Grey wolf algorithm flue gas waste heat control method and system introducing exhaust gas temperature prediction constraint

Abstract

The invention discloses a grey wolf algorithm flue gas waste heat control method and system introducing exhaust gas temperature prediction constraint, and the method comprises the steps: firstly building a thermodynamic simulation model of a flue gas waste heat system, collecting the working condition data of a unit in real time, and calibrating the model; under the targets of power supply coal consumption minimization, waste heat recovery rate maximization and the like, an exhaust gas temperature prediction result is used as a dynamic constraint, a grey wolf optimization algorithm is adopted to carry out multi-target optimization on parameters such as the opening degree of a flue gas baffle, the water supply distribution proportion of an economizer and the water flow of an air heater, and multiple regression or a BP neural network model is utilized to predict the exhaust gas temperature. If the predicted value is lower than the safety lower limit, dynamically adjusting the search direction to avoid an infeasible solution; and an optimization result is issued on line through a DCS system, and closed-loop control is realized. Compared with the prior art, on the premise that the safety of the exhaust gas temperature is guaranteed, the waste heat utilization rate is remarkably increased, power supply coal consumption is reduced, the capacity of adapting to different coal qualities and load fluctuations in a self-learning mode is achieved, and the good engineering application prospect is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of energy-saving optimization and operation control of thermal power units, and in particular to a flue gas waste heat control method and system using a grey wolf algorithm with exhaust temperature prediction constraints introduced. Background Art

[0002] Coal-fired power generation units, a crucial component of my country's baseload power supply, have always focused on energy conservation, emission reduction, and efficient operation. With rising environmental standards, the utilization of low-temperature waste heat from boiler exhaust is widely used to improve unit efficiency, reduce coal consumption, and minimize exhaust heat loss. In particular, the deployment of low-temperature economizers, air heaters, and flue gas bypass systems can significantly recover the heat energy from the boiler exhaust, improving thermal efficiency and lowering exhaust temperature.

[0003] However, optimizing the operation of flue gas waste heat utilization systems faces the challenges of multiple parameters and high coupling. Variables such as the flue gas bypass damper opening, economizer water distribution, and air preheater water flow rate synergistically influence waste heat utilization and exhaust temperature. These factors are also constrained by operating conditions such as main steam temperature, reheat temperature, and boiler load. In actual operation, if the exhaust temperature drops below the acid dew point, it can easily lead to operational risks such as low-temperature corrosion and ash accumulation. Therefore, a strict safety lower limit for the exhaust temperature is required.

[0004] Currently, the utilization of flue gas waste heat from thermal power units often relies on fixed strategies or manual experience-based adjustments. These control methods struggle to balance thermal economy and operational safety, especially under conditions of load fluctuations or changes in coal type, lacking dynamic prediction and coordinated control mechanisms. Furthermore, while some research has attempted to introduce intelligent optimization algorithms, such as genetic algorithms and particle swarm optimization, to optimize some unit parameters, these methods primarily focus on direct optimization of objective functions and lack predictive control capabilities for key constraints such as exhaust temperature. This can easily lead to optimized solutions failing to meet practical feasibility or impacting system safety.

[0005] The Gray Wolf Optimization Algorithm (GWA), a novel swarm intelligence optimization method, has been gradually applied to the optimization of complex energy systems due to its simple structure, strong global search capabilities, and rapid convergence. However, existing applications have not yet incorporated the exhaust gas temperature safety boundary in thermal systems for targeted modeling and prediction constraints. In particular, the algorithm's fitness model lacks a dynamic assessment mechanism for the flue gas system's safety lower limit, limiting its engineering applicability and control accuracy in deep waste heat recovery systems.

[0006] Therefore, there is an urgent need for an intelligent optimization method that can combine the physical boundaries of the thermal system and the operational safety constraints, effectively integrate the exhaust temperature prediction mechanism into the optimization algorithm, and ensure the safety and controllability of the unit operation while achieving energy-saving goals. Summary of the Invention

[0007] The purpose of the present invention is to overcome the shortcomings of existing coal-fired unit flue gas waste heat control optimization methods, such as the lack of exhaust temperature prediction mechanism, the optimization solution easily breaking through the safety boundary, and the poor system execution stability. A gray wolf algorithm flue gas waste heat control method and system with exhaust temperature prediction constraints are proposed to achieve energy-saving optimized operation of the unit while ensuring safe exhaust temperature.

[0008] In order to achieve the above object, the present invention adopts the following technical solutions:

[0009] The flue gas waste heat control method using the Grey Wolf algorithm with exhaust temperature prediction constraint includes the following steps:

[0010] Step S1: constructing a thermodynamic simulation model of the unit's flue gas waste heat system, the thermodynamic simulation model including the economizer, air heater, flue gas bypass channel and related heat exchange equipment, and performing thermal parameter simulation based on a simulation platform;

[0011] Step S2: collecting the current operating data of the unit and inputting it into the thermodynamic simulation model to calibrate the model parameters;

[0012] Step S3: constructing an exhaust gas temperature prediction model based on the calibrated thermodynamic simulation model and the historical operation data of the unit, wherein the exhaust gas temperature prediction model is a multiple linear regression model, a BP neural network model or a grey prediction model;

[0013] Step S4: setting an optimization objective function, including minimizing power supply coal consumption, maximizing flue gas waste heat recovery rate, and taking the exhaust gas temperature output by the prediction model as no less than a preset safety lower limit as a constraint condition;

[0014] Step S5: setting optimization variables, including: flue gas damper opening, economizer feed water distribution ratio, and heater heat medium water flow rate;

[0015] Step S6: performing multi-objective optimization on the objective function based on the Grey Wolf Optimization Algorithm, introducing the predicted exhaust temperature as a dynamic constraint term in the fitness function, and using a penalty function to constrain it not to be lower than a safety lower limit;

[0016] Step S7: When the predicted exhaust temperature corresponding to the candidate solution is lower than the safety lower limit, dynamically adjust the search path and step size parameters of the gray wolf individual to guide it away from the infeasible solution area;

[0017] Step S8: The final optimization result is used as a control instruction and sent to the corresponding actuators, including the flue gas damper, water supply regulating valve and variable frequency pump, through the DCS system to implement the control instruction.

[0018] A further improvement of the present invention is that the input parameters of the exhaust gas temperature prediction model include: flue gas flow rate, economizer outlet water temperature, low calorific value of coal, fuel sulfur content and secondary air and primary air distribution ratio.

[0019] A further improvement of the present invention is that the fitness function takes the following form:

[0020] J(X)=C 煤耗 (X)+λ·max{0,T 安全下限 -T 预测 (X)} 2 -η·R 余热 (X)

[0021] Among them, J(X) is the unit power supply coal consumption corresponding to the current control solution, C 煤耗 (X) is the unit power supply coal consumption corresponding to the current control solution, T 预测 (X) is the predicted exhaust temperature, T 安全下限 (X) is the set exhaust gas temperature safety lower limit, λ is the penalty coefficient, η is the weight factor of the waste heat recovery target, R 余热 (X) is the waste heat recovery rate or equivalent energy saving of the current solution.

[0022] A further improvement of the present invention is that the gray wolf optimization algorithm introduces an exhaust temperature safety boundary guidance mechanism during the search process. When the exhaust temperature predicted by the candidate solution is close to the safety lower limit boundary, the gray wolf individual is guided to move in the direction where the predicted value is within the safety range.

[0023] A further improvement of the present invention is that the final optimization result includes set values ​​for each major control link, including: flue gas damper opening, economizer water distribution ratio and heater heat medium water flow, which are automatically transmitted to the DCS system through the controller interface at a preset cycle for implementing closed-loop control.

[0024] A further improvement of the present invention is that the individual position update of the gray wolf optimization algorithm adopts the following method:

[0025]

[0026] Among them, X1, X2, and X3 represent the average position vectors of the gray wolf individuals' guidance results for the three types of alpha, β, and δ alpha alpha alpha, respectively. The coefficients used in the update process include the convergence factor a, whose value decreases linearly with the number of iterations.

[0027] A further improvement of the present invention is that the exhaust gas temperature prediction model is retrained or updated before each optimization run, and the model coefficients are corrected online according to the current coal type parameters and load conditions to maintain the prediction accuracy.

[0028] A further improvement of the present invention is that the control method has the ability to identify abnormal solutions. When the deviation between the predicted exhaust temperature and the actual operating value in two consecutive rounds of optimization exceeds a preset threshold, an early warning is automatically triggered, and the operating personnel are prompted to check the input data or switch to a safe and conservative mode of operation.

[0029] The flue gas waste heat control system with the Grey Wolf algorithm and exhaust temperature prediction constraint is introduced, including:

[0030] A simulation model construction unit constructs a thermodynamic simulation model of the unit's flue gas waste heat system, which includes an economizer, a heater, a flue gas bypass channel, and related heat exchange equipment, and performs thermal parameter simulation based on a simulation platform;

[0031] A model parameter calibration unit collects the current operating data of the unit and inputs it into the thermodynamic simulation model to calibrate the model parameters;

[0032] A prediction model building unit, which builds an exhaust gas temperature prediction model based on the calibrated thermodynamic simulation model and the historical operation data of the unit, wherein the exhaust gas temperature prediction model is a multiple linear regression model, a BP neural network model or a grey prediction model;

[0033] An optimization objective function setting unit is configured to set an optimization objective function, including minimizing power supply coal consumption, maximizing flue gas waste heat recovery rate, and taking the exhaust gas temperature output by the prediction model as not lower than a preset safety lower limit as a constraint condition;

[0034] An optimization variable setting unit for setting optimization variables, wherein the optimization variables include: flue gas damper opening, economizer feed water distribution ratio, and heater heat medium water flow rate;

[0035] A multi-objective optimization unit performs multi-objective optimization on the objective function based on a gray wolf optimization algorithm, introduces the predicted exhaust temperature as a dynamic constraint term in the fitness function, and uses a penalty function to constrain it not to be lower than a safety lower limit;

[0036] The judgment unit dynamically adjusts the search path and step size parameters of the gray wolf individuals to guide them away from the infeasible solution area when the predicted exhaust temperature corresponding to the candidate solution is lower than the safety lower limit;

[0037] The execution unit takes the final optimization result as a control instruction and sends it to the corresponding actuators, including the flue gas damper, water supply regulating valve and variable frequency pump, through the DCS system to implement the control instruction.

[0038] A further improvement of the present invention is that the input parameters of the flue gas temperature prediction model in the prediction model construction unit include: flue gas flow rate, economizer outlet water temperature, low calorific value of coal type, fuel sulfur content and secondary air and primary air distribution ratio.

[0039] Compared with the prior art, the present invention has at least the following beneficial technical effects:

[0040] Compared with the existing technology, the present invention introduces an exhaust temperature prediction mechanism as a dynamic constraint factor, which avoids the risk of the optimized solution breaking through the exhaust temperature safety lower limit under the energy-saving target, and improves the feasibility and safety of the optimization algorithm. By constructing a prediction model based on the current operating variables, the exhaust temperature response of the scheme can be evaluated in real time during the optimization process, forming a closed-loop correction mechanism. In addition, the penalty factor and search guidance mechanism introduced in the fitness function of the gray wolf algorithm enhance the algorithm's ability to adapt to constraints. At the system level, this method fully combines the physical structure of the thermal system with the DCS execution system, has good engineering deployment capabilities and scalability, is suitable for a variety of coal qualities and load fluctuation scenarios, and can achieve efficient, safe and intelligent control of the flue gas waste heat system. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0042] Figure 1 This is a schematic structural diagram of the flue gas waste heat system of a coal-fired unit according to the present invention;

[0043] Figure 2 This is a diagram of the optimization control system architecture of the method of the present invention;

[0044] Figure 3 Optimization flow chart of the Grey Wolf algorithm with exhaust temperature prediction constraints introduced;

[0045] Figure 4 Construct a flow chart for the exhaust temperature prediction model;

[0046] Figure 5 This is a structural block diagram of the flue gas waste heat control system using the Grey Wolf algorithm that introduces exhaust temperature prediction constraints. DETAILED DESCRIPTION

[0047] Hereinafter, only certain exemplary embodiments are briefly described. As will be appreciated by those skilled in the art, the described embodiments may be modified in various ways without departing from the spirit or scope of the present invention. Therefore, the drawings and description are to be considered as illustrative in nature and not restrictive.

[0048] In the description of the present invention, it is to be understood that when used in this specification and the appended claims, the terms “include” and “comprise” indicate the presence of the described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or collections thereof.

[0049] It should also be understood that the terms used in the present specification are only for the purpose of describing particular embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, the singular forms "a", "an", and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0050] It should be further understood that the term "and / or" used in the present description and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0051] The accompanying drawings illustrate various schematic diagrams of structures according to embodiments disclosed herein. These figures are not drawn to scale; for clarity, some details are exaggerated and some details may be omitted. The shapes of the various regions and layers shown in the figures, as well as their relative sizes and positional relationships, are merely exemplary and may deviate in practice due to manufacturing tolerances or technical limitations. Those skilled in the art may design regions / layers with different shapes, sizes, and relative positions as needed.

[0052] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0053] Example 1

[0054] The present invention provides a method for controlling flue gas waste heat using a Grey Wolf algorithm with exhaust temperature prediction constraints, comprising the following steps:

[0055] Step S1: constructing a thermodynamic simulation model of the unit's flue gas waste heat system, the thermodynamic simulation model including the economizer, air heater, flue gas bypass channel and related heat exchange equipment, and performing thermal parameter simulation based on a simulation platform;

[0056] Step S2: collecting the current operating data of the unit and inputting it into the thermodynamic simulation model to calibrate the model parameters;

[0057] Step S3: constructing an exhaust gas temperature prediction model based on the calibrated thermodynamic simulation model and the historical operation data of the unit, wherein the exhaust gas temperature prediction model is a multiple linear regression model, a BP neural network model or a grey prediction model;

[0058] Step S4: setting an optimization objective function, including minimizing power supply coal consumption, maximizing flue gas waste heat recovery rate, and taking the exhaust gas temperature output by the prediction model as no less than a preset safety lower limit as a constraint condition;

[0059] Step S5: setting optimization variables, including: flue gas damper opening, economizer feed water distribution ratio, and heater heat medium water flow rate;

[0060] Step S6: performing multi-objective optimization on the objective function based on the Grey Wolf Optimization Algorithm, introducing the predicted exhaust temperature as a dynamic constraint term in the fitness function, and using a penalty function to constrain it not to be lower than a safety lower limit;

[0061] Step S7: When the predicted exhaust temperature corresponding to the candidate solution is lower than the safety lower limit, dynamically adjust the search path and step size parameters of the gray wolf individual to guide it away from the infeasible solution area;

[0062] Step S8: The final optimization result is used as a control instruction and sent to the corresponding actuators, including the flue gas damper, water supply regulating valve and variable frequency pump, through the DCS system to implement the control instruction.

[0063] In this embodiment, the input parameters of the exhaust gas temperature prediction model include: flue gas flow rate, economizer outlet water temperature, low calorific value of coal, fuel sulfur content, and secondary air and primary air distribution ratio.

[0064] In this embodiment, the fitness function takes the following form:

[0065] J(X)=C 煤耗 (X)+λ·max{0,T 安全下限 -T 预测 (X)} 2 -η·R 余热 (X)

[0066] Among them, J(X) is the unit power supply coal consumption corresponding to the current control solution, C 煤耗 (X) is the unit power supply coal consumption corresponding to the current control solution, T 预测 (X) is the predicted exhaust temperature, T 安全下限 (X) is the set exhaust gas temperature safety lower limit, λ is the penalty coefficient, η is the weight factor of the waste heat recovery target, R 余热 (X) is the waste heat recovery rate or equivalent energy saving of the current solution.

[0067] In this embodiment, the gray wolf optimization algorithm introduces an exhaust temperature safety boundary guidance mechanism during the search process. When the exhaust temperature predicted by the candidate solution is close to the safety lower limit boundary, the gray wolf individual is guided to move in the direction where the predicted value is within the safety range.

[0068] In this embodiment, the optimization results include set values ​​for each major control link, specifically including: flue gas damper opening, economizer water distribution ratio, and heater heat medium water flow rate. These set values ​​are automatically transmitted to the DCS system through the controller interface at a preset cycle for implementing closed-loop control.

[0069] In this embodiment, the individual position update of the gray wolf optimization algorithm adopts the following method:

[0070]

[0071] Among them, X1+X2+X3 represent the average position vectors of the gray wolf individuals’ guidance results for the three types of alpha, β, and δ alpha alpha alpha, respectively. The coefficients used in the updating process include the convergence factor a, whose value decreases linearly with the number of iterations.

[0072] In this embodiment, the exhaust gas temperature prediction model is retrained or updated before each optimization run, and the model coefficients are corrected online according to the current coal type parameters and load conditions to maintain the prediction accuracy.

[0073] In this embodiment, the control method has the ability to identify abnormal solutions. When the deviation between the predicted exhaust temperature and the actual operating value in two consecutive rounds of optimization exceeds a preset threshold, the optimization control system automatically triggers an early warning and prompts the operator to check the input data or switch to a safe and conservative mode.

[0074] Example 2

[0075] The present invention provides a flue gas waste heat control method using a Grey Wolf algorithm with exhaust temperature prediction constraints, comprising:

[0076] (1) Build a simulation platform

[0077] like Figure 1 As shown, the flue gas waste heat system described in the present invention includes an economizer, a heater, a flue gas bypass baffle, an air preheater and related piping systems, which are the core components of the flue gas waste heat recovery of the unit. This embodiment takes a 1000MW supercritical coal-fired power generation unit as the application object to illustrate the method of the present invention. The unit is equipped with a flue gas waste heat recovery system, including high / low temperature economizers, air preheaters (heaters), flue gas bypass baffles and related heat exchange equipment. First, the EBSILON thermal system simulation software is used to construct a thermodynamic model of the unit's flue gas waste heat system. The model covers the key equipment and processes of the boiler tail heating surface (economizer, air preheater) and the flue gas heat exchange circuit, and calculates the thermal parameters under each working condition by setting boundary conditions.

[0078] After the model is established, the unit's current operating condition data is collected to calibrate the model (step S2). The operating condition data includes, but is not limited to, generator output power, boiler fuel feed and load rate, tail exhaust temperature, main steam temperature, reheat steam temperature, flue gas flow rate, air volume distribution ratio, and economizer inlet / outlet water temperature. The above measured data is input into the EBSILON model for simulation calculation, and the model parameters are adjusted to make the simulation results consistent with the unit's measured values, thereby ensuring that the model can accurately reflect the unit's actual characteristics. Through this calibrated simulation model, the various performance indicators of the unit before optimization can be obtained, providing a benchmark for comparison of subsequent optimization results.

[0079] (2) Constructing exhaust temperature prediction model

[0080] In order to achieve dynamic constraints on the lower limit of exhaust temperature safety, this embodiment introduces an exhaust temperature prediction model (step S3) during the optimization process. The exhaust temperature prediction model uses a multivariate linear regression model to achieve real-time estimation of the exhaust temperature at the tail end of the unit. The input parameters of the model are combined with the key variables affecting the exhaust temperature during unit operation, including: flue gas flow rate (or flue gas volume), economizer outlet feed water temperature, fuel characteristic parameters (such as calorific value, moisture, sulfur content, etc. of coal), and supply air / primary air volume distribution. Using the historical operation data of the unit and the calculation results of the simulation model as samples, the coefficients of the multivariate linear regression equation are trained and calibrated by the least squares method.

[0081] The constructed linear regression model is as follows:

[0082] T 排烟预测 =b0+b1X1+b2X2+…+b n X n

[0083] Among them, T 排烟预测 is the predicted exhaust gas temperature, X1, X2,…, X n are the selected independent variables affecting the exhaust gas temperature, such as flue gas flow rate, economizer outlet water temperature, coal low calorific value, air ratio, etc. b0 is the regression constant, b1~b n are the regression coefficients corresponding to the respective variables. This model can select linear terms, interaction terms, and other factors as needed to improve prediction accuracy. In practical applications, to ensure the model's applicability to different operating conditions, the regression model can be retrained or its coefficients updated before each optimization run. Model parameters can be calibrated online using the latest unit operating data to keep prediction errors within an acceptable range (e.g., an absolute error of no more than 5°C).

[0084] Through the above multivariate linear regression model, the exhaust gas temperature T after applying a certain control scheme can be quickly predicted according to the current unit operation status. 排烟预测This predicted value will be introduced as a dynamic constraint factor in the optimization algorithm to evaluate the safety of candidate solutions in real time and ensure that the optimization process does not sacrifice the exhaust temperature safety margin. Figure 4 As shown in Figure 2, the construction process of the exhaust temperature prediction model includes data collection, feature selection, model training and online correction.

[0085] (3) Setting optimization objectives and control variables

[0086] Before carrying out the optimization calculation, it is necessary to clarify the optimization objective function and the range of control variables (steps S4 and S5). The present invention sets three optimization goals for the utilization of waste heat from flue gas of the unit: one is to reduce the coal consumption for power supply and minimize the coal consumption rate of the unit; the second is to ensure that the exhaust temperature is not lower than the safety lower limit (that is, to meet the operation safety constraints); the third is to improve the waste heat recovery rate of flue gas and maximize the use of tail waste heat to increase the boiler feed water temperature or air temperature. In view of the above goals, this embodiment expresses the optimization problem as a multi-objective optimization, in which the coal consumption for power supply is used as the main optimization indicator, the waste heat recovery rate is used as a collaborative optimization indicator, and the lower limit constraint of the exhaust temperature is guaranteed by introducing a penalty function.

[0087] The control variables are selected from key adjustable parameters during unit operation, including the flue gas bypass damper opening, the high-pressure economizer flow control valve opening, and the speeds of variable frequency pumps I and II. The flue gas bypass damper opening refers to the opening of the damper valve that allows flue gas to bypass a portion of the air preheater. The high-pressure economizer flow control valve opening regulates the feedwater flow to the high-pressure economizer, with a range of 0 to 100%. The speed of variable frequency pump I regulates the condensate flow to the low-pressure economizer. The speed of variable frequency pump II regulates the heat transfer water flow to the low- and low-pressure economizers and air heaters, thereby controlling the air preheater inlet air temperature and the exhaust gas temperature. Each of these control variables has its own physical limits and operational constraints, and must be kept within safe and feasible ranges during the optimization process. For example, the damper opening must not exceed its mechanical limits, and the distribution ratio and water flow must not exceed the design values.

[0088] (4) Optimization search and prediction constraint fusion based on the gray wolf algorithm

[0089] like Figure 3As shown, the gray wolf optimization process with the exhaust temperature prediction constraint introduced includes the steps of initialization, fitness evaluation, individual position update and constraint judgment. After determining the objectives and variables, the gray wolf optimization algorithm (GWO) is used to optimize the multi-objective function (step S6). The gray wolf optimization algorithm is a bionic intelligent algorithm that performs a global search for the optimization problem by simulating the predation behavior and hierarchical division of labor in the gray wolf society. Specifically, each gray wolf individual is represented as a candidate solution vector, corresponding to a set of control variable values ​​X = (damper opening, water distribution ratio, heater water flow). During initialization, a number of gray wolf individuals (for example, 20) are randomly generated within the allowable range of each control variable as the initial population.

[0090] In order to evaluate the quality of each candidate solution, a fitness function of the optimization objective is established. For multiple objectives, each objective is usually weighted and combined into a single scalar objective function. In this embodiment, the fitness function is designed as:

[0091] J(X)=C 煤耗 (X)+Φ(T 安全下限 -T 排烟预测 (X))-η·R 余热 (X)

[0092] Among them, J(X) is the unit power supply coal consumption corresponding to the current control solution, C 煤耗 (X) represents the coal consumption of the power generation unit corresponding to the current solution X (the smaller the value, the better), R 余热 (X) represents the waste heat recovery rate or equivalent energy saving of the current solution (the larger the value, the better), and η is the weight coefficient used to balance the coal consumption and waste heat utilization goals; T 排烟预测 (X) is the predicted value of exhaust temperature under this solution calculated by the above regression model. The second term of the function is the penalty function of exhaust temperature constraint, T 安全下限 Indicates the set exhaust temperature safety lower limit. When the predicted exhaust temperature is higher than the safety lower limit, the value of this item is 0; when the predicted value is lower than the lower limit, the penalty function is triggered. The penalty function can be selected as a linear or quadratic increase as needed. For example, in this embodiment, a quadratic penalty form is used:

[0093]

[0094] Where λ is the penalty coefficient. A large value is chosen to ensure that solutions that violate the temperature constraint are penalized sufficiently, causing the fitness of such solutions to deteriorate dramatically. Therefore, during the optimization process, any candidate solution that causes the exhaust temperature to fall below the safe lower limit will be eliminated or guided to change due to the high penalty.

[0095] The Gray Wolf Algorithm updates its solutions based on the wolf pack's social hierarchy. In each iteration, the solution with the best fitness in the current population is defined as "α wolf," the second and third best solutions are defined as "β wolf" and "δ wolf," and the rest are ordinary "ω wolves." Its position update then follows the following model:

[0096] (a) Calculate the position vector differences and convergence factors between the three alpha, β, and δ alpha alpha and the rest of the individuals:

[0097] D α =|C1·X α -X|,

[0098] D β =|C2·X β -X|,

[0099] D δ =|C3·X δ -X|,

[0100] Among them, X is the position vector of the gray wolf individual to be updated, X α , X β , X δ are the position vectors of α, β, and δ wolves respectively; C1, C2, and C3 are coefficient vectors, usually defined as C i =2·r i , r i ∈[0,1] is a random vector.

[0101] (b) Calculate the temporary target position vectors X1, X2, X3, which represent the current individual's position relative to the prey being tracked by α, β, and δ wolves:

[0102] X1=X α -A1·D α ,

[0103] X2=X β -A2·D β ,

[0104] X3=X δ -A3·D δ ,

[0105] Among them, A i =2a·r i -a,r i The same as the above C i The same random factor is used; a is the convergence factor, which is initialized to 2 and decreases linearly to 0 with each iteration. The function of this coefficient a is to gradually narrow the search range, so that the algorithm smoothly transitions from the initial global exploration to the later local exploration.

[0106] (c) Update the current wolf position vector to the average of the three temporary positions mentioned above:

[0107]

[0108] Through this mechanism, individual gray wolves, guided by alpha, beta, and delta, approach their prey, continuously adjusting control variables to achieve a better solution. The algorithm iterates this process until a termination condition is met (e.g., the number of iterations reaches a preset limit or the improvement in fitness falls below a threshold).

[0109] In the search iteration process of the Gray Wolf Algorithm, the present invention integrates the exhaust temperature prediction constraint into the search guidance mechanism to form a dynamic adjustment strategy of "exhaust temperature safety boundary" (corresponding to step S7). Specifically, when each candidate solution is generated and evaluated, the prediction model is used to calculate the T corresponding to the solution. 排烟预测 If T appears 排烟预测 <T 安全下限 In the case of , the current candidate solution is determined to violate the safety constraint. In addition to giving a high penalty in the fitness function, the algorithm will also automatically adjust the position update strategy of the gray wolf individual: for example, reducing the exploration step length a of the individual or increasing the random coefficient r i The disturbance will make it search in the direction away from the default area (low temperature area) in the next step; or directly change the position X of the individual new Pull back to the nearest boundary value within the temperature safety range (equivalent to correcting the infeasible solution). At the same time, when it is detected that multiple wolf individuals are moving in a direction that violates the temperature constraint, the entire population's vigilance against the lower temperature limit is increased. That is, when updating, it tends to refer to solutions with good performance (high fitness and meeting the temperature constraint) within the current temperature safety range, prompting the population to gather towards the safe area. Through the above measures, the gray wolf algorithm achieves dynamic avoidance of the lower limit of the exhaust temperature during the optimization process, which is equivalent to establishing a "safety boundary" in the multidimensional search space, guiding the algorithm to always find the optimal solution within the safety boundary and avoiding the generation of infeasible control strategies. This search method that integrates predictive constraints improves the feasibility and safety of the optimized solution.

[0110] (5) Key parameter setting and logical judgment

[0111] Smoke exhaust temperature safety lower limit T 安全下限 The value of should be determined by comprehensively considering factors such as the material of the unit's heating surface, the flue gas dew point, and the sulfur content of the fuel. Generally speaking, the lower limit can be determined based on the unit's previous low-load operation tests or the manufacturer's recommendations. For example, the lower limit of a unit is about 90°C when burning low-sulfur coal, and about 120°C when burning high-sulfur coal. During operation, if the type of fuel coal changes or the corrosion margin requirements of the tail heating surface are adjusted, this lower limit can be dynamically modified. In this embodiment, T 安全下限 Take 120℃ to ensure that the air preheater and flue do not suffer from dew point corrosion.

[0112] This method implements optimized control strategies through the DCS, eliminating the need for fixed control cycle constraints. Optimization can be triggered based on changes in unit operating conditions. For example, when the load fluctuates significantly, the ambient temperature changes, or the coal quality changes, an optimization calculation is initiated to adjust the control parameters; when operating conditions are stable, the current optimization results are maintained. Furthermore, a periodic execution interval (e.g., every 5 to 10 minutes) can be set to verify the availability of newer, more optimal solutions. The optimization calculation itself can be performed online on a background computer, and the results are sent to the actuator via DCS instructions to implement closed-loop control adjustments.

[0113] In order to ensure the reliability of the prediction model, the system sets a prediction error monitoring threshold. When there is a significant deviation between the actual operating exhaust gas temperature and the model predicted value, corresponding processing needs to be triggered. In this embodiment, the threshold is set to a deviation of more than 5°C between the predicted value and the measured value in two consecutive rounds of optimization (approximately 3% to 5% of the typical exhaust gas temperature). Once this condition is triggered (refer to the abnormal solution discrimination capability of claim 8), the system will automatically issue an early warning signal and prompt the operator to check the sensor readings and model parameters through the human-machine interface; at the same time, the control strategy can temporarily fall back to a safe and conservative operating mode (for example, closing some air preheater bypass flue gas dampers to increase the exhaust gas temperature) to avoid unsafe operating conditions due to inaccurate predictions. After the problem is solved or the model is recalibrated, the optimization control operation is resumed.

[0114] The algorithm population size and number of iterations will affect the optimization effect and speed. In this embodiment, the number of gray wolf populations is 20, and the maximum number of iterations is 50; the convergence factor a decreases linearly from 2 to 0, and the penalty coefficient λ takes a larger value to strictly constrain the lower limit of temperature (for example, λ = 1000). These parameters can be adjusted appropriately according to the unit size and computing resources. After each round of optimization calculation is completed, the control parameters corresponding to the optimal solution are sent to the DCS, and the DCS sends the opening command and valve adjustment command to the on-site actuator to realize control.

[0115] (6) Control strategy implementation and effect verification

[0116] like Figure 2As shown, the method of the present invention is deployed in an online optimization control system, and a closed-loop control path is formed through the interaction of models, algorithms and DCS systems. After the optimal control parameters obtained by optimization calculation (including the opening of the flue gas bypass damper, the economizer feed water distribution ratio, and the heater heat medium water flow) are transmitted to each execution unit of the unit through the DCS system, the unit enters operation according to the optimal strategy. Since the present method has taken into account the exhaust gas temperature constraint during the optimization process, it ensures that the implemented control strategy will not cause the exhaust gas temperature to fall below the safety lower limit. The exhaust gas temperature monitored in actual operation is maintained above the safety value (about 125°C in this embodiment). At the same time, the optimized waste heat recovery strategy enables more flue gas waste heat to be used for heating feed water or air, thereby improving the thermal efficiency of the boiler. After the unit was put into optimized control, the power supply coal consumption was significantly reduced compared with before optimization, and the boiler efficiency was improved. For example, simulation calculations and actual operation results show that, under the same load conditions, optimized control reduces the unit's coal consumption from approximately 285g / kWh to around 280g / kWh, a reduction of approximately 1.7%. The exhaust temperature also drops from 130°C to approximately 125°C, maintaining a safe range with a slight increase to enhance heat transfer. Increased waste heat absorbed by the low-temperature economizer raises the feedwater temperature by approximately 5°C. These data fully demonstrate the effectiveness of the proposed method: while ensuring the exhaust temperature does not fall below the safety limit, it simultaneously improves the unit's economy and safety.

[0117] Example 3

[0118] like Figure 5 As shown, the present invention provides a flue gas waste heat control system with a Grey Wolf algorithm that introduces exhaust temperature prediction constraints, including:

[0119] A simulation model construction unit constructs a thermodynamic simulation model of the unit's flue gas waste heat system, which includes an economizer, a heater, a flue gas bypass channel, and related heat exchange equipment, and performs thermal parameter simulation based on a simulation platform;

[0120] A model parameter calibration unit collects the current operating data of the unit and inputs it into the thermodynamic simulation model to calibrate the model parameters;

[0121] A prediction model building unit, which builds an exhaust gas temperature prediction model based on the calibrated thermodynamic simulation model and the historical operation data of the unit, wherein the exhaust gas temperature prediction model is a multiple linear regression model, a BP neural network model or a grey prediction model;

[0122] An optimization objective function setting unit is configured to set an optimization objective function, including minimizing power supply coal consumption, maximizing flue gas waste heat recovery rate, and taking the exhaust gas temperature output by the prediction model as not lower than a preset safety lower limit as a constraint condition;

[0123] An optimization variable setting unit for setting optimization variables, wherein the optimization variables include: flue gas damper opening, economizer feed water distribution ratio, and heater heat medium water flow rate;

[0124] A multi-objective optimization unit performs multi-objective optimization on the objective function based on a gray wolf optimization algorithm, introduces the predicted exhaust temperature as a dynamic constraint term in the fitness function, and uses a penalty function to constrain it not to be lower than a safety lower limit;

[0125] The judgment unit dynamically adjusts the search path and step size parameters of the gray wolf individuals to guide them away from the infeasible solution area when the predicted exhaust temperature corresponding to the candidate solution is lower than the safety lower limit;

[0126] The execution unit takes the final optimization result as a control instruction and sends it to the corresponding actuators, including the flue gas damper, water supply regulating valve and variable frequency pump, through the DCS system to implement the control instruction.

[0127] In this embodiment, the input parameters of the flue gas temperature prediction model in the prediction model building unit include: flue gas flow rate, economizer outlet water temperature, low calorific value of coal type, fuel sulfur content, and secondary air and primary air distribution ratio.

[0128] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, from all points of view, the embodiments should be regarded as illustrative and non-restrictive. The scope of the present invention is defined by the appended claims rather than the above description, and it is intended that all changes that fall within the meaning and range of equivalents of the claims are included in the present invention. Any reference signs in the claims should not be construed as limiting the claim to which they relate.

[0129] In addition, it should be understood that although this specification describes the embodiments, not every embodiment contains only one independent technical solution. This description is for clarity only. Those skilled in the art should consider the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art. The above content is only for the purpose of illustrating the technical concept of the present invention and cannot be used to limit the scope of protection of the present invention. Any changes made based on the technical solution in accordance with the technical concept proposed by the present invention fall within the scope of protection of the claims of the present invention.

Claims

1. The Grey Wolf algorithm flue gas waste heat control method with exhaust temperature prediction constraint is introduced, which is characterized by: The following steps are involved: Step S1: constructing a thermodynamic simulation model of the unit's flue gas waste heat system, the thermodynamic simulation model including the economizer, air heater, flue gas bypass channel and related heat exchange equipment, and performing thermal parameter simulation based on a simulation platform; Step S2: collecting the current operating data of the unit and inputting it into the thermodynamic simulation model to calibrate the model parameters; Step S3: constructing an exhaust gas temperature prediction model based on the calibrated thermodynamic simulation model and the historical operation data of the unit, wherein the exhaust gas temperature prediction model is a multiple linear regression model, a BP neural network model or a grey prediction model; Step S4: setting an optimization objective function, including minimizing power supply coal consumption, maximizing flue gas waste heat recovery rate, and taking the exhaust gas temperature output by the prediction model as no less than a preset safety lower limit as a constraint condition; Step S5: setting optimization variables, including: flue gas damper opening, economizer feed water distribution ratio, and heater heat medium water flow rate; Step S6: performing multi-objective optimization on the objective function based on the Grey Wolf Optimization Algorithm, introducing the predicted exhaust temperature as a dynamic constraint term in the fitness function, and using a penalty function to constrain it not to be lower than a safety lower limit; Step S7: When the predicted exhaust temperature corresponding to the candidate solution is lower than the safety lower limit, dynamically adjust the search path and step size parameters of the gray wolf individual to guide it away from the infeasible solution area; Step S8: The final optimization result is used as a control instruction and sent to the corresponding actuators, including the flue gas damper, water supply regulating valve and variable frequency pump, through the DCS system to implement the control instruction.

2. The method for controlling flue gas waste heat using the Grey Wolf algorithm with exhaust temperature prediction constraint introduced according to claim 1 is characterized in that: The input parameters of the exhaust gas temperature prediction model include: flue gas flow rate, economizer outlet water temperature, low calorific value of coal type, fuel sulfur content, and secondary air and primary air distribution ratio.

3. The flue gas waste heat control method using the Grey Wolf algorithm with exhaust temperature prediction constraint introduced according to claim 1 is characterized in that: The fitness function takes the following form: J(X)=C 煤耗 (X)+λ·max{0,T 安全下限 -T 预测 (X)} 2 -η·R 余热 (X) Among them, J(X) is the unit power supply coal consumption corresponding to the current control solution, C 煤耗 (X) is the unit power supply coal consumption corresponding to the current control solution, T 预测 (X) is the predicted exhaust temperature, T 安全下限 (X) is the set exhaust gas temperature safety lower limit, λ is the penalty coefficient, η is the weight factor of the waste heat recovery target, R 余热 (X) is the waste heat recovery rate or equivalent energy saving of the current solution.

4. The method for controlling flue gas waste heat using the Grey Wolf algorithm with exhaust temperature prediction constraint introduced according to claim 1 is characterized in that: The gray wolf optimization algorithm introduces an exhaust temperature safety boundary guidance mechanism during the search process. When the exhaust temperature predicted by the candidate solution is close to the safety lower limit boundary, the gray wolf individual is guided to move in the direction where the predicted value is within the safety range.

5. The method for controlling flue gas waste heat using the Grey Wolf algorithm with exhaust temperature prediction constraint introduced according to claim 1 is characterized in that: The final optimization results include set values ​​for each major control link, including: flue gas damper opening, economizer water distribution ratio and heater heat medium water flow, which are automatically transmitted to the DCS system through the controller interface at a preset period for closed-loop control.

6. The flue gas waste heat control method using the Grey Wolf algorithm with exhaust temperature prediction constraint introduced according to claim 1 is characterized in that: The individual position update of the gray wolf optimization algorithm adopts the following method: Among them, X1, X2, and X3 represent the average position vectors of the gray wolf individuals' guidance results for the three types of alpha, β, and δ alpha alpha alpha, respectively. The coefficients used in the update process include the convergence factor a, whose value decreases linearly with the number of iterations.

7. The method for controlling flue gas waste heat using the Grey Wolf algorithm with exhaust temperature prediction constraint introduced according to claim 1 is characterized in that: The exhaust gas temperature prediction model is retrained or updated before each optimization run, and the model coefficients are corrected online according to the current coal type parameters and load conditions to maintain the prediction accuracy.

8. The method for controlling flue gas waste heat using the Grey Wolf algorithm with exhaust temperature prediction constraint introduced according to claim 1 is characterized in that: The control method has the ability to identify abnormal solutions. When the deviation between the predicted exhaust temperature and the actual operating value in two consecutive rounds of optimization exceeds a preset threshold, an early warning is automatically triggered, and the operator is prompted to check the input data or switch to a safe and conservative mode of operation.

9. The Grey Wolf algorithm flue gas waste heat control system with exhaust temperature prediction constraint is introduced, which is characterized by: include: A simulation model construction unit constructs a thermodynamic simulation model of the unit's flue gas waste heat system, which includes an economizer, a heater, a flue gas bypass channel, and related heat exchange equipment, and performs thermal parameter simulation based on a simulation platform; A model parameter calibration unit collects the current operating data of the unit and inputs it into the thermodynamic simulation model to calibrate the model parameters; A prediction model building unit, which builds an exhaust gas temperature prediction model based on the calibrated thermodynamic simulation model and the historical operation data of the unit, wherein the exhaust gas temperature prediction model is a multiple linear regression model, a BP neural network model or a grey prediction model; An optimization objective function setting unit is configured to set an optimization objective function, including minimizing power supply coal consumption, maximizing flue gas waste heat recovery rate, and taking the exhaust gas temperature output by the prediction model as not lower than a preset safety lower limit as a constraint condition; An optimization variable setting unit for setting optimization variables, wherein the optimization variables include: flue gas damper opening, economizer feed water distribution ratio, and heater heat medium water flow rate; A multi-objective optimization unit performs multi-objective optimization on the objective function based on a gray wolf optimization algorithm, introduces the predicted exhaust temperature as a dynamic constraint term in the fitness function, and uses a penalty function to constrain it not to be lower than a safety lower limit; The judgment unit dynamically adjusts the search path and step size parameters of the gray wolf individuals to guide them away from the infeasible solution area when the predicted exhaust temperature corresponding to the candidate solution is lower than the safety lower limit; The execution unit takes the final optimization result as a control instruction and sends it to the corresponding actuators, including the flue gas damper, water supply regulating valve and variable frequency pump, through the DCS system to implement the control instruction.

10. The Grey Wolf algorithm flue gas waste heat control system with exhaust temperature prediction constraint introduced according to claim 9 is characterized in that: The input parameters of the exhaust gas temperature prediction model in the prediction model building unit include: flue gas flow rate, economizer outlet water temperature, low calorific value of coal type, fuel sulfur content and secondary air and primary air distribution ratio.

Images