Multi-dimensional optimization method and system for improving primary frequency modulation capability of thermal power generating unit

Abstract

The invention discloses a multi-dimensional optimization method and system for improving the primary frequency modulation capability of a thermal power generating unit. The multi-dimensional optimization method comprises the steps that an annular data cache region is constructed by collecting power grid frequency signals, and the maximum frequency difference in a time window is dynamically extracted; in combination with cooperative judgment of an AGC instruction direction and a primary frequency modulation action direction, a load reference is intelligently selected, and a frequency difference function related to a maximum frequency difference is superposed to generate a real-time frequency modulation target value; constructing a primary frequency modulation performance comprehensive evaluation index based on the primary frequency modulation response speed and the electric quantity contribution degree; dynamically correcting a primary frequency modulation feedforward coefficient; establishing a valve opening-flow characteristic function model based on historical operation data, and dynamically generating an adaptive compensation amount according to a real-time working condition; and resetting parameters by adopting a gradient attenuation model. According to the method, full-chain optimization from frequency difference sensing to valve execution is realized, and the primary frequency modulation response quality of the thermal power generating unit and the qualification rate of power grid examination are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of thermal power generation, and particularly relates to a multi-dimensional optimization method and system for improving the primary frequency modulation capability of a thermal power unit. BACKGROUND

[0002] Power system frequency stability is a core index for safe operation of a power grid, and its essence is determined by dynamic balance between power generation and power consumption. When power generation is greater than power consumption, the frequency of the power grid rises; otherwise, the frequency decreases. As the first line of defense for power system frequency stability, primary frequency modulation is required to quickly respond to frequency deviation by adjusting the output power of a generator unit within several to tens of seconds when the load of the power grid changes, so as to maintain system stability. Under the background of new-type power system construction, with the continuous increase in the proportion of new energy installations and the continuous decrease in system inertia, higher requirements are put forward for the response speed and adjustment accuracy of primary frequency modulation of a thermal power unit.

[0003] The current primary frequency modulation control strategy commonly used by domestic thermal power units mainly has the following technical bottlenecks:

[0004] (1) The traditional scheme adopts real-time frequency correction. Since the inertia of a thermal power unit is relatively large, the response delay is relatively large, and the real-time frequency deviation signal is used for primary frequency modulation action, which is easy to result in a small actual frequency modulation amount.

[0005] (2) The AGC system is responsible for secondary frequency modulation, and its instruction period is usually 4-8 seconds, while the primary frequency modulation response is required to be completed within seconds. When the AGC instruction and the primary frequency modulation demand are in opposite directions, there is a lack of effective coordination strategy.

[0006] (3) The primary frequency modulation feedforward coefficient of the existing control system is usually fixedly set, and cannot adapt to changes in factors such as unit load, main steam pressure, coal quality and equipment characteristics.

[0007] (4) Although the flow characteristic linearization has been carried out in the current thermal power unit, it is found through field tests that the flow characteristics of most units are still nonlinear, and targeted flow characteristic compensation is needed.

[0008] (5) The existing control strategy only adjusts the theoretical frequency modulation target, and does not introduce network regulation related indicators to modify and compensate for units that fail in continuous examination.

[0009] (6) Although adaptive control is introduced in some units, when the unit equipment is aging, the coal quality changes or maintenance is carried out, the correction parameters are often no longer applicable, and the existing reset scheme mostly uses the way of simply restoring the initial value, which is easy to cause sudden changes in adjustment performance. SUMMARY

[0010] The technical problem to be solved by this invention is the shortcomings of thermal power units under the background of high proportion of new energy grid access, such as insufficient primary frequency regulation response speed, reduced regulation accuracy and low pass rate. It provides a multi-dimensional optimization method and system to improve the primary frequency regulation capability of thermal power units. It adopts a comprehensive solution that integrates frequency difference optimization, command coordination, parameter self-adaptation and valve compensation to meet the higher requirements of the new power system for the frequency regulation capability of thermal power units.

[0011] Therefore, one technical solution adopted by the present invention is: a multi-dimensional optimization method for improving the primary frequency regulation capability of thermal power units, comprising:

[0012] Step 1: Construct a ring-shaped data buffer by collecting power grid frequency signals, and dynamically extract the maximum frequency difference within the time window;

[0013] Step 2: By combining the coordinated judgment of the AGC command direction and the primary frequency regulation action direction, the load reference is intelligently selected, and the maximum frequency difference is superimposed to generate the real-time frequency regulation target value through the frequency difference-load function;

[0014] Step 3: Based on the measured power of the unit following the real-time frequency regulation target value, calculate the primary frequency regulation response speed and the frequency regulation power contribution, and construct a comprehensive evaluation index for primary frequency regulation performance accordingly.

[0015] Step 4: Dynamically adjust the primary frequency modulation feedforward coefficient based on the performance evaluation results of the comprehensive evaluation index of primary frequency modulation performance.

[0016] Step 5: Establish a valve opening-flow characteristic function model based on historical operating data, and dynamically generate adaptive compensation quantities according to real-time operating conditions;

[0017] Step 6: When the comprehensive evaluation index of primary frequency regulation performance exceeds the maximum threshold multiple times in a row, or the main steam pressure fluctuates drastically, or the load oscillates, the parameters are reset using the gradient decay model.

[0018] Furthermore, in step one, the time window T = K × T grid Where K∈[0.4,1] represents the floating coefficient, and T grid The assessment period is indicated; the circular data buffer adopts a circular queue storage structure with a queue length L = T / Δt, where Δt represents the sampling period. If the direction of a frequency modulation action changes, the circular data buffer is cleared and re-acquisition is performed.

[0019] Furthermore, in step two, the direction of the first frequency modulation action is defined as S1 = sign(K p ×Δf), the AGC command direction is defined as S2=sign(dP) AGC / dt), where K pP represents the theoretical frequency regulation coefficient corresponding to the frequency regulation load, where Δf represents the frequency difference within the time window; AGC Indicates unit load command;

[0020] If it is determined that S1 == S2, the unit load command P will be... AGC The real-time frequency regulation target value is generated by superimposing it with the theoretical frequency regulation load; otherwise, the AGC command is blocked, and the current real-time unit load P is set to... real The real-time frequency regulation target value is generated by superimposing it with the theoretical frequency regulation load:

[0021]

[0022] In the formula, |Δf max | represents the maximum frequency difference within the time window, ΔP max This indicates the real-time frequency modulation target value.

[0023] Furthermore, in step two, when |dP AGC When / dt|<δ, it is determined that there is no valid AGC instruction, and the direction judgment logic is skipped. The δ represents the AGC assessment dead zone.

[0024] Furthermore, in step three, the comprehensive evaluation index η of primary frequency regulation performance focuses on the primary frequency regulation response speed and the degree of power contribution, and is determined by the 15-second output response index ΔP. 15 % , 30-second output response index ΔP 30 The three-dimensional weighted model of the comprehensive evaluation index of primary frequency regulation performance, consisting of % and the power contribution index Q%, is expressed as follows:

[0025]

[0026] In the formula, coefficients α, β, and γ are the weighting coefficients of the 15-second output response index, the 30-second output response index, and the power contribution index, respectively, and α + β + γ = 1; A0 is the start time of the frequency modulation action, P0 is the initial power reference value, P(t) is the measured power at time t, and ΔP max T is the target value for real-time frequency modulation. 恢复 For the system frequency recovery time, P s (t) represents the actual output of the unit.

[0027] Furthermore, in step four, the dynamic correction method for the primary frequency regulation feedforward coefficient is as follows: when the primary frequency regulation is activated, a timing template is triggered. When the timing reaches the assessment cycle and the unit is operating normally, if the value of the comprehensive evaluation index η of the primary frequency regulation performance is greater than or equal to the preset value η0, the gradual correction loop of the primary frequency regulation feedforward coefficient is triggered for slow correction; if the value of the comprehensive evaluation index η of the primary frequency regulation performance is lower than the preset value η0, the emergency correction loop of the primary frequency regulation feedforward coefficient is triggered for rapid correction.

[0028] Furthermore, in step four, the dynamic correction formula for the primary frequency modulation feedforward coefficient is as follows:

[0029]

[0030] In the formula, K ff ′ represents the corrected first-order frequency modulation feedforward coefficient, K ff Here, ΔP represents the original primary frequency modulation feedforward coefficient, and ΔP represents the actual frequency modulation power, where ΔP ≤ ΔP. max ΔP max η is the real-time frequency modulation target value, λ is the correction coefficient of the asymptotic correction loop, and η is the value of the target value. s η is the benchmark value for the comprehensive evaluation index of primary frequency modulation performance, μ is the correction coefficient of the emergency correction loop, η is the comprehensive evaluation index of primary frequency modulation performance, and P is the benchmark value for the comprehensive evaluation index of primary frequency modulation performance. real This represents the real-time load of the generating unit.

[0031] Furthermore, in step five, the valve opening-flow characteristic function model is established through a data-driven approach: historical operating data is collected to construct a training dataset, a regression algorithm is used to train the valve opening-flow characteristic function model, and the model parameters are updated periodically to adapt to changes in valve opening-flow characteristics.

[0032] The adaptive compensation amount is as follows:

[0033]

[0034] In the formula, ΔV comp For adaptive compensation, V is the comprehensive valve position command, P is the inlet pressure, T is the main steam temperature, ΔQ is the main steam flow rate change range, and dΔQ / dt is the main steam flow rate change rate.

[0035] Furthermore, in step six, the parameters are reset according to the following gradient decay model:

[0036]

[0037] In the formula, k is the number of reset iterations. For the initial primary frequency modulation feedforward coefficient, For the frequency modulation feedforward coefficient under the k-th reset, S3 is the primary frequency modulation feedforward coefficient before the k-th reset, N is the number of consecutive non-compliances, S4 is the attenuation rate coefficient, and S5 is the attenuation dynamic factor coefficient.

[0038] This invention also provides another technical solution: a multi-dimensional optimization system for improving the primary frequency regulation capability of thermal power units, used to implement the above-mentioned multi-dimensional optimization method, comprising:

[0039] Frequency difference acquisition module: It constructs a ring-shaped data buffer by collecting power grid frequency signals and dynamically extracts the maximum frequency difference within a time window;

[0040] Command decision module: By combining the collaborative judgment of the AGC command direction and the primary frequency regulation action direction, the load reference is intelligently selected, and a frequency difference function with respect to the maximum frequency difference is superimposed to generate a real-time frequency regulation target value;

[0041] Performance evaluation module: Based on the actual power of the unit following the real-time frequency regulation target value, calculate the primary frequency regulation response speed and frequency regulation power contribution, and construct a comprehensive evaluation index of primary frequency regulation performance accordingly;

[0042] Feedforward optimization module: Dynamically corrects the primary frequency modulation feedforward coefficient based on the performance evaluation results of the comprehensive evaluation index of primary frequency modulation performance;

[0043] Valve compensation module: Based on historical operating data, a valve opening-flow characteristic function model is established, and adaptive compensation is dynamically generated according to real-time operating conditions;

[0044] Parameter reset module: When the comprehensive evaluation index of primary frequency regulation performance exceeds the maximum threshold multiple times in a row, or the main steam pressure fluctuates drastically, or the load oscillates, the gradient decay model is used to reset the parameters.

[0045] The beneficial effects of this invention are as follows: This invention realizes full-chain optimization from frequency difference sensing to valve execution, which can solve problems such as large load deviation, fixed parameters, and nonlinear valve adjustment deviation in traditional frequency regulation, and significantly improve the primary frequency regulation response quality of thermal power units and the pass rate of power grid assessment. Attached Figure Description

[0046] Figure 1 This is a flowchart of a multi-dimensional optimization method for the primary frequency regulation capability of a 1000MW double reheat unit according to Embodiment 1 of the present invention;

[0047] Figure 2 This is a schematic diagram of the collaborative decision-making mechanism between AGC commands and primary frequency regulation requirements in Embodiment 1 of the present invention;

[0048] Figure 3 This is a schematic diagram of the calculation process for the comprehensive evaluation index of primary frequency modulation performance in Embodiment 1 of the present invention;

[0049] Figure 4 This is a schematic diagram of the adaptive valve characteristic compensation strategy in Embodiment 1 of the present invention;

[0050] Figure 5 This is a diagram illustrating the collaborative working mechanism of primary frequency modulation feedforward optimization and parameter reset in Embodiment 1 of the present invention.

[0051] Figure 6This is a schematic diagram of the structure of a multi-dimensional optimization system for the primary frequency regulation capability of a 1000MW double reheat unit in Embodiment 2 of the present invention. Detailed Implementation

[0052] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0053] Example 1

[0054] Example 1 takes a 1000MW double reheat unit as an example. The load range of the primary frequency regulation closed-loop control is 300MW to 1000MW.

[0055] As attached Figure 1 As shown, this embodiment is a multi-dimensional optimization method for improving the primary frequency regulation capability of thermal power units, including the following steps:

[0056] Step 1: Construct a ring-shaped data buffer (depth 50 points, corresponding to a 5-second time window) by high-speed sampling of the power grid frequency signal at 50ms. Use a dynamic window algorithm to extract the instantaneous maximum frequency difference |Δf. max |Serves as a frequency modulation reference. When a change in frequency difference direction is detected, the ring data buffer is cleared and data is reacquired to avoid interference from historical data.

[0057] Step two, the direction of the primary frequency modulation action is defined as S1 = sign(Δf), and the direction of the AGC command is defined as S2 = sign(dP). AGC / dt).

[0058] If it is determined that S1 == S2, the unit load command P will be... AGC The real-time frequency regulation target value is generated by superimposing it with the theoretical frequency regulation load; otherwise, the AGC command is blocked, and the current real-time unit load P is set to... real The real-time frequency regulation target value is generated by superimposing it with the theoretical frequency regulation load.

[0059] In Example 1, the generator set speed non-uniformity rate is 5%, and the formula for calculating the real-time frequency regulation target value is as follows:

[0060]

[0061] Where, when |dP AGC When / dt|<2MW| (2MW is the AGC assessment dead zone), it is determined that there is no valid AGC instruction, and the direction judgment logic is skipped. Δf max

[0062] Appendix Figure 2 A flowchart of the collaborative decision-making mechanism between AGC instructions and primary frequency regulation requirements in this embodiment is provided.

[0063] Step 3: The comprehensive evaluation index η of primary frequency regulation performance focuses on the primary frequency regulation response speed and power contribution, and is determined by the 15-second output response index ΔP. 15 % , 30-second output response index ΔP 30 It consists of % and the power contribution index Q%.

[0064] In Example 1, the formula for calculating the primary frequency modulation performance index is as follows:

[0065]

[0066] Appendix Figure 3 A schematic diagram of the calculation process for the primary frequency modulation performance evaluation index in this embodiment is provided.

[0067] In step four, in Example 1, when a primary frequency regulation operation is performed, a timing template is triggered. When the timing reaches the grid assessment cycle and the unit is operating normally, if the value of η is not lower than the preset value of 0.7, the primary frequency regulation feedforward coefficient asymptotic correction loop is triggered for slow correction. If the value of η is lower than the preset value of 0.7, the primary frequency regulation feedforward coefficient emergency correction loop is triggered for rapid correction, and the main steam pressure is released to pull back the correction loop. The primary frequency regulation feedforward coefficient correction formula is as follows:

[0068]

[0069] In step five, the compensation function model is established using a data-driven approach: historical operating data is collected to construct a training dataset, a regression algorithm is used to train the functional relationship model, and the model parameters are periodically updated to adapt to changes in valve characteristics. The compensation function model is as follows:

[0070]

[0071] In Example 1, to facilitate deployment in the DCS, a multinomial regression algorithm is used to construct the compensation model:

[0072]

[0073] In the formula, θ1, θ2, θ3, θ4, θ5, and θ6 are polynomial coefficients.

[0074] Appendix Figure 4 A schematic diagram of the adaptive valve characteristic compensation strategy in this embodiment is provided.

[0075] In step six, when the comprehensive evaluation index of primary frequency regulation performance continuously exceeds the maximum threshold, or the main steam pressure fluctuates drastically, or the load oscillates, the parameters are reset using the following gradient decay model:

[0076]

[0077] AppendixFigure 5 This document records the collaborative working mechanism of frequency modulation feedforward optimization and parameter reset in this embodiment.

[0078] Example 2

[0079] This embodiment provides a multi-dimensional optimization system for improving the primary frequency regulation capability of thermal power units, used to implement the multi-dimensional optimization method described in Embodiment 1, and includes the following functional modules:

[0080] (1) Frequency difference acquisition module: By collecting power grid frequency signals, a ring data buffer is constructed to dynamically extract the maximum frequency difference |Δf within the time window. max |;

[0081] (2) Command decision module: Combines the collaborative judgment of the AGC command direction and the primary frequency regulation action direction, intelligently selects the load reference, and superimposes the frequency difference function about the maximum frequency difference to generate the real-time frequency regulation target value;

[0082] (3) Performance evaluation module: Based on the actual power of the unit following the real-time frequency regulation target value, calculate the primary frequency regulation response speed and frequency regulation power contribution, and construct a comprehensive evaluation index of primary frequency regulation performance accordingly;

[0083] (4) Feedforward optimization module: Dynamically correct the feedforward coefficient of primary frequency modulation based on the performance evaluation results of the comprehensive evaluation index of primary frequency modulation performance;

[0084] (5) Valve compensation module: Based on historical operating data, a valve opening-flow characteristic function model is established, and adaptive compensation is dynamically generated according to real-time operating conditions;

[0085] (6) Parameter reset module: When the comprehensive evaluation index of primary frequency regulation performance exceeds the maximum threshold multiple times, or the main steam pressure fluctuates drastically, or the load oscillates, the gradient attenuation model is used to reset the parameters.

[0086] Appendix Figure 6 A schematic diagram of a multi-dimensional optimization system structure for the primary frequency regulation capability of a 1000MW double reheat unit under Example 2 is provided.

[0087] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A multi-dimensional optimization method for improving the primary frequency regulation capability of thermal power units, characterized in that, include: Step 1: Construct a ring-shaped data buffer by collecting power grid frequency signals, and dynamically extract the maximum frequency difference within the time window; Step 2: By combining the coordinated judgment of the AGC command direction and the primary frequency modulation action direction, the load reference is intelligently selected, and the frequency difference function with respect to the maximum frequency difference is superimposed to generate the real-time frequency modulation target value; Step 3: Based on the measured power of the unit following the real-time frequency regulation target value, calculate the primary frequency regulation response speed and the frequency regulation power contribution, and construct a comprehensive evaluation index for primary frequency regulation performance accordingly. Step 4: Dynamically adjust the primary frequency modulation feedforward coefficient based on the performance evaluation results of the comprehensive evaluation index of primary frequency modulation performance. Step 5: Establish a valve opening-flow characteristic function model based on historical operating data, and dynamically generate adaptive compensation quantities according to real-time operating conditions; Step 6: When the comprehensive evaluation index of primary frequency regulation performance exceeds the maximum threshold multiple times in a row, or the main steam pressure fluctuates drastically, or the load oscillates, the parameters are reset using the gradient decay model.

2. The multi-dimensional optimization method for improving the primary frequency regulation capability of thermal power units according to claim 1, characterized in that, In step one, the time window T = K × T grid Where K∈[0.4,1] represents the floating coefficient, and T grid The assessment period is indicated; the circular data buffer adopts a circular queue storage structure with a queue length L = T / Δt, where Δt represents the sampling period. If the direction of a frequency modulation action changes, the circular data buffer is cleared and re-acquisition is performed.

3. The multi-dimensional optimization method for improving the primary frequency regulation capability of thermal power units according to claim 1, characterized in that, In step two, the direction of the first frequency modulation action is defined as S1 = sign(K p ×Δf), the direction of the AGC command is defined as... Among them, K p P represents the theoretical frequency regulation coefficient corresponding to the frequency regulation load, where Δf represents the frequency difference within the time window; AGC Indicates unit load command; If it is determined that S1 == S2, the unit load command P will be... AGC The real-time frequency regulation target value is generated by superimposing it with the theoretical frequency regulation load; otherwise, the AGC command is blocked, and the current real-time unit load P is set to... real The real-time frequency regulation target value is generated by superimposing it with the theoretical frequency regulation load: In the formula, |Δf max | represents the maximum frequency difference within the time window, ΔP max This indicates the real-time frequency modulation target value.

4. The multi-dimensional optimization method for improving the primary frequency regulation capability of thermal power units according to claim 3, characterized in that, In step two, when |dP AGC When / dt|<δ, it is determined that there is no valid AGC instruction, and the direction judgment logic is skipped. The δ represents the AGC assessment dead zone.

5. The multi-dimensional optimization method for improving the primary frequency regulation capability of thermal power units according to claim 1, characterized in that, In step three, the comprehensive evaluation index η of primary frequency regulation performance focuses on the primary frequency regulation response speed and the degree of power contribution, and is composed of the 15-second output response index ΔP. 15 % , 30-second output response index ΔP 30 The three-dimensional weighted model of the comprehensive evaluation index of primary frequency regulation performance, consisting of % and the power contribution index Q%, is expressed as follows: In the formula, coefficients α, β, and γ are the weighting coefficients of the 15-second output response index, the 30-second output response index, and the power contribution index, respectively, and α + β + γ = 1; A0 is the start time of the frequency modulation action, P0 is the initial power reference value, P(t) is the measured power at time t, and ΔP max T is the target value for real-time frequency modulation. 恢复 For the system frequency recovery time, P s (t) represents the actual output of the unit.

6. The multi-dimensional optimization method for improving the primary frequency regulation capability of thermal power units according to claim 1, characterized in that, In step four, the dynamic correction method for the primary frequency regulation feedforward coefficient is as follows: when the primary frequency regulation is activated, a timing template is triggered. When the timing reaches the assessment cycle and the unit is operating normally, if the value of the comprehensive evaluation index η of the primary frequency regulation performance is greater than or equal to the preset value η0, the gradual correction loop of the primary frequency regulation feedforward coefficient is triggered for slow correction; if the value of the comprehensive evaluation index η of the primary frequency regulation performance is lower than the preset value η0, the emergency correction loop of the primary frequency regulation feedforward coefficient is triggered for rapid correction.

7. A multi-dimensional optimization method for improving the primary frequency regulation capability of thermal power units according to claim 6, characterized in that, In step four, the dynamic correction formula for the primary frequency modulation feedforward coefficient is as follows: In the formula, K ff ′ represents the corrected first-order frequency modulation feedforward coefficient, K ff Here, ΔP represents the original primary frequency modulation feedforward coefficient, and ΔP represents the actual frequency modulation power, where ΔP ≤ ΔP. max ΔP max η is the real-time frequency modulation target value, λ is the correction coefficient of the asymptotic correction loop, and η is the value of the target value. s η is the benchmark value for the comprehensive evaluation index of primary frequency modulation performance, μ is the correction coefficient of the emergency correction loop, η is the comprehensive evaluation index of primary frequency modulation performance, and P is the benchmark value for the comprehensive evaluation index of primary frequency modulation performance. real This represents the real-time load of the generating unit.

8. The multi-dimensional optimization method for improving the primary frequency regulation capability of thermal power units according to claim 1, characterized in that, In step five, the valve opening-flow characteristic function model is established through a data-driven approach: historical operating data is collected to construct a training dataset, a regression algorithm is used to train the valve opening-flow characteristic function model, and the model parameters are updated periodically to adapt to changes in valve opening-flow characteristics. The adaptive compensation amount is as follows: In the formula, ΔV comp For adaptive compensation, V is the comprehensive valve position command, P is the inlet pressure, T is the main steam temperature, ΔQ is the main steam flow rate change range, and dΔQ / dt is the main steam flow rate change rate.

9. A multi-dimensional optimization method for improving the primary frequency regulation capability of thermal power units according to claim 1, characterized in that, In step six, reset the parameters according to the following gradient decay model: In the formula, k is the number of reset iterations. For the initial primary frequency modulation feedforward coefficient, For the frequency modulation feedforward coefficient under the k-th reset, S3 is the primary frequency modulation feedforward coefficient before the k-th reset, N is the number of consecutive non-compliances, S4 is the attenuation rate coefficient, and S5 is the attenuation dynamic factor coefficient.

10. A multi-dimensional optimization system for improving the primary frequency regulation capability of thermal power units, used to implement the multi-dimensional optimization method according to any one of claims 1-9, characterized in that, include: Frequency difference acquisition module: It constructs a ring-shaped data buffer by collecting power grid frequency signals and dynamically extracts the maximum frequency difference within a time window; Command decision module: By combining the collaborative judgment of the AGC command direction and the primary frequency regulation action direction, the load reference is intelligently selected, and a frequency difference function with respect to the maximum frequency difference is superimposed to generate a real-time frequency regulation target value; Performance evaluation module: Based on the actual power of the unit following the real-time frequency regulation target value, calculate the primary frequency regulation response speed and frequency regulation power contribution, and construct a comprehensive evaluation index of primary frequency regulation performance accordingly; Feedforward optimization module: Dynamically corrects the primary frequency modulation feedforward coefficient based on the performance evaluation results of the comprehensive evaluation index of primary frequency modulation performance; Valve compensation module: Based on historical operating data, a valve opening-flow characteristic function model is established, and adaptive compensation is dynamically generated according to real-time operating conditions; Parameter reset module: When the comprehensive evaluation index of primary frequency regulation performance exceeds the maximum threshold multiple times in a row, or the main steam pressure fluctuates drastically, or the load oscillates, the gradient decay model is used to reset the parameters.

Images