Iot-based highway tunnel electromechanical facility state sensing method and system

Abstract

The application discloses a kind of highway tunnel electromechanical facility state perception method and system based on Internet of Things, it is related to electromechanical equipment state monitoring technical field, including obtaining the overall layout information of tunnel equipment, and the multi-source heterogeneous perception data of tunnel equipment are collected;The multi-source heterogeneous perception data of tunnel equipment is preprocessed, and the multivariate time series data of tunnel equipment is obtained.The application realizes the multi-source heterogeneous perception of the operation data of tunnel electromechanical facility in whole area by Internet of Things collection network, relies on long short-term memory network to capture the time sequence evolution law of equipment state, further combines graph convolution network to excavate the linkage abnormal characteristics between equipment, finally realizes the accurate positioning of abnormal point, compared with traditional monitoring mode, can effectively improve the response speed and positioning accuracy of abnormal perception, can provide reliable support for the intelligent operation and maintenance of highway tunnel electromechanical system. That is, effectively realize the demand of global perception to the operation state of entire tunnel electromechanical system.

Description

Technical Field

[0001] This invention relates to the field of electromechanical equipment condition monitoring technology, and in particular to a method and system for sensing the condition of electromechanical facilities in highway tunnels based on the Internet of Things. Background Technology

[0002] The electromechanical facilities within highway tunnels, including lighting, ventilation, fire protection, and monitoring subsystems, are core infrastructure ensuring tunnel traffic safety. Due to the enclosed, humid, and dusty environment of tunnels, these facilities operate under harsh conditions for extended periods, making them prone to aging and malfunctions. Failure to detect and address these issues promptly can easily lead to traffic accidents. Traditional monitoring of tunnel electromechanical facilities relies heavily on a combination of regular manual inspections and individual point monitoring. Manual inspections are inefficient, dangerous, and unable to cover all facility locations within the tunnel, nor can they provide real-time fault warnings. Furthermore, existing monitoring systems are often distributed, with data from different manufacturers not being shared, hindering a comprehensive understanding of the entire tunnel's electromechanical system's operational status. Data transmission delays are high, errors are significant, and fault identification accuracy is low, failing to meet the demands of intelligent highway tunnel operation. Simultaneously, existing sensing solutions lack the capacity to process massive amounts of monitoring data, making it impossible to accurately assess potential facility failure risks and posing hidden dangers to tunnel operational safety. Summary of the Invention

[0003] The purpose of this invention is to address the difficulty in meeting the need for global perception of the operating status of the entire tunnel electromechanical system in existing technologies, and to propose a method and system for sensing the status of highway tunnel electromechanical facilities based on the Internet of Things.

[0004] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a method for sensing the status of electromechanical facilities in highway tunnels based on the Internet of Things, comprising: Acquire overall deployment information of tunnel equipment and collect multi-source heterogeneous sensing data of tunnel equipment; Preprocessing of multi-source heterogeneous sensing data of tunnel equipment yields multivariate time series data of tunnel equipment; Unsupervised feature extraction is performed on multivariate time series data of tunnel equipment to obtain low-dimensional high-level feature information of tunnel equipment. Collect multi-source heterogeneous sensing data of tunnel equipment at different time intervals, and construct low-dimensional high-level feature information of tunnel equipment at different time intervals; Based on the low-dimensional high-level feature information of tunnel equipment under different time series, the state evolution law of tunnel equipment is simulated. Real-time acquisition of multi-source heterogeneous sensing data from tunnel equipment, combined with the state evolution patterns of tunnel equipment, to perform health scoring on tunnel equipment and obtain the real-time usage status of tunnel equipment; Based on the overall deployment information of tunnel equipment, a relational topology diagram of tunnel equipment is constructed, and the tunnel equipment is classified according to its health score to obtain the current status level of each tunnel equipment. Based on the relationship topology of the tunnel equipment and combined with the status classification results of the tunnel equipment, a status linkage analysis of the tunnel equipment is performed to obtain abnormal linkage information of the tunnel equipment. Based on the abnormal linkage information of tunnel equipment and combined with the overall deployment information of tunnel equipment, the abnormal state sensing points of the target tunnel are determined.

[0005] In one feasible approach, the state evolution law of the simulated tunnel equipment includes: Arrange the low-dimensional high-level feature information under different time series in chronological order to obtain a time series sample set of device status; Based on the overall deployment information of tunnel equipment and combined with the status monitoring requirements of tunnel equipment, an evolution feature extraction network for tunnel equipment is constructed. The time series sample set of equipment status is input into the evolution feature extraction network of tunnel equipment to capture the evolution law of equipment status at different time scales and obtain the equipment status evolution model. Based on the equipment state evolution model, and using the low-dimensional high-level feature information of the current and historical moments, the equipment state characteristics of future moments are predicted.

[0006] In one feasible approach, the evolution feature extraction network for the tunnel equipment employs a long short-term memory network. Specifically, the Long Short-Term Memory (LSTM) network models temporal dependencies through forgetting gates, input gates, output gates, and cell states, at time steps... The update formula is: Equation 1; In Equation 1, For low-dimensional high-level feature information at time step Extracted feature vectors Forgotten Gate in Time Step The output value, For the input gate at time step The output value, For the output gate at time step The output value, For cell state at time step The updated value, For cell state at time step The value, For Long Short-Term Memory Networks at time steps Hidden state output, For Long Short-Term Memory Networks at time steps The hidden state, The weight matrix of the forget gate, The weight matrix of the input gate, The weight matrix of the output gate, The weight matrix of the gate is updated to reflect the cell state. The bias vector of the forget gate, The bias vector of the input gate, The bias vector of the output gate, Let be the bias vector for the cell renewal gate. for Activation function The hyperbolic tangent activation function is used. This indicates element-wise multiplication.

[0007] In one feasible approach, the construction of the tunnel equipment relationship topology diagram structure includes: The various electromechanical equipment inside the tunnel are defined as shown in the diagram. , where the set of nodes Each node in the diagram represents a piece of electromechanical equipment. The total number of mechanical and electrical equipment, edge set Indicates the relationships between devices; At this point, let's set up the adjacency matrix of each electromechanical device in the tunnel. for: Equation 2; Equation 3; In equations 2 and 3, For nodes in the device relationship topology graph With nodes The combined adjacency weight value between them For nodes With nodes Spatial distance relationship between them For nodes With nodes Spacing between devices For bandwidth parameters, For nodes With nodes Functional coupling relationship between them For nodes With nodes Electrical topology relationship between them These are the weighting coefficients for spatial distance relationships. Weighting coefficients for functional coupling relationships These are the weighting coefficients for electrical topology relationships.

[0008] In one feasible approach, determining the location of the abnormal state sensing point of the target tunnel includes: Obtain the health score of each tunnel device, classify it into levels, and obtain the current status level of each tunnel device; Based on the relationship topology of tunnel equipment, extract at least one equipment linkage link; Collect multi-source heterogeneous sensing data across the entire linkage of each device to obtain linkage execution information for each device linkage link; Anomaly analysis is performed on the linkage execution information of each device linkage link to obtain the abnormal linkage links of the tunnel equipment, and the execution status characteristics of any device in the current abnormal linkage link are determined respectively. Based on the execution status characteristics of any device in the current equipment linkage link, and combined with the current status level of each tunnel device and the overall deployment information of the tunnel devices, the abnormal status detection points of the target tunnel are determined.

[0009] In one feasible solution, the anomaly analysis of the linkage execution information for each device linkage link includes: Obtain low-dimensional high-level feature information, health score and current status level of each electromechanical device in any device linkage link as the initial feature of the node, and construct the link feature matrix. A graph convolutional network is used to perform multi-level neighborhood feature aggregation on each node in the current device linkage link, and to extract the abnormal linkage features of each node in the current device linkage link. Based on the abnormal linkage characteristics of each node in the current device linkage link, the abnormal probability value of the current device linkage link is output, and the deviation of the execution state of each device in the current device linkage link is determined.

[0010] Specifically, a graph convolutional network is used to perform multi-level neighborhood feature aggregation on each node in the device linkage link. The inter-layer propagation rule of the graph convolutional network is as follows: Equation 4; In Equation 4, For the first The node feature matrix of a layered graph convolutional network should be noted as follows: Each row corresponds to a feature vector of a node. This refers to the layer index of a graph convolutional network. To add a self-loop adjacency matrix, This is a comprehensive adjacency matrix of the device relationship topology graph, with diagonal elements equal to 1 and off-diagonal elements equal to 0, used to add self-loops. It is the identity matrix; for The degree matrix, whose diagonal elements , Adjacency matrix with self-loops The Middle line, number The specific weight values ​​of the column, when The value is 0 when the element is not a diagonal element; For the first Layer node feature matrix, ; For the first Layer-trainable weight matrix, , No. The dimension of the feature vector of the layer node. No. The dimension of the feature vector of the layer node. To show that the matrix is OK, A matrix of real numbers, where each row corresponds to a device node and each column corresponds to a feature dimension. The weight matrix is ​​a OK, A real matrix of columns, It is a non-linear activation function.

[0011] In one feasible approach, determining the location of the abnormal state sensing point of the target tunnel includes: Obtain the anomaly probability value of each device linkage link and the deviation of the execution state of each device in the linkage link; Based on the abnormal probability value of each device linkage link, all device linkage links are sorted by the degree of abnormality, and device linkage links with abnormal probability values ​​exceeding the preset abnormal threshold are marked as abnormal linkage links. For each abnormal linkage link, the execution state deviation of each node in the abnormal linkage link is extracted, and the node with the largest execution state deviation is identified as the abnormal source node of the abnormal linkage link. Based on the relational topology of the tunnel equipment, with the abnormal source node as the center, the multi-level neighboring nodes of the abnormal source node are traversed, and the current state level and execution state deviation of each neighboring node are obtained step by step. Based on the current status level and execution status deviation of each neighboring node, and combined with the spatial location coordinates of the equipment in the overall deployment information of the tunnel equipment, the boundary of the impact range of the abnormal state is determined. The spatial coordinates of the abnormal source node and the locations of all nodes within its influence range are determined as abnormal state sensing points, and the device identifier, spatial coordinates, and corresponding abnormal linkage link information of the abnormal state sensing points are output.

[0012] Specifically, determining the boundary of the influence range of the abnormal state includes: Set the initial search radius with the anomaly source node as the center node. The initial influence range is determined by the initial search radius, and the current state level of all neighboring nodes within the initial influence range is obtained. If there are neighboring nodes with a current status level of "Severe" or "Caution" within the initial influence range, expand the search radius to the current search radius. Continue to obtain the current status level of newly added neighboring nodes within the expanded influence range; If the current status level of newly added neighboring nodes within the expanded influence range is "severe" or "attention", continue to expand the search radius until the current status level of all newly added neighboring nodes within the current influence range is "good" or "qualified", then stop expanding. The boundary of the influence range determined by the search radius when expansion stops is taken as the boundary of the influence range of the abnormal state.

[0013] In a second aspect, the present invention also provides an Internet of Things (IoT)-based system for sensing the state of electromechanical facilities in highway tunnels, employing an IoT-based method for sensing the state of electromechanical facilities in highway tunnels as described in any one of the first aspects, and further comprising: The data acquisition module is used to acquire the overall deployment information of the tunnel equipment and complete the acquisition of multi-source heterogeneous sensing data of the tunnel equipment. The data preprocessing module is used to unify and clean the collected multi-source heterogeneous sensing data, and output multivariate time series data of the tunnel equipment. The feature extraction module is used to perform unsupervised feature extraction on multivariate time series data to obtain low-dimensional high-level feature information of tunnel equipment, and to organize the feature information of different time periods according to the collection time sequence. The health scoring module is used to score the health of the tunnel equipment based on real-time collected perception data and a trained state evolution model, thereby obtaining the real-time usage status of the tunnel equipment. The linkage analysis module is used to perform equipment status linkage analysis based on the relationship topology diagram structure and status classification results of tunnel equipment, and output abnormal linkage information of tunnel equipment. An anomaly location module is used to combine anomaly linkage information with the overall deployment information of tunnel equipment to ultimately determine the location of anomaly sensing points within the target tunnel.

[0014] The beneficial effects of this invention are as follows: This invention achieves multi-source heterogeneous perception of the operational data of electromechanical facilities throughout the tunnel through an Internet of Things (IoT) network. It leverages a Long Short-Term Memory (LSTM) network to capture the temporal evolution patterns of equipment states and further combines this with a graph convolutional network to mine the linkage anomaly characteristics between equipment, ultimately achieving precise location of anomaly points. Compared to traditional monitoring methods, this significantly improves the response speed and positioning accuracy of anomaly perception, providing reliable support for the intelligent operation and maintenance of highway tunnel electromechanical systems. In other words, it effectively meets the need for global perception of the operational status of the entire tunnel electromechanical system. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the overall process of a method for sensing the status of electromechanical facilities in a highway tunnel based on the Internet of Things, provided in an embodiment of the present invention. Figure 2 This is a schematic diagram illustrating the overall process of a method for sensing the status of electromechanical facilities in a highway tunnel based on the Internet of Things, as provided in an embodiment of the present invention. Figure 3 This is a schematic diagram illustrating the process of generating abnormal state sensing points in an IoT-based method for sensing the state of electromechanical facilities in a highway tunnel, as provided in an embodiment of the present invention. Figure 4 This is a schematic diagram of an IoT-based highway tunnel electromechanical facility status perception system architecture provided in an embodiment of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0017] Furthermore, if the embodiments of this invention involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the meaning of "and / or" throughout the text includes three parallel solutions; for example, "A and / or B" includes solution A, solution B, or a solution where both A and B are satisfied simultaneously. Furthermore, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0018] Reference Figures 1 to 3 As shown, to address the challenge of achieving comprehensive awareness of the operational status of the entire tunnel electromechanical system in existing technologies, this invention provides an Internet of Things (IoT)-based method for sensing the status of highway tunnel electromechanical facilities. This method utilizes an IoT data acquisition network to achieve multi-source heterogeneous sensing of operational data from all areas of the tunnel's electromechanical facilities. It leverages a Long Short-Term Memory (LSTM) network to capture the temporal evolution patterns of equipment status and further combines this with a graph convolutional network to mine the linkage anomaly characteristics between equipment, ultimately achieving precise location of anomaly points. Compared to traditional monitoring methods, this significantly improves the response speed and positioning accuracy of anomaly sensing, providing reliable support for the intelligent operation and maintenance of highway tunnel electromechanical systems. In other words, it effectively meets the requirement of comprehensive awareness of the operational status of the entire tunnel electromechanical system. In practical applications, an IoT data acquisition and transmission network is first pre-built according to the deployment location of the tunnel's electromechanical facilities. Corresponding sensing sensors are configured for different types of electromechanical equipment to ensure that various operating parameters such as current, voltage, temperature, running time, and output power of the equipment can be comprehensively collected. All collected data is transmitted to the back-end monitoring platform through a dedicated industrial Ethernet for tunnels. The back-end platform completes the entire process of data preprocessing, feature extraction, evolution modeling, and anomaly analysis, and finally outputs anomaly location information for operation and maintenance personnel to handle.

[0019] like Figures 1 to 2 As shown, an IoT-based method for sensing the status of electromechanical facilities in highway tunnels specifically includes: acquiring overall deployment information of tunnel equipment (such as the location of each electromechanical device, the electrical connection relationship between each electromechanical device, and the service life of each electromechanical device) via wired or wireless means, and collecting multi-source heterogeneous sensing data of the tunnel equipment; it should be noted that the multi-source heterogeneous sensing data of the tunnel equipment may include: three-phase current, three-phase voltage, power factor, and frequency collected by current transformers and voltage transformers; equipment shell temperature and cable sheath temperature collected by platinum resistance temperature sensors and distributed temperature measuring optical cables; equipment vibration acceleration and vibration velocity collected by triaxial accelerometers; tunnel temperature and humidity, CO concentration, and smoke concentration collected by environmental sensors; and one or more combinations of equipment appearance images and instrument panel readings collected by high-definition cameras.

[0020] Preprocessing of multi-source heterogeneous sensing data of tunnel equipment yields multivariate time-series data of the tunnel equipment. Outliers can be identified and removed using the interquartile range method, i.e., for any feature sequence in the multi-source heterogeneous sensing data, the first quartile of that feature sequence is calculated. and the third and fourth quartiles At this time, the interquartile range is , then an anomaly judgment interval can be set according to the actual data situation, data points exceeding the interval are determined as outliers and eliminated, for example, the anomaly judgment interval can be set as: ; data points exceeding this interval are determined as outliers and eliminated.

[0021] Then, each feature data after eliminating the outliers is normalized separately, and each feature data is mapped to a unified numerical interval; here, the min-max normalization method can be used to map each feature data to the interval: where is the original data, and are the minimum value and maximum value of this feature respectively; in addition, for data from sensors with different sampling frequencies, the linear interpolation method is used for resampling to a unified time frequency, and feature splicing is performed on multi-source data at the same moment.

[0022] After completing the above processing of the multivariate time series data of the tunnel equipment, unsupervised feature extraction is performed on the multivariate time series data of the tunnel equipment to obtain low-dimensional high-level feature information of the tunnel equipment; that is, the original high-dimensional multi-source heterogeneous sensing data (for example, the original feature dimension is d) can be subjected to nonlinear compression transformation through the encoder part of the stacked autoencoder, to obtain a feature vector whose dimension is much lower than the original dimension (p<<d) but has more abstract and rich semantic information. This feature representation retains the most discriminative essential information for equipment state identification in the original data, while eliminating redundant information and noise interference. Its specific physical meaning no longer corresponds to a single sensor reading (such as a current value or a temperature value at a certain moment), but corresponds to an abstract representation of the equipment operating state in the latent space, for example, it can represent high-level semantics such as "early bearing wear mode" or "insulation aging trend". Thereby, the massive multi-source heterogeneous sensing data of the tunnel equipment can be processed, and the subsequent data analysis intensity and the overall data volume can be reduced. Here, to facilitate the understanding of the content described above, the following description is provided: a stacked autoencoder network can be constructed, the stacked autoencoder is composed of an encoder and a decoder, the encoder maps the input vector to a low-dimensional latent representation, and the decoder reconstructs the latent representation back to the original space; the stacked autoencoder is trained with the goal of minimizing the reconstruction error, and the loss function of the reconstruction error consists of a mean square reconstruction error term and a weight decay regularization term; the multivariate time series data is input into the trained stacked autoencoder, and the low-dimensional high-level feature information is output after layer-by-layer encoding by the encoder. As an example: The input vector can be encoded by the encoder into a latent representation: , where , , where For encoding the weight matrix, For encoding bias, For encoding activation functions; The hidden layer representation is then reconstructed back to the original space using a decoder. ,in, To decode the weight matrix, For decoding bias, This is the decoding activation function; It should be noted that the training objective of the stacked autoencoder described above is to minimize the reconstruction error, and the loss function is: , In this equation, the first term is the mean squared reconstruction error, the second term is the weight decay regularization term, and λ is the regularization coefficient. The stacked autoencoder can be trained using this equation. After training, the encoder portion is retained, and the input multivariate time-series data is encoded into low-dimensional, high-level feature information for output. Then, by collecting multi-source heterogeneous sensing data from tunnel equipment at different time intervals, the sensing data collected in each time interval is processed according to the aforementioned preprocessing and encoding procedures to construct low-dimensional, high-level feature information for tunnel equipment corresponding to different operational stages.

[0023] Having completed the extraction of low-dimensional and high-level feature information of tunnel equipment under different time series, the state evolution law of tunnel equipment can be simulated based on its state changes under different time series. That is, the low-dimensional and high-level feature information under different time series can be arranged in chronological order to obtain a time series sample set of equipment state. Based on the overall deployment information of tunnel equipment and combined with the state monitoring requirements of tunnel equipment, an evolution feature extraction network for tunnel equipment is constructed. The time series sample set of equipment state is input into the evolution feature extraction network of tunnel equipment to capture the evolution law of equipment state at different time scales and obtain the equipment state evolution model. Based on the equipment state evolution model and the low-dimensional and high-level feature information of the current time and historical time, the equipment state characteristics of future time are predicted. In this embodiment, based on the monitoring needs and deployment location of the tunnel, an evolution feature extraction network for the tunnel equipment can be constructed using a Long Short-Term Memory (LSTM) network. It should be noted that the LSTM network performs time-series dependency modeling through forget gates, input gates, output gates, and cell states. Then, the time-series sample set of the tunnel equipment is input into the LSTM network for training, capturing the evolution patterns of the equipment state at different time scales to obtain an equipment state evolution model. Subsequently, the equipment state evolution model can be used to predict the equipment state characteristics of the tunnel equipment at future times based on low-dimensional high-level feature information from the current and historical times. To facilitate understanding how the LSTM network predicts the equipment state characteristics of the tunnel equipment at future times, the following example is provided: The Long Short-Term Memory (LSTM) network models temporal dependencies using forgetting gates, input gates, output gates, and cell states, at time steps... The update formula is: Equation 1; In Equation 1, For low-dimensional high-level feature information at time step Extracted feature vectors Forgotten Gate in Time Step The output value, For the input gate at time step The output value, For the output gate at time step The output value, For cell state at time step The updated value, For cell state at time step The value, For Long Short-Term Memory Networks at time steps Hidden state output, For Long Short-Term Memory Networks at time steps The hidden state, The weight matrix of the forget gate, The weight matrix of the input gate, The weight matrix of the output gate, The weight matrix of the gate is updated to reflect the cell state. The bias vector of the forget gate, The bias vector of the input gate, The bias vector of the output gate, Let be the bias vector for the cell renewal gate. for Activation function The hyperbolic tangent activation function is used. This indicates element-wise multiplication. In this embodiment, the state of each unit in the Long Short-Term Memory network is updated using Equation 1 above, thereby transmitting and retaining the state characteristics of the tunnel equipment at different time steps. Redundant information unrelated to the evolution of the equipment state is filtered out, while key dependencies that reflect the trend of state changes are retained. This accurately captures the gradual change pattern of the tunnel equipment state over time, providing a reliable model basis for subsequent health assessment and anomaly localization.

[0024] After constructing the evolution feature extraction network for the tunnel equipment as described above, the real-time usage status of the tunnel equipment can be obtained by collecting multi-source heterogeneous sensing data of the tunnel equipment in real time and combining it with the state evolution law of the tunnel equipment; that is, by analyzing the output of each time step based on the Long Short-Term Memory network... By introducing an attention mechanism to weighted aggregate the features of historical time steps, importance weights of features at different time steps are obtained. Then, based on these importance weights, the outputs of the Long Short-Term Memory (LSTM) network at all time steps are weighted and summed to obtain a weighted fused feature vector. This weighted fused feature vector is then input into a fully connected regression layer for mapping, outputting a health score for the tunnel equipment. To facilitate understanding of how the health score for the tunnel equipment is calculated, the following example illustrates how the output of the LSM network at each time step... The unnormalized attention score can be calculated using the following formula. : ; In the formula, and For trainable parameters, For bias. Then adopt The function's attention score for the unnormalized Normalization is performed to obtain attention weights. : ; The weighted fusion feature vector The weighted sum for all time steps: ; The weighted fusion feature vector is calculated. Input a fully connected regression layer, output a health score for the tunneling device: ; In the formula, This is the output layer weight matrix. For output layer bias, for Activation function. In this embodiment, the health score of each individual tunnel device is calculated using the above formula. The score range is normalized and mapped to the [0,1] interval. A score closer to 0 indicates a worse health condition, and a score closer to 1 indicates a better health condition. Multiple health thresholds can be pre-set based on the actual needs of tunnel operation and management; for example, the score results can be divided into... It was classified as "excellent" at that time. When it is classified as "qualified", When it is classified as "attention", When a situation is classified as "serious," operations and management personnel can then take corresponding inspection, maintenance, or emergency repair measures based on the different levels.

[0025] In this embodiment, to facilitate clear and intuitive observation of the linkage relationship between tunnel equipment and to facilitate rapid location of anomalies in subsequent detection, a relationship topology graph structure for the tunnel equipment can be constructed based on the overall deployment information of the tunnel equipment. Specifically, constructing the relationship topology graph structure for the tunnel equipment includes: obtaining the geographical location and spatial coordinates of each tunnel equipment based on the overall deployment information; then constructing a relationship topology graph for the tunnel equipment using each tunnel equipment as a node and the spatial adjacency relationship and / or equipment association relationship between the equipment as edges; subsequently, defining the adjacency matrix of the relationship topology graph based on the spatial distance, functional coupling relationship, and electrical topology relationship between the equipment; wherein, the adjacency matrix is ​​a weighted sum of the spatial distance relationship matrix, the functional coupling relationship matrix, and the electrical topology relationship matrix. Here, for ease of understanding how to construct the relationship topology graph structure for the tunnel equipment, an example is provided: Each piece of electromechanical equipment in the tunnel can be defined as shown in the diagram. , where the set of nodes Each node in the diagram represents a piece of electromechanical equipment. The total number of mechanical and electrical equipment, edge set Indicates the relationships between devices; At this point, let's set up the adjacency matrix of each electromechanical device in the tunnel. for: Equation 2; Equation 3; In equations 2 and 3, For nodes in the device relationship topology graph With nodes The combined adjacency weight value between them For nodes With nodes Spatial distance relationship between them For nodes With nodes Spacing between devices For bandwidth parameters, For nodes With nodes Functional coupling relationship between them For nodes With nodes Electrical topology relationship between them These are the weighting coefficients for spatial distance relationships. Weighting coefficients for functional coupling relationships These are the weighting coefficients for the electrical topology relationships. That is, the adjacency matrix integrating multi-dimensional associations can be calculated using Equations 2 and 3, thus completing the construction of the relational topology graph structure, where the weighting coefficients... , , It can be adaptively adjusted according to the actual tunnel deployment scenario and operational monitoring needs to adapt to highway tunnel monitoring scenarios of different scales and equipment configurations. It should be noted that, in this embodiment, in order to have a reference basis for quickly locating the abnormal sensing position of tunnel equipment, and at the same time, the tunnel equipment can be classified according to its health score to obtain the current status level of each tunnel equipment.

[0026] Reference Figure 3As shown, after completing the relationship topology diagram of the tunnel equipment and the current status level of each tunnel equipment, the current health score of each tunnel equipment can be used as the feature input of the corresponding node. Combined with the constructed relationship topology diagram, a graph convolutional network is used to aggregate and propagate the correlation features between equipment, fully mining the correlation information of equipment status in the topological space, thereby achieving accurate location of abnormal equipment. Specifically, based on the relationship topology diagram of the tunnel equipment and the status classification results, status linkage analysis of the tunnel equipment can be performed to obtain abnormal linkage information of the tunnel equipment. Then, based on the abnormal linkage information of the tunnel equipment and combined with the overall deployment information of the tunnel equipment, the abnormal status sensing points of the target tunnel can be determined. Specifically, determining the abnormal state sensing points of the target tunnel includes: obtaining the health score of each tunnel device and classifying it into levels to obtain the current state level of each tunnel device; then, based on the relationship topology of the tunnel devices, extracting at least one device linkage link; collecting multi-source heterogeneous sensing data for each device linkage link to obtain the linkage execution information of each device linkage link; that is, by collecting multi-source heterogeneous sensing data, low-dimensional high-level feature information, health score, and current state level of each electromechanical device in any device linkage link can be obtained, and then the low-dimensional high-level feature information, health score, and current state level of each electromechanical device in any device linkage link can be used as the initial features of the nodes to construct a link feature matrix. Subsequently, anomaly analysis can be performed on the linkage execution information of each device linkage link to obtain the abnormal linkage links of the tunnel devices, and the execution state characteristics of any device in the current abnormal linkage link can be determined respectively; based on the execution state characteristics of any device in the current device linkage link, combined with the current state level of each tunnel device and the overall deployment information of the tunnel devices, the abnormal state sensing points of the target tunnel can be determined. To facilitate understanding of how the above-mentioned anomaly analysis of the linkage execution information of each device linkage link is performed, the following explanation is provided: The anomaly analysis of the linkage execution information of each device linkage link includes: obtaining the low-dimensional high-level feature information, health score, and current status level of each electromechanical device in any device linkage link as the initial features of the node, and constructing a link feature matrix (i.e., for the initial features of the node, the feature dimension corresponds to the above-mentioned multi-dimensional feature information of each device, and the number of rows corresponds to the number of devices included in the link); then, by using a graph convolutional network to perform multi-order neighborhood feature aggregation on each node in the current device linkage link, the abnormal linkage features of each node in the current device linkage link are extracted respectively; then, based on the abnormal linkage features of each node in the current device linkage link, the abnormal probability value of the current device linkage link can be output, and the deviation of the execution status of each device in the current device linkage link can be determined.That is, when the abnormal probability value exceeds the preset abnormal threshold, the device linkage link is determined to be an abnormal linkage link, and the execution state deviation of each device in the link is output; it should be noted that the execution state deviation is the deviation between the actual operating parameters of the device and the predicted value of the device state evolution model.

[0027] To facilitate understanding of how to use a graph convolutional network to perform multi-level neighborhood feature aggregation on each node in the device linkage link, the following example illustrates this: A graph convolutional network is used to perform multi-level neighborhood feature aggregation on each node in the device linkage link. The inter-layer propagation rule of the graph convolutional network is as follows: Equation 4; In Equation 4, For the first The node feature matrix of a layered graph convolutional network should be noted as follows: Each row corresponds to a feature vector of a node. This refers to the layer index of a graph convolutional network. To add a self-loop adjacency matrix, This is a comprehensive adjacency matrix of the device relationship topology graph, with diagonal elements equal to 1 and off-diagonal elements equal to 0, used to add self-loops. It is the identity matrix; for The degree matrix, whose diagonal elements , Adjacency matrix with self-loops The Middle line, number The specific weight values ​​of the column, when The value is 0 when the element is not a diagonal element; For the first Layer node feature matrix, ; For the first Layer-trainable weight matrix, , No. The dimension of the feature vector of the layer node. No. The dimension of the feature vector of the layer node. To show that the matrix is OK, A matrix of real numbers, where each row corresponds to a device node and each column corresponds to a feature dimension. The weight matrix is ​​a OK, A real matrix of columns, This is a non-linear activation function. Equation 4 allows for two-layer graph convolutional propagation, enabling feature aggregation of all associated devices within the second-order neighborhood of the target node. This fully integrates the state information of spatially adjacent, functionally coupled, and electrically connected devices, effectively capturing the propagation effect of abnormal states in the topology link and avoiding misjudgments caused by occasional fluctuations in a single device. Finally, based on the node output features obtained after the two-layer feature aggregation, an anomaly score is calculated for each node. Devices with scores higher than the preset anomaly judgment threshold are identified as anomaly detection points. Subsequently, all identified anomaly points can be pushed to the tunnel operation and management platform, automatically generating anomaly alarm information to remind maintenance personnel to go to the corresponding locations for verification and handling.

[0028] In one feasible solution, to achieve faster detection of abnormal linkage locations and abnormal state sensing points in the tunnel, determining the abnormal state sensing points of the target tunnel includes: acquiring the abnormal probability value of each device linkage link and the execution state deviation of each device in the linkage link; then, based on the abnormal probability value of each device linkage link, ranking all device linkage links by their degree of abnormality, and marking the device linkage links whose abnormal probability values ​​exceed a preset abnormal threshold as abnormal linkage links; for each abnormal linkage link, extracting the execution state deviation of each node in the abnormal linkage link, and identifying the node with the largest execution state deviation as the abnormal linkage link. The abnormal source node of the tunnel is identified. Then, based on the relationship topology of the tunnel equipment, a breadth-first search algorithm is used to traverse the multi-level neighboring nodes of the abnormal source node, obtaining the current state level and execution state deviation of each neighboring node at each level. Based on the current state level and execution state deviation of each neighboring node, combined with the spatial coordinates of the equipment in the overall deployment information of the tunnel equipment, the boundary of the influence range of the abnormal state is determined. The spatial coordinates of the abnormal source node and the positions of all nodes within the boundary of its influence range are determined as abnormal state sensing points, and the equipment identifier, spatial coordinates, and corresponding abnormal linkage link information of the abnormal state sensing points are output. In other words, by using the above method, the source of the abnormality can be directly located and the influence range of the abnormality can be clearly defined, significantly reducing the scope and time of abnormality verification, helping maintenance personnel to quickly locate the core location of the fault, and improving the response efficiency of tunnel equipment abnormality handling.

[0029] To facilitate understanding of how the influence range boundary of an abnormal state is determined, as an example, determining the influence range boundary of an abnormal state includes: setting an initial search radius with the abnormal source node as the center node. The initial influence range is determined using the initial search radius, and the current status level of all neighboring nodes within this initial influence range is obtained. If there are neighboring nodes within the initial influence range with a current status level of "severe" or "attention", the search radius is expanded to the current search radius. The algorithm continues to acquire the current status level of newly added neighboring nodes within the expanded influence range. If any newly added neighboring nodes within the expanded influence range have a current status level of "Severe" or "Attention," the search radius is expanded until all newly added neighboring nodes within the current influence range have a current status level of "Excellent" or "Qualified." The boundary of the influence range determined by the search radius at which expansion stops is taken as the boundary of the influence range of the abnormal state. In other words, by traversing the multi-level neighboring nodes of the abnormal source node using the breadth-first search algorithm described above, and acquiring the current status level and execution status deviation of each neighboring node at each level, dynamic adaptation to the spread range of the abnormal state can be achieved. This avoids excessive expansion of the abnormal range, preventing unnecessary waste of maintenance resources, and also ensures that devices already affected by the abnormal propagation are not overlooked, guaranteeing the accuracy and completeness of the abnormal perception results. Subsequently, based on the defined abnormal influence range, an abnormal situation distribution map of tunnel equipment can be automatically generated, intuitively displaying the abnormal source, influence range, and status level of each affected device, providing intuitive decision-making reference for maintenance personnel to formulate handling plans.

[0030] Reference Figure 4In a second aspect, this invention also provides an Internet of Things (IoT)-based status sensing system for highway tunnel electromechanical facilities. This system employs the aforementioned IoT-based method for monitoring the status of highway tunnel electromechanical facilities, enabling intelligent sensing of their operational status and completing the entire status sensing process. The system supports direct integration with existing tunnel operation and management platforms, requiring no large-scale modifications to the existing tunnel sensing hardware network. It can be deployed simply by upgrading the algorithm module, adapting to the operational monitoring needs of highway tunnels at different construction stages and with varying equipment configurations. This effectively reduces the frequency of manual inspections and maintenance costs, improving the safety level and intelligent management capabilities of tunnel operations. Specifically, the sensing system includes: a data acquisition module, a data preprocessing module, a feature extraction module, a health scoring module, a linkage analysis module, and an anomaly localization module. The data acquisition module acquires the overall deployment information of the tunnel equipment and completes the collection of multi-source heterogeneous sensing data. The data preprocessing module unifies and cleans the collected multi-source heterogeneous sensing data, outputting multivariate time-series data of the tunnel equipment. The feature extraction module performs unsupervised feature extraction on the multivariate time-series data to obtain low-dimensional, high-level feature information of the tunnel equipment, and organizes the feature information for different time periods according to the acquisition sequence. The evolutionary modeling module trains an equipment state evolution model based on the organized low-dimensional, high-level feature information from different time periods and fits the model. The system describes the state evolution of tunnel equipment. The health scoring module, based on real-time collected sensing data and a trained state evolution model, scores the health of the tunnel equipment to obtain its real-time usage status. The topology construction module, based on the overall deployment information of the tunnel equipment, integrates spatial distance, functional coupling, and electrical topology relationships to construct a relational topology graph of the tunnel equipment and classifies the equipment status levels based on the health score. The linkage analysis module, based on the relational topology graph and status classification results, performs linkage analysis of the equipment status and outputs abnormal linkage information. The anomaly location module, combining the abnormal linkage information with the overall deployment information of the tunnel equipment, ultimately determines the location of abnormal state sensing points within the target tunnel. In this embodiment, the sensing system, through the orderly cooperation of its functional modules, combines multi-dimensional equipment relationships with graph convolution feature aggregation methods, overcoming the limitations of traditional single-point anomaly sensing that only focuses on single-device operating data. It effectively captures the transmission patterns of abnormal states in the equipment linkage network, significantly improving the accuracy and efficiency of anomaly sensing and location in highway tunnel electromechanical facilities, thus meeting the actual needs of intelligent operation and management of current highway tunnels. In actual deployment and application, operators can adjust core parameters such as weighting coefficients and anomaly judgment thresholds according to actual conditions such as tunnel length, total number of devices, and construction year, flexibly adapting to the perception needs of different scenarios, and possessing strong versatility and scalability.

[0031] In some implementations, the sensing system can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet), and end-to-end networks (e.g., ad hoc end-to-end networks), as well as any currently known or future-developed networks.

[0032] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), etc.

[0033] A third aspect of the present invention provides a computer-readable medium having a computer program stored thereon, wherein the program, when executed by a processor, implements a method for sensing the status of electromechanical facilities in a highway tunnel based on the Internet of Things, as described in any one of the first aspects. The computer-readable medium in this embodiment can be written in one or more programming languages ​​or a combination thereof to perform computer program code for carrying out operations of some embodiments of the present disclosure. These programming languages ​​include object-oriented programming languages—such as Java, Smalltalk, and C++—and conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can be executed entirely on a user's computer, partially on a user's computer, as a standalone software package, partially on a user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0034] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0035] In particular, according to some embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, some embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts.

[0036] A fourth aspect of the present invention provides an electronic device, comprising: one or more processors; a storage device storing one or more programs thereon; wherein, when the one or more programs are executed by the one or more processors, the one or more processors implement the Internet of Things-based method for sensing the state of electromechanical facilities in highway tunnels as described in the first aspect. The computer-readable medium may be included in the electronic device or may exist independently, i.e., not assembled into the electronic device. The computer-readable medium carries one or more programs, which, when executed by the electronic device, enable the electronic device to implement the Internet of Things-based method for sensing the state of electromechanical facilities in highway tunnels as described in the first aspect.

[0037] The fifth aspect of the present invention provides a computer program product, including a computer program that, when executed by a processor, implements a method for sensing the status of electromechanical facilities in a highway tunnel based on the Internet of Things as described in the first aspect.

[0038] The above description is merely a selection of preferred embodiments of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A method for sensing the status of electromechanical facilities in highway tunnels based on the Internet of Things, characterized in that, include: Acquire overall deployment information of tunnel equipment and collect multi-source heterogeneous sensing data of tunnel equipment; Preprocessing of multi-source heterogeneous sensing data of tunnel equipment yields multivariate time series data of tunnel equipment; Unsupervised feature extraction is performed on multivariate time series data of tunnel equipment to obtain low-dimensional high-level feature information of tunnel equipment. Collect multi-source heterogeneous sensing data of tunnel equipment at different time intervals, and construct low-dimensional high-level feature information of tunnel equipment at different time intervals; Based on the low-dimensional high-level feature information of tunnel equipment under different time series, the state evolution law of tunnel equipment is simulated. Real-time acquisition of multi-source heterogeneous sensing data from tunnel equipment, combined with the state evolution patterns of tunnel equipment, to perform health scoring on tunnel equipment and obtain the real-time usage status of tunnel equipment; Based on the overall deployment information of tunnel equipment, a relational topology diagram of tunnel equipment is constructed, and the tunnel equipment is classified according to its health score to obtain the current status level of each tunnel equipment. Based on the relationship topology of the tunnel equipment and combined with the status classification results of the tunnel equipment, a status linkage analysis of the tunnel equipment is performed to obtain abnormal linkage information of the tunnel equipment. Based on the abnormal linkage information of tunnel equipment and combined with the overall deployment information of tunnel equipment, the abnormal state sensing points of the target tunnel are determined.

2. The method for sensing the status of electromechanical facilities in highway tunnels based on the Internet of Things according to claim 1, characterized in that, The state evolution law of the simulated tunnel equipment includes: Arrange the low-dimensional high-level feature information under different time series in chronological order to obtain a time series sample set of device status; Based on the overall deployment information of tunnel equipment and combined with the status monitoring requirements of tunnel equipment, an evolution feature extraction network for tunnel equipment is constructed. The time series sample set of equipment status is input into the evolution feature extraction network of tunnel equipment to capture the evolution law of equipment status at different time scales and obtain the equipment status evolution model. Based on the equipment state evolution model, and using the low-dimensional high-level feature information of the current and historical moments, the equipment state characteristics of future moments are predicted.

3. The method for sensing the status of electromechanical facilities in highway tunnels based on the Internet of Things according to claim 2, characterized in that, The evolution feature extraction network for the tunnel equipment adopts a long short-term memory network. Specifically, the Long Short-Term Memory (LSTM) network models temporal dependencies through forgetting gates, input gates, output gates, and cell states, at time steps... The update formula is: Equation 1; In Equation 1, For low-dimensional high-level feature information at time step Extracted feature vectors Forgotten Gate in Time Step The output value, For the input gate at time step The output value, For the output gate at time step The output value, For cell state at time step The updated value, For cell state at time step The value, For Long Short-Term Memory Networks at time steps Hidden state output, For Long Short-Term Memory Networks at time steps The hidden state, The weight matrix of the forget gate, The weight matrix of the input gate, The weight matrix of the output gate, The weight matrix of the gate is updated to reflect the cell state. The bias vector of the forget gate, The bias vector of the input gate, The bias vector of the output gate, Let be the bias vector for the cell renewal gate. for Activation function The hyperbolic tangent activation function is used. This indicates element-wise multiplication.

4. The method for sensing the status of electromechanical facilities in highway tunnels based on the Internet of Things according to claim 3, characterized in that, The relationship topology diagram structure for constructing tunnel equipment includes: The various electromechanical equipment inside the tunnel are defined as shown in the diagram. , where the set of nodes Each node in the diagram represents a piece of electromechanical equipment. The total number of mechanical and electrical equipment, edge set Indicates the relationships between devices; At this point, let's set up the adjacency matrix of each electromechanical device in the tunnel. for: Equation 2; Equation 3; In equations 2 and 3, For nodes in the device relationship topology graph With nodes The combined adjacency weight value between them For nodes With nodes Spatial distance relationship between them For nodes With nodes Spacing between devices For bandwidth parameters, For nodes With nodes Functional coupling relationship between them For nodes With nodes Electrical topology relationship between them These are the weighting coefficients for spatial distance relationships. Weighting coefficients for functional coupling relationships These are the weighting coefficients for electrical topology relationships.

5. The method for sensing the status of electromechanical facilities in highway tunnels based on the Internet of Things according to claim 1, characterized in that, The determination of the abnormal state sensing points of the target tunnel includes: Obtain the health score of each tunnel device, classify it into levels, and obtain the current status level of each tunnel device; Based on the relationship topology of tunnel equipment, extract at least one equipment linkage link; Collect multi-source heterogeneous sensing data across the entire linkage of each device to obtain linkage execution information for each device linkage link; Anomaly analysis is performed on the linkage execution information of each device linkage link to obtain the abnormal linkage links of the tunnel equipment, and the execution status characteristics of any device in the current abnormal linkage link are determined respectively. Based on the execution status characteristics of any device in the current equipment linkage link, and combined with the current status level of each tunnel device and the overall deployment information of the tunnel devices, the abnormal status detection points of the target tunnel are determined.

6. The method for sensing the status of electromechanical facilities in highway tunnels based on the Internet of Things according to claim 5, characterized in that, The anomaly analysis of the linkage execution information for each device linkage link includes: Obtain low-dimensional high-level feature information, health score and current status level of each electromechanical device in any device linkage link as the initial feature of the node, and construct the link feature matrix. A graph convolutional network is used to perform multi-level neighborhood feature aggregation on each node in the current device linkage link, and to extract the abnormal linkage features of each node in the current device linkage link. Based on the abnormal linkage characteristics of each node in the current device linkage link, the abnormal probability value of the current device linkage link is output, and the deviation of the execution state of each device in the current device linkage link is determined.

7. The method for sensing the status of electromechanical facilities in a highway tunnel based on the Internet of Things according to claim 6, characterized in that, A graph convolutional network is used to perform multi-level neighborhood feature aggregation on each node in the device linkage link. The inter-layer propagation rule of the graph convolutional network is as follows: Equation 4; In Equation 4, For the first The node feature matrix of a layered graph convolutional network should be noted as follows: Each row corresponds to a feature vector of a node. This refers to the layer index of a graph convolutional network. To add a self-loop adjacency matrix, This is a comprehensive adjacency matrix of the device relationship topology graph, with diagonal elements equal to 1 and off-diagonal elements equal to 0, used to add self-loops. It is the identity matrix; for The degree matrix, whose diagonal elements , Adjacency matrix with self-loops The Middle line, number The specific weight values ​​of the column, when The value is 0 when the element is not a diagonal element; For the first Layer node feature matrix, ; For the first Layer-trainable weight matrix, , No. The dimension of the feature vector of the layer node. No. The dimension of the feature vector of the layer node. To show that the matrix is OK, A matrix of real numbers, where each row corresponds to a device node and each column corresponds to a feature dimension. The weight matrix is ​​a OK, A real matrix of columns, It is a non-linear activation function.

8. A method for sensing the status of electromechanical facilities in highway tunnels based on the Internet of Things, as described in any one of claims 6 to 7, characterized in that, The determination of the abnormal state sensing points of the target tunnel includes: Obtain the anomaly probability value of each device linkage link and the deviation of the execution state of each device in the linkage link; Based on the abnormal probability value of each device linkage link, all device linkage links are sorted by the degree of abnormality, and device linkage links with abnormal probability values ​​exceeding the preset abnormal threshold are marked as abnormal linkage links. For each abnormal linkage link, the execution state deviation of each node in the abnormal linkage link is extracted, and the node with the largest execution state deviation is identified as the abnormal source node of the abnormal linkage link. Based on the relational topology of the tunnel equipment, with the abnormal source node as the center, the multi-level neighboring nodes of the abnormal source node are traversed, and the current state level and execution state deviation of each neighboring node are obtained step by step. Based on the current status level and execution status deviation of each neighboring node, and combined with the spatial location coordinates of the equipment in the overall deployment information of the tunnel equipment, the boundary of the impact range of the abnormal state is determined. The spatial coordinates of the abnormal source node and the locations of all nodes within its influence range are determined as abnormal state sensing points, and the device identifier, spatial coordinates, and corresponding abnormal linkage link information of the abnormal state sensing points are output.

9. A method for sensing the status of electromechanical facilities in highway tunnels based on the Internet of Things, as described in claim 8, is characterized in that, The determination of the influence range boundary of the abnormal state includes: Set the initial search radius with the anomaly source node as the center node. The initial influence range is determined by the initial search radius, and the current state level of all neighboring nodes within the initial influence range is obtained. If there are neighboring nodes with a current status level of "Severe" or "Caution" within the initial influence range, expand the search radius to the current search radius. Continue to obtain the current status level of newly added neighboring nodes within the expanded influence range; If the current status level of newly added neighboring nodes within the expanded influence range is "severe" or "attention", continue to expand the search radius until the current status level of all newly added neighboring nodes within the current influence range is "good" or "qualified", then stop expanding. The boundary of the influence range determined by the search radius when expansion stops is taken as the boundary of the influence range of the abnormal state.

10. A status sensing system for electromechanical facilities in highway tunnels based on the Internet of Things, characterized in that, The method for sensing the status of electromechanical facilities in highway tunnels based on the Internet of Things, as described in any one of claims 1 to 9, includes: The data acquisition module is used to acquire the overall deployment information of the tunnel equipment and complete the acquisition of multi-source heterogeneous sensing data of the tunnel equipment. The data preprocessing module is used to unify and clean the collected multi-source heterogeneous sensing data, and output multivariate time series data of the tunnel equipment. The feature extraction module is used to perform unsupervised feature extraction on multivariate time series data to obtain low-dimensional high-level feature information of tunnel equipment, and to organize the feature information of different time periods according to the collection time sequence. The health scoring module is used to score the health of the tunnel equipment based on real-time collected perception data and a trained state evolution model, thereby obtaining the real-time usage status of the tunnel equipment. The linkage analysis module is used to perform equipment status linkage analysis based on the relationship topology diagram structure and status classification results of tunnel equipment, and output abnormal linkage information of tunnel equipment. An anomaly location module is used to combine anomaly linkage information with the overall deployment information of tunnel equipment to ultimately determine the location of anomaly sensing points within the target tunnel.

Images