A method and system for precise control of air-fuel ratio of a combustor based on hierarchical optimization

Abstract

The present application relates to the technical field of combustion control, in particular to a kind of burner air-fuel ratio precision control method and system based on layered optimization, the present application is by under the condition of flow closed loop operation acquisition multi-working condition data, constructs the feedforward control model with actuator control quantity and pipeline pressure as input, and through model verification ensures prediction accuracy.In the operating phase, using the layered control structure of monitoring optimization layer, setting decision layer, safety coordination layer and closed loop execution layer, the coordinated control of gas flow and air flow is realized.The setting decision layer generates flow set value based on the feedforward model, and combines the slow integral fine adjustment of flue gas oxygen content;The safety coordination layer ensures the combustion safety in the process of load change through cross limiting strategy;Closed loop execution layer completes flow accurate tracking control.At the same time, the smooth transition of control mode is realized through non-disturbance switching mechanism.The method can effectively improve the air-fuel ratio control precision, system response speed and operation safety.

Description

Technical Field

[0001] This invention relates to the field of combustion control technology, specifically to a method and system for precise control of the air-fuel ratio of a burner based on hierarchical optimization. Background Technology

[0002] In industrial combustion processes, precise control of the air-fuel ratio is crucial for improving combustion efficiency, reducing pollutant emissions, and ensuring safe equipment operation. Existing combustion control schemes suffer from the following shortcomings: First, while proportional linkage schemes using mechanical linkages or electronic follower circuits are simple in structure, the fixed and unadjustable air-fuel ratio cannot adapt to fluctuations in fuel calorific value, resulting in weak anti-disturbance capabilities and limited control precision. Second, closed-loop control schemes using flue gas oxygen content analyzers as feedback signals, while ensuring complete combustion under steady-state conditions, suffer from slow dynamic response due to the large pure time lag characteristic of online flue gas analysis. Under conditions of frequent load changes, the adjustment quality significantly deteriorates, making it difficult to suppress rapid disturbances. Third, schemes that equip both gas and air pipelines with online flow meters to achieve high control precision, while offering higher accuracy, significantly increase system hardware costs and long-term maintenance burdens, resulting in poor economic efficiency. Furthermore, existing systems generally use analog signal transmission, and system errors introduced by range setting errors, line losses, and electromagnetic interference cannot be ignored, further restricting the improvement of control precision. Regarding dynamic safety, existing solutions generally lack dedicated protection logic to prevent insufficient air supply (i.e., fuel-rich state) under rapid load changes, posing significant risks to dynamic safety. In summary, existing technologies struggle to simultaneously achieve high precision, high safety, strong adaptability, and good economy, necessitating a novel combustion control technology solution that overcomes these shortcomings.

[0003] To address this, a method and system for precise control of the air-fuel ratio in a burner based on hierarchical optimization is proposed. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for precise control of the air-fuel ratio of a burner based on hierarchical optimization. By constructing a feedforward control model and combining it with a hierarchical control structure, the method achieves coordinated regulation of gas and air flow. Furthermore, it improves control accuracy and system safety through oxygen content feedback fine-tuning and a safety limiting mechanism.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A method for precise control of burner air-fuel ratio based on hierarchical optimization includes: S1. When the combustion system is in a closed-loop flow operation state, collect load commands, actuator control quantities, measured flow values ​​and pipeline pressures under different load conditions. Based on the collected multi-dimensional data, construct a feedforward control model that describes the mapping relationship between actuator control quantities and flow. S2. Switch the control system to an operating mode based on a feedforward control model, and implement hierarchical optimization control of the air-fuel ratio through a four-layer hierarchical structure: The monitoring and optimization layer transmits the load setpoints to the setting and decision-making layer; The decision-making layer calculates the setpoints for gas flow and air flow based on the feedforward control model, and slowly fine-tunes the two setpoints based on the flue gas oxygen content to compensate for slow disturbances in fuel characteristics. The safety coordination layer implements cross-limiting processing for the gas flow setpoint and the air flow setpoint to ensure that the adjustment of the air flow setpoint precedes the gas flow setpoint during load changes, and outputs a safe setpoint. The closed-loop execution layer receives safety setpoints and forms independent closed-loop control for gas flow and air flow respectively.

[0006] Preferably, in step S1, the data acquisition process in the feedforward control model establishment stage is as follows: the data changes progressively in a step-by-step manner throughout the entire stroke range. After the system reaches a stable state under each step condition, the current actuator control quantity, measured flow rate, and pipeline pressure are recorded synchronously to form a multi-dimensional dataset covering the entire operating range. Using the actuator control quantity and pipeline pressure as input dimensions and the measured flow rate as output dimension, a segmented mapping feedforward control model is constructed based on the multi-dimensional dataset.

[0007] Preferably, a model verification step is included between steps S1 and S2: Under the closed-loop operation state of the flow, the predicted flow value calculated by the feedforward control model based on the current actuator control quantity and pipeline pressure is compared with the measured flow value, and the deviation between the predicted flow value and the measured flow value is statistically analyzed within the entire operating range; when the deviation within the entire operating range is lower than the preset accuracy threshold, the feedforward control model is determined to have passed the verification, and the process enters the execution stage switching in step S2; if there are operating points where the deviation exceeds the accuracy threshold, the process returns to step S1 to collect data for the corresponding operating conditions and revise the feedforward control model.

[0008] Preferably, the switching process from step S1 to step S2 adopts a non-disruptive switching method: before the switch, the output value of the feedforward control model continuously tracks the set value of the current flow closed-loop control, so that the output value of the feedforward control model is consistent with the flow closed-loop set value; when the switch is executed, the control power is transferred from the flow closed-loop mode to the feedforward model mode, and the gas flow set value and air flow set value before and after the switch remain continuous without generating a step; after the switch is completed, the temporary flow detection signal exits the control loop and enters the operation mode with the feedforward control model as the main mode and the flue gas oxygen content feedback fine adjustment as the auxiliary mode.

[0009] Preferably, the cross-limiting process of the safety coordination layer in step S2 is as follows: the safety coordination layer continuously judges the direction of change of the load command; when the load command is on an upward trend, the safety coordination layer first increases the air flow setpoint to the target value and adds a positive safety margin, and after the air flow closed-loop response confirms that the target range has been reached, the gas flow setpoint is then increased to the target value; when the load command is on a downward trend, the safety coordination layer first decreases the gas flow setpoint to the target value, and after the gas flow closed-loop response confirms that the target range has been reached, the air flow setpoint is then decreased to the target value; the output result of the cross-limiting process is passed down to the closed-loop execution layer as the safety setpoint.

[0010] Preferably, in step S2, the process of setting the decision layer to fine-tune the output value of the feedforward control model using flue gas oxygen content is as follows: taking the deviation between the measured value of flue gas oxygen content and the target oxygen content as the input signal, performing time integration on the deviation, and superimposing the integration result as a correction amount onto the air-fuel ratio reference value output by the feedforward control model; the time constant of the integration operation is much larger than the time scale of the load dynamic response, so that the correction amount only responds to the long-term drift of fuel characteristics without interfering with the rapid dynamic adjustment of load changes; when the correction amount exceeds the preset limit range, the correction amount limit protection is triggered to prevent the air-fuel ratio reference value from being erroneously adjusted due to abnormal flue gas oxygen content detection.

[0011] Preferably, the feedforward control model is stored in a combination of a segmented lookup table and a pipeline pressure correction factor: the lookup table uses the actuator control quantity as an index to record the reference flow value corresponding to each control quantity under standard pressure conditions; the pipeline pressure correction factor takes the difference between the current pipeline pressure and the standard reference pressure as input and outputs the correction amount for the reference flow value; when the decision layer calls the feedforward control model, it queries the reference flow value based on the actuator control quantity, and then superimposes the correction amount output by the pipeline pressure correction factor to obtain the predicted flow value under the current pipeline pressure conditions, and uses the predicted flow value as the basis for the flow setpoint transmitted to the safety coordination layer.

[0012] A precise air-fuel ratio control system for burners based on hierarchical optimization includes: a feedforward model building module, used to collect load commands, actuator control quantities, measured flow values ​​and pipeline pressures under different load conditions when the combustion system is in a flow closed-loop operation state, and to construct a feedforward control model describing the mapping relationship between actuator control quantities and flow based on the collected multi-dimensional data; The hierarchical optimization operation module is used to switch the control system to an operation mode based on the feedforward control model, and implements hierarchical optimization control of air-fuel ratio through a four-layer hierarchical structure: The monitoring and optimization layer transmits the load setpoints to the setting and decision-making layer; The decision-making layer calculates the setpoints for gas flow and air flow based on the feedforward control model, and slowly fine-tunes the two setpoints based on the flue gas oxygen content to compensate for slow disturbances in fuel characteristics. The safety coordination layer implements cross-limiting processing for the gas flow setpoint and the air flow setpoint to ensure that the adjustment of the air flow setpoint precedes the gas flow setpoint during load changes, and outputs a safe setpoint. The closed-loop execution layer receives safety setpoints and forms independent closed-loop control for gas flow and air flow respectively.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention achieves high-precision prediction and decoupled control of gas flow and air flow by constructing a feedforward control model based on actuator control quantity and pipeline pressure under closed-loop flow operation. Compared with traditional single feedback control methods, it can compensate for the effects of system nonlinearity and operating condition changes in advance, significantly reducing regulation lag and overshoot. Under multi-load conditions, the feedforward model can directly provide the matching flow setpoint, reducing dependence on feedback signals, improving system response speed and control stability, thereby making air-fuel ratio control more precise, effectively improving combustion efficiency and reducing energy consumption.

[0014] 2. This invention constructs a four-layer hierarchical control structure: a monitoring and optimization layer, a setting and decision-making layer, a safety coordination layer, and a closed-loop execution layer. This achieves hierarchical decoupling of control objectives and clear functional division. In particular, through the cross-limiting mechanism of the safety coordination layer, it ensures a strict sequence of air and gas regulation during load changes, fundamentally avoiding safety risks such as fuel overload or oxygen deficiency during combustion. Compared to the simple limiting or single-loop coordination methods in existing technologies, this structure has higher safety redundancy and control reliability under dynamic operating conditions, significantly improving the intrinsic safety level of the combustion system.

[0015] 3. This invention introduces a slow integral fine-tuning mechanism based on flue gas oxygen content, compensating only for long-term fuel characteristic drift without interfering with rapid dynamic adjustment due to load changes, thus achieving an optimized control strategy of "separation of fast and slow" control. Simultaneously, it combines model verification and a disturbance-free switching mechanism, performing full-condition accuracy verification before the model is put into operation, and achieving smooth switching through tracking consistency, avoiding the impact of control mode switching. Compared to existing technologies using direct switching or single closed-loop correction methods, this invention significantly improves the system's adaptability and robustness, ensuring stable and reliable air-fuel ratio control accuracy under long-term operation. Attached Figure Description

[0016] Figure 1 A schematic diagram of the combustor air-fuel ratio precision control method based on hierarchical optimization provided in an embodiment of the present invention; Figure 2A schematic diagram of the combustor air-fuel ratio precision control system based on hierarchical optimization provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the slow integral fine-tuning logic flow for flue gas oxygen content provided in an embodiment of the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention.

[0018] Example 1: Please see Figure 1 and Figure 2 This embodiment provides a method for precise air-fuel ratio control of a burner based on hierarchical optimization. It is applicable to precise air-fuel ratio control systems for burners based on hierarchical optimization. The burner is equipped with a gas regulating valve, an air regulating valve, a gas pipeline pressure transmitter, an air pipeline pressure transmitter, and a flue gas oxygen content analyzer installed in the flue. Each transmitter at the detection end supports digital communication protocols. The specific technical solution is as follows: a feedforward model building module is used to collect load commands, actuator control quantities, measured flow values, and pipeline pressures under different load conditions when the combustion system is in a flow closed-loop operation state. Based on the collected multi-dimensional data, a feedforward control model describing the mapping relationship between actuator control quantities and flow is constructed. The hierarchical optimization operation module is used to switch the control system to an operation mode based on the feedforward control model. It implements hierarchical optimization control of the air-fuel ratio through a four-layer hierarchical structure: the monitoring and optimization layer transmits the load setpoint to the setting decision layer; the setting decision layer calculates the gas flow setpoint and air flow setpoint based on the feedforward control model, and slowly fine-tunes the two setpoints based on the flue gas oxygen content to compensate for slow disturbances in fuel characteristics; the safety coordination layer performs cross-limiting processing on the gas flow setpoint and air flow setpoint to ensure that the adjustment of the air flow setpoint precedes the gas flow setpoint during load changes, and outputs a safety setpoint; the closed-loop execution layer receives the safety setpoint and forms independent closed-loop control for the gas flow and air flow respectively.

[0019] Furthermore, the data acquisition process in the feedforward control model establishment phase is as follows: the data changes progressively in a step-by-step manner throughout the entire stroke range. After the system reaches a stable state under each step of the operating condition, the current actuator control quantity, measured flow rate, and pipeline pressure are recorded synchronously to form a multi-dimensional dataset covering the entire operating condition range. Using the actuator control quantity and pipeline pressure as input dimensions and the measured flow rate as output dimension, a segmented mapping feedforward control model is constructed based on the multi-dimensional dataset.

[0020] Specifically, the feedforward control model establishment phase is carried out when the combustion system is in a flow closed-loop operation state. At this time, both the gas flow control loop and the air flow control loop are working normally, the flow sensor signal is effectively connected to the control system, and the system is operating in a normal process state.

[0021] The gas regulating valve is controlled by a monitoring system to gradually change its opening in a stepped manner across its entire stroke range, starting from the minimum opening and increasing step by step to the maximum opening. The step size is determined before commissioning based on the actuator's adjustment resolution and operating condition coverage requirements. After reaching each step, the system determines the stability condition: within a continuous sampling period (the length of which is determined based on the dynamic response characteristics of the controlled object, typically three to five times the system's step response time constant, measured by the commissioning engineer during system step response testing), if the changes in the measured flow and pressure values ​​are both lower than the stability criterion ratio for the corresponding range (this ratio is determined by process technicians based on control accuracy requirements, usually 0.5% to 1% of the corresponding range), the system is considered to have entered a steady state.

[0022] After steady-state confirmation, the current gas regulating valve control quantity, measured gas flow rate, and gas pipeline pressure are recorded synchronously to form a data sample point. The above process is repeated for the air-side actuator to collect an air-side data sample set. After traversing all step-by-step operating conditions, a multi-dimensional dataset covering the entire operating range of both the gas and air sides is obtained. Each sample point contains data in three dimensions: actuator control quantity, corresponding pipeline pressure, and corresponding measured flow rate.

[0023] Furthermore, the feedforward control model is stored in a combination of a segmented lookup table and a pipeline pressure correction factor: the lookup table uses the actuator control quantity as an index to record the reference flow value corresponding to each control quantity under standard pressure conditions; the pipeline pressure correction factor takes the difference between the current pipeline pressure and the standard reference pressure as input and outputs the correction amount for the reference flow value; when the decision layer calls the feedforward control model, it queries the reference flow value based on the actuator control quantity, and then superimposes the correction amount output by the pipeline pressure correction factor to obtain the predicted flow value under the current pipeline pressure conditions, and uses the predicted flow value as the basis for the flow setpoint transmitted to the safety coordination layer.

[0024] The feedforward control model is stored using a combination of a piecewise lookup table and a pipeline pressure correction factor. The lookup table uses the actuator control quantity as an index to store the baseline flow rate value corresponding to each control quantity under the standard reference pressure condition. The intermediate value between adjacent index points is obtained through linear interpolation. The standard reference pressure is selected as the pipeline pressure value with the highest frequency of occurrence during the data acquisition phase, ensuring that the pressure correction amount is close to zero under most operating conditions, reducing the impact of correction errors. The pipeline pressure correction factor takes the difference between the current pipeline pressure and the standard reference pressure as input and outputs the correction amount for the baseline flow rate value. The parameters of the correction factor are determined by fitting sample data obtained under different pressure conditions during the data acquisition phase.

[0025] Further, the model verification steps are as follows: Under the closed-loop operation state, the predicted flow value calculated by the feedforward control model based on the current actuator control quantity and pipeline pressure is compared with the measured flow value, and the deviation between the predicted flow value and the measured flow value is calculated within the entire operating range; when the deviation within the entire operating range is lower than the preset accuracy threshold, the feedforward control model is determined to have passed the verification and enters the hierarchical optimization operation execution stage; if there are operating points where the deviation exceeds the accuracy threshold, the feedforward model is returned to supplement the data of the corresponding operating conditions and the feedforward control model is corrected again.

[0026] Specifically, after the feedforward control model is constructed, the system maintains a closed-loop flow operation state to verify the model's accuracy under all operating conditions. The feedforward control model calculates the predicted flow rate in real time based on the current actuator control quantity and pipeline pressure. This predicted flow rate is then compared point by point with the measured value from the flow sensor, and the prediction deviation at each operating point within the entire operating range is statistically analyzed.

[0027] The method for obtaining the preset accuracy threshold is as follows: the allowable error range of air-fuel ratio control is determined according to the combustion system process specifications, and the allowable error is converted to the flow deviation dimension. The conversion is based on the product relationship between the target air-fuel ratio coefficient and the flow range. The boundary value of the conversion result is used as the accuracy threshold, which is confirmed by process technicians in the system commissioning documents according to combustion efficiency and emission compliance requirements.

[0028] When the prediction deviation of all operating points within the full operating range is lower than the accuracy threshold, the feedforward control model is deemed to have passed verification and enters the switching preparation stage. If there are operating points that exceed the accuracy threshold, additional data samples are collected near those operating points, the baseline flow rate value and pressure correction factor parameters of the corresponding operating point in the lookup table are corrected, and verification is performed again. This process is repeated until the accuracy across all operating conditions meets the requirements.

[0029] Furthermore, the switching process adopts a non-disruptive switching method: before the switch, the output value of the feedforward control model continuously tracks the set value of the current flow closed-loop control, so that the output value of the feedforward control model is consistent with the flow closed-loop set value; when the switch is executed, the control power is transferred from the flow closed-loop mode to the feedforward model mode, and the gas flow set value and air flow set value before and after the switch remain continuous without generating a step; after the switch is completed, the temporary flow detection signal exits the control loop and enters the operation mode with the feedforward control model as the main mode and the flue gas oxygen content feedback fine adjustment as the auxiliary mode.

[0030] Specifically, in the preprocessing stage before switching, the output of the feedforward control model continuously tracks the set value of the current flow closed-loop controller in a software tracking manner, so that the output value of the feedforward control model and the flow closed-loop set value remain consistent at any time, and the tracking deviation is within the resolution range of the control system.

[0031] During the switching process, control is transferred from the flow closed-loop controller to the feedforward control model. The gas flow setpoint and air flow setpoint do not change abruptly before and after the switching, maintaining a continuous transition.

[0032] After the switchover is complete, the pipeline flow sensor signal, which was originally used to build the feedforward model, is removed from the control calculation loop. The flow sensor remains online for process monitoring, but is no longer used as a control calculation input. The system enters normal operation mode with the feedforward control model as the primary source of setpoint calculations and flue gas oxygen content feedback integral fine-tuning as an auxiliary correction method.

[0033] Furthermore, the cross-limiting process of the safety coordination layer is as follows: the safety coordination layer continuously judges the direction of load command changes; when the load command shows an upward trend, the safety coordination layer first increases the air flow setpoint to the target value and adds a positive safety margin. After the air flow closed-loop response confirms that the target range has been reached, the gas flow setpoint is then increased to the target value; when the load command shows a downward trend, the safety coordination layer first decreases the gas flow setpoint to the target value. After the gas flow closed-loop response confirms that the target range has been reached, the air flow setpoint is then decreased to the target value; the output result of the cross-limiting process is passed down to the closed-loop execution layer as the safety setpoint.

[0034] Furthermore, the process of fine-tuning the output value of the feedforward control model using the oxygen content of flue gas is defined, referring to... Figure 3Specifically, the deviation between the measured value of flue gas oxygen content and the target oxygen content is used as the input signal. The deviation is integrated over time, and the integration result is added to the air-fuel ratio reference value output by the feedforward control model as a correction. The time constant of the integration operation is much larger than the time scale of the load dynamic response, so that the correction only responds to the long-term drift of fuel characteristics without interfering with the rapid dynamic adjustment of load changes. When the correction exceeds the preset limit range, the correction limit protection is triggered to prevent the air-fuel ratio reference value from being erroneously adjusted due to abnormal flue gas oxygen content detection.

[0035] Specifically, the system implements stratified air-fuel ratio optimization control through the following four-layer hierarchical structure: The monitoring and optimization layer receives load setpoints from the upper-level production scheduling system, expresses them in terms of thermal power or equivalent dimensions specified by the process, and transmits the load setpoints down to the setting decision layer. At the same time, it monitors the overall operating status of the system and records historical data.

[0036] After receiving the load setpoint, the decision-making layer calls the feedforward control model for calculation: First, the current gas regulating valve control quantity is used as an index to look up the table and obtain the reference gas flow rate value under the standard reference pressure condition; the current gas pipeline pressure is read, the difference between it and the standard reference pressure is calculated, and the correction amount is output through the pressure correction factor. The correction amount is added to the reference gas flow rate value to obtain the gas flow rate prediction value under the current operating condition, which is used as the gas flow rate setpoint; the air flow rate setpoint is obtained by multiplying the gas flow rate setpoint by the target air-fuel ratio coefficient determined by the process specification.

[0037] The decision-making layer is set to simultaneously receive the detection signal from the flue gas oxygen content analyzer. The difference between the measured value of flue gas oxygen content and the target oxygen content is used as the deviation signal. Time integration is performed on the deviation signal, and the integration result is superimposed on the gas flow rate setpoint and air flow rate setpoint as a correction amount to compensate for the slow drift of fuel characteristics.

[0038] The implementation of time integration is as follows: In each control cycle, the product of the current deviation value and the control cycle duration is added to the cumulative integral value of the previous cycle, and this process is repeated cyclically. The time constant for integration is determined as follows: In a step response test of the combustion system, the dynamic response time from the issuance of the load command to the completion of the flow response is measured. A value at least ten times this dynamic response time is taken as the lower limit of the integration time constant, ensuring that the change in correction amount is negligible during a complete load dynamic process. The specific value is determined by the control engineer based on the system dynamic characteristic test results and the typical timescale of fuel characteristic drift, typically ranging from several minutes to tens of minutes, and is confirmed in the control system commissioning documents.

[0039] The preset limit range is obtained by converting the maximum air-fuel ratio deviation that may be caused by fuel characteristics within the normal fluctuation range into the corresponding maximum reasonable flow correction amount, with the boundary value of this range serving as the limit value. When the absolute value of the correction amount exceeds the limit value, the integrator stops accumulating, the correction amount remains at the limit boundary value and no longer increases, and an alarm signal is sent to the monitoring layer, prompting the operator to check the working status of the flue gas oxygen content analyzer. Exceeding the limit range usually indicates an abnormality in the flue gas oxygen content detection system, rather than normal fuel characteristic drift.

[0040] The safety coordination layer receives the gas flow setpoint and air flow setpoint output by the setting decision layer, continuously judges the direction of change of the load command transmitted by the monitoring and optimization layer, and performs cross-limiting processing on the two setpoints.

[0041] When the load command shows an upward trend, the safety coordination layer first raises the air flow safety setpoint to the target air flow value and adds a positive safety margin (the positive safety margin is determined based on the maximum tracking deviation measured in the dynamic response test of the air flow closed-loop control loop, and is taken as 1.2 times the maximum tracking deviation to ensure that the measured air flow value always leads the gas flow target value under the safety margin protection). Then, the measured air flow value is continuously monitored. When it enters the confirmation bandwidth centered on the target air flow value and twice the measurement accuracy of the flow sensor (determined by the sensor technical specifications), it is determined that the air flow response has met the standard. After that, the gas flow safety setpoint is raised to the target gas flow value, completing the load increase cross-limiting.

[0042] When the load command shows a downward trend, the safety coordination layer first reduces the gas flow safety setpoint to the target gas flow value. After the actual gas flow value enters the aforementioned confirmation bandwidth, the air flow safety setpoint is then reduced to the target air flow value, completing the load reduction cross-limiting. The output result of the cross-limiting processing is passed down to the closed-loop execution layer as the safety setpoint.

[0043] The closed-loop execution layer receives the gas flow safety setpoint and air flow safety setpoint from the safety coordination layer, and forms independent actuator position closed-loop control loops for the gas regulating valve and the air regulating valve respectively. The gas side and the air side are adjusted independently and do not interfere with each other.

[0044] Each transmitter at the detection end (including pressure transmitters and flue gas oxygen content analyzers) transmits process detection values ​​directly to the data acquisition module at the decision-making level via fieldbus or industrial Ethernet digital communication protocol, in the form of engineering physical quantities (expressed in corresponding engineering units and actual values). Control quantities issued by the control layer to the gas regulating valve and air regulating valve are also transmitted via digital communication links in the form of engineering physical quantities. Digital communication eliminates the deviations in values ​​caused by impedance changes, signal attenuation, and induced interference along the transmission path of analog signals, ensuring consistency between the transmitted data and the actual engineering physical quantities.

[0045] Example 2: This embodiment, building upon the fundamental technical solutions described in Embodiment 1, including the four-layer hierarchical control architecture, feedforward control model establishment and verification, bumpless switching, and digital communication signal chain, further implements a three-dimensional feedforward control model with gas temperature correction, online model correction based on cumulative monitoring of correction values, and adaptive safety margin cross-limiting processing. The burner control system in this embodiment, based on the hardware configuration of Embodiment 1, adds a resistance temperature sensor (RTS) installed near the upstream pressure measuring point of the gas regulating valve. This sensor supports digital communication protocols and directly transmits the gas temperature measurement value to the set decision layer via a fieldbus.

[0046] The input dimension of the feedforward control model also includes gas temperature: During the data acquisition phase, the current gas temperature is recorded synchronously after each operating condition stabilizes, so that each data sample point includes three input dimensions: actuator control quantity, pipeline pressure, and gas temperature; a three-dimensional piecewise mapping feedforward control model is constructed with actuator control quantity, pipeline pressure, and gas temperature as three-dimensional inputs and measured flow rate as output; the lookup table uses actuator control quantity as the main index, and pipeline pressure correction factor and gas temperature correction factor are independently superimposed on the baseline flow rate value; when the decision layer calls the feedforward control model, it queries the baseline flow rate value based on the current actuator control quantity, and then sequentially superimposes the pressure correction quantity corresponding to the current pipeline pressure and the temperature correction quantity corresponding to the current gas temperature to obtain a predicted flow rate value that simultaneously eliminates the influence of pipeline pressure deviation and gas temperature deviation, and uses this predicted flow rate value as the basis for the flow rate setpoint.

[0047] Specifically, the data acquisition phase is expanded from the first embodiment to include: the actuator changes progressively in a step-by-step manner throughout its entire stroke. After each step of the operating condition stabilizes, data from four dimensions—the actuator control quantity, the measured flow rate, the pipeline pressure, and the gas temperature—are recorded simultaneously, so that each sample point contains three input variables and one output variable.

[0048] The model storage structure, based on a combination of segmented lookup tables and pressure correction factors, adds a gas temperature correction factor. The lookup table uses actuator control quantities as the primary index, storing the baseline flow rates corresponding to each control quantity under standard reference pressure and standard reference temperature conditions. The standard reference temperature is selected from the most frequently occurring gas temperature value during the data acquisition phase, typically the gas temperature corresponding to the annual average air temperature of the burner installation site.

[0049] The gas temperature correction factor takes the difference between the current gas temperature and the standard reference temperature as the input variable and outputs the correction amount for the reference flow rate. The parameter determination method of the temperature correction factor is as follows: using sample data obtained under different temperature conditions during the data acquisition phase, the relationship between temperature deviation and flow rate correction is linearly fitted, and the obtained linear coefficient is used as the proportional parameter of the temperature correction factor and stored in the control system parameter file; if the actual temperature variation range is large, causing the linear fitting residual to exceed the accuracy threshold, then the linear factor is divided into multiple segments according to the temperature range and stored in the form of a segmented lookup table, with the interpolation method consistent with the main lookup table.

[0050] The decision-making layer is configured to call the three-dimensional feedforward control model in the following order during each control cycle: Using the current actuator control quantity as an index, the main lookup table is queried to obtain the baseline flow rate value; the pressure correction is calculated using the difference between the current pipeline pressure and the standard reference pressure as input; the temperature correction is calculated using the difference between the current gas temperature and the standard reference temperature as input; the baseline flow rate value, pressure correction, and temperature correction are added together to obtain the predicted flow rate value under the current operating condition, which serves as the basis for calculating the gas flow rate setpoint. The air-side feedforward control model adopts the same extended structure, using air temperature instead of gas temperature as the temperature correction input.

[0051] The decision-making layer is configured to cumulatively monitor the correction amount of flue gas oxygen content to trigger online correction of the feedforward control model: the arithmetic mean of the correction amount is calculated within a preset time window; when the absolute value of the arithmetic mean exceeds a preset drift threshold, it is determined that there is a systematic deviation in the feedforward control model, triggering the online model correction program; the online model correction program uses the additive relationship between the output value of the feedforward control model and the reference flow value to convert the current correction amount mean into the correction amount of the reference flow value of the corresponding operating point lookup table, and updates the reference flow value of the corresponding operating point in the lookup table in the same direction; after the correction is completed, the integral of the correction amount is cleared to zero and the cumulative counting starts again; the preset drift threshold is determined based on the statistical characteristics of the measurement noise of flue gas oxygen content during normal system operation, with a preset multiple of the standard deviation of the measurement noise as the criterion value; the length of the preset time window is not less than five times the integral time constant of flue gas oxygen content; the online model correction program is only executed when the system is in a stable operating condition.

[0052] Specifically, before the system is officially put into operation, the following parameters need to be initialized and calibrated.

[0053] The length of the preset time window is determined as follows: it is more than five times the integral time constant of flue gas oxygen content to ensure that the average correction within the window can effectively filter out short-term fluctuations caused by the dynamic load process and retain only the low-frequency components caused by systematic drift; the upper limit of the window length is determined according to the requirements of the operation engineer for the model correction response speed, and is usually set to several hours in actual engineering.

[0054] The preset drift threshold is determined as follows: During the initial stable operation phase after the system commissioning is completed, when the combustion system is running under different fixed operating conditions, the arithmetic mean of the flue gas oxygen content correction amount under each operating condition is recorded within a preset time window. The standard deviation of this mean is calculated within multiple time windows. Two to three times this standard deviation is used as the preset drift threshold. The specific multiple is determined by the commissioning engineer based on the actual noise level of the system, so that the average correction amount caused by normal measurement noise does not exceed the threshold, while the average value caused by systematic drift can effectively trigger the correction procedure.

[0055] During system operation, the decision-making layer is set to update the arithmetic mean of the correction amount in a sliding window manner in each control cycle: the algebraic sum of all correction amount sampled values ​​within the time window is divided by the number of sample points to obtain the arithmetic mean of the correction amount within the current time window, and this mean is compared with the preset drift threshold.

[0056] When the absolute value of the mean exceeds the preset drift threshold, after confirming that the system is currently in a stable operating condition (the criterion is that the load change rate is lower than the preset rate threshold and the correction amount is within the normal limit range), the online model correction program is started. The online model correction program updates the lookup table according to the following logic: For the currently operating actuator control point, the current predicted flow rate is equal to the sum of the baseline flow rate, the pressure correction, the temperature correction, and the average correction amount; the average correction amount represents the systematic flow prediction deviation of the feedforward model under the current state of this operating point; this deviation is added to the baseline flow rate corresponding to this operating point in the lookup table (if the average correction amount is positive, the baseline flow rate is increased; if it is negative, it is decreased), so that the predicted flow rate of this operating point under the same input conditions in the next cycle increases (or decreases) by the corresponding order of magnitude, thereby causing the subsequent average correction amount to return to near zero. After updating all operating points with valid sampled data within the current time window, the integrator of the flue gas oxygen content correction amount is cleared, the historical data within the time window is cleared, and the cumulative counting restarts.

[0057] The safety coordination layer dynamically adjusts the positive safety margin in the cross-limiting process based on the rate of change of the load command: the load change rate is obtained by calculating the ratio of the change in the load setpoint to the control cycle duration within adjacent control cycles; when the load change rate exceeds the preset rate threshold, the positive safety margin increases proportionally to the basic margin based on the ratio of the load change rate to the preset rate threshold; when the load change rate does not exceed the preset rate threshold, the positive safety margin remains unchanged at the basic margin; the preset rate threshold is determined based on the maximum tracking rate of the airflow closed-loop control loop within the rated load range, so that the airflow safety setpoint after adding the dynamic safety margin at any load change rate does not exceed the tracking capability of the airflow closed-loop control loop; the basic margin is determined based on the steady-state tracking error of the airflow closed-loop control loop under the condition of uniform load change at the rate corresponding to the preset rate threshold.

[0058] Specifically, the safety coordination layer calculates the current load change rate in each control cycle by dividing the absolute value of the difference between the load setpoints of adjacent control cycles by the duration of the control cycle.

[0059] The calibration method for the preset rate threshold is as follows: During the commissioning phase, a ramp tracking test is performed on the air flow closed-loop control loop. The air flow setpoint is driven to change with a gradually increasing ramp rate. The tracking deviation of the measured air flow value is recorded as a function of the ramp rate. The ramp rate at which the tracking deviation begins to exceed the process tolerance range is identified. 60% to 80% of this rate is used as the preset rate threshold, leaving a response margin. The specific ratio is confirmed in the commissioning document.

[0060] The calibration method for the basic margin is as follows: drive the load setpoint to change at a ramp rate equal to the preset rate threshold, measure the maximum deviation between the measured air flow and the ideal tracking curve during the stable tracking phase, and use 1.2 times the maximum deviation as the basic margin.

[0061] The dynamic calculation process of positive safety margin: When the load change rate exceeds the preset rate threshold, the basic margin is multiplied by the ratio of the load change rate to the preset rate threshold to obtain the dynamic safety margin for this control cycle; when the load change rate does not exceed the preset rate threshold, the dynamic safety margin is equal to the basic margin.

[0062] During load increase, the safety coordination layer sets the airflow safety setpoint to the sum of the target airflow value and the current dynamic safety margin. Once the measured airflow value enters the confirmation bandwidth centered on the target airflow value and twice the measurement accuracy of the flow sensor, the gas flow safety setpoint is increased to the target gas flow value. The cross-limiting sequence during load decrease is the same as in Example 1. The dynamic safety margin is only valid in the load increase direction, and the safety margin parameters in the load decrease direction remain consistent with those in Example 1.

[0063] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for precise control of burner air-fuel ratio based on hierarchical optimization, characterized in that, include: S1. When the combustion system is in a closed-loop flow operation state, collect load commands, actuator control quantities, measured flow values ​​and pipeline pressures under different load conditions. Based on the collected multi-dimensional data, construct a feedforward control model that describes the mapping relationship between actuator control quantities and flow. S2. Switch the control system to an operating mode based on a feedforward control model, and implement hierarchical optimization control of the air-fuel ratio through a four-layer hierarchical structure: The monitoring and optimization layer transmits the load setpoints to the setting and decision-making layer; The decision-making layer calculates the setpoints for gas flow and air flow based on the feedforward control model, and slowly fine-tunes the two setpoints based on the flue gas oxygen content to compensate for slow disturbances in fuel characteristics. The safety coordination layer implements cross-limiting processing for the gas flow setpoint and the air flow setpoint to ensure that the adjustment of the air flow setpoint precedes the gas flow setpoint during load changes, and outputs a safe setpoint. The closed-loop execution layer receives safety setpoints and forms independent closed-loop control for gas flow and air flow respectively.

2. The method for precise control of burner air-fuel ratio based on hierarchical optimization according to claim 1, characterized in that: In step S1, the data acquisition process in the feedforward control model establishment stage is as follows: the data changes progressively in a stepwise manner throughout the entire stroke range. After the system reaches a stable state under each step condition, the current actuator control quantity, measured flow rate, and pipeline pressure are recorded synchronously to form a multi-dimensional dataset covering the entire operating range. Using the actuator control quantity and pipeline pressure as input dimensions and the measured flow rate as output dimension, a segmented mapping feedforward control model is constructed based on the multi-dimensional dataset.

3. The method for precise control of burner air-fuel ratio based on hierarchical optimization according to claim 2, characterized in that: The step between steps S1 and S2 includes a model verification step: Under the closed-loop operation state of the flow, the predicted flow value calculated by the feedforward control model based on the current actuator control quantity and pipeline pressure is compared with the measured flow value, and the deviation between the predicted flow value and the measured flow value is calculated within the entire operating range; when the deviation within the entire operating range is lower than the preset accuracy threshold, the feedforward control model is determined to have passed the verification, and the process proceeds to the execution stage of step S2; if there are operating points where the deviation exceeds the accuracy threshold, the process returns to step S1 to collect data for the corresponding operating conditions and revise the feedforward control model.

4. The method for precise control of burner air-fuel ratio based on hierarchical optimization according to claim 3, characterized in that: The transition from step S1 to step S2 is performed using a seamless transition method: Before the transition, the output value of the feedforward control model continuously tracks the setpoint of the current flow closed-loop control, ensuring that the output value of the feedforward control model is consistent with the flow closed-loop setpoint; during the transition, the control authority is transferred from the flow closed-loop mode to the feedforward model mode, and the gas flow setpoint and air flow setpoint remain continuous before and after the transition without any abrupt changes; after the transition is completed, the temporary flow detection signal exits the control loop and enters an operating mode that is primarily based on the feedforward control model and supplemented by fine-tuning based on flue gas oxygen content feedback.

5. The method for precise control of burner air-fuel ratio based on hierarchical optimization according to claim 1, characterized in that: The cross-limiting process of the safety coordination layer in step S2 is as follows: the safety coordination layer continuously judges the direction of change of the load command; when the load command is on the rise, the safety coordination layer first raises the air flow setpoint to the target value and adds a positive safety margin. After the air flow closed-loop response confirms that the target range has been reached, the gas flow setpoint is then raised to the target value. When the load command shows a downward trend, the safety coordination layer first reduces the gas flow setpoint to the target value. After the gas flow closed-loop response confirms that the target range has been reached, the air flow setpoint is then reduced to the target value. The output result of the cross-limiting processing is passed down to the closed-loop execution layer as the safety setpoint.

6. The method for precise control of burner air-fuel ratio based on hierarchical optimization according to claim 1, characterized in that: In step S2, the process of the decision-making layer fine-tuning the output value of the feedforward control model using flue gas oxygen content is as follows: the deviation between the measured value of flue gas oxygen content and the target oxygen content is used as the input signal, the deviation is integrated over time, and the integration result is superimposed on the air-fuel ratio reference value output by the feedforward control model as a correction amount; the time constant of the integration operation is much larger than the time scale of the load dynamic response, so that the correction amount only responds to the long-term drift of fuel characteristics without interfering with the rapid dynamic adjustment of load changes; when the correction amount exceeds the preset limit range, the correction amount limit protection is triggered to prevent the air-fuel ratio reference value from being erroneously adjusted due to abnormal flue gas oxygen content detection.

7. The method for precise control of burner air-fuel ratio based on hierarchical optimization according to claim 1, characterized in that: The feedforward control model is stored in the form of a combination of a segmented lookup table and a pipeline pressure correction factor: the lookup table is indexed by the actuator control quantity and records the reference flow value corresponding to each control quantity under standard pressure conditions. The pipeline pressure correction factor takes the difference between the current pipeline pressure and the standard reference pressure as input and outputs the correction amount for the reference flow rate value. When the decision-making layer calls the feedforward control model, it queries the baseline flow rate value based on the actuator control quantity, and then superimposes the correction amount output by the pipeline pressure correction factor to obtain the predicted flow rate value under the current pipeline pressure conditions. The predicted flow rate value is used as the basis for the flow rate setpoint value transmitted to the safety coordination layer.

8. A precise air-fuel ratio control system for burners based on hierarchical optimization, characterized in that, include: The feedforward model building module is used to collect load commands, actuator control quantities, measured flow values ​​and pipeline pressure under different load conditions when the combustion system is in a flow closed-loop operation state. Based on the collected multi-dimensional data, a feedforward control model describing the mapping relationship between actuator control quantities and flow is constructed. The hierarchical optimization operation module is used to switch the control system to an operation mode based on the feedforward control model, and implements hierarchical optimization control of air-fuel ratio through a four-layer hierarchical structure: The monitoring and optimization layer transmits the load setpoints to the setting and decision-making layer; The decision-making layer calculates the setpoints for gas flow and air flow based on the feedforward control model, and slowly fine-tunes the two setpoints based on the flue gas oxygen content to compensate for slow disturbances in fuel characteristics. The safety coordination layer implements cross-limiting processing for the gas flow setpoint and the air flow setpoint to ensure that the adjustment of the air flow setpoint precedes the gas flow setpoint during load changes, and outputs a safe setpoint. The closed-loop execution layer receives safety setpoints and forms independent closed-loop control for gas flow and air flow respectively.

Images