WSN intelligent clustering method and device for energy consumption optimization

Abstract

The invention discloses an intelligent clustering method and an intelligent clustering device for a wireless sensor network (WSN) for energy consumption optimization, and belongs to the field of wireless sensing. Aiming at the problems of energy imbalance, low coverage efficiency and the like of the existing WSN clustering algorithm, the method comprises the following steps: firstly, initializing a clustering scheme and calculating a fitness value, and comprehensively considering multiple factors such as energy consumption balance and the like by a fitness function; designing and calculating an adaptive factor; simulating the foraging behavior of the parrot-shaped parrot, wherein the foraging behavior comprises the steps of air search, diving predation and the like to adjust the cluster head position and the clustering scheme; the algorithm is prevented from falling into local optimum by enhancing search and avoiding a predation strategy; and finally, judging a termination condition to obtain an optimal solution. The device is composed of a processor, a memory and other modules, and is responsible for data processing, algorithm calculation, communication and energy monitoring. The method is mainly used for optimizing WSN clustering, improving the network coverage rate and the energy efficiency and accelerating the algorithm convergence speed.

Description

Technical Field

[0001] The present invention relates to the technical field of Wireless Sensor Network (WSN), and in particular, to a WSN clustering method and device of an Atlantic Puffin Optimization Algorithm (APOA) with an adaptive operator mechanism and a local search strategy, belonging to the technical field of WSN clustering. Background Art

[0002] A Wireless Sensor Network (WSN) consists of a large number of sensor nodes deployed in a monitoring area. These nodes form an ad-hoc network through wireless communication to collaboratively sense, collect, and process information in the monitoring area and send it to users. In a WSN, clustering is a crucial network organization method. By dividing sensor nodes into multiple clusters and electing cluster head nodes to be responsible for data collection and fusion within the clusters and then sending the fused data to the base station, it can effectively reduce the data transmission volume, reduce node energy consumption, and extend the network lifetime. However, traditional WSN clustering algorithms have many defects. For example, unreasonable clustering results lead to unbalanced loads of nodes within the cluster, seriously affecting network performance; when dealing with large-scale networks, the convergence speed of the algorithm is slow, and it is difficult to quickly find the optimal clustering scheme.

[0003] As a newly emerging bionic intelligent algorithm, the Atlantic Puffin Optimization Algorithm simulates the unique foraging behavior of Atlantic puffins and has advantages such as strong global search ability and fast convergence speed. Applying it to WSN clustering is expected to effectively alleviate the deficiencies of existing clustering technologies and significantly improve the overall performance of WSN. However, to successfully apply this algorithm to WSN clustering, it is necessary to make in-depth improvements according to the characteristics of WSN, including optimizing algorithm parameters, improving the adaptability and stability of the algorithm, etc. How to achieve this goal has become the focus and key of current WSN clustering technology research.

[0004] In view of the above problems, the present invention proposes a clustering method of an Atlantic Puffin Optimization Algorithm integrating an adaptive mechanism and a local search strategy, aiming to effectively alleviate problems such as energy imbalance and low coverage efficiency in WSN clustering. Summary of the Invention

[0005] The first technical solution is as follows:

[0006] An intelligent WSN clustering method for energy consumption optimization, comprising:

[0007] Initialize the puffin population and construct a WSN clustering model. Calculate the fitness values of each clustering scheme through a fitness function, design and calculate an adaptive operator, and simulate the behavior of puffins (such as aerial search, diving for prey, and aggregating for foraging) to optimize the clustering scheme until the optimal clustering scheme is output after meeting the termination conditions.

[0008] According to the above-mentioned initialization of the puffin population and construction of the WSN clustering model, it includes:

[0009] Initialize the puffin population, and adopt a coding method based on location information. Each individual represents a WSN clustering scheme. For a WSN with n sensor nodes, an individual can be represented as a vector containing m elements, where m is the expected number of cluster heads, and each element in the vector is a two-dimensional coordinate representing the position of the cluster head. Randomly generate m coordinates within the monitoring area as the initial cluster head positions, thereby determining the initial clustering scheme. This coding method intuitively reflects the position information of the cluster heads and provides a basis for subsequent clustering optimization.

[0010] According to the above-mentioned calculation of the fitness values of each clustering scheme through a fitness function, it includes:

[0011] The fitness function comprehensively considers multiple factors, such as the energy consumption balance E of nodes within the cluster balance , the distance D from the cluster head node to the base station toBS , the coverage redundancy R coverage , etc.;

[0012] Define the fitness function as: Fitness = α × E balance + β × D toBS + γ × (1 - R coverage )

[0013] Where α, β, and γ are weight coefficients, and satisfy α + β + γ = 1, α ≥ 0, β ≥ 0, γ ≥ 0. These weight coefficients are reasonably set according to the actual application scenarios and requirements. For example, in scenarios that are more sensitive to energy consumption, the value of α can be appropriately increased to highlight the importance of energy balance; in scenarios with higher requirements for coverage, the weight of γ can be increased to ensure a higher coverage rate of the network.

[0014] Calculate E balance , assume that there are N j nodes in the jth cluster, and the energy consumption of node k is e jk , then the average energy consumption of this cluster is:

[0015] Among them, m is the number of clusters. Through this calculation method, the balanced degree of energy consumption of nodes within the cluster can be accurately measured, providing a reliable energy balance index for the fitness function.

[0016] Calculate D toBS , Let the coordinates of the i-th cluster head node be (x hj , y hj ), and the base station coordinates be (x bs , y bs ), then:

[0017] This formula accurately calculates the distance from the cluster head node to the base station, reflecting the influence of the distance factor on the clustering scheme in the fitness function.

[0018] Calculate R coverage , Let the sensing radius of the sensor node be r. If the distance d il between node i and node l ≤ 2r, it is considered that their coverage areas overlap. By statistically calculating the proportion of the overlapping area, R coverage is obtained. This method of calculating the coverage redundancy can effectively evaluate the effectiveness of network coverage, avoid excessive coverage overlap, and improve the utilization efficiency of network resources.

[0019] Design and calculate the adaptive operator according to the above, including:

[0020] Define the adaptive operator λ, which is related to the current iteration number t, the maximum iteration number T, and the population fitness variance σ 2 . The population fitness variance σ 2 reflects the dispersion degree of individual fitness values in the population, and its calculation formula is:

[0021] Among them, P is the population size, is the fitness value of the i-th individual at time t, is the average fitness value of the population at time t.

[0022] The calculation formula of the adaptive operator λ is:

[0023] Among them, λ min and λ max are the minimum and maximum values of the adaptive operator, is the maximum value of the pre-set population fitness variance.

[0024] Optimize the clustering scheme according to the simulated behavior of the Atlantic puffin, including:

[0025] Step 1: Aerial Search; The individual puffins (clustering scheme) simulate the behavior of searching for food in the air and update their positions. In the WSN clustering scenario, position update means the adjustment of the cluster head positions. The individual updates its position according to its own position and the global optimal position, and the specific formula is:

[0026] where, represents the position of the i-th puffin at time t (i.e., the cluster head position of the current clustering scheme), represents the global optimal position at time t (i.e., the cluster head position of the optimal clustering scheme), and r1 is a random number in the range [0, 1]. This step simulates the process of puffins exploring different areas in the air to find a better food source (i.e., a better clustering scheme), enabling the algorithm to conduct a wide search in the solution space and continuously explore potential better clustering schemes.

[0027] Step 2: Dive for Prey; When the puffin discovers a potential food source (an area of a better clustering scheme), it performs a dive for prey action. In WSN clustering, this corresponds to locally optimizing the current clustering scheme. When the fitness value of an individual is better than the current population average fitness value to a certain extent, the clustering scheme corresponding to this individual is fine-tuned. For example, for a certain cluster head position, new position points are randomly generated within a certain range around it, and the fitness value is recalculated. If the new fitness value is better, then the cluster head position is updated. According to the idea of the local search strategy, if:

[0028] then:

[0029] where, is the population average fitness value at time t, and the calculation formula is:

[0030] where, p is the population size, r2 is a relatively small random number in the range [0, 0.1], and each element is updated using an independent r2 value; ΔX is the offset randomly generated around the cluster head position, and the offset ΔX j for the j-th cluster head position satisfies ΔX j =(Δx j , Δy j ), Δx j and Δy jThey are randomly generated within the range of [-δ], where δ is a preset offset range parameter. Through this local optimization operation, the clustering scheme can be finely adjusted to improve the quality of the clustering scheme. During the diving and predation stage, local optimization of individuals with better fitness values helps to further explore local optimal solutions and enhance the performance of the clustering scheme.

[0031] Step 3: Aggregate foraging; Arctic puffin individuals aggregate foraging through information sharing. In WSN clustering, this is reflected in the information exchange and integration between different clustering schemes. Each individual interacts with surrounding individuals to obtain their advantageous information. For example, for two adjacent individuals (clustering schemes), exchange partial cluster head position information, then recalculate the fitness value, and retain the new scheme with a better fitness value. Specifically, it can be expressed as: for individual i and individual j, randomly select some cluster head positions for exchange to obtain new individuals i' and j', calculate Fitness(i') and Fitness(j'), if: Fitness(i′)>Fitness(i)

[0032] Then

[0033] If: Fitness(j′)>Fitness(j)

[0034] Then This process is similar to the idea of multi-agent collaborative optimization. By collaborating among individuals, the quality of the overall clustering scheme is improved, making full use of the advantages of different clustering schemes to further optimize the clustering results. The aggregate foraging stage promotes information sharing and complementary advantages among clustering schemes, speeds up the convergence rate of the algorithm, and improves the overall performance of the clustering scheme.

[0035] Step 4: Strengthen search; As the iteration progresses, when the change in the fitness value of the population tends to be flat, enter the strengthen search stage. According to the strategy of balancing global search and local search, at this time, expand the search range and increase the randomness of the search. For example, in the position update formula, appropriately increase the value range of the random number r1 from [0,1] to [0,1.5], and adjust the proportion of local search individuals according to the adaptive operator λ. When λ is larger, increase the number of local search individuals to strengthen the search in the local area; when λ is smaller, reduce the number of local search individuals and conduct more global exploration. The specific implementation method can be to set a function f(λ) related to λ to determine the number of local search individuals, such as num local =int(p×f(λ)), where num localis the number of individuals in local search, and int() is the rounding function. In the enhanced search stage, by adjusting the random number range and increasing the number of individuals in local search, the intensity of global search and local search is balanced, enabling the algorithm to continue optimizing when approaching the optimal solution and improving the optimization ability of the algorithm.

[0036] Step 5: Avoid predation; Simulate the behavior of arctic puffins avoiding predators. In WSN clustering, to prevent the algorithm from falling into a local optimal solution, when an individual remains in a local optimal state for a long time (for example, the fitness value does not improve after consecutive iterations), a random perturbation is performed on it. According to the method of jumping out of the local optimum, a large random change is made to some of the cluster head positions of this individual, enabling it to jump out of the current local optimal region and search for a better solution again. For example, for an individual i whose fitness value has not improved after k consecutive iterations, randomly change 30% of its cluster head positions, that is:

[0037] where is the new clustering scheme after randomly generating some cluster head positions again. This strategy effectively expands the population search range and is an effective means to prevent the algorithm from falling into a local optimum, ensuring that the algorithm can continuously search in a complex solution space and improving the reliability and stability of the algorithm.

[0038] According to the above judgment, it is determined whether the termination condition is satisfied after each iteration, including:

[0039] After each iteration, it is judged whether the termination condition is satisfied. If the maximum number of iterations is reached (for example, the set maximum number of iterations is T, when t = T), or in k consecutive iterations, the change in the optimal fitness value is less than a certain minimum threshold ∈, the algorithm is terminated and the current optimal clustering scheme is output; otherwise, return to calculate the fitness values of each clustering scheme through the fitness function and continue the next iteration.

[0040] The second technical solution is as follows: An intelligent clustering device for WSN oriented to energy consumption optimization, used to execute the method described in the first technical solution, includes a processor, a memory, a data preprocessing module, a communication module, and an energy monitoring module:

[0041] The processor, as the core computing unit of the device, undertakes multiple key tasks. During the operation of the Arctic Puffin Optimization Algorithm (APOA), the processor is responsible for performing the calculations at each stage. In the initialization stage, it generates the Arctic Puffin population according to preset rules, determines the WSN clustering scheme represented by each individual, and calculates the corresponding fitness value. In subsequent stages such as aerial search, dive predation, aggregation foraging, enhanced search, and avoiding predation, the processor updates the positions of the clustering scheme, performs local optimization, conducts information interaction, adjusts the search range, and deals with local optima according to the corresponding formulas and strategies. At the same time, the processor also needs to screen different strategies, and judge when to adopt which strategy according to the real-time state of the algorithm operation and the preset conditions to ensure that the algorithm advances in a more optimal direction. When the algorithm may fall into a local optimal solution, the processor perturbs the relevant individuals randomly according to the avoiding predation strategy, so that the algorithm jumps out of the local optimal region and continues to search for a better solution;

[0042] The memory is mainly responsible for storing various key data, providing data support for the stable operation of the algorithm. Before the algorithm runs, the memory stores the relevant information of the sensor nodes, including the number of nodes, their positions, initial energy, etc. During the execution of the algorithm, it saves the clustering scheme generated in each round of iteration, as well as data such as the results of the calculated fitness function. These data are not only the basis for the algorithm to perform the next calculation and decision, but also facilitate the subsequent analysis and evaluation of the algorithm operation process. The memory also stores some intermediate variables and parameters during the algorithm operation to ensure that they can be quickly called and updated when needed, improving the operation efficiency of the algorithm;

[0043] The data preprocessing module preprocesses the raw data collected by the sensor nodes. First, it denoises the data, using a filtering algorithm to remove the noise data generated by environmental interference and other factors to ensure the accuracy of the data. Then it extracts features, extracting the key features related to clustering, such as the remaining energy and signal strength of the nodes. The data after this preprocessing will be used as the input for the subsequent clustering algorithm, improving the processing efficiency of the algorithm and the reliability of the clustering results;

[0044] The communication module is responsible for realizing data communication between the sensor nodes and the device, as well as between the device and the base station. It adopts an efficient wireless communication protocol and has the function of adaptively adjusting the communication power, dynamically adjusting the transmission power according to the distance between nodes and the signal strength to reduce energy consumption. It supports multi-channel communication, effectively avoiding communication conflicts and improving the stability and reliability of data transmission. During the clustering process, it is responsible for transmitting the information of each clustering scheme between different nodes to achieve information sharing and interaction, promoting the optimization of the clustering scheme;

[0045] The energy monitoring module monitors the energy consumption of sensor nodes in real time. By collecting and analyzing the node energy data, it accurately calculates the remaining energy of each node and feeds this information back to the processor. Based on the remaining energy of the nodes, the processor preferentially selects nodes with higher remaining energy as cluster heads during the clustering process to balance the network energy consumption and extend the network life cycle. It has an energy warning function. When the energy of some nodes in the network is lower than the set threshold, it sends a warning message to the processor in a timely manner so that the processor can take corresponding measures, such as readjusting the clustering scheme or starting the energy-saving mode;

[0046] The beneficial effects of the present invention are as follows: The present invention uses the Arctic puffin optimization algorithm (APOA) to fuse the adaptive mechanism and the local search strategy for clustering WSN. In terms of optimizing the cluster head selection and the formation of clusters, by simulating the unique foraging behavior of Arctic puffins, the clustering scheme is continuously adjusted, effectively improving the network coverage rate. At the same time, the fitness function comprehensively considers various factors such as the energy consumption balance of nodes within the cluster, the distance from the cluster head node to the base station, and the coverage redundancy, making the clustering result more reasonable, thereby reducing the node energy consumption and improving the energy efficiency. In terms of the algorithm performance, the adaptive mechanism and the local search strategy enhance the optimization ability of the algorithm. The local optimization operations during the diving predation and aggregating foraging processes can finely adjust the clustering scheme; the enhanced search phase expands the search range and increases the search randomness to avoid the algorithm falling into the local optimum prematurely; the avoiding predation strategy randomly perturbs individuals when the algorithm may fall into the local optimum solution, enabling it to jump out of the local optimum region and expanding the population search range, effectively improving the stability of the algorithm. Compared with the prior art, an intelligent clustering method and device for WSN oriented to energy consumption optimization perform better in terms of network coverage rate and energy efficiency, and the algorithm convergence speed is faster, which can quickly find a better clustering scheme, having significant advantages and application value in the field of WSN clustering technology. Description of the Drawings

[0047] Figure 1 It is a schematic flowchart of an intelligent clustering method for WSN oriented to energy consumption optimization;

[0048] Figure 2 It is a schematic structural diagram of an intelligent clustering device for WSN oriented to energy consumption optimization;

[0049] Figure 3 It is a schematic diagram of the fitness simulation comparison of three WSN clusterings;

[0050] Figure 4 It is the abstract drawing of an intelligent clustering method and device for WSN oriented to energy consumption optimization. Detailed Embodiments

[0051] In order to make the technical solutions and advantages in the embodiments of the present invention clearer and more understandable, the following further details the exemplary embodiments of the present invention with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than an exhaustive list of all embodiments. It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other.

[0052] Example 1: Refer to Figure 1 This embodiment is described in detail. An energy consumption optimization-oriented WSN intelligent clustering method specifically includes the following steps:

[0053] Step 1: Initialize the Arctic tern population. An encoding method based on location information is adopted, and each individual represents a WSN clustering scheme. For a WSN with 100 sensor nodes, an individual can be represented as a vector containing 10 elements, where 10 is the expected number of cluster heads, and each element in the vector is a two-dimensional coordinate representing the position of the cluster head. The population size of the algorithm is 90. Randomly generate 10 coordinates within the monitoring area of 400×400 as the initial cluster head positions, thereby determining the initial clustering scheme.

[0054] Step 2: Calculate the fitness value of each individual; the fitness function comprehensively considers the energy consumption balance E balance of the nodes within the cluster, the distance D toBS from the cluster head node to the base station, the coverage redundancy R coverage and other factors. Define the fitness function: Fitness = α×E balance +β×D toBS +γ×(1 - R coverage )

[0055] where α, β, and γ are weight coefficients, and satisfy α + β + γ = 1, α≥0, β≥0, γ≥0, and are reasonably set according to the actual application scenario and requirements to ensure the rationality of the clustering scheme.

[0056] E balance is the energy consumption balance of the nodes within the cluster, and is calculated in the following way: Let there be N j nodes in the jth cluster, and the energy consumption of node k is e jk , then the average energy consumption of this cluster is:

[0057] where m is the number of clusters.

[0058] D toBS is the distance from the cluster head node to the base station. Let the coordinates of the ith cluster head node be (x hj , y hj ) and the base station coordinates be (xbs , y bs ), then:

[0059] R coverage For the coverage redundancy, assume the sensing radius of the sensor node is r = 282.84. If the distance d between node i and node l il ≤ 2r, d il ≤ 565.68, then it is considered that their coverage areas overlap, and R is obtained by statistically calculating the proportion of the overlapping area coverage .

[0060] Step 3: Design and calculate the adaptive operator; Define the adaptive operator λ, which is related to the current iteration number t, the maximum iteration number T = 80, and the population fitness variance σ 2 The population fitness variance σ 2 reflects the degree of dispersion of the individual fitness in the population, and its calculation formula is:

[0061] where P is the population size, is the fitness value of the i-th individual at time t, is the average fitness value of the population at time t.

[0062] The calculation formula of the adaptive operator λ is:

[0063] where λ min = 0.2 and λ max = 0.9 are the minimum and maximum values of the adaptive operator, is the pre-set maximum value of the population fitness variance.

[0064] Step 4: Aerial search;

[0065] The individuals of Atlantic puffins (clustering scheme) simulate the behavior of searching for food in the air to update their positions. In the WSN clustering scenario, that is, to adjust the positions of cluster heads. In this stage, the individuals update their positions according to their own positions and the global optimal position. The specific formula is:

[0066] where, represents the position of the i-th Atlantic puffin at time t (i.e., the position of the cluster head of the current clustering scheme), represents the global optimal position at time t (i.e., the position of the cluster head of the optimal clustering scheme), and r1 is a random number in the range of [0, 1]. Through this operation, potential better clustering schemes are explored.

[0067] Step 5: Dive for prey;

[0068] When the Atlantic puffin discovers a potential food source (an area with a better clustering scheme), it performs a diving predation action. In WSN clustering, this corresponds to locally optimizing the current clustering scheme. When the fitness value of an individual is better than the current population average fitness value to a certain extent, the clustering scheme corresponding to this individual is fine-tuned. For example, for a certain cluster head position, new position points are randomly generated within a certain range around it, and the fitness value is recalculated. If the new fitness value is better, then the cluster head position is updated. According to the idea of the local search strategy, if:

[0069] Then:

[0070] Among them, is the population average fitness value at time t, and the calculation formula is:

[0071] Among them, p is the population size, which is 90. Among them, r2 is a relatively small random number, ranging from [0, 0.1], and each element is updated using an independent r2 value; ΔX is the offset randomly generated around the cluster head position. For the offset ΔX of the j-th cluster head position j satisfies ΔX j =(Δx j , Δy j ), Δx j and Δy j are respectively randomly generated within [-δ, δ], and δ is the preset offset range parameter, with a value of 0.1. Through this local optimization operation, the clustering scheme can be finely adjusted to improve the quality of the clustering scheme.

[0072] Step 6: Aggregate foraging;

[0073] Information exchange and fusion are carried out between different clustering schemes. After adjacent individuals exchange part of the cluster head position information, the fitness value is recalculated, and the better scheme is retained. For individuals i and j, after exchanging part of the cluster head positions to obtain new individuals i' and j', calculate Fitness(i') and Fitness(j'). If: Fitness(i)>Fitness(i)

[0074] Then

[0075] If: Fitness(j)>Fitness(j)

[0076] Then Improve the quality of the overall clustering scheme.

[0077] Step 7: Strengthen the search;

[0078] When the change of the population fitness value tends to be flat, enter the stage of strengthening the search. Expand the value range of the random number r1 from [0, 1] to [0, 1.5], and adjust the proportion of local search individuals according to the adaptive operator λ. When λ is large, increase the number of local search individuals to strengthen the search in the local area; when λ is small, reduce the number of local search individuals and conduct more global exploration. The specific implementation method can be to set a function f(λ) related to λ to determine the number of local search individuals, such as num local = int(p × f(λ)), where num locak is the number of local search individuals, and int() is the rounding function. In the stage of strengthening the search, by adjusting the random number range and increasing the number of local search individuals, the intensity of global search and local search is balanced, enabling the algorithm to continue to optimize when approaching the optimal solution and improving the optimization ability of the algorithm.

[0079] Step 8: Avoid predation;

[0080] When the fitness value of an individual does not increase for several consecutive iterations, perform random perturbation on it. For example, for an individual i whose fitness value has not increased for k consecutive iterations, randomly change 30% of its cluster head positions, that is:

[0081] where, is the new clustering scheme after randomly generating some cluster head positions again. This strategy effectively expands the population search range and improves the stability of the algorithm.

[0082] Step 9: Judge the termination condition: After each iteration is completed, judge whether the termination condition is met. If the maximum number of iterations is reached, that is, when t = T = 80, or in k = 5 consecutive iterations, the change of the optimal fitness value is less than a certain minimum threshold ∈ = 10 -6 , then terminate the algorithm and output the current optimal clustering scheme; otherwise, return to Step 2 to continue the next iteration.

[0083] Specifically, in this implementation case, the detection range size of the WSN clustering method based on the Arctic Tern optimization algorithm is set to 400m × 400m, the number of sensors is 100, the positions of the monitored targets and sensor nodes are randomly distributed, and the cluster head ratio is 0.1. The size of the Arctic Tern population is 90, and the number of iterations is 80 times;

[0084] Reference Figure 3, the comparative algorithms for the WSN clustering method based on the Arctic Puffin Optimization Algorithm (APOA) are the Genetic Algorithm (GA) and the Simulated Annealing Algorithm (SA). The population sizes of the above comparative algorithms are all 90, the number of iterations is 80, and the detection range sizes are all 400m×400m; the maximum value of the APOA adaptive adjustment coefficient is 0.9, and the minimum value is 0.2; the selection operation method adopted by GA is roulette wheel selection, the crossover probability is set to 0.8, and the mutation probability is 0.05; SA uses the exponential cooling method to update the temperature, the initial temperature is set to 100, the cooling coefficient is 0.90, generates neighborhood solutions, and decides whether to accept the new solution according to the Metropolis criterion. It can be observed from the simulation curve that the WSN clustering method of APOA basically flattens out at the 25th iteration, and the total energy consumption is the lowest, while GA and SA converge too slowly and the total energy consumption is large. APOA shows better convergence performance in this optimization problem, can find a better solution within fewer iterations, and the final optimal fitness value is also the lowest. Compared with GA and SA, the WSN clustering method based on APOA has better convergence performance and the minimum energy consumption.

[0085] Implementation Case 2: Refer to Figure 2 This embodiment is described in detail. An intelligent WSN clustering device for energy consumption optimization is used to execute the method described in Embodiment 1, and includes a processor, a memory, a data preprocessing module, a communication module, and an energy monitoring module;

[0086] The processor, as the core computing unit of the device, undertakes multiple key tasks. During the operation of the Arctic Puffin Optimization Algorithm (APOA), the processor is responsible for performing the computational work in each stage. In the initialization stage, it generates the Arctic Puffin population according to preset rules, determines the WSN clustering scheme represented by each individual, and calculates the corresponding fitness value. In subsequent stages such as aerial search, dive predation, aggregation foraging, enhanced search, and avoiding predation, the processor updates the position, performs local optimization, information interaction, adjusts the search range, and deals with the local optimum situation of the clustering scheme according to the corresponding formulas and strategies. At the same time, the processor also needs to screen different strategies, and according to the real-time state of the algorithm operation and the preset conditions, judge when to adopt which strategy to ensure that the algorithm advances in a better direction. When the algorithm may fall into a local optimum solution, the processor perturbs the relevant individuals randomly according to the avoiding predation strategy, so that the algorithm jumps out of the local optimum area and continues to search for a better solution;

[0087] The memory is mainly responsible for storing various types of key data, providing data support for the stable operation of the algorithm. Before the algorithm runs, the memory stores relevant information of the sensor nodes, including the number of nodes, their locations, initial energy, etc. During the execution of the algorithm, it saves the clustering schemes generated in each round of iteration, as well as data such as the results of the calculated fitness function. These data are not only the basis for the algorithm to perform the next calculation and decision-making, but also facilitate the subsequent analysis and evaluation of the algorithm operation process. The memory also stores some intermediate variables and parameters during the algorithm operation process to ensure that they can be quickly called and updated when needed, improving the operation efficiency of the algorithm;

[0088] The data preprocessing module preprocesses the raw data collected by the sensor nodes. First, it performs denoising on the data, using a filtering algorithm to remove the noise data generated due to environmental interference and other factors to ensure the accuracy of the data. Then, it extracts features, extracting key features related to clustering, such as the remaining energy and signal strength of the nodes. These preprocessed data will be used as the input for the subsequent clustering algorithm, improving the processing efficiency of the algorithm and the reliability of the clustering results;

[0089] The communication module is responsible for implementing data communication between the sensor nodes and the device, as well as between the device and the base station. It uses an efficient wireless communication protocol and has the function of adaptively adjusting the communication power. It dynamically adjusts the transmission power according to the distance and signal strength between nodes to reduce energy consumption. It supports multi-channel communication, effectively avoiding communication conflicts and improving the stability and reliability of data transmission. During the clustering process, it is responsible for transmitting the information of each clustering scheme between different nodes to achieve information sharing and interaction, promoting the optimization of the clustering scheme;

[0090] The energy monitoring module monitors the energy consumption of the sensor nodes in real time. By collecting and analyzing the node energy data, it accurately calculates the remaining energy of each node and feeds this information back to the processor. The processor preferentially selects nodes with higher remaining energy as cluster heads during the clustering process according to the remaining energy situation of the nodes to balance the network energy consumption and extend the network life cycle. It has an energy warning function. When the energy of some nodes in the network is lower than the set threshold, it timely sends a warning message to the processor so that the processor can take corresponding measures, such as readjusting the clustering scheme or starting an energy-saving mode.

[0091] Although the present invention has been described in terms of a limited number of embodiments, those skilled in the art, having the benefit of the foregoing description, will appreciate that other embodiments can be contemplated within the scope of the invention as thus described. In addition, it should be noted that the language used in this specification has been principally selected for readability and instructional purposes and not to limit or circumscribe the inventive subject matter. Accordingly, many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the appended claims. For the scope of the present invention, the disclosure herein is illustrative, and not restrictive, the scope of the invention being defined by the appended claims.

Claims

1. An intelligent clustering method and device for WSN oriented to energy consumption optimization, characterized in that It includes the following steps: Initialize the puffin population and construct a WSN clustering model, calculate the fitness values of each clustering scheme through a fitness function, design and calculate an adaptive operator, and simulate the behaviors of puffins (such as aerial search, dive predation, and aggregated foraging) to optimize the clustering scheme until the optimal clustering scheme is output after meeting the termination conditions.

2. The method for initializing the puffin population and constructing a WSN clustering model according to claim 1, wherein The puffin population is initialized using a coding method based on location information, and each individual represents a WSN clustering scheme; according to the number of sensor nodes and the expected number of cluster heads in the WSN, two-dimensional coordinates are randomly generated within the monitoring area as the initial cluster head positions to form an initial clustering scheme.

3. The method for calculating the fitness values of each clustering scheme through a fitness function according to claim 1, wherein Construct a fitness function and comprehensively consider the energy consumption balance of nodes in the cluster (E balance ), the distance from the cluster head node to the base station (D toBS ), coverage redundancy (R coverage ), the formula is: Fitness = α × E balance + β × D toBS + γ × (1 - R coverage ) Among them, α, β, and γ are weight coefficients, satisfying α + β + γ = 1, α ≥ 0, β ≥ 0, γ ≥ 0, and the weight coefficients are reasonably set according to the actual application scenarios and requirements.

4. The key parameter calculation method according to claim 3, wherein Let there be N j nodes in the j-th cluster, and the energy consumption of node k be e jk . Then the average energy consumption of this cluster is: The formula for the energy consumption balance of nodes within a cluster is: where m is the number of clusters; Let the coordinates of the \(i\)-th cluster head node be \((x hj , y hj \)), and the coordinates of the base station be \((x bs , y bs \)). Then the distance from the cluster head node to the base station is: Let the sensing radius of the sensor node be r. If the distance d between node i and node l il ≤ 2r, it is considered that their coverage areas overlap, and the coverage redundancy R is obtained by counting the proportion of the overlapping area coverage .

5. The design and calculation of the adaptive operator according to claim 1, wherein Define the adaptive operator λ, which is related to the current iteration number t, the maximum iteration number T, and the population fitness variance σ 2 and the population fitness variance σ 2 reflects the degree of dispersion of the individual fitness in the population, and its calculation formula is: where P is the population size, is the fitness value of the i-th individual at time t, is the population at time t Group average fitness value.

6. The adaptive operator λ according to claim 5, wherein The calculation formula is: Among them, λ min and λ max are the minimum and maximum values of the adaptive operator, which is the maximum value of the population fitness variance preset in advance.

7. The optimization of the clustering scheme by simulating the behaviors of puffins according to claim 1, wherein Step 1: Aerial search; Simulate the behavior of puffins searching for food in the air, adjust the cluster head positions in the clustering scheme, and perform position update operations according to the individual's own position and the global optimal position. The position update formula is: Among them, represents the position of the i-th puffin at time t, represents the global optimal position at time t, and r1 is a random number in the range of [0, 1]; Step 2: Dive predation; When the fitness value of an individual is better than the current population average fitness value, perform local optimization on the clustering scheme corresponding to the individual, randomly generate new position points within a certain range around the cluster head position, recalculate the fitness value, and update the cluster head position if the new value is better; If: Then: Among them, is the average fitness value of the population at time t, and the calculation formula is: where p is the population size, r2 is a small random number in the range of [0, 0.1], and each element is updated using an independent r2 value; ΔX is an offset randomly generated around the cluster head position, and the offset ΔX for the j-th cluster head position j satisfies ΔX j =(Δx j , Δy j ), where Δx j and Δy j are randomly generated within the range of [-δ, δ] respectively, and δ is a preset offset range parameter; Step 3: Aggregated foraging; Information exchange and fusion are carried out between different clustering schemes. Adjacent individuals exchange part of the cluster head position information and then recalculate the fitness value, retaining the better scheme; specifically, it can be expressed as: for individual i and individual j, randomly select part of the cluster head positions for exchange to obtain new individuals i' and j', calculate Fitness(i') and Fitness(j'), if: Fitness(i′)>Fitness(i) then If: Fitness(j′)>Fitness(j) Then Step 4: Enhanced search; When the change in the population fitness value tends to be flat, expand the search range, increase the search randomness, perform local search operations on more individuals, and adjust the proportion of locally searched individuals according to the adaptive operator λ to avoid the algorithm falling into a local optimum prematurely and continuously optimize the clustering scheme; Step 5: Avoid predation; when the fitness value of an individual does not improve after multiple consecutive iterations, randomly perturb it and randomly change the positions of some of its cluster heads to expand the population search range and improve the stability of the algorithm.

8. The algorithm termination condition according to claim 1 is characterized in that: If the change in the optimal fitness value is less than a certain minimum threshold ∈ when the maximum number of iterations is reached or in consecutive k iterations, the algorithm is terminated and the current optimal clustering solution is output; otherwise, return to claim 3 to continue the next iteration.

9. An energy consumption optimization-oriented WSN intelligent clustering device for implementing the method described in claims 1-6, characterized in that, include: Processor, memory, data preprocessing module, communication module and energy monitoring module: The processor: as a core computing unit, performs computing work at each stage of the Arctic puffin optimization algorithm, including generating Arctic puffin populations, determining clustering schemes, calculating fitness values, performing position updates, local optimization, information interaction, and adjusting the search range to adapt to local optimal situations, and is also responsible for selecting strategies according to the algorithm operation status and preset conditions; The memory is used to store the relevant information of the sensor nodes, the clustering scheme generated in each round of iteration, the fitness function results, the intermediate variables and parameters during the operation of the algorithm, and provide data support for the operation of the algorithm; The data preprocessing module performs denoising and feature extraction on the raw data collected by the sensor nodes, extracts key features related to clustering, and improves algorithm processing efficiency and reliability of clustering results; The communication module: realizes data communication between sensor nodes and devices and between devices and base stations, adopts efficient wireless communication protocols, has the functions of adaptively adjusting communication power and supporting multi-channel communication, and promotes clustering scheme optimization; The energy monitoring module monitors the energy consumption of sensor nodes in real time, calculates the remaining energy of each node and feeds it back to the processor. It has an energy early warning function and sends early warning information to the processor when the energy of some nodes is lower than the set threshold.

Images