A data optimization method and system based on an elephant-ant hybrid algorithm

Abstract

The application discloses a data optimization method and system based on an elephant trunk-wolf swarm hybrid algorithm, relates to the technical field of data optimization, and comprises the following steps: obtaining to-be-optimized data; constructing a hybrid optimization algorithm model based on fusion and cross of an elephant trunk search algorithm and a grey wolf algorithm; inputting the to-be-optimized data into the hybrid optimization algorithm model to obtain a data processing result after optimization. The application retains respective advantages and characteristics of the GWO and the BAS, the search agent in the hybrid algorithm is an elephant trunk, and the position updating mode of each search agent in the corresponding population in the process of searching for an optimal solution completely follows the position updating mode of the elephant trunk in the BAS, so that the search agent always moves towards a non-inferior solution in the position updating process, the appeal and the iteration step number satisfy the relationship of a Sigmoid function, the population pays attention to exploration in the early stage and development in the later stage in the optimization process, the balance between exploration and development is beneficial, and local optimization is avoided.

Description

Technical Field

[0001] This invention relates to the field of data optimization technology, and more specifically to a data optimization method and system based on a hybrid algorithm of longhorn beetle whiskers and wolf packs. Background Technology

[0002] In practical production applications, we often face numerous data optimization problems that minimize cost, risk, and time while maximizing efficiency, profit, and quality. These data optimization problems are widespread in agricultural production, mechanical design and processing, production scheduling, route planning, aerospace, water conservancy infrastructure, and other aspects of production and daily life, profoundly impacting our lives. To solve these data optimization problems, linear and nonlinear programming methods were initially developed, utilizing the functional characteristics or gradient information of the data optimization problem to find the optimal solution; these are currently the most widely used deterministic optimization methods. In contrast, nondeterministic (stochastic) optimization algorithms, due to their simplicity, ease of implementation, lack of need for gradient information during optimization, and good performance on multimodal functions, have received widespread attention and rapid development in recent decades, and are increasingly widely used in solving optimization problems. When using nondeterministic optimization algorithms to solve optimization problems, since we don't need to concern ourselves with the form of the objective function or calculate gradient information, the problem to be optimized can be treated as a black box. A deterministic input to the black box yields a deterministic output, without needing to concern ourselves with the details inside the black box, such as... Figure 1 As shown.

[0003] Among nondeterministic optimization algorithms, metaheuristic algorithms are the most prominent, relying on two key search mechanisms: exploration and exploitation. Exploration involves searching the entire feasible solution's unexplored regions globally, aiming to cover as many areas as possible. Exploration helps escape local optima. Exploration involves a detailed search of the explored regions, especially those most likely to contain the global optimum. Exploration improves the quality and accuracy of the optimization results. Exploration and exploitation are contradictory processes. Emphasizing exploration increases the probability of reaching the vicinity of the actual global optimum, but the quality and stability of the optimization results cannot be guaranteed. Emphasizing exploitation improves the quality of the optimization results, but increases the probability of getting trapped in local optima and premature convergence. Therefore, the core of metaheuristic algorithms lies in balancing exploration and exploitation to obtain or approach the actual optimum. Examples include genetic algorithms, differential evolution algorithms, gray wolf algorithms, simulated annealing algorithms, and particle swarm optimization algorithms. However, the "No free lunch" (NFL) theory states that no single metaheuristic optimization algorithm can provide the optimal solution for all types of data optimization problems; each algorithm has its own applicable scope. Some optimization algorithms perform well on unimodal functions but generally perform poorly on multimodal functions. Typical examples include the Grey Wolf Optimizer (GWO) and the Whale Optimizer. Conversely, other algorithms perform well on multimodal functions but generally perform poorly on unimodal functions. Typical examples include the Firefly Optimizer and the Beetle Antennae Search Algorithm (BAS).

[0004] Therefore, how to provide a data optimization method and system based on the beetle whisker-wolf hybrid algorithm, which integrates existing single-peak function optimization and multi-peak function optimization algorithms in the data optimization process, and improves the general data optimization capability of the optimization algorithm without violating NFL theory, is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] In view of this, this invention provides a data optimization method and system based on a beetle-antennae search algorithm-grey wolf hybrid algorithm. During data processing, it integrates existing algorithms with advantages in solving unimodal and multimodal function optimization problems, improving the general data optimization capability of the optimization algorithms without violating NFL theory. BAS (Beetle Antennae Search Algorithm) has a very simple form and approach, few parameters, and good performance in solving multimodal function optimization problems, exhibiting good global optimization results, but its local search capability is relatively poor. GWO (Grey Wolf Optimizer) has the characteristics of few parameters, simple form and approach, strong local search capability, but its global search capability is average. It can be seen that BAS and GWO complement each other and both possess the characteristics of simple form and few parameters. This invention integrates and improves the beetle-antennae search algorithm and the grey wolf algorithm, naming it the Beetle-Antennae search algorithm and Grey Wolf Optimizer hybrid optimization algorithm (BAGWO). This newly formed hybrid algorithm achieves good optimization results for both unimodal and multimodal optimization problems without violating NFL theory.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a data optimization method based on a hybrid algorithm of longhorn beetle whiskers and wolf packs, comprising: acquiring data to be optimized;

[0007] Construct a hybrid optimization algorithm model based on the fusion and crossover of the longhorn beetle whisker search algorithm and the gray wolf algorithm;

[0008] The data to be optimized is input into the hybrid optimization algorithm model to obtain the optimized data processing result.

[0009] A hybrid optimization algorithm model based on the fusion and crossover of the longhorn beetle whisker search algorithm and the gray wolf algorithm is constructed, including: defining the optimization objective function corresponding to the data to be optimized;

[0010] Initialize the parameters of the hybrid optimization algorithm model, including: assigning initial values ​​to the number of search agents, the initial whisker length of the search agents, the maximum number of iterations, the final appeal, and the initial number of local exploration steps in the parameter population of the hybrid optimization algorithm model;

[0011] The initial distribution of the population was sampled using the Latin hypercube sampling method to obtain the initial decision variables;

[0012] Longhorn beetles are used as search agents in the gray wolf algorithm, encouraging search agents in the population to move toward the historical best search agent.

[0013] The position of each search agent in the population is updated through the local development process in the longhorn beetle whisker search algorithm; and the historical best search agent is updated and recorded in real time.

[0014] The appeal and search agency need to be updated and extended;

[0015] Repeat the above steps until the maximum number of iterations is reached or other iteration convergence conditions are met.

[0016] The preferred steps of the partial development process are as follows: Initialize the orientation of the search agent;

[0017] Calculate the coordinates of the left and right whisker ends of the search agent;

[0018] Obtain the fitness functions corresponding to the left and right whisker ends of the search agent;

[0019] The new position of the search agent is calculated based on the fitness function.

[0020] Repeat the above steps until the local exploration is completed in k steps.

[0021] During local development, the search agent probes for local optima in its vicinity. BAGWO couples the number of local development steps with the number of iteration steps, making them satisfy a cosine function relationship. This is beneficial for the population to move towards the region where the global optimum is located, thus improving the stability of the optimization results.

[0022] Preferably, calculating the new location of the search agent based on fitness includes:

[0023] if

[0024]

[0025] if

[0026]

[0027] In the formula, S(X) is the sign function. This represents the fitness of the m-th search agent in the population at the i-th step of the global iteration and the (j-1)-th step of the local exploration; This indicates the position of the left whisker end of the m-th search agent. This represents the right whisker position of the m-th search agent; X is the decision variable in the optimization problem, and the upper and lower bounds of the decision variable X are Xm and Xm, respectively. u and X d ; θ represents the orientation of the initial search agent; c i This represents the length of the whiskers on one side of the search agent when the iteration step number is i.

[0028] Preferably, the appeal and the number of iterations are coupled, and the appeal is represented as:

[0029]

[0030] In the formula, ρ represents appeal, and N i N represents the current iteration step. u This represents the maximum number of iterations set by the program. g represents the shape coefficient. The larger g is, the more drastic the change in appeal ρ is, and vice versa. h represents the final appeal.

[0031] Preferably, the search agent must be represented as follows:

[0032]

[0033] In the formula, coefficients a and b are the front beard length factor and the back beard length factor, respectively; c u The initial beard length of the search agent; N s c represents the number of iterations at the switching point. s This indicates that the number of iteration steps is N. s The corresponding switching point must be long; N u This represents the maximum number of iterations set by the program, where i represents the current iteration number; coefficients a and b are the leading and trailing whisker length factors, respectively, specifically expressed as follows:

[0034] a = N u -1 ;

[0035]

[0036] Where p and q are constants.

[0037] Preferably, the number of iteration steps N at the switching point s Calculated based on the maximum number of iterations, expressed as:

[0038]

[0039] In the formula, w, s, and t are constants.

[0040] Preferably, the number of local development steps is coupled with the number of algorithm iterations, and the current number of local development steps is expressed as:

[0041]

[0042] In the formula, N i N represents the current iteration step; u k represents the maximum number of iterations set by the program. u The maximum number of exploration steps in the local area is a constant; k represents the number of local development steps corresponding to the current iteration step.

[0043] Preferably, a data optimization system based on a hybrid algorithm of longhorn beetle whiskers and wolf packs includes: a data acquisition module for acquiring data to be optimized;

[0044] The model building module is used to build a hybrid optimization algorithm model based on the fusion and crossover of the longhorn beetle whisker search algorithm and the gray wolf algorithm;

[0045] The result output module is used to input the data to be optimized into the hybrid optimization algorithm model to obtain the optimized data processing result.

[0046] As can be seen from the above technical solutions, compared with the prior art, the present invention discloses a data optimization method and system based on the beetle whisker-wolf pack hybrid algorithm, including: acquiring the data to be optimized; constructing a hybrid optimization algorithm model based on the fusion and crossover of the beetle whisker search algorithm and the gray wolf algorithm; inputting the data to be optimized into the hybrid optimization algorithm model to obtain the optimized data processing result. The present invention has the following beneficial effects: (1) The BAS algorithm and GWO are fused together to form BAGWO. In BAGWO, the search agent is replaced by the beetle instead of the gray wolf in GWO. The coordinate update method of the search agent in the development process adopts the BAS method, so that the search agent moves towards the non-dominated solution when updating the position, which improves the accuracy of optimization in the data optimization process;

[0047] (2) The α-wolf, β-wolf, and δ-wolf in GWO were unified and replaced by the globally best search agent. During the exploration process, the historical best search agent was used to call upon the search agents in the population. The degree of call was represented by the appeal power, where the appeal power and the number of iterations satisfy the relationship of the Sigmoid function.

[0048] (3) A shifting strategy was added to the beetle's beard length decay rate. When the number of iterations is small, a larger beard length decay rate is used, which gives the search agent a larger sensing distance and helps to quickly reach the region where the global optimum is located. When the number of iterations is large, a smaller beard length decay rate is used, which allows the search agent to perform fine-grained searches locally and helps to improve the accuracy and stability of the data optimization calculation results.

[0049] (4) During the local development process, the search agent probes for local optima in its vicinity. The hybrid algorithm couples the number of local development steps with the number of iteration steps, making them satisfy the cosine function relationship, which is beneficial for the population to move towards the region where the global optimum is located, and improves the stability of the data optimization calculation results. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0051] Figure 1 A schematic diagram of a black box model for solving optimization problems in the nondeterministic algorithm proposed in the background technology.

[0052] Figure 2 This invention provides a schematic flowchart of a data optimization method based on a hybrid algorithm of longhorn beetle whiskers and wolf packs.

[0053] Figure 3(a) is a schematic diagram of the global exploration process for updating the search agent location in an embodiment of the present invention.

[0054] Figure 3(b) is a schematic diagram of a partial development process for updating the search agent location according to an embodiment of the present invention.

[0055] Figure 4(a) is a schematic diagram showing the change of appeal with the number of iterations in an embodiment of the present invention.

[0056] Figure 4(b) is a schematic diagram showing the variation of the fiber length decay rate with the number of iteration steps in an embodiment of the present invention.

[0057] Figure 4(c) is a schematic diagram showing the variation of the number of local development steps with the number of iteration steps in an embodiment of the present invention.

[0058] Figure 5(a) is a schematic diagram of the initial distribution of different types of populations provided in the embodiments of the present invention.

[0059] Figure 5(b) is a schematic diagram of the initial population distribution after Latin hypercube sampling in an embodiment of the present invention.

[0060] Figure 6 The flowchart for calculating BAGWO is provided for an embodiment of the present invention.

[0061] Figure 7(a) is a schematic diagram of an optimization case of tension / compression spring design problem in an embodiment of the present invention.

[0062] Figure 7(b) is a schematic diagram of an optimization case of pressure tank design problem according to an embodiment of the present invention.

[0063] Figure 7(c) is a schematic diagram of an optimization case of the welded beam design problem in an embodiment of the present invention.

[0064] Figure 7(d) is a schematic diagram of an optimization case of the reducer design problem in an embodiment of the present invention.

[0065] Figure 7(e) is a schematic diagram of an optimization case of the triangular truss design problem according to an embodiment of the present invention.

[0066] Figure 7(f) is a schematic diagram of an optimization case of cantilever beam design problem according to an embodiment of the present invention.

[0067] Figure 7(g) is a schematic diagram of an optimization case of gear transmission design problem in an embodiment of the present invention.

[0068] Figure 7(h) is a schematic diagram of an optimization case of the stepped conical pulley problem according to an embodiment of the present invention. Detailed Implementation

[0069] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0070] Optimization refers to the act of finding the optimal solution from a decision space given certain constraints and objectives (single or multiple objectives). Optimization problems can be categorized into single-objective optimization and multi-objective optimization based on the number of objectives. Solving multi-objective optimization problems is often more complex than solving single-objective optimization problems. However, multi-objective optimization problems can be transformed into single-objective optimization problems using methods such as constraint methods, weighted sum methods, and objective programming. This invention focuses on algorithms for solving single-objective optimization problems. Optimization problems can also be categorized into constrained optimization problems and unconstrained optimization problems based on the presence or absence of constraints. The most common way to solve constrained optimization problems is to transform them into unconstrained optimization problems using penalty function methods. Finally, optimization problems can be categorized into linear optimization problems and nonlinear optimization problems based on the properties of the constraints and objective functions. In nonlinear optimization problems, the relationship between the constraints or objective function and the decision variables is nonlinear. Solving nonlinear optimization problems is often much more difficult than solving linear optimization problems, and many optimization problems encountered in real life are also nonlinear optimization problems. Optimization problems can be categorized into unimodal and multimodal functions based on the number of extrema within their feasible region. A unimodal function has only one global extremum within the feasible region, which is generally the optimal solution sought. A multimodal function has multiple extrema within the feasible region, and solving for the optimal value of a multimodal function can easily lead to local optima. Many continuous optimization problems in the real world have multimodal objective functions. There are also many other classifications of optimization problems; identifying and classifying optimization problems helps in selecting appropriate optimization methods to solve them.

[0071] Metaheuristic algorithms rely on two key search mechanisms during optimization: exploration and exploitation. Exploration involves searching the entire feasible solution's unexplored regions globally, aiming to cover as many areas as possible. Exploration helps avoid local optima. Exploration involves a detailed search of the explored regions, especially those most likely to contain the global optimum. Exploration improves the quality and accuracy of the optimization results. Exploration and exploitation are two contradictory search processes. Emphasizing exploration increases the probability of reaching the vicinity of the actual global optimum, but the quality and stability of the optimization results cannot be guaranteed. Emphasizing exploitation improves the quality of the optimization results, but increases the probability of getting trapped in local optima and premature convergence. This invention constructs a hybrid optimization algorithm model based on the fusion of the Bullfly Whiskers search algorithm and the Grey Wolf algorithm. This algorithm is designed to solve single-objective optimization problems with excellent comprehensive optimization capabilities. It is less prone to getting trapped in local optima, possesses good local and global search capabilities, and demonstrates good optimization performance for both unimodal and multimodal functions. It performs well in practical engineering optimization problems.

[0072] This invention discloses a data optimization method based on a hybrid algorithm of longhorn beetle whiskers and wolf packs, such as... Figure 2 As shown, this includes: obtaining the data to be optimized;

[0073] A hybrid optimization algorithm model based on the fusion and cross-engineering of the longhorn beetle whisker search algorithm and the gray wolf algorithm is constructed; the data to be optimized is input into the hybrid optimization algorithm model to obtain the optimized data processing result.

[0074] Specifically, a hybrid optimization algorithm model based on the fusion and crossover of the longhorn beetle whisker search algorithm and the gray wolf algorithm is constructed, including: defining the optimization objective function corresponding to the data to be optimized;

[0075] Initialize the parameters of the hybrid optimization algorithm model, including: assigning initial values ​​to the number of search agents, the initial whisker length of the search agents, the maximum number of iterations, the final appeal, and the initial number of local exploration steps in the parameter population of the hybrid optimization algorithm model;

[0076] The initial distribution of the population was sampled using the Latin hypercube sampling method to obtain the initial decision variables;

[0077] Longhorn beetles are used as search agents in the gray wolf algorithm, encouraging search agents in the population to move toward the historical best search agent.

[0078] The position of each search agent in the population is updated through the local development process in the longhorn beetle whisker search algorithm; and the historical best search agent is updated and recorded in real time.

[0079] The appeal and search agency need to be updated and extended;

[0080] Repeat the above steps until the maximum number of iterations is reached or other iteration convergence conditions are met.

[0081] Specifically, the embodiments of the present invention can be applied to data optimization problems that face a large number of factors, such as minimizing costs, risks and time, and maximizing efficiency, profit and quality. These data optimization problems are widely found in application fields such as agricultural production, mechanical design and processing, production scheduling, route planning, aerospace, and water conservancy infrastructure.

[0082] This invention integrates BAS and GWO to form a hybrid algorithm, named the Beetle Antennae search algorithm and Grey Wolf Optimizer hybrid optimization algorithm (BAGWO). BAGWO retains the advantages of both GWO and BAS, and makes several improvements to the exploration and development strategies. Figure 3(a) shows a schematic diagram of the BAGWO principle, where the search agent is replaced by a longhorn beetle instead of a grey wolf. The position update method of each search agent (i.e., the longhorn beetle) in the corresponding population during the search for the optimal solution completely follows the position update method of the longhorn beetle in BAS. This method ensures that the search agent always moves towards a non-dominated solution during the position update process. Unlike GWO, in BAGWO, each search agent is an equal individual with no social hierarchy. When updating the position of a search agent, it uniformly moves towards the position of the historically best search agent (HBSA), which is the search agent corresponding to the best fitness since the start of the optimization calculation. The historically best search agent is updated and recorded in real time during the optimization process.

[0083] Figure 3(b) illustrates the local exploration process of a single search agent in the BAGWO population. In the figure, d represents the step size of the search agent's movement, and k represents the number of steps in the local exploration. It can be seen that during local exploration, the search agent can only explore within a radius of kd centered on the starting point. Each search agent in the population performs this local exploration process. After all search agents in the population have completed their local exploration processes, the HBSA is updated and recorded, updating the current global optimum. The global optimum position at iteration step i is represented as... The corresponding fitness is The position of the search agent in the population after completing the partial development process is: The corresponding fitness is Then the global optimal position The update formula is shown in formula (1).

[0084]

[0085] In BAGWO, not only are the characteristics of BAS and GWO combined and integrated, but the concept of appeal is also newly introduced. Furthermore, various studies and improvements have been made to the patterns of appeal variation, beard length variation, local development step variation, and initial population distribution. Specific measures are as follows:

[0086] Specifically, appeal and its improvement: Appeal is a concept extracted from GWO, representing the leadership or influence of alpha, beta, and delta wolves on other gray wolves. Unlike GWO, the appeal mentioned in this embodiment represents the degree to which the HBSA (Hollow Headed Aspect) summons all search agents in the current population at a given iteration step. Appeal is a real number between 0 and 1. The closer the appeal is to 1, the greater the summoning of search agents, and the more they rush towards the HBSA position, especially when the appeal equals 1, the search agent rushes to the HBSA position at the current iteration step. Conversely, the closer the appeal is to 0, the smaller the summoning of search agents, and the closer the search agent is to its original position, especially when the appeal equals 0, the search agent remains in its original position. For swarm intelligence optimization algorithms, how to balance exploration and development is a crucial concern. Exploration represents the algorithm's global search capability, exploring unexplored areas, while development represents the algorithm's local search capability, involving detailed development of explored areas. Actual calculations show that focusing on exploration in the early stages and development in the later stages is beneficial for improving the algorithm's optimization ability and finding the actual optimal solution. Therefore, it is necessary to couple the appeal force with the number of iterations: a smaller appeal force is used when the number of iterations is small, and a larger appeal force is used when the number of iterations is large, up to a value equal to 1. Mathematically, the Sigmoid function is called a growth curve, with an S-shaped trend. It is often used as an activation function in neural networks and is well-suited for representing the appeal force ρ and the number of iterations N. i The relationship between the two is directly given here as a functional relationship, as shown in formula (2). The changing trend of appeal is shown in Figure 4(a).

[0087] Specifically, the appeal and the number of iterations are coupled, and the appeal is represented as:

[0088]

[0089] In the formula, ρ represents appeal, and N i N represents the current iteration step. uThis represents the maximum number of iterations set by the program. `g` represents the shape coefficient; the larger `g` is, the more drastic the change in appeal `ρ`, and vice versa. `h` represents the final appeal, a parameter that determines the final level of appeal. In this embodiment of the invention, the shape coefficient `g` is defaulted to 100, and the final appeal `h` is around 1, typically 0.99.

[0090] Specifically, the shifting strategy for the decay rate of the search agent's whisker length: The whisker length of the search agent represents its detection and sensing distance during the optimization process. To improve the accuracy of the optimization solution, the whisker length of the search agent needs to maintain a decreasing trend during the iterative optimization process. There are many ways to update the whisker length, among which the commonly used update formulas are shown in formulas (3) and (4), where c i σ represents the length of the beetle's whiskers on one side when the number of iterations is i, σ represents the decay rate of the beetle's whisker length, and c d This represents the minimum length of the longhorn beetle's whiskers. In some cases, c... d When the value is 0, formula (3) can be simplified to formula (4), where c u This represents the initial beetle beard length. Benchmark tests revealed that updating the beard length using the methods shown in formulas (3) and (4) has certain limitations. Some benchmark functions perform well when the beard length decay rate σ is large, while others perform well when the beard length decay rate σ is small.

[0091] c i+1 =σ·c i +c d #(3)

[0092] c i+1 =c u ·σ i #(4)

[0093] To address the matching problem of the beard length attenuation rate σ, the method shown in Figure 4(b) was adopted, at a certain iteration step N. s Previously, a smaller fiber length attenuation rate was used, N s Then, a larger beard length decay rate is adopted, and the number of iterations is N. s The corresponding beard length is the switching point beard length, denoted by c. s In actual parameter settings, since the solution accuracy is often related to the final beard length, people are more sensitive to the value of the beard length than the beard length decay rate σ. Therefore, it is more intuitive to express the beard length update formula as a function of the final beard length and the number of iteration steps, as shown in formula (5), where c s It can be calculated using formula (6). In formula (5), the coefficients a and b are the front beard length factor and the back beard length factor, respectively, which can be calculated using formulas (7) and (8). N sParameter selection and maximum number of iterations N u There is a relationship between them, when the maximum number of iterations N u When N is small, in order to ensure a sufficient number of exploration steps and avoid getting trapped in local optima, N s A larger value should be chosen, otherwise the maximum number of iterations N will be determined. u When N is large, the number of space exploration steps can be satisfied. s A relatively small value should be chosen to ensure sufficient space is developed. Formula (9) is N s and N u The empirical formula is given, where square brackets in the formula represent rounding up to the nearest integer.

[0094] Specifically, the search agent must be represented as follows:

[0095]

[0096]

[0097] In the formula, c u The initial beard length of the search agent; N s c represents the number of iterations at the switching point. s This indicates that the number of iteration steps is N. s The corresponding switching point must be long; N u This indicates the maximum number of iterations set by the program, where i represents the current iteration number.

[0098] Coefficients a and b are the front beard length factor and the back beard length factor, respectively, and are expressed as:

[0099] a = n u -1 #(7)

[0100]

[0101] Where p and q are constants, p = -0.7928 and q = -0.5031.

[0102] Specifically, the number of iteration steps n at the switching point s Calculated based on the maximum number of iterations, expressed as:

[0103]

[0104] Where w, s, and t are constants, w = 2, s = -0.6342, and t = 0.1775.

[0105] Specifically, the variation law of the number of local development steps of the search agent: In order to enable the algorithm to focus on exploration in the early stage, reach the vicinity of the actual optimal solution faster and better, and reduce the running cost of the algorithm to a certain extent, another useful improvement is to couple the number of local development steps with the number of algorithm iterations, so that the function variation law of the two conforms to the form of a cosine function, as shown in Figure 4(c). The specific functional relationship is shown in formula (10), where N in the formula i N represents the current iteration step. u k represents the maximum number of iterations set by the program. u The maximum number of exploration steps is represented by a constant, and k represents the number of local development steps corresponding to the current iteration step. The square brackets in formula (10) represent rounding up to the nearest integer. Therefore, the variation of the number of local development steps with the number of iteration steps is actually shown by the horizontal lines in Figure 4(c), and the number of horizontal lines is equal to k. u .

[0106] Specifically, the number of local development steps is coupled with the number of algorithm iterations. The current number of local development steps is represented as:

[0107]

[0108] In the formula, N i N represents the current iteration step; u k represents the maximum number of iterations set by the program. u The maximum number of exploration steps in the local area is a constant; k represents the number of local development steps corresponding to the current iteration step.

[0109] Specifically, the impact of initial population distribution: Tests show that BAGWO is sensitive to the initial population distribution, which can affect its optimization performance to some extent. When the initial population distribution is relatively uniform within the search region of the problem to be optimized, this effect is almost negligible. Figure 5(a) shows the possible initial population distribution when the decision variables of the problem to be optimized have a 2-dimensional dimension:

[0110] (1) The initial distribution of the population is concentrated in a certain place;

[0111] (2) The initial population distribution clustered in several locations;

[0112] (3) The initial distribution of the population is relatively uniform (this is the expected distribution).

[0113] To ensure a relatively uniform initial population distribution and prevent clustering in one or a few locations, the Latin hypercube sampling (LHS) method was introduced. First proposed by McKay et al. in 1979, LHS possesses the characteristic of uniform stratification, guaranteeing a relatively uniform distribution and covering the decision variable space as much as possible. Figure 5(b) illustrates BAGWO's use of LHS during population initialization, minimizing the problem of population clustering.

[0114] In one specific embodiment of the present invention, initializing the parameters of the hybrid optimization algorithm model includes: assigning initial values ​​to the number of search agents, the initial whisker length of the search agents, the maximum number of iteration steps, the final appeal, and the initial number of local exploration steps in the parameter population of the hybrid optimization algorithm model.

[0115] Among them, (1) the number of search agents B in the population: Generally speaking, the more search agents there are, the better the optimization performance of the algorithm, but the problem is that the time consumption increases. Considering the balance between the effectiveness of solving the optimization problem and the time consumption, the value of the number of search agents B is generally recommended to be between [5,50], and the value can generally be 30;

[0116] (2) The initial length of the search agent must be c. u This length is a relative length, with a value space of (0,1]. The larger the value, the larger the initial exploration space.

[0117] (3) Maximum number of iterations N u : The maximum number of iterations the algorithm can run. If this number is exceeded, the algorithm will stop running and output the calculation results.

[0118] (4) Final appeal h: When the number of iterations reaches the maximum number of iterations N u The appeal of a time is generally taken as 0.99;

[0119] (5) Initial local exploration steps k u : The number of exploration steps for each search agent during local exploration. A smaller parameter results in faster algorithm optimization, but with a corresponding decrease in optimization performance; conversely, a larger parameter leads to better optimization performance, but slower optimization speed. Considering the balance between the effectiveness and time consumption of the optimization problem solution, the initial local exploration steps k are... u The value of is generally recommended to be between [2, 20].

[0120] In practical calculations, the selection of algorithm parameters varies depending on the requirements of the task to be optimized. For tasks that are not sensitive to computation time, high-configuration parameters can be selected, in which case the optimization performance of BAGWO can be sufficiently released. For optimization tasks that are sensitive to computation time, low-configuration parameters can be selected. In this case, the optimization performance of the BAGWO algorithm is somewhat constrained, but there is still a good optimization effect. The specific parameter settings for different situations are shown in Table 1.

[0121] Table 1 Algorithm Parameter Settings

[0122]

[0123]

[0124] Specifically, the steps in the partial development process are as follows: Initialize the orientation of the search agent;

[0125] Calculate the coordinates of the left and right whisker ends of the search agent;

[0126] Obtain the fitness functions corresponding to the left and right whisker ends of the search agent;

[0127] The new position of the search agent is calculated based on the fitness function.

[0128] Specifically, the embodiments of the present invention also disclose the calculation steps of the hybrid optimization algorithm, as follows:

[0129] (1) Define the objective function f(X) to be solved, where X is the decision variable of the optimization problem, which is an n-dimensional vector, and the upper and lower bounds of the decision variable X are respectively Xi and Xj. u and X d ;

[0130] (2) Initialize the algorithm parameters, including the number of search agents B in the population and the initial length of the search agents c. u Maximum number of iterations N u Final Appeal h, Local Initial Exploration Steps k u Assign initial values;

[0131] (3) The initial distribution of the population was sampled using the Latin hypercube sampling method to obtain the initial decision variable value X0;

[0132] (4) For each search agent in the population, its position is updated by searching for longhorn beetles in BAS. For any search agent in the population, the steps of the local development process are as follows:

[0133] a) Randomly initialize the orientation of the search agent, represent this orientation with an n-dimensional vector, and normalize it;

[0134]

[0135] In the formula: r is the generated random n-dimensional vector; θ is the result of normalization; and the norm is the Euclidean norm.

[0136] b) Calculate the left whisker terminator of the search agent. and right whisker tip The coordinates of the location

[0137]

[0138] In the formula: X i The search agent center location is indicated by the subscript i, which represents the current iteration step, j, which represents the number of steps in the local exploration, and m, which represents the search agent number in the population.

[0139] c) The fitness corresponding to the tips of the left and right antennae of the longhorn beetle can then be calculated.

[0140] d) Calculate the new location of the search agent based on fitness:

[0141] if:

[0142]

[0143] if:

[0144]

[0145] In the formula: S(X) is the sign function, This represents the fitness of the m-th search agent in the population at the i-th step of the global iteration and the (j-1)-th step of the local exploration; This indicates the position of the left whisker end of the m-th search agent. Indicates the position of the right whisker end of the m-th search agent; θ represents the orientation of the initialized search agent; c i This represents the length of the whiskers on one side of the search agent when the iteration step number is i.

[0146] e) Repeat steps a) to d) until the local exploration of k steps is completed;

[0147] (5) Update the position of HBSA according to formula (1). and fitness In the subscript, i represents the current iteration step;

[0148] (6) Call on the search agents in the population to move towards HBSA. The formula for moving any search agent in the population is as follows:

[0149]

[0150] (7) Update the appeal according to formula (2);

[0151] (8) Update the whisker length of the search agent according to formula (5);

[0152] (9) Repeat the above process (4)-(8) until the maximum number of iterations is reached or other iteration convergence conditions are met.

[0153] This invention also provides pseudocode for a hybrid optimization algorithm (as shown in Table 2) and a computation flowchart (as shown in Table 2). Figure 6 (As shown). The algorithm can be implemented using different programming languages ​​based on the formulas provided in this scheme, and applied to the optimization and solution of practical problems.

[0154] Table 2. Pseudocode for the BAGWO algorithm

[0155]

[0156]

[0157] Specifically, this invention also discloses a data optimization system based on a hybrid algorithm of longhorn beetle whiskers and wolf packs, including: a data acquisition module for acquiring data to be optimized;

[0158] The model building module is used to build a hybrid optimization algorithm model based on the fusion and crossover of the longhorn beetle whisker search algorithm and the gray wolf algorithm;

[0159] The result output module is used to input the data to be optimized into the hybrid optimization algorithm model to obtain the optimized data processing result.

[0160] The embodiments of this invention have been validated and statistically analyzed through benchmarking and multiple real-world engineering optimization cases. The proposed hybrid optimization algorithm (BAGWO) demonstrates superior comprehensive optimization performance in single-objective optimization problems, with its stability and accuracy ranking among the best in its class. BAGWO is effective for most unimodal and multimodal optimization problems, and its general applicability is also among the best, making it particularly suitable for solving common single-objective optimization problems. Furthermore, BAGWO's relative optimization performance remains relatively stable as the dimensionality of the optimization problem changes. For time-sensitive optimization problems, adjusting the value of the local development step number k can reduce time consumption while maintaining good optimization results; while for time-insensitive problems, choosing a larger k value helps to obtain better results. These advantages are mainly attributed to the following reasons:

[0161] (1) The BAS algorithm and GWO were combined to form BAGWO. In BAGWO, the search agent was replaced by a longhorn beetle instead of a gray wolf in GWO. The coordinate update method of the search agent during the development process adopts the BAS method, so that the search agent always moves towards the non-dominated solution when updating its position.

[0162] (2) The α-wolf, β-wolf, and δ-wolf in GWO were unified and replaced by HBSA. The coordinates and fitness of HBSA were updated and recorded in real time during the optimization iteration process. During exploration, HBSA was used to call upon the search agents in the population, and the degree of call was represented by appeal strength. The appeal strength and the number of iterations satisfy the relationship of the Sigmoid function. This ensures that the population focuses on exploration in the early stages and development in the later stages of optimization, which is conducive to balancing exploration and development and avoids getting trapped in local optima.

[0163] (3) A shifting strategy was added to the beard length decay rate of the search agent. When the number of iteration steps is small, a larger beard length decay rate is used, which gives the search agent a larger perception distance and helps it quickly reach the region where the global optimum is located. When the number of iteration steps is large, a smaller beard length decay rate is used, which allows the search agent to perform a fine-grained search locally, which helps to improve the accuracy and stability of the optimization calculation results.

[0164] (4) During the local development process, the search agent probes for local optima in the vicinity of the local area. In BAGWO, the number of local development steps is coupled with the number of iteration steps, so that they satisfy the relationship of the cosine function, which is conducive to the population moving towards the region where the global optimum is located and improves the stability of the optimization calculation results;

[0165] (5) The initial position of the population is generated by the Latin hypercube sampling method, which is conducive to making the initial distribution of the population as uniform as possible and improving the stability of the optimization calculation results.

[0166] In one specific embodiment of the present invention, the optimization effect of the proposed BAGWO algorithm in the field of engineering optimization design is compared and verified. This embodiment selects eight commonly used engineering optimization design cases from the literature, and the schematic diagrams of the eight engineering optimization cases are shown below. Figures 7(a)-7(h) As shown, these problems are applied to the design problems of tension / compression springs, pressure tanks, welded beams, reducers, triangular trusses, cantilever beams, gear transmissions, and stepped conical pulleys. The characteristics of these eight engineering optimization problems are summarized in Table 3. BAGWO engineering optimization case studies include:

[0167] Table 3 Characteristics of Eight Engineering Optimization Problems

[0168]

[0169] The above eight engineering optimization problems are all constrained optimization problems. When solving constrained optimization problems, it is necessary to process the constraints. According to the research on constraint processing methods in evolutionary computation, constraint processing methods are mainly divided into: penalty function method, feasible rule method and multi-objective method. Due to its simple principle and ease of implementation, the penalty function method is one of the widely used constraint processing methods. The penalty function method adds a penalty function to the objective function to transform the constrained problem into an unconstrained problem. Equation (16) shows a common penalty function processing method, G i (X) is an inequality constraint, H j (X) represents equality constraints, p is the number of inequality constraints, q is the number of equality constraints, and a i and b j η is a positive constant, λ is the penalty function coefficient, and η and λ are equal to 1 or 2. For this penalty method shown in formula (16), the objective function value increases when a candidate solution violates any constraint, thus causing it to be discarded during the optimization process.

[0170]

[0171] st

[0172] G i (X)=max(0,g i (X)) η

[0173] H j (X)=|h j (X)| λ

[0174] In the parameter settings of all optimization algorithms used in the engineering optimization test, the maximum number of iterations was 500 and the population size was 50. Furthermore, all other parameter settings for the optimization algorithms used in the comparison were their default values.

[0175] Specifically, (a) the design problem of tension / compression springs (TCSD)

[0176] In the design problem of a tension / compression spring, the optimization objective is to minimize the weight of the spring while satisfying the constraints, as shown in Figure 7(a). This problem involves three decision variables: d, D, and P. The mathematical description of this optimization problem is as follows:

[0177] Optimization variables: x = [x1, x2, x3] = [d, D, P];

[0178] Optimize the objective function:

[0179] Constraints:

[0180]

[0181] Optimize the variable range as follows: 0.05≤x1≤2; 0.25≤x2≤1.3; 2≤x3≤15;

[0182] (b) Pressure Vessel Design Problems (PVD)

[0183] The pressure vessel design problem was proposed by Kannan et al. in 1994. The optimization objective is to minimize the manufacturing cost of the pressure vessel while satisfying constraints, as shown in Figure 7(b). This problem involves four decision variables: T s T h R, L. The mathematical description of this optimization problem is as follows:

[0184] Optimization variables: x = [x1, x2, x3, x4] = [T s ,T h [,R,L];

[0185] Optimize the objective function:

[0186] Constraint: g1(x) = -x1 + 0.0193x3 ≤ 0;

[0187] g2(x) = -x2 + 0.00954x3 ≤ 0;

[0188]

[0189] g4(x) = x4 - 240 ≤ 0;

[0190] Optimize variable range: 0 ≤ x i ≤99, i=1,2; 10≤x i ≤200, i=3,4;

[0191] (c) Welded Beam Design Problems (WBD)

[0192] The goal of the welded beam design problem is to minimize the manufacturing cost of the welded beam while satisfying constraints. Its structural schematic is shown in Figure 7(c). This problem involves four decision variables: h, l, t, and b; the constraints include τ, σ, and P. c , δ, etc. The mathematical description of this optimization problem is as follows:

[0193] Optimization variables: x = [x1, x2, x3, ..., x4] = [h, l, t, b];

[0194] Optimize the objective function:

[0195] Constraint: g1(x) = τ(x) - τ max ≤0,

[0196] g2(x)=σ(x)-σ max ≤0,

[0197] g3(x)=δ(x)-δ max ≤0,

[0198] g4(x) = x1 - x4 ≤ 0,

[0199] g5(x)=PP e (x)≤0,

[0200] g6(x) = 0.125 - x1 ≤ 0,

[0201]

[0202] Optimize variable range: 0.1 ≤ x i ≤2, i=1,4; 0.1≤x i ≤10, i=2,3;

[0203] Other variables:

[0204]

[0205] P = 6000, L = 14, δ max =0.25, E = 30 × 10 6 G = 12 × 10 6 ,

[0206] τ max =13600,σ max =30000;

[0207] (d) Reducer Design Issues (SRD)

[0208] The goal of the reducer design problem is to minimize the weight of the reducer while satisfying the constraints. Its structural diagram is shown in Figure 7(d). This problem involves seven decision variables: b, m, z, l1, l2, d1, and d2. The mathematical description of this optimization problem is as follows:

[0209] Optimization variables: x = [x1, x2, x3, x4, x5, x6, x7] = [b, m, z, l1, l2, d1, d1];

[0210] Optimize the objective function:

[0211] Constraints:

[0212]

[0213]

[0214] Optimize the variable range as follows: 2.6≤x1≤3.6, 0.7≤x2≤0.8, 17≤x3≤28, 7.3≤x4≤8.3, 7.3≤x5≤8.3, 2.9≤x6≤3.9, 5.0≤x7≤5.5;

[0215] (e) Triangular Truss Design Problem (TTD)

[0216] The objective of the triangular truss design problem is to minimize the total weight of the triangular truss while satisfying the constraints, as shown in Figure 7(e). This problem involves two decision variables: A1 and A2. The mathematical description of this optimization problem is as follows:

[0217] Optimization variables: x = [x1, x2] = [A1, A2];

[0218] Optimize the objective function:

[0219] Constraints:

[0220]

[0221] Optimize variable range: 0 ≤ x i ≤99, i=1,2;

[0222] Other variables: l = 100; P = 2; σ = 2;

[0223] (f) Cantilever beam design problems (CBD)

[0224] The objective of the cantilever beam design problem is to minimize the weight of a cantilever beam with five hollow blocks, as shown in Figure 7(f). This problem involves five decision variables, corresponding to different cantilever beam side lengths, denoted by x1, x2, x3, x4, and x5, from the fixed end to the suspended end. The mathematical description of this optimization problem is as follows:

[0225] Optimization variables: x = [x1, x2, x3, x4, x5];

[0226] The objective function to be optimized is: f(x) = 0.0624(x1+x2+x3+x4+x5);

[0227] Constraints:

[0228]

[0229] Optimize variable range: 0.01 ≤ x i ≤100, i=1,2,3,4,5;

[0230] (g) Gear Drive Design Problems (GTD)

[0231] The goal of the gear transmission design problem is to determine a suitable or optimal number of gear teeth to optimize the gear ratio, while satisfying constraints (here, the constraint is that the number of gear teeth is a positive integer), as shown in Figure 7(g). This problem involves four decision variables: T a ,T b ,T c ,T d The mathematical description of this optimization problem is as follows:

[0232] Optimization variables: x = [x1, x2, x3, x4] = [T a ,T b ,T d ,T f ];

[0233] Optimize the objective function:

[0234] Constraints: x i ∈N + i = 1, 2, 3, 4;

[0235] Optimize variable range: 12≤x i ≤60, i=1,2,3,4;

[0236] (h) Stepping Staircase Pulley Problem (SPD)

[0237] A step-step pulley is a series of pulleys forming a stepped cone. It is used in pairs to change the speed ratio of the shafts. The pulleys are used to transfer power from one shaft to another shaft at a certain distance via a belt or rope running on them. The objective of the triangular truss design problem is to minimize the total weight of the triangular truss while satisfying constraints, as shown in Figure 7(h). This problem involves five decision variables: w, d1, d2, d3, and d4. The mathematical description of this optimization problem is as follows:

[0238] Optimization variables: x = [x1, x2, x3, x4, x5] = [w, d1, d2, d3, d4];

[0239] Optimize the objective function:

[0240] Constraint: h1(x) = C1 - C2 = 0;

[0241] h2(x) = C1 - C3 = 0;

[0242] h3(x) = C1 - C4 = 0;

[0243] g i＝1,2,3,4 (x)(=-R i ≤2;

[0244] g i＝5,6,7,8 (x)=(0.75×745.6998)-P i-1 ≤0;

[0245] Optimize variable range: 0 ≤ x i ≤60, i=1,2; 0≤x i ≤90, i=3,4,5;

[0246] Other variables:

[0247]

[0248]

[0249] t=8, s=1.75, μ=0.35, ρ=7200, a=3.

[0250] Engineering Optimization Design Case Calculation Results: In this embodiment of the invention, the results of eight engineering optimization design cases, including 14 algorithms such as the BAGWO algorithm, are presented in tables. Table 4 shows the optimization calculation results for the TCSD optimization problem. The data in the table shows that BAGWO's optimization result is the best among the 14 algorithms compared. Similarly, Table 5 shows the optimization calculation results for the PVD optimization problem, Table 6 shows the optimization calculation results for the WBD optimization problem, Table 7 shows the optimization calculation results for the SRD optimization problem, Table 8 shows the optimization calculation results for the TTD optimization problem, Table 9 shows the optimization calculation results for the CBD optimization problem, Table 10 shows the optimization calculation results for the GTD optimization problem, and Table 11 shows the optimization calculation results for the SPD optimization problem. In Tables 4-11, the optimization values ​​with the best optimization effect are marked in bold black. Table 12 ranks the optimization effects of 14 optimization algorithms in eight engineering optimization cases. The ranking results show that BAGWO ranks first among the 14 optimization algorithms with an average ranking of 2.38, which verifies the effectiveness of BAGWO in optimizing practical engineering problems and demonstrates its outstanding competitiveness.

[0251] Table 4 Comparison of optimization results of different optimization algorithms for the TCSD problem

[0252]

[0253] Table 5 Comparison of optimization results of different optimization algorithms for the PVD problem

[0254]

[0255]

[0256] Table 6 Comparison of optimization results of different optimization algorithms for the WBD problem

[0257]

[0258] Table 7 Comparison of optimization results of different optimization algorithms for the SRD problem

[0259]

[0260]

[0261] Table 8 Comparison of optimization results of different optimization algorithms for the TTD problem

[0262]

[0263] Table 9 Comparison of optimization results of different optimization algorithms for the CBD problem.

[0264]

[0265]

[0266] Table 10 Comparison of optimization results of different optimization algorithms for the GTD problem

[0267]

[0268] Table 11 Comparison of optimization results of different optimization algorithms for the SPD problem

[0269]

[0270]

[0271] Table 12 Ranking of optimization performance of 14 participating algorithms in 8 engineering optimization problems

[0272]

[0273] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0274] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A data optimization method based on a hybrid algorithm of longhorn beetle whiskers and wolf packs, characterized in that, Applications include: (The text abruptly ends here, so the translation also ends here.) Obtain the data to be optimized; A hybrid optimization algorithm model based on the fusion and cross-engineering of the longhorn beetle whisker search algorithm and the gray wolf algorithm is constructed; the data to be optimized is input into the hybrid optimization algorithm model to obtain the optimized data processing result; In tension / compression spring design, the optimization objective is to minimize the weight of the spring while satisfying constraints. The optimization variables are: ; Optimize the objective function: ; Constraints: ; ； ； ； Optimize variable range: ; A hybrid optimization algorithm model based on the fusion and crossover of the longhorn beetle whisker search algorithm and the gray wolf algorithm is constructed, including: Define the objective function for the data to be optimized; Initialize the parameters of the hybrid optimization algorithm model, including: assigning initial values ​​to the number of search agents, the initial whisker length of the search agents, the maximum number of iterations, the final appeal, and the initial number of local exploration steps in the parameter population of the hybrid optimization algorithm model; The initial distribution of the population is sampled to obtain the initial decision variables; Longhorn beetles are used as search agents in the gray wolf algorithm, encouraging search agents in the population to move toward the historical best search agent. The position of each search agent in the population is updated through the local development process in the longhorn beetle whisker search algorithm; and the historical best search agent is updated and recorded in real time. The appeal and search agency need to be updated and extended; Repeat the above steps until the maximum number of iterations is reached or other iteration convergence conditions are met. The length of the search agent must be represented as follows: ； ； In the formula, The initial length of the search agent's whiskers; This indicates the number of iterations at the switching point; Indicates the number of iteration steps. The corresponding switching point must be long; This indicates the maximum number of iterations set by the program. Represents the current iteration step; coefficient and These are the front beard length factor and the back beard length factor, respectively, specifically expressed as follows: ； ； Where p and q are constants.

2. The data optimization method based on the beetle whisker-wolf hybrid algorithm according to claim 1, characterized in that, The steps of the partial development process are as follows: Initialize the orientation of the search agent; Calculate the coordinates of the left and right whisker ends of the search agent; Obtain the fitness functions corresponding to the left and right whisker ends of the search agent; The new position of the search agent is calculated based on the fitness function.

3. The data optimization method based on the longhorn beetle whisker-wolf pack hybrid algorithm according to claim 2, characterized in that, The new location of the search agent is calculated based on fitness, including: if : if : In the formula, For symbolic functions, The first in the representative population The search agent in the global iteration Step 1, Local Exploration Step adaptability; Indicates the first The left whisker end position of a search agent Indicates the first The right-hand end position of a search agent; These are the decision variables in an optimization problem. The corresponding upper and lower bounds are respectively and ; Indicates the orientation of the initial search agent; Indicates the number of iteration steps. The search agent's whisker length on one side.

4. The data optimization method based on the beetle whisker-wolf hybrid algorithm according to claim 1, characterized in that, The appeal is coupled with the number of iteration steps, where the appeal is represented as: In the formula, To demonstrate appeal, Indicates the current iteration step. This indicates the maximum number of iterations set by the program. Represents the shape factor. It represents ultimate appeal.

5. The data optimization method based on the beetle whisker-wolf hybrid algorithm according to claim 1, characterized in that, Number of iterations at the switching point Calculated based on the maximum number of iterations, expressed as: In the formula, w 、 s 、 t is a constant.

6. The data optimization method based on the beetle whisker-wolf hybrid algorithm according to claim 1, characterized in that, By coupling the number of local development steps with the number of algorithm iterations, the current number of local development steps is represented as: In the formula, Represents the current iteration step; This represents the maximum number of iterations set by the program. Representing the local maximum number of exploration steps, it is a constant; This represents the number of local development steps corresponding to the current iteration step.

7. A data optimization system based on a hybrid beetle whisker-wolf pack algorithm, employing the data optimization method based on a hybrid beetle whisker-wolf pack algorithm as described in any one of claims 1-6, characterized in that, include: The data acquisition module is used to acquire the data to be optimized. The model building module is used to build a hybrid optimization algorithm model based on the fusion and crossover of the longhorn beetle whisker search algorithm and the gray wolf algorithm; The result output module is used to input the data to be optimized into the hybrid optimization algorithm model to obtain the optimized data processing result.

Images