CN114599069B - Underwater wireless sensor network routing method based on energy self-collection - Google Patents
Underwater wireless sensor network routing method based on energy self-collection Download PDFInfo
- Publication number
- CN114599069B CN114599069B CN202210208771.4A CN202210208771A CN114599069B CN 114599069 B CN114599069 B CN 114599069B CN 202210208771 A CN202210208771 A CN 202210208771A CN 114599069 B CN114599069 B CN 114599069B
- Authority
- CN
- China
- Prior art keywords
- sensor network
- node
- underwater
- underwater sensor
- nodes
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 33
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 43
- 241000251468 Actinopterygii Species 0.000 claims abstract description 23
- 230000033001 locomotion Effects 0.000 claims abstract description 22
- 238000004088 simulation Methods 0.000 claims abstract description 20
- 238000005094 computer simulation Methods 0.000 claims abstract description 10
- 238000013135 deep learning Methods 0.000 claims abstract description 10
- 238000003062 neural network model Methods 0.000 claims abstract description 10
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 7
- 230000006870 function Effects 0.000 claims description 23
- 239000010410 layer Substances 0.000 claims description 23
- 230000000007 visual effect Effects 0.000 claims description 23
- 230000007246 mechanism Effects 0.000 claims description 15
- 238000000605 extraction Methods 0.000 claims description 14
- 238000012549 training Methods 0.000 claims description 11
- 230000008569 process Effects 0.000 claims description 9
- 230000016776 visual perception Effects 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 5
- 238000005303 weighing Methods 0.000 claims description 5
- 239000011229 interlayer Substances 0.000 claims description 4
- 230000004438 eyesight Effects 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 2
- 238000011176 pooling Methods 0.000 description 9
- 238000011156 evaluation Methods 0.000 description 7
- 230000006399 behavior Effects 0.000 description 6
- 230000005540 biological transmission Effects 0.000 description 6
- 230000008447 perception Effects 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 230000004927 fusion Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 238000013461 design Methods 0.000 description 3
- 238000005265 energy consumption Methods 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000019637 foraging behavior Effects 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 229920002430 Fibre-reinforced plastic Polymers 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000004134 energy conservation Methods 0.000 description 1
- 239000011151 fibre-reinforced plastic Substances 0.000 description 1
- 230000002431 foraging effect Effects 0.000 description 1
- 238000007499 fusion processing Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000002787 reinforcement Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000008054 signal transmission Effects 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W40/00—Communication routing or communication path finding
- H04W40/02—Communication route or path selection, e.g. power-based or shortest path routing
- H04W40/04—Communication route or path selection, e.g. power-based or shortest path routing based on wireless node resources
- H04W40/10—Communication route or path selection, e.g. power-based or shortest path routing based on wireless node resources based on available power or energy
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/06—Testing, supervising or monitoring using simulated traffic
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W40/00—Communication routing or communication path finding
- H04W40/02—Communication route or path selection, e.g. power-based or shortest path routing
- H04W40/20—Communication route or path selection, e.g. power-based or shortest path routing based on geographic position or location
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W84/00—Network topologies
- H04W84/18—Self-organising networks, e.g. ad-hoc networks or sensor networks
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Geometry (AREA)
- Computer Graphics (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an underwater wireless sensor network routing method based on energy self-collection, which comprises the following steps: laying an underwater sensor based on a water area to be measured to obtain an underwater sensor network node; constructing an underwater sensor network model based on the underwater sensor network nodes and the updated artificial fish swarm algorithm; constructing a dynamic three-dimensional simulation model based on the deep learning neural network model; and carrying out dynamic simulation on the underwater sensor network model based on the dynamic three-dimensional simulation model to realize intelligent control of the underwater sensor network. The invention can effectively simulate the working motion of the nodes in the whole network, and greatly enhances the scalability and robustness of the dynamic three-dimensional model.
Description
Technical Field
The invention belongs to the field of underwater wireless sensor networks, and particularly relates to an underwater wireless sensor network routing method based on energy self-collection.
Background
The underwater wireless sensor network is a multi-hop self-organizing network system formed by applying the wireless sensor network to the water, using an aircraft, a submarine or a surface ship to randomly arrange a large number (from hundreds to thousands) of inexpensive micro sensor nodes in a water area of interest, and cooperatively sensing, collecting and processing information of a sensing object in a network coverage area by the nodes through an underwater acoustic wireless communication mode and transmitting the information to a receiver.
In recent years, the technology of an underwater wireless sensor network is widely focused at home and abroad, and is widely used for marine data acquisition, pollution prediction, ocean exploitation, ocean monitoring and the like. For the research of the underwater wireless sensor network, relevant technicians of the navigation college of the northwest industrial university of China put forward the underwater wireless sensor network, the article analyzes a plurality of difficulties of node positioning, sensor network energy, target positioning and the like faced in the design of the underwater wireless sensor network, and based on the system energy problem, various topology algorithms are put forward for the effective calculation of the sensor energy, including a minimum energy path algorithm (LBMER), HCAN, CDS (Connecteddominating set), JSD, ant colony algorithm and the like, and further, the routing protocol of the wireless sensor network is put forward to be based on energy priority. The design goal of the routing protocol is to reduce energy consumption as much as possible, improve the effectiveness of energy usage, and avoid overuse of low-energy nodes, thereby prolonging the network life cycle. Meanwhile, the routing protocol cannot occupy a large amount of storage space, so that the calculated amount is reduced as much as possible. The researchers in China have paid attention to the key effect of the routing protocol on the energy conservation of the whole system, and the transmission capacity and reliability of the current underwater Wireless Sensor Network (WSN) are mostly limited by the following factors, namely the interference of the underwater noise environment, the complex underwater environment and the difficulty of replacing or charging the power of a battery. In a Wireless Sensor Network (WSN) system with energy self-collection capability, the difficulty of receiving and transmitting information is solved to a certain extent by energy accumulation of a single node, and the routing protocol with cooperation capability is particularly important to improve the working efficiency among a large number of large-scale nodes.
In the prior art, the underwater operation environment and the underwater operation characteristics are not considered, and due to the importance of the underwater sensor network to the development of future ocean industry, a technology is urgently needed to solve the problems of energy consumption of sensor nodes and transmission stability of the whole network in the existing underwater sensor network.
Disclosure of Invention
The invention coordinates the receiving and transmitting work of the wireless sensor network node with the energy self-collecting capability under water through a new energy optimization routing protocol, and improves the transmission efficiency and reliability of the wireless sensor network under water.
In order to achieve the above object, the present invention provides the following solutions: an underwater wireless sensor network routing method based on energy self-collection comprises the following steps:
a plurality of underwater sensors are arranged on the basis of a water area to be measured, and a plurality of underwater sensor network nodes are obtained; constructing an underwater sensor network model based on the underwater sensor network node and the updated artificial fish swarm algorithm; constructing a dynamic three-dimensional simulation model based on the deep learning neural network model; and carrying out dynamic simulation on the underwater sensor network model based on the dynamic three-dimensional simulation model to realize intelligent control of the underwater sensor network.
Preferably, the constructing the underwater sensor network model further comprises positioning, layering and clustering the underwater sensor network nodes to obtain inter-layer distances and node distances of the underwater sensor network nodes, and constructing the underwater sensor network model based on the inter-layer distances and the node distances.
Preferably, the interlayer distance is obtained by layering the distance from the underwater sensor network node to the route;
the node distance is the distance from the node of the same-layer underwater sensor network to the route.
Preferably, a data set is obtained by collecting working state images of the underwater sensor network nodes, the data set is divided into a training set and a verification set, the deep learning neural network model is trained, and a dynamic three-dimensional simulation model is constructed.
Preferably, the updating process of the artificial fish swarm algorithm comprises the following steps:
initializing artificial fish swarm algorithm parameters; calculating the self energy value of the underwater sensor network node and initializing a bulletin board; and carrying out iterative updating of the optimal node on the underwater sensor network node based on a dynamic weighing factor strategy, carrying out parameter secondary initialization based on the optimal node until the preset iterative times are reached, outputting an optimal solution in the bulletin board after the iteration of the artificial fish swarm algorithm is ended, obtaining an optimal focusing position according to the optimal solution, selecting a window, and finishing updating.
Preferably, the artificial fish swarm algorithm parameters include an underwater sensor network scale X, an initial position node sensing Visual field of each underwater sensor network node xi, a Step length Step of a node and a neighbor node, a node crowding degree delta in a specific area, a maximum Try number try_number of a route searching node, a current iteration number k and a maximum iteration number kmax.
Preferably, the process of constructing the dynamic three-dimensional simulation model based on the deep learning neural network model is that a moving image of the underwater sensor network node is collected, and a visual twin network model is constructed by simulating a three-dimensional environment; based on the vision twin network model, training the underwater sensor network model by setting a reward function, and constructing the dynamic three-dimensional simulation model.
Preferably, the dynamic simulation comprises a training phase and an reasoning phase;
the training stage is characterized in that a strategy function of a near-end strategy optimization algorithm is constructed based on a visual perception twin network algorithm of a multi-scale attention mechanism, and the underwater sensor network model is trained through the strategy function;
and the reasoning stage controls the motion of the underwater sensor network node through the dynamic three-dimensional simulation model to dynamically simulate the underwater sensor network model.
Preferably, the vision twin network model comprises an information extraction module and a decision module;
the visual twin network model processes the moving image through the information extraction module to obtain image extraction information;
and the visual twin network model fuses the image extraction information through the decision module, accelerates convergence through residual connection, and outputs the movement speed and movement direction of the underwater sensor network node.
The invention discloses the following technical effects:
according to the invention, an artificial fish swarm algorithm and a dynamic simulation algorithm are adopted, the artificial fish swarm algorithm can effectively find out the optimal solution in the whole network, on the basis of the improved artificial fish swarm algorithm, the description of the working path of the intelligent network node is realized by avoiding finding out the optimal solution on the basis of the original artificial fish swarm algorithm, and the convergence and the operation speed of the algorithm are effectively enhanced by setting dynamic weighing factors.
The self-balancing of the nodes and the relative balance among the nodes of the whole network can be accurately controlled based on the routing protocol set by the thought of improving the artificial fish swarm algorithm, the stability and the working efficiency of the whole network are facilitated, the dynamic simulation of the node motion state is performed based on a human left-right eye image recognition method through a deep learning neural network model, the working state simulation of each node in the whole network can be realized by combining the change trend of the network nodes in the path diagram through a system path diagram in the whole network, in the subsequent training, the working state of the node can be simulated based on the deep learning neural network model, the node state trend simulation is realized, and the intelligent control of the whole network is realized.
The invention takes the visual node image as input, is more similar to the real network node working movement, can effectively simulate the node working movement in the whole network, and effectively enhances the scalability and the robustness of the dynamic three-dimensional model.
The method is based on a three-dimensional dynamic simulation method, and according to the working characteristics of the existing node network, the method is used for carrying out targeted setting and creatively introducing strategy functions for data processing, so that the dynamic simulation of the node working state is realized.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method according to an embodiment of the present invention;
fig. 2 is an algorithm updating flow chart of the artificial fish swarm algorithm according to the embodiment of the invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In order that the above-recited objects, features and advantages of the present invention will become more readily apparent, a more particular description of the invention will be rendered by reference to the appended drawings and appended detailed description.
As shown in fig. 1, the invention provides an underwater wireless sensor network routing method based on energy self-collection, which comprises the following steps:
1. constructing underwater sensor network model
In an underwater sensor network with energy self-collection capability, the interaction of the rate of energy collection and the energy threshold of sensor-initiated transmission affects the transmission capability and reliability of the whole network, if each sensor of the underwater sensor network is regarded as a transmission node, the rate of energy collection and the rate of energy consumption of signal transmission of a single node need to ensure a relative balance, and a relative balance needs to be maintained between each node, so that the node positions of the underwater sensor network need to be positioned, and the priority between the relative balance between the nodes and the self-balance of the single node needs to be determined.
Node position location for underwater sensor network
The invention aims to better study the whole network, and aims to perform layering treatment on the whole network according to the distance from each node to the route to obtain the layer spacing, determine the node spacing based on the distance from the node to the route of each layer, perform underwater sensor network model construction through the layer spacing and the node spacing, and analyze the influence of the relative balance among the nodes on the running stability of the whole network based on the sensor coverage of each node and the condition of the node overlapping area of the content of a specific area.
Determining priority between relative balances between nodes and self-balances of individual nodes
For a specific area, if the area includes primary nodes and secondary nodes, the primary nodes can completely cover the specific area, and the secondary nodes are partially covered, when the primary nodes are in an energy shortage state, the secondary nodes are required to perform cooperative work, if the specific area includes two or more primary nodes and the primary nodes are also responsible for sensing work of other areas, the importance of each primary node needs to be judged, if the area covered by an independent primary node needs to be considered firstly, the self-balancing situation of the independent nodes needs to be considered, and if the area covered by a plurality of primary nodes needs to be considered, the relative balancing situation among the nodes needs to be considered, and if the self-balancing and the relative balancing situation is likely to exist, the invention aims at determining the optimal local nodes by determining a method of the optimal local solution, and calculating the working state of each node on the premise of ensuring smooth network operation according to the network operation situation.
Aiming at the underwater sensor network, one or more optimal nodes are necessarily present in the whole network, a section of path can be effectively connected in series by a method for sequentially searching the optimal nodes in adjacent nodes by starting from the optimal nodes, after the path is constructed, the optimal nodes in the network are restarted to be searched, repeated optimal nodes are removed, the optimal nodes in the network are continuously searched under the current condition, and the like, a plurality of paths are constructed by the thought, and the repeated nodes in the paths are removed, so that an underwater sensor network model can be constructed.
Step 1: according to the node condition of the existing underwater sensor network, the method defines the scale X= { X of the algorithm based on each node 1 ，x 2 ，...，x i ，...x n X is the scale of the entire underwater sensor network, X i (i=1, …, n) is each node of the underwater sensor network, and the router controls each node to enter through a routing protocolAnd (3) performing line work, wherein the nodes are route distribution nodes and correspond to the positions of the artificial fish.
Initializing parameters of a fish shoal algorithm: the scale X of the underwater sensor network, each node X i Is a part of the initial position of (a); node sensing Visual field range Visual; step of Step length of the node and the neighbor node; the node crowding degree delta in the specific area; the maximum Try number try_number of the route search node; current iteration number k; maximum number kmax of iterations.
Step 2: calculate the own energy value of each node and initialize the bulletin board: the value y=f (x) of the evaluation function of the node, i.e. the fitness value of the current node, is calculated. Setting up bulletin boards in the algorithm, and storing the optimal solution state. Initializing the bulletin board means comparing all the nodes obtained as described above to an optimal value and assigning the bulletin board. (the evaluation function is used for evaluating the priority of the node, the better the state of the node itself, the better the state in the specific area, and the larger the corresponding function value).
Step 3: a dynamic trade-off factor strategy is introduced: the Visual field and the Step size are adjusted in real time according to the iteration times, the early searching capacity and the later convergence capacity of the algorithm are improved, and the adjustment is shown in formulas (1) and (2):
Step＝Step-Step×α； (1)
Visual＝Visual-Visual×α； (2)
where alpha is a dynamic trade-off factor,
step 4: and each node updates its own position by searching for neighbor optimal nodes (foraging), mutual collaboration (clustering) among nodes in the area, node determining path depiction (rear-end collision) and random searching for optimal node (random) behaviors in the rest area in the network without circulating nodes in the area, after all nodes are updated, generating a new network, completing one iteration, selecting an optimal solution individual of the updated network, comparing with a bulletin board, and if the optimal solution individual is superior to the bulletin board, replacing the original state of the bulletin board. At this point, the bulletin board obtains the maximum value of the global fitness function that can be currently searched.
Foraging behavior: let x be i For the current state of the node individual, the corresponding evaluation function value is y i The method comprises the steps of carrying out a first treatment on the surface of the In the field of viewInternally randomly selecting a state x j The corresponding evaluation function value is y j ，
x j ＝x i +Visual*rand()(3)
If y j ＞y i Then look for according to this direction as follows:
otherwise reselecting the random state x j And continuously judging the fitness, and if the number of times of trial exceeds the maximum number of times of trial Try_number, executing random behaviors.
Clustering behavior: let x be i For node individual current state, x c For the center position of the companion of the current neighborhood, n f For the number of peers (d) ij ＜Visual，d ij ＝||x i -x j I, artificial fish x i And x j Distance between them), n is the total number of nodes. If the corresponding evaluation function value y c ＞y i And the representation is accompanied by a higher fitness function value and is not very crowded, the companion works as the primary node, otherwise, predation is performed:
rear-end collision behavior: let x be i For node individual current state, x j Is in the current field (d ij < Visual) individual states with maximum evaluation function values among all nodes, x j The corresponding evaluation function value is y j . If y j ＞y i And (2) andrepresenting companion x i With higher fitness function value and less crowded surrounding environment, go to companion x j The following movements are performed, otherwise predation is performed. (the aggregation behavior and the foraging behavior can be based on the actual applicationThe execution order is selected preferentially by the environment
Random behavior: a state is randomly selected in the field of view and then moved to this state.
Wherein the method comprises the steps ofFor the current node position, +.>To perform the position after the shoal of fish, x best To be the best fish location in the current overall situation, i.e. the location information stored by the bulletin board.
Step 5: secondary initialization parameters: after each iteration is completed, the optimal parameters of the initial algorithm (sensor network X; and corresponding each node X) are reset according to the position of the current optimal solution in the bulletin board i Is a position of (2); maximum number of iterations k max ) Avoiding local optimum values provides a more accurate optimum parameter in a relatively small area.
Step 6: checking termination conditions: if the iteration number k is greater than or equal to k max The algorithm iteration ends and the optimal solution (x in bulletin board is output best ，y best ) And selecting a window according to the obtained optimal focusing position, completing a window selection part in automatic focusing, if the termination condition is not met, assigning k+1 to k, and returning to the execution step 3.
The invention uses the area with the optimal node condition in the whole network as the selection basis of the focusing window, searches the whole network by utilizing an algorithm, ensures the focusing precision and effectively improves the efficiency. Mapping a network node optimizing strategy into an artificial fish swarm algorithm, and initializing parameters; initializing a bulletin board; introducing a dynamic weighing factor strategy; updating the node state; resetting the initialization parameters; the termination condition is checked. Introducing a dynamic weighing factor, and adjusting the step length and the visual field of the node in real time, so that the algorithm efficiency is improved; aiming at the problem of the local figure of merit of the algorithm, after each iteration, the initial parameters are reset according to the optimal solution, so that the real-time performance of focusing is effectively enhanced, and the higher convergence speed and the higher algorithm precision are ensured.
2. Construction of dynamic three-dimensional simulation model by deep learning neural network model
Visual simulation of the underwater sensor network can provide visual technical reference for researching the underwater sensor network, is beneficial to monitoring the working state of the underwater sensor by technicians so as to facilitate network adjustment in time, the three-dimensional reconstruction technology belongs to a mature reality simulation technology, and mainly carries out real three-dimensional mapping reconstruction through a computer technology.
The invention adopts a depth network model to process information, in particular to collect node working motion trail images of the whole network, obtain the motion speed and the motion direction of the images, combine the left eye perception image and the right eye perception image of a human body through simulating a three-dimensional environment, collect the left eye perception image and the right eye perception image, construct a left eye perception network model and a right eye perception network model, further construct a visual perception twin network model, have the same network model structure of the left eye and the right eye, train the network node motion model based on the visual perception twin network model, simulate the node working motion speed data and the motion direction, and construct a network node working motion simulation model for realizing the network node dynamic simulation of the simulated three-dimensional environment.
The method is divided into two stages, wherein in the training stage, a near-end strategy optimization reinforcement learning algorithm (PPO) is adopted to train the model, and a proposed visual perception twin network algorithm based on a multi-scale attention mechanism is adopted to construct a strategy (policy) function of the near-end strategy optimization algorithm. The motion of the network node is then controlled using the trained model in an inference phase.
Policy function-algorithm based on multiscale attention mechanism
The method comprises the steps that a strategy function in a motion model is constructed based on a visual perception twin network algorithm of a multi-scale attention mechanism, the input of a neural network is an RGB image collected by a network node in a simulated three-dimensional environment, namely, the collected image is used as a current state St, and the action t of the network node is output, wherein the action t comprises speed and direction. The backbone network employs a twin architecture, using two network branches with shared structures and parameters to process images acquired by the left and right eyes. The system comprises an information extraction module and a decision module, wherein the information extraction module is used for extracting information by processing an input image; the decision module fuses the extracted information, accelerates convergence through residual connection and outputs the speed and direction of the network node.
Information extraction module based on multiscale attention mechanism
An information extraction module based on a multi-scale attention mechanism is provided for processing information in the acquired image. The module consists of a scale attention mechanism and a space attention mechanism.
In the scale attention mechanism, feature pyramids and attention are structuredMechanism combination, and automatic acquisition of scale weight M by learning S (F) Promote important scale and inhibit unimportant scale, scale weight M S (F) Is calculated according to the formula:
wherein sigma refers to a sigmod function, after an RGB image passes through a convolution layer, the RGB image is input into pooling layers with different sizes, is downsampled into feature images with different scale information, is input into the convolution layer, and is recovered into the feature images with the original size through upsampling. These feature maps F of different scales i The feature block F with rich context information of different scales is obtained through cascading with the original features; on the other hand, these feature maps F of different scales i And obtaining the weight sizes of different scales through the attention module. Firstly, pooling is carried out by taking a scale as a unit through a global maximum pooling layer MaxPool and a global average pooling layer AvgPool to obtain pooling results of different scalesAnd->Respectively cascading the results of different scales of the maximum pooling layer and the average pooling layer to obtainAnd->Respectively inputting the two characteristic blocks into a convolutional layer Conv, using a sigmoid function sigma, and finally obtaining a scale weight M through addition s (F) And finally, carrying out dot product on the scale weights and the multi-scale feature map F, distributing the attention weights to different scales, and outputting a multi-scale feature block with scale attention. Different inputs->And->Weights W of shared convolutional layer Conv 0 。
The spatial attention mechanism of the information extraction module takes as input the characteristics of the scale attention mechanism output. The calculation formula of the spatial attention weight is as follows:
wherein G refers to the characteristics output by a scale attention mechanism, and the space attention mechanism distributes different weights for different space positions, so that AvgPool (G) and MaxPool (G) in the formula represent that the input characteristics G are pooled by taking a channel as a unit, and an average pooling result is outputAnd maximum pooling results->And cascade. The characteristics after cascading are input into a convolution layer Conv, W 1 Is a learnable parameter of a convolution layer, and is normalized by using a sigmoid function sigma to obtain a spatial attention weight M a (G) A. The invention relates to a method for producing a fibre-reinforced plastic composite Last spatial attention weight M a (G) And (5) dot product with the input feature G to obtain the output feature of the spatial attention mechanism.
Fusion decision module
After the visual images acquired by the network nodes are processed by the information extraction module, the obtained output characteristics are input into the fusion decision module, the fusion decision module is responsible for fusing the information of the left visual image and the right visual image, making decisions and outputting the speed and the direction (discrete value) of the network nodes.
The fusion decision module consists of four convolution layers and one full connection layer. The left and right network branches share structures and parameters, respectively process the characteristics of the left and right images, and fuse low-level and high-level information through jumper connection. And the middle network branch carries out fusion processing on the characteristics output by different convolution layers of the left branch and the right branch through cascading and four-layer convolution networks. The magnitude and direction of the predicted speed are output through the two fully connected layers, respectively.
The invention takes the visual node image as input, is more similar to the real network node working movement, can effectively simulate the node working movement in the whole network, and effectively enhances the scalability and the robustness of the dynamic three-dimensional model.
The above embodiments are only illustrative of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention, and various modifications and improvements made by those skilled in the art to the technical solutions of the present invention should fall within the protection scope defined by the claims of the present invention without departing from the design spirit of the present invention.
Claims (6)
1. An underwater wireless sensor network routing method based on energy self-collection is characterized by comprising the following steps:
a plurality of underwater sensors are arranged on the basis of a water area to be measured, and a plurality of underwater sensor network nodes are obtained; constructing an underwater sensor network model based on the underwater sensor network node and the updated artificial fish swarm algorithm; constructing a dynamic three-dimensional simulation model based on the deep learning neural network model; performing dynamic simulation on the underwater sensor network model based on the dynamic three-dimensional simulation model to realize intelligent control of the underwater sensor network;
the constructing of the underwater sensor network model further comprises the steps of positioning, layering and clustering the underwater sensor network nodes to obtain layer spacing and node spacing of the underwater sensor network nodes, and constructing the underwater sensor network model based on the layer spacing and the node spacing;
the interlayer spacing is obtained by layering the distance from the underwater sensor network node to the route;
the node distance is the distance from the node of the same-layer underwater sensor network to the route;
the process of constructing the dynamic three-dimensional simulation model based on the deep learning neural network model comprises the steps of collecting moving images of the underwater sensor network nodes and constructing a visual twin network model by simulating a three-dimensional environment; based on the vision twin network model, training the underwater sensor network model by setting a reward function, and constructing the dynamic three-dimensional simulation model.
2. The method for routing an underwater wireless sensor network based on energy self-collection according to claim 1, wherein,
acquiring a data set by acquiring working state images of the underwater sensor network nodes, dividing the data set into a training set and a verification set, training the deep learning neural network model, and constructing a dynamic three-dimensional simulation model.
3. The method for routing an underwater wireless sensor network based on energy self-collection according to claim 1, wherein,
the updating process of the artificial fish swarm algorithm comprises the following steps:
initializing artificial fish swarm algorithm parameters; calculating the self energy value of the underwater sensor network node and initializing a bulletin board; and carrying out iterative updating of the optimal node on the underwater sensor network node based on a dynamic weighing factor strategy, carrying out parameter secondary initialization based on the optimal node until the preset iterative times are reached, outputting an optimal solution in the bulletin board after the iteration of the artificial fish swarm algorithm is ended, obtaining an optimal focusing position according to the optimal solution, selecting a window, and finishing updating.
4. The method for routing an underwater wireless sensor network based on energy self-collection as claimed in claim 3, wherein,
the artificial fish swarm algorithm parameters comprise the scale X of the underwater sensor network and the node X of each underwater sensor network i Is a primary position node sensing field of viewVisual surrounding, step length of nodes and neighbor nodes, node crowding degree delta in a specific area, maximum Try number Try_number of route searching nodes, current iteration number k and maximum iteration number kmax.
5. The method for routing an underwater wireless sensor network based on energy self-collection according to claim 1, wherein,
the dynamic simulation comprises a training stage and an reasoning stage;
the training stage is characterized in that a strategy function of a near-end strategy optimization algorithm is constructed based on a visual perception twin network algorithm of a multi-scale attention mechanism, and the underwater sensor network model is trained through the strategy function;
and the reasoning stage controls the motion of the underwater sensor network node through the dynamic three-dimensional simulation model to dynamically simulate the underwater sensor network model.
6. The method for routing an underwater wireless sensor network based on energy self-collection according to claim 1, wherein,
the visual twin network model comprises an information extraction module and a decision module;
the visual twin network model processes the moving image through the information extraction module to obtain image extraction information;
and the visual twin network model fuses the image extraction information through the decision module, accelerates convergence through residual connection, and outputs the movement speed and movement direction of the underwater sensor network node.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210208771.4A CN114599069B (en) | 2022-03-04 | 2022-03-04 | Underwater wireless sensor network routing method based on energy self-collection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210208771.4A CN114599069B (en) | 2022-03-04 | 2022-03-04 | Underwater wireless sensor network routing method based on energy self-collection |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114599069A CN114599069A (en) | 2022-06-07 |
| CN114599069B true CN114599069B (en) | 2024-01-23 |
Family
ID=81815891
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210208771.4A Active CN114599069B (en) | 2022-03-04 | 2022-03-04 | Underwater wireless sensor network routing method based on energy self-collection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114599069B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115623494B (en) * | 2022-12-16 | 2023-03-14 | 中南大学 | NS 3-based multi-unmanned system distributed networking simulation method and system |
| CN117499883B (en) * | 2023-10-27 | 2024-07-19 | 江苏海洋大学 | Sensor control and management method for deep sea communication |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108459406A (en) * | 2018-03-15 | 2018-08-28 | 上海理工大学 | Microscope auto-focusing window selection method based on artificial fish-swarm algorithm |
| CN112597802A (en) * | 2020-11-25 | 2021-04-02 | 中国科学院空天信息创新研究院 | Pedestrian motion simulation method based on visual perception network deep learning |
| WO2021197244A1 (en) * | 2020-03-31 | 2021-10-07 | 长安大学 | Underwater sensor network-based method for detecting oil leakage point of oil transport pipeline and measuring oil leakage rate |
| WO2021259844A1 (en) * | 2020-06-25 | 2021-12-30 | Connaught Electronics Ltd. | Automatic visual perception using an artificial neural network |
-
2022
- 2022-03-04 CN CN202210208771.4A patent/CN114599069B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108459406A (en) * | 2018-03-15 | 2018-08-28 | 上海理工大学 | Microscope auto-focusing window selection method based on artificial fish-swarm algorithm |
| WO2021197244A1 (en) * | 2020-03-31 | 2021-10-07 | 长安大学 | Underwater sensor network-based method for detecting oil leakage point of oil transport pipeline and measuring oil leakage rate |
| WO2021259844A1 (en) * | 2020-06-25 | 2021-12-30 | Connaught Electronics Ltd. | Automatic visual perception using an artificial neural network |
| CN112597802A (en) * | 2020-11-25 | 2021-04-02 | 中国科学院空天信息创新研究院 | Pedestrian motion simulation method based on visual perception network deep learning |
Non-Patent Citations (3)
| Title |
|---|
| 一种基于黏性流体算法的水下传感器网络节点部署模型;田祥宏;南京理工大学学报;摘要及第2、4章 * |
| 一种鲁棒的无线传感器网络覆盖空洞修补方法;闫雒恒;贺昱曜;;计算机科学(第02期);全文 * |
| 基于人工鱼群的水下传感器网络优化算法;郑翔;汤磊;张文强;武欣嵘;童玮;;军事通信技术(第02期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114599069A (en) | 2022-06-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114599069B (en) | Underwater wireless sensor network routing method based on energy self-collection | |
| CN110750096B (en) | Mobile robot collision avoidance planning method based on deep reinforcement learning in static environment | |
| CN108600379A (en) | A kind of isomery multiple agent Collaborative Decision Making Method based on depth deterministic policy gradient | |
| CN112307622B (en) | Autonomous planning system and planning method for generating force by computer | |
| CN112036540B (en) | Sensor number optimization method based on double-population hybrid artificial bee colony algorithm | |
| CN101448267A (en) | Wireless sensor network node coverage optimization method based on particle swarm algorithm | |
| CN108036790A (en) | Robot path planning method and system based on mutillid algorithm under a kind of obstacle environment | |
| CN107229287A (en) | A kind of unmanned plane global path planning method based on Genetic Ant algorithm | |
| CN114942633A (en) | Multi-agent cooperative anti-collision picking method based on digital twins and reinforcement learning | |
| CN114092697B (en) | Building facade semantic segmentation method with attention fused with global and local depth features | |
| CN112469050B (en) | WSN three-dimensional coverage enhancement method based on improved wolf optimizer | |
| Yang et al. | A digital twins enabled underwater intelligent internet vehicle path planning system via reinforcement learning and edge computing | |
| CN109242183B (en) | Crowd simulation evacuation method and device based on artificial fish-swarm algorithm and target detection | |
| CN113242562B (en) | WSNs coverage enhancement method and system | |
| CN110109653A (en) | Land battle chess intelligent engine and operation method thereof | |
| CN109885082A (en) | The method that a kind of lower unmanned aerial vehicle flight path of task based access control driving is planned | |
| CN109299778A (en) | A kind of calculation method of the RCRSS rescue map subregion based on cuckoo searching algorithm | |
| CN116449863A (en) | Reinforced learning unmanned aerial vehicle cluster multi-target searching method based on pheromone | |
| CN116757249A (en) | Unmanned aerial vehicle cluster strategy intention recognition method based on distributed reinforcement learning | |
| CN113467481B (en) | Path planning method based on improved Sarsa algorithm | |
| CN117420849B (en) | Marine unmanned aerial vehicle formation granularity-variable collaborative search and rescue method based on reinforcement learning | |
| CN117932934A (en) | Situation-based unmanned cluster cooperative efficiency dynamic evaluation method | |
| CN116795098A (en) | Spherical amphibious robot path planning method based on improved sparrow search algorithm | |
| CN110334869A (en) | A kind of mangrove forest ecological health forecast training method based on dynamic colony optimization algorithm | |
| CN111626341A (en) | Feature level information fusion method for underwater target identification |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |