A wide load partition regulation system and method for a tubular air preheater

Abstract

The application discloses a wide-load partition regulation system and method for a tubular air preheater, and relates to the technical field of high-efficiency energy-saving industry.The model prediction control multi-target partition closed-loop control subsystem takes the minimization of the whole machine exhaust gas temperature as an optimization target, takes the fact that the cold end pipe wall temperature of each pipe box is not lower than the safety reference temperature dynamically output by the partition ammonium bisulfate phase change early warning model subsystem as a rolling constraint, and takes the opening degree of each partition air inlet regulating baffle and the opening degree of the hot air recirculation target regulating valve as the manipulated variables, so that a constrained quadratic programming optimization problem is solved on-line and the optimal control amount of each partition is output.The application realizes adaptive weight multi-target closed-loop optimization control of each partition through a model prediction control algorithm, dynamically coordinates the safety protection priority of each partition and the heat exchange efficiency of the whole machine, reduces the air leakage rate by more than 1.5%, and reduces the power consumption of the air supply and exhaust fan.

Description

Technical Field

[0001] This invention relates to the field of high-efficiency and energy-saving industrial technology, specifically to a wide-load zone regulation system and method for a tubular air preheater. Background Technology

[0002] Tubular air preheaters (hereinafter referred to as tubular air preheaters) are important waste heat recovery and heat exchange equipment in the tail flue of boilers in thermal power plants. They use metal tube bundles as heat transfer elements. High-temperature flue gas flows inside the tube bundle, and combustion air is laterally swept outside the tube bundle to complete heat exchange. The heat exchange effect of tubular air preheaters has a significant impact on the unit's thermal efficiency. Typically, for every 10°C reduction in flue gas temperature, the boiler's thermal efficiency can be increased by about 0.5% to 0.6%.

[0003] Many in-service thermal power units are required to frequently participate in deep peak shaving, operating for extended periods at 30% or even 20% below their rated output. Under this deep peak shaving condition, the flue gas mass flow rate is significantly reduced, the overall flue gas temperature decreases, and the outlet flue gas temperature of the selective catalytic reduction (SCR) denitrification system can drop to below 300°C. Consequently, the air preheater tube wall temperature drops significantly. At the same time, the spatial non-uniformity of combustion within the boiler furnace intensifies as the load decreases. The flue gas temperature difference between the edge tube box zone and the central tube box zone can reach 40°C to 80°C, and the ammonia escape concentration difference can reach more than 5 times. Frequent load increases and decreases also lead to different rates of heat change in each tube box, resulting in significant thermal expansion differences between tube boxes. This causes thermal fatigue cracking at the fillet welds of the tube sheet, and the air leakage rate rises rapidly.

[0004] During the SCR denitrification process, unreacted ammonia (NH3) flows into the air preheater with the flue gas. Within a temperature window of approximately 147°C to 230°C, it reacts with sulfur trioxide (SO3) and water vapor in the flue gas to generate liquid high-viscosity ammonium bisulfate (ABS, NH4HSO4). The liquid ABS adheres to the tube bundle surface, trapping fly ash and forming hard ash deposits. This causes blockage of the tube bundle flow section, resulting in a rapid increase in the pressure difference between the flue gas side and the air side of the air preheater. When the flue gas temperature further drops below the acid dew point, SO3 and water vapor condense into liquid sulfuric acid, causing rapid electrochemical corrosion of the tube bundle and significantly shortening its service life.

[0005] The main technical solution currently used in the project to address the above problems is a total hot air recirculation system: a total recirculation fan and a total recirculation regulating valve are installed on the total hot air outlet duct of the air preheater to draw part of the outlet hot air back to the total inlet cold air duct, thereby raising the overall inlet air temperature of the air preheater; the arithmetic mean of the temperatures at a small number of measuring points at the total flue gas outlet or total air outlet of the air preheater is used as the reference value for the overall unit temperature, and the opening of the recirculation valve is adjusted uniformly after being compared with a conservatively set unified safety lower limit temperature.

[0006] The aforementioned existing technical solutions have the following fundamental defects: Firstly, the adjustment granularity is a centralized control of the entire air preheater. A single adjustment mechanism cannot distinguish the differentiated thermal states of each tube box zone. The safety lower limit must be set according to the local tube box area with the lowest temperature and highest risk along the furnace width section. This results in the inlet air temperature of all tube box zones being forcibly raised to above the conservative safety lower limit. Under deep peak shaving conditions, the overall exhaust gas temperature of the unit is forced to rise by 15°C to 30°C, resulting in huge energy waste.

[0007] Secondly, centralized control cannot provide targeted protection for specific low-temperature, high-ammonia escape pipe boxes. Even if the overall average temperature meets safety requirements, specific "cold corner" pipe boxes may continue to fall below the ABS dew point or acid dew point, causing local pipe boxes to continuously experience ABS condensation and electrochemical corrosion. This is often not discovered until the system resistance increases significantly or the pipe bundle leaks, resulting in irreversible equipment damage and unplanned shutdowns.

[0008] Third, the existing scheme only makes a single parameter judgment based on the average temperature of a small number of point temperature measurement data. It fails to integrate the combined effects of multiple parameters such as ammonia escape concentration and SO3 concentration at the same spatial location on the critical temperature of ABS condensation, resulting in a large number of false alarms (unnecessary hot air recirculation, wasting energy) and false alarms (failure to provide protection when it should have, leading to corrosion).

[0009] Fourth, the overall hot air recirculation system uniformly raises the inlet air temperature of each tube box, and the air volume distribution of each tube box cannot be independently controlled. During rapid load increases and decreases, the rate of heat change of each tube box is different, resulting in thermal expansion differences between tube boxes. This causes thermal fatigue cracks at the fillet weld of the tube sheet to rapidly initiate and expand. Tube sheet cracking leads to a continuous increase in air leakage rate, which can exceed 5% in severe cases, and the power consumption of the induced draft fan and the forced draft fan increases significantly. Summary of the Invention

[0010] The technical problem this invention aims to solve is to address the fundamental deficiency of existing centralized hot air recirculation regulation methods, which cannot precisely intervene in the differentiated thermal states of each tube box zone in a tubular air preheater. This invention provides a wide-load zone regulation system and method for a tubular air preheater, enabling precise and independent regulation of the ventilation volume and inlet air temperature of each tube box zone with dual degrees of freedom. This eliminates coupled faults such as localized ABS condensation and ash blockage, low-temperature corrosion, and thermal stress cracking and air leakage in the tube sheet, while simultaneously reducing the overall exhaust gas temperature and improving boiler heat exchange efficiency.

[0011] A wide-load zoning regulation system for a tubular air preheater includes a tube box zoning actuator subsystem, a multi-dimensional measurement and sensing subsystem, a zoning ammonium bisulfate phase change early warning model subsystem, a model predictive control multi-objective zoning closed-loop control subsystem, and a distributed control system and advanced process control integrated communication subsystem.

[0012] The tube box partition actuator subsystem includes: N independent tube box partitions along the width of the boiler, where N is an integer not less than 2. Each tube box partition is independently equipped with a partition air inlet regulating baffle and a partition hot air recirculation targeted regulating valve. The partition air inlet regulating baffle is installed on the corresponding tube box partition cold air inlet branch pipe and is driven by an electric actuator to independently regulate the air inlet flow of that partition. The partition hot air recirculation targeted regulating valve guides the hot air outlet of the tube box partition back to the cold air inlet end of the same tube box partition through an independent hot air recirculation branch pipe, and after mixing with the cold air, it targets and increases the air inlet temperature of that partition.

[0013] The multidimensional measurement and sensing subsystem includes: an acoustic temperature measurement array arranged at the flue gas inlet section of the air preheater; a tunable diode laser absorption spectroscopy ammonia escape distribution measurement system arranged along the furnace width; a flue gas flow field measurement array at the cross-section of each tube box; a thermocouple array for the tube wall temperature at the cold end of each tube box; and SO3 concentration distribution feedback along the furnace width from the SCR denitrification system.

[0014] The partitioned ammonium bisulfate phase change early warning model subsystem, based on the real-time data of the multi-dimensional measurement and sensing subsystem, independently calculates the ABS dew point temperature and acid dew point temperature for each pipe box partition, compares the calculation results with the measured cold end pipe wall temperature of the partition, and outputs the partition risk level and risk trend prediction.

[0015] The model predictive control multi-objective zoned closed-loop control subsystem takes minimizing the overall exhaust temperature as the optimization objective, takes the cold end pipe wall temperature of each pipe box zone not lower than the safe reference temperature dynamically output by the zoned ammonium bisulfate phase change early warning model subsystem as the rolling constraint, and takes the opening degree of the air inlet regulating damper and the opening degree of the hot air recirculation targeted regulating valve of each zone as the manipulated variables. It solves the constrained quadratic programming optimization problem online and outputs the optimal control quantity of each zone.

[0016] The distributed control system and advanced process control integrated communication subsystem are responsible for real-time data exchange, control command issuance, and execution status feedback confirmation between the subsystems.

[0017] A wide-load zone adjustment method for a tubular air preheater is used to adjust N tube box zones along the width of the boiler, where N is an integer not less than 2. The method includes the following process: multi-dimensional real-time sensing process: reconstructing the two-dimensional temperature field of the flue gas inlet section of the air preheater through an acoustic temperature measurement array to obtain the representative flue gas temperature of each tube box zone; measuring the ammonia escape concentration distribution at the corresponding position of each tube box zone along the boiler width through a tunable diode laser absorption spectroscopy system; and obtaining the SO3 concentration distribution, flue gas velocity, and cold end tube wall temperature of each zone.

[0018] The zoned ammonium bisulfate phase change early warning process is as follows: For each zone, the ABS dew point temperature and acid dew point temperature are independently calculated based on the measured ammonia escape concentration and SO3 concentration of that zone; the safety margin between the measured cold end pipe wall temperature and the two critical temperatures of that zone is calculated; the three-level risk level is assessed based on the comparison between the safety margin and the safety margin threshold; linear regression is performed on the historical time series of the safety margin of each zone to predict the future risk change trend of each zone; and the risk status and trend prediction results of each zone are output to the model prediction controller.

[0019] Model predicts the closed-loop regulation process of each zone: the penalty weight for pipe wall temperature protection and the safe reference temperature of each zone are adaptively updated according to the risk level of each zone; the overall exhaust temperature is minimized as the optimization objective function, and the cold end pipe wall temperature of each zone is not lower than the corresponding safe reference temperature as a constraint. The constrained quadratic programming optimization problem is solved online in a rolling manner; the optimal air inlet regulating damper opening command and the hot air recirculation targeted regulating valve opening command of each zone are sent to the actuators of each zone in parallel, and the execution is confirmed to be in place.

[0020] The beneficial effects of this invention are as follows: 1. This invention achieves precise and independent adjustment of ventilation volume and inlet air temperature of each zone by independently configuring an air intake regulating baffle and a hot air recirculation targeted regulating valve in each pipe box zone. It implements targeted protection only for pipe box zones that are truly at risk, which can reduce the average exhaust gas temperature of the whole unit by 8°C to 15°C and improve the boiler thermal efficiency by about 0.5% to 0.9%.

[0021] 2. This invention combines an acoustic temperature measurement array with a tunable diode laser absorption spectroscopy system to obtain the spatially resolved temperature field and ammonia escape concentration distribution of each tube box zone. It also combines SO3 concentration data to perform multi-parameter coupled zoned ABS phase change risk early warning, which can eliminate ABS deposition dead zones and extend the service life of air preheater tube bundles and SCR catalysts by more than 30%.

[0022] 3. This invention achieves adaptive weighted multi-objective closed-loop optimization control for each zone through a model predictive control algorithm, dynamically coordinating the safety protection priority of each zone with the overall heat exchange efficiency, reducing the overall air leakage rate by more than 1.5%, and consequently reducing the power consumption of the supply and exhaust fans. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the overall three-layer architecture of the tubular air preheater wide-load zone regulation system of the present invention; Figure 2 This is a complete algorithm execution flowchart of the partitioned ammonium bisulfate phase change early warning model subsystem of the present invention; Figure 3 This is a flowchart illustrating the complete algorithm execution of the multi-objective partitioned closed-loop control subsystem of the model predictive control system of this invention. Figure 4 This is a complete full-process control logic diagram of the tubular air preheater wide-load zone regulation system of the present invention. Detailed Implementation

[0024] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0025] System Overall Architecture The tubular air preheater wide-load zoning regulation system proposed in this invention consists of five subsystems working together to form a complete "sensing-early warning-decision-execution-feedback" closed-loop regulation loop: the tube box zoning actuator subsystem (subsystem A) is located at the field execution layer and is responsible for the physical intervention of flow and temperature in each zone; the multi-dimensional measurement sensing subsystem (subsystem B) is located at the measurement sensing layer and collects five physical quantities in parallel with a 1-second sampling period; the zoning ammonium bisulfate phase change early warning model subsystem (subsystem C), the model predictive control multi-objective zoning closed-loop control subsystem (subsystem D), and the distributed control system and advanced process control integrated communication subsystem (subsystem E) together constitute the computing control layer, and the layers exchange data in real time through the industrial Ethernet OPC-UA protocol.

[0026] Subsystem A: Tube Box Zoning Actuator Subsystem: The tubular air preheater is divided into N independent flow tube box zones along the width of the boiler. It is recommended that N be 6 to 12, which is determined according to the physical dimensions of the air preheater, the boiler width, and the accuracy requirements. The boundaries of each zone are naturally formed by the existing physical tube box partitions in the air preheater body, without the need for large-scale internal structural modifications. The modification work is concentrated on adding branch pipe partitions and leading out independent inlet and outlet air branch pipes at the air duct nodes of each zone, and installing actuators on the branch pipes.

[0027] In each pipe box partition (the first) Partitioning, =1,2,...,N) A set of zoned air intake regulating damper assembly is installed on the cold air inlet branch pipe. It consists of parallel blade regulating dampers, an integrated electric actuator (including an absolute position encoder and torque limiting protection), a local control box, and signal cables. The regulating range is 0% to 100% of the full stroke, with a control accuracy of no less than ±0.5% of the full stroke. In the fully closed state, no more than 5% leakage is maintained to prevent localized overheating of the zoned pipe box due to complete ventilation shut-off. The response time is no more than 30 seconds. The actual opening value is fed back to the distributed control system and advanced process control platform in real time via a 4mA to 20mA signal. The opening command for each zone damper is recorded as... The value range is [0.05, 1.0].

[0028] A hot air recirculation branch pipe is led out at the corresponding node of the hot air outlet of each tube box zone. This branch pipe runs around the outside of the air preheater shell and connects to the cold air inlet branch pipe of the same tube box zone (located at the mixing section inlet downstream of the zone's air inlet regulating baffle). The branch pipe is equipped with a butterfly-type hot air recirculation targeted regulating valve (with an electric actuator), an insertion-type flow meter, and a thermal insulation layer. By adjusting the valve opening, the outlet hot air at approximately 280°C to 350°C can be drawn back and mixed with cold air at approximately 20°C to 40°C, targeting and raising the inlet air temperature of that zone by 20°C to 80°C, thereby correspondingly raising the cold end tube wall temperature of that zone. The opening command for each zone's hot air recirculation valve is recorded as follows: The value range is [0, ], It is determined by the pressure drop of the pipeline network.

[0029] The two types of actuators mentioned above work together to achieve precise and independent adjustment of the ventilation volume and inlet air temperature of each pipe box zone with two degrees of freedom. This is the fundamental structural feature that distinguishes this invention from the existing integrated hot air recirculation system (unified adjustment of the entire furnace).

[0030] Subsystem B: Multidimensional Measurement and Sensing Subsystem: An acoustic temperature measurement array is arranged at the flue gas inlet section of the air preheater (at an appropriate cross-sectional location between the SCR catalyst outlet and the economizer outlet), with no fewer than 6×6=36 pairs of transducers evenly distributed on the four walls of the flue. By measuring the one-way transit time of sound waves along each acoustic path, algebraic reconstruction technology is used to perform acoustic CT inversion calculations to reconstruct a two-dimensional flue gas temperature distribution map of the flue section. The propagation speed c of sound waves in the flue gas is related to the flue gas thermodynamic temperature. The relationship is: ;in, The adiabatic index of the flue gas is approximately 1.35. The universal gas constant is taken as 8.314 J / (mol·K); The temperature of the flue gas is in Kelvin (K). The average molar mass of the flue gas, expressed in kg / mol, is calculated based on fuel characteristics and excess air coefficient. The temperature value of each grid node can be inferred by measuring the sound wave transit time along a known distance path. The spatial resolution of the temperature field must match the number of pipe box sections N, ensuring that each section corresponds to at least four effective grid nodes in the temperature field diagram. The average node temperature serves as the representative flue gas temperature value for that section. The acoustic temperature measurement system has a temperature measurement accuracy of no less than ±5℃ and a sampling period of no more than 1 second.

[0031] A tunable diode laser absorption spectroscopy ammonia escape distribution measurement system is used to uniformly arrange no fewer than N laser measurement optical paths along the width of the air preheater. Each optical path traverses the flue gas flow channel of its corresponding tube box section. By detecting the differential absorption of laser light by NH3 molecules at specific near-infrared wavelengths, the volume concentration of NH3 in the flue gas of each section is measured in real time. The unit is ppm, the measurement range is 0 to 30 ppm, the measurement accuracy is not less than ±0.2 ppm, and the response time is not more than 1 second.

[0032] Simultaneously, SO3 concentration sampling and analysis points are arranged in every 2 to 3 zones along the same cross section to obtain SO3 concentration distribution data along the furnace width. The values ​​of each partition are filled by interpolation.

[0033] A multi-point array of Pitot tubes is uniformly arranged along the width of the flue gas inlet and outlet cross-section of the air preheater, and the flue gas velocity at the corresponding cross-sectional position of each tube box section is measured. Each pipe box section is equipped with no less than 2 flow velocity measurement points, with a measurement accuracy of no less than ±2% of the range and a sampling period of no more than 1 second.

[0034] On a representative tube bank of the cold-end tube bundle of the air preheater, at least three armored K-type thermocouples are arranged along the width of each tube box section to measure the outer wall temperature of the cold-end tube bundle in that section. The measurement accuracy is no less than ±2℃, the response time is no more than 5s, and the measured value of the pipe wall temperature is a direct verification quantity of the ABS phase change early warning model, as well as the core feedback quantity of the model predictive controller.

[0035] Subsystem C: Partitioned Ammonium Bisulfate Phase Change Early Warning Model Subsystem: This subsystem executes cyclically with a fixed period of 1 second. The complete algorithm flow is as follows: Figure 2 As shown.

[0036] Data Acquisition and Preprocessing: The system collects data from each sensor in subsystem B at a fixed 1-second interval. Current time of each pipe box partition The raw measurement data includes: the zone-representative flue gas temperature output from the acoustic thermography array. The volume concentration of NH3 in the zone output of a tunable diode laser absorption spectroscopy system (Unit: ppm) SO3 concentration output by the SO3 concentration sampling system for each zone (Unit: mg / Nm) 3 The zoned flue gas velocity output by the flue gas flow field array (Unit: m / s) and the zoned cold end pipe wall temperature output by the pipe wall thermocouple array (Unit: °C) The original collected data were processed sequentially with outlier removal (using the 3σ criterion, replacing data points that deviate from the mean by more than 3 times the standard deviation with the previous valid value) and moving average filtering (time window length is 30s, i.e. 30 sampling points) to obtain the representative parameters of each partition after smoothing. The above smoothing parameters are used as input for subsequent calculations.

[0037] Partial pressure calculation and critical temperature calculation: Read the current total flue gas pressure from the field pressure sensor. (Unit: atm), each pipe box is divided into zones. Calculate the partial pressure of NH3 using the following formula: ;in, For the first The NH3 partial pressure in the pipe box is measured in atm. The first step after moving average filtering Zoned NH3 volume concentration, in ppm; This represents the total pressure of the flue gas, measured in atm.

[0038] Calculate the partial pressure of SO3 using the following formula: ;in, For the first SO3 pressure distribution in the pipe box zone, unit: atm; The first step after moving average filtering SO3 mass concentration in the zone, unit: mg / Nm³ 3 ; The molar mass of SO3 is taken as 80 g / mol; The density of flue gas under standard conditions is expressed in g / Nm³. 3 .

[0039] Divide each pipe box into zones The dew point temperature of ABS was calculated using an improved Menasha engineering calculation formula. (Unit: °C) ;in, For the first The critical temperature at which the ABS in the pipe box section begins to undergo phase change and condensation, in °C; ln(·) is the natural logarithm operator; the coefficients 161.4, 2.0957, 1.5082, and 0.2541 are empirical fitting coefficients based on a large amount of experimental data, which will be corrected and calibrated in engineering applications in combination with the actual sulfur content and ammonia slip characteristics of the coal in this project.

[0040] Divide each pipe box into zones The acid dew point temperature is calculated using the following formula. (Unit: °C) ;in, For the first The critical temperature at which sulfuric acid begins to condense in the pipe box section, in °C; For the first The mass concentration of SO3 in the flue gas of the pipe box zone, in mg / Nm³ 3 ; The volume fraction of water vapor in the flue gas is a dimensionless decimal, determined by the boiler fuel characteristics (basic hydrogen content, moisture content) and excess air coefficient, and is the same for all zones under the same operating conditions; lg(·) is a commonly used logarithmic operator with base 10.

[0041] Safety margin calculation: for each pipe box zone Calculate the measured cold end pipe wall temperature of each zone. Safety margin relative to ABS dew point temperature: ; and the safety margin of the pipe wall temperature in this zone relative to the acid dew point temperature: A positive margin value indicates that the pipe wall temperature in that zone is above the corresponding corrosion critical temperature, and the zone is currently safe. A margin value approaching zero or being negative indicates that there is a risk of ABS deposition or acid corrosion in that zone, and a safety margin threshold should be set. The temperature is 15℃, which can be adjusted according to the actual coal quality and operating experience.

[0042] Level 3 risk assessment: For each pipe box zone... Taking into account both temperature safety margins, the risk level is assessed and assigned a value according to the following rules: If ≥ and ≥ If both conditions are met, the risk level of the partition is determined to be normal (green). =0; if 0≤ < or 0≤ < If either of the two conditions is met, it is considered a warning (yellow). =1; if <0 or If the value is less than 0 (either of the two conditions is met), then an alarm (red) is triggered. =2, summing the risk levels of all partitions into a partition risk state vector. .

[0043] Forecast trend analysis: for each pipe box zone Extracting data from 300 consecutive sampling points (corresponding to a 5-minute time window) in the past. and Historical time series data were fitted using least-squares linear regression to calculate the rate of change of the ABS safety margin over time for each partition. (Unit: °C / min) and the rate of change of acid dew point safety margin over time (Unit: °C / min) When the rate of change is negative, it indicates that the risk of the corresponding zone is deteriorating. Based on this, the trend of risk level change of each zone in the next 10 minutes is predicted and summarized into a trend deterioration rate vector.

[0044] Output warning results: The partition risk state vector at the current moment. Critical corrosion temperature boundary values ​​for each zone The deterioration rate vector is packaged and encapsulated, and sent to the model predictive control multi-objective partition closed-loop controller (subsystem D) through the communication interface of subsystem E. It waits until the next 1-second sampling period arrives before re-execution, so as to realize continuous online dynamic early warning of partition risk.

[0045] Subsystem D: Model Predictive Control Multi-Objective Partitioned Closed-Loop Control Subsystem: This subsystem is based on model predictive control theory and executes cyclically with a control cycle of 10 seconds. The complete algorithm flow is as follows: Figure 3 As shown.

[0046] The definitions of manipulated variable, controlled variable, and disturbance variable are as follows: The manipulated variable is the opening degree of the air inlet regulating damper for each pipe box zone. Hot air recirculation targeted regulating valve opening ( =1,...,N), a total of 2N manipulated variables; the controlled variable is the cold end pipe wall temperature of each pipe box zone. ( =1,...,N) and the total exhaust gas temperature of the whole machine (Calculated by area-weighted average of exhaust gas temperatures in each zone), totaling N+1 controlled variables; the disturbance variable is the unit's current load command. (Measurable disturbances) and flue gas inlet temperatures in each zone (Subsystem B measures in real time as a feedforward disturbance).

[0047] Regarding the first For each pipe box partition, a discrete-time state-space prediction model is established to describe the dynamic response of the cold end pipe wall temperature of the partition under control input and disturbance: ;in, For the first The predicted value of the cold end pipe wall temperature in the zone at the k+1 control step (after 10s), in °C; This is the current discrete control step number; For the first The temperature autoregressive decay coefficient corresponding to the heat capacity of the partitioned pipe wall, with a value range of (0,1), is determined by conducting a step response identification experiment on the partition. The closer the value is to 1, the larger the heat capacity and the slower the response. For the opening of the air intake damper The input gain coefficient for pipe wall temperature reflects the intensity of the influence of changes in airflow on pipe wall heat transfer, and is determined through system identification. Hot air recirculation valve opening The input gain coefficient for pipe wall temperature reflects the intensity of the influence of changes in hot air recirculation on inlet air temperature and thus on pipe wall temperature, and is determined through system identification. flue gas inlet temperature The disturbance gain coefficient for pipe wall temperature is determined through system identification; The current air temperature at the inlet of the main cooling duct is in °C, measured by a field temperature sensor. For the first The hot air temperature at the outlet of the zoned hot air recirculation branch pipe, in °C, is measured by the on-site temperature sensor. The model prediction error compensation term (bias compensation amount) is derived from the first... The difference between the measured pipe wall temperature and the predicted value in the previous step is calculated and updated to eliminate the steady-state prediction bias caused by model mismatch. It is initialized to 0.

[0048] The objective function of the multi-objective optimization problem is to minimize the overall exhaust temperature of the unit while penalizing deviations in the pipe wall temperature of each zone below the safe reference value and large changes in the control quantity. Rolling prediction optimization is performed with a prediction time domain of P=6 steps (corresponding to a 60s prediction window) and a control time domain of M=3 steps (corresponding to a 30s control window). The safe reference temperature for the pipe wall temperature of each zone is... The larger of the ABS dew point temperature and the acid dew point temperature output by subsystem C, plus the safety margin threshold. Calculation determined: ;in, and Subsystem C is in the first... The first step of the control step output Zoned ABS dew point temperature and acid dew point temperature, in °C; Safety margin threshold, unit: °C.

[0049] Pipe wall temperature protection penalty weight for each pipe box zone Based on the current risk level of this zone Dynamic assignment: when When =0 (normal), let =0, the optimizer can freely reduce the pipe wall temperature of this zone to reduce flue gas losses; when When =1 (warning), let = Initiate standard protection penalties to ensure that the pipe wall temperature in that zone does not fall below the safe reference temperature; when When =2 (alarm), let =10× A severe protection penalty is activated, forcing the optimizer to prioritize the protection of the pipe wall temperature in this zone to ensure that the pipe wall temperature in this zone is increased first. Here, represents the nominal protection penalty weight, and represents the design parameter. The constraints of the optimization problem include: the opening degree of the air intake regulating dampers in each zone. The lower limit is 0.05 and the upper limit is 1.0; the opening degree of the hot air recirculation targeted regulating valve in each zone. In 0 to The absolute value of the opening increment of the air intake regulating damper in each zone during a single control cycle shall not exceed 0.10; the absolute value of the opening increment of the hot air recirculation targeted regulating valve in each zone during a single control cycle shall not exceed 0.08; the sum of the openings of the air intake regulating dampers in all zones shall not be less than the average damper opening corresponding to the minimum ventilation requirement of the boiler. The product of this and the number of zones N is used to meet the minimum oxygen control requirements of the boiler.

[0050] In each control cycle Within this subsystem, the following operations are performed sequentially: The latest partition risk state vector and critical temperatures for each partition are read from subsystem C; the current actual unit load, measured cold-end pipe wall temperature for each partition, measured exhaust gas temperature for the entire unit, flue gas inlet temperature for each partition, cold air main pipe temperature, and actual opening degree of each actuator are read from subsystem E (distributed control system platform); the error compensation terms of the prediction model for each partition are updated using the difference between the measured values ​​and the previous predicted values; based on the updated error compensation terms and the current measurable disturbances, multi-step iterative prediction calculations are performed within the control time domain (M=3 steps) and prediction time domain (P=6 steps), expressing the prediction trajectory as an explicit linear function of the control variables; a quadratic programming solver using the active set method (such as qpOASES) is called to complete the solution within a single control cycle, typically... The solution time is no more than 200ms. Based on the principle of "rolling optimization, only the current best is executed in each step", the optimal control quantity at the current moment is extracted and sent to each zone actuator in parallel. If the solution fails (timeout), the instruction of the previous control cycle remains unchanged and an alarm is triggered. After each actuator adjusts to the target opening degree according to the instruction, the actual opening degree value is reported in real time through the position encoder. If an actuator failure is found (the actual opening degree deviates from the instruction for more than 5% for more than 30 seconds), the actuator of that zone is set to manual mode and an alarm is triggered. The other normal zones continue to participate in the model predictive control closed-loop control. When a certain pipe box zone is at the alarm risk level for 3 consecutive control cycles, the soot blowing program of that zone is interlocked and started. After the soot blowing is completed, the model predictive control closed-loop adjustment is automatically restored.

[0051] Prediction model parameters for each region { , , , On a weekly basis, online least squares identification and updates are performed using historical operating data stored in the real-time database. This automatically compensates for long-term model drift caused by decreased catalyst activity, increased ash accumulation in the tube bundles, and changes in coal quality, thus maintaining the long-term effectiveness of the prediction model.

[0052] Subsystem E: Integrated Communication Subsystem for Distributed Control Systems and Advanced Process Control This subsystem mainly consists of four parts: an advanced process control edge computing server (a high-performance computing platform for deploying model predictive control algorithms and ABS early warning models), a fieldbus communication network (using the OPC-UA protocol for bidirectional data exchange with the distributed control system), a real-time database (storing the real-time status and historical operating data of each zone, with historical data retained for more than 180 days), and an operator monitoring interface. The operator interface displays the risk level (green, yellow, and red codes), real-time temperature margin values, ABS dew point and pipe wall temperature comparison curves, and control effect trend graphs in real time in the form of zone heat maps. It supports manual intervention by operators and online adjustment of model predictive control parameters.

[0053] Complete operating logic: The system forms two nested closed loops: the inner closed loop has a 1-second response period. After parallel data acquisition by sensors in subsystem B, subsystem C executes the ABS phase change early warning algorithm and outputs the risk status of each zone to subsystem D; the outer closed loop has a 10-second control period. Subsystem D executes the model predictive control algorithm to generate the optimal control command. The actuators of each zone in subsystem A act independently. After the thermal state response of the pipe box is completed, the measured data is fed back to subsystem B. The complete full-process control logic is as follows: Figure 4 As shown.

[0054] When all zones meet the safe reference temperature, the system maintains normal optimized operation, keeping the overall exhaust temperature at the lowest value under the current operating conditions. When there is a warning or alarm zone, the model predictive controller automatically increases the opening of the hot air recirculation valve in that zone and optimizes the air intake distribution to accurately increase the local pipe wall temperature, while maintaining low exhaust temperature operation in other zones. When a zone is in an alarm state for three consecutive control cycles, the soot blowing program for that zone is interlocked and started. After soot blowing is completed, the model predictive control closed-loop regulation is automatically restored to achieve proactive fault prevention.

[0055] In a simplified scenario where existing units have limitations on the construction conditions for installing independent hot air recirculation branch pipes, local thermal state regulation can be achieved by independently adjusting the redistribution of air intake volume in each pipe box zone, based on retaining only a single actuator of the zone air intake regulating damper. Under the constraint of total air intake volume conservation, more air intake volume is redistributed to high-risk edge pipe box zones. The model predictive control optimization problem is simplified to an air intake volume redistribution problem, with minimal modification work. It can be used as a simplified implementation version of the complete solution of this invention and is suitable for scenarios with limited modification budgets. However, its effect on improving the pipe wall temperature of high-risk zones is weaker than the complete solution with a hot air recirculation targeted valve. For scenarios where the edge zone temperature is severely low under deep peak shaving conditions, the complete solution of this invention is still recommended.

[0056] Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make possible changes and modifications without departing from the spirit and scope of the present invention. Therefore, any modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall fall within the protection scope defined by the claims of the present invention.

Claims

1. A wide-load zone regulation system for a tubular air preheater, characterized in that, It includes a pipe box partitioning actuator subsystem, a multi-dimensional measurement and sensing subsystem, a partitioned ammonium bisulfate phase change early warning model subsystem, a model predictive control multi-objective partitioned closed-loop control subsystem, and a distributed control system and advanced process control integrated communication subsystem; The tube box partitioning actuator subsystem includes: N independent tube box partitions that divide the tubular air preheater along the width of the boiler, where N is an integer not less than 2, and each tube box partition is independently configured with a partition air inlet regulating baffle and a partition hot air recirculation targeted regulating valve. The partition air intake regulating baffle is installed on the cold air inlet branch pipe of the corresponding pipe box partition and is driven by an electric actuator to independently regulate the air intake flow of the partition. The zoned hot air recirculation targeted regulating valve draws the hot air from the zone outlet of the pipe box back to the cold air inlet of the same zone through an independent hot air recirculation branch pipe. After mixing with the cold air, it targets and increases the air temperature at the inlet of the zone. The multidimensional measurement and sensing subsystem includes: an acoustic temperature measurement array arranged at the flue gas inlet section of the air preheater, a tunable diode laser absorption spectroscopy ammonia escape distribution measurement system arranged along the furnace width, a flue gas flow field measurement array at the cross-section of each tube box, a tube wall temperature thermocouple array at the cold end of each tube box, and feedback of sulfur trioxide concentration distribution along the furnace width from the selective catalytic reduction denitrification system. The partitioned ammonium bisulfate phase change early warning model subsystem, based on real-time data from the multi-dimensional measurement and sensing subsystem, independently calculates the ammonium bisulfate dew point temperature and acid dew point temperature for each pipe box partition. The calculation results are compared with the measured cold-end pipe wall temperature of that partition, outputting the partition risk level and risk trend prediction. When calculating the ammonium bisulfate dew point temperature for the i-th pipe box partition, the partitioned ammonium bisulfate phase change early warning model subsystem uses a nonlinear logarithmic function relationship based on the partial pressure of ammonia and sulfur trioxide in that partition. The partial pressure of ammonia is obtained from a tunable diode laser absorption spectroscopy system. The measured ammonia escape volume concentration was converted from the total flue gas pressure. The sulfur trioxide partial pressure was calculated from the sulfur trioxide concentration detection data through molar mass and flue gas density. When calculating the acid dew point temperature for the i-th pipe box zone, the commonly used logarithmic function relationship based on the sulfur trioxide mass concentration and flue gas water vapor volume fraction of that zone was adopted. The zone ammonium bisulfate phase change early warning model subsystem cyclically executes six steps with a fixed period of 1 second: data acquisition and preprocessing, partial pressure calculation and critical temperature calculation, safety margin calculation, risk level assessment, prediction trend analysis, and result output. The model predictive control multi-objective zoned closed-loop control subsystem takes minimizing the overall exhaust temperature as the optimization objective, takes the cold end pipe wall temperature of each pipe box zone not lower than the safe reference temperature dynamically output by the zoned ammonium bisulfate phase change early warning model subsystem as the rolling constraint, and takes the opening of the air inlet regulating damper and the opening of the hot air recirculation targeted regulating valve of each zone as the manipulated variables. It solves the constrained quadratic programming optimization problem online and outputs the optimal control quantity of each zone. The distributed control system and advanced process control integrated communication subsystem are responsible for real-time data exchange, control command issuance, and execution status feedback confirmation between the subsystems.

2. The system according to claim 1, characterized in that, The adjustment range of the zone air intake regulating damper is from 0% to 100% of its full stroke. When fully closed, it retains no more than 5% leakage. The opening command for each zone air intake regulating damper is as follows: The value range is [0.05, 1.0]; the zoned hot air recirculation targeted regulating valve draws back the outlet hot air with a temperature of 280℃ to 350℃ and mixes it with the cold air with a temperature of 20℃ to 40℃, thereby targeting and increasing the inlet air temperature of the zone by 20℃ to 80℃.

3. The system according to claim 1, characterized in that, The acoustic temperature measurement array is arranged at the flue gas inlet section of the air preheater, with at least 6×6 pairs of transducers. A two-dimensional flue gas temperature distribution map of the flue section is reconstructed using algebraic reconstruction technology, ensuring that each pipe box section corresponds to at least 4 effective grid nodes in the temperature field map. The average node temperature is used as the representative flue gas temperature of that section, with a temperature measurement accuracy of not less than ±5℃ and a sampling period of not more than 1s. The tunable diode laser absorption spectroscopy ammonia slip distribution measurement system is arranged with at least N laser measurement optical paths along the width of the air preheater. The ammonia slip concentration measurement accuracy of each section is not less than ±0.2 ppm, and the response time is not more than 1s.

4. The system according to claim 1, characterized in that, The risk levels are divided into three levels: normal, warning, and alarm. The risk level is defined as follows: when the difference between the cold end pipe wall temperature of the i-th pipe box section and the ammonium bisulfate dew point temperature, as well as the difference between the cold end pipe wall temperature and the acid dew point temperature, is not lower than the safety margin threshold. When the difference is less than 1, it is considered normal; when either of the two differences is less than 1, it is considered normal. If the difference is not lower than zero, it is considered a warning; if either of the two differences is lower than zero, it is considered an alarm.

5. The system according to claim 1, characterized in that, The model predictive control multi-objective zoned closed-loop control subsystem has a control cycle of 10 seconds, with 6 steps in the prediction time domain and 3 steps in the control time domain. The cold end pipe wall temperature prediction model for each pipe box zone adopts a discrete-time state-space form, with the zone's inlet regulating damper opening, hot air recirculation targeted regulating valve opening, cold air main pipe temperature, hot air recirculation branch outlet temperature, and flue gas inlet temperature as inputs, and the predicted cold end pipe wall temperature of the zone as output. The model parameters are determined through step response identification experiments and are adaptively updated weekly through online least squares identification.

6. The system according to claim 5, characterized in that, In the model predictive control multi-objective zoned closed-loop control subsystem, the wall temperature protection penalty weight for each pipe box zone is dynamically assigned based on the current risk level of that zone: when the risk level is normal, the penalty weight is zero; when the risk level is warning, the penalty weight is the nominal protection penalty weight; when the risk level is alarm, the penalty weight is 10 times the nominal protection penalty weight; the safe reference temperature for the wall temperature of each zone is equal to the larger of the ammonium bisulfate dew point temperature and the acid dew point temperature of that zone, plus a safety margin threshold. .

7. A method for wide-load zone regulation of a tubular air preheater, applied to the wide-load zone regulation system of a tubular air preheater as described in any one of claims 1 to 6, for regulating N tube box zones divided along the width direction of the boiler in the tubular air preheater, where N is an integer not less than 2, characterized in that, Includes the following processes: Multidimensional real-time sensing process: The two-dimensional temperature field of the flue gas inlet section of the air preheater is reconstructed by an acoustic temperature measurement array to obtain the representative flue gas temperature of each tube box section; The ammonia escape concentration distribution at corresponding positions in each tube box section along the furnace width was measured using a tunable diode laser absorption spectroscopy system. Obtain the sulfur trioxide concentration distribution, flue gas velocity, and cold end pipe wall temperature of each zone; Zoned ammonium bisulfate phase change early warning process: For each zone of the pipe box, the ammonium bisulfate dew point temperature and acid dew point temperature are independently calculated based on the measured ammonia escape concentration and sulfur trioxide concentration of that zone. Calculate the safety margins for the measured cold end pipe wall temperature of this zone relative to the ammonium bisulfate dew point temperature and the acid dew point temperature, respectively; When the comparison between the safety margin and the safety margin threshold meets the preset conditions, it is assessed as one of the three risk levels: normal, warning, or alarm. Least squares linear regression was performed on the historical time series of safety margins for each zone to predict the future risk trends of each zone. Output the risk level and trend prediction results of each zone to the model prediction controller; Model predicts the closed-loop regulation process of each zone: the wall temperature protection penalty weight and safety reference temperature of each zone are adaptively updated according to the risk level of each zone. With the goal of minimizing the overall exhaust temperature and the constraint that the cold end pipe wall temperature of each zone should not be lower than the corresponding safe reference temperature, a constrained quadratic programming optimization problem is solved online in a rolling manner. The optimal air intake damper opening command and the hot air recirculation targeted regulating valve opening command for each zone are sent to the actuators of each zone in parallel, and the execution is confirmed through position feedback.

8. The method according to claim 7, characterized in that, The constraints of the constrained quadratic programming optimization problem include: The lower limit of the opening of the air intake regulating damper for each zone is 0.05, and the upper limit is 1.

0. The lower limit of the opening of the hot air recirculation targeted regulating valve in each zone is 0, and the upper limit is the corresponding upper limit value determined by the pressure drop calculation of the pipeline network in each zone; The absolute value of the opening increment of the air intake regulating damper in each zone during a single control cycle shall not exceed 0.10; The absolute value of the opening increment of the hot air recirculation targeted regulating valve in each zone during a single control cycle shall not exceed 0.08; The sum of the openings of all zone air intake regulating dampers shall not be less than the product of the average damper opening determined by the minimum ventilation requirement of the boiler and the number of zones N.

9. The method according to claim 7, characterized in that, When a certain pipe box zone is at the alarm risk level for three consecutive control cycles, the soot blowing program for that zone is interlocked and automatically resumed after the soot blowing is completed.

Images