AGC-based dynamic power regulation control method and system for thermal power generating unit

Abstract

The present application relates to the technical field of thermal power plant thermal automation control, in particular to a kind of dynamic power regulation control method and system of thermal power unit based on AGC.The present application obtains relative power error based on the difference between the automatic power generation control target power of current sampling time and the actual active power of generator, in combination with the rated active power of generator;Obtain the main steam pressure trend based on the difference between the main steam pressure of current sampling time and the adjacent sampling time;Determine the relative water level risk degree based on the relative relationship between the safety lower limit water level of deaerator and the actual water level of deaerator at current sampling time.Fuse the relative power error of current sampling time and historical sampling time, main steam pressure trend and turbine governing valve opening degree, in combination with the relative water level risk degree of current sampling time, determine the throttling exit condition index.Based on the time sequence evolution characteristics of throttling exit condition index, determine the delivery measures of the regulation control instruction of current sampling time.

Description

Technical Field

[0001] This invention relates to the field of thermal power plant automation control technology, specifically to a dynamic power regulation control method and system for thermal power units based on AGC. Background Technology

[0002] In the conventional operating environment where cogeneration thermal power units participate in deep peak shaving under the automatic generation control (AGC) of the grid, the units need to respond quickly to load increase commands issued by the grid. Since there is a delay of several minutes between adding fuel combustion on the boiler side and generating additional steam, the field generally adopts the throttling technology of closing the condensate regulating valve to reduce the flow of cold water, converting the extracted steam originally used to heat the cold water into active power of the turbine, thereby achieving a rapid increase in generator output.

[0003] Existing control technologies typically rely on a fixed static threshold for main steam pressure, attempting to end the throttling operation when the pressure rises back to the threshold. However, in the initial stage of AGC load increase response, to quickly coordinate the extraction of active power, the turbine control valves open significantly, rapidly consuming the existing steam in the pipeline network. This causes the main steam pressure to appear to show a significant short-term drop or stagnation. This physical appearance completely masks the true trend of the sluggish steam production accumulating on the boiler side, leading to a distortion in the timing of the throttling operation's determination by existing technologies. Summary of the Invention

[0004] To address the technical problem of distorted timing in determining the end of throttling operations in existing technologies for cogeneration thermal power units participating in deep peak shaving using Automatic Generation Control (AGC), this invention aims to provide a dynamic power regulation and control method and system for thermal power units based on AGC. The specific technical solution adopted is as follows: This invention proposes a dynamic power regulation control method for thermal power units based on AGC, the method comprising: The system acquires the generator's actual active power, automatic power generation control target power, main steam pressure before the turbine, turbine control valve opening, and deaerator water level at the current sampling time. Based on the difference between the automatic power generation control target power and the actual active power of the generator, and combined with the rated active power of the generator, the relative power error is obtained; based on the difference between the main steam pressure at the generator front at the current sampling time and the historical adjacent sampling time, the main steam pressure trend at the current sampling time is obtained; based on the relative relationship between the safe lower limit water level of the deaerator and the actual water level of the deaerator, the relative water level risk is determined. By integrating the relative power error between the current sampling time and the historical sampling time, the main steam pressure trend, and the turbine control valve opening, and combining the relative water level risk at the current sampling time, the throttling exit condition indicators at the current sampling time are determined. Based on the dynamic evolution characteristics of the throttling exit condition indicators at the current sampling time and the historical sampling time, the measures for issuing regulation and control commands are determined.

[0005] Furthermore, the process of obtaining the relative power error based on the difference between the automatic power generation control target power and the actual active power of the generator, combined with the generator's rated active power, includes: The difference between the automatic power generation control target power and the actual active power of the generator is taken as the current power deviation. The relative power error is obtained based on the relative relationship between the current power deviation and the generator's rated active power.

[0006] Furthermore, the step of obtaining the main steam pressure trend at the current sampling time based on the difference between the main steam pressure at the current sampling time and the historical adjacent previous sampling time includes: Based on the difference between the main steam pressure at the current sampling time and the main steam pressure at the previous adjacent sampling time in history, and combined with the time interval between the current sampling time and the previous adjacent sampling time in history, the trend of the main steam pressure at the current sampling time is obtained.

[0007] Furthermore, determining the relative water level risk based on the relative relationship between the lower safety limit water level of the deaerator and the actual water level of the deaerator includes: The ratio of the lower safety limit water level of the deaerator to the actual water level of the deaerator is used as the relative water level risk.

[0008] Furthermore, the fusion of the relative power error between the current sampling time and historical sampling times, the main steam pressure trend, and the turbine control valve opening, combined with the relative water level risk at the current sampling time, determines the throttling exit condition indicators for the current sampling time, including: Normalization is performed on the relative power error between the current sampling time and the historical sampling time within the preset backtracking time window to obtain the normalized error characteristics between the current sampling time and the historical sampling time. Normalization processing is performed on the main steam pressure trends at the current sampling time and the historical sampling time within the preset backtracking time window to obtain the normalized pressure characteristics at the current sampling time and the historical sampling time. Normalization processing is performed on the turbine valve opening at the current sampling time and the historical sampling time within the preset backtracking time window to obtain the normalized opening characteristics at the current sampling time and the historical sampling time. Based on the normalized error characteristics, normalized pressure characteristics, and normalized opening characteristics of the current sampling time and the historical sampling time within the preset backtracking time window, a multi-dimensional feature matrix is ​​constructed. Based on the state transition law of the multidimensional feature matrix, determine the characteristic value of the steam change trend at the current sampling time; By integrating the relative power error, main steam pressure trend, relative water level risk, and steam change trend characteristic value at the current sampling time, the throttling exit condition index at the current sampling time is obtained.

[0009] Furthermore, the construction of a multi-dimensional feature matrix based on the normalized error features, normalized pressure features, and normalized opening features of the current sampling time and historical sampling times within a preset backtracking time window includes: The normalized error features of the current sampling time and the historical sampling time are arranged in chronological order to construct an error row vector; The normalized pressure features of the current sampling time and the historical sampling time are arranged in chronological order to construct a pressure row vector; Construct an opening row vector by arranging the normalized opening features of the current sampling time and the historical sampling time in chronological order; A multidimensional feature matrix is ​​constructed based on the error row vector, pressure row vector, and opening row vector.

[0010] Further, determining the steam change trend feature value at the current sampling moment based on the state transition law of the multidimensional feature matrix includes: The multidimensional feature matrix is ​​processed using a dynamic mode decomposition algorithm to obtain continuous-time modal eigenvalues; The maximum value among the continuous time modal eigenvalues ​​is taken as the steam change trend eigenvalue at the current sampling time.

[0011] Furthermore, the determination of the measures for issuing regulation control commands based on the dynamic evolution characteristics of the throttling exit condition indicators at the current sampling time and historical sampling times includes: The difference between the throttling exit condition index at the current sampling time and the throttling exit condition index at the previous adjacent historical sampling time is used as the gradient of the throttling exit condition index at the current sampling time. If the gradients of the throttling exit condition indexes of the preset number of historical sampling times adjacent to the current sampling time are all less than or equal to 0, and the gradient of the throttling exit condition indexes of the current sampling time is greater than 0, then an adjustment control command is issued; otherwise, no adjustment control command is issued.

[0012] Furthermore, the adjustment control command includes: Map the current steam pressure trend at the sampling moment to the velocity increment of the condensate regulating valve; The speed increment is added to the preset reference safety speed to obtain the opening speed of the condensate regulating valve; The opening speed is used as an adjustment control command.

[0013] This invention also proposes a dynamic power regulation and control system for thermal power units based on AGC, the system comprising: The data acquisition module is used to obtain the actual active power of the generator, the target power of automatic power generation control, the main steam pressure in front of the turbine, the turbine control valve opening, and the actual water level of the deaerator at the current sampling time. The basic feature analysis module is used to obtain the relative power error based on the difference between the automatic power generation control target power and the actual active power of the generator, combined with the rated active power of the generator; to obtain the main steam pressure trend at the current sampling time based on the difference between the main steam pressure at the current sampling time and the previous adjacent sampling time; and to determine the relative water level risk based on the relative relationship between the lower safety limit water level of the deaerator and the actual water level of the deaerator. The throttling exit condition index analysis module is used to integrate the relative power error, main steam pressure trend and turbine control valve opening at the current sampling time and the historical sampling time, and combine them with the relative water level risk at the current sampling time to determine the throttling exit condition index at the current sampling time. The results analysis module is used to determine the measures for issuing regulation and control commands based on the dynamic evolution characteristics of the throttling exit condition indicators at the current sampling time and historical sampling times.

[0014] The present invention has the following beneficial effects: This invention, based on the difference between the automatic power generation control target power and the actual active power of the generator at the current sampling time, combined with the generator's rated active power, obtains the relative power error, transforming the fluctuating power gap into a dimensionless relative proportion under a unified benchmark, thus eliminating the magnitude differences in unit operating conditions. Based on the difference between the main steam pressure at the turbine inlet at the current sampling time and the historical adjacent sampling time, the trend of the main steam pressure at the current sampling time is obtained, reflecting the physical state of transient imbalance between steam production and consumption within the unit. Before a significant rebound in the absolute pressure reading, the inflection point of delayed steam inrush into the boiler is detected in advance, preventing a delay in ending condensate throttling operations. Based on the relative relationship between the deaerator's safe lower limit water level and the actual water level of the deaerator at the current sampling time, the relative water level risk is determined, transforming the absolute liquid level height of the deaerator into a dimensionless risk indicator. This provides a forced blocking intervention mechanism when the water level approaches the equipment's physical red line, preventing excessive throttling from causing feedwater pump burnout and shutdown accidents.

[0015] By integrating the relative power error between the current sampling time and historical sampling times, the main steam pressure trend, and the turbine control valve opening, combined with the relative water level risk at the current sampling time, the throttling exit condition indicators for the current sampling time are determined. By analyzing the joint evolution trajectory of the unit's multi-dimensional characteristic variables over time, the short-term interference caused by turbine control valve actions on pressure readings is removed at the pure data level. This extracts the implicit characteristics representing the true steam accumulation trend in the boiler, superimposing the unit's physical water level safety boundary onto these implicit characteristics, and objectively quantifying the comprehensive pressure exerted by the unit to continue maintaining throttling operations.

[0016] Based on the dynamic evolution characteristics of the throttling exit condition indicators at the current and historical sampling times, the measures for issuing regulation and control commands are determined. This enables accurate capture of the end time of condensate throttling operations and precise issuance of regulation and control commands, ensuring that the benefits of active power enhancement are maximized while preventing generator power overshoot and feedwater pump vaporization shutdown accidents. Attached Figure Description

[0017] Figure 1 A flowchart of a dynamic power regulation control method for thermal power units based on AGC is provided in one embodiment of the present invention; Figure 2 This is a flowchart of a method for obtaining throttling exit condition indicators according to an embodiment of the present invention. Detailed Implementation

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] The following description, in conjunction with the accompanying drawings, details the specific scheme of the dynamic power regulation and control method and system for thermal power units based on AGC provided by the present invention.

[0020] Please see Figure 1 The diagram illustrates a flowchart of a dynamic power regulation control method for thermal power units based on AGC, according to an embodiment of the present invention. The method includes: Step S101: Obtain the actual active power of the generator, the target power of automatic power generation control, the main steam pressure in front of the generator, the turbine control valve opening, and the actual water level of the deaerator at the current sampling time.

[0021] This invention receives data from the distributed control system (DCS) of a combined heat and power (CHP) thermal power unit (hereinafter referred to as the unit) at fixed sampling intervals. At each sampling time, the data includes the actual active power of the generator collected by the power transmitter on the generator outlet side, the latest automatic power generation control target power issued by the power grid dispatch center, the main steam pressure before the turbine collected by the pressure transmitter on the turbine main steam pipeline, the turbine valve opening collected by the displacement sensor on the turbine regulating valve servo mechanism, the actual water level of the deaerator collected by the differential pressure level gauge on the deaerator water tank, and the condensate regulating valve opening collected by the valve positioner on the condensate pipeline.

[0022] In one specific implementation of this invention, the fixed sampling interval is set to 1 second.

[0023] In this embodiment of the invention, a throttling intervention status bit is preset. This throttling intervention status bit is a flag variable used to latch the control status of the unit, and its value is 0 or 1. It is set to 0 during the initial operation of the unit, indicating that the unit is not in a state of performing condensate throttling operation. When the triggering condition for condensate throttling operation is met during operation, it is set to 1, representing that the unit is in a state of performing condensate throttling operation. The invention continuously executes algorithm analysis throughout the entire condensate throttling operation until the condensate throttling operation ends, at which point the throttling intervention status bit is reset to 0 to prevent the data flow and control logic from being interrupted midway.

[0024] Specifically, when the present invention detects that the target power of automatic power generation control at the current sampling time is greater than the target power of automatic power generation control at the previous adjacent sampling time, and the opening degree of the condensate regulating valve at the current sampling time is less than the opening degree of the condensate regulating valve at the previous adjacent sampling time, it determines that the power grid dispatch center requires the unit to increase power generation and that the unit is closing the opening degree of the condensate regulating valve in order to quickly increase power generation. At this time, the present invention sets the throttling intervention status bit to 1 and performs subsequent steps of analysis on all data at the current sampling time and subsequent sampling times until the throttling intervention status bit is 0. Otherwise, the present invention saves all data at the current sampling time and does not perform subsequent steps of analysis.

[0025] Step S102: Based on the difference between the automatic power generation control target power and the actual active power of the generator, and combined with the rated active power of the generator, the relative power error is obtained; based on the difference between the main steam pressure at the generator front at the current sampling time and the historical adjacent sampling time, the main steam pressure trend at the current sampling time is obtained; based on the relative relationship between the lower safety limit water level of the deaerator and the actual water level of the deaerator, the relative water level risk is determined.

[0026] The difference between the automatic generation control target power and the generator's actual active power represents the power gap between the generator's actual active power output and the active power required by the grid dispatch center. In other words, it represents the power shortfall in the generator's failure to meet grid load commands during transient operations. The generator's rated active power represents the baseline power for full-load operation. The relative power error obtained by combining the two represents the relative proportion between the generator's power shortfall and the baseline power, reflecting the relative underload degree of the generator.

[0027] Since both the target power of automatic generation control and the actual active power of the generator are dynamic and fluctuating absolute physical quantities, the difference between the two has different weights on the unit under different load conditions. Therefore, by introducing the static and constant rated active power of the generator as the denominator, the fluctuating power gap can be transformed into a dimensionless relative proportion under a unified benchmark, thus eliminating the difference in magnitude between the units under different operating conditions.

[0028] The difference in main steam pressure at the turbine inlet between the current sampling time and the previous adjacent historical sampling time represents the change in steam pressure inside the turbine's main steam pipeline over two consecutive sampling times. Since the turbine's main steam pipeline connects the boiler and the turbine, the change in steam pressure inside the pipeline essentially depends on the dynamic supply-demand difference between boiler steam production and turbine steam consumption. Therefore, analyzing the main steam pressure trend by examining the change in steam pressure inside the turbine's main steam pipeline over two consecutive sampling times can reflect the physical state of transient imbalance between steam production and consumption within the unit. This allows for early detection of the inflection point of delayed steam influx from the boiler before a significant rise in absolute pressure readings, preventing a delay in ending condensate throttling operations.

[0029] The lower safety limit water level of the deaerator represents the inherent physical baseline water level required to maintain normal feedwater pumping without vaporization damage. The relative water level risk, determined by the lower safety limit water level and the actual water level of the deaerator, represents the degree of risk of depletion of the deaerator's usable water capacity. By converting the absolute liquid level height of the deaerator into a dimensionless risk indicator, a forced shutdown intervention mechanism can be provided when the water level approaches the equipment's physical red line, preventing excessive throttling that could lead to feedwater pump burnout and shutdown.

[0030] Preferably, in some implementations of the present invention, the relative power error is obtained based on the difference between the automatic power generation control target power and the actual active power of the generator, combined with the rated active power of the generator, including: The difference between the automatic power generation control target power and the actual active power of the generator is taken as the current power deviation. The relative power error is obtained based on the relative relationship between the current power deviation and the generator's rated active power.

[0031] The current power deviation represents the number of megawatts of underload between the actual active power output of the generator at the current sampling time and the active power required by the power grid dispatch center. It directly reflects the power gap currently faced by the generator.

[0032] In this embodiment, the result of dividing the current power deviation by the generator's rated active power is taken as the relative power error; the result of dividing the current power deviation by the generator's rated active power reflects the relative relationship between the current power deviation and the generator's rated active power.

[0033] In one specific implementation of this invention, the rated active power of the generator is directly obtained from the generator's manufacturer's manual. As an example, for common deep peak-shaving cogeneration thermal power units, the rated active power of the generator is typically 300 MW.

[0034] In this embodiment of the invention, if the current power deviation is less than 0, it indicates that the actual active power of the generator has exceeded the target power of automatic power generation control and there is no power gap. At this time, the current power deviation is set to 0; if the current power deviation is greater than or equal to 0, the original value is retained.

[0035] Preferably, in some implementations of the present invention, the trend of the main steam pressure at the current sampling time is obtained based on the difference between the main steam pressure at the current sampling time and the historical adjacent sampling time, including: Based on the difference between the main steam pressure at the current sampling time and the main steam pressure at the previous adjacent sampling time in history, and combined with the time interval between the current sampling time and the previous adjacent sampling time in history, the trend of the main steam pressure at the current sampling time is obtained.

[0036] The change in steam pressure inside the main steam pipeline of the turbine within two consecutive sampling times is transformed into the dynamic rate of change of pressure over time, i.e. the trend of main steam pressure. This can eliminate the computational drift interference caused by the distributed control system of the unit under different sampling intervals, and accurately quantify the rate and direction of change of transient imbalance between steam production and steam consumption inside the unit.

[0037] Specifically, the main steam pressure trend at the current sampling time is obtained by subtracting the difference between the main steam pressure at the turbine inlet at the previous adjacent sampling time from the current sampling time, and then dividing by the time interval between the current sampling time and the previous adjacent sampling time.

[0038] It should be understood that the time interval between any two consecutive sampling moments is the fixed sampling interval of the distributed control system of the cogeneration thermal power unit in step S101. The unit of the main steam pressure trend is megapascals per second (MPa / s).

[0039] Preferably, in some implementations of the present invention, determining the relative water level risk level based on the relative relationship between the lower safety limit water level of the deaerator and the actual water level of the deaerator includes: The relative water level risk is calculated by dividing the lower safety limit water level of the deaerator by the actual water level of the deaerator at the current sampling time.

[0040] This embodiment uses the result of dividing the lower safety limit water level of the deaerator by the actual water level of the deaerator at the current sampling time to reflect the relative relationship between the lower safety limit water level of the deaerator and the actual water level of the deaerator at the current sampling time.

[0041] The relative water level risk level represents the degree of risk of depletion of the deaerator's available water capacity. The higher the relative water level risk level, the closer the actual water level of the deaerator is to the deaerator's safe lower limit water level, and the higher the degree of risk of depletion of the deaerator's available water capacity. The lower the relative water level risk level, the further the actual water level of the deaerator is from the deaerator's safe lower limit water level, and the lower the degree of risk of depletion of the deaerator's available water capacity.

[0042] In one specific implementation of this invention, the lower safety limit water level of the deaerator is an inherent anti-vaporization baseline parameter calibrated during the equipment's factory design and can be obtained from the deaerator's instruction manual. As an example, for common deep peak-shaving cogeneration thermal power units, the lower safety limit water level of the deaerator is typically 800 mm.

[0043] Step S103: Combine the relative power error, main steam pressure trend, and turbine valve opening between the current sampling time and historical sampling times, and combine them with the relative water level risk at the current sampling time to determine the throttling exit condition index at the current sampling time.

[0044] The throttling exit condition index represents the overall cost of continuing condensate throttling operation for the unit, integrating the risks and benefits of continuing condensate throttling operation. A decrease in the throttling exit condition index indicates that the benefits of filling the power gap through throttling outweigh the equipment safety risks; an increase in the throttling exit condition index indicates that the benefits of filling the power gap through throttling are less than the equipment safety risks.

[0045] By integrating the relative power error between the current sampling time and the historical sampling time, the main steam pressure trend, and the turbine control valve opening, it is possible to analyze the joint evolution trajectory of the unit's multi-dimensional characteristic variables in the time dimension. This allows us to remove the short-term interference caused by the turbine control valve action on the pressure reading at the pure data level and extract the implicit features that characterize the boiler's true steam accumulation trend.

[0046] By combining the relative water level risk at the current sampling moment, a physical water level safety boundary of the unit is superimposed on this implicit feature. The resulting throttling exit condition index can objectively quantify the comprehensive pressure the unit faces in continuing throttling operations under the combined effects of power deficit contraction, accelerated steam accumulation, and continuous water level decline. This operation transforms the power deficit, pressure fluctuations, and water level risk, which have different physical dimensions, into a unified dimensionless cost scale, eliminating the dimensional barriers of multiple physical field parameters and laying a unified data foundation for subsequently overcoming the blind spots of local variables and accurately capturing the timing of throttling exit.

[0047] Preferably, such as Figure 2 As shown, in some implementations of this invention, the relative power error between the current sampling time and historical sampling times, the main steam pressure trend, and the turbine control valve opening are integrated, along with the relative water level risk at the current sampling time, to determine the throttling exit condition indicators for the current sampling time, including: Step S201: Normalize the relative power error between the current sampling time and the historical sampling time within the preset backtracking time window to obtain the normalized error characteristics of the current sampling time and the historical sampling time.

[0048] Because the relative power error, main steam pressure trend, and turbine valve opening vary significantly in their numerical ranges and distribution intervals (for example, the relative power error is typically between 0 and 0.2; the main steam pressure trend is typically between -0.1 and 0.1; and the turbine valve opening is typically between 0 and 100), without normalization, larger variables will completely overwhelm smaller variables during subsequent matrix dimensionality reduction. Normalization maps different feature variables to the same dimensionless interval, ensuring that the implicit features of the actual boiler steam accumulation trend obtained subsequently can capture the dynamic evolution of each dimension variable in a balanced manner.

[0049] The normalized error characteristic represents the relative fluctuation distribution of the unit's relative power error within a preset backtracking time window, with a value ranging from 0 to 1. Specifically, the closer the normalized error characteristic is to 1, the larger the power gap between the corresponding sampling time and other sampling times within the preset backtracking time window; conversely, the closer the normalized error characteristic is to 0, the smaller the power gap between the corresponding sampling time and other sampling times within the preset backtracking time window.

[0050] In this embodiment of the invention, the current sampling time is taken as the last sampling time in the preset backtracking time window, and the relative power error, main steam pressure trend and turbine valve opening at the current sampling time and the historical sampling time in the preset backtracking time window are obtained for subsequent calculations.

[0051] Specifically, the maximum value among all relative power errors within the preset backtracking time window is taken as the maximum error; the minimum value among all relative power errors within the preset backtracking time window is taken as the minimum error; the maximum and minimum error values ​​are used to perform maximum and minimum value normalization on all relative power errors within the preset backtracking time window to obtain the normalized error characteristics at each sampling time.

[0052] It should be understood that the method of normalizing the original data using maximum and minimum value normalization is a well-known technique and will not be elaborated further.

[0053] It should be noted that during condensate throttling, the actual active power of the generator continuously approaches the target power of the automatic power generation control, and the relative power error will inevitably change over time. Theoretically, the maximum and minimum error values ​​will not be equal. However, if a sensor communication failure causes data freezing or the DCS system's precision is truncated, resulting in absolutely consistent collected values, the maximum error value may equal the minimum error value. In this case, to prevent the normalization fraction calculation from resulting in a zero denominator and causing program crashes, the above maximum and minimum value normalization calculation is not performed. Instead, the normalized error characteristic of all sampling times within the preset backtracking time window is directly set to 0.

[0054] In one specific implementation of this invention, the length of the preset backtracking time window is set to 60 sampling times.

[0055] In this embodiment of the invention, if the number of current sampling moments and historical sampling moments within the preset backtracking time window is less than the length of the preset backtracking time window during the initial stage of unit operation, the subsequent analysis and calculation are suspended, and only the data at each sampling moment is saved and accumulated until the number of current sampling moments and historical sampling moments within the preset backtracking time window reaches the length of the preset backtracking time window, at which point the subsequent analysis and calculation are performed.

[0056] Step S202: Normalize the main steam pressure trends at the current sampling time and the historical sampling time within the preset backtracking time window to obtain the normalized pressure characteristics at the current sampling time and the historical sampling time.

[0057] The normalized pressure characteristic represents the relative fluctuation distribution of the unit's main steam pressure trend within a preset retrospective time window, with a value ranging from 0 to 1. Specifically, a normalized pressure characteristic closer to 1 indicates a greater positive pressure rise rate at the corresponding sampling time relative to other sampling times within the preset retrospective time window, while a normalized pressure characteristic closer to 0 indicates a greater negative pressure fall rate at the corresponding sampling time relative to other sampling times within the preset retrospective time window.

[0058] Specifically, the maximum value among all main steam pressure trends within the preset backtracking time window is taken as the maximum pressure value; the minimum value among all main steam pressure trends within the preset backtracking time window is taken as the minimum pressure value; the maximum and minimum pressure values ​​are used to perform maximum and minimum value normalization on all main steam pressure trends within the preset backtracking time window to obtain the normalized pressure characteristics at each sampling time.

[0059] It should be understood that the method of normalizing the original data using maximum and minimum value normalization is a well-known technique and will not be elaborated further.

[0060] It should be noted that during condensate throttling, the main steam, being a high-temperature and high-pressure fluid, will inevitably experience slight fluctuations in readings due to internal disturbances and sensor measurement noise. Theoretically, the maximum and minimum pressure values ​​of the main steam should not be equal. However, if a sensor communication failure causes data freezing or the DCS system's accuracy is truncated, resulting in absolutely consistent collected values, the maximum pressure value may equal the minimum pressure value. In this case, to prevent the normalized fraction calculation from resulting in a zero denominator and causing program crashes, the above maximum and minimum value normalization calculation is not performed. Instead, the normalized pressure characteristic of all sampling times within the preset backtracking time window is directly set to 0.

[0061] Step S203: Normalize the turbine valve opening at the current sampling time and the historical sampling time within the preset backtracking time window to obtain the normalized opening characteristics at the current sampling time and the historical sampling time.

[0062] The normalized opening characteristic represents the relative fluctuation distribution of the turbine control valve opening within a preset backtracking time window, with a value ranging from 0 to 1. Specifically, a normalized opening characteristic closer to 1 indicates a greater degree of turbine control valve opening at the corresponding sampling time compared to other sampling times within the preset backtracking time window; conversely, a normalized opening characteristic closer to 0 indicates a smaller degree of turbine control valve opening at the corresponding sampling time compared to other sampling times within the preset backtracking time window.

[0063] Specifically, the maximum value among all turbine valve openings within the preset backtracking time window is taken as the maximum opening value; the minimum value among all turbine valve openings within the preset backtracking time window is taken as the minimum opening value; the maximum and minimum opening values ​​are used to perform maximum and minimum value normalization on all turbine valve openings within the preset backtracking time window to obtain the normalized opening feature at each sampling time.

[0064] It should be understood that the method of normalizing the original data using maximum and minimum value normalization is a well-known technique and will not be elaborated further.

[0065] It should be noted that, as a mechanical actuator, the turbine control valve may remain absolutely still for an extended period after its action is completed, or sensor communication may become stuck, causing the maximum opening value to equal the minimum opening value. In this case, to prevent the normalized fraction calculation from resulting in a zero denominator and causing program crashes, the above maximum and minimum value normalization calculation is not performed. Instead, the normalized opening characteristic of all sampling moments within the preset backtracking time window is directly set to 0.

[0066] Step S204: Based on the normalized error characteristics, normalized pressure characteristics, and normalized opening characteristics of the current sampling time and the historical sampling time within the preset backtracking time window, construct a multi-dimensional feature matrix.

[0067] When the generating unit responds to the automatic power generation control target issued by the power grid dispatch center, the large-scale operation of the turbine control valve consumes the existing steam in the turbine's main steam pipeline, masking or even reversing the true pressure appearance. Therefore, relying solely on pressure cannot reveal the boiler's delayed steam production trend. This embodiment constructs a multi-dimensional feature matrix based on the normalized error characteristics, normalized pressure characteristics, and normalized opening characteristics of the current sampling time and historical sampling times within a preset backtracking time window. This aims to combine these three physically strongly coupled and temporally misaligned variables into a unified matrix, establishing a complete mathematical mapping of the unit's transient thermal supply and demand. This lays a spatial foundation for subsequent matrix dimensionality reduction algorithms to filter out valve operation interference in multi-dimensional space and extract the true implicit trend representing the accumulation of boiler steam production.

[0068] The multidimensional feature matrix represents the set of joint time-series state evolutions of the unit within a preset backtracking time window, including the grid load demand response (relative power error), the pipeline charging and discharging status (main steam pressure trend), and the transient intervention of the actuator (turbine valve opening).

[0069] Furthermore, based on the normalized error characteristics, normalized pressure characteristics, and normalized opening characteristics of the current sampling time and historical sampling times within a preset backtracking time window, a multi-dimensional feature matrix is ​​constructed, including: The normalized error features of the current sampling time and the historical sampling time are arranged in chronological order to construct an error row vector; The normalized pressure features of the current sampling time and the historical sampling time are arranged in chronological order to construct a pressure row vector; Construct an opening row vector by arranging the normalized opening features of the current sampling time and the historical sampling time in chronological order; A multidimensional feature matrix is ​​constructed based on the error row vector, pressure row vector, and opening row vector.

[0070] Each column inside the multidimensional feature matrix represents a transient snapshot of an independent sampling moment within a preset backtracking time window, specifically including three elements: the normalized error feature, the normalized pressure feature, and the normalized opening feature of the unit at that sampling moment; each row inside the multidimensional feature matrix represents the evolution trajectory of the corresponding dimension feature in the continuous time series within the preset backtracking time window.

[0071] It should be noted that the multidimensional feature matrix is ​​a 3-row, N-column matrix, where N represents the number of sampling moments within the preset backtracking time window. The column vectors within the multidimensional feature matrix are strictly arranged in chronological order; however, the order of the error row vector, pressure row vector, and opening row vector within the multidimensional feature matrix is ​​not specific. For example, the error row vector can be placed in the first row, the pressure row vector in the second row, and the opening row vector in the third row.

[0072] In this embodiment of the invention, if the obtained multidimensional feature matrix is ​​a zero matrix, it indicates that the unit sensor communication has a global freeze fault. In order to ensure the safe operation of the unit, the invention immediately triggers the highest priority audible and visual alarm of the unit's distributed control system to prompt the operator to intervene manually; at the same time, the throttling intervention status is directly set to 0, and the subsequent calculation process is stopped.

[0073] Step S205: Determine the characteristic value of steam change trend at the current sampling time based on the state transition law of the multidimensional feature matrix.

[0074] The state transition law of the multidimensional feature matrix represents the transformation and mapping relationship of the three characteristic variables of the unit, namely the relative power error, the main steam pressure trend and the turbine control valve opening, in the multidimensional space between two adjacent sampling times. It reflects the inherent dynamic inertia and physical driving logic of the unit's thermodynamic system.

[0075] The characteristic value of steam variation trend represents the rate of divergence of the implicit imbalance where steam production exceeds steam consumption within the unit. A larger characteristic value indicates a stronger tendency for the actual accumulated, delayed steam production on the boiler side to surge into the turbine. The steam variation trend characteristic value obtained in this invention is a dimensionless value.

[0076] Based on the state transition law of the multidimensional feature matrix, the characteristic value of steam change trend is determined. It is possible to extract the most accurate true steam accumulation trend of the boiler from the aliased phenomenon that is disturbed by the large movement of the turbine control valve. This enables precise feedforward early warning of the delayed steam production of the unit, ensuring the stable response of the generator active power during the deep peak shaving process and avoiding serious power overshoot and performance violation risks.

[0077] Furthermore, based on the state transition law of the multidimensional feature matrix, the characteristic value of the steam change trend at the current sampling time is determined, including: The multidimensional feature matrix is ​​processed using a dynamic mode decomposition algorithm to obtain continuous-time modal eigenvalues; The maximum value among the continuous time modal eigenvalues ​​is taken as the steam change trend eigenvalue at the current sampling time.

[0078] Continuous-time modal eigenvalues ​​represent the evolutionary characteristics of the supply-demand imbalance between steam production and consumption within the unit, separated after dimensionality reduction and decoupling of the multidimensional feature matrix. The numerical values ​​of the continuous-time modal eigenvalues ​​characterize the exponential evolution rate of the supply-demand imbalance between steam production and consumption within the unit. When the continuous-time modal eigenvalue is positive and increasing, it indicates an accumulation and divergence trend where steam production exceeds steam consumption; when the continuous-time modal eigenvalue is negative or tends to zero, it indicates a consumption convergence trend where steam production is less than or equal to steam consumption, or a state of equilibrium.

[0079] During condensate throttling, the real threat to unit safety and the resulting power overshoot is the irreversible excess accumulation of stagnant steam production within the turbine's main steam pipe. The maximum value among the continuous-time modal eigenvalues ​​physically corresponds precisely to the core dominant mode in the unit where steam production exceeds steam consumption, resulting in the most intense and uncontrolled energy accumulation. Therefore, using the maximum value among the continuous-time modal eigenvalues ​​as the steam change trend characteristic value can accurately pinpoint this most serious threat of excess steam production accumulation, providing the most sensitive feedforward early warning signal for determining the timing of throttling exit.

[0080] Specifically, the steps for processing the multidimensional feature matrix using the dynamic mode decomposition algorithm to obtain continuous-time mode eigenvalues ​​are as follows: 1. Divide the multidimensional feature matrix into a feature matrix for the previous time step and a feature matrix for the next time step. The feature matrix for the previous time step consists of columns 1 to N-1 of the multidimensional feature matrix, and the feature matrix for the next time step consists of columns 2 to N of the multidimensional feature matrix. N is the total number of columns in the multidimensional feature matrix.

[0081] 2. Perform singular value decomposition on the characteristic matrix of the previous time step to obtain the left singular matrix, the singular value diagonal matrix, and the right singular matrix. Then, combine the singular value diagonal matrix with a very small positive number (e.g., 10). -5 After adding them together, find the inverse matrix to obtain the bias singular value inverse matrix. Then, perform matrix multiplication sequentially on the transpose of the left singular matrix, the characteristic matrix of the next time step, the right singular matrix, and the bias singular value inverse matrix to obtain the first-order state transition matrix.

[0082] 3. Perform eigenvalue decomposition on the first-order state transition matrix to obtain discrete eigenvalues. Perform a modulo operation on these discrete eigenvalues, taking their absolute values. To avoid logarithmic overflow, the resulting value is multiplied by a very small positive number (e.g., 10). -5The summation is performed, and the result is then logarithmically calculated to obtain the continuous-time modal eigenvalues. Using the logarithm with the natural constant e as the base, the above summation result is used as the argument for logarithmic operations.

[0083] It should be understood that the method of using dynamic mode decomposition algorithm to perform segmentation, dimensionality reduction and eigenvalue extraction on feature matrix composed of multidimensional discrete time series data in order to separate the eigenvalues ​​that characterize the dynamic evolution of the system is a well-known technology in this field and will not be described in detail here.

[0084] Step S206: By integrating the relative power error, main steam pressure trend, relative water level risk, and steam change trend characteristic value at the current sampling time, the throttling exit condition index at the current sampling time is obtained.

[0085] The throttling exit condition index, obtained by integrating the relative power error, main steam pressure trend, relative water level risk, and steam change trend characteristics at the current sampling moment, is a global safety and regulation benefit mapping boundary that integrates four dimensions: unit power gap, pipeline charging and discharging status, physical available water capacity baseline, and boiler steam implicit accumulation rate. It eliminates the dimensional barriers of multi-physics parameters, enabling cross-constraint of grid load command response requirements and internal hysteretic thermodynamic misalignment of the boiler system within the same dimension without relying on complex physical modeling. This achieves feedforward early warning intervention during condensate throttling, ensuring precise exit from throttling before generator power overshoot and feedwater pump vaporization accidents occur, effectively balancing active power enhancement benefits with equipment operational safety.

[0086] Specifically, taking the current sampling time as an example, the formula for calculating the throttling exit condition index is as follows: in, The throttling exit condition indicator represents the current sampling moment; Represents the preset pressure characteristic weight; e represents the natural constant; The characteristic value representing the steam change trend at the current sampling moment; This represents the trend of the main steam pressure at the current sampling moment; This represents the preset water level penalty weight; Represents the relative water level risk at the current sampling moment; Represents the preset power penalty weight; This represents the relative power error at the current sampling moment.

[0087] The pressure change term represents a trend amplification. This method nonlinearly amplifies the characteristic values ​​of steam change trends, giving the control algorithm extreme sensitivity to the process of implicit steam production accumulation in the boiler. This term represents the amplified effect of transient pressure fluctuations after the implicit steam production accumulation trend. In the middle and late stages of the condensate throttling process, when the boiler steam production lags and accumulates, causing the pressure to begin a slight recovery, this product term can utilize... The resulting exponential multiple amplifies the weak positive signal of the main steam pressure trend, thereby driving a rapid increase in the throttling exit condition index to prematurely end the condensate throttling operation. The preset pressure characteristic weight α not only eliminates the physical unit dimension of the main steam pressure trend, transforming it into a dimensionless pure numerical value to satisfy the addition operation logic, but also serves as an adjustment factor to determine the influence ratio of this pressure characteristic item in the overall evaluation index, with an exemplary value of 1.0 s / MPa.

[0088] The preset water level penalty weight is used to adjust the proportion of the relative water level risk in the throttling exit condition indicators, and its example value is 2.0.

[0089] The preset power penalty weight is used to adjust the proportion of the influence of relative power error in the throttling exit condition index, and its example value is 1.5.

[0090] It should be noted that when configuring the above weight parameters during the unit commissioning phase, an initial descent constraint condition must be met. Specifically, in the initial stage of condensate throttling operation, the absolute value of the product of the rate of decrease in relative power error and the power penalty weight must be greater than the absolute value of the product of the rate of increase in relative water level risk and the water level penalty weight. By configuring this constraint condition, it is ensured that the throttling exit condition indicator has a monotonically decreasing segment in the initial stage of intervention operation, thereby meeting the basic triggering requirement that the gradient of subsequent continuous historical indicators be less than or equal to 0.

[0091] Step S104: Based on the dynamic evolution characteristics of the throttling exit condition indicators at the current sampling time and the historical sampling time, determine the measures for issuing regulation and control commands.

[0092] The dynamic evolution of the throttling exit condition index reflects the dynamic game process and reversal trend between the benefits of filling the power gap and the risks of generator power overshoot and water level depletion during condensate throttling. Therefore, by analyzing the dynamic evolution of the throttling exit condition index, we can determine the critical state at which the throttling operation of the unit reverses from being more beneficial than harmful to being more harmful than beneficial. Using this critical state as a criterion, we can determine the issuance of regulation and control commands, accurately capturing the timing of the condensate throttling operation's termination and precisely issuing regulation and control commands. This ensures that while maximizing the active power increase benefit, it prevents generator power overshoot and feedwater pump vaporization shutdown accidents, overcoming the intervention lag and observation dead zone defects of traditional fixed threshold control.

[0093] Preferably, in some implementations of the present invention, the measures for issuing adjustment control commands are determined based on the dynamic evolution characteristics of the throttling exit condition indicators at the current sampling time and historical sampling times, including: The difference between the throttling exit condition index at the current sampling time and the throttling exit condition index at the previous adjacent historical sampling time is used as the gradient of the throttling exit condition index at the current sampling time. If the gradients of the throttling exit condition indexes of the preset number of historical sampling times adjacent to the current sampling time are all less than or equal to 0, and the gradient of the throttling exit condition indexes of the current sampling time is greater than 0, then an adjustment control command is issued; otherwise, no adjustment control command is issued.

[0094] The gradient of the throttling exit condition index represents the evolution direction of the overall cost of maintaining condensate throttling operation over time between two consecutive sampling moments. A gradient of less than or equal to 0 indicates that the benefits of continuing condensate throttling operation outweigh the risks; a gradient of greater than 0 indicates that the risks of continuing condensate throttling operation outweigh the benefits.

[0095] If the gradients of the throttling exit condition indicators for a preset number of historical sampling times adjacent to the current sampling time are all less than or equal to 0, and the gradient of the throttling exit condition indicator for the current sampling time is greater than 0, it indicates that after a period of continuous decline where the benefits outweigh the risks, the unit has experienced its first absolute reversal of risk outweighing benefits at the current sampling time. This combination of conditions physically corresponds to the critical red line where throttling intervention must be immediately terminated to prevent severe overshooting of the unit's generator power and vaporization of the feedwater pump, thus triggering the issuance of a regulation control command.

[0096] If the above conditions are not met, it means that the unit has not yet reached the critical inflection point, and may be in the initial downward throttling period dominated by the power gap, or in a period of gentle fluctuation. At this time, the benefits of continuing to maintain condensate throttling operation to fill the power gap still outweigh the potential risks, and the unit maintains its current throttling operation without issuing regulation control commands.

[0097] In one specific implementation of this invention, the preset quantity is set to 5.

[0098] Preferably, in some implementations of the embodiments of the present invention, the adjustment control command includes: Map the current steam pressure trend at the sampling moment to the velocity increment of the condensate regulating valve; The speed increment is added to the preset reference safety speed to obtain the opening speed of the condensate regulating valve; The opening speed is used as an adjustment control command.

[0099] The current steam pressure trend at the sampling moment reflects the intensity and accumulation rate of the delayed steam inflow into the turbine's main steam pipeline under the current operating conditions. Mapping the main steam pressure trend to the velocity increment of the condensate regulating valve enables adaptive dynamic matching between the valve's opening response and the internal steam accumulation intensity. This allows for real-time adjustment of the deaerator cooling water replenishment rate based on the speed of the boiler steam inflow, and also helps prevent generator power overshoot.

[0100] Specifically, the product of the current main steam pressure trend at the sampling moment and the preset valve opening ratio coefficient is used as the speed increment of the condensate regulating valve. The preset valve opening ratio coefficient reflects the equivalent conversion relationship between the unit's pressure volume characteristics and the regulating valve's hardware flow characteristics. Its unit is % / MPa (per megapascal), which is used to convert the main steam pressure trend dimension (MPa / s) into the valve opening speed dimension (% / s).

[0101] The preset baseline safety speed represents the minimum hardware safety rate at which the condensate regulating valve performs its opening operation. Adding this to the speed increment of the condensate regulating valve ensures that, at the moment the condensate throttling operation ends, even if the pipeline pressure recovery has not yet fully manifested (i.e., the main steam pressure trend is 0, resulting in a calculated speed increment of 0), the condensate regulating valve can still slowly open at a basic and safe minimum rate, prioritizing the preservation of the deaerator's water circulation margin that is nearing the red line.

[0102] To prevent the opening speed of the condensate regulating valve from exceeding the actual mechanical limit of the valve motor, which could lead to out-of-bounds errors or hardware damage, if the opening speed of the condensate regulating valve is greater than the preset maximum safe speed, then the preset maximum safe speed will be used as the opening speed of the condensate regulating valve; if the opening speed of the condensate regulating valve is less than or equal to the preset maximum safe speed, then the obtained opening speed of the condensate regulating valve will be retained.

[0103] The opening speed of the condensate regulating valve is used as the regulation and control command. The specific opening speed of the condensate regulating valve is clearly defined, so that the condensate regulating valve is freed from the lag feedback dependence on the generator power exceeding the limit. Condensate is introduced smoothly at a dynamic and safe opening speed. While quickly restoring the safe liquid level of the deaerator, the unit's operation is avoided from oscillation, thus completing the safe closed loop of the entire automatic power generation control deep peak shaving and throttling process.

[0104] In one specific implementation of this invention, the method for obtaining the preset valve opening ratio coefficient is as follows: during the unit commissioning phase, multiple sets of data on the change in the opening degree of a given condensate regulating valve and the corresponding trend data of the main steam pressure are collected; the least squares method is used to perform linear fitting on the above discrete data samples, and the slope of the calculated fitted line is used as the preset valve opening ratio coefficient.

[0105] The preset baseline safety speed and preset maximum safety speed are obtained from the hardware manual of the condensate regulating valve. As an example, for common deep peak-shaving cogeneration thermal power units, the preset baseline safety speed of the condensate regulating valve is usually 1.0% / s, and the preset maximum safety speed is usually 5.0% / s.

[0106] In this embodiment of the invention, before calculating the speed increment of the condensate regulating valve, it is first determined whether the main steam pressure trend at the current sampling time is less than 0. If the main steam pressure trend is less than 0, it indicates that the steam consumption inside the unit is greater than the steam production, and the unit is in a steam deficit stage. In order to prevent this negative value from causing the calculated speed increment to be negative, thereby causing the condensate regulating valve to close in reverse or stop moving, the main steam pressure trend is set to 0 at this time; if the main steam pressure trend is greater than or equal to 0, the original value is retained.

[0107] In this embodiment of the invention, after the regulation control command is issued, the invention monitors in real time whether the actual water level of the deaerator recovers and stabilizes at the deaerator's highest normal water level at each subsequent sampling time. For any subsequent sampling time, if the actual water level of the deaerator at that sampling time is greater than or equal to the deaerator's highest normal water level, it indicates that the available water circulation buffer capacity depleted due to condensate throttling operation has been fully compensated, and the unit is out of risk of water level depletion. At this time, the issued regulation control command is immediately revoked, and the throttling intervention status bit is reset to 0. If the actual water level of the deaerator at that sampling time is still less than the deaerator's highest normal water level, the issuance of the regulation control command continues.

[0108] In one specific implementation of this invention, the highest normal water level of the deaerator is the upper limit parameter of the standard operating liquid level specified in the equipment's factory design. This represents the safe buffer high level of the deaerator in a non-interventional state and can be obtained from the deaerator's instruction manual. As an example, for common deep peak-shaving cogeneration thermal power units, the highest normal water level of the deaerator is typically 2500 mm.

[0109] In summary, this invention obtains the relative power error based on the difference between the automatic power generation control target power and the actual active power of the generator at the current sampling time, combined with the generator's rated active power; it obtains the main steam pressure trend at the current sampling time based on the difference between the main steam pressure at the turbine inlet at the current sampling time and the previous adjacent historical sampling time; and it determines the relative water level risk based on the relative relationship between the deaerator's lower safety limit water level and the actual water level of the deaerator at the current sampling time. By integrating the relative power error, main steam pressure trend, and turbine valve opening at the current sampling time with the relative water level risk, the throttling exit condition index at the current sampling time is determined. Based on the dynamic evolution characteristics of the throttling exit condition index at the current sampling time and the historical sampling time, the measures for issuing regulation control commands are determined.

[0110] This invention also proposes a dynamic power regulation and control system for thermal power units based on AGC, the system comprising: The data acquisition module is used to obtain the actual active power of the generator, the target power of automatic power generation control, the main steam pressure in front of the turbine, the turbine control valve opening, and the actual water level of the deaerator at the current sampling time. The basic feature analysis module is used to obtain the relative power error based on the difference between the automatic power generation control target power and the actual active power of the generator, combined with the rated active power of the generator; to obtain the main steam pressure trend at the current sampling time based on the difference between the main steam pressure at the current sampling time and the previous adjacent sampling time; and to determine the relative water level risk based on the relative relationship between the lower safety limit water level of the deaerator and the actual water level of the deaerator. The throttling exit condition index analysis module is used to integrate the relative power error, main steam pressure trend and turbine control valve opening at the current sampling time and the historical sampling time, and combine them with the relative water level risk at the current sampling time to determine the throttling exit condition index at the current sampling time. The results analysis module is used to determine the measures for issuing regulation and control commands based on the dynamic evolution characteristics of the throttling exit condition indicators at the current sampling time and historical sampling times.

Claims

1. A dynamic power regulation control method for thermal power units based on AGC, characterized in that, The method includes: The system acquires the generator's actual active power, automatic power generation control target power, main steam pressure before the turbine, turbine control valve opening, and deaerator water level at the current sampling time. Based on the difference between the automatic power generation control target power and the actual active power of the generator, and combined with the rated active power of the generator, the relative power error is obtained; based on the difference between the main steam pressure at the generator front at the current sampling time and the historical adjacent sampling time, the main steam pressure trend at the current sampling time is obtained; based on the relative relationship between the safe lower limit water level of the deaerator and the actual water level of the deaerator, the relative water level risk is determined. By integrating the relative power error between the current sampling time and the historical sampling time, the main steam pressure trend, and the turbine control valve opening, and combining the relative water level risk at the current sampling time, the throttling exit condition indicators at the current sampling time are determined. Based on the dynamic evolution characteristics of the throttling exit condition indicators at the current sampling time and the historical sampling time, the measures for issuing regulation and control commands are determined. The adjustment control command includes: Map the current steam pressure trend at the sampling moment to the velocity increment of the condensate regulating valve; The speed increment is added to the preset reference safety speed to obtain the opening speed of the condensate regulating valve; The opening speed is used as an adjustment control command.

2. The dynamic power regulation and control method for thermal power units based on AGC according to claim 1, characterized in that, The relative power error, derived from the difference between the automatic power generation control target power and the actual active power of the generator, combined with the generator's rated active power, includes: The difference between the automatic power generation control target power and the actual active power of the generator is taken as the current power deviation. The relative power error is obtained based on the relative relationship between the current power deviation and the generator's rated active power.

3. The dynamic power regulation and control method for thermal power units based on AGC according to claim 1, characterized in that, The method of obtaining the main steam pressure trend at the current sampling time based on the difference between the main steam pressure at the current sampling time and the historical adjacent previous sampling time includes: Based on the difference between the main steam pressure at the current sampling time and the main steam pressure at the previous adjacent sampling time in history, and combined with the time interval between the current sampling time and the previous adjacent sampling time in history, the trend of the main steam pressure at the current sampling time is obtained.

4. The dynamic power regulation and control method for thermal power units based on AGC according to claim 1, characterized in that, The determination of relative water level risk based on the relative relationship between the lower safety limit water level of the deaerator and the actual water level of the deaerator includes: The ratio of the lower safety limit water level of the deaerator to the actual water level of the deaerator is used as the relative water level risk.

5. The dynamic power regulation and control method for thermal power units based on AGC according to claim 1, characterized in that, The method integrates the relative power error between the current sampling time and historical sampling times, the main steam pressure trend, and the turbine control valve opening, combined with the relative water level risk at the current sampling time, to determine the throttling exit condition indicators for the current sampling time, including: Normalization is performed on the relative power error between the current sampling time and the historical sampling time within the preset backtracking time window to obtain the normalized error characteristics between the current sampling time and the historical sampling time. Normalization processing is performed on the main steam pressure trends at the current sampling time and the historical sampling time within the preset backtracking time window to obtain the normalized pressure characteristics at the current sampling time and the historical sampling time. Normalization processing is performed on the turbine valve opening at the current sampling time and the historical sampling time within the preset backtracking time window to obtain the normalized opening characteristics at the current sampling time and the historical sampling time. Based on the normalized error characteristics, normalized pressure characteristics, and normalized opening characteristics of the current sampling time and the historical sampling time within the preset backtracking time window, a multi-dimensional feature matrix is ​​constructed. Based on the state transition law of the multidimensional feature matrix, determine the characteristic value of the steam change trend at the current sampling time; By integrating the relative power error, main steam pressure trend, relative water level risk, and steam change trend characteristic value at the current sampling time, the throttling exit condition index at the current sampling time is obtained.

6. The dynamic power regulation and control method for thermal power units based on AGC according to claim 5, characterized in that, The multi-dimensional feature matrix is ​​constructed based on the normalized error features, normalized pressure features, and normalized opening features of the current sampling time and historical sampling time within a preset backtracking time window, including: The normalized error features of the current sampling time and the historical sampling time are arranged in chronological order to construct an error row vector; The normalized pressure features of the current sampling time and the historical sampling time are arranged in chronological order to construct a pressure row vector; Construct an opening row vector by arranging the normalized opening features of the current sampling time and the historical sampling time in chronological order; A multidimensional feature matrix is ​​constructed based on the error row vector, pressure row vector, and opening row vector.

7. The dynamic power regulation and control method for thermal power units based on AGC according to claim 5, characterized in that, The step of determining the steam change trend characteristic value at the current sampling time based on the state transition law of the multidimensional feature matrix includes: The multidimensional feature matrix is ​​processed using a dynamic mode decomposition algorithm to obtain continuous-time modal eigenvalues; The maximum value among the continuous time modal eigenvalues ​​is taken as the steam change trend eigenvalue at the current sampling time.

8. The dynamic power regulation and control method for thermal power units based on AGC according to claim 1, characterized in that, The dynamic evolution characteristics of the throttling exit condition indicators based on the current sampling time and historical sampling times are used to determine the measures for issuing regulation control commands, including: The difference between the throttling exit condition index at the current sampling time and the throttling exit condition index at the previous adjacent historical sampling time is used as the gradient of the throttling exit condition index at the current sampling time. If the gradients of the throttling exit condition indexes of the preset number of historical sampling times adjacent to the current sampling time are all less than or equal to 0, and the gradient of the throttling exit condition indexes of the current sampling time is greater than 0, then an adjustment control command is issued; otherwise, no adjustment control command is issued.

9. A dynamic power regulation and control system for thermal power units based on AGC, characterized in that, The system includes: The data acquisition module is used to obtain the actual active power of the generator, the target power of automatic power generation control, the main steam pressure in front of the turbine, the turbine control valve opening, and the actual water level of the deaerator at the current sampling time. The basic feature analysis module is used to obtain the relative power error based on the difference between the automatic power generation control target power and the actual active power of the generator, combined with the rated active power of the generator; to obtain the main steam pressure trend at the current sampling time based on the difference between the main steam pressure at the current sampling time and the previous adjacent sampling time; and to determine the relative water level risk based on the relative relationship between the lower safety limit water level of the deaerator and the actual water level of the deaerator. The throttling exit condition index analysis module is used to integrate the relative power error, main steam pressure trend and turbine control valve opening at the current sampling time and the historical sampling time, and combine them with the relative water level risk at the current sampling time to determine the throttling exit condition index at the current sampling time. The results analysis module is used to determine the measures for issuing regulation and control commands based on the dynamic evolution characteristics of the throttling exit condition indicators at the current sampling time and the historical sampling time. The adjustment control command includes: Map the current steam pressure trend at the sampling moment to the velocity increment of the condensate regulating valve; The speed increment is added to the preset reference safety speed to obtain the opening speed of the condensate regulating valve; The opening speed is used as an adjustment control command.

Images