Search

CN-121974421-A - Dynamic regulation and control method and system for fusing multispectral monitoring time sequence characteristics

CN121974421ACN 121974421 ACN121974421 ACN 121974421ACN-121974421-A

Abstract

The invention relates to the technical field of energy and environmental protection, and particularly discloses a dynamic regulation and control method and a system for fusing multispectral monitoring time sequence characteristics, wherein a characteristic migration model based on contrast learning is constructed by continuously collecting visible light sequence images of slag falling from a furnace bottom, infrared temperature field dynamic distribution and 3D material level change data, slag falling thermal state reference characteristics under different load working conditions are extracted, a coupling time sequence prediction model of sensible heat of hot slag and evaporation of waste water is established by combining dynamic response characteristics of atomization evaporation of waste water in a cloud falling bed, and a real-time regulation and control strategy is generated by adopting a deep reinforcement learning algorithm; the invention breaks through the limitation of the traditional static parameter setting, and solves the contradiction between insufficient heat utilization and equipment scaling in the drying of the high-salt wastewater by adapting to the variable load working condition of the unit through the transfer learning of multispectral time sequence characteristics, thereby improving the evaporation efficiency of the desulfurization wastewater.

Inventors

  • MENG PENGQIANG
  • HUANG WENJING
  • LI FANG

Assignees

  • 南京通用电气装备有限公司

Dates

Publication Date
20260505
Application Date
20260120

Claims (10)

  1. 1.A dynamic regulation and control method for fusing multispectral monitoring time sequence features is characterized by comprising the following steps: Dividing monitoring area blocks of a slag falling bin, continuously collecting multispectral characteristic data of slag falling in each monitoring area block by using monitoring equipment, acquiring evolution characteristic data of atomization and evaporation of wastewater in a cloud falling bed, and preprocessing the multispectral characteristic data and the evolution characteristic data; constructing a feature migration model based on contrast learning according to multispectral characteristic data and evolution characteristic data, and extracting slag falling thermal state reference features and evaporation dynamic response features under different load working conditions; And a characteristic coupling time sequence prediction model is established through the slag falling thermal state reference characteristic and the evaporation dynamic response characteristic to predict thermal state change factors, a real-time regulation strategy is generated based on the thermal state change factors combined with a deep reinforcement learning algorithm, and the desulfurization wastewater injection parameters and the hot air regulation parameters are dynamically regulated.
  2. 2. The method for dynamically adjusting and controlling the fusion of the multispectral monitoring time sequence features according to claim 1, wherein the multispectral characteristic data comprise visible light sequence images, infrared temperature field distribution data and 3D material level change data, and the evolution characteristic data comprise droplet size distribution change data.
  3. 3. The dynamic regulation and control method integrating multispectral monitoring time sequence features according to claim 1, wherein a feature migration model for comparison learning is constructed according to multispectral feature data and evolution feature data, and slag falling thermal state reference features and evaporation dynamic response features under different load working conditions are extracted, and the specific steps are as follows: the extraction position is Is a monitoring area block of (1) Carrying out space-time registration and geometric calibration processing on the multispectral characteristic data and the evolution characteristic data to obtain first multispectral characteristic data and first evolution characteristic data; and obtaining a slag falling thermal state reference characteristic by carrying out fusion analysis on the first multispectral characteristic data of each monitoring area block, and obtaining an evaporation dynamic response characteristic based on the first evolution characteristic data analysis.
  4. 4. The method for dynamically adjusting and controlling the fused multispectral monitoring time sequence features according to claim 3, wherein the method for obtaining the slag falling thermal state reference features by carrying out fusion analysis on the first multispectral characteristic data of each monitoring area block comprises the following specific steps: Extracting visible light sequence image of monitoring area block Acquiring a motion trail of the falling slag, and capturing the motion speed of the falling slag based on the motion trail , wherein, To monitor regional blocks The jth slag falling position at the time t+1, To monitor regional blocks The j-th slag falling position at the time t, Monitoring slag flow accumulation uniformity according to the slag falling movement speed at time intervals; acquiring infrared temperature field distribution data of a monitoring area block, constructing a temperature distribution monitoring model, and calculating a temperature distribution coefficient of the monitoring area block; And acquiring 3D material level change data of the monitoring area block, and comprehensively obtaining slag falling thermal state reference characteristics of the monitoring area block based on slag flow accumulation uniformity, temperature distribution coefficients and 3D material level change data, wherein the slag flow accumulation uniformity of the monitoring area block is the temperature distribution coefficients of the monitoring area block, and the 3D material level change data of the monitoring area block is obtained.
  5. 5. The method for dynamically adjusting and controlling a multi-spectrum monitoring time sequence feature according to claim 3, wherein the evaporating dynamic response feature is obtained based on the analysis of the first evolution characteristic data, and the method comprises the following specific steps: Based on the division of monitoring area blocks of the slag falling bin, monitoring areas divided in a cloud falling bed plane are obtained, evolution characteristic data of each monitoring area are obtained, and drop particle size distribution change data in the evolution characteristic data are extracted; acquiring the droplet size distribution change data of each monitoring area in the monitoring time period, constructing a droplet distribution change evaluation model, calculating evaporation dynamic response characteristics, and evaluating whether the atomization efficiency of the wastewater is normal according to the evaporation dynamic response characteristics.
  6. 6. The dynamic regulation and control method integrating multispectral monitoring time sequence features according to claim 5 is characterized by evaluating whether the atomization efficiency of the wastewater is normal according to the evaporation dynamic response features, wherein the specific steps are that the evaporation dynamic response features are compared with preset atomization threshold values, and if the evaporation dynamic response features are larger than or equal to the preset atomization threshold values, the atomization efficiency of the wastewater is abnormal at the moment; if the evaporation dynamic response characteristic is smaller than a preset atomization threshold, the wastewater atomization efficiency is normal at the moment.
  7. 7. The method for dynamically regulating and controlling the fusion of the multispectral monitoring time sequence features according to claim 1, wherein the method for predicting the thermal state change factor by establishing a feature coupling time sequence prediction model through the slag falling thermal state reference features and the evaporation dynamic response features comprises the following specific steps: Constructing a first characteristic sequence by extracting a slag falling thermal state reference characteristic in a monitoring period, and constructing a second characteristic sequence by extracting an evaporation dynamic response characteristic The first characteristic sequence comprises a slag flow accumulation uniformity sequence Temperature distribution coefficient sequence 3D fill level change data sequence ; And constructing a characteristic coupling time sequence prediction model based on the first characteristic sequence and the second characteristic sequence, acquiring a thermal state change factor, and evaluating the matching degree of heat exchange between the falling of the hot slag and the atomization of the wastewater.
  8. 8. The method for dynamically adjusting and controlling a fused multispectral monitoring time sequence according to claim 7, wherein the slag flow stacking uniformity sequence is as follows , wherein, Monitoring area block for t moment Is characterized by the slag flow accumulation uniformity, Monitoring area block for T moment Slag flow accumulation uniformity of (C) and temperature distribution coefficient sequence of (C) , wherein, Monitoring area block for t moment Is used for the temperature distribution coefficient of the (c), Monitoring area block for T moment The 3D material level change data sequence is , Monitoring area block for t moment Is provided with 3D level change data of (c), Monitoring area block for T moment 3D fill level change data of (2); The second characteristic sequence is , wherein, Monitoring the area for time t Is characterized by an evaporation dynamic response of (a), Monitoring the area for time T Is a dynamic response characteristic of evaporation.
  9. 9. The dynamic regulation and control method integrating the multispectral monitoring time sequence characteristics according to claim 1 is characterized by obtaining a thermal state change factor and evaluating the matching degree of heat exchange between the falling of the thermal slag and the atomization of the wastewater, wherein the thermal state change factor is compared with a preset threshold value, if the thermal state change factor is larger than or equal to the preset threshold value, the matching degree of heat exchange between the falling of the thermal slag and the atomization of the wastewater is qualified, and if the thermal state change factor is smaller than the preset threshold value, the matching degree of heat exchange between the falling of the thermal slag and the atomization of the wastewater is not qualified.
  10. 10. A dynamic regulation and control system for fusing multispectral monitoring time sequence features, which is applied to a dynamic regulation and control method for fusing multispectral monitoring time sequence features according to any one of claims 1-9, and is characterized in that the system comprises a data acquisition module, a feature extraction module and a dynamic regulation and control module; The data acquisition module is used for continuously acquiring multispectral characteristic data of slag falling in each monitoring area block by dividing the monitoring area block of the slag falling bin, acquiring evolution characteristic data of wastewater atomization evaporation in the cloud falling bed, and preprocessing the multispectral characteristic data and the evolution characteristic data; The feature extraction module is used for constructing a feature migration model based on contrast learning according to the multispectral characteristic data and the evolution characteristic data, and extracting slag falling thermal state reference features and evaporation dynamic response features under different load working conditions; The dynamic regulation and control module is used for establishing a characteristic coupling time sequence prediction model to predict the thermal state change factor through the slag falling thermal state reference characteristic and the evaporation dynamic response characteristic, generating a real-time regulation and control strategy based on the thermal state change factor combined with a deep reinforcement learning algorithm, and dynamically regulating and controlling the desulfurization wastewater injection parameter and the hot air regulation and control parameter.

Description

Dynamic regulation and control method and system for fusing multispectral monitoring time sequence characteristics Technical Field The invention relates to the technical field of energy and environmental protection, in particular to a dynamic regulation and control method and system integrating multispectral monitoring time sequence characteristics. Background Along with the continuous promotion of the scale development and environmental protection requirement of thermal power generating units, the pulverized coal boiler generates a large amount of high-temperature furnace bottom slag and high-salinity high-hardness desulfurization wastewater in the long-term operation process. The bottom slag is usually discharged at a high temperature of 900 ℃ or above, contains a large amount of sensible heat, and meanwhile, the desulfurization wastewater contains high-concentration chloride ions, sulfate radicals and heavy metal ions, and severely pollutes the environment if being directly discharged. Therefore, how to couple and utilize the high-grade sensible heat carried by the furnace bottom slag and the desulfurization wastewater drying process to realize 'heat pollution control', not only can improve the energy utilization efficiency, but also can effectively solve the wastewater treatment problem, and has important engineering application value and significance of energy conservation and emission reduction. The existing system usually depends on a single sensor such as a thermocouple or a flowmeter, and the like, has limited acquired information, and is difficult to accurately capture the furnace bottom slag temperature field evolution, the wastewater mist evaporation rate and the dynamic change of a three-phase coupling area. In addition, the traditional method mostly adopts static threshold adjustment or rule control based on manual experience, and cannot cope with fluctuation of the slag discharge load of the furnace bottom, water quality and water quantity change of desulfurization wastewater and uncertainty of spraying working conditions in real time. Under the rapid change of load or extreme working condition, the problems of excessive spraying, insufficient evaporation or local overheating and the like often occur, so that the zero discharge effect of the wastewater is affected, and the energy efficiency of the system is reduced. Therefore, it is necessary to provide a dynamic regulation and control method and system for fusing multispectral monitoring time sequence features to solve the above technical problems, and in order to solve the above problems, a technical scheme is provided. Disclosure of Invention In order to overcome the defects in the prior art, the invention provides a dynamic regulation and control method and a system for fusing multispectral monitoring time sequence characteristics, which are used for solving the problems. In order to achieve the above purpose, the present invention provides the following technical solutions: A dynamic regulation and control method for fusing multispectral monitoring time sequence features comprises the following steps: Dividing monitoring area blocks of a slag falling bin, continuously collecting multispectral characteristic data of slag falling in each monitoring area block by using monitoring equipment, acquiring evolution characteristic data of atomization and evaporation of wastewater in a cloud falling bed, and preprocessing the multispectral characteristic data and the evolution characteristic data; constructing a feature migration model based on contrast learning according to multispectral characteristic data and evolution characteristic data, and extracting slag falling thermal state reference features and evaporation dynamic response features under different load working conditions; And a characteristic coupling time sequence prediction model is established through the slag falling thermal state reference characteristic and the evaporation dynamic response characteristic to predict thermal state change factors, a real-time regulation strategy is generated based on the thermal state change factors combined with a deep reinforcement learning algorithm, and the desulfurization wastewater injection parameters and the hot air regulation parameters are dynamically regulated. As a further scheme of the invention, the multispectral characteristic data comprise visible light sequence images, infrared temperature field distribution data and 3D material level change data, and the evolution characteristic data comprise droplet size distribution change data. As a further scheme of the invention, a characteristic migration model for comparison learning is constructed according to multispectral characteristic data and evolution characteristic data, and the slag falling thermal state reference characteristic and the evaporation dynamic response characteristic under different load working conditions are extracted, and the specific steps are as follows: the extrac