A data group optimization method, device, computer equipment and storage medium
By building an integrated algorithm library and population-based metaheuristic algorithm, a collection of multiple optimization problems in the wireless sensor network is solved, and the location of sensor nodes that maximize coverage is quickly found, solving the problem of difficult to effectively deploy in the existing technology. question.
Patent Information
- Application Number
- CN202210872495.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2022-01-24
- Filing Date
- 2022-07-20
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2042-07-20
AI Technical Summary
When building wireless sensor networks, it is difficult to find a general metaheuristic algorithm to effectively solve a collection of multiple optimization problems, making it difficult to obtain a satisfactory sensor node deployment scheme and resulting in a lack of maximum network coverage area.
By building an integrated algorithm library, it contains multiple optimization algorithms, using population-based metaheuristic algorithms and improved algorithms, virtual updates are made to virtually update virtual populations, select elite individuals to form elite groups, and materialize populations when specific conditions are met, to obtain The sensor node location that maximizes coverage.
By coordinating the exploration and utilization of entity populations and virtual populations, we can enhance population diversity, reduce local optimal traps, quickly obtain the optimal sensor node deployment solution, and improve the coverage of wireless sensor networks.
Smart Images

Figure CN115130641B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of computer technology, and in particular to a data group optimization method, device, computer equipment and storage medium. Background Art
[0002] When building a wireless sensor network, after the number of sensor nodes is determined, how to deploy sensor nodes while ensuring a certain quality of service and maximize the network coverage area is one of the basic problems of wireless sensor networks. Many optimization techniques have been proposed to complete optimization tasks. Metaheuristic algorithms, as one of the approximate methods, are general algorithms that are not targeted at specific optimization problems. Although a metaheuristic algorithm can be applied to a variety of optimization problems, the solution obtained for a specific optimization problem is not necessarily satisfactory. According to the no free lunch theorem, in theory, there is no algorithm that is better than other algorithms in solving all possible optimization problems. Therefore, in practice, researchers have designed a variety of metaheuristic algorithms to solve various optimization problems with different characteristics. When faced with a set of multiple optimization problems, it is difficult to find a general metaheuristic algorithm to obtain a satisfactory solution. Summary of the invention
[0003] Based on this, it is necessary to provide a data group optimization method, device, computer equipment and storage medium that can handle a set of optimization problems containing different characteristics in order to more quickly obtain a set of sensor node locations that can maximize the network coverage.
[0004] A data population optimization method, the method comprising:
[0005] Building an integrated algorithm library; the integrated algorithm library includes multiple optimization algorithms;
[0006] When optimizing the optimization problem, the optimization algorithm is used to virtually update the virtual population to be updated outputted in the previous iteration to obtain the virtual population corresponding to each optimization algorithm in the current iteration;
[0007] Select elite individuals from the virtual populations corresponding to the respective optimization algorithms, and form an elite group with the elite individuals;
[0008] Selecting the best elite individual from the elite group, and when the fitness value of the best elite individual is more than one order of magnitude smaller than the minimum fitness value in the historical iteration, materializing the virtual population where the best elite individual is located to obtain a physical population;
[0009] When the entity population meets the preset iterative output conditions, the population optimization result is output.
[0010] In one embodiment, the method further comprises:
[0011] When the fitness value of the best elite individual is smaller than the minimum fitness value in the historical iteration and does not exceed one order of magnitude, the best elite individual is added to other virtual populations and replaces an individual in the other virtual populations to obtain the virtual population to be updated output by the current iteration.
[0012] In one embodiment, the method further includes: when the fitness value of the best elite individual is not less than the minimum fitness value in the historical iteration, retaining all virtual populations, without exchanging information between the virtual populations, and directly using the virtual populations as the virtual populations to be updated output by the current iteration.
[0013] In one of the embodiments, the method further includes: when the fitness value of the best elite individual is smaller than the minimum fitness value in the historical iteration and does not exceed one order of magnitude, adding the best elite individual to other virtual populations and replacing one individual in the other virtual populations to obtain the virtual population to be updated output by the current iteration.
[0014] In one of the embodiments, the problem to be optimized is a wireless sensor network coverage optimization problem; each individual in the virtual population is a collection of sensor node locations;
[0015] The step of selecting elite individuals from the virtual populations corresponding to the respective optimization algorithms comprises:
[0016] Calculate the fitness values of all individuals in the virtual population corresponding to each optimization algorithm, and select the individual corresponding to the minimum fitness value in each virtual population as the elite individual; each individual corresponds to a position set of each group of sensor nodes; the fitness value is calculated based on the coverage area of the wireless sensor network composed of the set of sensor node positions corresponding to the individual; the sensing range of each sensor node is a circular area centered on each sensor node, and the radius of the circular area is obtained according to the characteristics of the sensor node;
[0017] If there is more than one individual with the minimum fitness value in a virtual population, one of them is randomly selected as the elite individual.
[0018] In one embodiment, the step of selecting the best elite individual from the elite group includes:
[0019] The fitness values of the elite individuals in the elite group are sorted, and the elite individual corresponding to the minimum fitness value in the elite group is selected as the best elite individual; if there is more than one elite individual with the minimum fitness value in the elite group, one of them is randomly selected as the best elite individual.
[0020] In one embodiment, the optimization algorithm includes a population-based meta-heuristic algorithm and an improved algorithm thereof.
[0021] In one embodiment, the preset iteration output condition is that the number of iterations reaches a preset maximum number of iterations.
[0022] The maximum number of iterations is 500*D, where D is the dimension of the problem to be optimized; the dimension is the number of sensor nodes of the wireless sensor network*the coordinate dimension of the sensor nodes.
[0023] A data group optimization device, the device comprising:
[0024] A construction module is used to construct an integrated algorithm library; the integrated algorithm library includes multiple optimization algorithms;
[0025] An updating module, used to virtually update the virtual population to be updated outputted in the previous iteration by using the optimization algorithm when optimizing the optimization problem to be optimized, so as to obtain the virtual population corresponding to each optimization algorithm in the current iteration;
[0026] A selection module is used to select elite individuals from the virtual populations corresponding to the optimization algorithms, and the elite individuals constitute an elite group;
[0027] A materialization module is used to select the best elite individual from the elite group, and when the fitness value of the best elite individual is more than one order of magnitude smaller than the minimum fitness value in the historical iteration, materialize the virtual population where the best elite individual is located, and output the physical population;
[0028] The output module is used to output the population optimization result when the entity population meets the preset iterative output conditions.
[0029] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:
[0030] Build an integrated algorithm library, which includes multiple optimization algorithms.
[0031] When optimizing the optimization problem, the optimization algorithm is used to virtually update the virtual population to be updated outputted in the previous iteration to obtain the virtual population corresponding to each optimization algorithm in the current iteration;
[0032] Select elite individuals from the virtual populations corresponding to the respective optimization algorithms, and form an elite group with the elite individuals;
[0033] Selecting the best elite individual from the elite group, and when the fitness value of the best elite individual is more than one order of magnitude smaller than the minimum fitness value in the historical iteration, materializing the virtual population where the best elite individual is located, and outputting the physical population;
[0034] When the entity population meets the preset iterative output conditions, the population optimization result is output.
[0035] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps:
[0036] Building an integrated algorithm library; the integrated algorithm library includes multiple optimization algorithms;
[0037] When optimizing the optimization problem, the optimization algorithm is used to virtually update the virtual population to be updated outputted in the previous iteration to obtain the virtual population corresponding to each optimization algorithm in the current iteration;
[0038] Select elite individuals from the virtual populations corresponding to the respective optimization algorithms, and form an elite group with the elite individuals;
[0039] Selecting the best elite individual from the elite group, and when the fitness value of the best elite individual is more than one order of magnitude smaller than the minimum fitness value in the historical iteration, materializing the virtual population where the best elite individual is located to obtain a physical population;
[0040] When the entity population meets the preset iterative output conditions, the population optimization result is output.
[0041] The above-mentioned data population optimization method, device, computer equipment and storage medium first construct an integrated algorithm library. When optimizing the optimization problem, the problem to be optimized refers to the wireless sensor network coverage optimization problem. The optimization algorithm is used to virtually update the virtual population to be updated output by the previous iteration to obtain the virtual population corresponding to each optimization algorithm of the current iteration, and the individuals in each virtual population are the set of sensor node positions; elite individuals are selected from the virtual populations corresponding to each optimization algorithm, and the elite groups are composed of elite individuals, and each individual corresponds to the position set of each group of sensor nodes; the best elite individual is selected from the elite group, and when the best elite individual When the fitness value is more than one order of magnitude smaller than the minimum fitness value in the historical iteration, the virtual population where the best elite individual is located is materialized to obtain a physical population, wherein the fitness value is calculated based on the coverage area of the wireless sensor network composed of a set of sensor node positions corresponding to the individual, and the perception range of each sensor node is a circular area centered on each sensor node, and the radius of the circular area is obtained based on the characteristics of the sensor node; when the physical population meets the pre-set iterative output conditions, the group optimization result is output. It can be understood that the group optimization result is a set of sensor node positions that can maximize the coverage of the wireless sensor network.
[0042] When performing population iteration, this application uses physical populations and virtual populations to coordinate exploration and utilization issues at the entire population level. All virtual populations originate from the same physical population, and each virtual population applies a population-based meta-heuristic optimization algorithm. Each algorithm will update the position of individuals in its corresponding virtual population according to its own selection rules to obtain individuals with better fitness values, that is, to obtain sensor node positions that contribute more to the coverage area of the wireless sensor network. Each virtual population uses an elite strategy to exchange information, which enhances the diversity of the population, enables the population to obtain satisfactory solutions faster during the update process, and reduces the situation of falling into local optimality, that is, to obtain the optimal sensor node deployment plan faster. BRIEF DESCRIPTION OF THE DRAWINGS
[0043] Figure 1 A schematic diagram of a flow chart of a data group optimization method in one embodiment;
[0044] Figure 2 is a schematic diagram of a P-meta-heuristic algorithm in one embodiment;
[0045] Figure 3 is a schematic diagram of an integrated P-meta-heuristic algorithm in one embodiment;
[0046] Figure 4 It is a structural block diagram of a data group optimization device in one embodiment;
[0047] Figure 5 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION
[0048] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0049] After years of development, there are already hundreds of optimization algorithms. For a specific optimization problem, especially when there is less information about the problem, it is impossible to determine which swarm intelligence optimization algorithm will be the best way to solve it. Although a better algorithm can be found through continuous attempts, it is time-consuming and labor-intensive. In addition, for solving some problems, a single algorithm search strategy cannot find a satisfactory solution. For example, it is better to use the search strategy of algorithm A at the beginning of the search, the search strategy of algorithm B in the middle of the search, and the search strategy of algorithm C at the end of the search. In this way, a satisfactory solution can be obtained, but people do not know the specific time of algorithm switching, or even which specific algorithms A, B, and C algorithms represent. This requires putting as many algorithms as possible into the algorithm library and integrating them. At the same time, in solving specific problems, each algorithm should be given as many opportunities as possible to play its own advantages.
[0050] In one embodiment, Figure 1 As shown, a data group optimization method is provided, comprising the following steps:
[0051] Step 102, building an integrated algorithm library.
[0052] When integrating multiple optimization algorithms, it is necessary to design an appropriate integration framework to enable them to give full play to their respective advantages, so as to solve various complex optimization problems. The more types and numbers of algorithms in the algorithm library, the better, so that multiple aspects of optimization problems can be covered.
[0053] This method regards a single metaheuristic algorithm as a whole, and realizes the integration of multiple algorithms without understanding the characteristics and internal mechanisms of each metaheuristic algorithm. The integrated algorithm library includes multiple optimization algorithms. Specifically, the optimization algorithm can be a population-based optimization algorithm.
[0054] Referring to the No Free Lunch (NFL) theorem, when the set of optimization problems is very large, no optimization algorithm can outperform all other algorithms. Integrating multiple optimization algorithms can be applicable to more optimization tasks.
[0055] Step 104, when optimizing the optimization problem, respectively use the optimization algorithms to virtually update the virtual population to be updated outputted in the previous iteration, and obtain the virtual population corresponding to each optimization algorithm in the current iteration.
[0056] The problem to be optimized refers to a problem that needs to be optimized using a swarm optimization algorithm, such as the wireless sensor network (WSN) coverage optimization in the present invention, or PID controller parameter optimization or image segmentation, etc. Each individual in the virtual population corresponds to a set of sensor node positions.
[0057] When optimizing an optimization problem, some optimization algorithms can be selected from the integrated algorithm library for iterative calculation, or all optimization algorithms in the integrated algorithm library can be used for iterative calculation. The optimization algorithm can also be adaptively selected from the integrated algorithm library according to different problems to be solved. Therefore, the selection of the optimization algorithm can be determined according to actual needs and is not specifically limited here.
[0058] According to the population including multiple groups of initial sensor node positions, namely the initial entity population, multiple sets including multiple groups of node positions, namely multiple virtual populations, are generated by using multiple selected optimization algorithms. Each virtual population N adopts the corresponding optimization algorithm N to perform virtual updates respectively. The update processes can be started at the same time without interfering with each other.
[0059] Step 106, selecting elite individuals from the virtual populations corresponding to the various optimization algorithms, and forming an elite group from the elite individuals.
[0060] The elite individuals are a set of sensor node locations corresponding to the largest coverage area of the wireless sensor network in each virtual population. It can be understood that the sensing range of each sensor node is a circular area centered on each sensor node, and the radius of the circular area is obtained according to the characteristics of the sensor node. If the sensor node locations are not well deployed, it will lead to a large overlap between the sensing ranges of the sensor nodes, that is, the number of sensor nodes is large but the coverage area of the wireless sensor network is small.
[0061] Step 108, selecting the best elite individual from the elite group, when the fitness value of the best elite individual is more than one order of magnitude smaller than the minimum fitness value in the historical iteration, the virtual population where the best elite individual is located is materialized to obtain a physical population.
[0062] When the fitness value of the best elite individual is more than one order of magnitude smaller than the minimum fitness value in the historical iteration, the fitness value of the best elite individual is set to the minimum fitness value in the historical iteration, and used as a comparison benchmark for the fitness value when the next iteration is needed. The fitness value is calculated based on the coverage area of the wireless sensor network composed of the set of sensor node positions corresponding to the individual.
[0063] Step 110: When the entity population meets the preset iterative output conditions, the population optimization result is output.
[0064] If the entity population does not meet the preset termination condition, the output entity population is virtualized again, that is, the same number of virtual populations are generated according to the corresponding number of optimization algorithms, and then the corresponding virtual populations are virtually updated according to each optimization algorithm to obtain a new round of iteration of the virtual population corresponding to each optimization algorithm, and then steps 106 to 110 are repeated until the output entity population meets the preset iteration output condition, and the group optimization result is output to solve the wireless sensor network (WSN) coverage optimization problem, and the whole process ends.
[0065] From the whole process, all virtual populations are derived from the same physical population. Then, the same operation is used to continuously virtualize and materialize the population, so that the population is continuously optimized until the iteration termination condition is reached. The preset iteration output condition can be that the number of iterations reaches the preset maximum number of iterations. For example, the maximum number of iterations is 500*D, where D is the dimension of the problem to be optimized, that is, the number of sensor nodes in the wireless sensor network*the coordinate dimension of the sensor nodes.
[0066] This method can also be applied to other technical fields, as detailed below:
[0067] 1. Taking the optimization of PID controller parameters as an example, it uses a linear combination of proportional unit P, integral unit I and differential unit D to form a control quantity to achieve control of the controlled object. In order to achieve a good control effect, appropriate control parameters P, I and D should be selected according to the production process. The objective function is designed based on factors such as input-output difference, control quantity and prevention of over-adjustment. Multiple sets containing multiple groups of (P, I, D) values, i.e., the generated virtual population, are iteratively calculated, and a set containing the optimal set of control parameter (P, I, D) values is output. This set of control parameters is the population optimization result, which is used to control the production process.
[0068] 2. When using the maximum entropy threshold segmentation method for image segmentation, it is necessary to find a set of optimal segmentation thresholds that maximize the total entropy of the image. Therefore, multiple selected optimization algorithms can be used to find a set of optimal thresholds. The specific operation is: set the number of thresholds to obtain the initial entity population, which includes multiple sets of thresholds. According to the initial entity population, multiple sets containing multiple sets of thresholds are generated, and the optimization boundary is set to the pixel range [0, 255] of the image. The parameters of the selected multiple optimization algorithms are set respectively, and the corresponding optimization algorithms are used for multiple sets for multiple iterative calculations. The best set of thresholds is output to segment the image into multiple unique areas, where a set of thresholds corresponds to an individual in the population, the number of threshold groups corresponds to the number of individuals in the population, and the generated multiple sets containing multiple sets of thresholds correspond to the generated multiple virtual populations. The output set containing the best set of thresholds corresponds to the final output entity population, and the best set of thresholds is the population optimization result.
[0069] The above-mentioned data population optimization method first constructs an integrated algorithm library. When optimizing the optimization problem, the problem to be optimized refers to the wireless sensor network coverage optimization problem. The optimization algorithm is used to virtually update the virtual population to be updated output by the previous iteration to obtain the virtual population corresponding to each optimization algorithm of the current iteration, and the individuals in each virtual population correspond to a set of sensor node positions; elite individuals are selected from the virtual populations corresponding to each optimization algorithm, and the elite groups are composed of elite individuals, and each individual corresponds to the position of each sensor node; the best elite individual is selected from the elite group, and when the fitness value of the best elite individual is more than one order of magnitude smaller than the minimum fitness value in the historical iteration, the virtual population where the best elite individual is located is materialized to obtain an entity population, wherein the fitness value is calculated based on the coverage area of the wireless sensor network composed of the set of sensor node positions corresponding to the individual, and the perception range of each sensor node is a circular area centered on each sensor node, and the radius of the circular area is obtained according to the characteristics of the sensor node; when the entity population meets the preset iterative output conditions, the group optimization result is output. It can be understood that the group optimization result is a set of sensor node positions that can maximize the coverage range of the wireless sensor network.
[0070] When performing population iteration, this application uses physical populations and virtual populations to coordinate exploration and utilization issues at the entire population level. All virtual populations originate from the same physical population, and each virtual population applies a population-based meta-heuristic optimization algorithm. Each algorithm will update the position of individuals in its corresponding virtual population according to its own selection rules to obtain individuals with better fitness values, that is, to obtain sensor node positions that contribute more to the coverage area of the wireless sensor network. Each virtual population uses an elite strategy to exchange information, which enhances the diversity of the population, enables the population to obtain satisfactory solutions faster during the update process, and reduces the situation of falling into local optimality, that is, to obtain the optimal sensor node deployment plan faster.
[0071] In one of the embodiments, the data population optimization method also includes: when the fitness value of the best elite individual is smaller than the minimum fitness value in the historical iteration and does not exceed one order of magnitude, the best elite individual is added to other virtual populations and replaces an individual in the other virtual populations to obtain the virtual population to be updated output by the current iteration.
[0072] Other virtual populations refer to virtual populations other than the virtual population where the best elite individual in the current iteration is located.
[0073] When the fitness value of the best elite individual is smaller than the minimum fitness value in the historical iteration and does not exceed one order of magnitude, the fitness value of the best elite individual is set to the minimum fitness value in the historical iteration, which is used as a comparison benchmark for the fitness value in the next iteration when it is needed.
[0074] The exchange of information among virtual populations can enhance the diversity of populations, enable populations to obtain satisfactory solutions more quickly during the updating process, and reduce the situation of falling into local optimality. This method adopts the elite strategy for communication between virtual populations. When the condition that "the fitness value of the best elite individual is smaller than the minimum fitness value in the historical iteration and does not exceed one order of magnitude" is met, the best elite individual is added to each virtual population to complete the information exchange between populations. This can minimize the interference with the virtual population, retain the continuity of the algorithm, and achieve population optimization with the help of the influence of the best elite individual in the population on the population.
[0075] In one embodiment, the data population optimization method further includes: when the fitness value of the best elite individual is not less than the minimum fitness value in the historical iteration, all virtual populations are retained, no information is exchanged between the virtual populations, and the virtual population is directly used as the virtual population to be updated output by the current iteration.
[0076] In one embodiment, selecting elite individuals from virtual populations corresponding to various optimization algorithms includes: calculating the fitness values of all individuals in the virtual populations corresponding to various optimization algorithms, and selecting the individual corresponding to the minimum fitness value in each virtual population as the elite individual.
[0077] If there is more than one individual with the minimum fitness value in a virtual population, one of them is randomly selected as the elite individual.
[0078] In one embodiment, selecting the best elite individual from the elite group includes:
[0079] Sort the fitness values of the elite individuals in the elite group, and select the elite individual corresponding to the minimum fitness value in the elite group as the best elite individual; if there is more than one elite individual with the minimum fitness value in the elite group, randomly select one of them as the best elite individual.
[0080] In order to minimize the interference with the virtual population, the individual with the smallest fitness value in each virtual population is selected as the elite individual of the virtual population. If there are multiple individuals with the smallest fitness value, one is randomly selected as the elite individual to obtain the updated elite group. Among these elite groups, the best elite individual is selected according to the size of the fitness value.
[0081] In one of the embodiments, the optimization algorithm includes a population-based metaheuristic algorithm and an improved algorithm thereof.
[0082] As one of the approximate methods, metaheuristic algorithms are general algorithms that are not targeted at specific optimization problems. According to the number of solutions processed in the optimization phase, metaheuristic algorithms can be divided into single solution-based algorithms (S-metaheuristics) and population-based algorithms (P-metaheuristics):
[0083] 1) In the generation phase, the S-metaheuristic algorithm generates a set of candidate solutions from the current solution, and in the replacement phase, it selects the most suitable solution from the generated set to replace the current solution until a satisfactory solution is obtained.
[0084] 2) If Figure 2As shown, a schematic diagram of the P-metaheuristic algorithm is provided, where Initial population position represents the position of individuals in the initialized population, Optimal position represents the position of the best individual in the population, Old represents the current population, New represents the candidate population generated based on the current population, Generate represents the generation process, Select represents the process of selecting the optimal population from the generated candidate population, and Final population position represents the optimized population finally output.
[0085] The P-metaheuristic algorithm starts the optimization process by generating a population, which represents a set of feasible solutions. Each individual in the population represents a solution. A new population is generated using a designed operator, and the original population individuals are replaced with the newly generated population individuals according to the selection strategy. In this way, the population will be continuously optimized, and this process continues until the set stopping condition is met. The difference between various P-metaheuristic algorithms lies in the way they perform the generation and selection process.
[0086] Based on the source of inspiration, P-metaheuristic algorithms can be divided into five categories: evolutionary algorithms, biologically based algorithms, human-based algorithms, physics-based algorithms, and mathematically based algorithms.
[0087] The search steps of these algorithms all include two stages: exploration and utilization. In the exploration stage, the diversity of the population should be maintained as much as possible so that the individuals in the population can fully explore various areas of the feature space. In the utilization stage, the search focus of the population should be placed on areas with high-quality solutions, and the search should be continuously conducted near their neighborhoods. Balancing exploration and utilization is an important issue, otherwise it will fall into a local optimum due to insufficient exploration or fail to converge due to insufficient utilization.
[0088] like Figure 3 As shown, a schematic diagram of the integrated P-meta-heuristic algorithm is provided, wherein Initial represents the process of initializing the positions of individuals in the initial population, during which the corresponding virtual population is generated according to the selected optimization algorithm; Generate represents the process of virtually updating each virtual population by iterative calculation using the corresponding optimization algorithm, and the updating processes do not interfere with each other; Select represents the process of selecting the virtual population to be updated in the next iteration among each virtual population; Final represents the process of obtaining the final output entity population.
[0089] Here, an embodiment of an integrated algorithm is provided. In the experimental part of this application, relatively new representative algorithms in recent years are selected for integration, which are Harris Hawk Optimization (HHO) algorithm, Seagull Optimization (SOA) algorithm, Slime Mold Optimization (SMA) algorithm, Salp Swarm Optimization (SSA) algorithm, and Manta Ray Foraging Optimization (MRFO) algorithm.
[0090] The DC algorithm is used to represent the integrated algorithm. In this embodiment, the Harris Hawk algorithm and the Seagull algorithm are integrated to obtain the DC2 algorithm, the Harris Hawk algorithm, the Seagull algorithm and the slime mold optimization algorithm are integrated to obtain the DC3 algorithm, the Harris Hawk algorithm, the Seagull algorithm, the slime mold optimization algorithm and the salp swarm optimization algorithm are integrated to obtain the DC4 algorithm, and the Harris Hawk algorithm, the Seagull algorithm, the slime mold optimization algorithm, the salp swarm optimization algorithm and the manta ray foraging optimization algorithm are integrated to obtain the DC5 algorithm. The parameters used in these integrated algorithms are the same as the parameter settings of the five original metaheuristic algorithms: the Harris Hawk Optimization (HHO) algorithm, the Seagull Optimization (SOA) algorithm, the slime mold optimization (SMA) algorithm, the salp swarm optimization (SSA) algorithm, and the manta ray foraging optimization (MRFO) algorithm.
[0091] In order to evaluate the performance of the algorithm, it is applied to two test sets. In this embodiment, two sets of benchmark functions are selected:
[0092] One set is 23 standard benchmark functions, detailed description is shown in Table 1. It contains function type, function name and function definition. In addition, Dims is the dimension of the function, Initialization range is the value range of the variable, and the optimal value corresponding to each function is the Global optimum. Among these functions, function F1-F13 is a high-dimensional problem with Dim = 30. Among them, F1-F7 functions are unimodal functions, and F8-F13 functions are multimodal functions, in which the number of local optimal solutions increases exponentially with the increase of the problem dimension. Functions F14-F23 are low-dimensional problems with only a few local minima.
[0093] Table 1.23 standard test functions
[0094]
[0095]
[0096] Another set of benchmark functions is the CEC2017 benchmark function, as detailed in Table 2. Among them, C1-C2 functions are unimodal functions, C3-C9 functions are simple multimodal functions, C10-C19 functions are mixed functions, and C20-C29 functions are composite functions. The dimensions can be set as needed. For the convenience of comparison, they are uniformly set to 30.
[0097] The mean values MEAN and standard deviations STD of the results obtained for all tested algorithms in 25 independent runs and 500 iterations on each dimension are recorded and compared and are recorded in Tables 3 and 4.
[0098] Table 3 shows the comparison results of the above nine different optimization algorithms on 23 standard test function sets. According to Friedman average ranking, DC5 ranks first, DC4 ranks second, MRFO ranks third, DC3 ranks fourth, DC2 ranks fifth, SMA ranks sixth, HHO ranks seventh, SSA ranks eighth, and SOA ranks ninth. The DC algorithms using the integrated framework have achieved good rankings. After integrating HHO and SOA algorithms, the DC2 algorithm surpasses the four meta-heuristic algorithms SMA, HHO, SSA, and SOA. The best ranked meta-heuristic algorithm is MRFO, which has excellent optimization ability. It not only surpasses the four algorithms SMA, HHO, SSA, and SOA, but also surpasses the two integrated algorithms DC2 and DC3. When facing the DC4 algorithm that integrates four meta-heuristic algorithms, the DC4 algorithm relies on the superiority of the integrated framework and surpasses the MRFO algorithm in the average optimization ability of the algorithm. In addition, according to the rankings of the five DC-type algorithms, DC2, DC3, DC4, and DC5, it can be seen that as the number of integrated metaheuristic algorithms increases, the average optimization ability of the algorithm continues to increase.
[0099] Table 2. CEC2017 test functions
[0100]
[0101] Table 3. Comparison results of different optimization algorithms on 23 standard test function sets
[0102]
[0103] Table 4 shows the comparison results of the above nine different optimization algorithms on the CEC2017 test function set. According to the Friedman average ranking, DC5 ranks first, DC4 ranks second, DC3 ranks third, MRFO ranks fourth, SMA ranks fifth, SSA ranks sixth, DC2 ranks seventh, HHO ranks eighth, and SOA ranks ninth. On the CEC2017 test set, the DC class algorithms using the integrated framework still maintain their advantages. For the comparison of metaheuristic algorithms, the rankings of HHO and SSA have changed, and HHO's performance is not good, indicating that the performance of the algorithms is different under different test sets. Therefore, the ranking of the DC2 algorithm has also dropped. According to the results on the CEC2017 test function set, the performance of the five DC class algorithms DC2, DC3, DC4, and DC5 still shows an increasing trend.
[0104] Table 4. Comparison results of different optimization algorithms on the CEC2017 test set
[0105]
[0106]
[0107] Table 5 shows the Wilcoxon test results between the ensemble algorithm and the metaheuristic algorithm and the ensemble algorithm on 23 standard test functions. The p-values obtained in the comparison between DC2 and HHO, DC3 and SMA are greater than 0.05 and less than 0.1, indicating that the results of the algorithms are different at the significance level of 0.1; the p-values obtained in the comparison between DC3 and DC2, DC4 and DC3, DC5 and MRFO are less than 0.05 and greater than 0.01, indicating that the results of the algorithms are different at the significance level of 0.05; in the comparison of other algorithms, the p-values are all less than 0.01, indicating that the results of the algorithms are different at the significance level of 0.01.
[0108] Table 5. Wilcoxon test results on 23 standard test functions
[0109]
[0110] Table 6 shows the Wilcoxon test results between the ensemble algorithm and the metaheuristic algorithm, and between the ensemble algorithms on the CEC2017 test function. The p-value obtained by comparing the DC2 and HHO algorithms is greater than 0.2, indicating that there is no significant difference between the results. In the comparison of other algorithms, the p-values are all less than 0.01, indicating that the results of the algorithms are different at the significance level of 0.01.
[0111] Table 6. Wilcoxon test results of the CEC2017 test function
[0112]
[0113] Through observation of the above experimental conditions, it can be seen that the DC5 algorithm based on the integrated framework proposed in this paper has a faster iterative optimization speed and can achieve better optimization results.
[0114] In one embodiment, Figure 4 As shown, a structural block diagram of a data group optimization device is provided, the device includes: a construction module, an update module, a selection module, an instantiation module and an output module, wherein:
[0115] A building module is used to build an integrated algorithm library; the integrated algorithm library includes multiple optimization algorithms;
[0116] An updating module is used to virtually update the virtual population to be updated outputted in the previous iteration by using the optimization algorithm when optimizing the optimization problem to be optimized, so as to obtain the virtual population corresponding to each optimization algorithm in the current iteration;
[0117] The selection module is used to select elite individuals from the virtual populations corresponding to each optimization algorithm, and the elite individuals form an elite group;
[0118] The materialization module is used to select the best elite individual from the elite group. When the fitness value of the best elite individual is more than one order of magnitude smaller than the minimum fitness value in the historical iteration, the virtual population where the best elite individual is located is materialized to obtain a physical population.
[0119] The output module is used to output the population optimization results when the entity population meets the preset iterative output conditions.
[0120] In one of the embodiments, the selection module is also used to calculate the fitness values of all individuals in the virtual populations corresponding to each optimization algorithm, and select the individual corresponding to the minimum fitness value in each virtual population as the elite individual. If there is more than one individual with the minimum fitness value in a virtual population, one of them is randomly selected as the elite individual.
[0121] In one embodiment, the instantiation module is further used to sort the fitness values of the elite individuals in the elite group, and select the elite individual corresponding to the minimum fitness value in the elite group as the best elite individual.
[0122] For the specific definition of the data group optimization device, please refer to the definition of the data group optimization method above, which will not be repeated here. Each module in the above-mentioned data group optimization device can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.
[0123] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 5As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store integrated algorithm library data. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a data group optimization method is implemented.
[0124] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0125] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method in the above embodiment when executing the computer program.
[0126] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method in the above embodiment are implemented.
[0127] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
[0128] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0129] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.
Claims
1. A data population optimization method, It is characterized in that The method comprises: Building an integrated algorithm library; the integrated algorithm library includes multiple optimization algorithms; When optimizing the optimization problem, the optimization algorithm is used to virtually update the virtual population to be updated outputted in the previous iteration to obtain the virtual population corresponding to each optimization algorithm in the current iteration; The problem to be optimized is a wireless sensor network coverage optimization problem; each individual of the virtual population corresponds to a set of sensor node positions; Selecting elite individuals from the virtual populations corresponding to the respective optimization algorithms, and forming an elite group from the elite individuals; selecting elite individuals from the virtual populations corresponding to the respective optimization algorithms, including: Calculate the fitness values of all individuals in the virtual populations corresponding to the optimization algorithms, and select the individual corresponding to the minimum fitness value in each virtual population as the elite individual; the fitness value is calculated based on the coverage area of the wireless sensor network composed of the set of sensor node positions corresponding to the individuals; the sensing range of each sensor node is a circular area centered on each sensor node, and the radius of the circular area is obtained according to the characteristics of the sensor node; If there is more than one individual with the minimum fitness value in a virtual population, one of them is randomly selected as the elite individual; Selecting the best elite individual from the elite group, and when the fitness value of the best elite individual is more than one order of magnitude smaller than the minimum fitness value in the historical iteration, materializing the virtual population where the best elite individual is located to obtain a physical population; When the entity population meets the preset iterative output conditions, the population optimization result is output.
2. The method according to claim 1, It is characterized in that The method further comprises: When the fitness value of the best elite individual is smaller than the minimum fitness value in the historical iteration and does not exceed one order of magnitude, the best elite individual is added to other virtual populations and replaces an individual in the other virtual populations to obtain the virtual population to be updated output by the current iteration.
3. The method according to claim 1, It is characterized in that The method further comprises: When the fitness value of the best elite individual is not less than the minimum fitness value in the historical iteration, all virtual populations are retained without information exchange between the virtual populations, and the virtual populations are directly used as the virtual population to be updated output by the current iteration.
4. The method according to claim 1, It is characterized in that Select the best elite individuals from the elite group, including: Sorting the fitness values of the elite individuals in the elite group, and selecting the elite individual corresponding to the minimum fitness value in the elite group as the best elite individual; If there is more than one elite individual with the minimum fitness value in the elite group, one of them is randomly selected as the best elite individual.
5. The method according to claim 1, It is characterized in that The optimization algorithm includes a population-based meta-heuristic algorithm and an improved algorithm thereof.
6. The method according to claim 1, It is characterized in that The preset iteration output condition is that the number of iterations reaches a preset maximum number of iterations; The maximum number of iterations is 500*D, where D is the dimension of the problem to be optimized; the dimension is the number of sensor nodes of the wireless sensor network*the coordinate dimension of the sensor nodes.
7. A data group optimization device, It is characterized in that The device comprises: A construction module is used to construct an integrated algorithm library; the integrated algorithm library includes multiple optimization algorithms; The updating module is used to perform virtual updating on the virtual population to be updated outputted in the previous iteration by using the optimization algorithm when optimizing the optimization problem to be optimized, so as to obtain the virtual population corresponding to each optimization algorithm in the current iteration; the problem to be optimized is the coverage optimization problem of the wireless sensor network; each individual of the virtual population corresponds to a set of sensor node positions; The selection module is used to select elite individuals from the virtual populations corresponding to the optimization algorithms, and the elite individuals constitute an elite group; the elite individuals are selected from the virtual populations corresponding to the optimization algorithms, including: calculating the fitness values of all individuals in the virtual populations corresponding to the optimization algorithms, and selecting the individual corresponding to the minimum fitness value in each virtual population as the elite individual; the fitness value is calculated based on the coverage area of the wireless sensor network composed of a set of sensor node positions corresponding to the individual; the sensing range of each sensor node is a circular area centered on each sensor node, and the radius of the circular area is obtained according to the characteristics of the sensor node; if there is more than one individual with the minimum fitness value in a virtual population, one of them is randomly selected as the elite individual; A materialization module is used to select the best elite individual from the elite group, and when the fitness value of the best elite individual is more than one order of magnitude smaller than the minimum fitness value in the historical iteration, materialize the virtual population where the best elite individual is located to obtain a physical population; The output module is used to output the population optimization result when the entity population meets the preset iterative output conditions.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program. It is characterized in that When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.
9. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.
