation of subsequent linkage control commands.

[0088] Further, step S20 includes:

[0089] Step S21: Obtain the unit rotor inertia constant and the unit's current operating parameters; the unit's current operating parameters include the unit's current speed, the unit's current load power, and the unit's current flue gas flow rate;

[0090] The rotor inertia constant H is obtained through the parameter reading interface of the unit control system. This parameter is an inherent characteristic parameter of the unit, calculated and determined by the manufacturer during the unit design phase based on the rotor material, structural dimensions, and moment of inertia. It is measured in seconds and reflects the rotor's ability to store kinetic energy. A larger H value means more kinetic energy is stored at the same angular velocity, resulting in more energy being released or absorbed during load changes, and consequently increasing the disturbance intensity to the flue gas system. Simultaneously, the current operating parameters of the unit, including the current unit speed n, are obtained through the unit's real-time measurement system. curren Current load power P of the unit current Current flue gas flow rate G of the unit current The current unit speed reflects the unit's real-time energy balance, the current load power characterizes the unit's current power generation output level, and the current flue gas flow rate reflects the real-time rate at which the combustion system delivers flue gas to the denitrification system. In existing technologies, traditional control methods often ignore the individual differences in the unit's rotor inertia constant H, using a uniform empirical value for coupling compensation calculations, leading to a mismatch between compensation accuracy and the unit's actual characteristics. Simultaneously, some methods rely solely on load parameters to determine operating conditions, ignoring the influence of speed and flue gas flow rate on the coupling relationship, resulting in quantitative deviations in coupling strength. Step S21 accurately obtains the unit's unique H value and three-dimensional current operating parameters including speed, load, and flue gas flow rate, providing fundamental data that aligns with the unit's actual characteristics for the subsequent establishment of the coupling coefficient function. This ensures that the quantitative analysis of coupling strength reflects the true coupling state under the current operating conditions and provides parameter dimension references for historical data modeling in step S22, maintaining parameter consistency between the historical model and the current operating condition analysis.

[0091] Step S22: Obtain the historical operating parameters of the unit, and establish the inertia-flue gas emission coupling coefficient function based on the unit rotor inertia constant, the historical operating parameters of the unit, and the response lag time function of the flue gas purification system.

[0092] Please see Figure 2 As shown, step S22 further includes:

[0093] Step S221: Obtain the unit's historical operating parameters, which include historical load sequences, historical speed sequences, and historical flue gas flow sequences;

[0094] Step S222: Substitute the historical load sequence into the response lag time function of the flue gas purification system to calculate and generate the historical lag time sequence;

[0095] Step S223: Perform multiple regression analysis on the unit rotor inertia constant, historical load sequence, historical speed sequence, historical flue gas flow sequence and historical lag time sequence to fit and obtain the inertia-flue gas emission coupling coefficient function.

[0096] Specifically, historical operating parameters of the generating unit are obtained through the query interface of the unit's historical database. These parameters need to cover common load variation conditions of the unit, and the time span needs to capture the coupling patterns under different seasons and fuel qualities. For example, the time span can be set to 3 months to ensure that the data volume can support the needs of statistical modeling. The historical operating parameter sequence specifically includes the historical load sequence P. hist Historical rotational speed sequence n hist Historical flue gas flow sequence G hist P hist Record the unit load power values ​​at each historical moment, n hist Record the unit speed values ​​at various historical moments, G hist Record the flue gas flow rate values ​​at each historical moment. During the acquisition process, time alignment of the three sequences needs to be achieved through timestamp comparison to avoid misjudgment of coupling relationships due to time misalignment. Existing technologies typically only use historical load data for modeling, ignoring the crucial roles of rotational speed and flue gas flow rate in the coupling transmission, resulting in the model failing to fully reflect the physical chain of "inertial response - combustion change - flue gas transmission". Step S221 solves the problems of incomplete historical data dimensions and time misalignment by acquiring multi-dimensional historical parameters and achieving time alignment, laying the foundation for subsequent extraction of complete coupling features. At the same time, the historical operating parameters of the unit are consistent with the current operating parameters in step S21 in terms of parameter type, ensuring that the historical model can be directly adapted to the current operating conditions for calculation.

[0097] Historical load sequence P hist The unit load power values ​​at each historical moment are successively substituted into the response lag time function τ=f(P) of the flue gas purification system to calculate the corresponding lag time at each historical moment, thus forming a historical lag time series. The core logic of this process is that the response lag characteristics of the denitrification system under historical operating conditions follow the same load dependence law as the current operating conditions. By using τ=f(P), historical load and historical lag time can be correlated, avoiding the complex operation of collecting historical lag time separately, such as tracing back historical NOx concentration and regulating valve opening data and recalculating the cross-correlation. In existing technologies, traditional methods often assume that the historical lag time is a fixed value and ignore the influence of load on lag characteristics, resulting in the lag time in historical coupling analysis being out of sync with the actual operating conditions. Step S222 generates historical lag time that matches the historical load through τ=f(P), ensuring that the dynamic characteristics of the lag time in the historical coupling relationship analysis are consistent with reality. At the same time, it establishes a collaborative correlation between historical data and the response lag time function of the flue gas purification system in step S12, so that the model system of the entire scheme forms a closed loop and avoids parameter conflicts caused by the independent models of each link.

[0098] For the unit rotor inertia constant H and historical load sequence P hist Historical rotational speed sequence n histHistorical flue gas flow sequence G hist Historical lag time series τ hist Multiple regression analysis was performed to fit the inertia-flue gas emission coupling coefficient function Kc=f(H,P,n,G,τ), where P is the unit load power, n is the unit speed, and G is the flue gas flow rate. The implementation is divided into three stages: the first stage is data preprocessing, initially processing P... hist n hist G hist τ hist Normalization was performed to eliminate the influence of parameter dimension differences on the regression results; the normalization formula used min-max standardization. Secondly, load change events were identified, and a load change rate threshold δP was set. rate ,δP rate The determination is based on the maximum load regulation capacity designed for the unit. For example, if the maximum load regulation rate of the unit is 10% of the rated load per minute, then δP rate It can be set to 3% of rated load / minute), when P hist The rate of change of load between adjacent time points is greater than δP rateWhen a load change event is identified, such as a load ramp-up or step event, only the historical operating parameters of the unit corresponding to the load change event are retained for modeling, excluding data without coupled disturbances under steady-state conditions, thus improving the model's relevance. The second stage is feature extraction, which extracts inertial response features and flue gas emission features from the historical operating parameters of the unit corresponding to the load change event: the inertial response feature uses the frequency change rate δf / δt, where f is the unit frequency, calculated from the rotational speed n, f=n / 60, δf is the frequency deviation between adjacent moments, and δt is the time interval, δf / δt reflects the speed of the inertial response; the flue gas emission feature uses the NOx concentration change rate δC / δt, where δC is the NOx concentration deviation between adjacent moments, obtained from historical flue gas monitoring data, δC / δt reflects the dynamic response of the flue gas purification system. The third stage is regression modeling. Based on the feature extraction results, the type of functional relationship is determined: if the scatter points of the inertial response characteristics and flue gas emission characteristics show an approximately linear distribution, then a multiple linear regression method is used; if the scatter points show a nonlinear distribution, such as a quadratic curve or an exponential curve, then a nonlinear regression method, such as the least squares method, is used. The regression coefficients are calculated using mathematical software or the built-in algorithm of the control system, with the minimum sum of squared residuals as the objective function, ensuring that the deviation between the fitted function and historical data is minimized. In existing technologies, traditional methods lack mathematical models to quantify coupling strength, relying on manual experience to set compensation amounts, resulting in low compensation accuracy and poor adaptability. Step S223 establishes an inertial-flue gas emission coupling coefficient function through multiple regression analysis, transforming the abstract coupling relationship into a calculable mathematical expression, solving the problem of the inability to quantify coupling strength. At the same time, through data preprocessing and feature extraction, it ensures that the model only focuses on operating conditions with coupling disturbances, improving model accuracy and generalization ability. This function also matches the current parameter dimension in step S21, providing a direct basis for the real-time compensation amount calculation in step S23.

[0099] Step S23: Calculate the amount of virtual power compensation to be injected under the current operating condition based on the unit rotor inertia constant, the current operating parameters of the unit, and the inertia-flue gas emission coupling coefficient function.

[0100] Further, step S23 includes:

[0101] Step S231: Extract the target power change from the AGC frequency regulation command, substitute the current load power in the unit's current operating parameters into the flue gas purification system response lag time function, and obtain the flue gas purification system response lag time under the current operating condition.

[0102] Step S232: Substitute the unit rotor inertia constant, the unit current speed, the unit current load power, the unit current flue gas flow rate, and the response lag time of the flue gas purification system under the current operating condition into the inertia-flue gas emission coupling coefficient function to calculate the inertia-flue gas emission coupling coefficient under the current operating condition.

[0103] Step S233: Based on the inertial-flue gas emission coupling coefficient, the target power change, and the response lag time of the flue gas purification system under the current operating conditions, calculate the amount of virtual power compensation that needs to be injected under the current operating conditions.

[0104] Specifically, the target power change δP in the AGC frequency modulation command is extracted through the AGC system's command parsing interface. agc δP agc The AGC command requires the unit to reduce its load from the current P. current Adjust to the difference from the target load; simultaneously adjust the unit's current load power P. current Substituting the response lag time function τ=f(P) of the flue gas purification system, the response lag time τ of the flue gas purification system under the current operating condition is calculated. current In existing technologies, traditional compensation methods often use a fixed target power change or ignore the dynamic changes in the response lag time of the current flue gas purification system, resulting in a mismatch between the virtual power compensation amount and the current frequency regulation requirements and lag characteristics; step S231 extracts δP in real time. agc With calculation τ current To ensure that the compensation amount can accurately match the current AGC frequency modulation amplitude and the lag state of the denitrification system, τ current The calculation continues the lag model from step S12, maintaining parameter consistency.

[0105] The rotor inertia constant H and the current rotational speed n of the unit are used. current Current load power P of the unit current Current flue gas flow rate G of the unit current and the calculated τ current Substituting all the values ​​into the inertial-flue gas emission coupling coefficient function Kc=f(H,P,n,G,τ), the inertial-flue gas emission coupling coefficient Kc under the current operating condition is calculated. current Kc current The coupling strength between the inertial response and the flue gas purification system under the current operating conditions was quantified. H reflects the inherent kinetic energy characteristics of the unit, and n current P reflects the real-time inertial response state. current With G current τ reflects the combustion-flue gas transfer state. current Reflecting the denitrification lag state, multiple parameters are input together to ensure Kc current It can comprehensively reflect the current coupling scenario. In existing technologies, traditional methods often use fixed coupling coefficient values, which cannot adapt to changes in coupling strength under different operating conditions; step S232 calculates Kc in real time. current This solves the problem of fixed coupling coefficients, enabling the quantification of coupling strength to dynamically match the current working conditions, thus ensuring the accuracy of subsequent compensation calculations.

[0106] Virtual power compensation δP under current operating conditions comp =Kc current ×δP agc ×τ current The logical basis for the formula design is: Kc current The strength of the coupling interference is determined by δP; the stronger the coupling, the greater the amount of compensation required. agc The magnitude of the power variation that determines AGC frequency modulation is higher; the larger the magnitude, the greater the risk of frequency fluctuations caused by coupling interference, requiring a larger compensation amount. current The duration of the coupling interference is determined by the hysteresis time. The longer the hysteresis time, the longer the interference effect lasts, and the hysteresis effect needs to be offset by extending the compensation amount. (The parameters and δP are then discussed.) comp The correlation conforms to physical laws: when Kc current As it increases, δP comp The value increases proportionally to ensure that stronger coupling interference can be countered; when δP agc As it increases, δP comp Synchronously increase to adapt to greater frequency modulation requirements; when τ current As it increases, δP comp This increases accordingly to compensate for the adjustment delay caused by the longer lag time. In existing technologies, traditional compensation formulas often only consider a single parameter (such as only δP). agc (Related), it cannot fully cover the combined effects of coupling strength, frequency modulation amplitude, and lag time, resulting in insufficient or excessive compensation; step S233 solves the problem of incomplete consideration of factors in the compensation amount through the design of a formula multiplied by three parameters, ensuring δP comp This formula can simultaneously balance coupling strength, frequency regulation requirements, and hysteresis characteristics, providing a precise compensation basis for correcting the load adjustment amplitude in step S31. Furthermore, this formula is compatible with the coupling coefficient function in step S22 and δP in step S231. agc and τ current This collaborative approach deeply integrates the compensation calculation with the parameters of the preceding model, avoiding deviations caused by independent calculations.

[0107] Step S20, through the technical path of "parameter acquisition - historical modeling - real-time calculation," solves the core problems of unquantifiable coupling strength between unit inertia and flue gas purification system, inadequate compensation amount adapting to real-time operating conditions, and disconnect between historical data and the current model, providing accurate virtual power compensation for step S30. The inertia-flue gas emission coupling coefficient function is universal; for different units of the same type, only the H value needs to be replaced with the unit's historical operating parameter sequence for refitting, quickly establishing a dedicated coupling model without redesigning the modeling logic, significantly improving the method's adaptability. The calculation formula for virtual power compensation uses parameter multiplication; by adjusting the weights of each parameter, such as adding a correction coefficient to the formula, it can adapt to the frequency regulation requirements of different power grids. For example, for power grids with higher frequency stability requirements, Kc can be increased. current The weighting coefficients further enhance the compensation effect. If step S20 is missing, the load adjustment range correction in step S31 will lack a quantitative compensation basis, resulting in the inability to offset the coupling interference of inertia and flue gas emissions, and the problems of frequency fluctuation and emission exceeding standards will still not be solved; at the same time, the response lag time function of the flue gas purification system established in step S12 will not be able to be associated with the inertial characteristics of the unit, the coordination of the entire scheme will be broken, and the closed loop of "lag prediction-coupling compensation-linkage control" cannot be realized.

[0108] Step S30: Based on the virtual power compensation amount that needs to be injected under the current operating conditions, generate a linkage control instruction set; and issue the linkage control instruction set to the unit control system and denitrification control system of the thermal power unit.

[0109] Step S30 aims to generate a set of linkage control instructions covering unit power regulation and denitrification system pre-regulation based on the virtual power compensation amount calculated in step S20. By issuing these instructions synchronously, it solves the problem of lack of coordination between the traditional AGC system and the flue gas purification system, which are independent controls. At the same time, it relies on a self-learning mechanism to update the core model and achieve continuous optimization of the dual objectives of frequency stability and emission compliance.

[0110] Please see Figure 3 As shown, step S30 further includes:

[0111] Step S31: Generate a set of linkage control instructions based on the virtual power compensation amount that needs to be injected under the current operating conditions;

[0112] Further, step S31 includes:

[0113] Step S311: The target power change extracted from the AGC frequency regulation command is superimposed with the virtual power compensation to generate a power dispatch command containing the corrected load adjustment magnitude.

[0114] Step S312: Analyze the cumulative lag time corresponding to each load point in the response lag prediction curve of the flue gas purification system to determine the pre-adjustment command of the denitrification system.

[0115] Step S313: Integrate the power scheduling command and the denitrification system pre-adjustment command to form a linkage control command set.

[0116] Specifically, the corrected load adjustment range δP total =δP agc +δP comp The core logic of the formula design lies in: δP agc This only reflects the power demand for grid frequency regulation and does not consider the coupling interference between unit inertia and the flue gas purification system. When AGC commands require the unit to adjust the load, the coupling effect will cause a deviation between the actual power output and the command. For example, changes in flue gas resistance caused by the lag of the denitrification system will indirectly affect the unit's power generation efficiency, while δP comp The amount of compensation required to offset this bias has been quantified, so adding the two together makes δP total It simultaneously meets both frequency modulation and coupling compensation requirements. Specifically, δP agc When δP is positive (load increase), if the coupling effect causes the actual power to be lower than the commanded value, δP comp If the summation is positive, then δP total Greater than δP agc This ensures that the actual power output covers the coupling losses; δP agc When δP is negative (load reduction), if the coupling effect causes the actual power to be higher than the commanded value, δP comp If the value is negative, the sum of δP total Less than δP agc This avoids power overshoot. In existing technologies, traditional power commands directly use δP. agc This leads to frequency fluctuations caused by coupling interference; step S312 achieves dynamic correction of power commands through formula superposition, so that the power commands executed by the unit control system are adapted to both grid frequency regulation and internal system coupling compensation, ensuring that frequency stability is not affected by coupling.

[0117] Step S313 determines the pre-adjustment command of the denitrification system by analyzing the cumulative lag time of each load point in the response lag prediction curve of the flue gas purification system. First, the target load power sequence P is identified. target The critical load point in (t) is selected based on the load change rate: when P target The load change rate at adjacent times in (t) exceeds the load change rate threshold δP rate At that time, the load point corresponding to that adjacent time is the critical load point. For each critical load point P... j Query the response lag prediction curve of the flue gas purification system to obtain the corresponding cumulative lag time τ. j , τ jThe time required for the denitrification system to reach the target NOx concentration response after receiving the adjustment command is defined as the time required. The command needs to be sent in advance to compensate for this lag, and j is the index variable of the critical load point in the target load power sequence. Therefore, the sending time of the denitrification system pre-adjustment command is set to the time when the unit control system executes the corresponding critical load point adjustment command minus τ. j In existing technologies, traditional denitrification regulation commands and unit load regulation commands are sent synchronously, resulting in NOx concentration response lagging behind load changes, which can easily lead to excessive emissions. Step S313 uses the cumulative lag time of the flue gas purification system response lag prediction curve to guide the advance sending of commands, solving the lag synchronization problem. This ensures that the NOx concentration regulation of the denitrification system and the unit power regulation are perfectly matched in time. Simultaneously, the selection of key load points ensures that pre-regulation only targets load change phases with significant lag effects, avoiding unnecessary regulation actions and reducing the energy consumption of the denitrification system. The power scheduling commands for the unit control system and the pre-regulation commands for the denitrification system are encapsulated into a linkage control command set. This linkage control command set uses a standardized communication protocol. The choice of communication protocol is based on the compatibility and real-time requirements of the industrial control system; for example, the Modbus-TCP protocol is selected. This protocol is widely used in the industrial field and can ensure that both the unit control system and the denitrification control system can accurately parse the commands, and the data transmission delay meets the requirements of collaborative control. Step S313 solves the problem of coordination failure caused by inconsistent command formats in traditional independent control, ensuring that the two control systems can receive and execute commands synchronously, and providing command transmission guarantee for the realization of dual synchronization objectives.

[0118] Step S32: Send a set of linkage control commands to the unit control system and denitrification control system of the thermal power unit, monitor the unit frequency deviation and flue gas NOx concentration deviation in real time, and update the response lag time function and inertia-flue gas emission coupling coefficient function of the flue gas purification system using the recursive least squares method based on the unit frequency deviation and flue gas NOx concentration deviation.

[0119] The calculation method for unit frequency deviation includes: determining the unit frequency reference value f. base and the unit's real-time frequency f real f base The frequency is determined based on the rated frequency of the power grid. For example, the rated frequency of the power grid is usually 50Hz, and in some areas it is 60Hz. base It must be consistent with the rated frequency of the power grid; the real-time frequency f of the unit real The speed signal is obtained through conversion from the unit's control system, and the conversion relationship is f. real =n current / 60, where n current The current unit speed; the formula for calculating the unit frequency deviation δf' is δf'=f real -f baseThe calculation method for NOx concentration deviation in flue gas includes: determining the NOx concentration baseline value C. base and real-time NOx concentration C real C base The determination is based on air pollutant emission standards and the emission control targets of the generating units. For example, a certain region stipulates that the NOx emission limit for thermal power units is a certain value of milligrams per cubic meter. base It can be set to a certain percentage of this limit (such as 90%) to reserve a certain control margin; NOx real-time concentration C real Data is acquired through the Continuous Flue Gas Monitoring System (CEMS) already installed in the thermal power unit. CEMS monitoring data must meet relevant metrological standards to ensure data accuracy. The formula for calculating the NOx concentration deviation δC' is δC' = C real -C base In the process of issuing joint control command sets to the unit control system and the denitrification control system, a timestamp calibration mechanism is adopted to ensure the synchronization of command issuance. A unified command execution timestamp is embedded in the joint control command set. After receiving the command, both the unit control system and the denitrification control system execute the command based on this timestamp, avoiding execution time differences caused by data transmission delays. The timestamp is generated based on the unified clock of the unit control system, which needs to be synchronized with the standard time via GPS or NTP protocol to ensure the accuracy of the timestamp. This method solves the problem of execution asynchrony caused by different transmission paths in traditional command issuance, and further enhances the coordination between unit power regulation and denitrification system pre-regulation.

[0120] In the step of updating the response lag time function and the inertia-flue gas emission coupling coefficient function of the flue gas purification system using the recursive least squares method, the triggering condition for model update is first determined, and the frequency deviation threshold δf is set. th NOx concentration deviation threshold δC th ,δf th The determination is based on the maximum allowable frequency deviation of the power grid. For example, if the power grid stipulates that the absolute value of the frequency deviation must not exceed 0.2Hz, then δf th It can be set to 0.1Hz, meaning an update is triggered when the absolute value of δf' exceeds 0.1Hz; δC th The determination is based on the allowable fluctuation range of the NOx concentration benchmark value, for example, C base 50mg / m 3 The allowable fluctuation range is ±5mg / m 3 Then δC th It can be set to 5mg / m 3 That is, when the absolute value of δC' exceeds 5 mg / m³ 3 An update is triggered when the absolute value of the real-time monitored δf' exceeds δf. th Or the absolute value of δC' exceeds δC thIf the model update process is triggered, then the model update process is activated; otherwise, there is no need to perform the model update operation.

[0121] The application process of recursive least squares method is as follows: For updating the response lag time function τ=f(P) of the flue gas purification system, the latest monitored unit current load power P is used. new Denitrification system regulating valve opening data subsequence V new (t), NOx concentration data subsequence C new (t) Substitute into the recursive least squares method to calculate the new cross-correlation coefficient R. new (τ), searching for R new (τ) The new lag time τ for reaching the maximum value new , will (P new , τ new ) as the new sample point, compared with the original sample point (P) i , τ i ( ) jointly fit the new flue gas purification system response lag time function τ=f new (P), the fitting process uses the minimum sum of squared residuals as the objective function to ensure that the deviation between the new function and the new and old sample data is minimized. For updating the inertial-flue gas emission coupling coefficient function Kc=f(H,P,n,G,τ), the latest monitored unit rotor inertia constant H (an inherent parameter of the unit, which remains unchanged if no equipment replacement has occurred) and the unit's current load power P are used. new Current rotational speed n new Current flue gas flow rate G new The newly calculated lag time τ new Substituting into the recursive least squares method, we can find (H,P) new ,n new G new ,τ new As new sample points, they are used together with the original sample points to fit a new inertial-flue gas emission coupling coefficient function Kc=f. new The fitting process (H, P, n, G, τ) also uses minimizing the sum of squared residuals as the objective function. The recursive least squares method is chosen because it can integrate new monitoring data in real time without recalculating all historical sample data, significantly reducing computational complexity and meeting the requirements for online real-time updates. Simultaneously, by assigning certain weights to new sample data (usually new samples have higher weights than old samples), this method allows the model to adapt to the latest operating conditions more quickly. This method solves the problem of traditional models being fixed and unable to adapt to dynamic changes in operating conditions (such as changes in fuel quality, aging of denitrification system equipment, and adjustments in unit load ranges), ensuring that the lag time function and coupling coefficient function can continuously and accurately reflect the system characteristics under the current operating conditions, providing reliable model support for the generation of subsequent linkage control command sets.

[0122] In traditional control systems, the AGC system focuses only on frequency regulation, and the denitrification system focuses only on emission control. The lack of a coordination mechanism between the two leads to frequency fluctuations caused by lag in denitrification regulation when the unit load changes. This, in turn, exacerbates emission exceedances, creating a vicious cycle. Step S30 generates a set of linked control instructions, associating power regulation with denitrification pre-regulation. Synchronous issuance and real-time monitoring ensure coordinated execution, completely breaking down the barriers of independent control. In traditional control systems, frequency and emission monitoring are often offline or low-frequency, failing to detect deviations promptly. By the time deviations are detected, they have often widened, leading to prolonged frequency exceedances or accumulated emission exceedances. Step S30 sets a monitoring cycle adapted to the operating conditions, capturing deviations in real time, triggering model updates and instruction adjustments, and preventing deviations from escalating.

[0123] The linkage control command set generated in step S31 compensates for coupling interference by correcting power commands and compensates for denitrification lag by pre-adjustment commands, ensuring that unit power regulation and NOx concentration regulation are synchronized in time. Real-time monitoring and model updates in step S32 further dynamically optimize command parameters, ensuring that the dual-synchronization state can be maintained even when operating conditions change, thereby reducing the number of frequency exceedances and the probability of emission exceedances. This directly achieves the core technical objective. The model update trigger condition is set based on the actual state of frequency and emission deviation, avoiding invalid updates and ensuring that the model maintains high accuracy throughout long-term operation. This solves the problem of traditional models becoming less accurate with use and extends the effective lifespan of the control scheme. The selection of key load points allows the denitrification system to pre-adjust only when the load change rate is large and the lag effect is significant, avoiding the energy waste of the denitrification system caused by traditional continuous adjustment. Model self-updating reduces the need for manual intervention, lowers the workload and skill requirements of maintenance personnel, and indirectly reduces maintenance costs. The frequency of model updates and the trend of deviation changes can indirectly reflect the operating status of the equipment. For example, when the response lag time function τ=f(P) of the flue gas purification system is updated frequently and τ continues to increase, it indicates that there may be faults in the denitrification system, such as stuck regulating valves or reduced output of spray pumps, resulting in a slower denitrification response speed. When the inertial-flue gas emission coupling coefficient function Kc=f(H,P,n,G,τ) is updated frequently and Kc fluctuates drastically, it indicates that there may be abnormalities in the unit rotor or combustion system (such as rotor imbalance or unstable fuel supply). This equipment status judgment based on model updates exceeds the expected goal of simply achieving dual synchronization, providing data support for preventive maintenance of the unit and reducing the risk of downtime and maintenance costs caused by equipment failures.

[0124] Example 2

[0125] This embodiment, based on Embodiment 1, provides an AGC frequency regulation and collaborative optimization control system for thermal power units, such as... Figure 4 As shown, it includes:

[0126] Response lag modeling module: used to collect flue gas NOx concentration data sequences of the flue gas purification system and denitrification system regulating valve opening data sequences under different load power; to perform cross-correlation analysis on the flue gas NOx concentration data sequences and denitrification system regulating valve opening data sequences, and to establish the response lag time function of the flue gas purification system;

[0127] Coupled modeling module: used to read the rotor inertia constant of the unit from the unit control system and obtain the historical operating parameters of the unit. Based on the rotor inertia constant, the historical operating parameters of the unit and the response lag time function of the flue gas purification system, the inertia-flue gas emission coupling coefficient function is established.

[0128] Power compensation calculation module: used to obtain the current operating parameters of the unit, and calculate the amount of virtual power compensation to be injected under the current operating conditions based on the unit rotor inertia constant, the current operating parameters of the unit and the inertia-flue gas emission coupling coefficient function;

[0129] Linkage control module: Based on the virtual power compensation amount that needs to be injected under the current operating conditions, it generates a linkage control instruction set; and issues the linkage control instruction set to the unit control system and denitrification control system of the thermal power unit.

[0130] Furthermore, in the response lag modeling module, the methods for establishing the response lag time function of the flue gas purification system include:

[0131] Step S121: Using the power value corresponding to the load power identifier as an index, divide the flue gas NOx concentration data sequence and the denitrification system regulating valve opening data sequence under different load powers into n1 flue gas NOx concentration data subsequences and n1 denitrification system regulating valve opening data subsequences.

[0132] Step S122: Calculate the cross-correlation coefficient between the NOx concentration data subsequence of flue gas and the opening data subsequence of the denitrification system regulating valve, and determine the response lag time of the flue gas purification system under the load power represented by the load power identifier;

[0133] Step S123: Fit the functional relationship between the response lag time of the flue gas purification system and the load power, and use it as the response lag time function of the flue gas purification system.

[0134] Furthermore, in the coupled modeling module, the method for establishing the inertia-flue gas emission coupling coefficient function includes: step S221, obtaining the unit's historical operating parameters, which include historical load sequences, historical speed sequences, and historical flue gas flow sequences;

[0135] Step S222: Substitute the historical load sequence into the response lag time function of the flue gas purification system to calculate and generate the historical lag time sequence;

[0136] Step S223: Perform multiple regression analysis on the unit rotor inertia constant, historical load sequence, historical speed sequence, historical flue gas flow sequence and historical lag time sequence to fit and obtain the inertia-flue gas emission coupling coefficient function.

[0137] Furthermore, in the power compensation calculation module, the method for calculating the amount of virtual power compensation to be injected under the current operating condition includes:

[0138] Step S231: Extract the target power change from the AGC frequency regulation command, substitute the current load power in the unit's current operating parameters into the flue gas purification system response lag time function, and obtain the flue gas purification system response lag time under the current operating condition.

[0139] Step S232: Substitute the unit rotor inertia constant, the unit current speed, the unit current load power, the unit current flue gas flow rate, and the response lag time of the flue gas purification system under the current operating condition into the inertia-flue gas emission coupling coefficient function to calculate the inertia-flue gas emission coupling coefficient under the current operating condition.

[0140] Step S233: Based on the inertial-flue gas emission coupling coefficient, the target power change, and the response lag time of the flue gas purification system under the current operating conditions, calculate the amount of virtual power compensation that needs to be injected under the current operating conditions.

[0141] The methods and systems of this application may be implemented in many ways. For example, they may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above-described order of steps for the method is for illustrative purposes only, and the steps of the method of this application are not limited to the order specifically described above, unless otherwise specifically stated.

[0142] In addition, the parts of the technical solutions provided in the embodiments of this application that are consistent with the implementation principles of the corresponding technical solutions in the prior art have not been described in detail, so as to avoid excessive elaboration.

[0143] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the invention. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for frequency regulation and collaborative optimization control of AGC in thermal power units, characterized in that, The method includes: Data sequences of NOx concentration in flue gas from the flue gas purification system and the opening sequence of the regulating valve in the denitrification system of thermal power units under different load power were collected; cross-correlation analysis was performed on the NOx concentration data sequence and the regulating valve opening sequence to establish the response lag time function of the flue gas purification system. The rotor inertia constant of the unit is read from the unit control system, and the historical operating parameters of the unit are obtained. Based on the rotor inertia constant, the historical operating parameters of the unit, and the response lag time function of the flue gas purification system, an inertia-flue gas emission coupling coefficient function is established. The current operating parameters of the unit are obtained, and the virtual power compensation amount to be injected under the current operating condition is calculated according to the rotor inertia constant, the current operating parameters of the unit, and the inertia-flue gas emission coupling coefficient function. Based on the virtual power compensation amount that needs to be injected under the current operating conditions, a set of linkage control instructions is generated; the set of linkage control instructions is then sent to the unit control system and denitrification control system of the thermal power unit. The NOx concentration data sequence of the flue gas and the opening degree data sequence of the denitrification system control valve are accompanied by the load power identifier of the corresponding time for each collection moment; the NOx concentration data sequence of the flue gas is formed by sorting the NOx concentration values ​​of the flue gas at different data set collection moments in time, and the opening degree data sequence of the denitrification system control valve is formed by sorting the NOx concentration values ​​of the flue gas at different data set collection moments in time. The method for cross-correlation analysis of flue gas NOx concentration data sequences and denitrification system regulating valve opening data sequences includes: using the power value corresponding to the load power identifier as an index, dividing the flue gas NOx concentration data sequences and denitrification system regulating valve opening data sequences under different load powers into n1 flue gas NOx concentration data subsequences and n1 denitrification system regulating valve opening data subsequences; forming n1 subsequence pairs from the n1 flue gas NOx concentration data subsequences and n1 denitrification system regulating valve opening data subsequences; performing cross-correlation analysis on each of the n1 subsequence pairs and calculating the cross-correlation coefficient of each subsequence pair; The method for dividing the flue gas NOx concentration data sequence and the denitrification system regulating valve opening data sequence under different load power into n1 flue gas NOx concentration data subsequences and n1 denitrification system regulating valve opening data subsequences includes: setting a load power division interval δP, classifying all flue gas NOx concentration values ​​that fall within the same δP load interval into one flue gas NOx concentration data subsequence, and classifying the regulating valve opening value at the corresponding time into one denitrification system regulating valve opening data subsequence; The method for establishing the response lag time function of the flue gas purification system includes: obtaining the response lag time of the flue gas purification system for each subsequence pair in the corresponding load interval based on the cross-correlation coefficient of each subsequence pair; fitting the response lag time of the flue gas purification system for the corresponding load interval of n1 subsequence pairs with the typical value of the corresponding load interval to obtain the response lag time function of the flue gas purification system. The historical operating parameters of the unit include historical load sequences, historical speed sequences, and historical flue gas flow sequences; the method for establishing the inertial-flue gas emission coupling coefficient function includes: substituting the historical load sequence into the response lag time function of the flue gas purification system to calculate and generate the historical lag time sequence; performing multiple regression analysis on the unit rotor inertial constant, historical load sequence, historical speed sequence, historical flue gas flow sequence, and historical lag time sequence to fit and obtain the inertial-flue gas emission coupling coefficient function.

2. The AGC frequency regulation and collaborative optimization control method for thermal power units according to claim 1, characterized in that, The current operating parameters of the unit include the current unit speed, the current unit load power, and the current unit flue gas flow rate; The method for calculating the amount of virtual power compensation to be injected under the current operating condition includes: Extract the target power change from the AGC frequency regulation command, substitute the current load power in the unit's current operating parameters into the flue gas purification system response lag time function, and obtain the flue gas purification system response lag time under the current operating conditions. Substitute the unit rotor inertia constant, unit current speed, unit current load power, unit current flue gas flow rate, and flue gas purification system response lag time under current operating conditions into the inertia-flue gas emission coupling coefficient function to calculate the inertia-flue gas emission coupling coefficient under current operating conditions. Based on the inertial-flue gas emission coupling coefficient, the target power change, and the response lag time of the flue gas purification system under the current operating conditions, the amount of virtual power compensation that needs to be injected under the current operating conditions is calculated.

3. The AGC frequency regulation and collaborative optimization control method for thermal power units according to claim 2, characterized in that, The method for generating the linkage control instruction set includes: The target power change is superimposed with the virtual power compensation to generate a power dispatch command that includes the corrected load adjustment range; Based on the response lag time function of the flue gas purification system, a response lag prediction curve of the flue gas purification system is generated; the cumulative lag time corresponding to each load point in the response lag prediction curve of the flue gas purification system is analyzed to determine the pre-adjustment command of the denitrification system. The power scheduling command and the denitrification system pre-adjustment command are integrated to form a linkage control command set.

4. The AGC frequency regulation and collaborative optimization control method for thermal power units according to claim 3, characterized in that, The method for generating the response hysteresis prediction curve of the flue gas purification system includes: Obtain the target load power sequence from the AGC frequency modulation command, substitute the target load power sequence into the flue gas purification system response lag time function, and calculate the flue gas purification system response lag time corresponding to each load point. The cumulative lag time of the flue gas purification system at each load point is summed to obtain the cumulative lag time of each load point. The cumulative lag time of each load point is then plotted as the flue gas purification system response lag prediction curve.

5. A thermal power unit AGC frequency regulation and collaborative optimization control system, used to implement the thermal power unit AGC frequency regulation and collaborative optimization control method according to any one of claims 1-4, characterized in that, The system includes: Response lag modeling module: used to collect flue gas NOx concentration data sequences of the flue gas purification system and denitrification system regulating valve opening data sequences under different load power; to perform cross-correlation analysis on the flue gas NOx concentration data sequences and denitrification system regulating valve opening data sequences, and to establish the response lag time function of the flue gas purification system; Coupled modeling module: used to read the rotor inertia constant of the unit from the unit control system and obtain the historical operating parameters of the unit. Based on the rotor inertia constant, the historical operating parameters of the unit and the response lag time function of the flue gas purification system, the inertia-flue gas emission coupling coefficient function is established. Power compensation calculation module: used to obtain the current operating parameters of the unit, and calculate the amount of virtual power compensation to be injected under the current operating conditions based on the unit rotor inertia constant, the current operating parameters of the unit and the inertia-flue gas emission coupling coefficient function; Linkage control module: Based on the virtual power compensation amount that needs to be injected under the current operating conditions, it generates a linkage control instruction set; and issues the linkage control instruction set to the unit control system and denitrification control system of the thermal power unit.

Images