A control parameter collaborative optimization method based on hybrid heuristic optimization

Abstract

The application discloses a control parameter collaborative optimization method based on a hybrid heuristic optimization, performs collaborative initialization on Ns basic optimization algorithms through a hybrid heuristic optimization thread, and performs parallel solving on a target optimization problem based on original iteration mechanisms of each algorithm; generates characteristic space data containing an algorithm winning colony, and extracts a current optimal solution of each algorithm; constructs a cross-algorithm winning colony interaction mechanism, imports the characteristic space data into a winning colony characteristic library, and outputs a target function optimization coefficient of the algorithm; and verifies stability and analyzes robustness of all target function optimization coefficients through a multi-target fusion decision unit, and determines a global optimal solution. The application combines multiple heuristic algorithms and realizes random exchange of winning colony individuals, maximally avoids influence of single algorithm defects on a solving result, and can realize fast initialization, parallel solving and iteration of multi-algorithm collaboration.

Description

Technical Field

[0001] This invention relates to a collaborative optimization method for control parameters based on hybrid heuristic optimization, which achieves control parameter optimization through computer technology. Background Technology

[0002] The optimization problem of control parameters in MPC controllers is actually the solution of a constrained optimization problem. When both the objective function and constraints are linear functions, the optimization problem is linear programming. When neither the objective function nor the constraints are linear functions, the optimization problem is nonlinear programming. The models involved in the optimized operation of air conditioning systems are mostly nonlinear models, and the objective functions established based on these nonlinear models are also nonlinear functions. Therefore, solving for the control parameters is solving a nonlinear constrained optimization problem. Commonly used methods for solving nonlinear constrained optimization problems can be broadly classified into two categories: traditional optimization methods and intelligent optimization methods. The specific optimization algorithms and advantages and disadvantages of each method are shown in Table 1.

[0003] Table 1. Solution methods for nonlinear constrained optimization problems

[0004]

[0005] As shown in Table 1, traditional optimization methods suffer from low computational efficiency and a tendency to get trapped in local optima. Furthermore, traditional methods struggle with non-convex feasible regions, feasible regions with multiple local extrema, disconnected feasible regions, and problems where all or part of the variables are discrete or integer. In contrast, intelligent optimization methods, due to their powerful global search capabilities and ability to handle nonlinear and complex non-convex optimization problems, have been widely applied in practical production processes and have achieved good results.

[0006] Heuristic algorithms are widely used in current research and applications of intelligent optimization methods. Heuristic algorithms are a class of problem-solving methods based on experience and intuition, guiding the optimization and decision-making process through heuristic rules, strategies, and search techniques. Common heuristic algorithms include genetic algorithms, ant colony optimization, simulated annealing, and particle swarm optimization. However, each individual heuristic algorithm has inherent drawbacks. For example, genetic algorithms have poor local search capabilities and are prone to premature convergence; ant colony optimization has long search times and its results are sensitive to parameters. Patent publication number CN116300430B discloses an MPC control parameter optimization method and its application in a parallel platform. By establishing an evaluation function for MPC control parameter optimization and using a real-number encoded differential evolution algorithm to perform meta-heuristic search in the parameter space, parameter optimization is achieved, enabling the model predictive controller to achieve better control performance. However, this method is still prone to falling into local search traps. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a collaborative optimization method for control parameters based on hybrid heuristic optimization. This method combines computer technology to minimize the impact of the limitations of each algorithm on the solution results, thereby increasing the probability of finding the global optimum. The core of this method lies in the ability to randomly swap individuals from the winning populations of different algorithms for optimization.

[0008] The objective of this invention is achieved through the following technical solution:

[0009] A collaborative optimization method for control parameters based on hybrid heuristic optimization, the method comprising:

[0010] The Ns basic optimization algorithms are co-initialized by a hybrid heuristic optimization thread, and the target optimization problem is solved in parallel based on the native iterative mechanism of each algorithm.

[0011] When any algorithm satisfies the first convergence condition or the initial iteration number iter1 reaches the Maxiter1 threshold, feature space data containing the winning community of the algorithms is generated, and the current optimal solution of each algorithm is extracted.

[0012] A cross-algorithm superior community interaction mechanism is constructed, the feature space data is imported into the superior community feature library, and feature sampling and cross-population embedding operations are performed from different superior communities based on the population diversity feature coefficient.

[0013] When a single algorithm satisfies the second convergence condition or the total number of iterations iter2 reaches the Maxiter2 threshold, the optimization coefficients of the objective function of the algorithm are output.

[0014] By using a multi-objective fusion decision unit to verify the stability and robustness of the optimization coefficients of all objective functions, the global optimal solution is determined.

[0015] As a preferred approach, the hybrid heuristic optimization thread is constructed by injecting a population feature transfer mechanism and a parameter adaptive adjustment strategy into the baseline optimization thread. The baseline optimization thread includes native operators of genetic algorithm, particle swarm optimization algorithm and firefly swarm optimization algorithm.

[0016] The cross-algorithm winner community interaction mechanism includes:

[0017] The feature space data is spliced ​​using a feature vector fusion unit to generate a superior community feature vector with cross-algorithm correlation.

[0018] The population diversity feature coefficient is calculated based on the feature vector of the superior community, and the sampling ratio and feature stratification interval for cross-population embedding are determined based on the coefficient.

[0019] The method for constructing the parameter adaptive adjustment strategy includes:

[0020] Monitor the rate of change of the objective function value of each algorithm within a continuous preset iteration window;

[0021] When the rate of change is lower than the dynamic adjustment threshold, the parameter perturbation module is triggered to perform nonlinear adjustment on the core parameters of the current algorithm.

[0022] Perform an exponential amplification operation on the mutation probability of the genetic algorithm;

[0023] Perform a piecewise decay operation on the inertia weights in the particle swarm optimization algorithm;

[0024] An adaptive recalibration operation of the luciferin coefficient is performed on the firefly algorithm.

[0025] As a preferred embodiment, the baseline optimization thread includes the native operators of genetic algorithm, particle swarm optimization algorithm, and firefly swarm optimization algorithm, specifically manifested as follows:

[0026] The genetic algorithm adopts a chromosome encoding mechanism. Its operators include gene recombination units based on roulette wheel selection operators, population evolution units based on single-point crossover operators, and gene perturbation units based on basic position mutation operators.

[0027] The particle swarm optimization algorithm adopts a velocity-position update model, and its operators include an individual optimal position tracking unit and a global optimal position fusion unit.

[0028] The firefly swarm optimization algorithm employs a brightness attraction mechanism, and its operators include a relative brightness calculation unit and a dynamic distance response unit.

[0029] As a preferred embodiment, the initialization process of the benchmark optimization thread includes:

[0030] Genetic algorithm dimension: Set a gene encoding parameter space including population size, chromosome length, crossover probability threshold, and mutation probability threshold, and generate an initial population that satisfies uniform distribution within the feasible region of control parameters;

[0031] Particle swarm optimization dimension: Define the kinematic parameter space including the number of particles, learning factor matrix, and inertia weight curve, and generate a position matrix with random initial velocity within the boundary of the feasible region of control parameters;

[0032] Firefly Algorithm Dimensions: Construct a biological excitation parameter space containing the number of fireflies, attribute dimensions, and search space topology, and generate an initial position tensor that satisfies a Gaussian distribution within a predefined multidimensional feasible region.

[0033] As a preferred approach, the iterative evolution process of native operators includes:

[0034] Genetic algorithms execute a three-stage evolutionary process of gene selection, crossover, and mutation, and drive iterative population updates through fitness function evaluation units;

[0035] The particle swarm optimization algorithm performs a two-stage motion process of velocity vector update and position vector correction, and realizes information interaction through individual historical best solution memory units and global best solution sharing units;

[0036] The firefly algorithm executes a three-stage biomimetic process of brightness calculation, position attraction, and parameter update, and achieves swarm intelligence evolution through a relative distance calculation unit and an attraction intensity adjustment unit.

[0037] As a preferred approach, methods for constructing the feature database of the winning community include:

[0038] For each winning community, a multidimensional feature label is established, including algorithm identifier, iterative stage feature code, gradient of objective function value change, and population diversity index;

[0039] The multidimensional feature labels are converted into traceable feature vectors using a feature encoding mechanism to construct a feature map of the superior community with spatiotemporal correlation.

[0040] As a preferred approach, feature sampling and cross-population embedding operations specifically include:

[0041] The feature extraction amount of each winning community is determined based on the dynamic sampling ratio calculation module, and the dynamic sampling ratio is positively correlated with the population diversity index.

[0042] A hierarchical feature sampling strategy is adopted to divide the victorious community into multiple feature subspaces according to the fitness distribution, and perform probability density-based adaptive sampling in each subspace.

[0043] The extracted feature samples are injected into the target population through a random position insertion algorithm, which includes a gene sequence recombination verification unit and a population capacity balancing unit.

[0044] As a preferred embodiment, the parameter perturbation module includes:

[0045] The genetic algorithm mutation enhancement unit uses an exponential mutation probability function;

[0046] The particle swarm inertia weighting unit employs a piecewise linear decay function;

[0047] The firefly luciferin regulation unit employs an adaptive update formula.

[0048] As a preferred approach, the methods for determining the second convergence condition include:

[0049] A joint decision-making model with dual convergence criteria is constructed, and the first convergence flag is triggered when the total number of iterations iter2>Miniter2 and the change in the objective function value Δf≤ε.

[0050] The second convergence flag is triggered when the number of iterations in which the algorithm gets stuck in a local optimum exceeds a preset threshold.

[0051] The dual convergence criterion joint decision-making model uses a weighted voting mechanism to comprehensively determine the termination time of the algorithm.

[0052] As a preferred approach, stability verification and robustness analysis include:

[0053] Establish a solution space volatility evaluation index, calculate the standard deviation σ and coefficient of variation CV of each candidate optimal solution in multiple independent runs; construct a multi-dimensional robustness evaluation matrix, including parameter sensitivity analysis, boundary condition testing and noise interference experiments;

[0054] The weight coefficients of each evaluation index are determined by the analytic hierarchy process (AHP), and the final optimal solution is determined by weighted comprehensive evaluation.

[0055] As a preferred approach, the method further includes optimizing the process traceability mechanism:

[0056] Construct an algorithm evolution log database to record multidimensional evolutionary data in real time, including population gene distribution, fitness change curves, and parameter adjustment trajectories.

[0057] Design a visualization analysis interface to convert the multidimensional evolution data into a visualization analysis report that includes a three-dimensional convergence surface, a parameter sensitivity heatmap, and a population diversity radar chart.

[0058] A control parameter collaborative optimization system based on hybrid heuristic optimization includes:

[0059] The system comprises a hybrid heuristic optimization engine module, a winner community interaction module, a parameter adaptive adjustment module, a multi-objective decision-making module, and an algorithm base library module. Specifically: the hybrid heuristic optimization engine module connects to both the algorithm base library module and the winner community interaction module, transmitting both the algorithm's native iterative data and the winner community feature space data; the parameter adaptive adjustment module is bidirectionally connected to the hybrid heuristic optimization engine module, transmitting the objective function's rate of change data in real time and providing feedback on parameter adjustment commands; and the multi-objective decision-making module receives optimization coefficient data from all algorithm channels and performs stability verification and robustness analysis.

[0060] The hybrid heuristic optimization engine module includes:

[0061] Collaborative initialization unit: Connects to the algorithm base library module, initializes the parameter space of Ns basic optimization algorithms, and establishes a parallel solution channel;

[0062] Native Iterative Execution Unit: Built-in genetic algorithm operator, particle swarm optimization operator and firefly swarm optimization operator, driving each algorithm to perform iterative calculations according to the native mechanism;

[0063] Convergence monitoring unit: connected to each algorithm channel, it determines the iteration termination condition through the convergence condition discrimination module and the Maxiter counter module;

[0064] Feature extraction unit: generates feature space data containing algorithmic winning communities and extracts the current optimal solution vector for each channel; the winning community interaction module includes:

[0065] Feature vector fusion unit: Receives feature space data from various algorithm channels and performs multi-dimensional feature concatenation to generate cross-algorithm related feature vectors;

[0066] Diversity analysis unit: Determines the feature sampling ratio and stratification interval based on the population diversity characteristic coefficient calculation module;

[0067] Cross-population embedding unit: connects to the hybrid heuristic optimization engine module to perform dynamic migration and population recombination operations of the winning feature vectors;

[0068] The parameter adaptive adjustment module includes:

[0069] Iterative window monitoring unit: Statistically tracks the rate of change of the objective function value of each algorithm using a sliding time window;

[0070] Disturbance triggering unit: When the rate of change is lower than the dynamic adjustment threshold, the parameter nonlinear adjustment channel is activated;

[0071] Algorithm regulator group: includes genetic algorithm mutation probability amplifier, particle swarm inertial weight attenuator, and firefly luciferin calibrator;

[0072] The multi-objective decision-making module includes:

[0073] Stability verification unit: The robustness of each optimization coefficient is tested using the Monte Carlo simulation module;

[0074] Decision fusion unit: The fuzzy comprehensive evaluation method is used to screen the global optimal solution of the Pareto front solution set;

[0075] Results output interface: Generates an analysis report containing the optimized parameter configuration scheme and the objective function response surface;

[0076] The algorithm's basic library module stores:

[0077] The core operators of genetic algorithms include selection operators, crossover operators, mutation operators, and fitness functions.

[0078] Particle swarm optimization component set: including velocity update calculator, position updater, and neighborhood topology template;

[0079] Firefly Algorithm Function Package: Includes a luciferin update module, a movement probability calculator, and an attraction calculation unit;

[0080] The feature transfer operations performed by the cross-population embedding unit include: feature space similarity calculation based on Mahalanobis distance; population recombination strategy using quantum rotation gate mechanism; and feature hierarchical sampling method based on information entropy weight.

[0081] The nonlinear adjustment channel for the parameters includes: an exponential amplification function for the mutation probability of the genetic algorithm; a piecewise decay strategy for the particle swarm inertia weight; and an adaptive calibration model for the firefly luciferin coefficient.

[0082] An electronic device, the electronic device comprising:

[0083] At least one processor; and a memory communicatively connected to said at least one processor; wherein,

[0084] The memory stores instructions that can be executed by the at least one processor, which enable the at least one processor to perform the steps of the previously described hybrid heuristic optimization-based control parameter collaborative optimization method.

[0085] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned collaborative optimization method for control parameters based on hybrid heuristic optimization.

[0086] This invention offers at least the following advantages: The core of this invention lies in combining multiple heuristic algorithms and implementing random exchanges of individuals within the winning community, thereby minimizing the impact of single-algorithm limitations on the solution results and significantly improving the probability of finding the global optimal solution. Utilizing computer technology, it enables rapid initialization and parallel solving / iteration of multi-algorithm collaboration. By extracting characteristics of the winning community, it ensures that the exchanged individuals possess high value. Attached Figure Description

[0087] To reveal the technical details of the embodiments of the present invention, the accompanying drawings involved in the embodiments will be briefly described below. It should be emphasized that these drawings only present several embodiments of the present invention and should not be considered as defining the scope of the invention. For those skilled in the art, other related drawings can still be derived based on these drawings without inventive effort.

[0088] Figure 1 Flowchart for optimizing the solution using a hybrid heuristic algorithm;

[0089] Figure 2 Here is a flowchart of the genetic algorithm computation process;

[0090] Figure 3 The flowchart for the particle swarm optimization algorithm is shown below.

[0091] Figure 4 Flowchart of the algorithm for optimizing firefly swarms. Detailed Implementation

[0092] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings, but the scope of protection of the present invention is not limited to the following description.

[0093] In the following description, embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. However, it should be understood that the present disclosure is not limited to the specific forms shown herein. Rather, it should be understood to encompass various variations, equivalents, and / or alternatives to the embodiments of the present disclosure. In illustrating the drawings, the same reference numerals will be used to denote similar components.

[0094] In the various embodiments of this disclosure, the terms "first," "second," "the first," or "the second" are intended to modify different components and not to indicate order and / or importance, nor do they constitute a limitation on the respective components. For example, a first user equipment and a second user equipment represent different user equipments, although they both fall under the category of user equipment. Similarly, a first component may be named a second component, and a second component may be named a first component, without changing their essential attributes within the scope of this disclosure.

[0095] In this disclosure, terminology is used to describe specific embodiments and does not constitute a limitation thereof. In this context, the use of the singular form also encompasses the plural form, unless otherwise expressly stated herein. In the course of description, terms such as “comprising” or “having” are intended to indicate the presence of features, quantities, steps, operations, structural components, parts, or combinations thereof, and do not preclude the possibility or addition of one or more other features, quantities, steps, operations, structural components, parts, or combinations thereof.

[0096] It should be clarified that while the following description provides detailed specific information to aid in a comprehensive understanding of the exemplary embodiments, those skilled in the art will recognize that the exemplary embodiments can be implemented even without these specific details. For example, the system may be illustrated using block diagrams to avoid excessive detail that could obscure the clarity of the example. In other cases, to maintain the clarity of the example, unnecessary details of well-known processes, structures, and techniques may be omitted.

[0097] A collaborative optimization method for control parameters based on hybrid heuristic optimization, the method comprising:

[0098] The Ns basic optimization algorithms are co-initialized by a hybrid heuristic optimization thread, and the target optimization problem is solved in parallel based on the native iterative mechanism of each algorithm.

[0099] When any algorithm satisfies the first convergence condition or the initial iteration number iter1 reaches the Maxiter1 threshold, feature space data containing the winning community of the algorithms is generated, and the current optimal solution of each algorithm is extracted.

[0100] A cross-algorithm superior community interaction mechanism is constructed. The feature space data is imported into a pre-set superior community feature library in the computer. Based on the population diversity feature coefficient, feature sampling and cross-population embedding operations are performed from different superior communities.

[0101] When a single algorithm satisfies the second convergence condition or the total number of iterations iter2 reaches or exceeds the Maxiter2 threshold, the optimization coefficients of the objective function of the algorithm are output.

[0102] By using a multi-objective fusion decision unit to verify the stability and robustness of the optimization coefficients of all objective functions, the global optimal solution is determined.

[0103] Hybrid heuristic optimization threads utilize computer multithreading technology to execute different optimization algorithms in parallel within the same program. This fully leverages the multi-core processing power of the computer, improving algorithm execution efficiency. Cooperative initialization allows for rapid initialization of the algorithm's parameter space, laying the foundation for parallel solution. By combining hybrid heuristic optimization and parallel solution, optimization efficiency and accuracy can be effectively improved, while reducing computational costs. Furthermore, the cooperative optimization method integrates the advantages of different algorithms, enhancing overall optimization performance.

[0104] In a preferred embodiment, the hybrid heuristic optimization thread is constructed by injecting a population feature transfer mechanism (or a cross-algorithm winner community interaction mechanism) and a parameter adaptive adjustment strategy into the baseline optimization thread. The baseline optimization thread includes native operators of genetic algorithms, particle swarm optimization algorithms, and firefly swarm optimization algorithms. The population feature transfer mechanism is an optimization algorithm strategy that extracts features or information from the population of one algorithm (source algorithm) and transfers them to another algorithm (target algorithm) to enhance the search capability and optimization performance of the target algorithm.

[0105] The cross-algorithm winner community interaction mechanism includes:

[0106] The feature space data is spliced ​​using a feature vector fusion unit to generate a superior community feature vector with cross-algorithm correlation.

[0107] The population diversity feature coefficient is calculated based on the feature vector of the superior community, and the sampling ratio and feature stratification interval for cross-population embedding are determined based on the coefficient.

[0108] Cross-algorithm winner community interaction mechanisms refer to the information exchange and feature sharing among winning populations of multiple algorithms to promote synergy and improve optimization results throughout the optimization process. This includes population feature transfer. Through the interaction of winning populations, collaborative optimization among algorithms is enhanced, avoiding the limitations of a single algorithm and improving overall optimization performance.

[0109] The method for constructing the parameter adaptive adjustment strategy includes:

[0110] Monitor the rate of change of the objective function value of each algorithm within a continuous preset iteration window;

[0111] When the rate of change is lower than the dynamic adjustment threshold, the parameter perturbation module is triggered to perform nonlinear adjustment on the core parameters of the current algorithm.

[0112] Perform an exponential amplification operation on the mutation probability of the genetic algorithm;

[0113] Perform a piecewise decay operation on the inertia weights in the particle swarm optimization algorithm;

[0114] An adaptive recalibration operation of the luciferin coefficient is performed on the firefly algorithm.

[0115] like Figure 1 As shown, there are Ns heuristic algorithms used for hybrid optimization. All algorithms are initialized and solved according to their own solution process. When any algorithm converges or the number of iterations (iter1) exceeds the maximum iteration limit (Maxiter1), a winning community is generated, and the optimal solution is calculated. The winning communities of different algorithms are extracted into a winning community database. Individuals are randomly selected from different winning communities and inserted into the random populations of other random algorithms for exchange, and then optimization continues. When the total number of iterations (iter2) of a single algorithm is greater than the total minimum number of iterations (Miniter2) and the change in the optimal solution of the objective function is less than or equal to a threshold or reaches the maximum number of iterations (Maxiter2), the optimal solution of that algorithm is output. Finally, the optimal solutions of all algorithms are compared to obtain the final optimal solution.

[0116] By injecting a mechanism into the benchmark optimization thread, population features are transferred between different algorithms, and parameters are dynamically adjusted. This mechanism can better utilize the advantages of different algorithms and improve optimization performance. The population feature transfer and adaptive parameter adjustment strategy can improve the algorithm's adaptability and robustness, enabling it to better adapt to different optimization problems and environmental changes. Simultaneously, this strategy can also prevent the algorithm from getting trapped in local optima, improving its global optimization capability.

[0117] In a preferred embodiment, the baseline optimization thread includes native operators of genetic algorithms, particle swarm optimization algorithms, and firefly swarm optimization algorithms, specifically manifested as follows:

[0118] The genetic algorithm adopts a chromosome encoding mechanism. Its operators include gene recombination units based on roulette wheel selection operators, population evolution units based on single-point crossover operators, and gene perturbation units based on basic position mutation operators.

[0119] The particle swarm optimization algorithm adopts a velocity-position update model, and its operators include an individual optimal position tracking unit and a global optimal position fusion unit.

[0120] The firefly swarm optimization algorithm employs a brightness attraction mechanism, and its operators include a relative brightness calculation unit and a dynamic distance response unit.

[0121] The baseline optimization thread includes native operators for genetic algorithms, particle swarm optimization (PSO), and firefly swarm optimization (FSO). The genetic algorithm employs a chromosome encoding mechanism, achieving evolution through gene recombination and population evolution. PSO uses a velocity and position update model, finding the optimal position through the interaction between individual units and the global optimum. FSO employs a brightness attraction mechanism, finding the optimal position by calculating relative brightness and dynamic distance response. By combining the advantages of different algorithms, this method can more comprehensively search the solution space, improving the accuracy and efficiency of optimization. Furthermore, the use of native operators for each algorithm ensures its stability and reliability.

[0122] In a preferred embodiment

[0123] The initialization process of the benchmark optimization thread includes:

[0124] Genetic algorithm dimension: Set a gene encoding parameter space including population size, chromosome length, crossover probability threshold, and mutation probability threshold, and generate an initial population that satisfies uniform distribution within the feasible region of control parameters;

[0125] Particle swarm optimization dimension: Define the kinematic parameter space including the number of particles, learning factor matrix, and inertia weight curve, and generate a position matrix with random initial velocity within the boundary of the feasible region of control parameters;

[0126] Firefly Algorithm Dimensions: Construct a biological excitation parameter space containing the number of fireflies, attribute dimensions, and search space topology, and generate an initial position tensor that satisfies a Gaussian distribution within a predefined multidimensional feasible region.

[0127] In a preferred embodiment

[0128] The iterative evolution process of native operators includes:

[0129] Genetic algorithms execute a three-stage evolutionary process of gene selection, crossover, and mutation, and drive iterative population updates through fitness function evaluation units;

[0130] The particle swarm optimization algorithm performs a two-stage motion process of velocity vector update and position vector correction, and realizes information interaction through individual historical best solution memory units and global best solution sharing units;

[0131] The firefly algorithm executes a three-stage biomimetic process of brightness calculation, position attraction, and parameter update, and achieves swarm intelligence evolution through a relative distance calculation unit and an attraction intensity adjustment unit.

[0132] The iterative evolution process of the native operators includes three stages: gene selection, crossover, and mutation in the genetic algorithm; two stages: velocity vector update and position vector correction in the particle swarm optimization algorithm; and three stages: brightness calculation, position attraction, and parameter update in the firefly algorithm. These processes drive the iterative update of the population through a fitness function evaluation unit, achieving self-optimization and evolution of the algorithm. The explicit evolutionary process and operators make the optimization process of the algorithm clearer and more controllable. Simultaneously, the iterative update of the population enhances the algorithm's adaptability and robustness.

[0133] In a preferred embodiment

[0134] Methods for constructing a database of characteristics of superior communities include:

[0135] For each winning community, a multidimensional feature label is established, including algorithm identifier, iterative stage feature code, gradient of objective function value change, and population diversity index;

[0136] A feature encoding mechanism is employed to convert the multidimensional feature labels into traceable feature vectors, constructing a spatiotemporally correlated feature map of the winning communities. The method for constructing the winning community feature library includes establishing multidimensional feature labels for each winning community. These labels contain information such as algorithm identifiers, iterative stage feature codes, gradients of objective function value changes, and population diversity indices. Then, a feature encoding mechanism is used to convert these multidimensional feature labels into traceable feature vectors, constructing a spatiotemporally correlated feature map of the winning communities. By constructing the winning community feature library, this method can record and trace the feature information of each winning community, providing a basis for subsequent cross-population embedding operations. Simultaneously, the spatiotemporally correlated feature map can better reflect the evolution and interaction processes of the winning communities.

[0137] In a preferred embodiment, the feature sampling and cross-population embedding operation specifically includes:

[0138] The feature extraction amount of each winning community is determined based on the dynamic sampling ratio calculation module, and the dynamic sampling ratio is positively correlated with the population diversity index.

[0139] A hierarchical feature sampling strategy is adopted to divide the victorious community into multiple feature subspaces according to the fitness distribution, and perform probability density-based adaptive sampling in each subspace.

[0140] The extracted feature samples are injected into the target population using a random position insertion algorithm, which includes a gene sequence recombination verification unit and a population capacity balancing unit. Feature sampling and cross-population embedding operations involve determining the feature extraction amount for each winning community based on a dynamic sampling ratio calculation module, dividing the winning communities into multiple feature subspaces according to their fitness distribution using a hierarchical feature sampling strategy, and performing probability density-based adaptive sampling within each subspace. Then, the extracted feature samples are injected into the target population using the random position insertion algorithm, achieving feature transfer and population recombination. The dynamic sampling and hierarchical feature sampling strategies enable the effective extraction and utilization of features from winning communities. Simultaneously, injecting feature samples into the target population using the random position insertion algorithm promotes population diversity and intelligence. This method also accelerates the convergence speed and accuracy of the optimization process.

[0141] In a preferred embodiment

[0142] The parameter perturbation module includes:

[0143] The genetic algorithm mutation enhancement unit uses an exponential mutation probability function;

[0144] The particle swarm inertia weighting unit employs a piecewise linear decay function;

[0145] The firefly luciferin regulation unit employs an adaptive update formula.

[0146] In a preferred embodiment

[0147] The methods for determining the second convergence condition include:

[0148] A joint decision model with dual convergence criteria is constructed. The first convergence flag is triggered when the total number of iterations iter2 > Miniter2 and the change in the objective function value Δf ≤ ε. The first convergence flag ensures timely termination when sufficient accuracy is achieved, avoiding resource waste. The first convergence flag can be used to initiate feature transfer.

[0149] A second convergence flag is triggered when the algorithm gets stuck in a local optimum for more than a preset threshold number of iterations. This second convergence flag serves as a fallback condition to prevent the algorithm from failing to terminate due to complex problems. The second convergence flag can be used to trigger the termination process.

[0150] The dual convergence criterion joint decision-making model employs a weighted voting mechanism to comprehensively determine the algorithm's termination timing. The weight of the first convergence flag can be set to 0.4, and the weight of the second convergence flag can be set to 0.6. By combining the two termination conditions, the algorithm can terminate when the conditions are met, avoiding wasting time on unnecessary iterations.

[0151] In a preferred embodiment

[0152] Stability verification and robustness analysis include:

[0153] Establish a solution space volatility evaluation index, calculate the standard deviation σ and coefficient of variation CV of each candidate optimal solution in multiple independent runs; construct a multi-dimensional robustness evaluation matrix, including parameter sensitivity analysis, boundary condition testing and noise interference experiments;

[0154] The weight coefficients of each evaluation index are determined by the analytic hierarchy process (AHP), and the final optimal solution is determined by weighted comprehensive evaluation.

[0155] In a preferred embodiment

[0156] The method also includes optimizing the process traceability mechanism:

[0157] Construct an algorithm evolution log database to record multidimensional evolutionary data in real time, including population gene distribution, fitness change curves, and parameter adjustment trajectories.

[0158] Design a visualization analysis interface to convert the multidimensional evolution data into a visualization analysis report that includes a three-dimensional convergence surface, a parameter sensitivity heatmap, and a population diversity radar chart.

[0159] A control parameter collaborative optimization system based on hybrid heuristic optimization includes:

[0160] The system comprises a hybrid heuristic optimization engine module, a winner community interaction module, a parameter adaptive adjustment module, a multi-objective decision-making module, and an algorithm base library module. Specifically: the hybrid heuristic optimization engine module connects to both the algorithm base library module and the winner community interaction module, transmitting both the algorithm's native iterative data and the winner community feature space data; the parameter adaptive adjustment module is bidirectionally connected to the hybrid heuristic optimization engine module, transmitting the objective function's rate of change data in real time and providing feedback on parameter adjustment commands; and the multi-objective decision-making module receives optimization coefficient data from all algorithm channels and performs stability verification and robustness analysis.

[0161] The hybrid heuristic optimization engine module includes:

[0162] Collaborative initialization unit: Connects to the algorithm base library module, initializes the parameter space of Ns basic optimization algorithms, and establishes a parallel solution channel;

[0163] Native Iterative Execution Unit: Built-in genetic algorithm operator, particle swarm optimization operator and firefly swarm optimization operator, driving each algorithm to perform iterative calculations according to the native mechanism;

[0164] Convergence monitoring unit: connected to each algorithm channel, it determines the iteration termination condition through the convergence condition discrimination module and the Maxiter counter module;

[0165] Feature extraction unit: Generates feature space data containing the algorithm's winning clusters and extracts the current optimal solution vector for each channel;

[0166] The superior community interaction module includes:

[0167] Feature vector fusion unit: Receives feature space data from various algorithm channels and performs multi-dimensional feature concatenation to generate cross-algorithm related feature vectors;

[0168] Diversity analysis unit: Determines the feature sampling ratio and stratification interval based on the population diversity characteristic coefficient calculation module;

[0169] Cross-population embedding unit: connects to the hybrid heuristic optimization engine module to perform dynamic migration and population recombination operations of the winning feature vectors;

[0170] The parameter adaptive adjustment module includes:

[0171] Iterative window monitoring unit: Statistically tracks the rate of change of the objective function value of each algorithm using a sliding time window;

[0172] Disturbance triggering unit: When the rate of change is lower than the dynamic adjustment threshold, the parameter nonlinear adjustment channel is activated;

[0173] Algorithm regulator group: includes genetic algorithm mutation probability amplifier, particle swarm inertial weight attenuator, and firefly luciferin calibrator;

[0174] The multi-objective decision-making module includes:

[0175] Stability verification unit: The robustness of each optimization coefficient is tested using the Monte Carlo simulation module;

[0176] Decision fusion unit: The fuzzy comprehensive evaluation method is used to screen the global optimal solution of the Pareto front solution set;

[0177] Results output interface: Generates an analysis report containing the optimized parameter configuration scheme and the objective function response surface;

[0178] The algorithm's basic library module stores:

[0179] The core operators of genetic algorithms include selection operators, crossover operators, mutation operators, and fitness functions.

[0180] Particle swarm optimization component set: including velocity update calculator, position updater, and neighborhood topology template;

[0181] Firefly Algorithm Function Package: Includes a luciferin update module, a movement probability calculator, and an attraction calculation unit;

[0182] The feature transfer operations performed by the cross-population embedding unit include: feature space similarity calculation based on Mahalanobis distance; population recombination strategy using quantum rotation gate mechanism; and feature hierarchical sampling method based on information entropy weight.

[0183] The nonlinear adjustment channel for the parameters includes: an exponential amplification function for the mutation probability of the genetic algorithm; a piecewise decay strategy for the particle swarm inertia weight; and an adaptive calibration model for the firefly luciferin coefficient.

[0184] This system is based on a hybrid heuristic optimization method, integrating multiple basic optimization algorithms and establishing parallel solution channels through a collaborative initialization unit. In the native iterative execution unit, genetic algorithm operators, particle swarm optimization operators, and firefly swarm optimization operators are built-in, driving each algorithm to perform iterative operations according to its native mechanism. The convergence monitoring unit is responsible for monitoring the iteration process of each algorithm, determining the iteration termination condition through a convergence condition discrimination module and a Maxiter counter module. The feature extraction unit generates feature space data containing the winning communities of the algorithms and extracts the current optimal solution vector for each channel. This feature space data is transmitted to the winning community interaction module, where the feature vector fusion unit performs multi-dimensional feature splicing to generate cross-algorithm related feature vectors. The diversity analysis unit determines the feature sampling ratio and stratification interval based on the population diversity feature coefficient calculation module, while the cross-population embedding unit performs dynamic migration and population recombination operations of the winning feature vectors to enhance the algorithm's search capability and avoid premature convergence. The parameter adaptive adjustment module uses the iterative window monitoring unit to statistically analyze the rate of change of the objective function value of each algorithm. When the rate of change is lower than the dynamic adjustment threshold, the perturbation trigger unit activates the parameter nonlinear adjustment channel. The algorithm regulator group contains regulators for different algorithms, such as a genetic algorithm mutation probability amplifier, a particle swarm inertia weight attenuator, and a firefly luciferin calibrator, used to adjust algorithm parameters in real time to optimize the search process. The multi-objective decision module receives optimization coefficient data from all algorithm channels, and the stability verification unit performs robustness tests on each optimization coefficient using a Monte Carlo simulation module. The decision fusion unit uses a fuzzy comprehensive evaluation method to screen for the global optimal solution from the Pareto front solution set, and finally generates an analysis report containing optimization parameter configuration schemes and the objective function response surface through the result output interface.

[0185] This invention selects genetic algorithm, particle swarm optimization algorithm, and firefly swarm optimization algorithm as the basic algorithms to establish a hybrid heuristic optimization solution algorithm. The basic principles and calculation process of each basic algorithm are described below.

[0186] (1) Genetic Algorithm

[0187] Genetic Algorithm (GA) is an adaptive global optimization algorithm that simulates the heredity and evolution of organisms in their natural environment. Its basic idea follows the principle of "natural selection and survival of the fittest" in biological evolution. The algorithm primarily uses encoding to represent individuals, evaluates the quality of individuals using a fitness function, and employs evolutionary operations such as selection, crossover, and mutation to achieve optimal search.

[0188] The computational flowchart of the genetic algorithm is as follows: Figure 2 As shown.

[0189] (2) Particle Swarm Optimization Algorithm

[0190] Particle Swarm Optimization (PSO) is a swarm intelligence optimization algorithm based on the foraging behavior of bird flocks. The algorithm analogizes the search space of the optimization problem to the flight space of the birds, and the optimal solution to the food they are searching for. Each bird is abstracted as a massless particle, containing two feature vectors: position and velocity. During the optimization process, the position and velocity of each particle are first randomly initialized, and then updated iteratively.

[0191] The computational flowchart of the particle swarm optimization algorithm is as follows: Figure 3 As shown.

[0192] (3) Firefly Swarm Optimization Algorithm

[0193] The Glowworm Swarm Optimization Algorithm (GSOA), proposed by Krishnanand et al., is based on the bioluminescence principle of fireflies. In the basic GSOA, each firefly is distributed within the defined space of the objective function. Fireflies carry their own luciferin and have their own field of vision. This field of vision is called the regional decision radius, and its size is affected by the number of surrounding fireflies. When the surrounding firefly density is low, the regional decision radius increases to facilitate finding more fireflies; conversely, the decision radius decreases. Ultimately, fireflies will gather around the firefly with the highest fitness value, thus determining the optimal value of the objective function.

[0194] The computational flowchart of the firefly optimization algorithm is as follows: Figure 4 As shown.

[0195] In one embodiment, a method for collaborative optimization of control parameters of a subway station air conditioning system based on hybrid heuristic optimization (Metro Station Air Conditioning System Control Parameter Collaborative Optimization Method) is proposed. This method uses a hybrid heuristic optimization thread to collaboratively initialize multiple basic optimization algorithms (including genetic algorithm, particle swarm optimization algorithm and firefly algorithm), and performs parallel solutions for key control parameters of the subway station air conditioning system such as outlet air temperature, wind speed and humidity.

[0196] When any algorithm satisfies the first convergence condition or the initial iteration count iter1 reaches the set threshold Maxiter1, feature space data containing the winning cluster is generated, and the optimal settings for outlet air temperature and wind speed in the current winning solution are extracted. The combination of these parameters directly affects the energy consumption and passenger comfort of the air conditioning system.

[0197] Next, an interaction mechanism for cross-algorithm winning communities is constructed. A feature vector fusion unit is used to perform multi-dimensional feature stitching on the feature space data, forming a winning community feature vector with cross-algorithm correlation. In this process, the advantages of each algorithm in controlling outlet air temperature and wind speed optimization are integrated to enhance the complementarity of overall parameters. This feature vector is then used to calculate the population diversity feature coefficient, thereby determining the sampling ratio and feature stratification interval for cross-population embedding, thus more effectively adjusting relevant control parameters.

[0198] When an algorithm satisfies the second convergence condition or the total number of iterations (iter2) reaches another set threshold (Maxiter2), the optimization coefficients of the algorithm's objective function are output, such as the energy consumption ratio and comfort index. These coefficients clearly reflect the optimal combination strategy of wind temperature and wind speed. Subsequently, a multi-objective fusion decision unit is used to verify the stability and robustness of all objective function optimization coefficients, ensuring that the configuration of key parameters such as outlet air temperature, wind speed, and humidity can maximize passenger comfort and system energy efficiency in the actual operating environment of a subway station.

[0199] To enhance the combined effect of parameters, the method incorporates a population feature transfer mechanism and an adaptive parameter adjustment strategy, monitoring in real time the rate of change of the objective function values ​​of each algorithm with respect to outlet air temperature, wind speed, and humidity within a specified iteration window. When the rate of change falls below the dynamic adjustment threshold, the parameter perturbation module is triggered to perform nonlinear adjustments to the core parameters. This includes dynamically optimizing the mutation probability of the genetic algorithm, piecewise decaying the inertia weight of the particle swarm optimization algorithm, and adaptively recalibrating the luciferin coefficient of the firefly algorithm, ensuring optimal adjustment of outlet air temperature and wind speed during the iteration process.

[0200] The overall solution, through the synergistic effect of various links, deepens the integration and optimization of control parameters in the subway station air conditioning system, in order to achieve the goals of reducing energy consumption, improving passenger comfort, and strengthening intelligent system management.

[0201] In another embodiment, we provide an alternative solution.

[0202] A collaborative optimization method for parameters of a subway station air conditioning system based on hybrid heuristic optimization (Collaborative Optimization Method for Parameters of Subway Station Air Conditioning System), the method comprising:

[0203] The genetic algorithm, particle swarm optimization algorithm and firefly swarm optimization algorithm are co-initialized by a hybrid heuristic optimization thread. Based on the native iteration mechanism of each algorithm, the multi-objective optimization problem of energy efficiency and comfort of air conditioning system is solved in parallel. The optimization parameters include supply air temperature setpoint, fan frequency, chilled water valve opening degree and fresh air ratio.

[0204] When any algorithm satisfies the first convergence condition or the initial iteration number iter1 reaches the Maxiter1 threshold, feature space data containing temperature field distribution characteristics, equipment energy consumption characteristics, and passenger density characteristics is generated, and the current optimal combination of equipment control parameters for each algorithm is extracted.

[0205] A cross-algorithm superior community interaction mechanism is constructed. The feature space data is imported into the superior community feature library containing historical operating condition features. The population diversity feature coefficient is calculated based on the heat load fluctuation coefficient and the passenger density change rate. Based on this, the air supply parameter sampling and cross-population embedding operation between different superior communities are performed.

[0206] When a single algorithm satisfies the second convergence condition or the total number of iterations iter2 reaches the Maxiter2 threshold, the equipment energy efficiency optimization coefficient and temperature uniformity index of the algorithm are output.

[0207] The stability of all optimization coefficients under dynamic load conditions is verified by a multi-objective fusion decision unit. The global optimal solution is determined based on energy consumption fluctuation tolerance and temperature deviation threshold. The verification process incorporates real-time meteorological parameters and passenger flow density monitoring data.

[0208] The hybrid heuristic optimization thread is constructed by injecting a device parameter migration mechanism and an environment parameter adaptive strategy, wherein:

[0209] The parameter migration mechanism enables the migration and recombination of wind turbine frequency parameter matrices between different algorithms, and the environmental parameter adaptive strategy includes:

[0210] Monitor the rate of change of COP value for each algorithm within three consecutive iteration windows, and trigger a dynamic adjustment when the rate of change falls below the threshold.

[0211] An exponential amplification operation is performed on the genetic algorithm to amplify the probability of air supply temperature variation, and the amplification factor is positively correlated with the real-time passenger flow density.

[0212] The particle swarm optimization algorithm performs a piecewise decay operation on the inertial weights, and the decay rate is linked to the rate of change of outdoor temperature and humidity.

[0213] An adaptive recalibration operation of the luciferin coefficient is performed on the firefly algorithm, with the calibration benchmark value being the best value of the same period in history for the current time period.

[0214] This invention also provides a collaborative optimization method for air conditioning cooling control parameters in subway stations based on hybrid heuristic optimization, namely, a collaborative optimization method for air conditioning cooling control parameters in subway stations, the method comprising the following steps:

[0215] Initialization phase: Parameterize the key control parameters of the subway station's air conditioning system, including the following parameters:

[0216] Temperature set point (T)set ),

[0217] Humidity set point (H) set ),

[0218] Air volume (Q) air ),

[0219] Chiller operating frequency (f) chiller ),

[0220] Compressor operating pressure (P) compressor ).

[0221] Hybrid heuristic optimization thread construction: Based on genetic algorithm, particle swarm optimization (PSO) algorithm, and firefly swarm optimization (FA) algorithm, an optimization population is initialized, and the control parameters of the air conditioning system are encoded as optimization variables. The initial population size and parameter encoding range are set, and a population feature transfer mechanism and parameter adaptive adjustment strategy are established.

[0222] Parallel solution and collaborative optimization:

[0223] Each optimization algorithm solves the control parameters of the air conditioning system in parallel based on its native iterative mechanism (such as gene recombination in genetic algorithms, velocity and position updates in PSO, and brightness attraction mechanism in FA).

[0224] When any algorithm satisfies the first convergence condition (such as the rate of change of the objective function value being lower than a threshold or the number of iterations reaching a preset value), the current optimal control parameter configuration of the algorithm is extracted, that is, the winning group feature data of the algorithm is extracted, and a multi-dimensional feature vector containing the current optimal solution of the air conditioning system (such as the optimal temperature set point and the optimal air supply volume) is generated.

[0225] Cross-algorithm winner community interaction and embedding (cross-algorithm winner community interaction mechanism):

[0226] The winning community feature data of each algorithm are imported into the winning community feature library. The feature vector fusion unit is used to perform multi-dimensional feature splicing on the feature space data to generate a winning community feature vector with cross-algorithm correlation. Based on the winning community feature vector, the population diversity feature coefficient is calculated, and the sampling ratio and feature stratification interval of cross-population embedding are dynamically determined according to this coefficient.

[0227] The feature sampling ratio is dynamically adjusted based on the population diversity coefficient, and representative control parameter configurations are extracted from the winning population. A random position insertion algorithm is used to inject the extracted control parameter configurations into the target algorithm population, achieving cross-algorithm knowledge transfer and enhancing population diversity.

[0228] Parameter adaptive adjustment strategy: Monitor the rate of change of the objective function value of each algorithm within a continuous preset iteration window. When the rate of change is lower than the dynamic adjustment threshold, trigger the parameter perturbation module to perform nonlinear adjustment on the core parameters of the current algorithm.

[0229] Optimization termination and confirmation of the global optimal solution:

[0230] When a single algorithm reaches Maxiter2 in the total number of iterations iter2 or the change in the objective function value Δf ≤ ε, ε is the convergence threshold, and the optimal control parameter configuration of the algorithm is output.

[0231] By using a multi-objective fusion decision unit to verify the stability and robustness of the optimal control parameter configurations of each algorithm (e.g., standard deviation < 0.15), the optimal combination of subway station air conditioning control parameters is finally determined to ensure the efficient operation of the system and the comfortable experience of passengers.

[0232] In another embodiment, a different method for collaborative optimization of temperature control parameters of subway station air conditioning systems based on hybrid heuristic optimization is provided, namely, a method for collaborative optimization of air conditioning refrigeration control parameters.

[0233] Controlled parameters: Station hall ambient temperature (target parameter: 26±1℃, constraint range: 24-28℃); Air conditioning supply volume (target parameter: dynamic adjustment, constraint range: 2000-5000m³). 3 / h)

[0234] Hybrid optimization thread initialization and parameter modeling

[0235] Benchmark Optimization Thread Construction

[0236] Genetic Algorithm (GA) Dimensions

[0237] Control parameters: Air conditioner start / stop threshold, air volume adjustment step size

[0238] Initialization: Population size = 50, chromosome coding length = 10 (5-bit temperature parameter + 5-bit airflow parameter), crossover probability = 0.8, mutation probability = 0.05

[0239] Particle Swarm Optimization (PSO) Dimensions

[0240] Control parameters: Temperature PID coefficients (Kp, Ki, Kd), airflow rate of change

[0241] Initialization: Particle count = 30, learning factors c1 = 1.5 (temperature parameter weight), c2 = 1.2 (airflow parameter weight), inertia weight ω = 0.9

[0242] Firefly Algorithm (FA) Dimensions

[0243] Control parameters: Temperature setpoint offset, air volume dynamic response time

[0244] Initialization: Firefly number = 40, luciferin attenuation coefficient = 0.4, attraction coefficient β = 0.3

[0245] Overall optimization goals:

[0246]

[0247] Treal: Actual temperature, Tset: Set temperature (26℃)

[0248] Qactual: Actual air volume; Qmax: Maximum air volume (5000m³ / h) 3 / h)

[0249] Epower: Real-time power of the air conditioning system (kW)

[0250] Weighting coefficients: α = 0.6 (temperature priority), β = 0.3 (airflow efficiency), γ = 0.1 (energy consumption)

[0251] Superior Community Interactions and Parameter Optimization

[0252] Cross-algorithm feature transfer

[0253] GA Superior Community Extraction: Select the top 10% of individuals with fitness values ​​(temperature fluctuation <0.5℃ and optimal airflow adjustment step size).

[0254] PSO Superior Community Extraction: Selecting the globally optimal particle (PID parameter combination to achieve temperature overshoot <5%)

[0255] FA (Flame Analyzer) winning community extraction: Select fireflies with the highest brightness values ​​(air supply response time < 30 seconds).

[0256] Feature embedding and parameter tuning

[0257] GA→PSO migration: Insert the GA airflow regulation step size gene sequence into the PSO particle velocity update formula.

[0258] FA→GA migration: The temperature offset parameter of FA is introduced into the GA chromosome as a mutation operator.

[0259] Parameter adaptive adjustment strategy

[0260] Triggering conditions: Temperature fluctuation > 0.8℃ for 5 consecutive iterations or air supply volume exceeds limit.

[0261] Genetic Algorithm: Mutation Probability Adjusted by Exponential Growth

[0262] Where t is the number of consecutive invalid iterations;

[0263] Particle Swarm Optimization Algorithm: Segmented Weight Decay

[0264] Firefly Algorithm: Dynamic Recalibration of Fluorescein Coefficient

[0265] Convergence determination and global optimal solution generation

[0266] Convergence condition:

[0267] First convergence criterion: Total number of iterations > 50 and change in objective function value Δf ≤ 0.01

[0268] Second convergence criterion: The air supply volume does not exceed the constraint range for 10 consecutive iterations and the temperature standard deviation is <0.3℃.

[0269] Verification of the global optimal solution:

[0270] Stability analysis: Three candidate solutions (temperature-airflow parameter combinations) were selected, and a 72-hour field test was conducted at a subway station during the evening peak hours (passenger flow 2000 people / hour).

[0271]

[0272] After robust testing, it meets the requirements for temperature and air volume.

[0273] Although preferred embodiments of the present invention have been described in detail, those skilled in the art, upon grasping the basic inventive concept, can make further changes and modifications to these embodiments. Therefore, the appended claims are intended to cover these preferred embodiments and all changes and modifications falling within the scope of the present invention. The above descriptions are merely preferred embodiments of the present invention and are not intended to limit its scope. It should be noted that any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for collaborative optimization of control parameters based on hybrid heuristic optimization, characterized in that, The method includes: The genetic algorithm, particle swarm optimization algorithm and firefly swarm optimization algorithm are co-initialized by a hybrid heuristic optimization thread. Based on the native iteration mechanism of each algorithm, the multi-objective optimization problem of energy efficiency and comfort of air conditioning system is solved in parallel. The optimization parameters include supply air temperature setpoint, fan frequency, chilled water valve opening degree and fresh air ratio. The hybrid heuristic optimization thread is constructed by injecting a population feature transfer mechanism and a parameter adaptive adjustment strategy into the baseline optimization thread; When any algorithm satisfies the first convergence condition or the initial iteration number iter1 reaches the Maxiter1 threshold, feature space data containing temperature field distribution characteristics, equipment energy consumption characteristics, and passenger density characteristics is generated, and the current optimal combination of equipment control parameters for each algorithm is extracted. A cross-algorithm superior community interaction mechanism is constructed. The feature space data is imported into the superior community feature library containing historical operating condition features. Based on the population diversity feature coefficient, feature sampling and cross-population embedding operations are performed from different superior communities. Specifically, the population diversity feature coefficient is calculated based on the heat load fluctuation coefficient and the passenger density change rate. Based on this, the air supply parameter sampling and cross-population embedding operations between different superior communities are performed. When a single algorithm satisfies the second convergence condition or the total number of iterations iter2 reaches the Maxiter2 threshold, the equipment energy efficiency optimization coefficient and temperature uniformity index of the algorithm are output. The stability of all optimization coefficients under dynamic load conditions is verified by a multi-objective fusion decision unit. The global optimal solution is determined based on energy consumption fluctuation tolerance and temperature deviation threshold. Real-time meteorological parameters and passenger flow density monitoring data are introduced in the verification process. Feature sampling and cross-population embedding operations specifically include: The feature extraction amount of each winning community is determined based on the dynamic sampling ratio calculation module, and the dynamic sampling ratio is positively correlated with the population diversity index. A hierarchical feature sampling strategy is adopted to divide the victorious community into multiple feature subspaces according to the fitness distribution, and perform probability density-based adaptive sampling in each subspace. The extracted feature samples are injected into the target population through a random position insertion algorithm, which includes a gene sequence recombination verification unit and a population capacity balancing unit.

2. The control parameter collaborative optimization method based on hybrid heuristic optimization according to claim 1, characterized in that: Cross-algorithm winner community interaction mechanisms include: The feature space data is spliced ​​using a feature vector fusion unit to generate a superior community feature vector with cross-algorithm correlation. The population diversity feature coefficient is calculated based on the feature vector of the superior community, and the sampling ratio and feature stratification interval for cross-population embedding are determined based on the coefficient. The method for constructing the parameter adaptive adjustment strategy includes: Monitor the rate of change of the objective function value of each algorithm within a continuous preset iteration window; When the rate of change is lower than the dynamic adjustment threshold, the parameter perturbation module is triggered to perform nonlinear adjustment on the core parameters of the current algorithm. Perform an exponential amplification operation on the mutation probability of the genetic algorithm; Perform a piecewise decay operation on the inertia weights in the particle swarm optimization algorithm; An adaptive recalibration operation of the luciferin coefficient is performed on the firefly algorithm.

3. The control parameter collaborative optimization method based on hybrid heuristic optimization according to claim 2, characterized in that: The native operators of genetic algorithms, particle swarm optimization algorithms, and firefly swarm optimization algorithms are specifically manifested as follows: The genetic algorithm adopts a chromosome encoding mechanism. Its operators include gene recombination units based on roulette wheel selection operators, population evolution units based on single-point crossover operators, and gene perturbation units based on basic position mutation operators. The particle swarm optimization algorithm employs a velocity-position update model, whose operators include an individual optimal position tracking unit and a global optimal position fusion unit. The firefly swarm optimization algorithm employs a brightness attraction mechanism, and its operators include a relative brightness calculation unit and a dynamic distance response unit.

4. The control parameter collaborative optimization method based on hybrid heuristic optimization according to claim 3, characterized in that: The iterative evolution process of native operators includes: Genetic algorithms execute a three-stage evolutionary process of gene selection, crossover, and mutation, and drive iterative population updates through fitness function evaluation units; The particle swarm optimization algorithm performs a two-stage motion process of velocity vector update and position vector correction, and realizes information interaction through individual historical optimal solution memory units and global optimal solution sharing units; The firefly algorithm executes a three-stage biomimetic process of brightness calculation, position attraction, and parameter update, and achieves swarm intelligence evolution through a relative distance calculation unit and an attraction intensity adjustment unit.

5. The control parameter collaborative optimization method based on hybrid heuristic optimization according to any one of claims 1-4, characterized in that: Methods for constructing a database of characteristics of superior communities include: For each winning community, a multidimensional feature label is established, including algorithm identifier, iterative stage feature code, gradient of objective function value change, and population diversity index; The multidimensional feature labels are converted into traceable feature vectors using a feature encoding mechanism to construct a feature map of the superior community with spatiotemporal correlation.

6. The control parameter collaborative optimization method based on hybrid heuristic optimization according to claim 2, characterized in that: The parameter perturbation module includes: The genetic algorithm mutation enhancement unit uses an exponential mutation probability function; The particle swarm inertia weighting unit employs a piecewise linear decay function; The firefly luciferin regulation unit employs an adaptive update formula.

7. The control parameter collaborative optimization method based on hybrid heuristic optimization according to claim 1, characterized in that: The methods for determining the second convergence condition include: A joint decision-making model with dual convergence criteria is constructed. The first convergence flag is triggered when the total number of iterations iter2 > the total minimum number of iterations Miniter2 and the change in the objective function value Δf ≤ 𝜀, where 𝜀 is the convergence threshold. The second convergence flag is triggered when the number of iterations in which the algorithm gets stuck in a local optimum exceeds a preset threshold. The dual convergence criterion joint decision-making model uses a weighted voting mechanism to comprehensively determine the termination time of the algorithm.

8. A control parameter collaborative optimization system based on hybrid heuristic optimization, characterized in that, The method for collaborative optimization of control parameters based on hybrid heuristic optimization as described in any one of claims 1-7 includes: The system comprises a hybrid heuristic optimization engine module, a winner community interaction module, a parameter adaptive adjustment module, a multi-objective decision-making module, and an algorithm base library module. Specifically: the hybrid heuristic optimization engine module connects to both the algorithm base library module and the winner community interaction module, transmitting both the algorithm's native iterative data and the winner community feature space data; the parameter adaptive adjustment module is bidirectionally connected to the hybrid heuristic optimization engine module, transmitting the objective function's rate of change data in real time and providing feedback on parameter adjustment commands; and the multi-objective decision-making module receives optimization coefficient data from all algorithm channels and performs stability verification and robustness analysis. The hybrid heuristic optimization engine module includes: Collaborative initialization unit: Connects to the algorithm base library module, initializes the parameter space of Ns basic optimization algorithms, and establishes a parallel solution channel; Native Iterative Execution Unit: Built-in genetic algorithm operator, particle swarm optimization operator and firefly swarm optimization operator, driving each algorithm to perform iterative calculations according to the native mechanism; Convergence monitoring unit: connected to each algorithm channel, it determines the iteration termination condition through the convergence condition discrimination module and the Maxiter counter module; Feature extraction unit: Generates feature space data containing the algorithm's winning clusters and extracts the current optimal solution vector for each channel; The superior community interaction module includes: Feature vector fusion unit: Receives feature space data from various algorithm channels and performs multi-dimensional feature concatenation to generate cross-algorithm related feature vectors; Diversity analysis unit: Determines the feature sampling ratio and stratification interval based on the population diversity characteristic coefficient calculation module; Cross-population embedding unit: connects to the hybrid heuristic optimization engine module to perform dynamic migration and population recombination operations of the winning feature vectors; The parameter adaptive adjustment module includes: Iterative window monitoring unit: Statistically tracks the rate of change of the objective function value of each algorithm using a sliding time window; Disturbance triggering unit: When the rate of change is lower than the dynamic adjustment threshold, the parameter nonlinear adjustment channel is activated; Algorithm regulator group: includes genetic algorithm mutation probability amplifier, particle swarm inertial weight attenuator, and firefly luciferin calibrator; The multi-objective decision-making module includes: Stability verification unit: The robustness of each optimization coefficient is tested using the Monte Carlo simulation module; Decision fusion unit: The fuzzy comprehensive evaluation method is used to screen the global optimal solution of the Pareto front solution set; Results output interface: Generates an analysis report containing the optimized parameter configuration scheme and the objective function response surface; The algorithm's basic library module stores: The core operators of genetic algorithms include selection operators, crossover operators, mutation operators, and fitness functions. Particle swarm optimization component set: including velocity update calculator, position updater, and neighborhood topology template; Firefly Algorithm Function Package: Includes a luciferin update module, a movement probability calculator, and an attraction calculation unit; The feature transfer operations performed by the cross-population embedding unit include: feature space similarity calculation based on Mahalanobis distance; population recombination strategy using quantum rotation gate mechanism; and feature hierarchical sampling method based on information entropy weight. The nonlinear adjustment channel for the parameters includes: an exponential amplification function for the mutation probability of the genetic algorithm; a piecewise decay strategy for the particle swarm inertia weight; and an adaptive calibration model for the firefly luciferin coefficient.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the steps of a control parameter collaborative optimization method based on hybrid heuristic optimization as described in any one of claims 1 to 7.

Images