A self-healing mechanism for a sensor node communication system and method

Abstract

The application discloses a kind of sensing node communication systems and methods with self-healing mechanism, including node detection module, node communication module, pre-processing module, hybrid training module and self-healing control module.System is achieved by configuring multiple sensing nodes the periodic collection of multidimensional environmental parameters, based on self-organizing network sharing state information, construct local network health state atlas.Data after pre-processing is used for hybrid neural network training, long-term dependence features and local abnormal fluctuation features of node state are extracted therefrom to assess failure risk.When risk score exceeds threshold value, the system automatically performs dynamic routing reconstruction, standby link activation and other self-healing operations.The application improves the stability and adaptive capacity of underwater sensor network, and has wide application value.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of Internet of Things, and in particular to a sensing node communication system with a self-healing mechanism and a method thereof. BACKGROUND

[0002] With the rapid development of Internet of Things, edge computing and intelligent sensing technology, distributed sensing systems have been widely applied in environmental monitoring, industrial control, smart facilities, disaster warning and other fields. Such systems are usually composed of a large number of sensing nodes, which cooperatively collect and transmit multi-source environmental data through wireless communication networks to realize continuous sensing and state management of target areas.

[0003] In practical applications, especially in some complex or extreme environments, such as high humidity, high pressure, high corrosive atmosphere or underwater environment, the deployment and communication of sensing nodes often face severe challenges. Sensing networks in these environments not only need to have strong structural adaptability and energy control ability, but also have higher requirements for network stability, communication reliability and fault recovery capability.

[0004] However, existing sensing communication systems mostly adopt static routing structure and centralized control architecture, which are difficult to identify and dynamically respond in time when node failures or communication link abnormalities occur, easily leading to network paralysis, data loss or partial system failure. In addition, most current systems lack real-time analysis and prediction capability of node state and environmental data, and cannot effectively warn before failure occurs, and lack control mechanisms with adaptive and self-healing characteristics.

[0005] Therefore, there is an urgent need for a sensing node communication system with fault prediction capability and self-healing control logic, which can continuously and stably operate in various complex environments, significantly improve the adaptive ability and overall robustness of the network, and meet the actual needs of high-reliability communication and intelligent fault tolerance. SUMMARY

[0006] In view of the deficiencies in the prior art, the purpose of the present application is to provide a sensing node communication system with a self-healing mechanism and a method thereof, for improving the stability, adaptive ability and fault response efficiency of underwater sensing networks.

[0007] To achieve the above-mentioned purpose, the present application provides the following technical solutions: a sensing node communication system with a self-healing mechanism, comprising:

[0008] a node detection module configured to configure a plurality of sensing nodes, each of the sensing nodes comprising a plurality of sensors, each of the sensors being configured to collect multi-dimensional detection data at a preset period;

[0009] A node communication module connected to the node detection module, configured to establish a communication connection for each of the sensor nodes in the self-organizing multi-hop network, and to form a local network health state atlas through a state sharing protocol based on the detection data detected by each of the sensor nodes, the local network health state atlas comprising a plurality of node state data, the node state data comprising node self-checking data, physical environment parameters, and chemical environment parameters;

[0010] A preprocessing module connected to the node communication module, configured to sequentially perform preprocessing and normalization processing on the node self-checking data, the physical environment parameters, and the chemical environment parameters, to construct a training data set;

[0011] A hybrid training module connected to the preprocessing module, configured to train a hybrid neural network based on the training data set, the hybrid neural network comprising a time series modeling layer, a local feature extraction layer, and a fully connected layer, the time series modeling layer being configured to capture long-term dependence features of each node state in the training data set by using a long short-term memory network, the local feature extraction layer being configured to extract local abnormal fluctuation features within a time window in the training data set by using a convolutional neural network layer, and the fully connected layer being configured to fuse the long-term dependence features and the local abnormal fluctuation features to output a fault risk score;

[0012] A self-healing control module connected to the hybrid training module and the node communication module, configured to automatically trigger a self-healing response measure based on the node state data of the adjacent sensor nodes when the fault risk score at the current sensor node is greater than a preset self-healing risk threshold, the self-healing response measure comprising dynamic routing reconstruction, backup link activation, and data load redistribution.

[0013] Further, the preprocessing module comprises:

[0014] A time series division unit configured to record the node self-checking data, the physical environment parameters, and the chemical environment parameters in a time sequence, and to segment the data by using a sliding window technology to form continuous time series data samples;

[0015] A preprocessing unit connected to the time series division unit, configured to sequentially perform data cleaning, outlier detection, and missing value filling on the time series data samples to obtain preprocessed data samples;

[0016] A normalization unit connected to the preprocessing unit, configured to perform normalization processing on multi-dimensional data in the preprocessed data samples based on a preset normalization method to form the training data set.

[0017] Further, the node self-checking data comprises node voltage, node current, node energy consumption state, signal-to-noise ratio, data packet loss rate, and signal delay;

[0018] The physical environment parameters include temperature, ambient pressure, salinity, dissolved oxygen concentration, ambient turbidity;

[0019] The chemical environment parameters include pH value and specific ion concentration.

[0020] Further comprising:

[0021] A fault detection module connected to the node communication module, configured to perform fault detection on the node state data and generate corresponding fault detection results, and process the fault detection results to obtain fault actual scores;

[0022] An incremental training module connected to the fault detection module and the hybrid training module, configured to process the fault actual scores and the fault risk scores to obtain a predicted deviation value, and perform dynamic adjustment on the weight parameters of the hybrid neural network based on the predicted deviation value and then retrain to perform incremental training and update the hybrid neural network.

[0023] Further, the function expression of the hybrid neural network comprises:

[0024] ;

[0025] ;

[0026] ;

[0027] ;

[0028] ; ;

[0029] wherein, represents the fault risk score, represents the length of the time window, represents the time variable, represents a Gamma weight function constructed by the current voltage and the temperature of the node, represents the current voltage of the node, represents the standard deviation of the temperature, represents the mean value of the temperature, represents a current disturbance modulation term, represents a composite excitation factor composed of the product of the current and the data packet loss rate, represents a sensitivity adjustment factor of ambient pressure change, represents the historical current mean value, represents the change interval of the ambient turbidity, represents the change amount of the signal-to-noise ratio, represents a communication and activation modeling function, represents a first kind of first order Bessel function measuring environmental disturbance amplitude, represents a second kind of zero order Bessel function representing node redundancy connectivity strength, represents a salinity disturbance frequency factor, represents a complexity parameter of current network communication topology, represents a rate of change of the node energy consumption, represents a time window width of the PH value, represents a delay decay and ion anomaly suppression function, represents a current signal maximum delay, represents a historical average signal delay, represents a data channel congestion factor, represents a time-dependent adjustment coefficient, represents a mean value of the specific ion concentration, represents a standard deviation of the specific ion concentration, represents a self-healing trigger suppression function, represents a self-healing response time activation coefficient, represents a state consistency measurement factor between adjacent nodes, represents a natural logarithm function, represents a complexity quantitative indicator of ion species in the current environment.

[0030] Further, the node state data further includes underwater acoustic interference data and underwater optical interference data, the underwater acoustic interference data including power spectral density, the underwater optical interference data including optical image blurriness, color cast drift, and image noise index.

[0031] Further, a function optimization module is further included, connected to the node communication module and the hybrid training module, for optimizing the communication and activation modeling function according to the power spectral density, the optical image blurriness, and the image noise index, and optimizing the self-healing trigger suppression function according to the image blurriness and the color cast drift, and further updating the fault risk score according to the optimized communication and activation modeling function and the optimized self-healing trigger suppression function.

[0032] Further, the optimized communication and activation modeling function is configured as:

[0033] ; wherein, represents the optimized communication and activation modeling function, represents the normalized power spectral density value, represents the image noise index, denotes a blur index adjustment factor, denotes a Bessel disturbance frequency term for simulating the periodic disturbance of the PSD in underwater acoustic propagation;

[0034] The optimized self-healing trigger suppression function is configured as:

[0035] ;

[0036] wherein, denotes the optimized self-healing trigger suppression function, denotes the image blur index, denotes a blur index amplification coefficient for controlling the influence of blur on the activated interference force, denotes a color deviation drift difference, respectively denote the color difference fluctuation coefficients of the G and R channels;

[0037] The function expression of the updated fault risk score is configured as:

[0038] ;

[0039] wherein, denotes the updated fault risk score.

[0040] A sensor node communication method with a self-healing mechanism, applied to the sensor node communication system with a self-healing mechanism described above, comprising:

[0041] Step S1, the node detection module configures a plurality of sensor nodes, each of the sensor nodes includes a plurality of sensors, and each of the sensors is used to collect multi-dimensional detection data at a preset period;

[0042] Step S2, the node communication module establishes a communication connection for each of the sensor nodes in the self-organizing multi-hop network, and the detection data detected by each of the sensor nodes forms a local network health state map through a state sharing protocol, the local network health state map includes a plurality of node state data, and the node state data includes node self-checking data, physical environment parameters, and chemical environment parameters;

[0043] Step S3, the preprocessing module sequentially performs preprocessing and normalization processing on the node self-checking data, the physical environment parameters, and the chemical environment parameters to construct a training data set;

[0044] Step S4, the hybrid training module trains a hybrid neural network according to the training data set, the hybrid neural network comprising a time series modeling layer, a local feature extraction layer and a fully connected layer, the time series modeling layer adopts a long short-term memory network to capture long-term dependence features of node states in the training data set, the local feature extraction layer adopts a convolutional neural network layer to extract local abnormal fluctuation features within a time window in the training data set, and the fully connected layer fuses the long-term dependence features and the local abnormal fluctuation features to output a fault risk score;

[0045] Step S5, when the fault risk score of the current sensing node is greater than a preset self-healing risk threshold, the self-healing control module automatically triggers a self-healing response measure according to the node state data near the sensing node, the self-healing response measure comprising dynamic routing reconstruction, backup link activation and data load redistribution.

[0046] Advantages of the present application:

[0047] The present application collects node voltage, current, energy consumption state, signal-to-noise ratio, data packet loss rate and signal delay and other key indicators, realizes node-level self-diagnosis and health assessment, and significantly enhances the timeliness and accuracy of fault identification;

[0048] The present application can comprehensively monitor the changes of underwater environment by integrating temperature, environmental pressure, salinity, dissolved oxygen, pH value, turbidity and specific ion concentration and other physical and chemical parameters;

[0049] The present application also adopts a hybrid neural network architecture that fuses LSTM and CNN, which can not only capture the time dependence in the sensing data, but also identify short-term local fluctuation abnormalities, significantly improving the depth and robustness of fault prediction;

[0050] After identifying high-risk nodes, the present application can automatically trigger multi-level response strategies including dynamic routing reconstruction, backup link activation and data load redistribution based on the state of adjacent nodes, effectively preventing network paralysis or data interruption;

[0051] In the face of complex and variable physical and chemical environments underwater, the present application has real-time adaptation and reconstruction capabilities, significantly improving the stability and fault tolerance of the sensing network in underwater operations. BRIEF DESCRIPTION OF DRAWINGS

[0052] Figure 1 is a structural schematic diagram of the sensing node communication system in the present application.

[0053] Label: 1, node detection module; 2, node communication module; 3, preprocessing module; 31, time sequence division unit; 32, preprocessing unit; 33, normalization unit; 4, hybrid training module; 5, self-healing control module; 6, fault detection module; 7, incremental training module; 8, function optimization module. DETAILED DESCRIPTION

[0054] The application will be further described in detail below in conjunction with the accompanying drawings and examples. Identical parts are denoted by identical reference numerals. It should be noted that the words "front", "back", "left", "right", "up" and "down" used in the following description refer to the directions in the drawings, and the words "bottom surface" and "top surface", "inner" and "outer" refer to the directions towards or away from the geometric center of a particular part.

[0055] Example 1, Reference Figure 1 As a first embodiment of the application, the embodiment provides a sensing node communication system with a self-healing mechanism, which can improve the stability, self-adaptability and fault response efficiency of the underwater sensing network, comprising:

[0056] The node detection module 1 is configured to configure a plurality of sensing nodes, each sensing node comprising a plurality of sensors, each sensor being configured to collect multi-dimensional detection data at a predetermined period;

[0057] The node communication module 2 is connected to the node detection module 1 and is configured to form a self-organizing multi-hop network to establish communication connections for each sensing node. The detection data detected by each sensing node forms a local network health state map through a state sharing protocol. The local network health state map comprises a plurality of node state data, and the node state data comprises node self-checking data, physical environment parameters and chemical environment parameters.

[0058] The preprocessing module 3 is connected to the node communication module 2 and is configured to sequentially preprocess and normalize the node self-checking data, the physical environment parameters and the chemical environment parameters to construct a training data set.

[0059] The hybrid training module 4 is connected to the preprocessing module 3 and is configured to train a hybrid neural network according to the training data set. The hybrid neural network comprises a time sequence modeling layer, a local feature extraction layer and a full connection layer. The time sequence modeling layer uses a long short-term memory network to capture long-term dependence features of each node state in the training data set. The local feature extraction layer uses a convolutional neural network layer to extract local abnormal fluctuation features within a time window in the training data set. The full connection layer fuses the long-term dependence features and the local abnormal fluctuation features to output a fault risk score.

[0060] The self-healing control module 5 is connected to the hybrid training module 4 and the node communication module 2, and is used to automatically trigger a self-healing response measure according to the node state data at the adjacent sensor nodes when the fault risk score at the current sensor node is greater than a preset self-healing risk threshold, the self-healing response measure including dynamic route reconstruction, backup link activation, and data load redistribution.

[0061] Specifically, in the embodiment, the node self-checking data includes node voltage, node current, node energy consumption state, signal-to-noise ratio, data packet loss rate, and signal delay.

[0062] The physical environment parameters include temperature, environmental pressure, salinity, dissolved oxygen concentration, and environmental turbidity.

[0063] The chemical environment parameters include pH value and specific ion concentration.

[0064] Working principle of embodiment 1:

[0065] In a certain offshore shipwreck site protection project, a sensor node communication system based on embodiment 1 is deployed to build an underwater sensing network covering key areas around the site. The system is composed of multiple sensor nodes, each node integrating multiple sensors and intelligent communication modules.

[0066] Node data acquisition:

[0067] Each sensor node is configured with sensing devices to detect the following detection data:

[0068] Node self-checking data: including node voltage, current, energy consumption state, signal-to-noise ratio, data packet loss rate, signal delay, and other self-checking indicators, reflecting the real-time running health status of the node device itself;

[0069] Physical environment parameters: including temperature, environmental pressure, salinity, dissolved oxygen concentration, and environmental turbidity, etc. information, used to evaluate environmental changes;

[0070] Chemical environment parameters: including the pH value and specific ion concentration (such as chloride ion, sodium ion, etc.) of the water body, used to judge the environmental corrosiveness and potential impact on cultural relics.

[0071] The sensor nodes synchronously collect the above multi-dimensional detection data at preset time intervals, and build a self-organizing multi-hop network through underwater acoustic communication to realize data sharing and state coordination.

[0072] Data processing and model training:

[0073] All data are first normalized and abnormal values are removed by the preprocessing module 3 to form a structured training data set. The system uses a hybrid neural network model:

[0074] The input data of the mixed neural network model consists of: multiple time series of sensor data of a sensor node within a certain time window (e.g., 30 minutes);

[0075] Data dimensions:

[0076] Node self-checking data (e.g., voltage, current, packet loss rate, signal delay, signal-to-noise ratio) - a total of 5 dimensions;

[0077] Physical parameters (e.g., temperature, environmental pressure, salinity, dissolved oxygen, turbidity) - a total of 5 dimensions;

[0078] Chemical parameters (e.g., pH value, ion concentration) - a total of 2 dimensions;

[0079] Total: 12-dimensional data channels x time series, forming an input tensor.

[0080] The time series modeling layer uses an LSTM network to capture the long-term trends of time series signals such as node voltage, current, and signal-to-noise ratio, forming long-term dependency features. The network structure of the time series modeling layer includes 1 to 2 layers of stacked LSTM units, each containing 64 or 128 LSTM units.

[0081] The local feature extraction layer uses a CNN to extract features such as PH value, salinity, and packet loss rate within a short time window;

[0082] The fully connected layer includes a feature fusion layer that concatenates the long-term dependency features extracted by the time series modeling layer and the local features extracted by the local feature extraction layer, and then uses a Flatten layer to convert them into a vector. Then, it is input into the fully connected structure for fusion modeling to output the fault risk score of each node.

[0083] Self-healing response mechanism:

[0084] When a node's risk score exceeds the self-healing risk threshold due to environmental changes or equipment aging, the system immediately triggers the self-healing mechanism based on the state data of adjacent nodes in the graph:

[0085] If the node's signal-to-noise ratio continues to decline and the signal delay increases, the system determines that the communication is unstable and automatically performs dynamic routing reconstruction;

[0086] If the node voltage significantly decreases and the energy consumption state is abnormal, the backup communication link is activated;

[0087] If the turbidity and PH value of the node's monitoring area abnormally increase, affecting data acquisition quality, the data load redistribution mechanism is started, temporarily switching tasks to adjacent healthy nodes.

[0088] The whole process does not need manual intervention, realizes continuous, stable and safe acquisition of key underwater data, and guarantees efficient development of site environment evaluation and protection work.

[0089] Preferably, the preprocessing module 3 comprises:

[0090] a time sequence division unit 31, configured to record node self-checking data, physical environment parameters and chemical environment parameters in time sequence, and perform data segmentation by using a sliding window technology to form continuous time sequence data samples;

[0091] a preprocessing unit 32 connected to the time sequence division unit 31, configured to sequentially perform data cleaning, outlier detection and missing value filling on the time sequence data samples to obtain preprocessed data samples;

[0092] a normalization unit 33 connected to the preprocessing unit 32, configured to perform normalization processing on multi-dimensional data in the preprocessed data samples according to a preset normalization method to form a training data set.

[0093] Specifically, in the embodiment, the time sequence division unit 31 records data once every 10 seconds, adopts a sliding window mechanism, and is set as follows:

[0094] window length: 60 (i.e. 600 seconds = 10 minutes);

[0095] sliding step: 30 (half-overlapping window, 30 samples for one “step”). Each window constructs a sample sequence, and each sample sequence contains 60-time continuous sequences of 12 types of data including node voltage, current, delay, SNR, temperature, environmental pressure, turbidity, salinity, dissolved oxygen, PH value, specific ion concentration and data packet loss rate.

[0096] The preprocessing unit 32 sequentially executes the following preprocessing procedures on each window sample:

[0097] data cleaning: removing obviously incorrect data (such as invalid abnormal values of voltage < 2V and PH < 1); marking points exceeding the set boundary (such as z-score > 3).

[0098] outlier detection: using an algorithm based on median absolute deviation (MAD) to detect mutation points; a single sudden jump value is not deleted, but is replaced by the median in the window.

[0099] missing value filling: using a bidirectional interpolation method (bidirectional linear + short-term mean) to fill in missing data; ensuring the continuity of the time sequence and providing input integrity for subsequent modeling of the LSTM.

[0100] Example: if the PH value is missing at the 38th second, the linear extrapolated mean of the 37th and 39th seconds is used to complete it; if the consecutive missing is > 3, the PH mean in the window is used to replace it.

[0101] Normalization unit 33: After preprocessing, the system performs normalization operation: all variables are uniformly standardized using Z-score method, and the normalized data is saved in the training data set.

[0102] The preprocessing module 3 can reduce more than 85% of abnormal data input; the normalization strategy eliminates the scale deviation between various sensor data, making the model learning more stable; the sliding window mechanism increases the sample size, improving the training sufficiency and model robustness; the embodiment ensures that the model can effectively operate in the real complex marine environment and accurately predict the failure trend of the underwater node.

[0103] Embodiment 2, as the second embodiment of the application, is different from the previous embodiment in that the embodiment provides a fault detection module 6 and an incremental training module 7, which can realize adaptive correction and retraining of model prediction, and optimize the overall risk prediction performance of the system, which includes:

[0104] The fault detection module 6 is connected to the node communication module 2, and is used for fault detection according to the state data of each node, and generates a corresponding fault detection result, and processes the fault actual score based on the fault detection result;

[0105] The incremental training module 7 is connected to the fault detection module 6 and the mixed training module 4, and is used for processing the predicted deviation value according to the fault actual score and the fault risk score, and retraining after dynamically adjusting the weight parameters of the mixed neural network based on the predicted deviation value, to perform incremental training and update of the mixed neural network.

[0106] Working principle of embodiment 2:

[0107] The fault detection module 6 is connected to the node communication module 2, and its main task is to convert the running state data of each node into a "actual fault degree" score value that can be determined by the system. Specifically, it includes:

[0108] Data acquisition and integration: receiving node self-checking data (such as voltage, current, delay, signal-to-noise ratio, packet loss rate), environmental parameters (temperature, PH, salinity, image quality, etc.) and communication state feedback of other adjacent nodes;

[0109] Rule recognition and fault determination: based on the set detection rules and models (for example: if the voltage is lower than the set limit and the delay is high, then the fault; if the image appears strong blur and color shift, it is determined to be abnormal), logical judgment is made on the current node state;

[0110] Score generation: mapping the judgment result to the actual fault score , which represents the fault severity of the node in the current period. Among them: Normal; : light to moderate abnormality; : severe failure, need self-healing response.

[0111] Result output: actual score of failure Failure risk score with current node Transmitted together to the incremental training module.

[0112] The incremental training module 7 connects the failure detection module 6 and the hybrid training module 4, and its purpose is to realize online adjustment and adaptive learning of the model, and maintain prediction accuracy. Specifically, it includes:

[0113] Prediction bias calculation: receive the prediction score of the current output of the hybrid neural network, and compare the difference with the actual score output by the failure detection module to calculate the prediction bias : .

[0114] Error judgment and activation mechanism: the system sets a deviation tolerance threshold, when Greater than the deviation tolerance threshold: and continuously exceeds the set period (such as 3 rounds), the system determines that the current model produces "prediction deviation", and needs to update the weight.

[0115] Training sample construction: temporarily store such high deviation samples (including input sequence and error label) into the "incremental training sample pool".

[0116] Weight fine-tuning and incremental training: based on high deviation samples, perform Mini-batch Backpropagation; use a low learning rate (such as 0.0005) to fine-tune the weights of the LSTM layer, CNN layer or fully connected layer in the network; After training, output the updated parameter weights.

[0117] Model update and iteration: the new weights are automatically loaded into the hybrid neural network model; the updated model is used to continue scoring and prediction in the next period; form a continuous closed-loop model update mechanism.

[0118] This embodiment ensures that the model not only has high prediction ability, but also can continuously correct its reasoning deviation according to actual operation feedback, so that the model has the intelligent characteristics of "self-learning", "self-correction" and "self-evolution", and maintains high-precision prediction performance and adaptability in long-term operation.

[0119] Preferably, the function expression of the hybrid neural network includes:

[0120] ;

[0121] ;

[0122] ;

[0123] ;

[0124] ; ;

[0125] wherein, represents the failure risk score, represents the length of time window, represents the time variable, represents the Gamma weight function constructed from the current voltage and temperature of the node, represents the current voltage of the node, represents the standard deviation of temperature, represents the mean of temperature, represents the current perturbation modulation term, represents the compound excitation factor composed of the product of current and data packet loss rate, represents the sensitivity adjustment factor of environmental pressure change, represents the historical current mean, represents the change interval of environmental turbidity, represents the change amount of signal-to-noise ratio, represents the communication and activation modeling function, represents the first-order Bessel function of the first kind measuring the amplitude of environmental disturbance, represents the zero-order Bessel function of the second kind representing the strength of node redundancy connectivity, represents the salinity disturbance frequency factor, represents the complexity parameter of the current network communication topology, represents the rate of change of node energy consumption, represents the time window width of PH value, represents the delay attenuation and ion abnormal suppression function, represents the current maximum signal delay, represents the historical average signal delay, represents the data channel congestion factor, represents the time-dependent adjustment coefficient, represents the mean of specific ion concentration, represents the standard deviation of specific ion concentration, represents the self-healing trigger suppression function, represents the self-healing response time activation coefficient, represents the state consistency measurement factor between adjacent nodes, represents the natural logarithmic function, represents the complexity quantitative index of ion species in the current environment.

[0126] Specifically, in this embodiment, a set of underwater self-healing control systems based on the Internet of Things is deployed in a Song Dynasty shipwreck site area in a certain sea area of the South China Sea by the protection unit, which contains 20 distributed intelligent underwater nodes. Node A12 is located at a water depth of 11.8 meters and is specially used to monitor local water chemical changes. The node presents high-frequency fluctuations for three consecutive days, which is identified as a potential abnormal node by the system model, triggering risk score evaluation.

[0127] The data collected by the node (average value, unit has been standardized):

[0128] Voltage , current , average temperature , standard deviation , environmental pressure factor , historical average current , turbidity interval , signal-to-noise ratio fluctuation , salinity excitation frequency , communication topology complexity , node energy consumption rate , PH window , maximum signal delay , average delay , congestion coefficient , current specific ion concentration average , standard deviation , self-healing trigger coefficient , node state consistency , ion complexity index , time integration window , the system calculates the integral with an interval of one second.

[0129] Fit a Gamma-Gaussian mixed curve between voltage and temperature to reflect the influence of temperature on node power supply fluctuations; Construct a disturbance enhancement model by the logarithmic relationship between current and environmental pressure, turbidity, and signal-to-noise ratio changes; Simulate nonlinear interference and PH-related excitation in the communication process to make the score sensitive to link conditions; Couple specific ion concentration anomalies and signal delay problems to model as key risk factors; Inhibit fast response actions through hyperbolic secant to prevent frequent false triggering of the system; Accumulate the full dynamic risk curve within 10 minutes in integral form to obtain the node's failure risk score.

[0130] After model integral operation, the final score calculation result of node A12 is:

[0131] ;

[0132] System score judgment logic as follows:

[0133]

[0134] Since , the system judges that the node is in the "high-risk abnormal early warning interval (0.7 to 1.0)", and the following self-healing strategies are automatically executed:

[0135] Dynamic routing reconstruction: switch the original data path forwarding task of A12 node to adjacent A11 and A13;

[0136] Backup link activation: enable the low-frequency link of the acoustic channel to take over part of its data communication;

[0137] Task load redistribution: distribute the water chemical monitoring tasks of the node to adjacent nodes in the proportions of 20%, 30%, and 50%;

[0138] Training data increment storage: add this data feature to the model "risk abnormal sample library" for subsequent training model optimization.

[0139] Embodiment 3, referring to the figure, is the second embodiment of the application, which is different from the previous embodiment in that it provides a function optimization module 8 that can accurately respond to hidden faults caused by image degradation, acoustic channel instability, etc., and achieve higher accuracy in risk prediction and system adaptive ability. The node state data further includes underwater acoustic interference data and underwater optical interference data, the underwater acoustic interference data includes power spectral density, and the underwater optical interference data includes optical image blurriness, color shift drift, and image noise index.

[0140] It also includes a function optimization module 8 connected to the node communication module 2 and the mixed training module 4, which is used to optimize the communication and activation modeling function according to the power spectral density, optical image blurriness, and image noise index, and optimize the self-healing trigger suppression function according to the image blurriness and color shift drift, and then update the fault risk score according to the optimized communication and activation modeling function and the optimized self-healing trigger suppression function.

[0141] Working principle of embodiment 3:

[0142] In complex underwater environments, non-structural changes such as underwater image quality fluctuations and acoustic channel interference often cannot be captured in time by traditional feature models, resulting in a large deviation between risk prediction results and actual node state.

[0143] To this end, the function optimization module 8 introduces acoustic and optical interference factors and embeds them in the form of a complex function into the fault risk score function, enabling the system to accurately respond to nonlinear, multidimensional disturbance signals.

[0144] The core task of the function optimization module 8 is to embed the above interference factors into the scoring function (communication and activation modeling function and self-healing trigger suppression function) in a complex mathematical expression, thereby dynamically adjusting the input of complex function structures such as activation function, Bessel function, and error function during scoring calculation, and then adjusting the output score.

[0145] When the interference term fluctuates rapidly (for example: the image noise index doubles in a short time), the function optimization module 8 will directly amplify the local fluctuation response term in the scoring function, causing the model's risk score to rise rapidly in the short term, thereby achieving early warning of implicit abnormalities.

[0146] At the same time, by dynamically adjusting the amplification coefficient and normalization scale in the function structure, the system can automatically adapt to different levels of interference according to different water areas and node deployment situations, maintaining the balance between the stability and sensitivity of the score.

[0147] By introducing the function optimization module 8 and structurally embedding the underwater interference factors into the risk score function, the system can effectively improve the recognition accuracy of implicit faults such as image degradation and unstable sound channels; at the same time, it avoids misjudging optical or acoustic quality fluctuations as ordinary physical abnormalities; and further realizes higher confidence fault score output, improving the prediction accuracy and self-adaptive ability of the overall system.

[0148] Preferably, the optimized communication and activation modeling function is configured as: ; wherein, represents the optimized communication and activation modeling function, represents the normalized power spectral density value, represents the image noise index, represents the blur index adjustment factor, represents the Bessel disturbance frequency term, which is used to simulate the periodic disturbance of PSD in underwater sound propagation;

[0149] The optimized self-healing trigger suppression function is configured as:

[0150] ;

[0151] wherein, represents the optimized self-healing trigger suppression function, represents the image blur index, represents the blur index amplification coefficient, which is used to control the disturbance intensity of the blur on activation, represents the color shift drift difference value, Represents the color difference fluctuation coefficients of the G and R channels, respectively. The above optimization makes the self-healing trigger suppression function more sensitive and adjustable to blur and color shift, and avoids false positives that cause frequent self-healing activations.

[0152] The function expression configuration of the updated fault risk score is:

[0153] ;

[0154] in, Indicates the updated fault risk score.

[0155] Specifically, in this embodiment, a team deployed the IoT-based underwater self-healing control system of the present invention in a Qing Dynasty shipwreck site conservation project in the East China Sea. The system consists of multiple intelligent sensor nodes. Node A09 was installed near the port side of the main hull and was responsible for underwater image acquisition, environmental monitoring, and acoustic link maintenance in key cultural relic areas.

[0156] The node has experienced frequent image blur, unstable links, and severe color deviation in the past two days. The system automatically calls the optimized risk scoring function. , combining image and acoustic data to comprehensively assess its failure risk.

[0157] Key input parameter values ​​collected by the node (after standardization):

[0158] Signal-to-noise ratio fluctuations , power spectral density , signal delay , topological complexity , image blur index , image noise index , color shift , color difference fluctuation coefficient , image channel consistency index ,Voltage , current , pH value window , mean temperature , temperature standard deviation , energy consumption rate , specific ion concentration , specific ion concentration , fuzzy index adjustment coefficient , image fuzzy magnification factor , salinity frequency term , Bessel perturbation frequency term , PSD interference amplitude term ;

[0159] The above parameters are brought into the function expression of the updated fault risk score, and the following can be obtained .

[0160] The system score judgment logic is as follows:

[0161]

[0162] Since , the A09 node is in the high risk interval (0.91), and the system immediately performs the following self-healing control:

[0163] Image acquisition migration: suspend the video acquisition task of A09, and perform the same angle retake by the adjacent node A08;

[0164] Data path reconstruction: reset the image and sound signal data stream, and bypass A09 using the standby communication link (audio 2);

[0165] Node state notification broadcast: through the state sharing protocol, the A09 fault state is conveyed to all nodes in the network;

[0166] Model feedback learning: the input and result of this scoring function are packaged and stored in the model incremental training data set;

[0167] Administrator alarm push: the system pops up a high-risk prompt on the control terminal, and generates a scoring curve chart for operation and maintenance analysis.

[0168] The embodiment shows that after introducing acoustic and optical interference factor modeling, the scoring function can accurately respond to implicit faults caused by image degradation, sound channel instability, etc., and realize higher accuracy of risk prediction and system adaptive ability.

[0169] A sensing node communication method with a self-healing mechanism, applied to the sensing node communication system with a self-healing mechanism, comprising:

[0170] Step S1, the node detection module 1 configures a plurality of sensing nodes, each sensing node comprising a plurality of sensors, each sensor being configured to collect multi-dimensional detection data at a predetermined period;

[0171] Step S2, the node communication module 2 establishes a communication connection for each sensing node in the self-organizing multi-hop network, and the detection data detected by each sensing node forms a local network health state map through a state sharing protocol, the local network health state map comprising a plurality of node state data, the node state data comprising node self-checking data, physical environment parameters, and chemical environment parameters;

[0172] Step S3, the preprocessing module 3 sequentially pre-processes and normalizes the node self-checking data, the physical environment parameters, and the chemical environment parameters to construct a training data set;

[0173] In step S4, the hybrid training module 4 trains the hybrid neural network including the time series modeling layer, the local feature extraction layer and the full connection layer according to the training data set. The time series modeling layer adopts the long short-term memory network to capture the long-term dependence features of the node states in the training data set. The local feature extraction layer adopts the convolutional neural network layer to extract the local abnormal fluctuation features within the time window in the training data set. The full connection layer fuses the long-term dependence features and the local abnormal fluctuation features to output the fault risk score.

[0174] In step S5, the self-healing control module 5 automatically triggers the self-healing response measures according to the node state data at the adjacent sensing nodes when the fault risk score at the current sensing node is greater than the preset self-healing risk threshold. The self-healing response measures include dynamic routing reconstruction, backup link activation and data load redistribution.

[0175] The above is only the preferred embodiment of the present application, and the protection scope of the present application is not limited to the above-mentioned embodiments. Any technical solutions falling within the concept of the present application shall be considered as falling within the protection scope of the present application. It should be noted that, for ordinary skilled persons in the art, some improvements and refinements without departing from the principles of the present application shall also be considered as falling within the protection scope of the present application.

Claims

1. A self-healing sensor node communication system, characterized in that, The application relates to a self-healing network health state detection method and device. The application comprises: a node detection module (1) configured to configure a plurality of sensing nodes, each of the sensing nodes comprising a plurality of sensors, each of the sensors being configured to collect multi-dimensional detection data at a preset period; a node communication module (2) connected to the node detection module (1) and configured to establish a communication connection for each of the sensing nodes by constructing a self-organizing multi-hop network, detection data detected by each of the sensing nodes forming a local network health state atlas by a state sharing protocol, the local network health state atlas comprising a plurality of node state data, the node state data comprising node self-checking data, physical environment parameters and chemical environment parameters; a preprocessing module (3) connected to the node communication module (2) and configured to sequentially perform preprocessing and normalization processing on the node self-checking data, the physical environment parameters and the chemical environment parameters to construct a training data set; a hybrid training module (4) connected to the preprocessing module (3) and configured to train a hybrid neural network according to the training data set, the hybrid neural network comprising a time series modeling layer, a local feature extraction layer and a full connection layer, the time series modeling layer being configured to capture long-term dependence features of each node state in the training data set by using a long short-term memory network, the local feature extraction layer being configured to extract local abnormal fluctuation features in a time window in the training data set by using a convolutional neural network layer, and the full connection layer being configured to fuse the long-term dependence features and the local abnormal fluctuation features to output a fault risk score; 2. The self-healing mechanism enabled sensor node communication system according to claim 1, wherein, a self-healing control module (5) connected to the hybrid training module (4) and the node communication module (2) and configured to automatically trigger a self-healing response measure according to the node state data of adjacent sensing nodes when the fault risk score at the current sensing node is greater than a preset self-healing risk threshold, the self-healing response measure comprising dynamic routing reconstruction, backup link activation and data load redistribution. The preprocessing module (3) comprises: a time series division unit (31) configured to record the node self-checking data, the physical environment parameters and the chemical environment parameters in time series and perform data segmentation by using a sliding window technology to form continuous time series data samples; a preprocessing unit (32) connected to the time series division unit (31) and configured to sequentially perform data cleaning, abnormal value detection and missing value filling on the time series data samples to obtain preprocessed data samples; 3. The self-healing sensor node communication system of claim 1, wherein: a normalization unit (33) connected to the preprocessing unit (32) and configured to perform normalization processing on multi-dimensional data in the preprocessed data samples according to a preset normalization method to form the training data set. The node self-checking data comprises node voltage, node current, node energy consumption state, signal-to-noise ratio, data packet loss rate and signal delay; The physical environment parameters comprise temperature, environmental pressure, salinity, dissolved oxygen concentration and environmental turbidity; 4. The self-healing sensor node communication system of claim 1, wherein, The chemical environment parameters comprise PH value and specific ion concentration. The application further comprises: a fault detection module (6) connected to the node communication module (2), configured to perform fault detection on each node state data and generate a corresponding fault detection result, and to process the fault actual score based on the fault detection result; an incremental training module (7) connected to the fault detection module (6) and the hybrid training module (4), configured to process a predicted deviation value based on the fault actual score and the fault risk score, and to perform dynamic adjustment of the weight parameters of the hybrid neural network based on the predicted deviation value and then retrain the hybrid neural network to perform incremental training and update of the hybrid neural network.

5. The self-healing sensor node communication system of claim 3, wherein: The function expression of the hybrid neural network includes: ； ； ； ； ； ； wherein, represents the failure risk score, represents the time window length, represents the time variable, represents the Gamma weight function constructed from the voltage and the temperature of the node, represents the voltage of the node, represents the standard deviation of the temperature, represents the mean of the temperature, represents the current perturbation modulation term, represents the compound excitation factor constructed from the current and the data packet loss rate, represents the sensitivity adjustment factor of the environmental pressure change, represents the historical current mean, represents the change interval of the environmental turbidity, represents the change amount of the signal-to-noise ratio, represents the communication and activation modeling function, represents the first-order Bessel function of the first kind measuring the environmental disturbance amplitude, represents the zero-order Bessel function of the second kind representing the strength of the node redundancy connectivity, represents the salinity disturbance frequency factor, represents the complexity parameter of the current network communication topology, represents the energy consumption change rate of the node, represents the time window width of the pH value, represents the delay attenuation and ion abnormal suppression function, represents the current maximum signal delay, represents the historical average signal delay, represents the data channel congestion factor, represents the time-dependent adjustment coefficient, represents the mean of the specific ion concentration, represents the standard deviation of the specific ion concentration, represents the self-healing trigger inhibition function, represents the self-healing response time activation coefficient, represents the state consistency measurement factor between adjacent nodes, represents the natural logarithm function, represents the complexity quantitative index of the ion species in the current environment.

6. The self-healing sensor node communication system of claim 5, wherein: The node state data further includes underwater acoustic interference data and underwater optical interference data, the underwater acoustic interference data includes power spectral density, and the underwater optical interference data includes optical image blurriness, color deviation drift, and image noise index.

7. The self-healing sensor node communication system according to claim 6, wherein, Further comprising a function optimization module (8) connected to the node communication module (2) and the hybrid training module (4), configured to optimize the communication and activation modeling function based on the power spectral density, the optical image blurriness, and the image noise index, and to optimize the self-healing trigger suppression function based on the image blurriness and the color deviation drift, and to update the fault risk score based on the optimized communication and activation modeling function and the optimized self-healing trigger suppression function.

8. The self-healing mechanism enabled sensor node communication system according to claim 7, wherein: The optimized communication and activation modeling function is configured as: ; wherein, represents the optimized communication and activation modeling function, represents the normalized power spectral density value, represents the image noise index, represents the blur index adjustment factor, represents the Bessel perturbation frequency term, used to model the periodic perturbation of the PSD in underwater acoustic propagation; The optimized self-healing trigger suppression function is configured as: ； wherein, represents the optimized self-healing trigger inhibition function, represents the image blurriness index, represents the blurriness index amplification coefficient, for controlling the blurriness on the activated interference strength, represents the color cast drift difference, respectively represent the color difference fluctuation coefficient of G, R channel; The function expression of the updated fault risk score is configured as: ； wherein, represents the updated failure risk score.

9. A method for communication of a sensing node with self-healing mechanism, applied to the sensing node communication system with self-healing mechanism as claimed in any one of claims 1-8, characterized in that, including: Step S1, the node detection module (1) configures a plurality of sensing nodes, each of the sensing nodes includes a plurality of sensors, and each of the sensors is configured to collect multi-dimensional detection data at a preset period; Step S2, the node communication module (2) establishes communication connections for each of the sensing nodes in the self-organizing multi-hop network, and the detection data detected by each of the sensing nodes forms a local network health state map through a state sharing protocol, the local network health state map includes a plurality of node state data, and the node state data includes node self-checking data, physical environment parameters, and chemical environment parameters; Step S3, the preprocessing module (3) sequentially performs preprocessing and normalization processing on the node self-checking data, the physical environment parameters, and the chemical environment parameters to construct a training data set; Step S4, the hybrid training module (4) trains a hybrid neural network based on the training data set, the hybrid neural network includes a time series modeling layer, a local feature extraction layer, and a full connection layer, the time series modeling layer uses a long short-term memory network to capture long-term dependence features of each node state in the training data set, the local feature extraction layer uses a convolutional neural network layer to extract local abnormal fluctuation features within a time window in the training data set, and the full connection layer fuses the long-term dependence features and the local abnormal fluctuation features to output a fault risk score; Step S5, the self-healing control module (5) automatically triggers a self-healing response measure according to the node status data of the neighboring sensor nodes when the fault risk score at the current sensor node is greater than the preset self-healing risk threshold, and the self-healing response measure includes dynamic routing reconstruction, backup link activation and data load redistribution.

Images