Intelligent diagnosis method and system for node failure of intelligent space sensor network

Abstract

The application provides a node fault intelligent diagnosis method and system for an intelligent space sensing network, real-time monitoring data sets of all sensing nodes in a target sensing network are acquired, state feature extraction processing is performed on the real-time monitoring data sets, state feature sets of each sensing node are generated, multi-dimensional fault matching processing is performed on the state feature sets based on a preset fault diagnosis strategy model, a candidate fault type set of the sensing node is determined, a node fault processing strategy set of the sensing node is generated according to a strategy mapping relationship between the candidate fault type set and a preset fault processing knowledge base, node optimization operation is performed on a fault node in the target sensing network based on the node fault processing strategy set, and the optimized node operation parameters are fed back to a control terminal of the target sensing network. The application effectively avoids the problem of system stability decline caused by traditional single-point repair, and ensures the continuous and reliable operation of the overall network performance after fault repair.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing and fault diagnosis, and in particular to a node fault intelligent diagnosis method and system of an intelligent space sensor network. BACKGROUND

[0002] In the technical field of intelligent space sensor network, node fault diagnosis is the core link to ensure the reliable operation of the network, and the key lies in realizing accurate positioning and repair of faults through multi-dimensional data analysis. The existing technology usually adopts a single-dimensional diagnosis mechanism, such as monitoring equipment operation parameter abnormalities or network communication interruption events alone for fault judgment, or linear compensation processing of environmental interference through a fixed threshold method; such methods can realize basic diagnosis in a single fault scenario, but in actual application, they have serious diagnostic blind spots when facing composite faults caused by coupling of hardware performance degradation, network topology fluctuation and environmental interference: traditional single-dimensional data collection leads to the loss of associated characteristics of equipment load abnormalities and communication link degradation, so that network performance degradation caused by hardware faults is misjudged as an independent event; at the same time, the static environmental interference compensation mechanism cannot capture the nonlinear influence of temperature gradient and electromagnetic noise superposition on node stability, resulting in a high intermittent fault missed detection rate. In addition, the existing step-by-step repair strategy lacks timing coordination in the hardware reset and network reconstruction process, which easily causes parameter calibration conflicts, resulting in a decrease in system stability after node optimization. How to realize feature fusion of multi-source heterogeneous data, cross-dimensional accurate identification of composite faults and adaptive collaborative execution of repair strategies has become a key technical bottleneck to improve the fault diagnosis efficiency of intelligent space sensor networks. SUMMARY

[0003] Therefore, the embodiments of the present application provide at least a node fault intelligent diagnosis method and system of an intelligent space sensor network.

[0004] The technical scheme of the embodiments of the present application is as follows:

[0005] On the one hand, the embodiments of the present application provide a node fault intelligent diagnosis method of an intelligent space sensor network, comprising:

[0006] obtaining a real-time monitoring data set of all sensor nodes in a target sensor network, the real-time monitoring data set including equipment operation parameters, network communication parameters and environmental interference parameters of each sensor node;

[0007] performing state feature extraction processing on the real-time monitoring data set to generate a state feature set of each sensor node, the state feature set including equipment operation state features, network connection stability features and environmental interference associated features;

[0008] Based on a preset fault diagnosis strategy model, multi-dimensional fault matching processing is performed on the state feature set to determine a candidate fault type set of the sensing node, and the candidate fault type set contains at least one candidate fault type associated with the state feature set;

[0009] According to a strategy mapping relationship between the candidate fault type set and a preset fault processing knowledge base, a node fault processing strategy set of the sensing node is generated, and the node fault processing strategy set contains an optimal processing strategy for each candidate fault type;

[0010] Based on the node fault processing strategy set, a node optimization operation is performed on a fault node in the target sensing network, and an optimized node running parameter is fed back to a control terminal of the target sensing network.

[0011] In another aspect, an embodiment of the present application provides a computer system, including a memory and a processor, the memory storing a computer program capable of running on the processor, and the processor implements the steps in the above method when executing the program.

[0012] The node fault intelligent diagnosis method of the intelligent space sensing network provided by the present application constructs a multi-dimensional monitoring data set by synchronously collecting device running parameters, network communication parameters and environmental interference parameters in real time, and generates a state feature set containing device running state features, network connection stability features and environmental interference correlation features based on state feature extraction processing. The above features are processed by a fault diagnosis strategy model for multi-dimensional fault matching, accurately identifying a composite fault type set under the coupling action of hardware performance degradation, network link abnormalities and environmental interference, and then generating a node optimization strategy that integrates hardware parameter calibration, network topology reconstruction and environmental resistance enhancement, thereby realizing cross-dimensional collaborative diagnosis and adaptive repair of sensing node fault causes under complex working conditions. Through joint modeling of device energy consumption trends, network packet loss correlation matrices and interference superposition effects in the feature extraction process, the fault diagnosis process can simultaneously capture the nonlinear correlation between the internal hardware state evolution of the node, the external network communication quality fluctuation and the environmental interference propagation path, significantly improving the detection sensitivity of intermittent faults and hidden faults. At the same time, the generation and execution mechanism of the node optimization strategy optimizes the timing of hardware reset instructions and communication link reconstruction protocols in a coordinated manner, effectively avoiding the problem of system stability decline caused by traditional single-point repair, and ensuring the continuous and reliable operation of the overall network performance after fault repair.

[0013] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, but not limiting the technical solutions of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 An implementation flowchart of a node fault intelligent diagnosis method of an intelligent space sensing network provided by an embodiment of the present application is shown in the figure.

[0015] Figure 2 A hardware entity diagram of a computer system provided by an embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0016] An embodiment of the present application provides a node fault intelligent diagnosis method of an intelligent space sensing network, which can be executed by a processor of a computer system. The computer system can be a server, a notebook computer, a tablet computer, a desktop computer, or any device with data processing capability.

[0017] Figure 1 An implementation flowchart of a node fault intelligent diagnosis method of an intelligent space sensing network provided by an embodiment of the present application is shown in the figure. Figure 1 As shown in the figure, the method comprises the following steps:

[0018] Step S100: Obtain a real-time monitoring data set of all sensing nodes in a target sensing network, wherein the real-time monitoring data set comprises device running parameters, network communication parameters, and environmental interference parameters of each sensing node.

[0019] The device running parameters are various data describing the running state of the sensing node itself, such as the energy consumption of the node and the load condition of the hardware components. The energy consumption reflects the energy consumed by the node during running, and the load condition reflects the working intensity of the hardware components. The network communication parameters are data related to the communication of the sensing node in the network, such as the packet loss rate, signal strength, and communication interruption frequency. The packet loss rate indicates the proportion of lost data packets during data transmission, the signal strength reflects the strength of the communication signal between nodes, and the communication interruption frequency reflects the frequency of communication connection interruption. The environmental interference parameters are related parameters of the environment in which the sensing node is located, such as temperature gradient, electromagnetic noise, and humidity. The temperature gradient indicates the rate of change of temperature in space, electromagnetic noise can interfere with the communication and running of the node, and the change of humidity can affect the signal attenuation.

[0020] The specific implementation of obtaining the real-time monitoring data set can be achieved by installing various sensors on the sensing nodes to collect corresponding data. For the device operation parameters, current sensors and voltage sensors can be used to measure the energy consumption of the nodes, and load sensors can be used to monitor the load of the hardware components. For the network communication parameters, network monitoring software can be used to count data such as packet loss rate, signal strength, and communication interruption frequency. For the environmental interference parameters, temperature sensors can be used to measure temperature, electromagnetic sensors can be used to detect electromagnetic noise, and humidity sensors can be used to obtain humidity data. For example, in a sensing network of a smart building, each room's sensing node will be equipped with temperature sensors, humidity sensors, current sensors, etc. Real-time collection of device operation parameters, network communication parameters, and environmental interference parameters is then performed through these sensors, and these data are transmitted to the data center for aggregation to form the real-time monitoring data set.

[0021] Step S200: performing state feature extraction processing on the real-time monitoring data set to generate a state feature set of each sensing node, the state feature set including device operation state features, network connection stability features, and environmental interference correlation features.

[0022] The state feature extraction processing is an analysis and processing of the real-time monitoring data set to extract key features that can reflect the state of the sensing node. The device operation state features are features that describe the running condition of the sensing node device, such as time series-based energy consumption trend and hardware component load fluctuation curve. The energy consumption trend reflects the change of node energy consumption over time, and the hardware component load fluctuation curve reflects the dynamic change of hardware component load. The network connection stability features are used to measure the connection stability of the sensing node in the network, including the correlation matrix between packet loss rate and signal strength and the communication interruption frequency distribution graph. The correlation matrix represents the mutual relationship between the packet loss rate and the signal strength, and the communication interruption frequency distribution graph shows the distribution of communication interruptions at different time periods. The environmental interference correlation features are features that reflect the influence of environmental interference on the sensing node, including the interference superposition effect of temperature gradient and electromagnetic noise and the quantitative influence coefficient of humidity change on signal attenuation. The interference superposition effect reflects the influence of the combined action of temperature gradient and electromagnetic noise on the node, and the quantitative influence coefficient is used to measure the specific influence degree of humidity change on signal attenuation.

[0023] As an implementation, the step S200 of performing state feature extraction processing on the real-time monitoring data set to generate a state feature set of each sensing node can specifically include the following steps S210-S250:

[0024] Step S210: Perform periodic fluctuation pattern recognition processing on the device operation parameters to generate device operation state features, which include time series-based energy consumption change trends and hardware component load fluctuation curves.

[0025] Periodic fluctuation pattern recognition processing is an analysis of device operation parameters to identify periodic fluctuation patterns therein. Time series-based energy consumption change trends refer to arranging energy consumption data of a node in chronological order and analyzing its change trend over time. Hardware component load fluctuation curves are curves showing the change of hardware component load data over time.

[0026] In specific implementation, a time series analysis algorithm such as ARIMA (Autoregressive Integrated Moving Average Model) can be used to process device operation parameters. First, the collected device operation parameters are preprocessed, including data cleaning, normalization, and other operations to remove noise and outliers. Then, the ARIMA model is used to model and predict energy consumption data and hardware component load data to obtain energy consumption change trends and hardware component load fluctuation curves. For example, in a sensor network in an industrial production environment, periodic fluctuation pattern recognition processing is performed on the device operation parameters of a certain sensor node. Energy consumption data and hardware component load data of the node within a week are collected and analyzed using the ARIMA model to obtain periodic change trends of energy consumption at different time periods of the day and corresponding fluctuation curves of hardware component load, which can be used to determine whether the device operation state of the node is normal.

[0027] Step S220: Perform link abnormal event correlation analysis processing on the network communication parameters to extract network connection stability features, which include an association matrix between packet loss rate and signal strength and a communication interruption frequency distribution graph.

[0028] Link abnormal event correlation analysis processing is an analysis of link abnormal events in network communication parameters to find their correlation. The association matrix between packet loss rate and signal strength is a two-dimensional matrix representing the degree of association between packet loss rate and signal strength at different values. The communication interruption frequency distribution graph is a graph showing the frequency of communication interruptions according to time or other dimensions.

[0029] In a specific implementation, a link abnormal event correlation analysis process can be performed using an association rule mining algorithm, such as the Apriori algorithm. First, the collected network communication parameters are discretized, and the packet loss rate and signal strength are divided into different intervals. Then, the Apriori algorithm is used to mine the association rules between the packet loss rate and the signal strength, and an association matrix is generated. For the communication interruption frequency distribution graph, statistical analysis methods can be used to count the time and frequency of communication interruptions, and a distribution graph is drawn. For example, in a wireless sensor network, the network communication parameters of a certain sensor node are subjected to a link abnormal event correlation analysis process. The packet loss rate, signal strength, and communication interruption frequency data of the node within a month are collected, and the Apriori algorithm is used to mine the association rules that the packet loss rate will significantly increase when the signal strength is below a certain threshold, and generate an association matrix. At the same time, the time and frequency of communication interruptions are counted, and a communication interruption frequency distribution graph is drawn. Through these features, the network connection stability of the node can be evaluated.

[0030] Step S230: Perform multi-source interference coupling analysis processing on the environmental interference parameters, and construct the environmental interference correlation features, which include the interference superposition effect of temperature gradient and electromagnetic noise and the quantitative influence coefficient of humidity change on signal attenuation.

[0031] Multi-source interference coupling analysis processing is an analysis of multiple interference sources in environmental interference parameters, considering their coupling effects. The interference superposition effect of temperature gradient and electromagnetic noise refers to the influence of the combined action of temperature gradient and electromagnetic noise on the sensor node. The quantitative influence coefficient of humidity change on signal attenuation is a numerical value that measures the specific influence of humidity change on signal attenuation.

[0032] In a specific implementation, a multiple regression analysis method can be used to analyze and process the multi-source interference coupling. First, historical data of environmental interference parameters are collected, including temperature gradient, electromagnetic noise, humidity, etc. Then, a multiple regression model is established, with signal attenuation as the dependent variable, temperature gradient, electromagnetic noise, and humidity as the independent variables. The parameters of the model are estimated by the least squares method, and the quantitative influence coefficient of humidity change on signal attenuation is obtained. For the superimposed effect of temperature gradient and electromagnetic noise, the experimental design method can be used to conduct experiments under different temperature gradient and electromagnetic noise conditions, measure the performance indicators of the nodes, and analyze their superimposed effect. For example, in an intelligent home sensor network, the environmental interference parameters of a certain sensor node are analyzed and processed by the multi-source interference coupling. The temperature gradient, electromagnetic noise, humidity, and signal attenuation data of the node under different environmental conditions are collected, a model is established using the multiple regression analysis method, and the quantitative influence coefficient of 0.5 dB signal attenuation per 1% increase in humidity is obtained. At the same time, through experimental design, it is analyzed that under the superimposed conditions of high temperature and strong electromagnetic noise, the communication performance of the node is significantly reduced.

[0033] Step S240: time-domain alignment processing of the energy consumption change trend and the hardware component load fluctuation curve to generate a first fusion feature vector; spatial mapping processing of the correlation matrix and the communication interruption frequency distribution map to generate a second fusion feature vector; weight superposition processing of the interference superposition effect and the quantitative influence coefficient to generate a third fusion feature vector.

[0034] The time-domain alignment processing is to align the energy consumption change trend and the hardware component load fluctuation curve in the time dimension, so that they have corresponding values at the same time point. The first fusion feature vector is a feature vector obtained by fusing the energy consumption change trend and the hardware component load fluctuation curve after time-domain alignment. The spatial mapping processing is to map the correlation matrix and the communication interruption frequency distribution map in the spatial dimension, so that they can be represented in the same space. The second fusion feature vector is a feature vector obtained by fusing the correlation matrix and the communication interruption frequency distribution map after spatial mapping. The weight superposition processing is to assign different weights to the interference superposition effect and the quantitative influence coefficient, and then superimpose them. The third fusion feature vector is a feature vector obtained by fusing the interference superposition effect and the quantitative influence coefficient after weight superposition.

[0035] In a specific implementation, for time domain alignment processing, a time synchronization algorithm such as NTP (Network Time Protocol) can be used to ensure that the timestamps of the energy consumption trend and the hardware component load fluctuation curve are consistent. Then, the aligned data is normalized, and then vector splicing is performed to generate a first fused feature vector. For spatial mapping processing, a spatial transformation algorithm such as PCA (Principal Component Analysis) can be used to map the correlation matrix and the communication interruption frequency distribution map to a low-dimensional space, and then perform vector splicing to generate a second fused feature vector. For weight superposition processing, the weights of the interference superposition effect and the quantization influence coefficient can be determined according to experience or experiment, and then they are weighted and summed to generate a third fused feature vector. For example, in an intelligent agricultural sensor network, the features of a certain sensor node are fused. The NTP protocol is used to time-domain align the energy consumption trend and the hardware component load fluctuation curve, and then the normalized data is spliced into a first fused feature vector. The PCA algorithm is used to map the correlation matrix and the communication interruption frequency distribution map to a two-dimensional space, and splice them into a second fused feature vector. According to the experiment, the weight of the interference superposition effect is 0.6, and the weight of the quantization influence coefficient is 0.4. They are weighted and summed to generate a third fused feature vector.

[0036] Step S250: generating the state feature set containing multi-dimensional association indicators according to the feature complementary relationship between the first fused feature vector, the second fused feature vector, and the third fused feature vector.

[0037] The feature complementary relationship means that the first fused feature vector, the second fused feature vector, and the third fused feature vector each contain different information, which can complement each other. The multi-dimensional association indicators mean that the state feature set contains multi-dimensional association information, which can comprehensively reflect the state of the sensor node.

[0038] In a specific implementation, a feature fusion algorithm such as a neural network fusion algorithm can be used to generate the state feature set. First, the first fused feature vector, the second fused feature vector, and the third fused feature vector are input into a multi-layer perceptron (MLP) neural network. The MLP neural network includes an input layer, a hidden layer, and an output layer. The input layer receives the three fused feature vectors, the hidden layer performs nonlinear transformation on the input, and the output layer outputs the state feature set containing multi-dimensional association indicators. For example, in an intelligent traffic sensor network, the first fused feature vector, the second fused feature vector, and the third fused feature vector of a certain sensor node are input into an MLP neural network with two hidden layers. After training and learning, the neural network outputs a state feature set containing multi-dimensional association indicators such as device operation, network connection, and environmental interference, which can more comprehensively understand the state of the node.

[0039] Step S300: based on the preset fault diagnosis strategy model, multi-dimensional fault matching processing is performed on the state feature set to determine a candidate fault type set of the sensor node, and the candidate fault type set contains at least one candidate fault type having an association relationship with the state feature set.

[0040] The preset fault diagnosis strategy model is a model for diagnosing faults of a sensor node and is pre-constructed, and contains association relationships between various fault types and state features. The multi-dimensional fault matching processing is to analyze the state feature set from multiple dimensions to find fault types matching the state feature set. The candidate fault type set is a set of possible fault types determined according to the matching result.

[0041] As an implementation manner, in the step S300, based on the preset fault diagnosis strategy model, multi-dimensional fault matching processing is performed on the state feature set to determine a candidate fault type set of the sensor node, and the step S300 can specifically include the following steps S310-S350.

[0042] Step S310: inputting the first fusion feature vector into a hardware degradation analysis module of the fault diagnosis strategy model to generate a hardware-related fault set by matching a preset hardware failure mode library, and the hardware failure mode library contains power module abnormality, sensor drift, and circuit aging feature mode.

[0043] The hardware degradation analysis module is a module in the fault diagnosis strategy model specially used for analyzing hardware degradation. The preset hardware failure mode library is a database containing various hardware failure mode features. The power module abnormality refers to a situation that a power module cannot work normally, such as unstable voltage and insufficient power. The sensor drift refers to a phenomenon that a measurement value of a sensor deviates. The circuit aging feature mode refers to a feature mode of performance degradation of a circuit with an increase in use time.

[0044] In a specific implementation, the hardware degradation analysis module can use a rule-based matching algorithm. First, compare the first fusion feature vector with each feature pattern in the hardware failure mode library, and calculate the similarity between them. Similarity can be calculated using Euclidean distance, cosine similarity, and other measurement methods. When the similarity exceeds a certain threshold, it is considered that the feature pattern matches the first fusion feature vector, and the corresponding hardware failure type is added to the hardware-related fault set. For example, in a smart medical sensor network, input the first fusion feature vector of a certain sensor node into the hardware degradation analysis module. The module compares the first fusion feature vector with the power module anomaly, sensor drift, and circuit aging feature patterns in the hardware failure mode library. If the similarity between the first fusion feature vector and the power module anomaly feature pattern is 0.8, which exceeds the preset threshold of 0.7, then add the power module anomaly to the hardware-related fault set.

[0045] Step S320: input the second fusion feature vector into the network topology analysis module of the fault diagnosis strategy model, and generate a network-related fault set by identifying communication link breakage features and routing table anomaly features. The communication link breakage features include neighbor node unreachable events and data retransmission number mutation patterns.

[0046] The network topology analysis module is a module in the fault diagnosis strategy model for analyzing network topology and communication. Communication link breakage features refer to related features of communication link interruption, neighbor node unreachable events indicate that a node cannot communicate with adjacent nodes, and data retransmission number mutation patterns refer to sudden increases in data retransmission number. Routing table anomaly features refer to features of errors or inconsistencies in routing table information.

[0047] In a specific implementation, the network topology analysis module can use a machine learning classification algorithm, such as support vector machine (SVM). First, use historical data to train the SVM, with the second fusion feature vector as input and the communication link breakage features and routing table anomaly features as output categories. After training, input the current second fusion feature vector into the trained SVM, and determine whether there is a communication link breakage or routing table anomaly based on the output result. If so, add the corresponding network failure type to the network-related fault set. For example, in a smart grid sensor network, input the second fusion feature vector of a certain sensor node into the network topology analysis module. The module uses the trained SVM to classify the second fusion feature vector and determines that there is a neighbor node unreachable event, adding the communication link breakage-related fault type to the network-related fault set.

[0048] Step S330: inputting the third fusion feature vector into an environmental resistance analysis module of the fault diagnosis strategy model, generating an environmental related fault set by analyzing the deviation between the interference superposition effect and the node anti-interference threshold, and adjusting the node anti-interference threshold according to stable operation data under historical environmental parameters.

[0049] The environmental resistance analysis module is a module in the fault diagnosis strategy model for analyzing the resistance of nodes to environmental interference. The deviation between the interference superposition effect and the node anti-interference threshold refers to the difference between the actual value of the interference superposition effect and the anti-interference threshold that the node can withstand. The node anti-interference threshold is determined according to stable operation data of the node under historical environmental parameters, and it will be adjusted as the environment changes.

[0050] In specific implementation, the environmental resistance analysis module can use a threshold comparison algorithm. First, compare the interference superposition effect value in the third fusion feature vector with the node anti-interference threshold. If the interference superposition effect value exceeds the node anti-interference threshold, it is considered that there is an environmental related fault, and the corresponding fault type is added to the environmental related fault set. For example, in an intelligent meteorological sensor network, input the third fusion feature vector of a certain sensor node into the environmental resistance analysis module. The module compares the interference superposition effect value with the node anti-interference threshold adjusted according to historical environmental data, and finds that the interference superposition effect value exceeds the threshold, so the fault type caused by environmental interference is added to the environmental related fault set.

[0051] Step S340: fault coupling degree evaluation is performed on the hardware related fault set, the network related fault set and the environmental related fault set, and the conditional trigger probability between any two fault types is calculated.

[0052] Fault coupling degree evaluation is to analyze the correlation degree between different fault types in the hardware related fault set, the network related fault set and the environmental related fault set. The conditional trigger probability refers to the probability of occurrence of another fault type under the condition of occurrence of a fault type.

[0053] In specific implementation, Bayesian network can be used for fault coupling degree evaluation. First, construct a Bayesian network, take the fault types in the hardware related fault set, the network related fault set and the environmental related fault set as nodes, and determine the conditional probability distribution between nodes according to historical fault data. Then, use Bayesian inference algorithm to calculate the conditional trigger probability between any two fault types. For example, in an intelligent industrial sensor network, a Bayesian network is constructed using historical fault data, and power module abnormality, communication link breakage and environmental interference fault are taken as nodes. Through Bayesian inference algorithm, it is calculated that the conditional trigger probability of communication link breakage under the condition of power module abnormality is 0.3.

[0054] Step S350: According to the conditional trigger probability, the fault types with strong correlation are combined and optimized to generate the candidate fault type set containing the main fault type and the derived fault type.

[0055] Strong correlation refers to a high conditional trigger probability between two fault types. The combined and optimized processing is to combine and optimize the fault types with strong correlation to determine the main fault type and the derived fault type. The main fault type is the main cause of other faults, and the derived fault type is other faults caused by the main fault type.

[0056] In specific implementation, a clustering algorithm such as K-Means algorithm can be used for combined and optimized processing. First, a similarity matrix between fault types is constructed according to the conditional trigger probability. Then, the K-Means algorithm is used to cluster the fault types, and the fault types with high conditional trigger probability are clustered into a class. In each class, the fault type with the maximum conditional trigger probability is selected as the main fault type, and the other fault types are selected as the derived fault types. For example, in an intelligent logistics sensor network, a similarity matrix is constructed according to the conditional trigger probability evaluated by fault coupling degree, and the K-Means algorithm is used to cluster the fault types into two classes. In one of the classes, the conditional trigger probability of the power module anomaly is the largest, and it is selected as the main fault type, and the communication link breakage and sensor drift are selected as the derived fault types, to generate a candidate fault type set containing the main fault type and the derived fault type.

[0057] Step S400: According to the strategy mapping relationship between the candidate fault type set and the preset fault handling knowledge base, a node fault handling strategy set of the sensor node is generated, and the node fault handling strategy set contains the optimal handling strategy for each candidate fault type.

[0058] The preset fault handling knowledge base is a database storing handling strategies corresponding to various fault types. The strategy mapping relationship refers to the correspondence between the fault types in the candidate fault type set and the handling strategies in the fault handling knowledge base. The node fault handling strategy set is a set of optimal handling strategies for each candidate fault type generated according to the strategy mapping relationship.

[0059] As an implementation, the step S400, according to the strategy mapping relationship between the candidate fault type set and the preset fault handling knowledge base, generates a node fault handling strategy set of the sensor node, which can specifically include the following steps S410-S440:

[0060] Step S410: retrieving a set of baseline handling strategies corresponding to the main fault type from the fault handling knowledge base, the set of baseline handling strategies including a hardware reset instruction, a communication link reestablishment protocol, and an environment adaptive calibration scheme.

[0061] The set of baseline handling strategies is a series of basic handling strategies pre-established for the main fault type. The hardware reset instruction is an instruction for resetting the hardware device to restore it to the initial state. The communication link reestablishment protocol is a protocol for reestablishing the communication link to ensure normal data transmission. The environment adaptive calibration scheme is a scheme for automatically adjusting the node parameters according to environmental changes to adapt to the environment.

[0062] In specific implementation, a keyword matching algorithm can be used to retrieve the set of baseline handling strategies corresponding to the main fault type from the fault handling knowledge base. The name of the main fault type is taken as the keyword, and a search is performed in the fault handling knowledge base to find the handling strategies containing the keyword, which are added to the set of baseline handling strategies. For example, in an intelligent education sensor network, the main fault type is power module abnormality, and a keyword matching algorithm is used to search the fault handling knowledge base for handling strategies containing "power module abnormality", and the hardware reset instruction, the communication link reestablishment protocol, and the environment adaptive calibration scheme are added to the set of baseline handling strategies.

[0063] Step S420: performing strategy derivation processing on the set of baseline handling strategies according to the derived fault type: embedding a voltage compensation parameter for circuit aging characteristics in the hardware reset instruction; adding a path optimization rule based on the load state of neighbor nodes in the communication link reestablishment protocol; and configuring an interference shielding threshold in the environment adaptive calibration scheme.

[0064] The strategy derivation processing is a further optimization and extension of the set of baseline handling strategies according to the derived fault type. The voltage compensation parameter for circuit aging characteristics is a parameter set to compensate for voltage drop caused by circuit aging. The path optimization rule based on the load state of neighbor nodes is a rule for selecting the optimal path according to the load state of neighbor nodes when reestablishing the communication link. The interference shielding threshold is a threshold for shielding environmental interference.

[0065] In a specific implementation, for the hardware reset instruction, according to the analysis result of the circuit aging feature, the specific value of the voltage compensation parameter is determined and embedded into the hardware reset instruction. For the communication link reconstruction protocol, by monitoring the load state of the neighbor node, the path optimization rule is formulated and added to the communication link reconstruction protocol. For the environment adaptive calibration scheme, according to the type and intensity of the environmental interference, the interference shielding threshold is configured. For example, in an intelligent tourism sensor network, the derived fault types include circuit aging and communication link instability. For the hardware reset instruction, after analyzing the circuit aging feature, the voltage compensation parameter is determined as 5V and embedded into the hardware reset instruction. For the communication link reconstruction protocol, after monitoring the load condition of the neighbor node, the path optimization rule is formulated as selecting the neighbor node with the lightest load as the communication path and added to the communication link reconstruction protocol. For the environment adaptive calibration scheme, according to the intensity of the electromagnetic noise, the interference shielding threshold is configured as 10dB.

[0066] Step S430: performing execution priority evaluation on the derived processing strategies, the execution priority evaluation being based on a product index of a strategy execution time consumption and a fault repair probability.

[0067] The execution priority evaluation is to sort the derived processing strategies and determine their execution order. The strategy execution time consumption is the time required for executing a certain processing strategy. The fault repair probability is the probability of successfully repairing a fault by a certain processing strategy. The product index is the product of the strategy execution time consumption and the fault repair probability.

[0068] In a specific implementation, the execution time consumption and the fault repair probability of each processing strategy can be statistically obtained from historical data. Then, the product index of each processing strategy is calculated and the processing strategies are sorted in the order of the product index from small to large. For example, in an intelligent financial sensor network, there are three derived processing strategies: strategy A, strategy B and strategy C. According to the historical data, the execution time consumption of strategy A is 10 minutes and the fault repair probability is 0.8; the execution time consumption of strategy B is 15 minutes and the fault repair probability is 0.6; the execution time consumption of strategy C is 20 minutes and the fault repair probability is 0.5. The product index of strategy A is calculated as 8, the product index of strategy B is calculated as 9, and the product index of strategy C is calculated as 10. The processing strategies are sorted in the order of the product index from small to large as strategy A, strategy B and strategy C.

[0069] Step S440: selecting a processing strategy that meets a preset repair success rate according to the sorting result to generate the node fault processing strategy set, and configuring an execution condition constraint and a rollback mechanism for each processing strategy.

[0070] The preset repair success rate is a preset minimum probability of success of a repair of a fault by a processing strategy. The execution condition constraint refers to a condition that needs to be met for execution of a processing strategy. The rollback mechanism refers to a mechanism for restoring a node to a state before execution when execution of a processing strategy fails.

[0071] In a specific implementation, processing strategies are selected from the sorting result one by one until the cumulative repair success rate of the selected processing strategies reaches the preset repair success rate. The selected processing strategies are added to the node fault processing strategy set. For each processing strategy, an execution condition constraint is configured according to its characteristics and requirements. For example, for a hardware reset instruction, the execution condition constraint can be that the hardware state of the node is normal. At the same time, a rollback mechanism is set for each processing strategy, which can restore the node to the state before execution when the processing strategy fails. For example, in an intelligent entertainment sensor network, the preset repair success rate is 0.9. The sorting result is strategy A, strategy B, and strategy C. The fault repair probability of strategy A is 0.8, and the fault repair probability of strategy B is 0.6. Strategy A is selected first, and the cumulative repair success rate is 0.8, which does not reach the preset repair success rate. Strategy B is selected, and the cumulative repair success rate is 0.8 + (1-0.8) x 0.6 = 0.92, which reaches the preset repair success rate. Strategy A and strategy B are added to the node fault processing strategy set. Strategy A is configured with an execution condition constraint that the hardware temperature of the node is within a normal range, and a rollback mechanism is set for strategy A and strategy B, which restores the hardware configuration and communication parameters of the node when execution fails.

[0072] Step S500: performing a node optimization operation on a fault node in the target sensor network based on the node fault processing strategy set, and feeding back optimized node running parameters to a control terminal of the target sensor network.

[0073] The node optimization operation is an operation of repairing and optimizing a fault node according to the node fault processing strategy set. The optimized node running parameters refer to device running parameters, network communication parameters, and environmental interference parameters of the node after the node optimization operation. The control terminal is a terminal device for monitoring and managing the target sensor network.

[0074] As an implementation manner, the step S500 of performing a node optimization operation on a fault node in the target sensor network based on the node fault processing strategy set can specifically include the following steps S510-S550:

[0075] Step S510: selecting a target processing strategy with the highest priority according to the execution priority evaluation result, and analyzing a hardware reset instruction in the target processing strategy and injecting the voltage compensation parameter.

[0076] The target processing strategy is the highest priority processing strategy selected from the set of node fault processing strategies. The hardware reset instruction is parsed by extracting and understanding the specific operation steps and parameters in the hardware reset instruction. The voltage compensation parameter is injected into the hardware reset instruction by adding the previously configured voltage compensation parameter for the aging characteristics of the circuit.

[0077] In a specific implementation, according to the execution priority evaluation result, the highest priority processing strategy is selected as the target processing strategy. The hardware reset instruction in the target processing strategy is parsed using a parsing algorithm to extract the hardware devices to be reset and the operation steps. Then, the voltage compensation parameter is injected into the hardware reset instruction. For example, in an intelligent energy sensor network, the execution priority evaluation result shows that the priority of strategy A is the highest, and strategy A is selected as the target processing strategy. The hardware reset instruction in strategy A is parsed to obtain the hardware devices to be reset, which are sensor modules, and the operation steps, which are to turn off the power supply, wait for 5 seconds, and then turn on the power supply. The previously configured voltage compensation parameter 3V is injected into the hardware reset instruction, so that an additional 3V voltage compensation is provided for the sensor modules when the power supply is turned on.

[0078] Step S520: After the hardware reset operation is performed on the fault node, the recovery state of the energy consumption change trend and the load fluctuation curve is monitored in real time, and if the preset stable interval is not reached, the communication link reconstruction protocol is triggered.

[0079] The recovery state of the energy consumption change trend and the load fluctuation curve refers to whether the energy consumption change trend and the load fluctuation curve return to the normal state after the hardware reset operation is performed on the fault node. The preset stable interval is the normal range of the energy consumption change trend and the load fluctuation curve that is set in advance. The communication link reconstruction protocol is a protocol configured in advance for reestablishing the communication link.

[0080] In a specific implementation, after the hardware reset operation is performed on the fault node, the sensor is used to monitor the energy consumption change trend and the load fluctuation curve in real time. The monitored data is compared with the preset stable interval, and if the preset stable interval is not reached, the communication link reconstruction protocol is triggered. For example, in an intelligent traffic sensor network, after the hardware reset operation is performed on the fault node, the energy consumption change trend and the load fluctuation curve are monitored in real time. The preset stable interval is that the energy consumption change rate is within ±5%, and the fluctuation range of the load fluctuation curve is within ±10%. The monitored energy consumption change rate is 8%, which exceeds the preset stable interval, and the communication link reconstruction protocol is triggered.

[0081] Step S530: According to the path optimization rule, the data transmission link between the fault node and the neighbor node is reestablished, and it is verified whether the packet loss rate is reduced to the preset communication quality threshold.

[0082] The path preference rule is a previously configured path preference rule based on a load state of a neighbor node. The data transmission link is a link between the failed node and the neighbor node for transmitting data. The preset communication quality threshold is a preset maximum allowed value of a packet loss rate.

[0083] In implementation, according to the path preference rule, an optimal neighbor node is selected as a target node for data transmission. A data transmission link between the failed node and the neighbor node is re-established using a network protocol. After the link is established, the packet loss rate is monitored and compared with the preset communication quality threshold. For example, in an intelligent industrial sensor network, the path preference rule is to select a neighbor node with the lightest load as a communication path. The failed node re-establishes a data transmission link with the neighbor node with the lightest load through a network protocol. After the link is established, the packet loss rate is monitored to be 8%, and the preset communication quality threshold is 5%. The packet loss rate does not decrease to the preset communication quality threshold.

[0084] Step S540: If the packet loss rate is still higher than the communication quality threshold, an interference shielding threshold in the environment adaptive calibration scheme is activated, and a signal transmission power and a receiving sensitivity parameter are adjusted synchronously.

[0085] The interference shielding threshold is a previously configured threshold for shielding environmental interference. The signal transmission power refers to a power of a signal transmitted by the failed node. The receiving sensitivity parameter refers to a sensitivity of a signal received by the failed node.

[0086] In implementation, when the packet loss rate is still higher than the communication quality threshold, the interference shielding threshold in the environment adaptive calibration scheme is activated. According to a requirement of the interference shielding threshold, the signal transmission power and the receiving sensitivity parameter are adjusted synchronously. For example, in an intelligent agricultural sensor network, the packet loss rate is 8%, which is higher than the preset communication quality threshold of 5%. The interference shielding threshold in the environment adaptive calibration scheme is activated to be 10 dB. According to the requirement of the interference shielding threshold, the signal transmission power is increased by 2 dB, and the receiving sensitivity parameter is adjusted to be able to receive a signal with an intensity of 10 dB or less.

[0087] Step S550: The optimized node running parameters and a network topology structure are encapsulated into a configuration update package, and the configuration update package is broadcast to all associated nodes through a distributed synchronization mechanism of the target sensor network.

[0088] The optimized node running parameters are device running parameters, network communication parameters, and environmental interference parameters of the failed node after node optimization operation. The network topology structure is a connection relationship between nodes in the target sensor network. The configuration update package is a data package in which the optimized node running parameters and the network topology structure are encapsulated. The distributed synchronization mechanism is a mechanism for synchronizing node configurations in the target sensor network.

[0089] In a specific implementation, the optimized node operation parameters and network topology structure are sorted and encapsulated to generate a configuration update package. The configuration update package is broadcast to all associated nodes using the distributed synchronization mechanism of the target sensor network. For example, in an intelligent medical sensor network, the optimized node operation parameters and network topology structure are encapsulated into a configuration update package in JSON format. The configuration update package is broadcast to all associated nodes related to the faulty node through the distributed synchronization mechanism of the target sensor network, so that they can update their own configuration information.

[0090] As an implementation, the method provided by the embodiment of the present application further includes the following steps S560-S5100:

[0091] Step S560: After the broadcast of the configuration update package is completed, a cross-node collaborative verification process is started: the neighbor node sends a verification data package to the faulty node and records the end-to-end transmission delay.

[0092] The cross-node collaborative verification process is a process for verifying whether the node optimization operation is successful. The verification data package is a data package for verifying the communication performance of the node. The end-to-end transmission delay refers to the time required for the verification data package to be sent from the neighbor node to the faulty node.

[0093] In a specific implementation, after the broadcast of the configuration update package is completed, the neighbor node sends a verification data package to the faulty node according to a preset rule. The sending time and receiving time of the verification data package are recorded using a timestamp, and the end-to-end transmission delay is calculated. For example, in an intelligent education sensor network, after the broadcast of the configuration update package is completed, the neighbor node A sends a verification data package to the faulty node B. The neighbor node A records the sending time as t1, the faulty node B records the receiving time as t2, and the end-to-end transmission delay is t2-t1.

[0094] Step S570: If the transmission delay exceeds a preset delay tolerance, it is determined that the current optimization operation has not completely repaired the communication link breakage feature in the network-related fault set, and the rollback mechanism is triggered to recover to the last stable configuration.

[0095] The preset delay tolerance is a preset maximum allowed value of the end-to-end transmission delay. The rollback mechanism is a mechanism that restores the node to the state before execution when the processing strategy execution fails.

[0096] In a specific implementation, the calculated end-to-end transmission delay is compared with the preset delay tolerance. If the transmission delay exceeds the preset delay tolerance, it is determined that the current optimization operation does not completely repair the communication link breakage feature, and a rollback mechanism is triggered to restore the node to the last stable configuration. For example, in an intelligent tourism sensor network, the preset delay tolerance is 100 ms, and the calculated end-to-end transmission delay is 120 ms, which exceeds the preset delay tolerance. It is determined that the current optimization operation does not completely repair the communication link breakage feature, and a rollback mechanism is triggered to restore the hardware configuration and communication parameters of the node to the last stable configuration.

[0097] Step S580: If the transmission delay meets the requirements, real-time running data of the faulty node is collected to generate a verification feature set, and the verification feature set is input into the fault diagnosis strategy model for secondary fault matching.

[0098] The verification feature set is a feature set generated according to the real-time running data of the faulty node for verifying whether the fault is repaired. The secondary fault matching is to input the verification feature set into the fault diagnosis strategy model again for fault matching.

[0099] In a specific implementation, when the transmission delay meets the requirements, the real-time running data of the faulty node is collected using sensors, including device running parameters, network communication parameters, and environmental interference parameters. Feature extraction processing is performed on these data to generate a verification feature set. The verification feature set is input into the fault diagnosis strategy model for secondary fault matching. For example, in an intelligent financial sensor network, the transmission delay is 80 ms, which meets the preset delay tolerance. The real-time running data of the faulty node is collected, and feature extraction processing is performed to generate a verification feature set. The verification feature set is input into the fault diagnosis strategy model for secondary fault matching to check whether there is still another fault type.

[0100] Step S590: When there is no fault type in the candidate fault type set in the secondary matching result, it is confirmed that the node optimization operation is successful, and the current configuration parameters are persistently stored in the fault handling knowledge base.

[0101] The secondary matching result refers to the result obtained after the verification feature set is input into the fault diagnosis strategy model for secondary fault matching. The persistent storage is to save the current configuration parameters in the fault handling knowledge base for subsequent use.

[0102] In a specific implementation, when none of the candidate fault types in the secondary matching result exists, it indicates that the node optimization operation is successful. The configuration parameters of the current node, including device operation parameters, network communication parameters, and environmental interference parameters, are saved to the fault handling knowledge base. For example, in an intelligent entertainment sensor network, the secondary matching result shows that none of the candidate fault types exists. It is confirmed that the node optimization operation is successful, and the configuration parameters of the current node are saved to the fault handling knowledge base, so that they can be referred to in subsequent similar faults.

[0103] Step S5100: If a new fault type is detected, incremental strategy generation is performed based on the similarity of the new fault type and the historical handling strategies, and the incremental strategy is added to the node fault handling strategy set.

[0104] The new fault type is a fault type that has not occurred before and is detected in the secondary fault matching process. The incremental strategy generation is to generate a new handling strategy according to the similarity of the new fault type and the historical handling strategies. The incremental strategy is the newly generated handling strategy.

[0105] In a specific implementation, when a new fault type is detected, a similarity calculation algorithm is used to calculate the similarity of the new fault type and the fault types in the historical handling strategies. The closest historical handling strategy is selected according to the similarity, and is modified and optimized to generate an incremental strategy. The incremental strategy is added to the node fault handling strategy set. For example, in an intelligent energy sensor network, the secondary fault matching detects a new fault type of sensor precision decline. A cosine similarity algorithm is used to calculate the similarity of the fault type and the fault types in the historical handling strategies, and it is found that the similarity with the sensor drift fault type is the highest. The handling strategy of the sensor drift fault is modified and optimized, and a step of sensor calibration is added to generate an incremental strategy. The incremental strategy is added to the node fault handling strategy set.

[0106] As an implementation, the method provided by the embodiment of the present application further includes the following steps S5110-S5150:

[0107] Step S5110: After the configuration update package broadcast is completed, the cooperative verification data set of the fault node and its neighbor nodes is obtained, and the cooperative verification data set contains real-time monitoring results of the optimized device operation parameters, network communication parameters, and environmental interference parameters.

[0108] The cooperative verification data set is a data set for verifying the effect of the node optimization operation, and contains real-time monitoring results of the fault node and its neighbor nodes. The optimized device operation parameters, network communication parameters, and environmental interference parameters are corresponding parameters of the node after the node optimization operation.

[0109] In a specific implementation, after the configuration update package broadcast is completed, real-time operation data of the faulty node and its neighbor nodes are obtained using sensors and monitoring devices, including device operation parameters, network communication parameters, and environmental interference parameters, etc. These data are sorted and summarized to generate a collaborative verification dataset. For example, in an intelligent traffic sensor network, after the configuration update package broadcast is completed, device operation parameters, network communication parameters, and environmental interference parameters of the faulty node and its neighbor nodes are obtained using current sensors, network monitoring software, and environmental sensors. These data are sorted into a table form collaborative verification dataset.

[0110] Step S5120: fault elimination verification processing is performed on the collaborative verification dataset: the real-time monitoring results are input into the fault diagnosis strategy model to regenerate a verification candidate fault type set.

[0111] The fault elimination verification processing is an analysis of the collaborative verification dataset to verify whether the node optimization operation has eliminated the fault. The verification candidate fault type set is a set of possible fault types regenerated according to the real-time monitoring results.

[0112] In a specific implementation, the real-time monitoring results in the collaborative verification dataset are input into the fault diagnosis strategy model, and a verification candidate fault type set is regenerated according to the previous fault matching process. For example, in an intelligent industrial sensor network, the real-time monitoring results of the collaborative verification dataset are input into the fault diagnosis strategy model, and a verification candidate fault type set is regenerated after processing by the hardware degradation analysis module, the network topology analysis module, and the environmental resistance analysis module.

[0113] Step S5130: when there is an intersection between the verification candidate fault type set and the candidate fault type set, it is determined that the node optimization operation has not completely eliminated the target fault type, triggering an incremental strategy optimization process.

[0114] The incremental strategy optimization process is a process of further optimizing the processing strategy when the node optimization operation has not completely eliminated the target fault type.

[0115] In a specific implementation, a set intersection algorithm is used to calculate the intersection of the verification candidate fault type set and the candidate fault type set. If the intersection is not empty, it is determined that the node optimization operation has not completely eliminated the target fault type, triggering the incremental strategy optimization process. For example, in an intelligent agricultural sensor network, the verification candidate fault type set is {power module abnormality, communication link breakage}, and the candidate fault type set is {power module abnormality, sensor drift}. The intersection is {power module abnormality}, which is not empty, so it is determined that the node optimization operation has not completely eliminated the target fault type, triggering the incremental strategy optimization process.

[0116] Step S5140: Extracting an optimized compensation strategy of a historical failure handling case from the failure handling knowledge base based on the residual fault types corresponding to the intersection, and generating a compensatory node optimization instruction combined with the state feature set.

[0117] The residual fault types refer to the fault types in the intersection of the verification candidate fault type set and the candidate fault type set, which represent the faults that are not completely eliminated after the node optimization operation. The optimized compensation strategy of the historical failure handling case is extracted from the past cases stored in the failure handling knowledge base, which are the cases of handling the same or similar faults but not successful, and is used to make up for the improvement strategy of the previous handling strategy. The state feature set contains device running state features, network connection stability features, and environmental interference correlation features, etc., which can reflect the current overall state of the node.

[0118] In specific implementation, first, the residual fault types corresponding to the intersection are analyzed to determine their fault features and possible causes. Then, the historical failure handling cases related to the residual fault types are searched in the failure handling knowledge base by keyword matching and similarity search methods. For the found cases, the optimized compensation strategies are extracted, which may include adjusting hardware parameters, modifying communication protocols, enhancing environmental adaptability measures, etc. Then, combined with the state feature set of the current node, the device running condition, network connection state, and environmental interference factors of the node are comprehensively considered to further adjust and optimize the extracted optimized compensation strategies, and to generate compensatory node optimization instructions. For example, in an intelligent logistics sensor network, the residual fault type is "power module abnormality", and the historical failure handling case is found from the failure handling knowledge base, which is that the fault is not solved due to the unreasonable setting of the voltage regulation range of the power module. The extracted optimized compensation strategy is to expand the voltage regulation range and increase the voltage fluctuation monitoring and automatic adjustment mechanism. Combined with the state feature set of the current node, it is found that the device load is high and the environmental temperature is unstable, and the strategy is further optimized, that is, based on the expansion of the voltage regulation range, the voltage regulation parameters are dynamically adjusted according to the device load and environmental temperature, and the compensatory node optimization instruction is generated.

[0119] Step S5150: After executing the compensatory node optimization instruction, the verification data set is collected again, and until there is no intersection between the verification candidate fault type set and the candidate fault type set, the final effective optimization strategy parameters and corresponding monitoring data are added to the incremental learning module of the failure handling knowledge base.

[0120] The execution of the compensatory node optimization instruction is a further optimization operation on the fault node according to the operation steps and parameters specified in the instruction. The re-collection of the verification data set is the reacquisition of the collaborative verification data set of the fault node and its neighbor nodes after the execution of the optimization instruction, including the real-time monitoring results of the optimized device operation parameters, network communication parameters and environmental interference parameters. The incremental learning module is a part of the fault handling knowledge base, which is used to store new optimization strategy parameters and monitoring data for subsequent continuous learning and improvement of the fault handling strategy.

[0121] In specific implementation, the fault node is operated according to the compensatory node optimization instruction, for example, adjusting the parameters of the hardware device, modifying the network configuration, starting the environmental adaptation measures, etc. After the operation is completed, the real-time operation data of the fault node and its neighbor nodes are collected again using sensors and monitoring devices to generate a new verification data set. The new verification data set is input into the fault diagnosis strategy model to regenerate the verification candidate fault type set. Then, it is checked whether there is an intersection between the verification candidate fault type set and the candidate fault type set. If there is an intersection, it means that there is still a residual fault that has not been solved, and the incremental strategy optimization process needs to be executed again to generate a new compensatory node optimization instruction and execute it. If there is no intersection, it means that the node optimization operation successfully eliminates the target fault type. At this time, the final effective optimization strategy parameters, such as hardware configuration parameters, network protocol settings, environmental adaptation thresholds, etc., and the corresponding monitoring data, such as device operation state data, network communication performance data, environmental interference parameters, etc., are added to the incremental learning module of the fault handling knowledge base. For example, in an intelligent medical sensor network, after the execution of the compensatory node optimization instruction, the verification data set is collected again, and the verification candidate fault type set is generated through the fault diagnosis strategy model. After multiple iterations of optimization, the verification candidate fault type set has no intersection with the candidate fault type set. The final adjusted hardware parameters, network communication protocol settings and the collected device energy consumption, signal strength, environmental temperature and other monitoring data are added to the incremental learning module of the fault handling knowledge base.

[0122] As an implementation manner, the method provided by the embodiment of the present application further includes the following steps S5131-S5135:

[0123] Step S5131: In the incremental strategy optimization process, the change trend of the environmental interference parameter is monitored in real time, and when it is detected that the intensity of the interference source exceeds the coverage range of the current interference shielding threshold, the environmental resistance reinforcement mode is triggered.

[0124] The change trend of environmental interference parameters refers to the changes of environmental interference parameters (such as temperature gradient, electromagnetic noise, humidity, etc.) over time. The interference source intensity refers to the intensity of interference generated by the interference source (such as electromagnetic radiation source, high temperature area, etc.). The current interference shielding threshold refers to the threshold set by the node to shield environmental interference. The environmental resistance strengthening mode is a special working mode in which the node takes more powerful measures to resist environmental interference.

[0125] In the incremental strategy optimization process, environmental sensors are used to monitor environmental interference parameters in real time, such as temperature sensors to monitor temperature gradients, electromagnetic sensors to monitor electromagnetic noise intensity, and humidity sensors to monitor humidity changes. The monitored data is analyzed to calculate the change trend of the interference parameters, for example, by calculating the change rate over a period of time to determine whether the interference intensity is increasing or decreasing. The current interference source intensity is compared with the current interference shielding threshold. If the interference source intensity exceeds the coverage of the current interference shielding threshold, it means that the current interference shielding measure is insufficient to cope with environmental interference, and the environmental resistance strengthening mode is triggered. For example, in a smart industrial workshop sensor network, the electromagnetic noise intensity is monitored in real time by electromagnetic sensors in the incremental strategy optimization process. When the electromagnetic noise intensity is detected to suddenly increase and exceed the current set interference shielding threshold, the environmental resistance strengthening mode is triggered.

[0126] Step S5132: In the environmental resistance strengthening mode, the signal modulation scheme of the faulty node is redistributed according to the type of interference source: frequency hopping modulation strategy is adopted for electromagnetic noise interference, and thermal stability compensation algorithm is enabled for temperature gradient interference.

[0127] The signal modulation scheme refers to the way and method of modulating the signal. Different signal modulation schemes can improve the transmission performance of the signal in different environments. The frequency hopping modulation strategy is a signal modulation strategy that quickly switches the transmission frequency of the signal between different frequency bands to avoid electromagnetic noise interference in the target frequency band. The thermal stability compensation algorithm is an algorithm for compensating the influence of temperature gradient on signal transmission and device performance. It can dynamically adjust the parameters of the device according to temperature changes to maintain the stability of the device.

[0128] In a specific implementation, first, the type of the interference source is determined through analysis of the environmental interference parameters. It could be electromagnetic noise interference, temperature gradient interference, or other types of interference. If it is electromagnetic noise interference, the frequency band hopping modulation strategy is adopted. The implementation of this strategy can be achieved by pre-setting a set of available frequency bands. When it is detected that a certain frequency band is interfered by electromagnetic noise, it will automatically switch to another frequency band for signal transmission. For example, in a wireless sensor network, 10 available frequency bands are pre-set. When it is detected that frequency band 3 is severely interfered by electromagnetic noise, the signal transmission is switched to frequency band 7. If it is temperature gradient interference, the thermal stability compensation algorithm is enabled. This algorithm can dynamically adjust the working parameters of the device according to the temperature changes monitored by the temperature sensor in real time, such as adjusting the signal transmission power, adjusting the working frequency of the hardware components, etc., to compensate for the influence of temperature gradient on the performance of the device. For example, when the temperature rises, the signal transmission power is appropriately reduced to reduce the signal distortion caused by the temperature rise.

[0129] Step S5133: Reconstruct the physical layer communication protocol between the faulty node and the neighbor nodes based on the updated signal modulation scheme, and synchronously adjust the data frame length and the check code redundancy ratio.

[0130] The physical layer communication protocol refers to the rules and standards followed in communication at the physical layer, which specifies the transmission mode, coding mode, modulation mode, etc. of the signal. The data frame length refers to the length of each data frame in data transmission, and the check code redundancy ratio refers to the proportion of check code in the data frame. The check code is used to detect and correct errors in data transmission.

[0131] In a specific implementation, the physical layer communication protocol between the faulty node and the neighbor nodes is reconstructed according to the updated signal modulation scheme. For example, if the frequency band hopping modulation strategy is adopted, the frequency band selection and switching rules in the physical layer communication protocol need to be modified; if the thermal stability compensation algorithm is enabled, the signal parameters and device working mode in the protocol need to be adjusted. At the same time, the data frame length and the check code redundancy ratio are synchronously adjusted. When the environmental interference is enhanced, appropriately increasing the data frame length can improve the data transmission efficiency, but it may increase the error rate, so the check code redundancy ratio needs to be increased accordingly to improve the reliability of data transmission. The specific adjustment can be determined according to experiments and experience. For example, in a sensor network of a smart building, the physical layer communication protocol is reconstructed according to the updated signal modulation scheme, and the original fixed frequency band selection rule is changed to a dynamic frequency band hopping rule. At the same time, the data frame length is increased from the original 100 bytes to 120 bytes, and the check code redundancy ratio is increased from the original 10% to 15%.

[0132] Step S5134: Collect the reconstructed network communication parameters to generate an anti-interference performance evaluation index. When the evaluation index does not reach a preset resistance threshold, iteratively update the parameter combination of the frequency hopping modulation strategy and the thermal stability compensation algorithm.

[0133] The anti-interference performance evaluation index is an index for evaluating the performance of the node in terms of anti-interference. It can be obtained by collecting the reconstructed network communication parameters such as packet loss rate, signal strength, bit error rate, etc. and performing comprehensive calculation. The preset resistance threshold is the minimum requirement for anti-interference performance. The parameter combination of the frequency hopping modulation strategy and the thermal stability compensation algorithm refers to the combination of parameters such as frequency switching rule and switching frequency in the frequency hopping modulation strategy, and parameters such as temperature compensation coefficient and parameter adjustment range in the thermal stability compensation algorithm.

[0134] In specific implementation, the network monitoring device is used to collect the reconstructed network communication parameters, such as calculating the packet loss rate by counting the number of lost packets within a set time, measuring the signal strength using a signal strength sensor, and calculating the bit error rate by analyzing the error conditions in data transmission, etc. According to the collected network communication parameters, the anti-interference performance evaluation index is calculated, for example, the weighted average method can be used to weight and sum the packet loss rate, signal strength, bit error rate, etc. according to the preset weight to obtain a comprehensive anti-interference performance evaluation index. The calculated evaluation index is compared with the preset resistance threshold. If the evaluation index does not reach the preset resistance threshold, it means that the current parameter combination of the frequency hopping modulation strategy and the thermal stability compensation algorithm cannot meet the anti-interference requirements, and the parameter combination needs to be iteratively updated. Optimization algorithms such as genetic algorithm, particle swarm algorithm, etc. can be used to search and optimize the parameter combination to find a better parameter combination. For example, in an intelligent traffic sensor network, the reconstructed network communication parameters are collected, and the anti-interference performance evaluation index is calculated to be 0.7, and the preset resistance threshold is 0.8. The genetic algorithm is used to iteratively optimize the frequency switching frequency of the frequency hopping modulation strategy and the temperature compensation coefficient of the thermal stability compensation algorithm. After multiple iterations, a better parameter combination is found, which improves the anti-interference performance evaluation index to 0.85.

[0135] Step S5135: Encapsulate the optimized parameter combination and the physical layer communication protocol into an environmental resistance reinforcement package, and distribute it to the node group of the same type of interference area through the distributed synchronization mechanism.

[0136] The environmental resistance enhancement package is a data package that encapsulates the optimized frequency hopping modulation strategy, the parameter combination of the thermal stability compensation algorithm, and the reconstructed physical layer communication protocol. The node group in the same type of interference area refers to a group of nodes that are subjected to the same type of environmental interference, such as electromagnetic noise interference, temperature gradient interference, etc. The distributed synchronization mechanism is a mechanism for synchronizing node configuration and data in the network.

[0137] In a specific implementation, the optimized frequency hopping modulation strategy, the parameter combination of the thermal stability compensation algorithm, and the reconstructed physical layer communication protocol are sorted and encapsulated to generate an environmental resistance enhancement package. The package can be stored in a set file format, such as JSON, XML, etc. Then, through the distributed synchronization mechanism, the environmental resistance enhancement package is distributed to the node group in the same type of interference area. In the distribution process, broadcast, multicast, etc. can be used to ensure that each node can receive the environmental resistance enhancement package. For example, in a large factory sensor network, the optimized parameter combination and physical layer communication protocol are encapsulated into a JSON format environmental resistance enhancement package. Through the distributed synchronization mechanism, the package is distributed to the node group subjected to electromagnetic noise interference, so that these nodes can update their signal modulation scheme and physical layer communication protocol, improving the anti-interference ability.

[0138] As an implementation, the method provided by the embodiment of the present application can further include the following steps S5136-S51310:

[0139] Step S5136: deploying a cross-node cooperative maintenance mechanism in the node group in the same type of interference area, periodically collecting the state feature set of each node in the group to generate a group health degree atlas.

[0140] The cross-node cooperative maintenance mechanism is a mechanism for implementing cooperative maintenance and management in a node group, which enables nodes to cooperate with each other to jointly cope with faults and environmental interference. The group health degree atlas is a graph for visually displaying the overall health status of a node group, which is obtained by collecting the state feature set of each node in the group, including device running state features, network connection stability features, and environmental interference correlation features, and performing comprehensive analysis and visualization processing.

[0141] In a specific implementation, a cross-node collaborative maintenance mechanism is established in a node group in the same type of interference area. The collaboration between nodes can be achieved by setting a dedicated collaborative management node or using a distributed algorithm. The state feature set of each node in the group is periodically collected, and the collection period can be set according to actual conditions, such as every hour, every day, etc. The collected state feature set is comprehensively analyzed, for example, the average device operating parameters, network communication performance indicators, and environmental interference intensity of the nodes in the group are calculated. The analysis results are visualized in the form of a graph and a group health graph is generated. For example, in a smart city sensor network, a cross-node collaborative maintenance mechanism is deployed in a node group affected by temperature gradient interference. The state feature set of each node in the group is collected every 6 hours, and the average energy consumption, average packet loss rate, and average temperature gradient of the nodes are calculated. These indicators are displayed in the form of a heat map to generate a group health graph, and the health status of different nodes is represented by the depth of color.

[0142] Step S5137: Topological vulnerability analysis is performed on the group health graph to identify a set of key nodes that have communication link overload or hardware load imbalance.

[0143] Topological vulnerability analysis is an analysis of the topology of a node group to find parts that are prone to failure or performance bottlenecks. Communication link overload refers to the load on a communication link exceeding its carrying capacity, resulting in data transmission delays, packet loss, and other problems. Hardware load imbalance refers to uneven distribution of hardware component loads among nodes, with some nodes having excessively high hardware loads and others having excessively low hardware loads. The set of key nodes refers to a set of nodes that have a significant impact on network performance and stability in the node group.

[0144] In a specific implementation, graph theory and network analysis methods are used to perform topological vulnerability analysis on the group health graph. By analyzing the connection relationships and communication traffic between nodes, the communication load and hardware load of each node are calculated. According to a pre-set threshold, it is determined whether there is a communication link overload or hardware load imbalance. If the communication load of a node exceeds the carrying capacity of the communication link, or the hardware load of a node is excessively high, the node is identified as a key node. For example, in a smart energy grid sensor network, topological vulnerability analysis is performed on the group health graph. By calculating the communication traffic and hardware component load between nodes, it is found that node A's communication traffic exceeds the carrying capacity of its communication link, and node B's hardware CPU load is excessively high. Nodes A and B are identified as key nodes and form a set of key nodes.

[0145] Step S5138: According to the spatial distribution characteristics of the key node set, generate load migration strategy and link reconstruction strategy: migrate the data forwarding task of the overloaded node to the idle node, and increase the redundant communication link on the critical path.

[0146] The spatial distribution characteristics of the key node set refer to the geographical location distribution of the key nodes in the node group. The load migration strategy is a strategy for balancing node load, which migrates the data forwarding task of the overloaded node to the idle node to reduce the burden of the overloaded node. The link reconstruction strategy is a strategy for optimizing network topology, which increases the redundant communication link on the critical path to improve the reliability and fault tolerance of the network.

[0147] In specific implementation, first analyze the spatial distribution characteristics of the key node set to understand the geographical location of the key nodes and their connection relationship with other nodes. For the key nodes with communication link overload or hardware load imbalance, generate the load migration strategy. The idle node can be determined through network monitoring and load prediction algorithm, and the data forwarding task of the overloaded node can be migrated to the idle node. For example, by monitoring the load of the nodes in real time, it is found that the load of node A is too high, while the load of node C is relatively low, part of the data forwarding task of node A is migrated to node C. At the same time, according to the location of the key nodes and the network topology, generate the link reconstruction strategy. Increase the redundant communication link on the critical path to improve the reliability of the network. For example, increase a standby communication link between node A and node B on the critical path, when the main link fails, automatically switch to the standby link.

[0148] Step S5139: During the execution of the load migration strategy, monitor the hardware load fluctuation curve of the migration target node in real time, and when the load mutation is detected to be out of the hardware stability interval, automatically rollback to the last stable topology configuration.

[0149] The execution process of the load migration strategy is the process of migrating the data forwarding task of the overloaded node to the idle node. The migration target node refers to the idle node that receives the data forwarding task. The hardware load fluctuation curve refers to the change curve of the hardware load of the migration target node over time. The hardware stability interval refers to the load range within which the hardware of the migration target node can operate stably.

[0150] In a specific implementation, in the process of executing the load migration strategy, the hardware load fluctuation curve of the migration target node is monitored in real time using a hardware load sensor. The monitored curve is analyzed to calculate the change rate and fluctuation amplitude of the load. When a load mutation exceeding the hardware stability interval is detected, it indicates that the hardware of the migration target node cannot withstand the sudden increase in load, which may cause hardware failure or performance degradation. At this time, the last stable topology configuration is automatically rolled back, and the data forwarding task is redistributed to the original node. For example, in a sensor network of an intelligent logistics warehouse, the load migration strategy is executed to migrate the data forwarding task of node A to node C. In the migration process, the hardware CPU load fluctuation curve of node C is monitored in real time. When it is detected that the CPU load suddenly increases from 20% to 80%, exceeding the hardware stability interval (30%-70%), the last stable topology configuration is automatically rolled back, and the data forwarding task is redistributed to node A.

[0151] Step S51310: integrating the successfully implemented load migration strategy and the link reconstruction strategy into a group maintenance knowledge base, and establishing a strategy mapping channel between the group maintenance knowledge base and the fault handling knowledge base.

[0152] The group maintenance knowledge base is a knowledge base for storing and managing node group maintenance strategies, which includes successfully implemented load migration strategies and link reconstruction strategies. The strategy mapping channel is a channel for establishing a connection between the group maintenance knowledge base and the fault handling knowledge base, through which information sharing and strategy coordination between the two knowledge bases can be achieved.

[0153] In a specific implementation, the successfully implemented load migration strategy and the link reconstruction strategy are sorted and summarized and stored in the group maintenance knowledge base. The group maintenance knowledge base can be stored in the form of a database, and each strategy can include the name of the strategy, the applicable scenario, the specific operation steps, the effect evaluation, etc. At the same time, a strategy mapping channel is established between the group maintenance knowledge base and the fault handling knowledge base. Information sharing between the two knowledge bases can be achieved by establishing an association table or an interface. For example, when a fault related to node load or network topology occurs in the fault handling knowledge base, the corresponding load migration strategy or link reconstruction strategy can be obtained from the group maintenance knowledge base through the strategy mapping channel for fault handling. In a sensor network of an intelligent medical park, the successfully implemented load migration strategy and the link reconstruction strategy are stored in the group maintenance knowledge base, and a strategy mapping channel is established between the group maintenance knowledge base and the fault handling knowledge base. When a node load is too high, the load migration strategy can be obtained from the group maintenance knowledge base for processing.

[0154] As an implementation manner, the method provided by the embodiment of the present application can further include the following steps S51311-S51315:

[0155] Step S51311: When the group health atlas continuously displays that the target node is in a high load state, start a preventive node maintenance process: predict the hardware remaining life based on the state feature set, and generate a maintenance priority list.

[0156] The target node refers to the node that is continuously in a high load state in the group health atlas. The preventive node maintenance process is a process of maintaining the node in advance to avoid node failure. The hardware remaining life refers to the time that the hardware device of the node can still work normally under the current usage state. The maintenance priority list is generated according to the hardware remaining life and importance of the node, etc., and is used to determine the order of node maintenance.

[0157] In specific implementation, when the group health atlas continuously displays that the target node is in a high load state, it indicates that the node may have the risk of hardware aging or performance decline, and the preventive node maintenance process is started. Based on the state feature set of the target node, a machine learning algorithm or a physical model is used to predict the hardware remaining life. For example, a support vector machine (SVM) algorithm can be used, and the device running parameters (such as energy consumption, temperature, hardware load, etc.) in the state feature set are taken as input to train a prediction model to predict the hardware remaining life. According to the prediction result and the importance of the node in the network, etc., a maintenance priority list is generated. The importance can be evaluated according to the function of the node, the number of connected nodes, etc. For example, in a sensor network of a smart traffic hub, node A is continuously in a high load state in the group health atlas. Using the SVM algorithm, the hardware remaining life of node A is predicted to be 6 months according to the state feature set of node A. Considering that node A is connected to multiple important monitoring devices, it is placed in the front row in the maintenance priority list.

[0158] Step S51312: According to the maintenance priority list, inject a hardware self-check instruction to the target node to obtain memory bad block distribution data and power module output stability curve.

[0159] The hardware self-check instruction is an instruction for the hardware device of the node to perform self-checking. The memory bad block distribution data refers to the distribution of damaged storage blocks in the memory of the node. The power module output stability curve refers to the curve of the output voltage or current of the power module with respect to time, which reflects the output stability of the power module.

[0160] In a specific implementation, according to the maintenance priority list, hardware self-check instructions are sequentially injected into the target node. The hardware self-check instructions can be sent to the target node through a network communication protocol, and the target node starts the hardware self-check program after receiving the instructions. During the self-check process, memory bad block distribution data and power module output stability curves are obtained. For memory bad block distribution data, a memory self-check algorithm can be used to test the read and write of each memory block, mark the damaged memory block, and record its location and quantity. For the power module output stability curve, a voltage sensor and a current sensor can be used to monitor the output voltage and current of the power module in real time, and the monitoring data is recorded in chronological order to form the output stability curve. For example, in a smart industrial automation sensor network, according to the maintenance priority list, hardware self-check instructions are injected into the target node B. The target node B starts the hardware self-check program and detects 3 bad blocks in the memory through the memory self-check algorithm, which are located at addresses 0x100, 0x200 and 0x300. At the same time, the output voltage and current of the power module are monitored using a voltage sensor and a current sensor, and the output stability curve of the power module is generated.

[0161] Step S51313: Perform address remapping operation on the memory bad block distribution data, and apply feedback voltage regulation to the power module output stability curve.

[0162] The address remapping operation refers to remapping the addresses of bad blocks in the memory to other available memory blocks to avoid read and write operations on the bad blocks and improve the reliability of the memory. The feedback voltage regulation refers to adjusting the output voltage of the power module in real time according to the power module output stability curve to keep it stable.

[0163] In a specific implementation, for the memory bad block distribution data, the address of the bad block is remapped using an address remapping algorithm. First, scan the available storage blocks in the memory to find enough free space to replace the bad block. Then, modify the address mapping table of the memory to map the address of the bad block to the address of the available storage block. In subsequent read and write operations, the system will automatically access the remapped address. For the power module output stability curve, a feedback control algorithm such as a PID (Proportional-Integral-Derivative) controller is used to adjust the input parameters of the power module in real time according to the changes in the output stability curve to maintain the stability of the output voltage. For example, when the output voltage is lower than the set value, increase the input voltage of the power module; when the output voltage is higher than the set value, reduce the input voltage of the power module. In a smart home sensor network, the address remapping operation is performed on the memory bad block distribution data of the target node C, and the address of the bad block 0x100 is remapped to the address of the available storage block 0x500. At the same time, a PID controller is used to perform feedback voltage regulation on the power module output stability curve, and when the output voltage decreases, the input voltage of the power module is increased in time to keep the output voltage within a stable range.

[0164] Step S51314: Recalculate the hardware remaining life prediction value after completing the self-maintenance operation, and if the prediction value does not reach the preset safety threshold, trigger a node retirement warning and send a data backup instruction to the adjacent node.

[0165] The self-maintenance operation refers to the maintenance operations such as address remapping operation and feedback voltage regulation performed on the target node. The hardware remaining life prediction value is the hardware remaining life predicted again based on the state feature set of the node after completing the self-maintenance operation. The preset safety threshold is the lowest safety value of the hardware remaining life preset in advance. The node retirement warning is a reminder signal used to prompt that the hardware remaining life of the node is insufficient and needs to be retired. The data backup instruction is an instruction for the adjacent node to backup the data of the target node.

[0166] In a specific implementation, after completing the self-maintenance operation, the state feature set of the target node is collected again, and the hardware remaining life prediction value is recalculated using the previous prediction model. The prediction value is compared with the preset safety threshold, and if the prediction value does not reach the preset safety threshold, it means that the hardware remaining life of the node is insufficient to ensure its normal work, and a node retirement warning is triggered. At the same time, a data backup instruction is sent to the adjacent node to backup the data of the target node to prevent data loss. For example, in a smart educational campus sensor network, after completing the self-maintenance operation on the target node D, the hardware remaining life prediction value is recalculated to be 3 months, and the preset safety threshold is 6 months. Since the prediction value does not reach the preset safety threshold, a node retirement warning is triggered, and a data backup instruction is sent to the adjacent node, and the adjacent node starts to backup the data of the target node D.

[0167] Step S51315: During the node retirement warning effective period, gradually migrate the data forwarding task of the target node to the backup node until the communication link load is reduced to zero, and perform a safe power-off operation, and update the topology data of the group health map.

[0168] The node retirement warning effective period refers to the time period from triggering the node retirement warning to completing the node retirement processing. The backup node refers to a node that is set in advance to replace the target node. The communication link load refers to the data transmission amount of the target node on the communication link. The safe power-off operation refers to turning off the power of the node under the premise of ensuring the safety of the node data. The topology data of the group health map refers to the data reflecting the connection relationship between nodes and the network topology in the group health map.

[0169] In specific implementation, during the node retirement warning effective period, the data forwarding task of the target node is gradually migrated to the backup node according to a preset strategy. A phased migration manner can be adopted, that is, first migrating part of non-critical data forwarding tasks, and then migrating critical data forwarding tasks. During the migration process, the communication link load of the target node is monitored in real time, and when the communication link load is reduced to zero, it indicates that the target node no longer undertakes the data forwarding task, and a safe power-off operation can be performed. The safe power-off operation can be performed according to the shutdown process of the node to ensure that the data of the node is properly saved. At the same time, the topology data of the group health map is updated, the target node is removed from the map, and the connection relationship related to the target node is updated. For example, in a sensor network of an intelligent tourist attraction, during the node retirement warning effective period, the data forwarding task of the target node E is gradually migrated to the backup node F. After a period of migration, the communication link load of the target node E is reduced to zero, and a safe power-off operation is performed. At the same time, the topology data of the group health map is updated, the target node E is deleted from the map, and the connection relationship between the adjacent nodes and the backup node F is updated.

[0170] It can be understood that the various algorithms involved in the above introduction of the embodiments of the present application, such as genetic algorithm, cosine similarity algorithm, time synchronization algorithm, etc., can be known from related contents in the prior art. In order to save space, the above-mentioned algorithms will not be expanded in the embodiments of the present application. In addition, those skilled in the art can supplement the details according to the common knowledge in the art when implementing the scheme of the present application, for example, the dimension conflict before feature fusion can be eliminated by normalization, the dimension difference can be eliminated by interpolation, the threshold can be reasonably set based on historical data, experience or business scene requirements, the model can be trained based on a general model training method, etc. The present application will not introduce the redundant implementation process in too much detail.

[0171] Figure 2 A hardware entity schematic diagram of a computer system provided for an embodiment of the present application is shown in FIG. 10. The hardware entity of the computer system 1000 includes a processor 1001 and a memory 1002, wherein the memory 1002 stores a computer program executable on the processor 1001, and the processor 1001 implements the steps in the method of any of the above embodiments when executing the program. Figure 2

[0172] The memory 1002 stores a computer program executable on the processor, and the memory 1002 is configured to store instructions and applications executable by the processor 1001, and can also cache data (for example, image data, audio data, voice communication data and video communication data) to be processed or having been processed by the processor 1001 and each module in the computer system 1000, which can be realized by a FLASH or a Random Access Memory (RAM).

[0173] The processor 1001 implements the steps of the intelligent fault diagnosis method of the node of the intelligent space sensor network of any of the above embodiments when executing the program. The processor 1001 generally controls the overall operation of the computer system 1000.

[0174] The above is only an embodiment of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application.​

Claims

1. A method for intelligent diagnosis of node faults of an intelligent space sensor network, characterized in that, The method comprises the following steps: acquiring a real-time monitoring data set of all sensor nodes in a target sensor network, the real-time monitoring data set comprising device operation parameters, network communication parameters and environmental interference parameters of each sensor node; performing state feature extraction processing on the real-time monitoring data set to generate a state feature set of each sensor node, the state feature set comprising device operation state features, network connection stability features and environmental interference correlation features; based on a preset fault diagnosis strategy model, performing multi-dimensional fault matching processing on the state feature set to determine a candidate fault type set of the sensor node, the candidate fault type set comprising at least one candidate fault type that is correlated with the state feature set; according to a strategy mapping relationship between the candidate fault type set and a preset fault handling knowledge base, generating a node fault handling strategy set of the sensor node, the node fault handling strategy set comprising an optimal handling strategy for each candidate fault type; based on the node fault handling strategy set, performing node optimization operation on a fault node in the target sensor network, and feeding back the optimized node operation parameters to a control terminal of the target sensor network; wherein the state feature extraction processing on the real-time monitoring data set to generate a state feature set of each sensor node comprises: performing periodic fluctuation mode recognition processing on the device operation parameters to generate the device operation state features, the device operation state features comprising a time series-based energy consumption change trend and a hardware component load fluctuation curve; performing link abnormal event correlation analysis processing on the network communication parameters to extract the network connection stability features, the network connection stability features comprising an association matrix between packet loss rate and signal strength and a communication interruption frequency distribution graph; performing multi-source interference coupling analysis processing on the environmental interference parameters to construct the environmental interference correlation features, the environmental interference correlation features comprising a disturbance superposition effect of temperature gradient and electromagnetic noise and a quantitative influence coefficient of humidity change on signal attenuation; performing time domain alignment processing on the energy consumption change trend and the hardware component load fluctuation curve to generate a first fusion feature vector; performing spatial mapping processing on the association matrix and the communication interruption frequency distribution graph to generate a second fusion feature vector; and performing weight superposition processing on the disturbance superposition effect and the quantitative influence coefficient to generate a third fusion feature vector; based on the feature complementary relationship between the first fusion feature vector, the second fusion feature vector and the third fusion feature vector, generating the state feature set comprising multi-dimensional correlation indexes; the multi-dimensional fault matching processing on the state feature set based on the preset fault diagnosis strategy model to determine the candidate fault type set of the sensor node comprises: inputting the first fusion feature vector into a hardware degradation analysis module of the fault diagnosis strategy model to generate a hardware-related fault set by matching a preset hardware failure mode library, the hardware failure mode library comprising power module abnormality, sensor drift and circuit aging feature mode; inputting the second fusion feature vector into a network topology analysis module of the fault diagnosis strategy model, generating a network-related fault set by identifying communication link breakage features and routing table abnormality features, the communication link breakage features including neighbor node unreachable events and data retransmission number mutation patterns; inputting the third fusion feature vector into an environmental resistance analysis module of the fault diagnosis strategy model, generating an environment-related fault set by analyzing the deviation between interference superposition effects and node anti-interference threshold values, the node anti-interference threshold values being adjusted according to stable operation data under historical environmental parameters; performing fault coupling degree evaluation on the hardware-related fault set, the network-related fault set and the environment-related fault set, and calculating the conditional triggering probability between any two fault types; performing combination optimization processing on fault types with strong correlation relationships according to the conditional triggering probability, and generating the candidate fault type set including a main fault type and a derived fault type; generating the node fault handling strategy set of the sensing node according to the strategy mapping relationship between the candidate fault type set and a preset fault handling knowledge base, including: retrieving a benchmark handling strategy set corresponding to the main fault type from the fault handling knowledge base, the benchmark handling strategy set including a hardware reset instruction, a communication link reconstruction protocol and an environmental adaptive calibration scheme; performing strategy derivation processing on the benchmark handling strategy set according to the derived fault type: embedding a voltage compensation parameter for circuit aging characteristics in the hardware reset instruction; adding a path optimization rule based on neighbor node load state in the communication link reconstruction protocol; configuring an interference shielding threshold value in the environmental adaptive calibration scheme; performing execution priority evaluation on the derived handling strategy, the execution priority evaluation being based on the product index of strategy execution time consumption and fault repair probability for sorting; selecting a handling strategy meeting a preset repair success rate according to the sorting result to generate the node fault handling strategy set, and configuring an execution condition constraint and a rollback mechanism for each handling strategy; performing node optimization operation on a fault node in the target sensing network based on the node fault handling strategy set, including: selecting a target handling strategy with the highest priority according to the evaluation result of the execution priority, analyzing the hardware reset instruction in the target handling strategy and injecting the voltage compensation parameter; after performing a hardware reset operation on the fault node, monitoring the recovery state of the energy consumption change trend and the load fluctuation curve in real time, and triggering the communication link reconstruction protocol if a preset stable interval is not reached; reestablishing the data transmission link between the fault node and the neighbor node according to the path optimization rule, and verifying whether the packet loss rate has dropped to a preset communication quality threshold value; if the packet loss rate is still higher than the communication quality threshold value, activating the interference shielding threshold value in the environmental adaptive calibration scheme to synchronously adjust the signal transmission power and the receiving sensitivity parameter; The optimized node operating parameters and the network topology structure are encapsulated into a configuration update package, and the configuration update package is broadcast to all associated nodes through a distributed synchronization mechanism of the target sensor network.

2. The method of claim 1, wherein, The method further comprises: After the broadcast of the configuration update package is completed, a cross-node collaborative verification process is started: the neighbor nodes send verification data packets to the faulty node and record the end-to-end transmission delay; If the transmission delay exceeds a preset delay tolerance, it is determined that the current optimization operation has not completely repaired the communication link breakage feature in the set of network-related faults, and the rollback mechanism is triggered to restore to the last stable configuration; If the transmission delay meets the requirements, real-time operating data of the faulty node are collected to generate a set of verification features, and the set of verification features are input into the fault diagnosis strategy model for secondary fault matching; When there is no fault type in the set of candidate fault types in the secondary matching result, it is confirmed that the node optimization operation is successful, and the current configuration parameters are persistently stored in the fault handling knowledge base; If a new fault type is detected, an incremental strategy is generated based on the similarity of the new fault type and the historical handling strategy, and the incremental strategy is added to the set of node fault handling strategies.

3. The method of claim 2, wherein, The method further comprises: After the broadcast of the configuration update package is completed, a cross-node collaborative verification process is started: the neighbor nodes send verification data packets to the faulty node and record the end-to-end transmission delay; If the transmission delay exceeds a preset delay tolerance, it is determined that the current optimization operation has not completely repaired the communication link breakage feature in the set of network-related faults, and the rollback mechanism is triggered to restore to the last stable configuration; If the transmission delay meets the requirements, real-time operating data of the faulty node are collected to generate a set of verification features, and the set of verification features are input into the fault diagnosis strategy model for secondary fault matching; When there is no fault type in the set of candidate fault types in the secondary matching result, it is confirmed that the node optimization operation is successful, and the current configuration parameters are persistently stored in the fault handling knowledge base; If a new fault type is detected, an incremental strategy is generated based on the similarity of the new fault type and the historical handling strategy, and the incremental strategy is added to the set of node fault handling strategies.

4. The method of claim 3, wherein, The method further comprises: In the incremental strategy optimization process, the change trend of the environmental interference parameter is monitored in real time, and when it is detected that the intensity of the interference source exceeds the coverage range of the current interference shielding threshold, an environmental resistance reinforcement mode is triggered; In the environmental resistance reinforcement mode, the signal modulation scheme of the faulty node is re-allocated according to the type of the interference source: frequency band hopping modulation strategy is adopted for electromagnetic noise interference, and a thermal stability compensation algorithm is enabled for temperature gradient interference; Based on the updated signal modulation scheme, the physical layer communication protocol between the faulty node and the neighbor nodes is reconstructed, and the data frame length and the check code redundancy ratio are synchronously adjusted; The reconstructed network communication parameters are collected to generate an anti-interference performance evaluation index, and when the evaluation index does not reach a preset resistance threshold, the parameter combination of the frequency band hopping modulation strategy and the thermal stability compensation algorithm is iteratively updated; The optimized parameter combination is encapsulated into an environmental resistance reinforcement package together with the physical layer communication protocol, and is distributed to a node group in a same type of interference area through the distributed synchronization mechanism.

5. The method of claim 4, wherein, The method further includes: A cross-node cooperative maintenance mechanism is deployed in the node group in the same type of interference area, and a group health degree atlas is generated by periodically collecting state feature sets of each node in the group; Topology vulnerability analysis is performed on the group health degree atlas to identify a key node set in which communication links are overloaded or hardware loads are unbalanced; According to spatial distribution characteristics of the key node set, a load migration strategy and a link reconstruction strategy are generated: data forwarding tasks of overloaded nodes are migrated to idle nodes, and redundant communication links are added on critical paths; During execution of the load migration strategy, hardware load fluctuation curves of migration target nodes are monitored in real time, and when a load mutation that exceeds a hardware stability interval is detected, the last stable topology configuration is automatically rolled back to; The successfully implemented load migration strategy and link reconstruction strategy are integrated into a group maintenance knowledge base, and a strategy mapping channel between the group maintenance knowledge base and the fault processing knowledge base is established.

6. A computer system comprising a memory and a processor, said memory storing a computer program operable on the processor, characterised in that, The processor implements the steps in the method of any one of claims 1 to 5 when executing the program.

Images