Combustion optimization control system based on multi-target optimization

Abstract

The invention relates to the technical field of industrial automation and intelligent manufacturing, and discloses a combustion optimization control system based on multi-target optimization, which comprises a data collection unit, a sensor fault diagnosis unit, a data preprocessing unit, a multi-target optimization unit, a multi-target online regulation and control unit, a control instruction generation unit and an online learning and self-adaption unit, the data collection unit is responsible for collecting combustion system and extreme working condition parameters, the sensor fault diagnosis unit is responsible for diagnosing sensor null drift and failure faults, the data preprocessing unit outputs high-quality multi-source data, and the multi-target optimization unit calculates a sensor i compensation value, boiler efficiency, a NOx emission value and a combustion stability index based on the high-quality multi-source data. The multi-target online regulation and control unit outputs a Pareto optimal solution set based on a calculation result, the control instruction generation unit generates a specific control instruction which can be directly issued to the DCS, and the online learning and self-adaption unit updates model parameters and decision rules online.

Description

Technical Field

[0001] This invention relates to the field of industrial automation and intelligent manufacturing technology, specifically to a combustion optimization control system based on multi-objective optimization. Background Technology

[0002] In the field of industrial boiler combustion process control, achieving efficient, clean, and stable operation has always been a core objective. Traditional combustion optimization control systems mostly employ single-loop regulation strategies based on classic PID controllers or rely on empirical models under static operating conditions, which have the following significant limitations: First, the control system is often designed around a single objective (such as economy or environmental protection), making it difficult to balance economy (boiler efficiency) and environmental protection (NOx) simultaneously. x The system faces several challenges. First, it needs to achieve a dynamic balance between multiple conflicting objectives, such as pollutant emissions and safety (combustion stability). Second, the system is highly dependent on the measurement accuracy and reliability of field sensors, lacking effective fault diagnosis and data compensation mechanisms. If a sensor experiences zero drift or failure, it will directly cause the optimization results to deviate from reality, and may even lead to control instability. Third, existing optimization models are mostly static models, lacking online self-learning capabilities and unable to adapt to dynamic disturbances such as fuel composition fluctuations, load changes, and equipment aging, resulting in long-term performance degradation. Therefore, developing an intelligent combustion optimization system that can balance multiple objectives, possess high reliability, and continuously adapt has become an urgent need in the fields of industrial automation and intelligent manufacturing technology. Summary of the Invention

[0003] (a) Technical problems to be solved

[0004] To address the shortcomings of existing technologies, this invention provides a combustion optimization control system based on multi-objective optimization, which has the advantages of multi-objective collaborative balancing, data-driven fault tolerance, and online adaptive optimization, and solves the problems of traditional systems in terms of multiple objective trade-offs, data reliability assurance, and dynamic operating condition adaptability.

[0005] (II) Technical Solution

[0006] To achieve the above objectives, the present invention provides the following technical solution: a combustion optimization control system based on multi-objective optimization, comprising a data collection unit, a sensor fault diagnosis unit, a data preprocessing unit, a multi-objective optimization unit, a multi-objective online control unit, a control command generation unit, and an online learning and adaptation unit;

[0007] The data collection unit is responsible for collecting parameters of the combustion system, including fuel, air, status, emissions, equipment, and extreme operating conditions.

[0008] The sensor fault diagnosis unit is responsible for diagnosing sensor zero drift and failure faults, and compensating for the missing abnormal data area through the redundant sensor weighted compensation calculation formula.

[0009] The data preprocessing unit performs data alignment and noise reduction, feature engineering extraction and working condition pattern recognition based on the compensated abnormal data, and outputs high-quality multi-source data.

[0010] The multi-objective optimization unit calculates the sensor i compensation value based on high-quality multi-source data. Boiler efficiency NO x Emissions and combustion stability index ;

[0011] Based on the calculation results, the multi-objective online control unit uses a third-generation non-dominated sorting genetic algorithm to search the five-dimensional decision variables of primary air volume, secondary air volume, burnout air volume, coal feed rate and burner swing angle in real time, dynamically adjusts the air-coal ratio and flame position, achieves a balance of three objectives of high efficiency, low emissions and high stability, and outputs a Pareto optimal solution set.

[0012] The control command generation unit combines the Pareto optimal solution set output by the multi-objective online control unit with the current DCS range and the actuator limit, and converts it into a 4-20mA standard signal and valve opening percentage to generate specific control commands that can be directly sent to the DCS.

[0013] The online learning and adaptive unit monitors the optimization effect online and updates model parameters and decision rules online.

[0014] Preferably, the data collection unit includes a fuel parameter acquisition terminal, an air system acquisition terminal, a combustion status acquisition terminal, an emission parameter acquisition terminal, an equipment status acquisition terminal, an extreme operating condition acquisition terminal, and a load demand acquisition terminal.

[0015] Preferably, the fuel parameter acquisition terminal is responsible for acquiring fuel type, fuel flow rate, fuel calorific value and fuel moisture content; the air system acquisition terminal is responsible for acquiring primary air volume / pressure, secondary air volume / pressure and excess air coefficient.

[0016] Preferably, the combustion status acquisition terminal is responsible for acquiring multi-point furnace temperature, furnace negative pressure, flame intensity / frequency, and flue gas temperature; the emission parameter acquisition terminal is responsible for acquiring flue gas NO. x Concentration, SO2 concentration, and smoke and dust concentration.

[0017] Preferably, the equipment status acquisition terminal is responsible for acquiring the induced draft fan / forced draft fan speed, fuel valve opening degree, and sensor operating voltage; the extreme condition acquisition terminal is responsible for acquiring the furnace pressure sudden change value, flame signal interruption duration, and fuel supply interruption early warning signal; and the load demand acquisition terminal is responsible for acquiring the real-time grid load and boiler steam pressure.

[0018] Preferably, when the sensor fault diagnosis unit diagnoses a fault in sensor i, it takes data from redundant sensors j and k of the same type, and then calculates the compensation value for sensor i using the weighted compensation formula for redundant sensors. The calculation formula is as follows: In the formula, This represents the compensation value for faulty sensor i. This represents the weighting coefficient of redundant sensor j. This represents the weighting coefficient of the redundant sensor k. This represents the real-time data acquired by redundant sensor j, and the real-time output of sensor j of the same type in the data collection unit. This represents the real-time data collected by the redundant sensor k;

[0019] By calculating the compensation value of sensor i As real-time alternative data to fault sensor i, it can be incorporated into the standardized process of the data preprocessing unit to enable continuous data chain connection.

[0020] Preferably, the multi-objective optimization unit calculates boiler efficiency based on high-quality multi-source data. The calculation formula is as follows: In the formula, , These represent the specific enthalpy of steam at the outlet and the inlet, respectively. Indicates the main steam flow rate. Indicates the instantaneous flow rate of fuel. This indicates the received lower heating value of the fuel.

[0021] Preferably, the multi-objective optimization unit calculates NO based on high-quality multi-source data. x Emissions The calculation formula is as follows: In the formula, Indicates measured dry basis NO x Volume concentration, This indicates the measured dry basis oxygen content. Standard reference oxygen content, Smoke density under standard conditions.

[0022] Preferably, the multi-objective optimization unit calculates the combustion stability index based on high-quality multi-source data. The calculation formula is as follows: In the formula, , These represent the mean and standard deviation of the temperature at multiple points in the furnace, respectively. , This represents the mean and standard deviation of the negative pressure in the furnace. Indicates the cumulative duration of the flame signal interruption. Indicates the statistical period. , , These represent the normalized weighting coefficients for temperature fluctuation, pressure fluctuation, and flame interruption, respectively, satisfying... .

[0023] Preferably, the multi-objective online control unit is based on boiler efficiency. NO x Emissions and combustion stability index The specific operation to output the Pareto optimal solution set is as follows:

[0024] S1. Decision space initialization and population generation: Based on the current boiler load and safe operation boundary, determine the feasible range of five decision variables, and randomly generate an initial population containing multiple individuals within this range.

[0025] S2. Multi-objective performance evaluation and ranking: Substitute the combination of decision variables for each individual in the population into the boiler efficiency. NO x Emissions and combustion stability index In the prediction model, the values ​​of its three objective functions are calculated; then, based on the core mechanism of NSGA-III, the individuals in the population are non-dominated and sorted to distinguish their superior and inferior levels, and the reference point mechanism is used to perform adaptive normalization and association operations on individuals in the same non-dominated level.

[0026] S3. Elite Preservation and Iterative Evolution: Through tournament selection, simulated crossover and mutation genetic operations, a new offspring population is generated. Combined with the parent population, an elite preservation strategy is implemented to select the next generation population with better performance. This process is iterated until the convergence condition is met or the maximum number of iterations is reached. Finally, a uniformly distributed Pareto optimal solution set is output for the decision-making subunits within the multi-objective online control unit to select.

[0027] Compared with the prior art, the present invention provides a combustion optimization control system based on multi-objective optimization, which has the following beneficial effects:

[0028] 1. This invention calculates the compensation value of sensor i. As real-time substitute data for faulty sensor i, it is incorporated into the standardized process of the data preprocessing unit, enabling continuous data chain connection and avoiding calculation interruptions or deviations caused by sensor failure; simultaneously, by real-time monitoring of the compensation value of sensor i... The numerical range can also determine the severity of sensor malfunctions, distinguish between temporary fluctuations and permanent failures, and replace failed sensors in a timely manner, reducing manual inspection costs. Ultimately, this achieves the beneficial effect of ensuring the continuity and reliability of input data, providing high-quality data support for subsequent optimization decisions.

[0029] 2. This invention calculates boiler efficiency. As a core evaluation indicator of the economic efficiency of a combustion system, substituting it into the three-objective equilibrium model of a multi-objective optimization unit enables targeted optimization of parameters such as the air-coal ratio and quantitative evaluation of the energy conversion efficiency of the combustion process; simultaneously, by monitoring boiler efficiency... The numerical range can also determine the current level of combustion economy, ultimately achieving the beneficial effect of continuously improving boiler thermal efficiency and reducing unit steam production energy consumption.

[0030] 3. This invention calculates NO x Emissions As a key indicator for evaluating the environmental performance of combustion systems, incorporating it into the Pareto optimization model of a multi-objective online control unit enables precise control of parameters such as the excess air coefficient and quantitative constraints on pollutant emission levels. Simultaneously, by monitoring NO... x Emissions The numerical range can also be used to determine the risk of failing to meet environmental standards, ultimately achieving the beneficial effect of balancing environmental requirements with combustion efficiency and avoiding fines for exceeding standards.

[0031] 4. This invention calculates the combustion stability index. As a comprehensive evaluation index for the safety of combustion systems, it can be incorporated into the three-objective constraint model of a multi-objective optimization unit to enable coordinated control of parameters such as furnace negative pressure and flame intensity, and to quantitatively characterize the stability of the combustion process. Simultaneously, by monitoring the combustion stability index... The numerical range can also determine the level of operational risk, provide risk warnings and implement corrective measures to prevent flameout and deflagration accidents, and ultimately achieve the beneficial effects of ensuring the safe operation of the boiler and extending the service life of the equipment. Attached Figure Description

[0032] Figure 1 This is a flowchart illustrating the system operation of the present invention. Detailed Implementation

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] Please see Figure 1 The combustion optimization control system based on multi-objective optimization includes a data collection unit, a sensor fault diagnosis unit, a data preprocessing unit, a multi-objective optimization unit, a multi-objective online control unit, a control command generation unit, and an online learning and adaptation unit.

[0035] The data collection unit is responsible for collecting parameters of the combustion system, including fuel, air, status, emissions, equipment, and extreme operating conditions, in order to address the problem of scarce extreme samples.

[0036] The end is responsible for collecting sensor fault diagnosis unit, which is responsible for diagnosing sensor zero drift and failure faults. It compensates for the missing abnormal data area through redundant sensor weighted compensation calculation formula to solve the problem of low input data reliability.

[0037] The terminal is responsible for collecting data. The preprocessing unit performs data alignment and noise reduction, feature engineering extraction and working condition pattern recognition based on the compensated abnormal data, and outputs high-quality multi-source data to provide a foundation for subsequent analysis and calculation.

[0038] The end is responsible for collecting multi-objective optimization unit data and calculating sensor i compensation value based on high-quality multi-source data. Boiler efficiency NO x Emissions and combustion stability index ;

[0039] The terminal is responsible for collecting data from the multi-objective online control unit. Based on the calculation results, it uses the third-generation non-dominated sorting genetic algorithm (NSGA-Ⅲ) to search for five-dimensional decision variables in real time, including primary air volume, secondary air volume, burnout air volume, coal feed rate and burner tilt angle. It dynamically adjusts the air-coal ratio and flame position to achieve a balance of three objectives: high efficiency, low emissions and high stability, and outputs a Pareto optimal solution set.

[0040] The terminal is responsible for collecting control command generation units. It combines the Pareto optimal solution set output by the multi-objective online control unit with the current DCS range and actuator limit, and converts it into a 4-20mA standard signal and valve opening percentage to generate specific control commands that can be directly sent to the DCS.

[0041] The terminal is responsible for collecting the online learning and adaptive unit online monitoring and optimization effects, and updating model parameters and decision rules online, so that the system can adapt to long-term drift due to equipment aging and fuel changes, and solve the problem of insufficient dynamic adaptability.

[0042] The data collection unit includes fuel parameter acquisition terminals, air system acquisition terminals, combustion status acquisition terminals, emission parameter acquisition terminals, equipment status acquisition terminals, extreme operating condition acquisition terminals, and load demand acquisition terminals.

[0043] The fuel parameter acquisition terminal is responsible for collecting fuel type, fuel flow rate, fuel calorific value, and fuel moisture content; the air system acquisition terminal is responsible for collecting primary air volume / pressure, secondary air volume / pressure, and excess air coefficient.

[0044] The combustion status acquisition terminal is responsible for collecting multi-point furnace temperature, furnace negative pressure, flame intensity / frequency, and flue gas temperature; the emission parameter acquisition terminal is responsible for collecting flue gas NO. x Concentration, SO2 concentration, and smoke and dust concentration.

[0045] The equipment status acquisition terminal is responsible for collecting the speed of the induced draft fan / forced draft fan, the opening degree of the fuel valve, and the working voltage of the sensor; the extreme operating condition acquisition terminal is responsible for collecting the sudden change value of the furnace pressure, the duration of the flame signal interruption, and the early warning signal of the fuel supply interruption; the load demand acquisition terminal is responsible for collecting the real-time load of the power grid and the boiler steam pressure.

[0046] The advantages are: by establishing collection terminals in categories through the above-mentioned data collection units, multi-dimensional full-condition data coverage (covering 7 major categories of core parameters such as fuel characteristics, combustion state, and emission indicators), special collection of extreme condition samples (targeted capture of scarce samples such as sudden changes in furnace pressure), and spatiotemporal synchronized data association and integration (ensuring that the timestamps of data from different collection terminals are aligned), the beneficial effect of providing a complete data foundation for subsequent fault diagnosis and optimization calculation is achieved, solving the problem of insufficient model generalization ability caused by incomplete data dimensions and missing extreme samples in traditional systems from the source.

[0047] When the sensor fault diagnosis unit diagnoses a fault in sensor i, it takes data from redundant sensors j and k of the same type, and then calculates the compensation value for sensor i using the weighted compensation formula for redundant sensors. The calculation formula is as follows: In the formula, This represents the compensation value for faulty sensor i (consistent with the measurement dimension of sensor i, such as temperature in °C and concentration in mg / m³, and the final output faulty sensor replacement data is used in the subsequent data preprocessing unit). The weighting coefficient of redundant sensor j is represented (unitless, only reflecting the weight ratio, calculated from the historical operating data of sensor j; the higher the accuracy, the larger the weighting coefficient). This represents the weighting coefficient of redundant sensor k (unitless, reflecting only the weighting percentage, similar to...). (Calculated based on historical data of sensor k, ensuring that the high-precision sensor contributes more to the compensation value). This represents the real-time data collected by redundant sensor j, and the real-time output of sensor j of the same type in the data collection unit (which has undergone preliminary signal filtering but has not been compensated for faults). This represents the real-time data acquired by redundant sensor k (with the same measurement dimension as sensor i, and so on). (The real-time signal comes from sensor k of the same type in the data collection unit).

[0048] The advantage is that by calculating the compensation value of sensor i... As real-time substitute data for faulty sensor i, it is incorporated into the standardized process of the data preprocessing unit, enabling continuous data chain connection and avoiding calculation interruptions or deviations caused by sensor failure; simultaneously, by real-time monitoring of the compensation value of sensor i... The numerical range (such as deviation from normal data of the same period in history exceeding ±5%) can also determine the severity of sensor i failure, distinguish between temporary fluctuations and permanent failures, replace failed sensors in a timely manner, reduce manual inspection costs, and ultimately achieve the beneficial effect of ensuring the continuity and reliability of input data, providing high-quality data support for subsequent optimization decisions.

[0049] The multi-objective optimization unit calculates boiler efficiency based on high-quality multi-source data. The calculation formula is as follows: In the formula, , These represent the specific enthalpy of the steam at the outlet and the inlet (kJ·kg⁻¹). -1 ), Indicates the main steam flow rate (kg·s) -1 ), Indicates the instantaneous flow rate of fuel (kg·s) -1 ), This indicates the lower heating value of the fuel on a received basis (kJ·kg). -1 ).

[0050] The advantage is that it allows for the calculation of boiler efficiency. As a core evaluation indicator of the economic efficiency of a combustion system, substituting it into the three-objective equilibrium model of a multi-objective optimization unit enables targeted optimization of parameters such as the air-coal ratio and quantitative evaluation of the energy conversion efficiency of the combustion process; simultaneously, by monitoring boiler efficiency... The numerical range can also be used to determine the current combustion economy level: when the boiler efficiency... When the efficiency is greater than 90% (design optimal value), maintain the current control parameters; when the boiler efficiency... When the boiler efficiency is in the 85%–90% range, fine-tune the ratio of secondary air volume to coal feed rate; when the boiler efficiency is... When the efficiency is less than 85%, deep optimization (expanding the optimization range) is triggered to avoid energy waste and ultimately achieve the beneficial effect of continuously improving boiler thermal efficiency and reducing unit steam production energy consumption.

[0051] The multi-objective optimization unit calculates NO based on high-quality multi-source data. x Emissions The calculation formula is as follows: In the formula, Indicates measured dry basis NO x Volume concentration (ppm) This indicates the measured dry basis oxygen content (%). Standard reference oxygen content (6% for coal-fired boilers). Flue gas density under standard conditions (kg·m³) -3 According to 1.34 kg·m -3 calculate).

[0052] The advantage is: by calculating NO x Emissions As a key indicator for evaluating the environmental performance of combustion systems, incorporating it into the Pareto optimization model of a multi-objective online control unit enables precise control of parameters such as the excess air coefficient and quantitative constraints on pollutant emission levels. Simultaneously, by monitoring NO... x Emissions The numerical range can also be used to determine the risk of failing to meet environmental standards: when NO x Emissions When NO is less than 50 mg / m³ (ultra-low emission standard), maintain the current air-fuel ratio; when NO x Emissions When NO is in the range of 50–100 mg / m³, increase the secondary air volume to reduce local high-temperature areas; when NO x Emissions When the concentration exceeds 100 mg / m³, the coal feed rate is temporarily reduced and an early warning is triggered to ensure that emissions meet standards. Ultimately, this achieves the beneficial effect of balancing environmental protection requirements with combustion efficiency and avoiding fines for exceeding standards.

[0053] The multi-objective optimization unit calculates the combustion stability index based on high-quality multi-source data. The calculation formula is as follows: In the formula, , These represent the mean and standard deviation (°C) of the temperature at multiple points in the furnace. , This represents the mean and standard deviation of the negative pressure in the furnace (Pa). Indicates the cumulative duration (s) of the flame signal interruption. Indicates the statistical period (taken as 300-600s). , , These represent the normalized weighting coefficients for temperature fluctuation, pressure fluctuation, and flame interruption, respectively, satisfying... .

[0054] The advantage is that it allows for the calculation of the combustion stability index. As a comprehensive evaluation index for the safety of combustion systems, it can be incorporated into the three-objective constraint model of a multi-objective optimization unit to enable coordinated control of parameters such as furnace negative pressure and flame intensity, and to quantitatively characterize the stability of the combustion process. Simultaneously, by monitoring the combustion stability index... The numerical range of the combustion stability index can also determine the operational risk level: when the combustion stability index... When the value is greater than 0.9 (high stability state), maintain the current burner tilt angle and fan parameters; when When the pressure is in the range of 0.7-0.9, adjust the induced draft fan speed to stabilize the negative pressure in the furnace; when When the value is less than 0.7, emergency combustion stabilization measures (such as increasing the primary air volume) are triggered and an alarm is sounded to prevent accidents such as flameout and deflagration, ultimately achieving the beneficial effect of ensuring the safe operation of the boiler and extending the service life of the equipment.

[0055] Multi-objective online control unit based on boiler efficiency NO x Emissions and combustion stability index Output the Pareto optimal solution set, the specific operation of which is as follows:

[0056] S1. Decision space initialization and population generation: Based on the current boiler load and safe operation boundary, determine the feasible range of five decision variables (primary air volume, secondary air volume, burnout air volume, coal feed rate, and burner sway angle), and randomly generate an initial population containing multiple individuals (i.e., different combinations of control parameters) within this range.

[0057] S2. Multi-objective performance evaluation and ranking: Substitute the combination of decision variables for each individual in the population into the boiler efficiency. NO x Emissions and combustion stability index In the prediction model, the values ​​of its three objective functions are calculated; then, based on the core mechanism of NSGA-III, the individuals in the population are non-dominated and sorted to distinguish their superior and inferior levels, and the reference point mechanism is used to perform adaptive normalization and association operations on individuals in the same non-dominated level in order to maintain the diversity of the solution set on the Pareto front.

[0058] S3, Elite Preservation and Iterative Evolution: Through tournament selection, simulated crossover and mutation genetic operations, a new offspring population is generated, and combined with the parent population, an elite preservation strategy is implemented to select the next generation population with better performance; this process is iterated until the convergence condition is met or the maximum number of iterations is reached, and finally a uniformly distributed Pareto optimal solution set is output for the decision subunits within the multi-objective online control unit to select.

[0059] The advantages are: by systematic optimization, balanced decision-making and linear adaptation, the limitations of traditional single-objective or fixed-weight optimization are overcome, enabling the system to provide a series of operating schemes that achieve the best balance between boiler thermal efficiency, environmental emissions and operational stability under complex and variable operating conditions, thereby significantly improving the overall performance of the combustion process.

[0060] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A combustion optimization control system based on multi-objective optimization, characterized in that, It includes a data collection unit, a sensor fault diagnosis unit, a data preprocessing unit, a multi-objective optimization unit, a multi-objective online control unit, a control command generation unit, and an online learning and adaptation unit; The data collection unit is responsible for collecting parameters of the combustion system, including fuel, air, status, emissions, equipment, and extreme operating conditions. The sensor fault diagnosis unit is responsible for diagnosing sensor zero drift and failure faults, and compensating for the missing abnormal data area through the redundant sensor weighted compensation calculation formula. The data preprocessing unit performs data alignment and noise reduction, feature engineering extraction and working condition pattern recognition based on the compensated abnormal data, and outputs high-quality multi-source data. The multi-objective optimization unit calculates the sensor i compensation value based on high-quality multi-source data. Boiler efficiency NO x Emissions and combustion stability index ; Based on the calculation results, the multi-objective online control unit uses a third-generation non-dominated sorting genetic algorithm to search the five-dimensional decision variables of primary air volume, secondary air volume, burnout air volume, coal feed rate and burner swing angle in real time, dynamically adjusts the air-coal ratio and flame position, achieves a balance of three objectives of high efficiency, low emissions and high stability, and outputs a Pareto optimal solution set. The control command generation unit combines the Pareto optimal solution set output by the multi-objective online control unit with the current DCS range and the actuator limit, and converts it into a 4-20mA standard signal and valve opening percentage to generate specific control commands that can be directly sent to the DCS. The online learning and adaptive unit monitors the optimization effect online and updates model parameters and decision rules online.

2. The combustion optimization control system based on multi-objective optimization according to claim 1, characterized in that: The data collection unit includes a fuel parameter acquisition terminal, an air system acquisition terminal, a combustion status acquisition terminal, an emission parameter acquisition terminal, an equipment status acquisition terminal, an extreme operating condition acquisition terminal, and a load demand acquisition terminal.

3. The combustion optimization control system based on multi-objective optimization according to claim 2, characterized in that: The fuel parameter acquisition terminal is responsible for collecting fuel type, fuel flow rate, fuel calorific value and fuel moisture content; the air system acquisition terminal is responsible for collecting primary air volume / pressure, secondary air volume / pressure and excess air coefficient.

4. The combustion optimization control system based on multi-objective optimization according to claim 2, characterized in that: The combustion status acquisition terminal is responsible for collecting multi-point furnace temperature, furnace negative pressure, flame intensity / frequency, and flue gas temperature; the emission parameter acquisition terminal is responsible for collecting flue gas NO. x Concentration, SO2 concentration, and smoke and dust concentration.

5. The combustion optimization control system based on multi-objective optimization according to claim 2, characterized in that: The equipment status acquisition terminal is responsible for acquiring the speed of the induced draft fan / forced draft fan, the opening degree of the fuel valve, and the operating voltage of the sensor; the extreme condition acquisition terminal is responsible for acquiring the sudden change value of the furnace pressure, the duration of the flame signal interruption, and the early warning signal of the fuel supply interruption; the load demand acquisition terminal is responsible for acquiring the real-time load of the power grid and the boiler steam pressure.

6. The combustion optimization control system based on multi-objective optimization according to claim 1, characterized in that: When the sensor fault diagnosis unit diagnoses a fault in sensor i, it takes data from redundant sensors j and k of the same type, and then calculates the compensation value for sensor i using the weighted compensation formula for redundant sensors. The calculation formula is as follows: In the formula, This represents the compensation value for faulty sensor i. This represents the weighting coefficient of redundant sensor j. This represents the weighting coefficient of the redundant sensor k. This represents the real-time data acquired by redundant sensor j, and the real-time output of sensor j of the same type in the data collection unit. This represents the real-time data collected by the redundant sensor k; By calculating the compensation value of sensor i As real-time alternative data to fault sensor i, it can be incorporated into the standardized process of the data preprocessing unit to enable continuous data chain connection.

7. The combustion optimization control system based on multi-objective optimization according to claim 1, characterized in that: The multi-objective optimization unit calculates boiler efficiency based on high-quality multi-source data. The calculation formula is as follows: In the formula, , These represent the specific enthalpy of steam at the outlet and the inlet, respectively. Indicates the main steam flow rate. Indicates the instantaneous flow rate of fuel. This indicates the received lower heating value of the fuel.

8. The combustion optimization control system based on multi-objective optimization according to claim 1, characterized in that: The multi-objective optimization unit calculates NO based on high-quality multi-source data. x Emissions The calculation formula is as follows: In the formula, Indicates measured dry basis NO x Volume concentration, This indicates the measured dry basis oxygen content. Standard reference oxygen content, Smoke density under standard conditions.

9. The combustion optimization control system based on multi-objective optimization according to claim 1, characterized in that: The multi-objective optimization unit calculates the combustion stability index based on high-quality multi-source data. The calculation formula is as follows: In the formula, , These represent the mean and standard deviation of the temperature at multiple points in the furnace, respectively. , This represents the mean and standard deviation of the negative pressure in the furnace. Indicates the cumulative duration of the flame signal interruption. Indicates the statistical period. , , These represent the normalized weighting coefficients for temperature fluctuation, pressure fluctuation, and flame interruption, respectively, satisfying... .

10. The combustion optimization control system based on multi-objective optimization according to claim 1, characterized in that: The multi-objective online control unit is based on boiler efficiency. NO x Emissions and combustion stability index The specific operation to output the Pareto optimal solution set is as follows: S1. Decision space initialization and population generation: Based on the current boiler load and safe operation boundary, determine the feasible range of five decision variables, and randomly generate an initial population containing multiple individuals within this range. S2. Multi-objective performance evaluation and ranking: Substitute the combination of decision variables for each individual in the population into the boiler efficiency. NO x Emissions and combustion stability index In the prediction model, the values ​​of its three objective functions are calculated; then, based on the core mechanism of NSGA-III, the individuals in the population are non-dominated and sorted to distinguish their superior and inferior levels, and the reference point mechanism is used to perform adaptive normalization and association operations on individuals in the same non-dominated level. S3. Elite Preservation and Iterative Evolution: Through tournament selection, simulated crossover and mutation genetic operations, a new offspring population is generated. Combined with the parent population, an elite preservation strategy is implemented to select the next generation population with better performance. This process is iterated until the convergence condition is met or the maximum number of iterations is reached. Finally, a uniformly distributed Pareto optimal solution set is output for the decision-making subunits within the multi-objective online control unit to select.

Images