Bridge health real-time monitoring system based on intelligent sensing and big data analysis

Abstract

The invention discloses a bridge health real-time monitoring system based on intelligent sensing and big data analysis. The bridge health real-time monitoring system comprises an intelligent sensing layer, a data transmission layer, an edge computing layer, a cloud computing layer and an application layer, the intelligent sensing layer deploys a multi-modal sensor on a bridge in a self-adaptive manner; the data transmission layer can perform data transmission; the edge calculation layer carries out preprocessing and real-time analysis on the data; the cloud computing layer is used for fusing multi-source data based on a constructed federal learning driven big data analysis platform and carrying out long-term performance prediction on a bridge; and the application layer provides a visual monitoring interface and intelligent decision support, and realizes autonomous optimization of a bridge maintenance strategy based on reinforcement learning. According to the invention, advanced technologies such as quantum sensing, dynamic digital twinning, federated learning and the like are integrated, real-time preprocessing and analysis of data are realized, and the response speed of a monitoring system is obviously improved; and multi-source data fusion and collaborative analysis are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of bridge monitoring, and in particular to a real-time bridge health monitoring system based on intelligent sensing and big data analysis. Background Art

[0002] Traditional bridge health monitoring methods rely primarily on manual inspections, periodic testing, and the collection and analysis of data from a limited number of sensors. These methods have numerous limitations, including low monitoring frequency, limited data coverage, poor real-time performance, and insufficient analytical accuracy. As bridge structures become increasingly complex, traditional monitoring methods are no longer able to meet the demands of modern bridge safety and maintenance.

[0003] In recent years, with the rapid development of technologies such as the Internet of Things, big data, and artificial intelligence, the field of bridge health monitoring has ushered in new technological innovations. The introduction of dynamic digital twin technology has made real-time simulation and prediction of bridge structure and behavior a reality. Although these technologies have made significant progress in their respective fields, their integration and application in bridge health monitoring still face numerous challenges. Further research is needed to accurately integrate multi-source heterogeneous data and build efficient and reliable monitoring networks. Summary of the Invention

[0004] To this end, the present invention provides a real-time bridge health monitoring system based on intelligent sensing and big data analysis to solve the problems in the prior art.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] A real-time bridge health monitoring system based on intelligent sensing and big data analysis, including intelligent sensing layer, data transmission layer, edge computing layer, cloud computing layer and application layer;

[0007] The intelligent sensing layer determines the monitoring section of the bridge through finite element analysis, optimizes sensor positions using an improved particle swarm algorithm, and adaptively deploys multimodal sensors on the bridge; the multimodal sensors include node sensors with data transmission capabilities;

[0008] The data transmission layer performs data transmission based on the constructed low-power wide area network;

[0009] The edge computing layer pre-processes the data and performs real-time analysis. Several edge computing layers are deployed on the bridge.

[0010] The cloud computing layer, based on a built federated learning-driven big data analysis platform, integrates multi-source data and performs long-term performance predictions on bridges.

[0011] The application layer provides a visual monitoring interface and intelligent decision support. The application layer also realizes autonomous optimization of bridge maintenance strategies based on reinforcement learning.

[0012] Further: Particle swarm algorithm optimizes the sensor position, and its calculation formula is as follows:

[0013]

[0014] Among them, σ i (x) and ∈ i (x) are the stress and strain sensitivities of the i-th node, c i is the deployment cost, w1 and w2 are weight coefficients, C max For the budget constraint.

[0015] Furthermore: the data transmission layer adopts LoRaWAN and 5G-A fusion networking, LoRaWAN is used for low-frequency environmental data transmission, and 5G-A is used for high-frequency structural response data transmission. The calculation formula is as follows:

[0016]

[0017] Among them, T total is the total transmission delay, D LoRa 、D 5G is the amount of data, R LoRa 、R 5G is the transmission rate.

[0018] Further: the vibration signal is compressed at level 3 based on wavelet transform.

[0019] Furthermore, the edge computing layer implements dynamic allocation of computing tasks between edge nodes and the cloud based on reinforcement learning. The specific calculation formula is as follows:

[0020]

[0021] Among them, Q(s,a) is the Q value (state-action value function) of performing action a in the current state s, which represents the expected long-term return of taking this action in this state; α is the learning rate, r is the immediate reward, γ is the discount factor, s′ is the new state reached after performing action a, and a′ represents the possible actions that can be taken in the new state s′. All actions need to be traversed to find the maximum value.

[0022] Furthermore, the edge computing layer can also use a long short-term memory network to learn historical data patterns and generate an adaptive warning threshold, as shown in the following formula:

[0023]

[0024] in, is the current moment prediction value, σ is the activation function, W f 、W i is the weight matrix, is the input sequence.

[0025] Further: Building a digital twin framework for federated learning is achieved through the following steps:

[0026] (1) Local model training:

[0027] First, a dynamic digital twin is constructed based on the fusion of the bridge finite element model and the measured data. The formula is as follows:

[0028] M twin =α□M FEM +(1-α)□M data ;

[0029] Among them, M twin is the fused digital twin model, α is the fusion coefficient, which controls the weight of the finite element model and the measured data, M FEM is a bridge model based on finite element analysis, M data A bridge model constructed based on measured data;

[0030] (2) Local model optimization: The Adam optimizer is used to update the model parameters, and the learning rate dynamic adjustment strategy is as follows:

[0031]

[0032] Among them, η t is the current learning rate, η0 is the initial learning rate, β is the decay rate, and t is the number of training iterations;

[0033] (3) Secure aggregation: Differential privacy protection technology is used to perturb the local model parameters. The calculation formula is as follows:

[0034]

[0035] in, is the perturbation parameter; w k are the original model parameters; N(0, σ 2 I) is Gaussian noise with a mean of 0; σ 2 I is the variance; I is the identity matrix; σ controls the noise intensity;

[0036] (4) Global model update: Aggregate the global model through the FedAvg algorithm. The formula is as follows:

[0037]

[0038] Among them, Wglobal is the global model parameter, which is obtained by weighted average of all local model parameters; K is the number of bridges participating in federated learning; nk is the local data volume of the kth bridge; n is the total data volume of all bridges, are the local model parameters of the kth bridge.

[0039] Furthermore, the cloud computing layer needs to build a long-term performance prediction system. The specific steps are as follows:

[0040] (1) Feature extraction, including time domain features, frequency domain features, and time-frequency domain features;

[0041] (2) Construct a prediction model based on a temporal convolutional network, using causal convolution and dilated convolution to capture temporal dependencies. The formula is as follows:

[0042] y t =σ(W*x t +b);

[0043] Where yt is the output value at time t, * is the convolution operation, σ is the activation function, W is the convolution kernel weight matrix, xt is the input sequence at time t, and b is the bias term;

[0044] The self-attention mechanism is introduced to strengthen key features. The calculation formula is as follows:

[0045]

[0046] Among them, Q is the displacement query matrix, K is the key matrix, V is the value matrix, dk is the key vector dimension, and softmax is the normalized exponential function.

[0047] Furthermore, the application layer constructs a state-action-reward model to optimize the allocation of maintenance resources, and the calculation is as follows:

[0048]

[0049] Among them, μθ(s) is the policy function, Q μ (s,a) is the action value function

[0050] Based on the digital twin simulation results, the bridge load limit is dynamically adjusted according to the following formula:

[0051]

[0052] Among them, L limit is the dynamic load limit, L design is the design load limit, ∈ actual is the measured strain, ∈ threshold is the warning threshold.

[0053] Going further: for abnormal data detected by multimodal sensors, the intelligent sensing layer transmits the data directly to the application layer based on node sensors with transmission capabilities, and triggers an early warning; when an early warning is generated, maintenance personnel are required to confirm on-site to first confirm whether there is a fault in the multimodal sensor, and then further detection; when multiple continuously arranged multimodal sensors emit abnormal data, an emergency warning is triggered.

[0054] The present invention has the following advantages: it integrates cutting-edge technologies such as quantum sensing, dynamic digital twins, and federated learning to build a new generation of bridge health monitoring system, and the deployment of adaptive edge computing nodes realizes real-time preprocessing and analysis of data, significantly improving the response speed of the monitoring system.

[0055] This federated learning-driven big data analytics platform enables multi-source data fusion and collaborative analysis, supporting cross-bridge data collaborative analysis. A federated averaging algorithm aggregates global models, improving model accuracy and generalization. A reinforcement learning-based autonomous decision-making system, through the construction of a state-action-reward model, enables autonomous optimization of bridge maintenance strategies, improving resource utilization and maintenance efficiency.

[0056] Other features and advantages of the present invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] To more intuitively illustrate the prior art and the present application, exemplary drawings are provided below. It should be understood that the specific shapes and structures shown in the drawings should not generally be considered as limiting conditions for implementing the present application; for example, based on the technical concepts disclosed in this application and the exemplary drawings, those skilled in the art are capable of easily making routine adjustments or further optimizations to the addition / reduction / attribution division, specific shapes, positional relationships, connection methods, dimensional ratios, etc. of certain units (components).

[0058] Figure 1 This is a system block diagram of a real-time bridge health monitoring system based on intelligent sensing and big data analysis provided in one embodiment of the present application. DETAILED DESCRIPTION

[0059] The following specific embodiments illustrate the implementation of the present invention. People familiar with this technology can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. It should be understood that these embodiments are only to further illustrate the present invention and cannot be understood as limiting the scope of protection of the present invention. Technical engineers in this field can make some non-essential improvements and adjustments to the present invention based on the content of the above invention; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0060] See also Figure 1 The real-time bridge health monitoring system based on intelligent sensing and big data analysis includes intelligent sensing layer, data transmission layer, edge computing layer, cloud computing layer and application layer.

[0061] a. The intelligent sensing layer deploys a multimodal sensor network to collect real-time data on bridge structural response, environmental parameters, and traffic loads. Adaptive sensor deployment is based on bridge structural topology optimization theory, using finite element analysis to identify key monitoring sections (such as main beams, pylons, and cable anchorage areas). Multimodal sensors include node sensors with data transmission capabilities.

[0062] In this embodiment, the multimodal sensor types include: a quantum diamond NV color center strain gauge with an accuracy of ±0.1με and a dynamic range of 0-5000με, suitable for high-precision strain monitoring; a TENG self-powered displacement sensor with a resolution of 0.01mm, based on triboelectric nano-electrical generation technology, requiring no external power supply; and a fiber Bragg grating (FBG) strain gauge with a central wavelength of 1550nm and a temperature compensation error of <±0.5%. The quantum sensor (quantum diamond NV color center strain gauge) is encapsulated in epoxy resin and has an operating temperature range of -30°C to 70°C and a humidity protection level of IP68. The TENG self-powered displacement sensor undergoes laser calibration every six months to ensure accuracy.

[0063] Combined with the improved particle swarm optimization (PSO) to optimize the sensor position,

[0064]

[0065] Among them, σ i (x) and ∈ i (x) are the stress and strain sensitivities of the i-th node, c i is the deployment cost, w1 and w2 are weight coefficients, C max For the budget constraint.

[0066] A dynamic inertia weight adjustment strategy is employed, setting the inertia weight to decrease with the number of iterations t. A genetic algorithm mutation operation is also introduced to randomly reset particle positions with a 5% probability, enhancing global search capabilities. In this embodiment, the weight coefficients w1 and w2 are based on the Pareto frontier of stress sensitivity, strain sensitivity, and deployment cost, ultimately set to w1 = 0.6 and w2 = 0.4.

[0067] b. Data transmission layer: building a low-power wide area network (LPWAN) to achieve efficient transmission of sensor data.

[0068] LoRaWAN and 5G-A (5G Advanced) are used for integrated networking. LoRaWAN is used for low-frequency environmental data transmission (such as temperature and humidity), and 5G-A is used for high-frequency structural response data transmission (such as vibration and strain). The formula is as follows:

[0069]

[0070] Among them, T total is the total transmission delay, D LoRa 、D 5G is the amount of data, R LoRa 、R 5G is the transmission rate.

[0071] In addition, the vibration signal is compressed at three levels based on wavelet transform (WT), and the AES-256 encryption algorithm is used to ensure data security.

[0072] In this embodiment, the converged networking coordination mechanism is specifically as follows:

[0073] Frequency band allocation: LoRaWAN uses the EU 868MHz band (uplink: 868.0-868.6MHz), and 5G-A uses the n78 band (3.3-3.8GHz);

[0074] Switching strategy: When the data volume D>1MB or real-time requirement T delay When the transmission time is less than 10ms, it will automatically switch to 5G-A transmission.

[0075] Anti-interference measures: Frequency hopping spread spectrum (FHSS) technology is used, and the LoRaWAN channel switching cycle is 1s.

[0076] c. Edge computing layer: deploy several edge computing nodes on the bridge to realize data preprocessing and real-time analysis.

[0077] Based on reinforcement learning (Q-learning), computing tasks are dynamically allocated between edge nodes and the cloud. The specific calculation formula is as follows:

[0078]

[0079] Among them, Q(s,a) is the Q value (state-action value function) of performing action a in the current state s, which represents the expected long-term return of taking this action in this state; α is the learning rate, r is the immediate reward, γ is the discount factor, s′ is the new state reached after performing action a, and a′ represents the possible actions that can be taken in the new state s′. All actions need to be traversed to find the maximum value.

[0080] At the same time, a long short-term memory network (LSTM) is used to learn historical data patterns (data source: three years of historical monitoring data of a suspension bridge, sampling frequency 100Hz; training set: 80% of the data is used for training, 20% for testing) to generate an adaptive warning threshold. The formula is as follows:

[0081]

[0082] in, is the current moment prediction value, σ is the activation function, W f 、W i is the weight matrix, is the input sequence.

[0083] d. Cloud computing layer: build a federated learning-driven big data analysis platform to achieve multi-source data fusion and long-term performance prediction.

[0084] Build a multi-bridge federated learning system to achieve collaborative analysis of cross-bridge data.

[0085] The federated digital twin framework is implemented through the following steps:

[0086] (1) Local model training:

[0087] First, a dynamic digital twin is constructed based on the fusion of the bridge finite element model (FEM) and the measured data. The formula is as follows:

[0088] M twin =α□M FEM +(1-α)□M data ;

[0089] Among them, M twin is the fused digital twin model, α is the fusion coefficient, which controls the weight of the finite element model and the measured data, M FEM is a bridge model based on finite element analysis, M data A bridge model constructed based on measured data.

[0090] (2) Local model optimization: The Adam optimizer is used to update the model parameters, and the learning rate dynamic adjustment strategy is as follows:

[0091]

[0092] Among them, η t is the current learning rate, η0 is the initial learning rate, β is the decay rate, and t is the number of training iterations.

[0093] (3) Secure aggregation: Differential privacy protection technology is used to perturb the local model parameters. The formula is as follows:

[0094]

[0095] in, is the perturbation parameter; w k are the original model parameters; N(0,σ 2 I) is Gaussian noise with a mean of 0; σ 2 I is the variance; I is the identity matrix; σ controls the noise intensity.

[0096] (4) Global model update: Aggregate the global model through the FedAvg algorithm. The formula is as follows:

[0097]

[0098] Among them, w global is the global model parameter, which is obtained by weighted average of all local model parameters; K is the number of bridges participating in federated learning; nk is the local data volume of the kth bridge; n is the total data volume of all bridges, are the local model parameters of the kth bridge.

[0099] The specific steps to build a long-term performance prediction system are as follows:

[0100] 1) Perform feature extraction, including time domain features, frequency domain features, and time-frequency domain features.

[0101] Time domain feature extraction: Calculate the mean, variance, peak and other statistical quantities of the data.

[0102] Frequency domain feature extraction: Obtain spectrum features through fast Fourier transform (FFT).

[0103] Time-frequency domain feature extraction: Apply wavelet transform to extract multi-scale time-frequency features.

[0104] 2) Build a prediction model

[0105] Based on the temporal convolutional network, causal convolution and dilated convolution are used to capture temporal dependencies. The formula is as follows:

[0106] y t =σ(W*x t +b);

[0107] Among them, yt is the output value at time t, * is the convolution operation, σ is the activation function, W is the convolution kernel weight matrix, xt is the input sequence at time t, and b is the bias term.

[0108] The self-attention mechanism is introduced to strengthen key features. The formula is as follows:

[0109]

[0110] Among them, Q is the displacement query matrix, K is the key matrix, V is the value matrix, dk is the key vector dimension, and softmax is the normalized exponential function.

[0111] e. Application layer: provides a visual monitoring interface and intelligent decision-making support, and realizes autonomous optimization of bridge maintenance strategies based on reinforcement learning.

[0112] Develop a 3D bridge model based on Unity3D, map sensor data (such as strain and displacement) in real time, and support multi-scale viewing (global to local).

[0113] Build a state-action-reward model to optimize maintenance resource allocation. The formula is as follows:

[0114]

[0115] Among them, μθ(s) is the policy function, Q μ (s,a) is the action value function

[0116] Based on the digital twin simulation results, the bridge load limit (such as vehicle weight and speed) is dynamically adjusted. The formula is as follows:

[0117]

[0118] Among them, L limit is the dynamic load limit, L design is the design load limit, ∈ actual is the measured strain, ∈ threshold is the warning threshold.

[0119] For example, a high-pier, long-span bridge in a mountainous area has a main pier height of 180m, a main span of 400m, and a design life of 100 years. The monitoring content includes: main pier inclination (±0.005°), main beam deflection (±10mm), anchor cable stress (±1% error) and ambient wind speed (0-30m / s).

[0120] Four quantum diamond NV color-center strain gauges were placed atop the main piers, three TENG self-powered displacement sensors were installed mid-span in the main beams, and eight fiber Bragg grating (FBG) strain gauges were installed in the anchorage areas. Two NVIDIA Jetson Xavier NX edge servers, each equipped with a Volta-based GPU (with a computing power of 21TOPS), were deployed. A federated learning system was constructed in conjunction with five surrounding bridges, using a local ResNet-18 model with a 24-hour global model iteration cycle. Deployed in a private cloud environment, the system comprises 10 computing nodes (each equipped with two Intel Xeon Gold 6248 CPUs, 384GB of memory, and an NVIDIA A100 GPU) with a storage capacity of 1PB. This deployment enables long-term or short-term health monitoring of the bridge.

[0121] In addition, for abnormal data detected by sensors, the intelligent sensing layer transmits the data directly to the application layer based on node sensors with transmission capabilities, and triggers an early warning; when an early warning is generated, maintenance personnel are required to confirm on-site to first confirm whether there is a sensor fault, and then further detection; when multiple consecutive sensors emit abnormal data, the highest early warning is triggered.

[0122] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A real-time bridge health monitoring system based on intelligent sensing and big data analysis, characterized by: It includes intelligent sensing layer, data transmission layer, edge computing layer, cloud computing layer and application layer; The intelligent sensing layer determines the monitoring section of the bridge through finite element analysis, optimizes sensor positions using an improved particle swarm algorithm, and adaptively deploys multimodal sensors on the bridge. The multimodal sensors include node sensors with data transmission capabilities. The data transmission layer performs data transmission based on the constructed low-power wide area network; The edge computing layer pre-processes the data and performs real-time analysis. Several edge computing layers are deployed on the bridge. The cloud computing layer, based on a built federated learning-driven big data analysis platform, integrates multi-source data and performs long-term performance predictions on bridges. The application layer provides a visual monitoring interface and intelligent decision support. The application layer also realizes autonomous optimization of bridge maintenance strategies based on reinforcement learning.

2. The real-time bridge health monitoring system based on intelligent sensing and big data analysis according to claim 1 is characterized in that: The particle swarm algorithm optimizes the sensor position, and its calculation formula is as follows: Among them, σ i (x) and ∈ i (x) are the stress and strain sensitivities of the i-th node, c i is the deployment cost, w1 and w2 are weight coefficients, C max For the budget constraint.

3. The real-time bridge health monitoring system based on intelligent sensing and big data analysis according to claim 1 is characterized in that: The data transmission layer adopts LoRaWAN and 5G-A fusion networking. LoRaWAN is used for low-frequency environmental data transmission, and 5G-A is used for high-frequency structural response data transmission. The calculation formula is as follows: Among them, T total is the total transmission delay, D LoRa 、D 5G is the amount of data, R LoRa 、R 5G is the transmission rate.

4. The real-time bridge health monitoring system based on intelligent sensing and big data analysis according to claim 3 is characterized in that: The vibration signal is compressed at three levels based on wavelet transform.

5. The real-time bridge health monitoring system based on intelligent sensing and big data analysis according to claim 1 is characterized in that: The edge computing layer implements dynamic allocation of computing tasks between edge nodes and the cloud based on reinforcement learning. The specific calculation formula is as follows: Among them, Q(s,a) is the Q value (state-action value function) of performing action a in the current state s, which represents the expected long-term return of taking this action in this state; α is the learning rate, r is the immediate reward, γ is the discount factor, s′ is the new state reached after performing action a, and a′ represents the possible actions that can be taken in the new state s′. All actions need to be traversed to find the maximum value.

6. The real-time bridge health monitoring system based on intelligent sensing and big data analysis according to claim 1 is characterized in that: The edge computing layer can also use long-term short-term memory networks to learn historical data patterns and generate adaptive warning thresholds. The formula is as follows: in, is the current moment prediction value, σ is the activation function, W f 、W i is the weight matrix, is the input sequence.

7. The real-time bridge health monitoring system based on intelligent sensing and big data analysis according to claim 1 is characterized in that: Building a digital twin framework for federated learning is achieved through the following steps: (1) Local model training: First, a dynamic digital twin is constructed based on the fusion of the bridge finite element model and the measured data. The formula is as follows: M twin =α□M FEM +(1-α)□M data ; Among them, M twin is the fused digital twin model, α is the fusion coefficient, which controls the weight of the finite element model and the measured data, M FEM is a bridge model based on finite element analysis, M data A bridge model constructed based on measured data; (2) Local model optimization: The Adam optimizer is used to update the model parameters, and the learning rate dynamic adjustment strategy is as follows: Among them, η t is the current learning rate, η0 is the initial learning rate, β is the decay rate, and t is the number of training iterations; (3) Secure aggregation: Differential privacy protection technology is used to perturb the local model parameters. The calculation formula is as follows: in, is the perturbation parameter; w k are the original model parameters; N(0,σ 2 I) is Gaussian noise with a mean of 0; σ 2 I is the variance; I is the identity matrix; σ controls the noise intensity; (4) Global model update: Aggregate the global model through the FedAvg algorithm. The formula is as follows: Among them, w global is the global model parameter, which is obtained by weighted average of all local model parameters; K is the number of bridges participating in federated learning; nk is the local data volume of the kth bridge; n is the total data volume of all bridges, are the local model parameters of the kth bridge.

8. The real-time bridge health monitoring system based on intelligent sensing and big data analysis according to claim 7 is characterized in that: The cloud computing layer needs to build a long-term performance prediction system. The specific steps are as follows: (1) Feature extraction, including time domain features, frequency domain features, and time-frequency domain features; (2) Construct a prediction model based on a temporal convolutional network, using causal convolution and dilated convolution to capture temporal dependencies. The formula is as follows: y t =σ(W*xx t +b); Where yt is the output value at time t, * is the convolution operation, σ is the activation function, W is the convolution kernel weight matrix, xt is the input sequence at time t, and b is the bias term; The self-attention mechanism is introduced to strengthen key features. The calculation formula is as follows: Among them, Q is the displacement query matrix, K is the key matrix, V is the value matrix, dk is the key vector dimension, and softmax is the normalized exponential function.

9. The real-time bridge health monitoring system based on intelligent sensing and big data analysis according to claim 1 is characterized in that: The application layer constructs a state-action-reward model to optimize the allocation of maintenance resources. The calculation is as follows: Among them, μθ(s) is the policy function, Q μ (s,a) is the action value function Based on the digital twin simulation results, the bridge load limit is dynamically adjusted according to the following formula: Among them, L limit is the dynamic load limit, L design is the design load limit, ∈ actual is the measured strain, ∈ threshold is the warning threshold.

10. The real-time bridge health monitoring system based on intelligent sensing and big data analysis according to any one of claims 1 to 9, characterized in that: In response to abnormal data detected by multimodal sensors, the intelligent sensing layer transmits the data directly to the application layer based on node sensors with transmission capabilities, and triggers an early warning. When an early warning is generated, maintenance personnel are required to confirm on-site whether there is a fault in the multimodal sensor before further testing. When multiple continuously arranged multimodal sensors emit abnormal data, an emergency warning is triggered.

Images