CN-122006439-A - Intelligent control method and system for wet desulfurization of thermal power plant

CN122006439ACN 122006439 ACN122006439 ACN 122006439ACN-122006439-A

Abstract

An intelligent control method and system for wet desulfurization of a thermal power plant relate to the technical field of wet desulfurization of thermal power plants and are used for improving the accuracy of limestone slurry supply. The method comprises the steps of intercepting historical data from a real-time operation data sequence of a desulfurization system to construct a rolling estimation window, establishing a double-process decoupling mechanism model comprising a gas-liquid mass transfer fast dynamic sub-process and a limestone dissolution slow dynamic sub-process, calculating a slurry chemical buffer index in the rolling estimation window, assigning a dynamic lag time in the limestone dissolution slow dynamic sub-process based on a preset nonlinear mapping relation and the slurry chemical buffer index to obtain a correction lag time, substituting the correction lag time as a known parameter into the double-process decoupling mechanism model in the rolling estimation window, solving an optimized gas-liquid mass transfer coefficient and an optimized limestone activity coefficient, updating the double-process decoupling mechanism model according to the optimization coefficient and the correction lag time, and calculating a future limestone slurry supply flow instruction based on the model.

Inventors

WANG HONGWEI
JIAO YUMING
SUI XIAOHUA
WANG DESHENG
NIU LEI

Assignees

国电内蒙古东胜热电有限公司

Dates

Publication Date: 20260512
Application Date: 20251206

Claims (10)

1. An intelligent control method for wet desulfurization of a thermal power plant, which is characterized by being applied to a control system, the method comprising: Intercepting historical data from a real-time operation data sequence of the desulfurization system to construct a rolling estimation window; Establishing a double-process decoupling mechanism model comprising a gas-liquid mass transfer fast dynamic sub-process and a limestone dissolution slow dynamic sub-process, wherein the gas-liquid mass transfer fast dynamic sub-process comprises a gas-liquid mass transfer coefficient used for representing gas-liquid contact mass transfer efficiency, and the limestone dissolution slow dynamic sub-process comprises a limestone activity coefficient used for representing limestone raw material reactivity and a variable dynamic lag time; Calculating a slurry chemical buffer index in the rolling estimation window based on the nonlinear sensitivity characteristic of the pH value of the slurry to the concentration of the reactant, wherein the slurry chemical buffer index represents the inertial magnitude of the slurry for resisting the pH value change; Assigning the dynamic lag time in the slow dynamic sub-process of limestone dissolution based on a preset nonlinear mapping relation and the slurry chemical buffer index to obtain a corrected lag time; Substituting the correction lag time as a known parameter into the double-process decoupling mechanism model in the rolling estimation window, and solving an optimized gas-liquid mass transfer coefficient and an optimized limestone activity coefficient which minimize the deviation between the output track and the historical data track of the double-process decoupling mechanism model by taking the gas-liquid mass transfer coefficient and the limestone activity coefficient as optimization variables; updating the double-process decoupling mechanism model according to the optimized gas-liquid mass transfer coefficient, the optimized limestone activity coefficient and the correction lag time; and calculating a future limestone slurry supply flow instruction based on the updated double-process decoupling mechanism model.
2. The method according to claim 1, wherein the step of calculating a slurry chemical buffer index within the rolling estimation window based on the nonlinear sensitivity characteristic of slurry pH to reactant concentration, specifically comprises: Constructing an acid-base buffer characteristic curve function of the pH value of the slurry along with the addition amount change of the added acid-base reagent based on the ionization balance relation of the carbonate and sulfite system; Substituting the pH value of the real-time slurry in the rolling estimation window into the acid-base buffer characteristic curve function, and calculating the reciprocal of the tangential slope of the acid-base buffer characteristic curve function at the pH value point at each moment; and normalizing the reciprocal to obtain the slurry chemical buffer index.
3. The method according to claim 2, wherein the step of constructing an acid-base buffer characteristic curve function of the slurry pH value according to the addition amount of the added acid-base reagent based on the ionization balance relation of the carbonate and sulfite systems comprises the following steps: constructing the acid-base buffer characteristic curve function based on a buffer capacity definition formula, wherein the acid-base buffer characteristic curve function is configured to represent the ratio relation between the differential variation of the added amount of the added acid-base reagent and the differential variation of the pH value causing the variation; The acid-base buffer characteristic curve function comprises a mathematical expression taking the hydrogen ion concentration as an independent variable, and the mathematical expression comprises a bisulfide ionization balance constant term and a bicarbonate ionization balance constant term as denominator parameters; When the concentration of hydrogen ions corresponding to the pH value of the slurry approaches to the minimum value, the chemical buffer index of the slurry is in a preset peak value interval, and the slurry is characterized as being in a strong buffer state.
4. The method according to claim 1, wherein the step of assigning the dynamic lag time in the slow dynamic sub-process of limestone dissolution based on a preset nonlinear mapping relationship and the slurry chemical buffer index to obtain a corrected lag time comprises: Taking the ratio of the liquid level volume of the slurry tank of the absorption tower to the flow of the slurry circulating pump as a physical mixing hysteresis reference value; substituting the slurry chemical buffer index as an input variable into the preset nonlinear mapping relation, and calculating to obtain a chemical damping coefficient; And determining the multiplication value of the physical mixing hysteresis reference value and the chemical damping coefficient as the correction hysteresis time.
5. The method of claim 4, wherein the preset nonlinear mapping relationship specifically comprises: When the chemical buffer index of the slurry is increased, the chemical damping coefficient obtained according to the preset nonlinear mapping relation is in monotonic nonlinear increment, and when the chemical buffer index of the slurry is in a preset peak value interval, the chemical damping coefficient reaches a maximum threshold value.
6. The method of claim 5, wherein after the step of determining the multiplication value of the physical mixture hysteresis reference value and the chemical damping coefficient as the corrected hysteresis time, the method further comprises: In the rolling estimation window, calculating a cross-correlation function of the actual measured slurry pH value track relative to the historical limestone slurry supply flow track; Extracting an actual measurement lag time based on a peak position of the cross-correlation function; calculating a time deviation between the corrected lag time and the actual measured lag time; When the actual measurement lag time is larger than the correction lag time and the time deviation is larger than a preset deviation range, increasing the growth rate adjusting parameter in the preset nonlinear mapping relation according to a preset step length so as to improve the output chemical damping coefficient; And when the actual measurement lag time is smaller than the correction lag time and the time deviation is larger than the preset deviation range, reducing the growth rate adjusting parameter in the preset nonlinear mapping relation according to a preset step length.
7. The method according to claim 1, wherein the step of calculating a future limestone slurry feed flow command based on the updated dual process decoupling mechanism model specifically comprises: constructing a prediction model containing the corrected lag time based on the updated double-process decoupling mechanism model; Constructing an objective function of model predictive control based on environmental emission constraint conditions and economic operation cost indexes in a future preset period, wherein the objective function at least comprises a first performance index item for representing control target deviation and a second performance index item for representing limestone slurry supply flow variation amplitude, and the second performance index item is configured with a variable smoothing weight coefficient; acquiring the current slurry chemical buffer index at the optimal solving time of each rolling control period; If the current slurry chemical buffer index is in a preset non-high value interval, setting the smoothing weight coefficient to be a preset reference weight value; if the current slurry chemical buffer index is in a preset high value interval, calculating the difference value between the slurry chemical buffer index and the lower limit value of the high value interval; Calculating a weight increment based on the difference value and a preset monotonically increasing function; And determining the sum of the reference weight value and the weight increment as the current smoothing weight coefficient.
8. An intelligent control system comprising one or more processors and memory coupled to the one or more processors, the memory to store computer program code comprising computer instructions that the one or more processors invoke to cause the intelligent control system to perform the method of any of claims 1-7.
9. A computer program product containing instructions which, when run on an intelligent control system, cause the intelligent control system to perform the method of any of claims 1-7.
10. A computer readable storage medium comprising instructions which, when run on an intelligent control system, cause the intelligent control system to perform the method of any of claims 1-7.

Description

Intelligent control method and system for wet desulfurization of thermal power plant Technical Field The application relates to the technical field of wet desulfurization of thermal power plants, in particular to an intelligent control method and system for wet desulfurization of a thermal power plant. Background With the increasing strictness of environmental regulations, limestone-gypsum wet flue gas desulfurization technology is generally adopted by coal-fired power plants to reduce sulfur dioxide emission in flue gas so as to meet emission standards. In the actual operation of the desulfurization system, in order to achieve both the environmental protection standard and the economical efficiency, the flow rate of limestone slurry entering the absorption tower needs to be precisely controlled. The ideal control target is to stably maintain the pH value of slurry in the absorption tower and the concentration of sulfur dioxide at the outlet in an optimal range while adapting to the fluctuation of coal quality and the load change of a unit, thereby reducing the waste of limestone raw materials and the power consumption of equipment. In order to solve the problems of multivariable coupling and large inertia hysteresis involved in the desulfurization process, a control system of the related art generally adopts a multi-scale dynamic optimization scheme based on model predictive control. The specific implementation mode of the scheme is that the control system respectively establishes a fast dynamic submodel for describing the absorption of the gas-phase sulfur dioxide and a slow dynamic submodel (commonly called as a physical mixing model) based on the fluid dynamics mechanism of the slurry pool by taking the characteristics of fast gas-phase transmission and slow liquid-phase reaction in the desulfurization reaction into consideration. Aiming at the slow dynamic process, the related technology sets a linearly-changing physical lag time constant in the physical mixing model according to physical parameters such as slurry tank volume, fluid flow rate, mixing power of a stirrer and the like, and introduces a Smith predictor or a time lag compensation module in a controller. In the running process, the controller predicts the response track of the pH value of the slurry to the slurry supply flow according to the physical mixing model, and corrects the control instruction in advance by utilizing the hysteresis compensation module so as to overcome the response delay caused by large volume and realize the advanced adjustment of the slurry supply. However, the desulfurization slurry is a complex carbonate-sulfite chemical buffer system, the response resistance (i.e. buffer capacity) of the desulfurization slurry to the pH value change does not keep linear change as described in a physical mixing model of the related art, but rather shows severe nonlinear fluctuation (such as a strong buffer platform region between pH 5.0 and 6.0) along with the difference of the current pH value, when the slurry is in the strong buffer platform region, the response of the pH value to the slurry supply amount can become sluggish, and a chemical pseudo-lag is shown, at this time, the linear compensation model of the related art can judge the chemical buffer phenomenon as insufficient slurry supply amount or incomplete physical mixing, and further continuously accumulate and output excessive slurry supply instructions, so that the slurry supply instruction of the limestone slurry is inaccurate. Disclosure of Invention The application provides an intelligent control method and system for wet desulfurization of a thermal power plant, which are used for improving the accuracy of limestone slurry supply. According to the method, historical data are intercepted from a real-time operation data sequence of a desulfurization system to construct a rolling estimation window, a double-process decoupling mechanism model comprising a gas-liquid mass transfer fast dynamic sub-process and a limestone dissolution slow dynamic sub-process is established, the gas-liquid mass transfer fast dynamic sub-process comprises a gas-liquid mass transfer coefficient used for representing gas-liquid contact mass transfer efficiency, the limestone dissolution slow dynamic sub-process comprises a limestone activity coefficient used for representing limestone raw material reaction activity and a variable dynamic lag time, a slurry chemical buffer index in a rolling estimation window is calculated based on the nonlinear sensitivity characteristic of slurry pH value to reactant concentration, the slurry chemical buffer index represents the inertia of slurry resistance to pH value change, a correction lag time is obtained by assigning the correction lag time to the limestone dissolution slow dynamic sub-process based on a preset nonlinear mapping relation and the slurry chemical buffer index, the correction lag time is substituted into the double-process decoupling mec