Large power station boiler inferior coal low-nitrogen combustion optimization control system and method based on machine learning and multi-objective optimization

Abstract

The invention discloses a large power station boiler inferior coal low-nitrogen combustion optimization control system and method based on machine learning and multi-objective optimization, and relates to the technical field of coal-fired power generation and clean combustion control. The control system is composed of a combustor subsystem, a detection subsystem and a control subsystem, the combustor subsystem constructs a subarea air supply structure with stable combustion at the bottom layer, low nitrogen at the middle and upper layers and burnout at the upper part through layered combustors and OFA nozzles, and the detection subsystem completes online monitoring of coal quality and online analysis of combustion state parameters; and the control subsystem comprises coal quality adaptive decision, multi-objective optimization and closed-loop feedback. According to the control method, inferior coal is recognized through machine learning, layered air distribution is activated in a self-adaptive mode, the optimal air volume and the rotational flow angle are optimized and solved online with efficiency and NOx as multiple targets, and finally precise execution and fine adjustment are conducted through closed-loop feedback. According to the invention, wide-load stable combustion, continuous low NOx level and burnout degree improvement can be realized, and the method is suitable for engineering application of combustion of inferior coal in a large-scale unit.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of coal-fired power generation and clean combustion control, and relates to the technical field of combustion optimization and pollutant control of large-scale power station boilers, and in particular to a large-scale power station boiler substandard coal combustion optimization control system and method based on machine learning and multi-objective optimization. BACKGROUND

[0002] Coal occupies an important position in the energy system, and large-capacity, high-parameter coal-fired generating units, as the basic force of power supply, play a core role in ensuring power balance and improving system regulation capacity. Under the background of energy structure transformation, improving the operation flexibility of such units and realizing clean and efficient combustion have become an important development direction. Limited by the characteristics of coal resource endowment and market fluctuations, a large number of power station boilers are designed to burn substandard coal to reduce fuel costs. However, the characteristics of low volatile matter and high fixed carbon content of substandard coal result in poor ignition performance and insufficient combustion stability, making it difficult to adapt to rapid load change requirements; the high nitrogen content exacerbates the generation of NOx, which is contrary to the requirements of green and low-carbon development; the low calorific value characteristics require a larger coal injection amount to maintain the load, resulting in an increase in flue gas volume and exhaust gas loss, which not only reduces the boiler efficiency but also affects the economic efficiency of unit peak shaving. Therefore, in the process of building a new power system, realizing efficient and stable combustion of substandard coal and effectively controlling pollutant emissions has become a key technical breakthrough for improving the operation flexibility of units and promoting the coordinated development of coal-fired power and new energy.

[0003] The existing combustion control of large-scale boilers mainly relies on traditional staged combustion and flue gas recirculation technology, which distributes air volume at different heights in the furnace to suppress the generation of NOx. However, the traditional equal air distribution and open-loop regulation method cannot effectively solve the above-mentioned contradictions: if the oxygen content in the main combustion zone is reduced to suppress the generation of NOx, it will further aggravate the ignition delay of substandard coal, leading to incomplete combustion and an increase in stable combustion load; if the oxygen content in the main combustion zone is increased to improve ignition and burnout, it will cause a sharp increase in the generation of NOx.

[0004] To solve the above problems, there have been many attempts in the prior art to solve the above problems. For example, Chinese patent CN106895435A proposes a hierarchical air distribution method, but mainly for single type burners, without fully considering the stable combustion demand of low-quality coal, and the NOx control effect is limited. CN119687477A focuses on dynamic regulation of combustion, but does not involve oxygen distribution mechanism among different types or different levels of burners. CN115875690A proposes a dual-fuel burner and its automatic control method, the core of which is to stabilize the heat output through automatic switching and compensation of main and auxiliary fuels. Although this method improves the stability of heat value, it does not involve the complex wind-coal matching and nitrogen oxide control mechanism in coal-fired boilers. CN113188120A discloses a double-tuned burner with independently adjustable air volume and swirl intensity, which enhances the adaptability of the burner to coal quality changes. Although this design improves the adjustment flexibility, its control logic is still focused on the wind field organization within a single burner, and does not achieve global optimization of air distribution among multiple burners. In addition, existing combustion control systems mostly use static air distribution strategies based on experience or preset models, which are difficult to effectively respond to rapid fluctuations in incoming coal quality and deep changes in boiler load, resulting in a lag in the adjustment of combustion parameters and the inability to achieve coordinated control of combustion stability, thermal efficiency and NOx emissions.

[0005] In summary, although there have been many improvement schemes in the prior art, a control method that can simultaneously optimize combustion stability, load response rate and NOx emission concentration when large-scale power station boilers burn low-quality coal has not been formed. Therefore, there is an urgent need to develop a new control system and method that can simultaneously solve the contradiction between unstable combustion of low-quality coal and high NOx emission, to meet the requirements of the new power system for flexible, clean and efficient operation of coal-fired units. SUMMARY

[0006] (I) Invention purposes The purpose of the present application is to overcome the shortcomings of the prior art and provide a large-scale power station boiler low-quality coal low-nitrogen combustion optimization control system and method based on machine learning and multi-objective optimization. Through the deep integration of online coal quality sensing, multi-parameter dynamic monitoring and intelligent control algorithm, real-time and accurate matching of air distribution strategy and combustion state is achieved, ultimately achieving efficient and low-nitrogen clean combustion of low-quality coal, and meeting the strict requirements of the new power system for deep peak shaving capacity and ultra-low emission operation of large-scale coal-fired units.

[0007] (II) Technical solutions To achieve the purpose of the present application and solve its technical problems, the present application adopts the following technical solutions: The first inventive objective of the present application is to provide a large power station boiler low-nitrogen combustion optimization control system based on machine learning and multi-objective optimization, which is used for realizing stable combustion in a wide load range, high combustion efficiency and low nitrogen oxide emission by synergistically optimizing the burner structure, stratified air distribution strategy and intelligent control algorithm in the furnace under the condition of using low-quality coal in a large power station boiler, and at least comprises: A burner subsystem, comprising burners arranged in a staggered stratified manner along the height direction of the front and rear walls of the furnace, first-type swirl pulverized coal burners integrated with plasma ignition and stable combustion functions arranged in the bottom layer of the furnace, second-type double-air-regulating swirl pulverized coal burners arranged in the middle layer and the top layer, and an over fire air (OFA) nozzle arranged above the burners, which is used for constructing a partitioned air supply channel of a main combustion zone and a post-combustion zone and capable of adjusting the ratio and swirl intensity of the inner and outer secondary air and OFA; A detection subsystem, which is used for providing online sensing data required for control and at least comprises (i) a coal quality online monitoring unit, which is used for measuring the ash content, moisture content, volatile content and calorific value of the coal fed into the furnace in real time and generating coal quality characteristic parameters; (ii) a combustion state sensing unit, which is used for acquiring the oxygen concentration, temperature distribution and NOx concentration at the furnace outlet of each layer of the combustion zone in real time; and (iii) an air volume and air pressure monitoring unit, which is used for monitoring the flow and resistance characteristics of each primary air, secondary air and OFA air duct in real time; A control subsystem, which is used for implementing stratified differential air distribution and global synergistic optimization control based on coal quality online sensing and combustion state feedback, and comprises a central controller and intelligent control modules and execution units in communication connection with the central controller, wherein the intelligent control modules at least comprise (i) a coal quality adaptive decision module, which is internally provided with a coal quality-air distribution mapping model based on machine learning, receives the coal quality characteristic parameters and outputs stratified air distribution parameter suggestions for low-quality coal, including the target of the excess air coefficient of each layer, the target of the proportion of OFA air volume and the target of the stratified swirl vane angle; (ii) a multi-objective optimization control module, which takes the maximization of combustion efficiency and the minimization of NOx emission as optimization objectives, combines the unit load instruction and the burner subsystem constraint, solves the optimal combination of the opening degree of each stratified air door and the swirl vane angle, so that the bottom stable combustion zone, the upper low-nitrogen zone and the OFA combustion zone are coupled optimally in the whole furnace; and (iii) a combustion state closed-loop feedback module, which fine-tunes the optimization solution according to the real-time deviation of the oxygen concentration, temperature distribution and NOx concentration monitoring signals, so as to suppress the disturbance and maintain stable combustion and ultra-low emission; and wherein the execution unit is in communication connection with the air door and vane adjusting mechanism of each layer, and is used for finely adjusting the air volume and air pressure of each burner region.

[0008] The second inventive objective of the present application is to provide a large power station boiler low-nitrogen combustion optimization control method, which is based on the above-mentioned control system of the present application and at least comprises the following steps: SS1. Real-time acquisition and preprocessing of multi-source data: Based on the coal quality online monitoring unit, the ash content, moisture content, volatile matter and calorific value of the incoming coal are measured in real time to generate coal quality characteristic parameters. Based on the combustion state perception unit, the oxygen concentration, temperature distribution of each layer of the combustion area and the NOx concentration at the furnace outlet are obtained in real time. Based on the air volume and air pressure monitoring unit, the flow and resistance characteristics of each primary air, secondary air and OFA air duct are monitored in real time, and the collected data are time-synchronized, abnormal data are removed and features are extracted; SS2. Coal quality characteristic determination and air distribution mode triggering: The coal quality adaptive decision module determines the coal quality based on the received coal quality characteristic parameters. When the ash content is greater than 25% and the low calorific value is 18 MJ / kg, the inferior coal layered differential air distribution mode is triggered, and the target of each layer excess air coefficient, the target of OFA proportion and the suggestion of inner and outer secondary air blade angle are generated. When the coal quality returns to the set upper limit and meets the hysteresis and minimum residence time conditions, the switch back to the conventional equal air distribution mode is triggered. 18MJ / kg, the inferior coal layered differential air distribution mode is triggered, and the target of each layer excess air coefficient, the target of OFA proportion and the suggestion of inner and outer secondary air blade angle are generated. When the coal quality returns to the set upper limit and meets the hysteresis and minimum residence time conditions, the switch back to the conventional equal air distribution mode is triggered. SS3. Multi-objective optimization to solve the optimal control strategy: Under the inferior coal layered differential air distribution mode, the output of step SS2 is used as the optimization initial value and boundary. Based on the multi-objective optimization control module, a rolling horizon optimization problem is constructed with the maximum combustion efficiency and the minimum NOx emission as the main target, and the uniformity of the furnace temperature distribution and the fly ash carbon content constraint as the auxiliary. The decision variables include at least the opening of each layered air door and the angle of each layered swirl blade. The hard constraint conditions include at least the fan capacity, the air duct pressure drop, the safe operation window of the burner and the load instruction. The predictive control or fuzzy reasoning algorithm is used to solve and output the adjustment instruction of each air door opening and blade angle in real time, and the executable target trajectory is output for the execution unit to track. SS4. Execution response and closed-loop feedback rolling optimization: The central controller issues the allocation instruction of primary air, inner and outer secondary air and OFA and the blade angle instruction to the execution unit, and completes the coordinated adjustment of the bottom stable combustion zone, the upper low-nitrogen zone and the post-combustion zone. The combustion state closed-loop feedback module continuously feeds back the oxygen concentration, temperature distribution of each layer of the combustion area and the NOx concentration at the furnace outlet based on the combustion state perception unit. The multi-objective optimization solution is fine-tuned and amplitude-limited corrected by using the rolling optimization strategy, and the OFA proportion and the outer secondary air blade angle are dynamically adjusted, and the opening of each layer air door is adjusted in linkage, so as to suppress the load disturbance and parameter drift, and maintain stable combustion and ultra-low emission. SS5. Safety protection and mode smooth switching: When the flame is retracted, the local oxygen is rich or supercooled, the NOx is over-limit or the sensor health is insufficient, the interlocking and safety rollback are triggered, the preset safety air distribution curve is enabled and the change rate and boundary are tightened. When switching from the conventional equal air distribution to the inferior coal differential air distribution or vice versa, the smooth switching is completed according to the preset gradual change trajectory and the minimum residence time, so as to ensure that the wall temperature and the furnace temperature rise are not over-limit. SS6. Adaptive update and performance maintenance (preferred step): Perform online incremental learning or periodic offline retraining of the coal quality-fuel distribution mapping model using closed-loop operational data; Periodically re-tune the optimization weights and active constraint set, update the mode threshold and hysteresis bandwidth according to real-time bias and disturbance estimates; Continuously achieve the goals of low NOx emission, low fly ash carbon content and high boiler efficiency under poor coal conditions through the synergistic adaptation of feedforward-optimization-feedback.

[0009] (Three) Technical effects Compared with the prior art, the large power station boiler low-nitrogen combustion optimization control system and method based on machine learning and multi-objective optimization of poor coal of the present application have the following beneficial and significant technical effects: (1) The present application creates a bottom stable combustion zone, middle and upper low-nitrogen zone, and upper OFA burnout zone through the synergy of double-burner structure design and stratified differential air distribution control, taking into account the dual needs of poor coal ignition and stable combustion and low NOx emission. The measured data show that, compared with before application: the NOx emission concentration at the furnace outlet is reduced from 194.97 mg / Nm 3 to 165.9 mg / Nm 3 , with a decrease of 14.9%; the fly ash carbon content is stably at an excellent level of below 1.80%; the minimum stable combustion load is greatly reduced from 30%THA to 18%THA, significantly improving the low-load peak shaving capability and operation flexibility of the boiler.

[0010] (2) The present application establishes an intelligent decision-making mechanism based on machine learning and multi-objective optimization, realizes feedforward control by constructing a rapid mapping from coal quality parameters to operating variables, and outputs the optimal combination of damper opening and inner and outer swirl vane angle by introducing a rolling horizon multi-objective optimization oriented to efficiency-emission-stability, unifying coal quality determination results, equipment limits and safety interlocks into a solvable constraint set; Cooperate with closed-loop feedback to implement small-step fine-tuning and amplitude-limiting correction on oxygen concentration / temperature distribution / NOx concentration bias, significantly enhance the robustness to load steps, coal quality mutations and measurement noise, reduce overshoot and oscillation, shorten disturbance recovery time, and improve full-condition tracking capability and executability. BRIEF DESCRIPTION OF DRAWINGS

[0011] Figure 1 Fig. 1 shows the architecture diagram of the large power station boiler low-nitrogen combustion optimization control system based on machine learning and multi-objective optimization provided by the embodiment of the present application; Figure 2 Fig. 2 shows the schematic diagram of the arrangement of the boiler burner and OFA nozzle along the furnace width direction; Figure 3 Fig. 3 shows the schematic diagram of the structure of (a) the first type of swirl pulverized coal burner, (b) the second type of swirl pulverized coal burner and (c) the OFA nozzle; Figure 4 The diagram shown is an implementation flowchart of the low-NOx combustion optimization control method for inferior coal in large power plant boilers provided in an embodiment of the present invention.

[0012] Explanation of reference numerals in the attached drawings: 10 for the front / rear wall of the furnace, 20 for the first type of swirl pulverized coal burner, 21 for the primary air pulverized coal passage, 22 for the primary air inlet elbow, 23 for the inner secondary air passage, 24 for the inner adjustable axial swirl blade, 25 for the outer secondary air passage, 26 for the outer adjustable axial swirl blade, 30 for the second type of swirl pulverized coal burner, 31 for the central direct air passage, 32 for the primary air pulverized coal passage, 33 for the primary air inlet elbow, 34 for the guide, 35 for the inner secondary air passage, 36 for the inner fixed axial swirl blade, 37 for the outer secondary air passage, 38 for the outer adjustable axial swirl blade, 40 for the OFA nozzle, 41 for the central direct air passage, 42 for the outer ring swirl air duct, and 43 for the adjustable axial swirl blade. Detailed Implementation

[0013] This invention aims to provide a low-NOx combustion optimization control system and method for low-quality coal in large power plant boilers based on machine learning and multi-objective optimization. To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions in the embodiments of this invention will be described in more detail below with reference to the accompanying drawings. The described embodiments are some, but not all, embodiments of this invention, and are exemplary, intended to explain the invention, and should not be construed as limiting the invention.

[0014] Example 1: Control System like Figure 1 As shown in the embodiments of the present invention, the low-NOx combustion optimization control system for low-quality coal in large power plant boilers, based on machine learning and multi-objective optimization, is used to achieve real-time and accurate matching of air distribution strategy and combustion state in large power plant boilers burning low-quality coal through the deep integration of online coal quality sensing, multi-parameter dynamic monitoring, and intelligent control algorithms, ultimately achieving efficient, low-NOx, and clean combustion of low-quality coal. Specifically, the low-NOx combustion optimization control system for low-quality coal in large power plant boilers of the present invention mainly includes three main parts: a burner subsystem, a detection subsystem, and a control subsystem, wherein: The burner subsystem includes two types of dual-adjustable swirl pulverized coal burners arranged in a staggered, layered manner along the height of the furnace front / rear walls 10. The bottom layer of the furnace houses the first type of swirl pulverized coal burner 20, integrating plasma ignition and combustion stabilization functions. Several layers of the first type of swirl pulverized coal burner can be installed at the bottom layer as needed, for example, two layers with eight burners per layer. The middle and top layers of the furnace house the second type of swirl pulverized coal burner 30. Several layers of the second type of swirl pulverized coal burner can be installed as needed, for example, four layers with eight burners per layer. Several layers of OFA (overburning air) nozzles 40 are installed above the burners to construct a zoned air supply channel between the main combustion zone and the post-burnout zone, and to adjust the ratio and swirl intensity of the internal and external secondary air and OFA. The arrangement structure of the burners and OFA nozzles along the furnace width in the height direction is described in [reference needed]. Figure 2 .

[0015] More specifically, such as Figure 3 As shown in Figure (a), the first type of swirl pulverized coal burner 20 is used to form a strong backflow stable combustion zone at the bottom of the furnace to ensure reliable ignition of low-quality coal. It includes: (i) a primary air pulverized coal channel 21, which is located in the central area of ​​the burner and extends axially. Its inlet end is provided with a primary air inlet elbow 22, and its outlet end forms a primary air pulverized coal nozzle, which is used to transport the primary air pulverized coal airflow to the center of the combustion zone in a direct, non-swirling manner; (ii) an inner secondary air channel 23, which is arranged around the outside of the primary air pulverized coal channel. The channel is provided with adjustable axial swirl blades 24, which are used to guide the inner secondary air. The secondary airflow forms a central high-temperature recirculation zone to ensure reliable ignition of low-quality coal in the bottom stable combustion zone; (iii) an external secondary air channel 25 is arranged around the outside of the internal secondary air channel, and an adjustable axial swirl vane 26 is provided in the channel to supplement the air required for the complete combustion of pulverized coal and to adjust the combustion intensity and oxygen distribution under the premise of ensuring the stability of bottom ignition; (iv) a plasma ignition device is integrated in the burner nozzle area and arranged axially relative to the primary air pulverized coal nozzle to provide high-temperature plasma ignition energy during boiler start-up or under low load conditions.

[0016] Preferably, the first type of swirl pulverized coal burner 20 is a burner that combines plasma ignition and stable combustion functions. Adjustable axial swirl blades 24 and 26 are installed in both its inner and outer secondary air channels. The installation angle of the inner secondary air swirl blade 24 is 20°. ~60 The initial setting is 40. The left and right sides are used to guide the inner airflow to form a central recirculation zone; the installation angle of the outer secondary air blade 26 is 40 degrees. ~80 The initial setting is 60. The left and right sides are used to enhance the rotational intensity of the peripheral airflow and the shear mixing effect; a small amount of inner secondary air is used to ignite the coal powder, and a large amount of outer secondary air is used to supplement the air required for coal powder burnout. The composite rotational flow formed by the inner and outer secondary air blades interacts at the combustion outlet, generating high-intensity vortex and negative pressure backflow area, thereby establishing a stable ignition core area in the front part of the furnace, and realizing the rapid ignition and stable combustion support of low-volatile and high-ash low-quality coal. The burner structure can be operated in cooperation with the plasma ignition device, so that when the excess air coefficient of the stable combustion zone at the bottom is controlled in the range of 0.85-0.95 (further preferably 0.85-0.88), the rapid ignition and stable combustion of high-ash low-quality coal powder can still be ensured.

[0017] As Figure 3 As shown in the middle (b) figure, the second type of rotational flow coal powder burner 30 is used to form a low-NOx combustion zone with strong reducing atmosphere in the middle and top layers of the furnace and to realize coal powder burnout control, and the whole adopts a concentric multi-ring layered air supply structure, at least including: (i) a central straight flow air passage 31 arranged in the central region of the burner and extending along the axial direction, used to convey central straight flow air in the form of non-rotational jet flow to stabilize the flame root and the backflow area boundary; (ii) a primary air coal powder passage 32 coaxially arranged around the central straight flow air passage 31, and the inlet end thereof is communicated with a primary air inlet elbow 33 with a guide 34, and the outlet end forms an annular primary air coal powder injection port, used to inject the primary air coal powder airflow in the form of annular jet flow into the combustion zone and realize bidirectional ignition and entrainment from inside to outside and from outside to inside through the inner and outer secondary air; (iii) an inner secondary air passage 35 arranged around the outer side of the primary air coal powder passage 32, and fixed axial rotational flow vanes 36 are arranged in the passage, used to form an inner layer of weak rotational flow to promote the early ignition of the inner side coal powder airflow; (iv) an outer secondary air passage 37 arranged around the outer side of the inner secondary air passage 35, and adjustable axial rotational flow vanes 38 are arranged in the passage, used to form outer secondary air with adjustable rotational flow intensity to enhance peripheral shear mixing and supplement the air required for burnout.

[0018] As a preferred, the second type of rotational flow coal powder burner 30 adopts central straight flow, which is surrounded by annular straight flow primary air coal powder injection port, rotational flow inner secondary air injection port and rotational flow outer secondary air injection port in sequence. By virtue of the design of central straight flow area and inner and outer secondary rotational flow area, the primary air coal powder mixture of the annular primary air coal powder injection port is ignited and ignited from inside to outside and from outside to inside at the same time. The rotational flow vanes 36 arranged in the inner secondary air passage 35 are arranged at a fixed angle to form a stable weak inner rotational flow and limit the size and position of the near-injection port backflow core, and the installation angle is preferably 20 The left and right sides are used to enhance the rotational intensity of the peripheral airflow and the shear mixing effect; a small amount of inner layer secondary air is used to ignite the coal powder, and a large amount of outer layer secondary air is used to supplement the air required for coal powder burnout. The composite rotational flow formed by the inner and outer secondary air blades interacts at the combustion outlet, generating high-intensity vortex and negative pressure backflow area, thereby establishing a stable ignition core area in the front part of the furnace, and realizing the rapid ignition and stable combustion support of low-volatile and high-ash low-quality coal. The burner structure can be operated in cooperation with the plasma ignition device, so that when the excess air coefficient of the stable combustion zone at the bottom is controlled in the range of 0.85-0.95 (further preferably 0.85-0.88), the rapid ignition and stable combustion of high-ash low-quality coal powder can still be ensured. The small-angle swirling flow on the left and right sides enables rapid mixing and ignition while avoiding excessive swirling shear force that could cause the flame to be blown out or the coal powder jet to disperse. Furthermore, the inner secondary air, as the first layer of the outer ignition air, works in conjunction with the central direct air to form a bidirectional heating envelope for the annular coal powder jet. Its weak swirling characteristics reserve space for the establishment of a reducing atmosphere and avoid premature introduction of excessive oxygen while ensuring reliable ignition.

[0019] The swirl blades 38 in the external secondary air duct 37 are adjustable in angle, with an adjustment range of 40°. -80 The initial angle is set to 60 degrees. The left and right sides are used to generate adjustable swirl intensity secondary air, serving as the second layer of supplemental oxygen supply air. This supplemental oxygen supply and entrainment mixing to the combustion zone from the outside inwards is achieved by adjusting the blade angle to change the swirl intensity. When the blade angle is set to a smaller value (close to 40 degrees), the swirl intensity is adjusted. When the blade angle is set to a large value (close to 80 degrees), it generates a strong swirling flow, enhancing airflow mixing intensity and turbulence, shortening combustion completion time, making it suitable for high-load conditions or operating scenarios requiring rapid burnout; When the swirling intensity is reduced, the residence time of the airflow in the combustion zone is extended and the volume of the combustion zone is expanded, forming a deep reducing atmosphere. This is suitable for operation scenarios where low-NOx combustion control is required or where low-quality coal needs to be fully burned.

[0020] Overall, the second-type swirl pulverized coal burner 30 controls the excess air coefficient in the middle and upper low-NOx zones within the range of 0.75-0.85 (more preferably 0.78-0.82), and in conjunction with the dynamic adjustment of the external secondary air blade angle, creates an oxygen-deficient, strongly reducing atmosphere in the main combustion zone. This reduces the generated nitrogen oxides to nitrogen in an oxygen-deficient environment, while simultaneously inhibiting the generation of new nitrogen oxides from the generation mechanism. This achieves a dual low-NOx combustion effect of air staging and fuel staging, significantly reducing the concentration of nitrogen oxide emissions while ensuring stable combustion of low-quality coal.

[0021] like Figure 3 As shown in Figure (c), the OFA nozzle 40 adopts a coaxial composite structure of central direct-flow secondary air and outer ring swirl secondary air. This structure is used to construct a staged air supply post-combustion zone above the main combustion zone of the furnace and to achieve coordinated control of penetration and mixing. It includes at least: (i) a central direct-flow air channel 41, located at the center of the nozzle and extending axially, used to eject direct-flow secondary air in the form of a non-swirling jet, forming an axial penetrating airflow above the main combustion zone, guiding the high-temperature flue gas in the furnace to entrain and promote the secondary oxidation reaction of residual combustible components; and (ii) an outer ring swirl air channel 42, surrounding the central direct-flow air channel, with adjustable axial swirl blades 43 (preferably with an adjustment range of 20°). -90 ), for generating adjustable swirl intensity of outer ring swirl secondary air, forming controllable tangential momentum and swirl flow field at the nozzle outlet by the swirl effect, optimizing the penetration and mixing intensity of the overfire air, and ensuring complete combustion of the pulverized coal.

[0022] As preferred, the central straight flow in the OFA nozzle 40 and the outer ring swirl flow form a straight flow-swirl-backflow triple airflow coupling structure at the nozzle outlet, which on the one hand introduces air into the deep part of the furnace through the penetration effect of the central straight flow jet, and on the other hand promotes the reburning reaction of residual CO and unburned carbon particles through the radial entrainment effect of the outer ring swirl flow. The air volume of the OFA nozzle accounts for 25%-35% (further preferably 27%-30%) of the total secondary air volume, and under the action of the multi-objective optimization algorithm of the control system, the swirl blade angle and injection intensity are automatically adjusted in combination with real-time NOx concentration and temperature distribution feedback to maintain the air staging effect of the burnout zone and the minimum generation rate of NOx, realizing efficient burnout and ultra-low emission under the condition of poor coal.

[0023] As shown in Figure 1 The control system of the present application further includes a detection subsystem for providing online sensing data required for control, including: (i) a coal quality online monitoring unit for real-time measurement of ash content, moisture content, volatile matter and calorific value of the coal entering the furnace and generation of coal quality characteristic parameters; (ii) a combustion state sensing unit for real-time acquisition of oxygen concentration, temperature distribution and NOx concentration at the furnace outlet of each layer of combustion area; (iii) a wind volume and pressure monitoring unit for real-time monitoring of the flow and resistance characteristics of each primary air, secondary air and OFA air duct.

[0024] As preferred, the coal quality online monitoring unit adopts a multi-modal sensing fusion analysis device, including at least two or more measurement means among microwave, laser and X-ray for real-time measurement of ash content, moisture content, volatile matter and low calorific value of the coal entering the furnace, and through self-calibration and fault diagnosis algorithms to eliminate and compensate for drift and abnormal data, finally output coal quality characteristic parameters for control, when the detected ash content is greater than 25% and the calorific value is less than 18 MJ / kg, triggering the system to enter the layered differential air distribution mode; when the coal quality parameters return to the set upper limit, return to the conventional equal air distribution mode.

[0025] As preferred, the combustion state sensing unit comprises a hierarchically arranged wide-range oxygen concentration sensor array, a high-sensitivity temperature sensor array, and an online NOx concentration monitoring device arranged at the furnace outlet and / or the back pass, forming a full-process combustion state sensing network covering the main combustion zone, the burnout zone, and the flue gas discharge path, the sampling frequency of each sensor is not less than 1 Hz, and the data quality control is realized through signal conditioning, time synchronization, and outlier rejection, and the collected data form the linkage observation quantity of oxygen field-temperature field-outlet NOx after position mapping and feature extraction, which is used by the intelligent control module for rolling optimization and closed-loop correction of the hierarchical air distribution parameters.

[0026] As preferred, the air volume and air pressure monitoring unit comprises high-precision air volume sensors, air pressure transmitters, and valve position / air damper angle position feedback devices arranged on the primary air, secondary air, and OFA air ducts of each first-type and second-type swirl pulverized coal burner and OFA nozzle, and is linked with the calibration module for online correction of measurement errors caused by dust, wear, and temperature drift to monitor the air volume and resistance characteristics of each air duct in real time, and the collected data are fed back to the intelligent control module as constraints and verification quantities for multi-objective optimization and closed-loop correction.

[0027] As shown in Figure 1 The control system of the present application further comprises a control subsystem for implementing hierarchical differential air distribution and global collaborative optimization control based on coal quality online sensing and combustion state feedback, including a central controller and an intelligent control module and an execution unit in communication connection therewith, the intelligent control module comprising: (i) a coal quality adaptive decision-making module, which is built-in with a coal quality-air distribution mapping model based on machine learning, receives coal quality characteristic parameters and outputs hierarchical air distribution parameter suggestions for inferior coal, including each layer excess air coefficient target, OFA air volume proportion target, and hierarchical swirl vane angle target; (ii) a multi-objective optimization control module, which takes the maximization of combustion efficiency and the minimization of NOx emission as optimization objectives, combines with unit load instructions and burner subsystem constraints to solve the optimal combination of each hierarchical air damper opening and swirl vane angle, so that the bottom stable combustion zone, the upper low-nitrogen zone, and the OFA burnout zone achieve coupled optimization in the full furnace range; (iii) a combustion state closed-loop feedback module, which fine-tunes the optimization solution according to the real-time deviations of oxygen concentration, temperature distribution, and NOx concentration monitoring signals to suppress disturbances and maintain stable combustion and ultra-low emission; the execution unit is in communication connection with the air damper and vane adjustment mechanisms of each layer for fine adjustment of the air volume and air pressure of each burner region.

[0028] As preferred, the coal quality self-adaptive decision module is used to generate a feedforward target of stratified air distribution and swirl setting under the condition of coal quality fluctuation, the coal quality-air distribution mapping model built in the module is trained by using a supervised learning or ensemble learning method based on machine learning, the input of the model is the coal quality characteristic parameters output by the coal quality online monitoring unit, and the output includes stratified excess air coefficient target, OFA air volume ratio target and stratified swirl vane angle target; and the model maintains self-adaptability to coal quality fluctuation through online incremental learning or regular offline retraining, and sets confidence evaluation and fallback strategy, when the model confidence is lower than the threshold, switches to a conservative safe air distribution curve to ensure stable combustion; when the online determination of ash content 25% and low calorific value 18MJ / kg triggers the poor coal stratified air distribution mode and introduces the parameter hysteresis to avoid frequent jitter; when the coal quality recovers to the set upper limit and meets the hysteresis condition, it is smoothly switched back to the conventional equal air distribution mode.

[0029] Further, the coal quality-air distribution mapping model is trained by using a supervised learning or ensemble learning method based on machine learning, taking the coal quality characteristic parameter vector under the historical operation condition as the input feature, and taking the excess air coefficient, OFA air volume ratio and inner and outer swirl vane angle of each layer under the corresponding condition optimized by expert experience or offline multi-objective optimization as the output label for supervised learning training; under the poor coal stratified differential air distribution mode, the mapping model recommendation value strengthens the lower air distribution ratio to prolong the residence time of poor coal in the high temperature zone and improve the burnout rate, while weakening the upper air distribution ratio and increasing the OFA air volume ratio to establish a deep staged combustion structure, effectively reducing the NOx generation concentration under the premise of ensuring stable combustion.

[0030] As preferred, the multi-objective optimization control module takes the feedforward target and mode state of the coal quality self-adaptive decision module as the initial value and soft constraint, constructs an optimization problem with maximum combustion efficiency and minimum NOx emission as the main target, and with uniformity of furnace temperature distribution and fly ash carbon content constraint as the auxiliary, the decision variables at least include the air door opening degree of each layer and the swirl vane angle of each layer, the hard constraint conditions at least include the fan capacity, the air duct pressure drop, the safe operation window of the burner and the load instruction, the predictive control or fuzzy reasoning algorithm is used to solve and output the adjustment instruction of each air door opening degree and vane angle in real time, and the executable target trajectory is output for the execution unit to track.

[0031] As preferred, the combustion state closed-loop feedback module is communicatively connected to the multi-objective optimization control module and the combustion state monitoring unit, compares the received real-time monitoring data of the combustion state with the set value, implements small-step fine-tuning and amplitude limiting correction on the multi-objective optimization solution based on the deviation and trend, adopts layered and partitioned gain scheduling and dead zone anti-saturation strategy, and preferentially allocates the adjustment amount to the control channels that are most sensitive to the target and have low mutual coupling; when it is observed that the NOx is close to the upper limit or there is a local overheating or oxygen deficiency trend, the OFA proportion and the outer secondary air blade angle are preferentially adjusted, and the opening degrees of the air dampers of each layer are simultaneously corrected to suppress the disturbance and parameter drift; when the unit rapidly changes load, the linkage adjustment weight of the OFA swirl blade angle and the upper outer secondary air guide vane angle is increased to suppress flame shrinkage or skew and maintain a low NOx level; if a sensing abnormality or data quality warning is detected, the current optimization amount is frozen and switched to a conservative air distribution curve, and after the observation is restored, the optimal trajectory is asymptotically regressed to maintain stable combustion and ultra-low emission targets.

[0032] As preferred, the execution unit includes electric or pneumatic servo mechanisms connected to the air dampers of the primary air, secondary air and OFA air ducts of each layer, and angle servo actuators connected to the adjustable swirl vanes of each burner and OFA nozzle; and the execution unit has position closed-loop and speed feedforward functions to complete tracking adjustment of air distribution and vane angle in small steps and short cycles, and to achieve stuck detection and fault switching through state monitoring, thereby ensuring the reachability and dynamic stability of the optimization control strategy.

[0033] It should be noted that the control system of the present application is not a simple superposition of components, but cooperates in an integrated architecture of layered burners-OFA nozzles-sensing-decision-optimization-feedback-interlocking: The bottom stable combustion zone and the middle and upper low-nitrogen zones form a three-section air staging channel under the cooperation of the OFA coaxial composite nozzle; the coal quality-air distribution mapping provides a feedforward setting, the multi-objective optimization unifies the efficiency and emission targets in a rolling time domain, and the closed-loop feedback is small-step corrected based on the oxygen concentration / temperature distribution / NOx concentration deviation, and is protected by equipment constraints and safety fallback. This architecture enables the system to automatically trigger the layered differential air distribution mode under the condition of low-quality coal, to realize spatial partitioning and cooperative regulation inside the furnace by air distribution strengthening in the bottom stable combustion zone, air distribution weakening in the upper low-nitrogen zone, and OFA staged oxygen supplementation in the post-combustion zone, to control the NOx emission concentration within the ultra-low emission standard while maintaining the boiler combustion efficiency and load response performance on the premise of ensuring stable ignition and full burnout of low-quality coal with high ash content and low calorific value.

[0034] Embodiment 2: Control method On the basis of the above-mentioned embodiment 1, the large-capacity power station boiler low-nitrogen combustion optimization control method based on the above-mentioned large-capacity power station boiler low-nitrogen combustion optimization control system for low-quality coal is further described, such asFigure 4 As shown, the method mainly includes the following steps when implemented: SS1. Real-time acquisition and preprocessing of multi-source data: Based on the coal quality online monitoring unit, the ash content, moisture content, volatile matter and calorific value of the incoming coal are measured in real time to generate coal quality characteristic parameters. Based on the combustion state perception unit, the oxygen concentration, temperature distribution and NOx concentration at the furnace outlet of each layer of the combustion area of the boiler are obtained in real time. Based on the air volume / air pressure monitoring unit, the flow and resistance characteristics of each primary air, secondary air and OFA air duct are monitored in real time, and the collected data are time-synchronized, abnormal data are excluded, and features are extracted.

[0035] SS2. Coal quality characteristic determination and air distribution mode triggering: The coal quality adaptive decision module determines the coal quality based on the received coal quality characteristic parameters. When it is detected that the ash content is greater than 25% and the low calorific value 18MJ / kg, the inferior coal layered differential air distribution mode is triggered, and the target of the excess air coefficient of each layer, the target of the OFA proportion, and the suggestion of the inner and outer secondary air blade angle are generated. When the coal quality returns to the set upper limit and meets the hysteresis and minimum residence time conditions, the conventional equal air distribution mode is triggered. The mode flag is transmitted as an explicit constraint to the subsequent steps.

[0036] As an option, the coal quality characteristic determination and air distribution mode triggering further introduces a hysteresis criterion and a minimum residence time strategy. A double-threshold interval is used for mode switching management of the ash content and the low calorific value. When it is detected that the ash content is greater than the first threshold value 25% and the low calorific value the second threshold value is 18MJ / kg, and the duration of this coal quality state exceeds the coal quality determination delay threshold, the inferior coal layered differential air distribution mode is triggered. When the coal quality returns to the ash content the third threshold value is 22%, or the low calorific value the fourth threshold value is 20MJ / kg, and the duration of this coal quality state exceeds the coal quality recovery minimum residence time, and meets the hysteresis determination condition, the conventional equal air distribution mode is triggered. The first threshold value and the third threshold value form an ash content determination hysteresis interval, and the second threshold value and the fourth threshold value form a calorific value determination hysteresis interval. At the same time, the maximum switching frequency and the minimum residence time constraints are applied to the mode switching, and the gradual change trajectory of the layered excess air coefficient and the OFA proportion is generated in the switching transition process to avoid thermal shock and furnace temperature mutation.

[0037] SS3. Multi-objective optimization to solve the optimal control strategy: Under the mode of poor coal layering and differential air distribution, the output of step SS2 is used as the initial value and boundary for optimization, and a rolling horizon optimization problem is constructed based on the multi-objective optimization control module, with the maximization of combustion efficiency and minimization of NOx emission as the main objectives, and the uniformity of furnace temperature distribution and the constraint of fly ash carbon content as the auxiliary objectives. The decision variables include at least the opening of each layer air door and the angle of each layer swirl vane, the hard constraint conditions include at least the fan capacity, the air duct pressure drop, the safe operation window of the burner and the load instruction, and the predictive control or fuzzy reasoning algorithm is used to solve and output the adjustment instructions of the air door opening and the vane angle in real time, and the executable target trajectory is output for the execution unit to track.

[0038] As preferred, the solution of the multi-objective optimization problem is centered on model predictive control, and the distribution of primary air, inner and outer secondary air and OFA and the angle of each layer swirl vane are optimized in the rolling horizon; the cost function includes the negative term of thermal efficiency, the NOx emission term, the CO and fly ash carbon penalty term, the temperature uniformity and flame stability penalty term, and the adaptive redistribution mechanism of target weight is set; the hard constraints include the fan power and pressure head, the air duct pressure drop, the jet momentum flux ratio, the minimum stable combustion air volume of the burner, the wall temperature and the upper limit of the furnace temperature rise, and the upper limit of the load tracking error, to ensure the feasibility and safety of the solution.

[0039] In addition, under the mode of poor coal layering and differential air distribution, the asymmetric air distribution template and the variable slope constraint are added, a higher minimum air volume and a smaller allowable downward adjustment slope are set for the bottom stable combustion zone, a lower target excess air coefficient is set for the upper main combustion zone and the upward adjustment slope is limited, and the linkage constraint of OFA ratio and outer secondary air swirl angle is set to maintain the jet momentum flux ratio; under the mode of conventional equal air distribution, the layering air distribution deviation penalty is introduced to maintain symmetry and economy.

[0040] SS4. Execution response and closed-loop feedback rolling optimization: The central controller issues the distribution instructions of primary air, inner and outer secondary air and OFA and the vane angle instructions to the execution unit, and completes the coordinated adjustment of the bottom stable combustion zone, the upper low-nitrogen zone and the post-combustion zone; the combustion state closed-loop feedback module continuously feeds back the oxygen concentration, temperature distribution and furnace outlet NOx concentration of each layer combustion zone based on the combustion state perception unit, and uses the rolling optimization strategy to implement small-step fine-tuning and amplitude-limiting correction of the multi-objective optimization solution, dynamically adjusts the OFA ratio and the outer secondary air vane angle, and adjusts the air door opening of each layer in linkage, to suppress load disturbance and parameter drift, and maintain stable combustion and ultra-low emission.

[0041] As preferred, the synergistic adjustment strategy of each layer combustion zone is: the bottom layer stable combustion zone corresponding to the bottom of the furnace adopts air distribution strengthening strategy, increases the primary air speed and the secondary air distribution volume, increases the local oxygen concentration and the combustion temperature, ensures the rapid ignition of low-quality coal and the complete analysis of volatile matter; the middle main combustion zone corresponding to the middle region of the furnace adopts moderate air distribution strategy, maintains a low oxygen concentration environment and appropriately reduces the cyclone intensity, prolongs the residence time of coal powder in the high temperature zone, and promotes the complete combustion of fixed carbon; the upper low nitrogen zone corresponding to the upper region of the furnace adopts air distribution weakening strategy, reduces the secondary air distribution volume and reduces the outer ring cyclone blade angle, maintains a lean oxygen environment to inhibit the generation of nitrogen oxides NOx; the top afterburning zone adopts OFA air distribution strengthening strategy, layered and gradual oxygen supplement, lower layer strengthens oxygen supplement, upper layer moderate oxygen supplement, ensures the fly ash burnout while avoiding excessive oxygen leading to NOx emission rebound, realizes the synergistic optimization of stable combustion and low nitrogen emission.

[0042] Further, the combustion state closed-loop feedback adopts a rolling optimization fine-tuning mechanism, based on the oxygen concentration, temperature distribution and furnace outlet NOx concentration of each layer combustion zone continuously fed back by the combustion state sensing unit, the deviation of the actual value and the target value of the performance index is calculated in real time, including the combustion efficiency deviation, the nitrogen oxide emission deviation, the temperature distribution uniformity deviation, the fly ash carbon content deviation; when the performance deviation exceeds the preset threshold or the load disturbance and coal quality fluctuation are detected, the rolling optimization fine-tuning process is triggered, the OFA proportion and the outer secondary air blade angle which respond fast are preferentially dynamically adjusted, small step adjustment is implemented and amplitude limiting correction is carried out; the opening degree of each layer damper which responds slowly is simultaneously linked and fine-tuned, through the fast-slow variable layered adjustment strategy, the control stability is maintained while the working condition change is quickly responded, the influence of load disturbance and parameter drift on combustion performance is effectively inhibited, the stable combustion of low-quality coal and the ultra-low nitrogen oxide emission target are maintained.

[0043] SS5. Safety protection and mode smooth switching: When the flame recession, local oxygen enrichment or supercooling, NOx overrun or sensor health is insufficient are monitored, the interlocking and safety rollback is triggered, the preset safety air distribution curve is enabled and the change rate and boundary are tightened; when switching from conventional equal air distribution to low-quality coal differential air distribution or vice versa, the smooth switching is completed according to the preset gradual change trajectory and minimum residence time, ensuring that the wall temperature and furnace temperature rise are not over-limit.

[0044] As preferred, the safety protection strategy includes flame recession detection, local oxygen enrichment or supercooling criterion, NOx and CO double threshold linkage, wall temperature rate protection and sensor degradation identification; when any safety protection condition is triggered, the system enters the safety rollback mode, converges according to the preset safety air distribution curve, and freezes part of the weight and tightens the constraint boundary; the safety rollback removal needs to meet the multi-variable joint recovery condition and the minimum residence time, and the mode smooth switching is realized through the preset gradual change trajectory, avoiding thermal stress impact and temperature oscillation.

[0045] SS6. Adaptive update and performance maintenance (preferred steps): Perform online incremental learning or periodic offline retraining of the coal quality-air distribution mapping model with closed-loop operational data; periodically re-tune the optimization weights and active constraint set, update the mode threshold and hysteresis bandwidth according to real-time bias and disturbance estimation; continuously achieve the goals of low NOx emission, low fly ash carbon content and high boiler efficiency under poor coal conditions through the synergistic adaptation of feedforward-optimization-feedback.

[0046] It should be noted that the control method shown in Embodiment 2 constructs a full-process closed-loop control system of data acquisition-mode determination-optimization solving-execution feedback-safety protection-adaptive update through six steps. The method automatically triggers the air distribution mode switching according to the coal quality characteristic parameters, solves the optimal solution of the opening degree of each layer air door and the blade angle through a multi-objective optimization algorithm, and implements rolling optimization fine-tuning based on real-time feedback of the combustion state, realizing the coordinated regulation of the furnace partition. The hysteresis criterion and minimum residence time strategy introduced in the method avoid mode jitter, the fast-slow variable layered regulation strategy balances the response speed and stability, and the safety rollback and gradual trajectory mechanism guarantees the switching smoothness, which simultaneously meets the control targets of stable combustion, ultra-low emission and high burnout rate under poor coal conditions.

[0047] Through the above embodiments, the purposes of the present application are completely and effectively achieved. Those skilled in the art can understand that the present application includes but is not limited to the contents described in the drawings and the above specific embodiments. Although the present application has been described in relation to the presently preferred embodiments thereof, it is to be understood that the application is not limited to the disclosed embodiments, and any modification not departing from the functional and structural principles of the present application will be included in the scope of the claims.

Claims

1. A low-NOx combustion optimization control system for low-quality coal in a large power plant boiler based on machine learning and multi-objective optimization, characterized in that, At least including: The burner subsystem includes burners arranged in layers along the height direction on the front and rear walls of the furnace. The bottom layer of the furnace is equipped with a first-class swirl pulverized coal burner with integrated plasma ignition and stable combustion functions. The middle and top layers are equipped with a second-class swirl pulverized coal burner with dual air adjustment. OFA nozzles are set above the burners. The detection subsystem is used to provide online sensing data required for control, including at least (i) a coal quality online monitoring unit for real-time measurement of ash, moisture, volatile matter and calorific value of coal fed into the furnace and generating coal quality characteristic parameters; (ii) Combustion status sensing unit, used to acquire oxygen concentration, temperature distribution and NOx concentration at furnace outlet in each combustion zone in real time; (iii) Air volume and pressure monitoring unit, used to monitor the flow and resistance characteristics of each primary air, secondary air and OFA duct in real time. The control subsystem, used for implementing stratified differential air distribution and global collaborative optimization control based on online coal quality sensing and combustion status feedback, includes a central controller, an intelligent control module, and an execution unit. The intelligent control module includes at least (i) a coal quality adaptive decision-making module, which has a built-in coal quality-air distribution mapping model based on machine learning, receives coal quality characteristic parameters, and outputs suggestions for stratified air distribution parameters for low-quality coal; (ii) a multi-objective optimization control module, which takes maximizing combustion efficiency and minimizing NOx emissions as optimization objectives, and solves the optimal combination of damper opening and swirl blade angle for each stratum by combining unit load commands and burner subsystem constraints; and (iii) a combustion status closed-loop feedback module, which fine-tunes the optimization solution based on the real-time deviation of oxygen concentration, temperature distribution, and NOx concentration monitoring signals. The execution unit is communicatively connected to the damper and blade adjustment mechanisms of each stratum to finely adjust the air volume and air pressure of each burner area.

2. The control system according to claim 1, characterized in that, The first type of swirl pulverized coal burner is used to form a strong backflow stable combustion zone at the bottom of the furnace to ensure reliable ignition of low-quality coal, including: (i) a primary air pulverized coal channel, which is set in the central area of ​​the burner and extends axially, with a primary air inlet elbow at its inlet end and a primary air pulverized coal nozzle at its outlet end; (ii) an inner secondary air channel, which is arranged around the outside of the primary air pulverized coal channel, with adjustable axial swirl blades inside the channel to guide the inner secondary air flow to form a central high-temperature backflow zone; (iii) an outer secondary air channel, which is arranged around the outside of the inner secondary air channel, with adjustable axial swirl blades inside the channel to supplement the air required for pulverized coal combustion; and (iv) a plasma ignition device, which is integrated into the burner nozzle area.

3. The control system according to claim 1, characterized in that, The second type of swirl pulverized coal burner is used to form a strong reducing atmosphere in the middle and top layers of the furnace and to achieve pulverized coal burnout control. It includes at least: (i) a central direct current air channel, which is set in the central area of ​​the burner and extends axially to transport central direct current air; (ii) a primary air pulverized coal channel, which is arranged coaxially around the central direct current air channel and whose inlet end is connected to the primary air inlet elbow, and whose outlet end forms an annular primary air pulverized coal nozzle; (iii) an inner secondary air channel, which is arranged around the outside of the primary air pulverized coal channel and has fixed axial swirl blades in the channel to form a weak swirl airflow in the inner layer; and (iv) an outer secondary air channel, which is arranged around the outside of the inner secondary air channel and has adjustable axial swirl blades in the channel to form an outer secondary air with adjustable swirl intensity, enhance peripheral shear mixing, and supplement the air required for burnout.

4. The control system according to claim 1, characterized in that, The OFA nozzle is located above the main combustion zone of the furnace and is used to construct a post-combustion zone with staged air supply. It includes at least: (i) a central direct current air channel, which is arranged at the center of the nozzle and extends axially to spray direct current secondary air and form an axial penetrating airflow above the main combustion zone; (ii) an outer ring swirl air channel, which is arranged around the outside of the central direct current air channel. The channel is equipped with adjustable axial swirl blades to generate an outer ring swirl secondary air with adjustable swirl intensity, optimize the penetration and mixing intensity of the combustion air, and ensure complete combustion of pulverized coal.

5. The control system according to claim 1, characterized in that, The online coal quality monitoring unit employs a multimodal sensor fusion analysis device to measure the ash content, moisture, volatile matter, and lower heating value of the coal fed into the furnace in real time. It also uses self-calibration and fault diagnosis algorithms to eliminate and compensate for drift and abnormal data, and finally outputs coal quality characteristic parameters for control. When the detected ash content is greater than 25% and the calorific value is less than 18 MJ / kg, the system is triggered to enter the stratified differential air distribution mode; when the coal quality parameters recover to the set upper limit, it returns to the conventional equal air distribution mode.

6. The control system according to claim 1, characterized in that, The combustion state sensing unit includes a layered array of wide-range oxygen sensors, a high-sensitivity temperature sensor array, and an online NOx concentration monitoring device installed at the furnace outlet or tail flue, forming a full-process combustion state sensing network covering the main combustion zone, burnout zone, and flue gas emission path. The collected data is then mapped and feature extracted to form a linked observation of oxygen field, temperature field, and outlet NOx.

7. The control system according to claim 1, characterized in that, The air volume and pressure monitoring unit includes high-precision air volume sensors, air pressure transmitters, and valve position / damper angle feedback devices installed on the primary air, secondary air, and OFA ducts of each type I swirl pulverized coal burner, type II swirl pulverized coal burner, and OFA nozzle, for real-time monitoring of the air volume and resistance characteristics of each duct.

8. The control system according to claim 1, characterized in that, The coal quality adaptive decision-making module is used to generate feedforward targets for stratified air distribution and swirl settings under coal quality fluctuation conditions. Its built-in coal quality-air distribution mapping model is trained using supervised learning or ensemble learning methods based on machine learning. The model's input is the coal quality characteristic parameters output by the online coal quality monitoring unit, and the output includes the stratified excess air coefficient target, OFA air volume ratio target, and stratified swirl blade angle target. When determining ash content online... 25% and low calorific value When the coal quality reaches 18 MJ / kg, the stratified air distribution mode for low-quality coal is triggered; when the coal quality recovers to the set upper limit and meets the hysteresis condition, it smoothly switches back to the normal equal air distribution mode.

9. The control system according to claim 8, characterized in that, The coal quality-air distribution mapping model is trained using supervised learning or ensemble learning methods based on machine learning. The coal quality characteristic parameter vector under historical operating conditions is used as the input feature, and the excess air coefficient, OFA air volume ratio, and inner and outer swirl blade angle of each layer obtained by expert experience optimization or offline multi-objective optimization under the corresponding operating conditions are used as the output labels for supervised learning training. Under the differential air distribution mode for low-quality coal, the mapping model recommends strengthening the lower air distribution ratio to prolong the residence time of low-quality coal in the high-temperature zone and improve the burnout, while weakening the upper air distribution ratio and increasing the proportion of OFA air volume to establish a deep-stage combustion structure, thereby reducing NOx generation concentration while ensuring stable combustion.

10. The control system according to claim 1, 8, or 9, characterized in that, The multi-objective optimization control module uses the feedforward objective and mode state of the coal quality adaptive decision module as initial values ​​and soft constraints to construct an optimization problem with maximizing combustion efficiency and minimizing NOx emissions as the main objectives and constraining furnace temperature distribution uniformity and fly ash carbon content as secondary objectives. The decision variables include at least the opening degree of each layer damper and the angle of each layer swirl blade. The hard constraints include at least the fan capacity, duct pressure drop, burner safe operation window and load command. The module uses predictive control or fuzzy inference algorithm to solve the problem in real time and output the adjustment commands for each damper opening degree and blade angle, and outputs the executable target trajectory for the execution unit to track.

11. The control system according to claim 1, characterized in that, The combustion state closed-loop feedback module compares the received real-time combustion state monitoring data with the set value, and performs small-step fine-tuning and amplitude-limiting correction on the multi-objective optimization solution based on deviation and trend. When NOx is observed to be close to the upper limit or there is a trend of local overheating or oxygen deficiency, the OFA ratio and the external secondary air blade angle are adjusted, and the opening of each layer of dampers is corrected simultaneously to suppress disturbances and parameter drift. When the unit changes load rapidly, the linkage adjustment weight of OFA swirl blade angle and upper external secondary air guide vane angle is increased to suppress flame retraction or deflection and maintain a low NOx level. If a sensor abnormality or data quality alarm is detected, the current optimization quantity is frozen and switched to a conservative air distribution curve. After the observation recovers, the optimal trajectory is asymptotically returned.

12. The control system according to claim 1, characterized in that, The execution unit includes an electric or pneumatic servo mechanism connected to the primary air, secondary air, and OFA duct dampers of each layer, and an angle servo actuator connected to the adjustable swirl blades of each burner and OFA nozzle; and the execution unit has position closed-loop and speed feedforward functions to complete the tracking and adjustment of air volume distribution and blade angle within a small step size and short cycle, and realizes jam detection and fault switching through status monitoring.

13. A method for optimizing and controlling low-NOx combustion of inferior coal in a large power plant boiler, based on the control system described in any one of claims 1 to 12, characterized in that, At least the following steps are included: SS1. Real-time measurement of ash, moisture, volatile matter and calorific value of coal fed into the furnace and generation of coal quality characteristic parameters; real-time acquisition of oxygen concentration, temperature distribution and NOx concentration at the furnace outlet in each combustion zone of the boiler; real-time monitoring of flow rate and resistance characteristics of each primary air, secondary air and OFA duct. SS2. When the ash content is detected to be greater than 25% and the lower heating value is... When the coal quality reaches 18 MJ / kg, the low-quality coal stratified air distribution mode is triggered, generating the excess air coefficient, OFA ratio, and secondary air blade angle suggestions for each layer; when the coal quality recovers to the set upper limit and meets the hysteresis and minimum residence time, it switches back to the normal equal air distribution mode. SS3. In the low-quality coal stratified air distribution mode, an optimization problem is constructed with the main objectives of maximizing combustion efficiency and minimizing NOx emissions, and the secondary objectives of furnace temperature uniformity and fly ash carbon content. The decision variables include at least the opening degree of each stratified damper and the swirl blade angle. The hard constraints include at least the fan capacity, duct pressure drop, burner safe operation window and load command. The system solves and outputs the adjustment commands for each damper opening degree and blade angle in real time, and outputs the executable target trajectory for the execution unit to track. SS4. Issue adjustment commands for the opening degree of each damper and the blade angle to the execution unit to complete the coordinated allocation of each combustion zone; Based on continuous feedback of oxygen concentration, temperature distribution and NOx concentration at furnace outlet, the multi-objective optimization solution is fine-tuned with small steps and limited correction to maintain stable combustion and ultra-low emissions. SS5. When flame retraction, local oxygen enrichment or supercooling, NOx exceeding the limit or insufficient sensor health are detected, interlocking and safety backoff are triggered, the preset safety air distribution curve is activated and the rate of change and boundary are tightened; when switching from conventional equal air distribution to differential air distribution for inferior coal or back-cutting in the opposite direction, the smooth switching is completed according to the preset gradual trajectory and minimum residence time.

14. The control method according to claim 13, characterized in that, In step SS2, the coal quality characteristic determination and air distribution mode triggering further introduce hysteresis criteria and minimum residence time strategies. A dual-threshold range is used to manage mode switching for ash content and lower heating value. When the ash content is detected to be greater than the first threshold by 25% and the lower heating value... When the second threshold is 18 MJ / kg, and the duration of this coal quality condition exceeds the coal quality determination delay threshold, the inferior coal stratification differential air distribution mode is triggered; when the coal quality recovers to the ash content... The third threshold is 22% or the lower heating value. When the coal quality reaches the fourth threshold of 20 MJ / kg and the duration of this coal quality state exceeds the minimum residence time for coal quality recovery, the hysteresis judgment condition is met, triggering a switchback to the conventional equal air distribution mode. Among them, the ash content judgment hysteresis interval is formed between the first and third thresholds, and the calorific value judgment hysteresis interval is formed between the second and fourth thresholds. At the same time, the maximum switching frequency and minimum residence time constraints are applied to the mode switching, and the switching transition process generates a gradual trajectory of the stratified excess air coefficient and OFA ratio.

15. The control method according to claim 13, characterized in that, In step SS3, the multi-objective optimization problem is solved with model predictive control as the core, simultaneously optimizing the allocation of primary air, internal and external secondary air, OFA, and the swirl blade angle of each layer in the rolling time domain. The cost function includes negative terms for thermal efficiency, NOx emission, CO and fly ash carbon content penalties, temperature uniformity and flame stability penalties, and an adaptive redistribution mechanism for objective weights is set. Hard constraints include fan power and head, duct pressure drop, nozzle momentum-to-flux ratio, minimum stable combustion air volume for burners, upper limits for wall temperature and furnace temperature rise, and upper limit for load tracking error.

16. The control method according to claim 15, characterized in that, In step SS3, under the differential air distribution mode for low-quality coal stratification, an asymmetric air distribution template and a variable slope constraint are added. A higher minimum air volume and a smaller allowable downward adjustment slope are set for the bottom stable combustion zone, and a lower target excess air coefficient and a restriction on its upward adjustment slope are set for the upper main combustion zone. At the same time, linkage constraints are set for OFA ratio and external secondary air swirl angle to maintain the nozzle momentum flux ratio. Under the conventional equal air distribution mode, a stratified air distribution deviation penalty is introduced.

17. The control method according to claim 13, characterized in that, In step SS4, the coordinated allocation strategy for each combustion zone is as follows: The bottom stable combustion zone corresponds to the lower part of the furnace, and an enhanced air distribution strategy is adopted to increase the primary air velocity and secondary air volume, improve the local oxygen concentration and combustion temperature, and ensure the rapid ignition and complete release of volatile matter from low-quality coal; the middle main combustion zone corresponds to the middle part of the furnace, and a moderate air distribution strategy is adopted to maintain a low oxygen concentration environment and appropriately reduce the swirl intensity, prolong the residence time of pulverized coal in the high-temperature zone, and promote the complete combustion of fixed carbon; the upper low-NOx zone corresponds to the upper part of the furnace, and a weakened air distribution strategy is adopted to reduce the secondary air volume and lower the angle of the outer ring swirl blades, maintain an oxygen-deficient environment, and inhibit the formation of nitrogen oxides (NOx); the top post-burnout zone adopts an enhanced OFA air distribution strategy, gradually supplementing oxygen in layers, with enhanced oxygen supplementation in the lower layer and moderate oxygen supplementation in the upper layer, ensuring the burnout of fly ash while avoiding excessive oxygen leading to a rebound in NOx emissions, thus achieving synergistic optimization of stable combustion and low NOx emissions.

18. The control method according to claim 13 or 17, characterized in that, In step SS4, the combustion state closed-loop feedback adopts a rolling optimization fine-tuning mechanism. Based on the oxygen concentration, temperature distribution, and furnace outlet NOx concentration continuously fed back by the combustion state sensing unit, the deviation between the actual and target values ​​of performance indicators is calculated in real time, including combustion efficiency deviation, nitrogen oxide emission deviation, temperature distribution uniformity deviation, and fly ash carbon content deviation. When the performance deviation exceeds the preset threshold or load disturbance or coal quality fluctuation is detected, the rolling optimization fine-tuning process is triggered. Priority is given to dynamically adjusting the OFA ratio and the external secondary air blade angle, which have a fast response speed, implementing small step adjustments and amplitude limit corrections. Simultaneously, the opening of the dampers in each layer with a slower response speed is fine-tuned. Through the hierarchical adjustment strategy of fast and slow variables, the control stability is maintained while responding quickly to changes in operating conditions.

19. The control method according to claim 13, characterized in that, In step SS5, the safety protection strategies include flame retraction detection, local oxygen enrichment or supercooling criteria, NOx and CO dual threshold linkage, wall temperature rate protection, and sensor degradation identification. When any safety protection condition is triggered, the system enters the safety rollback mode, converges according to the preset safety air distribution curve, and freezes some weights and tightens the constraint boundaries. The safety rollback release requires the satisfaction of multi-variable joint recovery conditions and minimum dwell time, and achieves smooth mode switching through preset gradual trajectory to avoid thermal stress shock and temperature oscillation.

20. The control method according to claim 13, characterized in that, It also includes the adaptive update and performance maintenance step SS6, which uses closed-loop operating data to perform online incremental learning or periodic offline retraining on the coal quality-air distribution mapping model; and periodically re-optimizes the weights and active constraint set based on real-time deviation and disturbance estimates, and updates the mode threshold and hysteresis bandwidth. Through a synergistic adaptive approach of feedforward, optimization, and feedback, the goals of low NOx emissions, low fly ash, and high boiler efficiency under low-quality coal conditions are continuously achieved.

Images