Grey wolf optimization algorithm unit operation optimization method and system with adaptive memory mechanism

Abstract

The invention discloses a grey wolf optimization algorithm unit operation optimization method and system with a self-adaptive memory mechanism, and is applied to energy-saving optimization and operation control of a coal-fired power generation unit. The method comprises the following steps: firstly, constructing a thermodynamic simulation model of a unit, and setting a multi-objective optimization function; global optimization is carried out on the target function through a grey wolf optimization algorithm, an adaptive memory mechanism is introduced, a historical working condition memory bank is established, and an optimal solution is extracted from the memory bank in combination with similarity matching to guide population search. During each iteration, the memory bank content is dynamically updated according to the working condition change, the target weight can be adjusted according to the real-time working condition in the optimization process, and the stability and adaptability of the optimization result are improved. And an optimization result is issued to an execution mechanism through the DCS system, so that closed-loop control is realized, and the overall economy and the energy utilization rate of the unit are improved. The method can effectively solve the problems that a traditional optimization method is poor in adaptability, poor in target coordination and the like under complex working conditions, and has high engineering application value.

Description

Technical Field

[0001] The present invention relates to the technical field of energy-saving optimization and operation control of coal-fired power generation units, and in particular to a method and system for optimizing the operation of a coal-fired power generation unit using a Grey Wolf optimization algorithm with an adaptive memory mechanism. Background Art

[0002] With the continuous increase in global energy demand and increasingly stringent environmental protection policies, energy conservation and efficient operation of coal-fired power generation units have become an important way to improve energy utilization efficiency and reduce carbon emissions. However, traditional coal-fired unit operation optimization methods usually rely on empirical rules and preset control strategies. These methods have many shortcomings in practical applications. First, traditional methods often rely on empirical adjustments or adjustments based on set control curves. The optimization process often only focuses on a specific parameter or local performance and lacks systematic global optimization. Due to the complex coupling relationship between the various subsystems of the unit, traditional methods often find it difficult to fully consider the overall efficiency of the unit, resulting in limited optimization effects and failure to fully improve the energy utilization and operating efficiency of the unit.

[0003] Secondly, existing methods for multi-objective optimization problems typically use manually set target weights to balance the relationships between various indicators. While this approach can solve some basic optimization tasks, because the target weights are often static and cannot be adaptively adjusted based on real-time operating conditions, they often struggle to effectively respond to complex operating conditions during unit operation. This often results in traditional methods remaining stuck at local optimal solutions during the optimization process, lacking globality and flexibility, and failing to achieve long-term, stable optimization of the unit under varying operating conditions.

[0004] Furthermore, traditional optimization methods have limited adaptability to external disturbances and operating fluctuations. In actual operation, coal-fired power generation units often face complex factors such as fluctuating coal quality, varying loads, and external environmental influences, all of which directly impact their operating status and performance. If optimization methods fail to adjust control strategies in real time, this can often lead to problems such as main steam pressure fluctuations and inadequate waste heat utilization, impacting the overall economic efficiency and operational stability of the unit.

[0005] Furthermore, existing optimization methods often fail to fully utilize historical operating data, lacking an effective mechanism to store and utilize optimal solutions from historical operating conditions. Traditional optimization methods typically start from scratch, ignoring the accumulation of historical experience and data. This results in a waste of computing resources and time. This is especially true when encountering similar operating conditions, as traditional optimization algorithms are unable to extract useful information from historical records to guide the current optimization process.

[0006] Therefore, how to combine modern intelligent optimization algorithms and effective data memory mechanisms to solve the limitations of traditional methods, especially how to adaptively adjust optimization goals and strategies in dynamic environments, has become an urgent problem to be solved in the optimization control technology of coal-fired power generation units. Summary of the Invention

[0007] The purpose of this invention is to provide a unit operation optimization method and system using the Gray Wolf optimization algorithm with an adaptive memory mechanism. This method addresses the problems of traditional optimization methods, such as reliance on empirical adjustments, poor target coordination, weak adaptability, and insufficient use of historical data. By introducing an adaptive memory mechanism and combining it with the Gray Wolf optimization algorithm for multi-objective optimization control, this method can achieve coordinated adjustment of optimization objectives while ensuring efficient unit operation, thereby improving the overall economic performance and energy efficiency of coal-fired power generation units.

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

[0009] The unit operation optimization method of the Grey Wolf optimization algorithm with an adaptive memory mechanism includes the following steps:

[0010] Step S1: constructing a thermodynamic simulation model of the flue gas waste heat system of a coal-fired generator unit, including an economizer, a heater, a flue gas bypass channel, and related heat exchange equipment;

[0011] Step S2: setting the objective function of the optimized thermodynamic simulation model, including the operating performance indicators of power supply coal consumption, plant power consumption rate, exhaust gas temperature and waste heat utilization rate, and determining the optimized control variables;

[0012] Step S3: Call the Gray Wolf Optimization Algorithm as the main optimizer to perform multi-parameter global optimization on the objective function;

[0013] Step S4: During the execution of the Grey Wolf Optimization Algorithm, an adaptive memory mechanism is introduced to establish a memory library containing historical optimal solutions, operating status labels, and fitness scores; the adaptive memory mechanism is used to record the optimal solutions under different operating conditions of the unit and to guide the search process of the Grey Wolf Optimization Algorithm;

[0014] Step S5: In each iteration, based on the similarity between the current working condition characteristics and the historical working condition, the optimal solution individual with the highest matching degree is extracted from the memory bank and used as the guide wolf in the initial population, or used to replace the low fitness individual;

[0015] Step S6: Adaptively update the memory library content according to the change of the unit status, including adding new operating conditions, replacing failed solutions and adjusting memory priorities;

[0016] Step S7: Based on the optimal solution obtained by the convergence of the Grey Wolf Optimization Algorithm, the optimal operating parameters under the current working conditions are output and executed through the DCS control system to achieve closed-loop operation optimization control.

[0017] A further improvement of the present invention is that the adaptive memory mechanism includes the following submodules: a working condition label extraction module, a similarity matching module, a memory library management module, and a memory guidance control module; wherein:

[0018] The operating condition label extraction module is used to extract the key feature values ​​of the current operating condition of the unit and form an operating condition label vector;

[0019] The similarity matching module is used to calculate the similarity between the current working condition label and the historical working condition labels in the memory library;

[0020] The memory management module is used to maintain the memory in the Grey Wolf optimization algorithm and perform addition, update and invalid replacement of optimal solutions;

[0021] The memory-guided control module guides the gray wolf optimization algorithm to generate the initial population or replace low-fitness individuals based on the extracted historical optimal solution.

[0022] A further improvement of the present invention is that the memory bank is a data storage structure in an adaptive memory mechanism, comprising a plurality of historical operation data entries, each memory entry comprising the following fields:

[0023] Working condition label: a feature vector representing the historical operating status;

[0024] Optimal control variable vector: the optimal solution obtained by the Grey Wolf optimization algorithm under the corresponding working conditions;

[0025] Fitness function value: used to measure the quality of the control variable solution;

[0026] Timestamp: records the creation time of the memory entry;

[0027] Memory weight: reflects the reference value and frequency of use of the entry.

[0028] A further improvement of the present invention is that the similarity matching module uses weighted Euclidean distance to calculate the similarity between the current working condition label and each historical working condition label in the memory library. The calculation formula is as follows:

[0029]

[0030] Among them, S j is the similarity between the current working condition and the jth historical working condition, x i is the i-th eigenvalue in the current working condition label vector, The i-th eigenvalue corresponding to the j-th historical record in the memory bank, wi is the weight coefficient of the i-th working condition dimension, and n is the dimension of the working condition label vector.

[0031] A further improvement of the present invention is that, during the initial population generation process of the gray wolf optimization algorithm, the memory guidance control module uses the optimal solution whose current working condition label is closest to the historical label in the memory library as the guide wolf, and randomly perturbs some population individuals with the guide wolf as the center, thereby accelerating the convergence speed of the search process.

[0032] A further improvement of the present invention is that the memory management module adopts a limited capacity management strategy. When the number of memory entries exceeds the set capacity upper limit, it is eliminated and updated according to the fitness score and memory time decay function, and memory entries with higher fitness and newer time are retained first.

[0033] A further improvement of the present invention is that the memory-guided control module is applicable to the following three operating scenarios in the Grey Wolf Optimization Algorithm:

[0034] (1) The current working condition is a historical repeated working condition, and the memory solution is directly reused;

[0035] (2) The current operating conditions are similar to the historical ones, but need to be corrected and adjusted, and local perturbation optimization is adopted;

[0036] (3) The current working condition is a novel scene, the memory bank is empty or the matching degree is low, and the global search mode is triggered.

[0037] The unit operation optimization system based on the Gray Wolf optimization algorithm with an adaptive memory mechanism includes:

[0038] Model building unit, which builds a thermodynamic simulation model of the flue gas waste heat system of a coal-fired generator unit, including economizers, air heaters, flue gas bypass channels, and related heat exchange equipment;

[0039] The objective function setting unit sets the objective function of the optimized thermodynamic simulation model, including the operating performance indicators of power supply coal consumption, plant power consumption rate, exhaust gas temperature and waste heat utilization rate, and determines the optimized control variables;

[0040] The multi-parameter global optimization unit calls the Gray Wolf Optimization Algorithm as the main optimizer to perform multi-parameter global optimization on the objective function;

[0041] A memory library establishment unit, which introduces an adaptive memory mechanism during the execution of the Grey Wolf optimization algorithm to establish a memory library containing historical optimal solutions, operating status labels, and fitness scores; the adaptive memory mechanism is used to record the optimal solutions under different operating conditions of the unit and to guide the search process of the Grey Wolf optimization algorithm;

[0042] Iteration unit, in each iteration, extracts the optimal solution individual with the highest matching degree from the memory bank based on the similarity between the current working condition characteristics and the historical working condition, and uses it as the guide wolf in the initial population, or to replace the low fitness individual;

[0043] The memory update unit adaptively updates the memory contents according to the changes in the unit status, including adding new operating conditions, replacing failed solutions, and adjusting memory priorities;

[0044] The execution unit outputs the optimal operating parameters under the current working conditions based on the optimal solution obtained by the convergence of the Grey Wolf optimization algorithm, and sends them for execution through the DCS control system to achieve closed-loop operation optimization control.

[0045] A further improvement of the present invention is that the adaptive memory mechanism in the memory library establishment unit includes the following submodules: a working condition label extraction module, a similarity matching module, a memory library management module and a memory guidance control module; wherein:

[0046] The operating condition label extraction module is used to extract the key feature values ​​of the current operating condition of the unit and form an operating condition label vector;

[0047] The similarity matching module is used to calculate the similarity between the current working condition label and the historical working condition labels in the memory library;

[0048] The memory management module is used to maintain the memory in the Grey Wolf optimization algorithm and perform addition, update and invalid replacement of optimal solutions;

[0049] The memory-guided control module guides the gray wolf optimization algorithm to generate the initial population or replace low-fitness individuals based on the extracted historical optimal solution.

[0050] A further improvement of the present invention is that the memory bank in the memory bank establishment unit is a data storage structure in an adaptive memory mechanism, comprising a plurality of historical operation data entries, each memory entry comprising the following fields:

[0051] Working condition label: a feature vector representing the historical operating status;

[0052] Optimal control variable vector: the optimal solution obtained by the Grey Wolf optimization algorithm under the corresponding working conditions;

[0053] Fitness function value: used to measure the quality of the control variable solution;

[0054] Timestamp: records the creation time of the memory entry;

[0055] Memory weight: reflects the reference value and frequency of use of the entry.

[0056] Compared with the prior art, the present invention has at least the following beneficial technical effects:

[0057] This invention, by integrating an adaptive memory mechanism with the Gray Wolf optimization algorithm, addresses the challenges of traditional optimization methods, such as reliance on empirical adjustments, poor target coordination, poor adaptability, and insufficient utilization of historical data. By constructing a thermodynamic simulation model and integrating it with the actual operating conditions of the unit, the optimization objective function not only achieves coordinated optimization of multiple objectives, including power supply coal consumption, plant power utilization, exhaust temperature, and waste heat utilization, but also effectively utilizes historical operating data during the optimization process through the adaptive memory mechanism, significantly reducing computing resource consumption and improving optimization efficiency. During the Gray Wolf optimization process, the wolf group is guided to dynamically adjust target weights, enabling the optimization results to adapt to changes in the unit's real-time operating conditions, enhancing the flexibility and stability of the optimization process. Ultimately, the optimized control parameters are distributed and implemented through the DCS control system, achieving closed-loop optimization control of the unit, effectively improving its operating efficiency, economy, and energy utilization, and demonstrating greater adaptability and robustness under complex operating conditions. This method not only has promising engineering applications but also provides effective support for the efficient and energy-saving operation of coal-fired power generation units. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] 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.

[0059] Figure 1 This is a schematic diagram of the thermal flow of the ultra-supercritical unit system, showing the main equipment such as the boiler, economizer, air preheater, steam turbine, heater and their thermal connection relationship, which is the system foundation of the optimization object of this invention;

[0060] Figure 2 This is a flow chart of the gray wolf optimization algorithm of the present invention, showing the complete optimization process from population initialization, fitness calculation, individual position update, introduction of adaptive memory mechanism to output of optimal solution;

[0061] Figure 3 This is a structural block diagram of the adaptive memory mechanism of the present invention, which includes a working condition label extraction module, a similarity matching module, a memory guidance control module, and a history memory management module, which are used to guide the gray wolf optimization algorithm to converge quickly;

[0062] Figure 4 This is a logic diagram of the weighted Euclidean distance for similarity calculation in the present invention, showing the structural path and weighting strategy for distance calculation between working condition label vectors;

[0063] Figure 5This is a schematic diagram of the spatial behavior of the gray wolf's position update process in the present invention, illustrating the search and update strategy of the individual under the guidance of the three guiding solutions α, β, and δ;

[0064] Figure 6 This is a structural block diagram of the unit operation optimization system using the Grey Wolf optimization algorithm with an adaptive memory mechanism of the present invention. DETAILED DESCRIPTION

[0065] 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.

[0066] In the description of the present invention, when used in this specification and the appended claims, the terms "include" and "comprising" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or collections thereof.

[0067] 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.

[0068] 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.

[0069] 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.

[0070] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0071] Example 1

[0072] The present invention provides a unit operation optimization method using a Gray Wolf optimization algorithm with an adaptive memory mechanism, which is applied to the operation optimization of the boiler electrical system of a coal-fired power generation unit, comprising the following steps:

[0073] Step S1: constructing a thermodynamic simulation model of the flue gas waste heat system of the unit, including an economizer, a heater, a flue gas bypass channel and related heat exchange equipment;

[0074] Step S2: setting the optimization objective function, including the operation performance indicators such as power supply coal consumption, plant power consumption rate, exhaust gas temperature and waste heat utilization rate, and determining the optimization control variables;

[0075] Step S3: Call the Gray Wolf Optimization Algorithm as the main optimizer to perform multi-parameter global optimization on the objective function;

[0076] Step S4: During the execution of the Grey Wolf Optimization Algorithm, a historical operating condition memory mechanism is introduced to establish an adaptive memory library containing historical optimal solutions, operating status labels, and fitness scores; the adaptive memory mechanism is used to record the optimal solutions under different operating conditions of the unit and to guide the search process of the Grey Wolf Optimization Algorithm;

[0077] Step S5: In each iteration, based on the similarity between the current working condition characteristics and the historical working condition, the optimal solution individual with the highest matching degree is extracted from the memory bank and used as the initial guide of the wolf pack or to replace the low fitness individual;

[0078] Step S6: Adaptively update the memory library content according to the change of the unit status, including adding new operating conditions, replacing failed solutions, and adjusting memory priorities;

[0079] Step S7: Based on the optimal solution obtained by the convergence of the Grey Wolf Optimization Algorithm, the optimal operating parameters under the current working conditions are output and executed through the DCS control system to achieve closed-loop operation optimization control.

[0080] In this embodiment, the adaptive memory mechanism includes the following submodules: a working condition label extraction module, a similarity matching module, a history library management module, and a memory guidance control module; wherein:

[0081] The operating condition label extraction module is used to extract the key feature values ​​of the current operating condition of the unit and form an operating condition label vector;

[0082] The similarity matching module calculates the similarity between the current working condition and the historical working condition and selects the historical optimal solution with the highest matching degree;

[0083] The history library management module is used to maintain the memory library in the Grey Wolf optimization algorithm and perform the addition, update and invalid replacement of the optimal solution;

[0084] The memory-guided control module guides the gray wolf optimization algorithm to generate the initial population or replace low-fitness individuals based on the extracted historical optimal solution.

[0085] In this embodiment, each memory entry in the historical memory library includes the following fields: operating condition label, optimal control variable vector, fitness function value, timestamp, memory weight, and the memory library can be dynamically updated according to changes in operating conditions.

[0086] In this embodiment, the similarity matching module uses weighted Euclidean distance to calculate the similarity between the current working condition and each historical working condition in the memory library. The calculation formula is as follows:

[0087]

[0088] Among them, S j is the similarity between the current working condition and the jth historical working condition, x i is the i-th eigenvalue in the current working condition label vector, The i-th eigenvalue corresponding to the j-th historical record in the memory bank, w i is the weight coefficient of the i-th working condition dimension, and n is the dimension of the working condition label vector.

[0089] In this embodiment, during the initial population generation process of the gray wolf optimization algorithm, the memory guidance control module uses the optimal solution whose current working condition label is closest to the historical label in the memory library as the guide wolf, and randomly perturbs some population individuals with the guide wolf as the center, thereby accelerating the convergence speed of the search process.

[0090] In this embodiment, the memory management module adopts a limited capacity management strategy. When the number of memory entries exceeds the set capacity upper limit, it is eliminated and updated according to the fitness score and memory time decay function, and memory entries with higher fitness and newer time are retained first.

[0091] In this embodiment, the memory guidance mechanism is applicable to the following three scenarios:

[0092] (1) The current working condition is a historical repeated working condition, and the memory solution is directly reused;

[0093] (2) The current operating conditions are similar to the historical ones, but need to be corrected and adjusted, and local perturbation optimization is adopted;

[0094] (3) The current working condition is a novel scene, the memory bank is empty or the matching degree is low, and the global search mode is triggered.

[0095] In this embodiment, the unit operation optimization method constitutes a periodically operating optimization system, which calls the Gray Wolf optimization algorithm and adaptive memory mechanism every 5 minutes to complete the optimization process, and performs batch updates and compression operations on the memory library when the daily load changes drastically to enhance the dynamic adaptability and stability of the system.

[0096] Example 2

[0097] (1) System structure and optimization goals

[0098] This embodiment takes a 1000MW ultra-supercritical coal-fired power generation unit as the object. The unit typically includes a boiler, a steam turbine, a generator, a flue gas waste heat system (economizer, air preheater, heater, etc.) and an auxiliary system (forced draft fan, induced draft fan, feed water pump, variable frequency pump, etc.). The established thermodynamic simulation model covers the boiler, steam turbine and the main heat exchange equipment of the unit, and is used to predict operating indicators such as power supply coal consumption, exhaust temperature, and plant power rate under different working conditions. The optimization goal is to minimize the unit's power supply coal consumption and plant power rate while meeting safety and environmental protection constraints (such as exhaust temperature, unit equilibrium state, etc.), improve thermal efficiency, and achieve economic operation and energy conservation and emission reduction. Figure 1 As shown, it is a schematic diagram of the system thermal flow of a 1000MW ultra-supercritical coal-fired power generation unit, showing the main equipment such as boiler, economizer, air preheater, steam turbine, heater and their thermal connection relationship, which constitutes the basic structure of the thermodynamic simulation modeling of the present invention.

[0099] (2) Simulation model and optimization variables

[0100] A thermodynamic simulation model is established for a 1000MW ultra-supercritical coal-fired unit. The model includes the thermodynamic parameter relationships of each important link of the boiler, steam turbine and generator, and is used to simulate performance output indicators such as coal consumption, exhaust temperature, and plant power consumption rate under different operating conditions. The simulation model is calibrated through the heat balance equation and empirical coefficients combined with the unit test data to achieve accurate prediction of the operating status indicators. In this embodiment, the air preheater bypass flue gas flow damper opening x1, high-pressure economizer feed water flow x2, low-pressure economizer condensate flow x3 and heater heat medium water flow of the unit are used as optimization variables x4, and the variables are adjusted by the air preheater bypass flue gas damper, flow control valve, variable frequency pump I speed and variable frequency pump II speed respectively. Let the optimization variables form a vector:

[0101] x={x1,x2,x3,x4}

[0102] After the simulation model inputs f(x), the corresponding coal consumption Y can be calculated and output coal (x), exhaust gas temperature T ex (x) and plant power consumption rate k aux (x) and other indicators are used to evaluate optimization results. The simulation model was verified using data from both classic load and variable operating conditions, with the error of each key output parameter exceeding ±0.5%, meeting the accuracy requirements for operational optimization. This simulation model can be used to generate the performance data required for the iterative optimization process and to evaluate the quality of the optimization results.

[0103] (3) Grey Wolf Optimization Algorithm

[0104] like Figure 2 The figure shows a flow chart of the gray wolf optimization algorithm adopted by the present invention, including key steps such as population initialization, fitness calculation, individual position update, introduction of adaptive memory mechanism, and output of optimal solution, which reflects the optimization process structure of the present method. This embodiment adopts the gray wolf optimization algorithm (GWO) as a global searcher to achieve multi-parameter global optimization. The gray wolf population size is set to about 40 (the interval of 30 to 50 can be taken), and the maximum number of iterations is about 150 times (the interval of 100 to 200 can be taken) to ensure the adequacy of the search and the efficiency of convergence. In the algorithm, the control parameter a determines the search strategy. In the traditional GWO, the value of a decreases linearly from 2 to 0; in order to make the algorithm more exploratory in the early stage, the control parameter a of this embodiment adopts a nonlinear decreasing method, and the formula is as follows:

[0105]

[0106] Where t is the current iteration number, T is the maximum number of iterations, and a0 is the initial value (usually 2). This allows the value of a to decrease nonlinearly along the iteration process, thereby better balancing global search and local development.

[0107] In each iteration, the three gray wolves with the best fitness are denoted as α, β, and δ, which represent the best, second-best, and third-best solutions found so far. The other individuals jointly update their positions based on the positions of α, β, and δ. Specifically, the distance D between each wolf and α, β, and δ is defined as α 、D β 、D δ , and calculate the corresponding position update components X1, X2, X3, and then take the average value to update the next generation position. Figure 5 As shown in the figure, the spatial behavior diagram of the gray wolf's position update process is shown, which illustrates the position adjustment path of the individual under the joint action of the three guiding solutions α, β, and δ. The update formula is as follows:

[0108] D α =|C1·X α -X|,X1=X α -A1·D α

[0109] D β ·|C2·X β -X|,X2=X β -A2·D β

[0110] D δ =|C3·X δ -X|,X3=X δ -A3·D δ

[0111]

[0112] Among them, X α , X β , X δ are the positions of the three individuals with the best fitness in the current population; X is the position of the current individual; A i =2a·r i -a, C i =2·r i ', r i ,r i '∈[0,1] is a random number. This introduces random perturbations into the search process, preventing it from falling into local optima. When |A|>1, the wolf approaches the prey, and the algorithm focuses on local exploration. When |A|<1, the wolf moves away from the prey, and the algorithm focuses on global exploration.

[0113] The present invention constructs a multi-objective optimization function to balance unit economics and constraints. A common approach is to normalize each objective (such as power supply coal consumption, exhaust temperature, and plant power consumption rate), assign weights, and then sum them to form a comprehensive objective function. For example, each objective value can be normalized by subtracting the expected / reference value and dividing it by the range. Weights can then be assigned based on desired preferences to achieve balanced optimization of multiple objectives. For example:

[0114]

[0115] Among them, f1, f2, f3 are coal consumption, plant power consumption rate, and exhaust gas temperature respectively; w1, w2, w3 are weight coefficients of each target (satisfying w1+w2+w3=1); f i min ,f i max are the normalized interval endpoints of each objective function.

[0116] (4) Adaptive memory mechanism

[0117] like Figure 3 Figure 2 shows the block diagram of the adaptive memory mechanism of the present invention, which includes a working condition label extraction module, a similarity matching module, a memory guidance and control module, and a memory library management module. These modules work together to guide the optimization process to rapid convergence. The introduction of the adaptive memory mechanism can use historical optimization results to guide the search, thereby accelerating convergence and improving stability. Specifically, it includes the following elements:

[0118] ① Operating condition label extraction: by analyzing the current operating status of the unit (such as load, supply temperature and other key features), a label vector representing the operating condition characteristics is generated.

[0119] ②Similarity calculation: calculate the weighted Euclidean distance between the current working condition label and each historical working condition label in the memory library to obtain the similarity index. Figure 4 The figure shows the weighted Euclidean distance logic diagram of the similarity calculation of the present invention, which shows the weighted distance calculation path and weight distribution structure between the current working condition label vector and the historical label in the memory library. The weighted Euclidean distance formula is:

[0120]

[0121] where x i is the eigenvalue of the current working condition, is the characteristic value corresponding to the historical working condition, w i is the feature weight; n is the number of dimensions of the working condition label. The smaller the similarity, the closer the working conditions are.

[0122] ③ Memory entry structure: Each entry in the memory database records a historical operating condition, including the operating condition label, the corresponding optimal control variable vector, the solution's fitness score, the recording timestamp, and the memory weight. The memory weight can be set based on the solution's quality and frequency of use.

[0123] ④ Memory update and elimination: adopt a limited capacity management strategy and set the maximum memory capacity L (e.g. 100 entries). When the number of memories exceeds L, based on the fitness score F and the time decay function comprehensive evaluation, eliminate the entries with the most serious decay or the lowest fitness to ensure that the information in the memory is relevant and representative of the current operating environment. The time decay function calculation formula is:

[0124] ρ(t)=e -λt

[0125] Where λ is the time decay factor (e.g., 0.01–0.05), and t is the time interval between the memory entry and the current moment. Memory elimination can be sorted based on the value of F·ρ(t), prioritizing entries with high fitness and "freshness."

[0126] ⑤ The current operating condition can be divided into three typical cases: when there is a perfect match, the corresponding historical optimal solution is directly called; when there are similar but not identical operating conditions, the historical optimal control variable is fine-tuned as an initial reference; for new scenarios, random initialization is used and gradual exploration is carried out. In each iteration, based on the similarity between the current operating condition and the memory library, the most matching individual from the historical optimal solution is selected as the "guide wolf". This individual is used as the center to generate part of the initial population or replace individuals with poor fitness to accelerate search convergence.

[0127] Example 3

[0128] like Figure 6 As shown, the present invention provides a unit operation optimization system using a Grey Wolf optimization algorithm with an adaptive memory mechanism, comprising:

[0129] Model building unit, which builds a thermodynamic simulation model of the flue gas waste heat system of a coal-fired generator unit, including economizers, air heaters, flue gas bypass channels, and related heat exchange equipment;

[0130] The objective function setting unit sets the objective function of the optimized thermodynamic simulation model, including the operating performance indicators of power supply coal consumption, plant power consumption rate, exhaust gas temperature and waste heat utilization rate, and determines the optimized control variables;

[0131] The multi-parameter global optimization unit calls the Gray Wolf Optimization Algorithm as the main optimizer to perform multi-parameter global optimization on the objective function;

[0132] A memory library establishment unit, which introduces an adaptive memory mechanism during the execution of the Grey Wolf optimization algorithm to establish a memory library containing historical optimal solutions, operating status labels, and fitness scores; the adaptive memory mechanism is used to record the optimal solutions under different operating conditions of the unit and to guide the search process of the Grey Wolf optimization algorithm;

[0133] Iteration unit, in each iteration, extracts the optimal solution individual with the highest matching degree from the memory bank based on the similarity between the current working condition characteristics and the historical working condition, and uses it as the guide wolf in the initial population, or to replace the low fitness individual;

[0134] The memory update unit adaptively updates the memory contents according to the changes in the unit status, including adding new operating conditions, replacing failed solutions, and adjusting memory priorities;

[0135] The execution unit outputs the optimal operating parameters under the current working conditions based on the optimal solution obtained by the convergence of the Grey Wolf optimization algorithm, and sends them for execution through the DCS control system to achieve closed-loop operation optimization control.

[0136] In this embodiment, the adaptive memory mechanism in the memory library establishment unit includes the following submodules: a working condition label extraction module, a similarity matching module, a memory library management module, and a memory guidance control module; wherein:

[0137] The operating condition label extraction module is used to extract the key feature values ​​of the current operating condition of the unit and form an operating condition label vector;

[0138] The similarity matching module is used to calculate the similarity between the current working condition label and the historical working condition labels in the memory library;

[0139] The memory management module is used to maintain the memory in the Grey Wolf optimization algorithm and perform addition, update and invalid replacement of optimal solutions;

[0140] The memory-guided control module guides the gray wolf optimization algorithm to generate the initial population or replace low-fitness individuals based on the extracted historical optimal solution.

[0141] In this embodiment, the memory in the memory building unit is a data storage structure in an adaptive memory mechanism, which includes multiple historical operation data entries, and each memory entry includes the following fields:

[0142] Working condition label: a feature vector representing the historical operating status;

[0143] Optimal control variable vector: the optimal solution obtained by the Grey Wolf optimization algorithm under the corresponding working conditions;

[0144] Fitness function value: used to measure the quality of the control variable solution;

[0145] Timestamp: records the creation time of the memory entry;

[0146] Memory weight: reflects the reference value and frequency of use of the entry.

[0147] 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.

[0148] 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 unit operation optimization method of the Grey Wolf optimization algorithm with an adaptive memory mechanism is characterized by: The following steps are involved: Step S1: constructing a thermodynamic simulation model of the flue gas waste heat system of a coal-fired generator unit, including an economizer, a heater, a flue gas bypass channel, and related heat exchange equipment; Step S2: setting the objective function of the optimized thermodynamic simulation model, including the operating performance indicators of power supply coal consumption, plant power consumption rate, exhaust gas temperature and waste heat utilization rate, and determining the optimized control variables; Step S3: Call the Gray Wolf Optimization Algorithm as the main optimizer to perform multi-parameter global optimization on the objective function; Step S4: During the execution of the Grey Wolf Optimization Algorithm, an adaptive memory mechanism is introduced to establish a memory library containing historical optimal solutions, operating status labels, and fitness scores; the adaptive memory mechanism is used to record the optimal solutions under different operating conditions of the unit and to guide the search process of the Grey Wolf Optimization Algorithm; Step S5: In each iteration, based on the similarity between the current working condition characteristics and the historical working condition, the optimal solution individual with the highest matching degree is extracted from the memory bank and used as the guide wolf in the initial population, or used to replace the low fitness individual; Step S6: Adaptively update the memory library content according to the change of the unit status, including adding new operating conditions, replacing failed solutions and adjusting memory priorities; Step S7: Based on the optimal solution obtained by the convergence of the Grey Wolf Optimization Algorithm, the optimal operating parameters under the current working conditions are output and executed through the DCS control system to achieve closed-loop operation optimization control.

2. The unit operation optimization method using the Grey Wolf optimization algorithm with an adaptive memory mechanism according to claim 1 is characterized in that: The adaptive memory mechanism includes the following submodules: a working condition label extraction module, a similarity matching module, a memory library management module, and a memory guidance and control module; wherein: The operating condition label extraction module is used to extract the key feature values ​​of the current operating condition of the unit and form an operating condition label vector; The similarity matching module is used to calculate the similarity between the current working condition label and the historical working condition labels in the memory library; The memory management module is used to maintain the memory in the Grey Wolf optimization algorithm and perform addition, update and invalid replacement of optimal solutions; The memory-guided control module guides the gray wolf optimization algorithm to generate the initial population or replace low-fitness individuals based on the extracted historical optimal solution.

3. The unit operation optimization method using the Grey Wolf optimization algorithm with an adaptive memory mechanism according to claim 2 is characterized in that: The memory library is a data storage structure in the adaptive memory mechanism, which contains multiple historical operation data entries. Each memory entry contains the following fields: Working condition label: a feature vector representing the historical operating status; Optimal control variable vector: the optimal solution obtained by the Grey Wolf optimization algorithm under the corresponding working conditions; Fitness function value: used to measure the quality of the control variable solution; Timestamp: records the creation time of the memory entry; Memory weight: reflects the reference value and frequency of use of the entry.

4. The unit operation optimization method using the Grey Wolf optimization algorithm with an adaptive memory mechanism according to claim 2 is characterized in that: The similarity matching module uses weighted Euclidean distance to calculate the similarity between the current working condition label and each historical working condition label in the memory library. The calculation formula is as follows: Among them, S j is the similarity between the current working condition and the jth historical working condition, x i is the i-th eigenvalue in the current working condition label vector, The i-th eigenvalue corresponding to the j-th historical record in the memory bank, w i is the weight coefficient of the i-th working condition dimension, and n is the dimension of the working condition label vector.

5. The unit operation optimization method using the Grey Wolf optimization algorithm with an adaptive memory mechanism according to claim 2 is characterized in that: During the initial population generation process of the gray wolf optimization algorithm, the memory guidance control module uses the optimal solution whose current working condition label is closest to the historical label in the memory library as the guide wolf, and randomly perturbs some population individuals around the guide wolf, thereby accelerating the convergence speed of the search process.

6. The method for optimizing unit operation using the Grey Wolf optimization algorithm with an adaptive memory mechanism according to claim 2 is characterized in that: The memory management module adopts a limited capacity management strategy. When the number of memory entries exceeds the set capacity upper limit, it is eliminated and updated according to the fitness score and memory time decay function, and memory entries with higher fitness and newer time are retained first.

7. The method for optimizing unit operation using the Grey Wolf optimization algorithm with an adaptive memory mechanism according to claim 2 is characterized in that: The memory-guided control module is applicable to the following three operating scenarios in the Grey Wolf Optimization Algorithm: (1) The current working condition is a historical repeated working condition, and the memory solution is directly reused; (2) The current operating conditions are similar to the historical ones, but need to be corrected and adjusted, and local perturbation optimization is adopted; (3) The current working condition is a novel scene, the memory bank is empty or the matching degree is low, and the global search mode is triggered.

8. The unit operation optimization system of the Grey Wolf optimization algorithm with adaptive memory mechanism is characterized by: include: Model building unit, which builds a thermodynamic simulation model of the flue gas waste heat system of a coal-fired generator unit, including economizers, air heaters, flue gas bypass channels, and related heat exchange equipment; The objective function setting unit sets the objective function of the optimized thermodynamic simulation model, including the operating performance indicators of power supply coal consumption, plant power consumption rate, exhaust gas temperature and waste heat utilization rate, and determines the optimized control variables; The multi-parameter global optimization unit calls the Gray Wolf Optimization Algorithm as the main optimizer to perform multi-parameter global optimization on the objective function; A memory library establishment unit, which introduces an adaptive memory mechanism during the execution of the Grey Wolf optimization algorithm to establish a memory library containing historical optimal solutions, operating status labels, and fitness scores; the adaptive memory mechanism is used to record the optimal solutions under different operating conditions of the unit and to guide the search process of the Grey Wolf optimization algorithm; Iteration unit, in each iteration, extracts the optimal solution individual with the highest matching degree from the memory bank based on the similarity between the current working condition characteristics and the historical working condition, and uses it as the guide wolf in the initial population, or to replace the low fitness individual; The memory update unit adaptively updates the memory contents according to the changes in the unit status, including adding new operating conditions, replacing failed solutions, and adjusting memory priorities; The execution unit outputs the optimal operating parameters under the current working conditions based on the optimal solution obtained by the convergence of the Grey Wolf optimization algorithm, and sends them for execution through the DCS control system to achieve closed-loop operation optimization control.

9. The unit operation optimization system using the Grey Wolf optimization algorithm with an adaptive memory mechanism according to claim 8 is characterized in that: The adaptive memory mechanism in the memory library establishment unit includes the following submodules: a working condition label extraction module, a similarity matching module, a memory library management module, and a memory guidance and control module; wherein: The operating condition label extraction module is used to extract the key feature values ​​of the current operating condition of the unit and form an operating condition label vector; The similarity matching module is used to calculate the similarity between the current working condition label and the historical working condition labels in the memory library; The memory management module is used to maintain the memory in the Grey Wolf optimization algorithm and perform addition, update and invalid replacement of optimal solutions; The memory-guided control module guides the gray wolf optimization algorithm to generate the initial population or replace low-fitness individuals based on the extracted historical optimal solution.

10. The unit operation optimization system using the Grey Wolf optimization algorithm with an adaptive memory mechanism according to claim 9 is characterized in that: The memory in the memory building unit is a data storage structure in the adaptive memory mechanism, which contains multiple historical operation data entries. Each memory entry contains the following fields: Working condition label: a feature vector representing the historical operating status; Optimal control variable vector: the optimal solution obtained by the Grey Wolf optimization algorithm under the corresponding working conditions; Fitness function value: used to measure the quality of the control variable solution; Timestamp: records the creation time of the memory entry; Memory weight: reflects the reference value and frequency of use of the entry.

Images