Underwater acoustic sensor network topology clustering method, system, device, medium and product

Abstract

The application discloses a kind of underwater acoustic sensor network topology clustering method, system, equipment, medium and product, it is related to sensor network topology management field, the method includes: obtaining the state feature of multiple survival nodes in underwater acoustic sensor network;And construct shared Q value table;Further determine the cluster head selection preference score of each survival node;According to the preset fixed cluster head number, based on node constraint, and using greedy Top-K algorithm to determine cluster head set;Determine the individualized reward and state transition information of each survival node in underwater acoustic sensor network;Based on time series difference method, the shared Q value table is trained, and the trained shared Q value table is obtained;According to the current state feature of multiple survival nodes in underwater acoustic sensor network, the current cluster head set is determined using the trained shared Q value table;And according to cluster head set, underwater data is collected.The application can improve the coverage and throughput of underwater acoustic sensor network, realize large-scale, long-period underwater monitoring.

Description

Technical Field

[0001] This application relates to the field of sensor network topology management, and in particular to a method, system, device, medium and product for topology clustering of underwater acoustic sensor networks. Background Technology

[0002] With the continuous development of fields such as marine observation, resource exploration, and underwater environmental monitoring, underwater acoustic sensor networks, as a critical infrastructure, are becoming increasingly large and complex, placing higher demands on network energy efficiency, coverage, and real-time data transmission. Traditional underwater network topology clustering methods, such as cluster head selection strategies based on random rotation, fixed rules, or greedy heuristics, while computationally simple, often face problems such as unstable cluster head rotation, an increase in orphan nodes, and uneven energy consumption in dynamic, time-varying, and energy-constrained underwater environments, leading to incomplete network coverage, decreased throughput, and increased transmission latency.

[0003] Adaptive and intelligent topology management methods, as an emerging technology, can dynamically adjust cluster head selection strategies based on network conditions to optimize overall performance. However, existing methods based on optimization algorithms or multi-agent reinforcement learning still have certain limitations: such as reliance on prior knowledge, difficulty in online updates, susceptibility to energy hotspots when nodes move or have uneven density, slow convergence speed, high communication overhead, or routing instability and packet loss in the early stages of learning due to policy non-convergence.

[0004] Therefore, based on the above problems, there is an urgent need to provide a topology clustering method for underwater acoustic sensor networks, which can improve the coverage and throughput of underwater acoustic sensor networks and realize large-scale, long-term underwater monitoring. Summary of the Invention

[0005] The purpose of this application is to provide a method, system, device, medium and product for topology clustering of underwater acoustic sensor networks, which can improve the coverage and throughput of underwater acoustic sensor networks and realize large-scale, long-term underwater monitoring.

[0006] To achieve the above objectives, this application provides the following solution: In a first aspect, this application provides a method for topology clustering of underwater acoustic sensor networks, including: Acquire the state characteristics of multiple surviving nodes in an underwater acoustic sensor network; the state characteristics include: remaining energy, distance characteristics, energy consumption rate, and depth information; Based on the state characteristics, a shared Q-value table is constructed; and based on the shared Q-value table, the cluster head selection preference score of each surviving node is determined. Based on the cluster head selection preference score and a preset fixed number of cluster heads, the cluster head set is determined using a greedy Top-K algorithm based on node constraints; the node constraints include cluster head spacing constraints and cooling time constraints. Network communication is performed based on the cluster head set, and individualized reward and state transition information for each surviving node in the underwater acoustic sensor network is determined. Based on individualized reward and state transition information, the shared Q-value table is trained using the temporal difference method to obtain a trained shared Q-value table. Based on the current state characteristics of multiple surviving nodes in the underwater acoustic sensor network, the current cluster head set is determined using a pre-trained shared Q-value table; and underwater data is collected based on the cluster head set.

[0007] Secondly, this application provides an underwater acoustic sensor network topology clustering system, comprising: A state feature determination module is used to acquire the state features of multiple surviving nodes in an underwater acoustic sensor network; the state features include: remaining energy, distance features, energy consumption rate, and depth information. A shared Q-value table construction module is used to construct a shared Q-value table based on the state characteristics; and to determine the cluster head selection preference score of each surviving node based on the shared Q-value table; The cluster head set determination module is used to determine the cluster head set based on the cluster head selection preference score and a preset fixed number of cluster heads, using a greedy Top-K algorithm based on node constraints; the node constraints include cluster head spacing constraints and cooling time constraints. The reward module is used to conduct network communication based on the cluster head set and to determine the individualized reward and state transition information for each surviving node in the underwater acoustic sensor network. The shared Q-value table training module is used to train the shared Q-value table based on the temporal difference method according to individualized reward and state transition information, so as to obtain the trained shared Q-value table. The data acquisition module is used to determine the current cluster head set based on the current state characteristics of multiple surviving nodes in the underwater acoustic sensor network and a pre-trained shared Q-value table; and to acquire underwater data based on the cluster head set.

[0008] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described underwater acoustic sensor network topology clustering method.

[0009] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described underwater acoustic sensor network topology clustering method.

[0010] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described underwater acoustic sensor network topology clustering method.

[0011] According to the specific embodiments provided in this application, this application has the following technical effects: This application provides a topology clustering method for underwater acoustic sensor networks. By acquiring the state characteristics of multiple surviving nodes in the underwater acoustic sensor network and constructing a shared Q-value table, it further determines the cluster head selection preference score of each surviving node, quantifying the cluster head selection and optimizing its quality. Based on a preset fixed number of cluster heads and node constraints, it uses a greedy Top-K algorithm to determine the cluster head set, avoiding nodes from continuously serving as cluster heads under the condition of a fixed number of cluster heads, thus achieving stable optimization of cluster head selection. Finally, it determines the individualized reward and state transition information for each surviving node in the underwater acoustic sensor network. Based on the temporal difference method, a shared Q-value table is trained, and the current cluster head set is determined using the trained shared Q-value table. Data is then collected. The introduction of individualized rewards effectively reduces the existence of orphan nodes, and optimizes the overall performance of coverage, throughput, and latency while balancing energy consumption. In summary, this application overcomes the problems of unstable cluster head selection, uneven energy, and low coverage efficiency in dynamic underwater environments of traditional clustering methods, and significantly improves the adaptability, stability, and energy efficiency of topology clustering. It is suitable for data collection and topology management scenarios of underwater sensor networks such as marine monitoring and seabed exploration. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a flowchart illustrating a method for topology clustering of an underwater acoustic sensor network according to an embodiment of this application. Figure 2 This is a flowchart illustrating a specific method for topology clustering of an underwater acoustic sensor network in one embodiment of this application. Detailed Implementation

[0014] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0015] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0016] In one exemplary embodiment, such as Figure 1 and Figure 2 As shown, a method for topology clustering of underwater acoustic sensor networks is provided, including the following steps S1 to S6. Wherein: S1: Obtain the state characteristics of multiple surviving nodes in the underwater acoustic sensor network.

[0017] The initial state features of multiple surviving nodes are obtained and normalized to obtain state features, which include remaining energy, distance features, energy consumption rate, and depth information. The distance feature is the nearest distance between the current node and the last selected cluster head. State features can be represented by feature vectors, the expression of which is as follows: ; in, for Nodes that are always alive State feature vector, for Nodes that are always alive The remaining energy, for Nodes that are always alive Distance characteristics, for Nodes that are always alive The rate of energy consumption, for Nodes that are always alive Depth information.

[0018] After obtaining the state vector, the feature vector is uniformly discretized into discrete state indices.

[0019] S2: Construct a shared Q-value table based on the state characteristics; and determine the cluster head selection preference score for each surviving node based on the shared Q-value table.

[0020] Let the discretized state space in the underwater acoustic sensor network be... The action space is And construct a dimension based on the state space and action space. Two-dimensional matrix ,in, This represents the total number of discrete states in the state space. This represents the total number of discrete actions in the action space. In this application, the action space only includes two discrete actions: acting as a cluster head and not acting as a cluster head. In the initial stage, a two-dimensional matrix is ​​used... All state-action pairs If the values ​​are uniformly initialized to 0 (or a very small random constant), then the initial shared Q-value table is... This indicates that there has been no interaction with the underwater environment at this stage, and any action taken by a surviving node in any state is without any preset preference. As the shared Q-value table is updated, these Q-values, initially set to 0, will be continuously corrected based on the node's individualized rewards and state transition information, using a temporal difference method.

[0021] The cluster head selection preference score for each surviving node can be determined using the shared Q-value table. This score is composed of a Q-value dominance term (the Q-value of selecting the right cluster head versus the Q-value of selecting the wrong cluster head), a time-decreasing exploration incentive term, and random small perturbations. The formula for calculating the cluster head selection preference score is as follows: ; in, Select a preference score for the cluster head. Current state The next choice is the Q value of the cluster head. Current state The Q value of an inappropriately selected cluster head. For exploration incentives that decay over time, For random small perturbations (small random perturbations that break a tie).

[0022] The corresponding Q value can be obtained by looking up the shared Q value table using the discrete state index.

[0023] S3: Based on the cluster head selection preference score and the preset fixed number of cluster heads, determine the cluster head set using a greedy Top-K algorithm based on node constraints.

[0024] Node constraints include cluster head spacing constraints and cooling time constraints. Depending on the specific circumstances, node constraints can be selected from either cluster head spacing constraints or cooling time constraints, or both. In this application, both cluster head spacing constraints and cooling time constraints are used to determine the cluster head set. S3 specifically includes: S31: Select candidate cluster heads from surviving nodes based on the cooldown time constraint.

[0025] The cooldown time constraint is used to limit the continuous holding of a cluster head by a surviving node. Specifically, if a surviving node is elected as the cluster head in this round, its cooldown time is set to a positive integer. If not selected, the cooldown time decreases by 1 in each iteration until it reaches 0. Only surviving nodes with a cooldown time of 0 satisfy the cooldown time constraint and are eligible to be candidates for cluster head in the next round. This application selects only surviving nodes with a cooldown time of 0 as candidate cluster heads to ensure fairness in the rotation.

[0026] S32: Based on the candidate cluster heads, and considering the cluster head selection preference score and cluster head spacing constraint, the initial cluster head set is determined using the greedy Top-K algorithm.

[0027] Among the candidate cluster heads, a greedy Top-K algorithm is used to select cluster heads whose cluster head selection preference scores are greater than a preset score threshold and who meet the cluster head spacing constraint, and an initial cluster head set is constructed. The cluster head selection preference score is used as the ranking criterion for cluster head selection, and the cluster head spacing constraint is used to remove cluster heads whose cluster head spacing is greater than a preset distance threshold.

[0028] S33: In the initial cluster head set, when the number of cluster heads is less than the preset fixed number of cluster heads, the cluster head spacing constraint is adjusted until the number of cluster heads is the same as the preset fixed number of cluster heads, thus obtaining the cluster head set.

[0029] Specifically, under the condition of satisfying the cooling time constraint (time cooling mechanism), the surviving nodes with high cluster head selection preference scores and satisfying the cluster head spacing constraint are selected as cluster heads first. If the preset distance threshold in the cluster head spacing constraint makes it impossible to select K cluster heads, then the process enters the relaxation phase, and the preset distance threshold is adjusted to allow the selection of cluster heads to continue under the premise of moderately relaxing the preset distance threshold, so as to ensure that the number of cluster heads reaches the preset fixed number of cluster heads K.

[0030] S4: Perform network communication based on the cluster head set and determine the individualized reward and state transition information for each surviving node in the underwater acoustic sensor network.

[0031] S4 specifically includes: S41: Based on the cluster head set, network communication is performed to determine the network parameters of the underwater acoustic sensor network and the state transition information of each surviving node.

[0032] Network communication and data collection are performed based on the cluster head set to obtain the state characteristics of multiple surviving nodes in the next round and the network parameters of the underwater acoustic sensor network. Network parameters include network coverage, network throughput, and latency. Based on the state characteristics of multiple surviving nodes in the current round and the next round, the state transition information for each surviving node can be obtained.

[0033] S42: Determine the global reward of the underwater acoustic sensor network based on the network parameters.

[0034] S43: Based on the global reward, use vectorized reward shaping technology to determine the individualized reward for each surviving node.

[0035] This application, based on the global reward, utilizes vectorized reward shaping technology to add a penalty to orphan nodes and provides each surviving node with an individualized reward related to network coverage, network throughput, and latency metrics, thereby optimizing credit allocation and subsequent updates to the shared Q-value table. The individualized reward calculation formula is as follows: ; in, Individualized rewards for each surviving node. For the global reward of the underwater acoustic sensor network, As an additional reward based on node performance, , As the first preset positive weight, For network throughput, As the second preset positive weight, Data transmission delay, also known as latency metric. and This is to ensure that underwater acoustic sensor networks tend to seek out topologies with high throughput and low latency. This is a penalty for orphaned nodes, applied when a surviving node becomes an orphan node because it fails to establish a communication connection with any cluster head in the current topology. Take a preset negative constant; when a surviving node successfully connects to the cluster head, Take 0.

[0036] This application decomposes the global performance metric (global reward) into node-level learning signals (individualized rewards), which improves the stability and convergence efficiency of policy updates.

[0037] S5: Based on individualized reward and state transition information, the shared Q-value table is trained using the temporal difference method to obtain a trained shared Q-value table.

[0038] Based on individualized rewards and the state transition information determined in S4, the shared Q-value table is trained in multiple rounds using the temporal difference method in a multi-round simulation environment. During the training process, a sleep perception model and a probabilistic acoustic link model are used to simulate the dynamic performance and energy consumption process of the underwater acoustic sensor network. The steps S1-S5 are repeated multiple times to form an interactive-trial-update closed-loop iterative process, continuously optimizing the cluster head selection strategy of surviving nodes until the shared Q-value table converges, thus obtaining the trained shared Q-value table.

[0039] S6: Based on the current state characteristics of multiple surviving nodes in the underwater acoustic sensor network, determine the current cluster head set using a trained shared Q-value table; and collect underwater data based on the cluster head set.

[0040] The trained shared Q-value table is deployed on the actual underwater acoustic sensor network to determine the current cluster head set, and the current cluster head set is used to collect underwater data, achieving low-overhead and highly stable online topology clustering.

[0041] This application achieves dynamic optimization and stable operation of underwater acoustic sensor network topology clustering through cluster head selection preference scoring and a greedy Top-K algorithm. Utilizing a lightweight Q-learning strategy, combined with cluster head spacing constraints and cooling time constraints, this application adaptively selects cluster heads, effectively reducing orphan nodes, balancing the energy consumption of the underwater acoustic sensor network, and thus synergistically improving network coverage and throughput while reducing transmission latency.

[0042] In one exemplary embodiment, the underwater acoustic sensor network topology clustering method consists of two parts: an offline training phase and an online deployment phase. In the offline training phase, the underwater acoustic sensor network is first initialized, and the following steps are performed in each iteration: The state features of all surviving nodes are collected and converted into feature vectors, which are then discretized into discrete state indices; based on the discrete state indices and a shared Q-value table, the cluster head selection preference score of each surviving node is determined; based on the cluster head selection preference score and a preset fixed number of cluster heads, a cluster head set is determined using a greedy Top-K algorithm, based on node constraints; network communication is performed, and the network parameters of the underwater acoustic sensor network (network coverage, network throughput, and latency) are determined; the global reward is calculated based on the network parameters, and the individualized reward and state transition information of each surviving node are further determined using a vectorized reward shaping technique; finally, the shared Q-value table is updated using a temporal difference method to complete the policy optimization; this process is repeated multiple times until the shared Q-value table converges. During the online deployment phase, a pre-trained shared Q-value table is used. Only the steps of state feature acquisition, cluster head selection preference score calculation, and cluster head selection are performed. The pre-trained shared Q-value table is pre-configured or distributed to the memory of each surviving node in the form of a lightweight lookup table. During online operation, only low-complexity table lookup and basic arithmetic operations are involved. There is no need for floating-point number updates and Q-value information exchange between nodes, which realizes lightweight and stable online cluster head selection and completes continuous topology optimization of the dynamic underwater acoustic sensor network.

[0043] In one exemplary embodiment, the method of this application, the Hybrid, Energy-Efficient, Distributed (HEED) protocol, and the Low-Energy Adaptive Clustering Hierarchy (LEACH) protocol were used to traverse the surviving nodes (10-100). The traversal test results show that, compared with HEED and LEACH, this application improves the coverage by 5.4% and 3.7%, respectively, increases the throughput by 12.2% and 4.9%, respectively, and reduces the latency by 8.4% and 7.0%, respectively. That is, this application has a better trade-off between coverage, throughput, and latency.

[0044] Compared to traditional fixed-rule or centralized optimization methods, this application not only significantly enhances the adaptability and robustness of topology management in dynamic and time-varying underwater environments, but also balances algorithm performance and deployment feasibility through offline training and low-overhead online decision-making mechanisms. It is particularly suitable for large-scale underwater long-term monitoring scenarios with limited energy and complex channels, and has important practical value and broad application prospects.

[0045] The selection and quantification methods of the above state features, the variations and replacements of algorithms related to the shared Q-value table, the selection and adjustment of node constraints, the shaping method of the reward function, and the topology construction strategy can all be adapted and modified according to the properties of other sensor networks or mobile ad hoc networks.

[0046] Based on the same inventive concept, this application also provides an underwater acoustic sensor network topology clustering system applied to the underwater acoustic sensor network topology clustering method. The solution provided by this system is similar to the implementation scheme described in the above-described underwater acoustic sensor network topology clustering method. Therefore, the specific limitations of one or more underwater acoustic sensor network topology clustering system embodiments provided below can be found in the limitations of the underwater acoustic sensor network topology clustering method described above, and will not be repeated here.

[0047] In one exemplary embodiment, an underwater acoustic sensor network topology clustering system is provided, comprising: A state feature determination module is used to acquire the state features of multiple surviving nodes in an underwater acoustic sensor network; the state features include: remaining energy, distance features, energy consumption rate, and depth information. A shared Q-value table construction module is used to construct a shared Q-value table based on the state characteristics; and to determine the cluster head selection preference score of each surviving node based on the shared Q-value table; The cluster head set determination module is used to determine the cluster head set based on the cluster head selection preference score and a preset fixed number of cluster heads, using a greedy Top-K algorithm based on node constraints; the node constraints include cluster head spacing constraints and cooling time constraints. The reward module is used to conduct network communication based on the cluster head set and to determine the individualized reward and state transition information for each surviving node in the underwater acoustic sensor network. The shared Q-value table training module is used to train the shared Q-value table based on the temporal difference method according to individualized reward and state transition information, so as to obtain the trained shared Q-value table. The data acquisition module is used to determine the current cluster head set based on the current state characteristics of multiple surviving nodes in the underwater acoustic sensor network and a pre-trained shared Q-value table; and to acquire underwater data based on the cluster head set.

[0048] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal. The computer device includes a processor, memory, input / output interfaces (I / O), and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is connected to the system bus via the I / O interfaces. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the computer device stores underwater acoustic sensor network topology clustering data. The input / output interfaces of the computer device are used for exchanging information between the processor and external devices. The communication interface of the computer device is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements an underwater acoustic sensor network topology clustering method.

[0049] In one exemplary embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps in the embodiment of the underwater acoustic sensor network topology clustering method.

[0050] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the embodiment of the underwater acoustic sensor network topology clustering method.

[0051] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the embodiment of the underwater acoustic sensor network topology clustering method.

[0052] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0053] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0054] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, etc., and are not limited to these.

[0055] The technical features of the above embodiments can be combined in any way. For the sake of brevity, 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.

[0056] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for topology clustering of underwater acoustic sensor networks, characterized in that, The underwater acoustic sensor network topology clustering method includes: Acquire the state characteristics of multiple surviving nodes in an underwater acoustic sensor network; the state characteristics include: remaining energy, distance characteristics, energy consumption rate, and depth information; Based on the state characteristics, a shared Q-value table is constructed; and based on the shared Q-value table, the cluster head selection preference score of each surviving node is determined. Based on the cluster head selection preference score and a preset fixed number of cluster heads, the cluster head set is determined using a greedy Top-K algorithm based on node constraints; the node constraints include cluster head spacing constraints and cooling time constraints. Network communication is performed based on the cluster head set, and individualized reward and state transition information for each surviving node in the underwater acoustic sensor network is determined. Based on individualized reward and state transition information, the shared Q-value table is trained using the temporal difference method to obtain a trained shared Q-value table. Based on the current state characteristics of multiple surviving nodes in the underwater acoustic sensor network, the current cluster head set is determined using a pre-trained shared Q-value table; and underwater data is collected based on the cluster head set.

2. The underwater acoustic sensor network topology clustering method according to claim 1, characterized in that, The acquisition of the state characteristics of multiple surviving nodes in the underwater acoustic sensor network specifically includes: Using formula Determine state characteristics; in, for Nodes that are always alive State feature vector, for Nodes that are always alive The remaining energy, for Nodes that are always alive Distance characteristics, for Nodes that are always alive The rate of energy consumption, for Nodes that are always alive Depth information.

3. The underwater acoustic sensor network topology clustering method according to claim 1, characterized in that, The determination of the cluster head selection preference score for each surviving node based on the shared Q-value table specifically includes: Using formula Determine the cluster head selection preference score for each surviving node; in, Choose a preference score for the cluster head. Current state The next choice is the Q value of the cluster head. Current state The Q value of an inappropriately selected cluster head. For exploration incentives that decay over time, It represents a random, small perturbation.

4. The underwater acoustic sensor network topology clustering method according to claim 1, characterized in that, The process of determining the cluster head set based on cluster head selection preference scores and a preset fixed number of cluster heads, using node constraints and a greedy Top-K algorithm, specifically includes: Based on the cooldown time constraint, candidate cluster heads are selected from the surviving nodes; Based on the candidate cluster heads, and considering the cluster head selection preference scores and cluster head spacing constraints, the greedy Top-K algorithm is used to determine the initial cluster head set. In the initial cluster head set, when the number of cluster heads is less than the preset fixed number of cluster heads, the cluster head spacing constraint is adjusted until the number of cluster heads is the same as the preset fixed number of cluster heads, thus obtaining the cluster head set.

5. The underwater acoustic sensor network topology clustering method according to claim 1, characterized in that, The step of performing network communication based on the cluster head set and determining the individualized reward and state transition information for each surviving node in the underwater acoustic sensor network specifically includes: Network communication is performed based on the cluster head set to determine the network parameters of the underwater acoustic sensor network and the state transition information of each surviving node; the network parameters include: network coverage, network throughput, and latency. Based on the network parameters, determine the global reward of the underwater acoustic sensor network; Based on the global reward, a vectorized reward shaping technique is used to determine the individualized reward for each surviving node.

6. The underwater acoustic sensor network topology clustering method according to claim 5, characterized in that, The step of determining the individualized reward for each surviving node based on the global reward using vectorized reward shaping technology specifically includes: Using formula Determine the individualized reward for each surviving node; in, Individualized rewards for each surviving node. For the global reward of the underwater acoustic sensor network, As an additional reward based on node performance, , For orphan nodes, penalty items, As the first preset positive weight, For network throughput, As the second preset positive weight, This refers to data transmission latency.

7. A topology clustering system for underwater acoustic sensor networks, characterized in that, The underwater acoustic sensor network topology clustering system includes: A state feature determination module is used to acquire the state features of multiple surviving nodes in an underwater acoustic sensor network; the state features include: remaining energy, distance features, energy consumption rate, and depth information. A shared Q-value table construction module is used to construct a shared Q-value table based on the state characteristics; and to determine the cluster head selection preference score of each surviving node based on the shared Q-value table. The cluster head set determination module is used to determine the cluster head set based on the cluster head selection preference score and a preset fixed number of cluster heads, using a greedy Top-K algorithm based on node constraints; the node constraints include cluster head spacing constraints and cooling time constraints. The reward module is used to conduct network communication based on the cluster head set and to determine the individualized reward and state transition information for each surviving node in the underwater acoustic sensor network. The shared Q-value table training module is used to train the shared Q-value table based on the temporal difference method according to individualized reward and state transition information, so as to obtain the trained shared Q-value table. The data acquisition module is used to determine the current cluster head set based on the current state characteristics of multiple surviving nodes in the underwater acoustic sensor network and a trained shared Q-value table; and to acquire underwater data based on the cluster head set.

8. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the underwater acoustic sensor network topology clustering method according to any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the underwater acoustic sensor network topology clustering method as described in any one of claims 1-6.

10. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the underwater acoustic sensor network topology clustering method as described in any one of claims 1-6.

Images