Expressway frame hydraulic monitoring control system based on internet of things

Abstract

The application relates to the technical field of control and discloses a highway frame hydraulic monitoring control system based on Internet of Things, which comprises a hydraulic dynamic monitoring module, a multi-dimensional stress analysis module, an abnormal area positioning module and a self-adaptive control strategy module. The hydraulic dynamic monitoring module collects real-time hydraulic data of key nodes based on Internet of Things sensors and outputs theoretical hydraulic values in combination with a hydraulic dynamic calculation model. The multi-dimensional stress analysis module performs multi-dimensional difference analysis on the time-domain fluctuation deviation, frequency energy deviation and pressure wave peak sequence similarity of the theoretical and measured hydraulic values, and generates a node-level stress difference matrix. The abnormal area positioning module inputs the stress difference matrix into a spatial topology analysis network, generates an abnormal probability distribution diagram of an abnormal pressure propagation path in combination with frame structure parameters and node position information, and positions an abnormal area. The self-adaptive control strategy module configures control parameters according to the abnormal probability distribution diagram, enables a high-frequency hydraulic monitoring mode for a high-probability abnormal area, and applies pressure disturbance testing to adjacent nodes. The hydraulic state of the frame is comprehensively sensed and self-adaptive control is realized.

Description

Technical Field

[0001] This invention relates to the field of control technology, specifically to a hydraulic monitoring and control system for highway structures based on the Internet of Things. Background Technology

[0002] As a key component of transportation infrastructure, highway structures endure multiple stresses over long periods, including vehicle loads, environmental erosion, and geological changes. Their structural stability directly impacts driving safety and the normal operation of the transportation network. Currently, hydraulic monitoring of highway structures primarily employs traditional single-point sensing technology, relying on periodic manual inspections or data collection from fixed monitoring points. This approach suffers from limitations in monitoring range and real-time performance.

[0003] Traditional monitoring methods struggle to cover all critical nodes of the structure, often only selecting a few typical locations for monitoring, resulting in a one-sided understanding of the overall hydraulic status of the structure. In terms of data processing, existing systems primarily focus on collecting and displaying single hydraulic values, lacking in-depth analysis of hydraulic data and failing to effectively identify structural stress anomalies behind hydraulic changes.

[0004] Due to the complex structure of highways and the close mechanical connections between their nodes, local hydraulic anomalies can trigger a chain reaction, making it difficult for traditional monitoring methods to quickly locate abnormal areas and trace the propagation path of anomalies. Furthermore, existing control systems often employ preset, fixed control strategies, failing to dynamically adjust control parameters based on real-time monitoring data, thus hindering precise protection and maintenance of the highway structure.

[0005] With the continuous growth of highway traffic flow and the increase in service life, the safety risks faced by the structure are constantly accumulating. Traditional monitoring and control methods are no longer able to meet the needs of modern transportation infrastructure safety management. There is an urgent need for an integrated solution that can achieve comprehensive perception, in-depth analysis, precise positioning and adaptive control. Summary of the Invention

[0006] The purpose of this invention is to provide an Internet of Things-based hydraulic monitoring and control system for highway structures to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides an Internet of Things-based hydraulic monitoring and control system for highway overpasses, the system comprising:

[0008] Hydraulic dynamic monitoring module: Based on IoT sensors, it collects hydraulic data of key nodes of highway structure in real time, including hydraulic value, hydraulic change rate and pressure fluctuation frequency, and outputs theoretical hydraulic value through hydraulic dynamic calculation model;

[0009] Multidimensional stress analysis module: Performs multidimensional difference analysis on the theoretical hydraulic value and the measured hydraulic value. The multidimensional difference analysis includes time domain fluctuation deviation, frequency domain energy shift and pressure peak sequence similarity, and generates a node-level stress difference matrix.

[0010] Anomaly area location module: Input the stress difference matrix into the spatial topology analysis network, combine the frame structure parameters and node location information to generate an anomaly probability distribution map of the abnormal pressure propagation path, and locate the abnormal frame physical area;

[0011] Adaptive control strategy module: Configure control parameters according to the anomaly probability distribution map, including enabling high-frequency hydraulic monitoring mode for high-probability anomaly areas and applying pressure disturbance test to adjacent nodes.

[0012] Preferably, the hydraulic dynamic monitoring module specifically includes:

[0013] Historical data feature mining: Multi-dimensional decomposition processing of historical hydraulic data of key nodes of highway structure, including using empirical mode decomposition to extract the energy ratio of steady-state and transient components of hydraulic waveform, establishing the correlation matrix between pressure fluctuation frequency and structure load through pressure fluctuation correlation analysis, and using dynamic time warping algorithm to align pressure peak sequence patterns under different working conditions.

[0014] Hydraulic dynamic calculation model construction: The processed historical hydraulic data is input into the hybrid prediction network. The hybrid prediction network includes an LSTM time series prediction unit based on the aging curve of the frame material to generate basic hydraulic prediction values, a fully connected network with embedded frequency domain attention mechanism to correct prediction deviations caused by pressure fluctuations, and a peak feature compensator to dynamically adjust prediction weights according to the real-time collected pressure peak sequence.

[0015] The distortion rate and phase jitter parameters of the hydraulic waveform, the frequency components and phase angle distribution of pressure fluctuations, the amplitude abrupt change gradient and time interval entropy value of the pressure peak sequence are captured synchronously by IoT sensors.

[0016] Theoretical value calculation: Input the real-time collected data into the hydraulic dynamic calculation model to obtain the theoretical hydraulic value.

[0017] Preferably, the calculation of the theoretical hydraulic value includes performing:

[0018] Based on adaptive filtering during the frame's working phase, measurement noise caused by ambient temperature and humidity is eliminated;

[0019] The correlation features of hydraulic pressure, pressure fluctuation, and peak sequence are integrated through a spatiotemporal feature fusion algorithm.

[0020] The output includes the theoretical hydraulic pressure value within the normal operating condition fluctuation range, and the theoretical hydraulic pressure value is dynamically updated according to the aging state of the frame.

[0021] Preferably, the adaptive control strategy module specifically includes:

[0022] Obtain the stress difference value of each node in the stress difference matrix, and determine the initial control parameters based on the relationship between the stress difference value and the preset first stress difference threshold and second stress difference threshold;

[0023] When the stress difference value is less than or equal to the second stress difference threshold, or greater than or equal to the first stress difference threshold, the preset control parameter is determined as the initial control parameter.

[0024] When the stress difference value is greater than the second stress difference threshold and less than the first stress difference threshold, the average value of the first stress difference threshold and the second stress difference threshold is obtained, an adjustment coefficient is determined according to the ratio of the stress difference value to the average value, and the preset control parameter adjusted according to the adjustment coefficient is used as the initial control parameter.

[0025] Wherein, the first stress difference threshold is less than the second stress difference threshold.

[0026] Preferably, it also includes an inspection path planning module, used to formulate inspection paths based on the importance and anomaly levels of the frame nodes, including:

[0027] The importance level of the frame node is used as the first selection factor, the abnormality level of the frame node is used as the second selection factor, and the distance between the inspection starting point and the frame node is used as the third selection factor.

[0028] The inspection starts from the starting point. If there are multiple structural nodes of equal importance around the starting point, the structural node with the higher abnormality level will be selected first for inspection.

[0029] If the importance level and abnormality level of the frame nodes are equal, then select the frame node closest to the starting point to begin the inspection;

[0030] When the inspection reaches a frame node, if the node is connected to only one node to be inspected, then the node to be inspected is inspected directly.

[0031] If the node is connected to two or more nodes to be inspected, the node with the higher importance level will be inspected first. When the importance levels of multiple nodes to be inspected are equal, the node with the higher anomaly level will be inspected first. When the importance level and the anomaly level are equal, a node to be inspected will be randomly selected for inspection.

[0032] If the node is not connected to other nodes to be inspected, and there are uninspected frame nodes, then the nearest uninspected frame node will be inspected first.

[0033] Once all frame nodes have been inspected, return to the starting point;

[0034] The inspection path is obtained according to the above rules.

[0035] Preferably, it also includes a data transmission priority module for determining the priority of data transmission, including:

[0036] Identify data acquisition nodes, analyze the type and degree of abnormality of the hydraulic data acquired by the nodes, filter out acquisition nodes containing abnormal data, determine the range of data nodes that need to be prioritized for transmission, and obtain a list of abnormal nodes;

[0037] Based on the list of abnormal nodes, the data distribution among the affected nodes is analyzed, the degree of data correlation and interaction frequency between nodes are calculated, the impact range and priority of data transmission are determined, and a data transmission priority index is generated.

[0038] This includes calculating the degree of data association and interaction frequency between nodes, including obtaining the data flow and request frequency of nodes in the data dimension, calculating the data interaction frequency based on the data flow and request frequency between nodes, and determining the impact range and priority of data transmission based on the data interaction frequency.

[0039] Preferably, the system further includes a feedback adjustment module, used to detect the adjusted frame hydraulic state data through Kalman filtering, and to feed back the processed hydraulic parameters to the hydraulic dynamic monitoring module and the adaptive control strategy module to adjust the control strategy, including:

[0040] Collect hydraulic status data of key nodes of the frame after adjustment;

[0041] The collected hydraulic state data is processed by Kalman filtering to obtain the processed hydraulic parameter data;

[0042] The processed hydraulic parameter data is fed back to the hydraulic dynamic monitoring module and the adaptive control strategy module to identify the changing trend and periodic fluctuation of the hydraulic parameters and obtain the hydraulic parameter range.

[0043] The hydraulic parameter range is compared with the expected range, and the control strategy is adjusted based on the comparison results.

[0044] Preferably, the system further includes a security response module, used to determine abnormal frame nodes based on the abnormal location results, adjust the monitoring parameters of the nodes, allocate control resource ratios, update monitoring and control methods, and generate security control measures.

[0045] Preferably, the system further includes an execution control module, used to convert the control parameters generated by the adaptive control strategy module into execution instructions, specifically including:

[0046] Receive the control parameters configured by the adaptive control strategy module, and divide the control region hierarchy according to the control parameter type;

[0047] Match the hydraulic actuators corresponding to each control area level to generate an execution command chain that includes pressure adjustment range, response timing and action cycle;

[0048] The execution command chain is sent to the regional hydraulic execution terminal through the Internet of Things communication protocol, driving the actuator to perform pressure regulation operation according to the command chain parameters;

[0049] The execution control module interacts with the abnormal region location module to obtain updated data of the abnormal probability distribution map in real time and dynamically adjust the execution cycle of the execution instruction chain.

[0050] Preferably, the system further includes a comprehensive feedback module, used to calculate the frame health index based on hydraulic monitoring data and control strategy execution results, and adjust the monitoring frequency and control intensity of the entire system according to the health index to form closed-loop control.

[0051] Compared with the prior art, the beneficial effects of the present invention are:

[0052] This IoT-based highway frame hydraulic monitoring and control system uses a hydraulic dynamic monitoring module to collect hydraulic data from key nodes in real time based on IoT sensors. Combined with a hydraulic dynamic calculation model, it outputs theoretical hydraulic values, changing the situation of incomplete data collection and lack of theoretical analysis in traditional monitoring. This allows for a more comprehensive understanding of the frame's hydraulic state and a more accurate representation of the actual structural stress conditions.

[0053] The multidimensional stress analysis module performs multidimensional difference analysis between theoretical and measured hydraulic values, covering time-domain fluctuation deviation, frequency-domain energy shift, and pressure peak sequence similarity, generating a node-level stress difference matrix. This breaks through the limitations of traditional analysis that only focuses on single-dimensional differences, and can mine the structural stress information hidden in hydraulic data from multiple perspectives, making the identification of structural stress anomalies more accurate.

[0054] The abnormal area location module inputs the stress difference matrix into the spatial topology analysis network, combines the frame structure parameters and node location information to generate an abnormal probability distribution map of the abnormal pressure propagation path and locate the abnormal frame physical area. This solves the problem of difficulty in quickly locating abnormal areas and tracing propagation paths in traditional methods, and can promptly lock down areas with potential safety hazards, providing direction for subsequent maintenance work.

[0055] The adaptive control strategy module configures control parameters based on the anomaly probability distribution map, activates high-frequency hydraulic monitoring mode for high-probability anomaly areas, and applies pressure disturbance tests to adjacent nodes. It breaks free from the constraints of traditional fixed control strategies and can dynamically adjust the control mode according to the anomalies monitored in real time, making the protection of the frame structure more targeted and timely. This helps to optimize the allocation of maintenance resources while ensuring the stability of the frame structure. Attached Figure Description

[0056] Figure 1 This is a schematic diagram illustrating the working principle of the Internet of Things-based highway frame hydraulic monitoring and control system described in this invention.

[0057] Figure 2 A schematic diagram illustrating the working principle of the abnormal area location module;

[0058] Figure 3 This is a schematic diagram of the working principle of the adaptive control strategy module;

[0059] Figure 4 This is a schematic diagram illustrating the working principle of the data transmission priority module. Detailed Implementation

[0060] 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 some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0061] Please see Figure 1 and Figure 2 This invention provides an Internet of Things-based hydraulic monitoring and control system for highway overpasses, the system comprising:

[0062] The hydraulic dynamic monitoring module uses an IoT sensor network to collect real-time hydraulic parameters of key nodes in the frame, including dynamic indicators such as hydraulic pressure value, hydraulic pressure change rate, and pressure fluctuation frequency. The collected data is input into a hydraulic dynamic calculation model, which outputs theoretical hydraulic values. This model integrates material aging characteristics and operating conditions. A multi-dimensional stress analysis module performs time-domain fluctuation deviation, frequency-domain energy shift, and pressure peak sequence similarity analysis between theoretical and measured values, generating a node-level stress difference matrix. An anomaly location module, combined with the frame's spatial topology, converts the stress difference matrix into an anomaly probability distribution map, accurately identifying anomaly propagation paths. An adaptive control strategy module dynamically configures monitoring frequency and pressure disturbance parameters based on anomaly distribution characteristics, forming a closed-loop control system.

[0063] The structural topology of the framework is a node-connection network constructed based on the CAD model and finite element model of the highway framework design, including: ① Geometric topology: 3D coordinates of nodes (with the origin at the framework start point, X-axis extending along the highway, Y-axis perpendicular to the highway, and Z-axis vertical), node spacing (2m for beam nodes, 3m for column nodes), and node type (load-bearing nodes are marked C, connecting nodes are marked J); ② Physical connection topology: connection method (bolted connection is marked B, welding is marked W), and connection stiffness parameters (bolted connection stiffness). Welded connection stiffness ); ③ Mechanical topology: force transmission path (directed edge marking load-bearing node → beam node → column node), load transfer coefficient (node To the node Transmission coefficient ,like This means that 80% of the hydraulic load at node A is transferred to node B.

[0064] The spatial topology analysis network structure adopts a hybrid network of graph neural network (GNN) and finite element mechanical constraints. The structure includes: ① Input layer: stress difference matrix (dimensions). , The total number of nodes, with elements representing time-domain, frequency-domain, and peak similarity weighted difference values ​​(weights of 0.4, 0.3, and 0.3 respectively), and the adjacency matrix (dimension...). , Represents a node and ① Load transfer relationship); ② Hidden layer: 2 convolutional layers (convolution kernel) Activation function ReLU), 1 fully connected layer (number of neurons) ); ③ Output layer: node anomaly probability (range [0,1], activation function Sigmoid).

[0065] Transformation steps and calculation logic: Step 1: Node association mapping: Map the nodes in the stress difference matrix Difference value , and nodes in the adjacency matrix Connection node Association, forming a set Step 2: Anomaly Probability Correction: Obtain the initial probability using GNN. Corrected by incorporating mechanical constraints: ( For the transmission coefficient, For the difference value of the connecting nodes, the correction logic is: the greater the difference between the connecting nodes, the higher the probability of the current node being abnormal); Step 3: Distribution map generation: based on the 3D model of the frame, according to... Mapped colors ( green, yellow, (Red) The red arrow marks the abnormal propagation path (high probability node → connection node with the largest propagation coefficient).

[0066] The time-domain fluctuation deviation is calculated using the root mean square error (RMSE), with a time window of 30 seconds (sampling frequency 1Hz, number of sampling points). ),formula:

[0067]

[0068] ( for Real-time measured hydraulic pressure value (unit: MPa) for Theoretical hydraulic pressure value at any given time (unit: MPa)).

[0069] Frequency domain energy shift calculation: Step 1: Frequency domain transformation, for and Perform a Fast Fourier Transform (FFT) to obtain the frequency domain sequence. , (Frequency range 0-10Hz); Step 2: Energy calculation, calculate the energy of the characteristic frequency band (1-5Hz): Step 3: Offset quantization (Relative deviation reflects the difference in energy distribution in the frequency domain).

[0070] Pressure peak sequence similarity calculation: Step 1: Peak extraction, threshold set to... ,when and At that time, it was determined to be a peak, forming a measured sequence. Theoretical sequence Step 2: Similarity calculation using the Dynamic Time Warping (DTW) algorithm to calculate the sum of path distances. Normalized similarity: ( The similarity ranges from [0,1], with the closest similarity being 1.

[0071] Stress difference matrix generation: matrix dimensions, ( The number of key nodes is represented by three columns, each corresponding to a different key node. , , ); matrix element, the first OK (MPa) (%) .

[0072] Example 1: Specific implementation of the hydraulic dynamic monitoring module, which consists of two main stages: historical data feature mining and real-time calculation. Historical data processing employs a multi-level decomposition architecture. First, signal decomposition technology is used to decompose the original hydraulic waveform into multiple components with different frequency characteristics. The energy distribution characteristics of each component are calculated, and a feature database describing the steady-state and transient behavior of the hydraulic system is established. In the correlation analysis stage, statistical methods are used to study the relationship between pressure fluctuation frequency and external load changes, constructing a multi-dimensional feature matrix that reflects the dynamic characteristics of the system. For pressure waveform changes under different operating conditions, a sequence matching algorithm is used to extract typical waveform features, forming a comparable reference template library.

[0073] The hydraulic dynamic calculation model adopts a composite network structure design, comprising three main parts: a time-series prediction unit, a frequency domain analysis unit, and a feature compensation unit. The time-series prediction unit uses a neural network architecture with memory capabilities. The input layer receives preprocessed historical operating data, and through a multi-layer network structure, it extracts long-term dependencies in the time series, outputting basic predicted values. The frequency domain analysis unit is designed with a network structure capable of dynamic weight adjustment, automatically adjusting the attention weights of different frequency components based on real-time acquired frequency domain features. The feature compensation unit employs pattern recognition technology, generating corresponding compensation coefficients by comparing the similarity between real-time acquired peak sequence features and a pre-stored template library.

[0074] The theoretical value calculation process employs an embedded processing architecture. The system first monitors environmental parameters in real time, dynamically adjusting filtering parameters based on collected temperature and humidity data to eliminate interference from environmental factors on the measurement results. In the feature fusion stage, a weighted integration method is used to process feature information from multiple dimensions, including hydraulic pressure, fluctuation frequency, and peak interval. The weighting coefficients are dynamically adjusted based on the real-time changes of each feature. The calculation process comprehensively considers the impact of long-term factors such as material aging and structural deformation on system characteristics, ultimately outputting a theoretical hydraulic pressure prediction result containing a confidence interval.

[0075] The historical data feature mining process employs a phased processing strategy. The first phase decomposes the original hydraulic waveform into multiple scales, extracting signal feature components at different time scales, calculating the energy distribution ratio of each component, and establishing quantitative indicators describing the system's steady-state and transient behavior. The second phase analyzes the correlation between pressure fluctuation characteristics and external load changes, constructing a feature matrix reflecting the system's dynamic characteristics using statistical methods, with the matrix dimensions dynamically adjusted according to system complexity. The third phase processes pressure waveform changes under different operating conditions, employing a sequence alignment algorithm to extract representative waveform feature patterns, forming a comparable set of reference templates.

[0076] The network structure design of the hydraulic dynamic calculation model emphasizes multi-feature fusion. The time-series prediction part employs network units with long short-term memory capabilities. The input layer receives denoised historical operating data, extracts feature patterns from the time series through multi-layer nonlinear transformations, and outputs basic predicted values. The frequency domain analysis part is designed with an adaptive network structure, capable of dynamically adjusting the attention distribution of different frequency components based on real-time frequency domain features. The feature compensation part uses pattern matching technology, comparing real-time waveform features with a template library through similarity calculations to generate corresponding correction coefficients.

[0077] The theoretical value calculation process is dynamically adjusted in real time. The system continuously monitors changes in environmental parameters and dynamically optimizes the filtering algorithm parameters based on real-time data from temperature and humidity sensors to effectively suppress measurement noise. In the multi-feature fusion stage, a weighted integration method based on an attention mechanism is designed to automatically adjust the contribution weights of different feature dimensions. The calculation process comprehensively considers the long-term effects of material performance degradation and cumulative structural deformation, and adapts to the slow changes in system characteristics by periodically updating model parameters, ultimately outputting theoretical prediction values ​​with probability distribution characteristics.

[0078] Real-time data processing employs a pipelined architecture. Raw data collected by the sensor network first enters a preprocessing stage for noise suppression and outlier removal. The processed data then enters three analysis channels in parallel: a time-domain analysis channel calculates the statistical characteristics of the hydraulic pressure value; a frequency-domain analysis channel extracts the spectral characteristics of pressure fluctuations; and a waveform analysis channel identifies the spatiotemporal patterns of wave peak sequences. The outputs from each channel are input to a feature fusion unit, which generates a comprehensive feature representation through a dynamic weight allocation mechanism. Finally, this representation is input into a prediction model to calculate the theoretical hydraulic pressure value.

[0079] The model update mechanism employs incremental learning. The system periodically adds new operational data to the training set, updating the model parameters without complete retraining. The update process considers the distribution differences between the old and new data, maintaining the model's adaptability through importance weighting. For slowly changing factors such as material aging, the system establishes a dedicated degradation model, predicting performance change trajectories through long-term trend analysis, and incorporating the prediction results into the parameter adjustment process of the main model.

[0080] The degradation model was established using a materials mechanics model + LSTM data-driven model, with the following steps: Step 1: Data acquisition, collecting material aging test data (Q345 steel, hydraulic load-bearing capacity decay data from 0-10 years at 25℃ and 60% humidity), and 5-year historical hydraulic data of the frame (sampling frequency 1 time / day); Step 2: Sub-model construction, mechanical sub-model (material aging curve). ( For service years, For initial load-bearing capacity, (Attenuation coefficient), output aging coefficient LSTM sub-model input: "Hydraulic data from the past year" ", output (Network structure: 128 neurons in the input layer, 2 hidden layers each with 64 neurons, 1 neuron in the output layer, activation function Linear); Step 3: Hybridization. (Weights are determined based on fitting error; the smaller the fitting error of the mechanical model, the higher the weight); Step 4 verification uses actual aging data from the past year to ensure... .

[0081] Prediction results are integrated into the main model (hydraulic dynamic calculation model): The LSTM time-series prediction unit in the main model of the object (outputs basic hydraulic prediction values) ); integration method, adjusting the basic forecast value, formula: ( (This is the final theoretical hydraulic value); update cycle: the degenerate model is retrained every 3 months, and the main model is updated synchronously. value.

[0082] The system implementation employs a distributed computing architecture. After sensor nodes complete data acquisition and preliminary processing, they upload the feature data to edge computing nodes for model inference. Lightweight model versions are deployed on the edge nodes, responsible for real-time prediction tasks. The central server maintains the complete model, periodically receives runtime data from the edge nodes, optimizes the model and updates parameters, and then distributes the updated model parameters back to the edge nodes. This architectural design ensures both the system's real-time responsiveness and continuous model optimization.

[0083] The anomaly detection mechanism employs a multi-level threshold strategy. The system calculates confidence intervals based on theoretical predictions and sets multiple levels of anomaly judgment thresholds. When a measured value exceeds the first-level threshold, an early warning is triggered, and the anomaly characteristics are recorded, but no immediate control action is taken. When the value exceeds the second-level threshold, a cause analysis process is initiated to identify possible sources of the anomaly. When the value exceeds the third-level threshold, an emergency control protocol is immediately triggered. This hierarchical strategy effectively balances system sensitivity and false alarm rate.

[0084] The data processing workflow employs time window management. The system maintains multiple time windows of varying lengths to capture characteristic changes at different time scales. Short windows detect instantaneous anomalies, medium windows analyze periodic fluctuations, and long windows track slow trend changes. The analysis results from each window are input into a comprehensive judgment module to generate a complete assessment of the system status. The window length is dynamically adjusted based on the system's operating status; under stable conditions, the window is appropriately extended to improve the signal-to-noise ratio, while under transitional conditions, the window is shortened to enhance response speed.

[0085] The system calibration mechanism employs multi-source data fusion. In addition to hydraulic sensor data, the system also integrates multi-modal monitoring data such as structural strain and vibration acceleration, improving the accuracy of condition assessment through data fusion technology. The calibration process establishes a correlation model between hydraulic parameters and other monitoring indicators. When inconsistencies occur, a cross-validation process is initiated to identify potential sensor faults or model biases and trigger corresponding correction measures.

[0086] The hardware implementation employs a modular design. The sensor module includes a pressure sensing unit, signal conditioning circuitry, and a communication interface, featuring self-diagnostic and temperature compensation functions. The data acquisition module supports multi-channel synchronous sampling, with configurable sampling rate and resolution. The processing module utilizes a heterogeneous computing architecture, combining general-purpose processors and dedicated accelerators to balance computational performance and power consumption requirements.

[0087] The software architecture employs a microservices design. Each functional module is encapsulated as an independent service, interacting through well-defined interfaces. Services communicate via an event-driven mechanism; state changes are broadcast through a message bus, and relevant services respond based on their subscription relationships. This architecture improves the system's scalability and maintainability, facilitating independent upgrades and replacements of functional modules.

[0088] The system deployment takes into account practical engineering constraints. The sensor placement scheme is based on finite element analysis results, with monitoring nodes set at critical structural points and stress concentration areas. The cabling scheme balances signal integrity and ease of construction, using standardized connectors to simplify installation and maintenance. The power supply scheme combines line power supply and energy harvesting technologies, configuring backup power supplies at critical nodes to ensure continuous system operation. The protective design meets engineering environmental requirements, including dustproof, waterproof, and corrosion-resistant measures.

[0089] Example 2: See Figure 3 The adaptive control strategy module is implemented by employing a dynamic threshold adjustment mechanism to achieve precise control of the highway frame hydraulic system. The system establishes a hierarchical response system, generating control parameter sets for different anomaly levels by quantitatively analyzing the deviation of stress differences from preset thresholds. The control strategy formulation process comprehensively considers material properties, structural response characteristics, and system stability requirements, resulting in a closed-loop control scheme with adaptive capabilities.

[0090] The stress difference threshold is set based on material mechanical properties and engineering experience. The system maintains two key threshold parameters, each corresponding to a different response level. The first threshold is directly related to the material's safe operating stress, reflecting the critical point at which the system enters a warning state. The second threshold, set as a multiple of the first threshold, marks the boundary where the system enters an emergency state. These two thresholds constitute a three-level response range, dividing the system state into a normal operating zone, a transition zone, and an emergency zone. The threshold parameters are periodically checked and adjusted based on actual test data of the frame structure to maintain consistency with the actual structural condition.

[0091] Stress difference value calculation (weighted fusion): Step 1: Index normalization, unifying the three dimensions of the index into the [0,1] interval: ,in, This represents the normalized time-domain fluctuation bias. This represents the original time-domain fluctuation deviation. This represents the normalized frequency domain energy shift. This represents the original frequency domain energy shift. The difference in the normalized peak sequence. This represents the similarity of the original peak sequences. 0.5 MPa is the maximum allowable time-domain deviation, and 20% is the maximum allowable frequency-domain offset. The reverse reaction is positively correlated with the degree of difference; step 2 weighted fusion yields the final stress difference value. (The weights are determined based on the degree of influence of the indicators on structural stress anomalies, with time-domain deviation having the greatest impact).

[0092] Threshold determination basis and value: First stress difference threshold Based on the determination of the elastic limit stress of the material, Q345 steel corresponds to... (Normal-Warning Boundary); Second Stress Difference Threshold Based on the determination of the yield strength stress of the material, the corresponding stress for Q345 steel is... (Early Warning - Emergency Boundary); Verify 100 sets of historical fault data, More than 95% of the time, the anomalies are real, ensuring the reliability of the threshold.

[0093] Quantitative analysis application rules: : Under normal conditions, use preset control parameters (monitoring frequency 1 time / minute); Warning status, control parameter adjustment coefficient Control parameter = preset parameter × ; Emergency state: Use preset emergency control parameters (monitoring frequency 1 time / second).

[0094] The control parameter generation algorithm employs a standardization method. The system first normalizes the stress differences calculated at each node to eliminate dimensional influences and ensure comparability of states across different nodes. The algorithm establishes a mapping between standard scores and control parameters, automatically selecting a preset parameter combination when the score falls within a specific range. Within the transition range, the system uses a continuous adjustment mechanism, generating smoothly changing control parameters by calculating the relative relationship between the current state and the baseline state. This design avoids system oscillations caused by abrupt parameter changes, achieving a smooth transition of the control state.

[0095] The parameter adjustment process incorporates a nonlinear interpolation method. Multiple feature points are established within the transition interval, defining a smooth parameter variation curve using these points. The interpolation function is chosen considering the system's response characteristics, employing a gentler rate of change near the threshold region and allowing for larger adjustment amplitudes in the intermediate region. The adjustment coefficient is calculated by comprehensively considering the current state value, historical trends, and the states of adjacent nodes to avoid unnecessary parameter adjustments caused by accidental fluctuations in isolated nodes. The parameter update cycle is set based on the system's dynamic characteristics, appropriately extending the update interval under stable operating conditions and shortening the response time during rapid change phases.

[0096] The control parameter set includes multiple adjustable dimensions. The monitoring frequency parameter is dynamically adjusted based on the anomaly probability, increasing data acquisition density in potentially risky areas and reducing the sampling rate in stable areas to conserve system resources. The pressure disturbance amplitude is matched to the structural response capability, with multiple adjustment levels set within the material's elastic range. The sampling period parameter is coordinated with the monitoring frequency to ensure the integrity and timeliness of data acquisition. Interlocking relationships are established between the parameters to avoid parameter conflicts caused by independent adjustments.

[0097] The emergency recalculation mechanism is designed as an event-triggered mode. The system continuously monitors the rate of change of node states. When the rate of change exceeds a preset limit, the regular calculation process is immediately interrupted, and an emergency assessment procedure is initiated. The recalculation process employs simplified models and optimized algorithms to shorten response time while ensuring the reliability of the results. The verification of emergency calculation results uses cross-validation methods, confirming the rationality of the calculation results through multi-faceted data consistency analysis. Verified control parameters are immediately issued for execution, shortening the duration of the abnormal state.

[0098] The control strategy is implemented using a distributed architecture. Parameter calculation units are deployed at control nodes in each region to enable localized execution of control decisions. The central node is responsible for maintaining the global strategy and threshold parameters, and periodically pushes updates to each region node. This architecture reduces the impact of communication latency on control timeliness while maintaining the consistency of the overall system strategy. Region nodes possess a certain degree of autonomous decision-making capability, and can maintain basic control functions based on local data in the event of communication interruptions or other abnormal situations.

[0099] The parameter adjustment process is logged in a complete operation log. The system records in detail the triggering conditions, calculation process, and execution results for each parameter change, forming a traceable control history. Log data is used for subsequent strategy optimization and system evaluation; analysis of historical operation records identifies areas for improvement in control strategies. Log management employs a circular storage mechanism, optimizing storage resource usage while ensuring data integrity.

[0100] The system establishes a control effectiveness evaluation mechanism. After each parameter adjustment, the system continuously monitors the response characteristics of the controlled nodes and evaluates the effectiveness of the control strategy by comparing the expected effect with the actual response. The evaluation results are fed back to the parameter generation algorithm for dynamically optimizing the control parameter mapping relationship. Statistical analysis of long-term operating data reveals the distribution patterns of optimal control parameters under different operating conditions, providing data support for strategy optimization.

[0101] The anomaly handling process employs a tiered response strategy. For different types of anomalies, the system predefines multiple contingency plans. General anomalies trigger adjustments to routine monitoring parameters, medium-risk anomalies initiate localized stress disturbance testing, and high-risk anomalies activate regional coordinated control strategies. Contingency plan selection is based on anomaly type identification results. The system uses pattern matching to categorize the current anomaly into predefined anomaly patterns and invokes the corresponding handling procedures.

[0102] The system maintenance module enables dynamic updates to the control strategy. As the system architecture ages and the external environment changes, the applicability of the existing control strategy is periodically evaluated. The update process employs an incremental learning approach, adjusting specific parameters while retaining the core strategy framework. Before major changes, simulation verification is performed, testing the control effect of the new strategy using historical data playback. Updates are only implemented after confirming safety and effectiveness.

[0103] At the hardware level, redundant design ensures reliability. Critical control nodes are equipped with backup computing units, seamlessly switching to the backup unit in case of primary unit failure. Important parameters are stored in multiple non-volatile memories to prevent data loss. Communication links employ multi-path transmission, automatically routing to backup channels in case of partial line failure. The power system design considers different power outage requirements, with uninterruptible power supplies (UPS) provided for critical equipment.

[0104] The software implementation prioritizes real-time performance and determinism. The control algorithm employs a time-deterministic coding style to ensure real-time requirements are met even in worst-case scenarios. Critical computational tasks are allocated fixed time slices to avoid uncertainties introduced by task scheduling. Interrupt handling utilizes a layered design, keeping high-priority interrupt service routines concise and efficient. Memory management pre-allocates critical data structures to reduce uncertainties arising from dynamic memory allocation.

[0105] The human-machine interface design supports transparent management of control strategies. Operators can view the basis and effective range of current control parameter settings and intervene manually when necessary. The system provides parameter adjustment suggestion functions, recommending possible optimization directions based on the current system status. The historical status playback function supports traceability analysis of the control process, assisting in fault diagnosis and strategy evaluation.

[0106] System integration testing employs a layered verification approach. Unit tests verify the independent operational characteristics of each functional module, integration tests examine the interaction logic between modules, and system tests evaluate the overall control effectiveness. Test case design covers typical operating conditions and boundary conditions, with particular attention to the smoothness and stability of state transition processes. Test data includes laboratory simulation data and field-collected data to ensure the representativeness of test results.

[0107] An operation and maintenance mechanism ensures the system's sustainable operation. Regular health checks assess the operational status of each functional module, and preventative maintenance plans are developed based on equipment operating time and environmental conditions. A fault diagnosis system monitors equipment for abnormal signs in real time and provides early warnings of potential failures. Spare parts management employs intelligent predictive methods, predicting the remaining lifespan of critical components based on equipment aging models and optimizing spare parts inventory.

[0108] Example 3: The inspection path planning module employs a multi-objective optimization algorithm. The importance level of frame nodes is divided into 5 levels based on structural mechanics analysis results, and the anomaly level is divided into 4 levels based on the stress difference matrix. The path planning engine maintains three priority queues: the critical node queue is arranged in descending order of importance level, the anomaly node queue is arranged in descending order of anomaly level, and the adjacent node queue is arranged in ascending order of spatial distance. The inspection robot uses decision tree logic to select target nodes: first, it checks whether there are nodes with a critical level ≥ 4 within a 15-meter radius of the current node; if so, it prioritizes inspection; second, it checks nodes with anomaly level ≥ 3; when multiple candidate nodes have the same conditions, the Dijkstra algorithm is used to calculate the shortest reachable path. For path selection at branch nodes, the system calculates the cumulative risk value of each branch in real time (importance level × anomaly level / distance), and selects the branch with the highest risk value for priority inspection. After completing the inspection of the current area, the system uses A... The algorithm plans the optimal path to the nearest uninspected node, with path weights taking into account factors such as ground flatness, obstacle distribution, and signal strength. Example 3 describes the implementation of the inspection path planning module, which uses a multi-dimensional decision algorithm to determine the optimal inspection route for the highway frame hydraulic monitoring system. The system establishes a triple evaluation system based on node importance, anomaly degree, and spatial location, and employs a dynamic priority queue to manage nodes to be inspected, achieving reasonable allocation of inspection resources and scientific path planning. The path generation process integrates graph theory algorithms and heuristic rules to optimize inspection efficiency while meeting full coverage requirements.

[0109] The node assessment system employs a hierarchical quantification method. Importance levels are divided into five levels based on structural mechanics analysis results, with higher values ​​indicating a more critical role for the node in structural safety. Anomaly levels are based on stress difference matrix calculations and are categorized into four severity levels, reflecting the node's immediate health status. Spatial location information is recorded as three-dimensional coordinates and preprocessed into a topological connectivity diagram, marking reachable paths and actual distances between nodes. These three dimensions of assessment data constitute the foundational input for path planning.

[0110] The path planning engine maintains three dynamic data structures. A priority queue for critical nodes is arranged in descending order of importance, ensuring that critical parts receive priority for inspection. An anomaly queue is sorted according to its anomaly level, enabling timely identification and handling of serious anomalies. A neighboring node queue stores spatial location information and uses a min-heap structure for fast retrieval of the nearest node. These queues are dynamically updated during the inspection process, reflecting completed inspections and newly added anomalies.

[0111] The inspection robot's decision-making logic employs a tree-like structure. The system first checks the node distribution within its current location's radius to determine if a high-priority target exists within the search area. If multiple candidate nodes exist, a comprehensive evaluation metric is used for selection.

[0112]

[0113] Where R represents the overall priority of the node, I is the importance level, A is the anomaly level, and D represents the Euclidean distance to the current node. α and β are weighting coefficients, set to 0.6 and 0.4 respectively, reflecting a balance between the emphasis on structural importance and the degree of anomaly. This calculation formula ensures that, when nodes are close in distance, the more important or more anomalous nodes are prioritized.

[0114] Branch node processing employs a risk accumulation assessment method. When an inspection reaches a node connecting multiple branches, the system calculates the cumulative risk value for each branch path. The calculation process involves accumulating the product of the importance level and the anomaly level for each node along the potential path, and then dividing by the path length. This assessment method considers both the criticality of nodes on the path and inspection efficiency, avoiding path detours caused by simply prioritizing a single point.

[0115] Inter-region transfers employ a heuristic search algorithm. After completing the inspection of the current region, the system plans the optimal path to the uninspected region using an improved A algorithm. The path cost function comprehensively considers factors such as actual distance, ground traffic conditions, and signal coverage quality. Based on the standard A algorithm, a dynamic weight adjustment mechanism is introduced, emphasizing distance factors in open areas and increasing the weight of traffic difficulty in complex environments, ensuring that the generated path is both theoretically optimal and engineering feasible.

[0116] A real-time route adjustment mechanism responds to changes in system status. When the monitoring system reports a new anomaly or an escalation of an anomaly, the inspection planning module immediately reassesses the current route. Anomaly information triggers queue priority updates, which may lead to the interruption or modification of currently executing routes. Adjustment decisions consider the balance between interruption costs and response benefits, avoiding efficiency losses caused by frequent changes. The system sets a minimum completion percentage threshold; the current route segment will only be abandoned when a new anomaly reaches a sufficiently severe level.

[0117] A communication delay compensation mechanism ensures the reliability of remote control. In areas with unstable wireless signals, the system predicts potential communication interruption periods and pre-caches critical commands and data. Route planning avoids known signal blind spots, or, if necessary, downloads detailed map data and inspection plans in advance. Command transmission employs redundant coding and acknowledgment retransmission mechanisms to ensure accurate delivery of control commands.

[0118] The inspection process is recorded using a spatiotemporal tagging method. The system comprehensively records the inspection time, operation content, and observation results for each node, and the data is stored in association with the 3D model of the structure. The recorded information includes text descriptions, numerical measurements, and multimedia evidence, forming a complete and traceable inspection archive. This data is used for subsequent trend analysis and system optimization, and also provides historical reference for anomaly diagnosis.

[0119] The anomaly handling process is deeply integrated with route planning. When an anomaly is confirmed during inspection, the system immediately assesses the resource requirements and time estimates for anomaly handling. If on-site repair is required, a route including the replenishment of necessary tools and materials is planned; if further inspection is required, a route for specialized inspection equipment is arranged. During the handling process, subsequent routes are dynamically adjusted, appropriately increasing the inspection priority of related nodes.

[0120] The human-computer interface supports visualized management of the inspection process. Operators can view the inspection progress, current path, and distribution of nodes to be inspected in real time. The system provides path suggestions and modification tools, allowing experienced engineers to adjust automatically generated routes. An emergency intervention function supports quick jumps to designated nodes to handle unexpected situations. The interface design takes into account the display characteristics of mobile devices, optimizing visibility and ease of operation in outdoor environments.

[0121] The system adopts a modular software architecture. The core path planning algorithm is encapsulated as an independent service, interacting with modules such as status monitoring and exception handling through well-defined interfaces. Computationally intensive tasks are deployed on edge servers to reduce the burden on central processing. Mobile terminals run lightweight clients, primarily responsible for data display and command reception, maintaining responsiveness.

[0122] The hardware platform design takes into account engineering environment requirements. The inspection robot is equipped with a multi-sensor system, including a positioning module, environmental monitoring equipment, and a dedicated flaw detection instrument for the frame. The mobile chassis adapts to different road conditions and has a certain obstacle-crossing and hill-climbing ability. The communication system supports multiple wireless protocols and automatically selects the optimal connection method based on site conditions. The power system meets the requirements for continuous operation and is equipped with fast charging and backup battery solutions.

[0123] Testing and verification employed a combined virtual and real-world approach. In a laboratory environment, digital twin technology was used to simulate various structural configurations and anomaly distributions to verify the algorithm's coverage and efficiency. Field testing involved actual inspections of typical sections to verify the system's performance in a real-world environment. Test data included metrics such as path execution time, anomaly detection rate, and ease of operation, which were used to continuously optimize algorithm parameters.

[0124] A maintenance and update mechanism ensures the long-term effectiveness of the system. As the system architecture changes and monitoring needs adjust, the system regularly updates node importance and connection relationship data. Algorithm parameters are fine-tuned based on accumulated inspection experience to adapt to changes in actual engineering conditions. Software components support remote upgrades, facilitating functional expansion and problem fixing. Hardware maintenance plans are developed based on usage intensity and environmental impact to prevent equipment performance degradation.

[0125] Example 4: See Figure 4 The collaborative working mechanism of the data transmission priority module and the feedback adjustment module is demonstrated through a specific example, showcasing the data processing flow and control strategy optimization process of the system in actual operation. An anomaly was detected by the overpass monitoring system in the section from K23+500 to K25+800 of a certain highway, and the system initiated the complete data transmission and feedback adjustment process.

[0126] During the abnormal node identification phase, the system sets a 30-second sliding detection window to monitor the deviation between the node's hydraulic data and the theoretical value. When node A-7 shows a deviation exceeding the threshold of 15% for three consecutive sampling cycles, it is marked as a primary abnormal node. Subsequently, the system detects that adjacent nodes A-6 and A-8 successively exhibit similar anomalies in the following two cycles, forming an abnormal node cluster. The system generates a list of abnormal nodes containing the following information:

[0127] Table 1: Example of an abnormal node list generated by the system containing the following information.

[0128]

[0129] The data transmission priority calculation process employs a modified PageRank algorithm. The system first analyzes the topological relationships between nodes, determining that node A-7 is located on the critical path, affecting data parsing of the three downstream nodes. The algorithm assigns A-7 a high initial weight of 0.45, while the edge node B-4 has an initial weight of 0.15. The data freshness coefficient is calculated based on the most recent data update time, with a higher coefficient for more recent updates. The final priority score considers three factors: node weight, anomaly severity, and downstream impact range, determining that node A-7 has the highest data transmission priority and must complete transmission within 50ms.

[0130] The network scheduler divides the data stream into three channels based on priority scoring. The real-time streaming channel is allocated to the complete dataset for nodes A-7, containing raw waveforms, feature extraction results, and intermediate anomaly diagnostic data. The priority packet channel transmits compressed feature data from nodes A-6 and A-8, preserving core anomaly features while reducing data volume. The general data channel processes routine monitoring data from nodes B-4 using a batch transmission mode. This hierarchical transmission mechanism ensures that critical anomaly data arrives at the analysis center first, supporting rapid decision-making.

[0131] The feedback adjustment module initiates a dual Kalman filter processing flow. The first filter processes the raw sensor data, employing a dedicated noise model designed for the pulse interference unique to hydraulic systems, effectively suppressing measurement noise while preserving anomalous features. The second filter analyzes the residual sequence, identifying a persistent negative deviation trend at node A-7, which is judged as a potential structural anomaly. The filter result update cycle is set to 60 seconds, with each update including complete parameter verification and model validation steps.

[0132] The control strategy optimization employs a state-based reinforcement learning framework. The system encodes the current frame state as a 128-dimensional feature vector, containing parameters such as pressure values, fluctuation characteristics, and temperature compensation coefficients for each node. The action space is defined as a combination of three dimensions: monitoring frequency adjustment, pressure disturbance amplitude, and sampling period. The reward function design considers a balance between anomaly recovery speed, energy efficiency, and system stability. The learning process adopts an offline training and online fine-tuning mode, first training the basic policy on historical data, and then adaptively adjusting it based on real-time feedback.

[0133] During implementation, upon detecting an anomaly at node A-7, the system immediately increased the monitoring frequency of that node to three times the normal value, while simultaneously reducing the sampling rate of non-critical nodes in adjacent areas to balance the system load. Stress disturbance testing employed a stepped increase strategy, with the initial amplitude set at 5% of the normal operating pressure, subsequently gradually increasing to 15% based on the response. The sampling period was dynamically matched to the monitoring frequency; a shorter period was used during high-speed sampling periods to ensure data continuity, while the period was appropriately extended during regular monitoring periods to conserve resources.

[0134] The anomaly diagnosis process incorporates a multi-evidence fusion method. The system cross-compares hydraulic data, strain gauge readings, and accelerometer signals, revealing a slight change in vibration frequency at node A-7 when the anomaly occurred. The diagnostic engine searches its historical case database, finding a high probability of loose bolts corresponding to similar feature combinations. The system automatically generates inspection suggestions, prompting a focused inspection of the tightness of the connection at this node.

[0135] The communication quality monitoring module evaluates data transmission efficiency in real time. The system records metrics such as data transmission success rate, latency, and packet loss rate for each node. When a decline in transmission quality is detected at a C-segment node with a weak wireless signal, the system automatically switches to a backup communication frequency band and adjusts the data packetization strategy, increasing error correction coding overhead to improve transmission reliability. For nodes experiencing persistent communication difficulties, the system activates a local caching mechanism to prioritize the transmission of backlogged critical data once communication is restored.

[0136] The human-computer interface presents multi-level system status information. Operators can view a global anomaly distribution heatmap to quickly locate problem areas. After drilling down to a specific node, detailed multi-dimensional data analysis results and diagnostic suggestions are displayed. The control strategy adjustment interface provides parameter suggestion sliders, allowing engineers to fine-tune the parameters based on the system's recommended values. All operations are logged in a complete audit log, recording the initiator, time, and reason for parameter modifications.

[0137] The system maintenance module implements preventative health management. Regular self-test procedures verify the operational status of each functional module, including sensor calibration checks, communication link testing, and computing resource monitoring. Maintenance plans are intelligently generated based on equipment runtime and environmental conditions, increasing the frequency of electrical component checks during hot and humid seasons and focusing on battery performance in low-temperature environments. Fault prediction models analyze equipment aging trends and provide early warnings of potential hardware problems.

[0138] The version update mechanism supports continuous system evolution. Minor algorithm optimizations are implemented through parameter configuration updates, allowing deployment without system downtime. Major feature upgrades employ a canary release strategy, first testing on select nodes to verify stability before full rollout. The update rollback function retains multiple historical versions, enabling rapid revert to a stable version in case of compatibility issues.

[0139] In the implementation case, the system successfully identified a loose flange connection bolt fault at node A-7 using the aforementioned mechanism. The entire process, from initial anomaly detection to accurate problem localization, took less than 8 minutes. During this time, the system automatically adjusted its monitoring strategy, prioritizing the real-time transmission of critical data while optimizing resource allocation to avoid system overload. Continuous calibration of the feedback adjustment module ensured the accuracy of the diagnostic data, ultimately guiding maintenance personnel to efficiently handle the fault. Throughout the process, monitoring functions in non-critical areas remained operational, and no global performance degradation occurred due to local anomalies.

[0140] Example 5: The collaborative working mechanism of the safety response module, execution control module, and integrated feedback module constitutes the closed-loop management architecture of the highway frame hydraulic monitoring and control system. The system manages the abnormal response process through a state machine model, transforms control strategies into executable instructions, and dynamically optimizes overall control parameters based on feedback data, realizing a complete closed loop from abnormal detection to handling and effect evaluation.

[0141] The safety response module uses a finite state machine model to manage abnormal nodes. The system defines three core states: normal state (corresponding to routine operating conditions where hydraulic parameters fluctuate within permissible ranges), warning state (indicating parameters exceed thresholds but have not yet reached dangerous levels), and emergency state (indicating the system has detected a significant anomaly that may endanger structural safety). State transition conditions are based on a probabilistic risk assessment model; when the anomaly probability exceeds a preset threshold, state transition is automatically triggered. In the warning state, the system activates an enhanced monitoring mode, increasing the sampling frequency to three times the base value and shortening the data transmission interval to one-fifth of the original value. Upon entering the emergency state, the system activates a regional linkage mechanism, adjusting not only the parameters of the abnormal node itself but also reconfiguring the monitoring strategies and control logic of adjacent nodes to form a collaborative response network.

[0142] The execution control module accurately translates control strategies into execution instructions. The system employs layered compilation technology to process control parameters. The high-level strategy describes abstract control objectives, such as "stabilizing node pressure" or "suppressing abnormal fluctuations." The middle layer decomposes the strategy into a sequence of time-series actions, including basic operations such as valve opening and closing, pump speed adjustment, and delay waiting. The bottom layer generates specific control signals, determining the operating parameters and timing relationships of the actuators. The instruction chain verification process uses a formal method to check the completeness of the timing logic and the safety of parameter combinations, eliminating instruction combinations that may lead to system conflicts or resource contention. Verified instruction chains are distributed to regional execution terminals via industrial IoT protocols, ensuring accurate synchronization of control actions.

[0143] A dynamic adjustment mechanism is introduced during execution. The system receives real-time updates of the probability distribution map from the anomaly area location module and adjusts the command parameters based on the anomaly propagation trend. For rapidly spreading anomaly patterns, the control cycle is shortened and the adjustment amplitude is increased; for locally stable anomalies, a gradual adjustment strategy is adopted to avoid system oscillation. Execution effect monitoring data is fed back to the control strategy generation module, forming a closed-loop optimization circuit. Execution records meticulously store the expected goals and actual effects of each operation, providing case support for subsequent strategy optimization.

[0144] The integrated feedback module performs system-level health status assessments. The assessment model considers sixteen characteristic parameters, covering multiple dimensions such as hydraulic stability, control response characteristics, and energy efficiency. Parameter weights are dynamically adjusted based on the frame's service life, emphasizing performance indicators in the newly constructed stage and strengthening structural safety considerations in the aging stage. The health index is calculated using a random forest algorithm to handle the nonlinear relationships between different parameters, outputting a standardized score from 0 to 100. The score results intuitively reflect the overall system status and guide the adjustment direction of the global control strategy.

[0145] The health index is directly related to the system's operating mode. When the index is above 80, the system enters energy-saving optimization mode, appropriately extending the monitoring interval and reducing the sampling frequency of non-critical nodes to reduce energy consumption while ensuring basic monitoring needs are met. When the index is between 60 and 80, a standard monitoring strategy is adopted to maintain normal control intensity. When the index is below 60, an enhanced control mode is activated to improve the response speed of actuators, increase backup power preparation, and, if necessary, call upon redundant control resources to deal with potential risks. A buffer zone is set between mode switching intervals to avoid frequent state jumps near critical values.

[0146] The system implementation adopts a distributed event-driven architecture. Each module is encapsulated as an independent service, interacting loosely through a message bus. An event publish-subscribe mechanism ensures that state changes are propagated to relevant modules in a timely manner; for example, an anomaly detection event can simultaneously trigger multiple services such as security response, execution control, and data logging. Service discovery and load balancing mechanisms automatically allocate computing resources, concentrating resources to ensure core functions during periods of high anomaly incidence, and allocating resources to support background analysis tasks during normal periods.

[0147] The hardware design prioritizes reliability and fault tolerance. Critical control nodes employ a dual-processor redundancy design, maintaining state synchronization between primary and backup units for seamless failover. Execution terminals are equipped with local caches, allowing continued execution of pre-stored instruction sequences even during network interruptions. The power system features multi-level backup solutions, including uninterruptible power supplies, fast-switching backup circuits, and dedicated batteries for critical equipment. Communication links implement multi-path redundancy, automatically selecting the optimal transmission channel.

[0148] The human-computer interface supports multi-role collaboration. On-site operators view a concise status overview and operation instructions for rapid response to emergencies. Technical engineers access detailed parameters and analysis tools for in-depth diagnostics and strategy fine-tuning. Management personnel obtain comprehensive reports and trend charts to understand the system's long-term operational status. The interface design follows context-aware principles, automatically adapting the display layout and operation methods to different environments and devices.

[0149] The system maintenance mechanism enables full lifecycle management. The configuration management database records the versions and parameters of all hardware devices and software components, supporting rapid fault location and recovery. Predictive maintenance models analyze equipment aging trends and schedule preventative maintenance before performance reaches critical levels. The knowledge base system accumulates anomaly handling cases and optimization experience, using semantic retrieval to assist decision-making. Software updates support incremental releases and canary rollouts, minimizing the impact of upgrades on system continuity.

[0150] Testing and verification employ a full-scenario coverage strategy. Unit tests verify the basic functionality of each service, integration tests check the consistency of interfaces between modules, and system tests evaluate the effectiveness of the overall control flow. Test case design considers typical operating conditions, boundary conditions, and fault scenarios, with particular attention to the timing characteristics of anomaly propagation and control response. Stress tests simulate system performance under high load conditions to verify the reliability of the resource scheduling algorithm.

[0151] In a typical application scenario, the system detects that a hydraulic parameter at a certain node continuously deviates from the theoretical value. The safety response module assesses the anomaly probability as reaching 0.75, triggering a warning state. The execution control module generates an adjustment command consisting of three steps: first, opening the bypass valve to divert some pressure; second, adjusting the main pump speed to compensate for the flow change; and finally, confirming the effect after a 300-millisecond delay. The comprehensive feedback module monitors the system response during the adjustment process, calculating that the health index slightly increases from 72 to 78. After confirming the effectiveness of the control strategy, it continues to maintain the enhanced monitoring mode for 2 hours until the system is fully stable, at which point it resumes normal operation. The entire process is completed automatically; operators only need to monitor the system's suggestions and confirm critical operations.

[0152] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0153] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An expressway frame hydraulic monitoring control system based on Internet of Things, characterized in that, The method comprises the following steps: Hydraulic dynamic monitoring module: based on the Internet of Things sensor, real-time acquisition of the hydraulic data of the key nodes of the highway frame body, including the hydraulic value, the hydraulic change rate and the pressure fluctuation frequency, and output of the theoretical hydraulic value through the hydraulic dynamic calculation model; Multi-dimensional stress analysis module: multi-dimensional difference analysis of the theoretical hydraulic value and the measured hydraulic value, including time domain fluctuation deviation, frequency domain energy deviation and pressure wave peak sequence similarity, and generation of a stress difference matrix at the node level; Abnormal area positioning module: input of the stress difference matrix into a spatial topology analysis network, combination of the frame body structure parameters and the node position information, generation of an abnormal probability distribution map of the abnormal pressure propagation path, and positioning of the abnormal frame body physical area; Adaptive control strategy module: configuration of control parameters according to the abnormal probability distribution map, including enabling a high-frequency hydraulic monitoring mode for a high-probability abnormal area and applying a pressure disturbance test to adjacent nodes; The hydraulic dynamic monitoring module specifically comprises: Historical data feature mining: multi-dimensional decomposition processing of the historical hydraulic data of the key nodes of the highway frame body, including extraction of the energy proportion of the steady-state component and the transient component of the hydraulic waveform by using the empirical mode decomposition, establishment of a correlation matrix of the pressure fluctuation frequency and the frame body load by pressure fluctuation correlation analysis, and alignment of the pressure wave peak sequence mode under different working conditions by using the dynamic time warping algorithm; Hydraulic dynamic calculation model construction: input of the processed historical hydraulic data into a hybrid prediction network, the hybrid prediction network comprising an LSTM time series prediction unit based on the aging curve of the frame body material, a fully connected network embedded with a frequency domain attention mechanism, and a wave peak feature compensator for dynamically adjusting the prediction weight according to the real-time collected pressure wave peak sequence; Synchronous capture of the distortion rate and phase jitter parameters of the hydraulic waveform, the frequency component and phase angle distribution of the pressure fluctuation, and the amplitude mutation gradient and time interval entropy value of the pressure wave peak sequence by the Internet of Things sensor; Theoretical value calculation: input of the real-time collected data into the hydraulic dynamic calculation model to obtain the theoretical hydraulic value.

2. The highway structure hydraulic monitoring control system based on Internet of Things according to claim 1, characterized in that, In the calculation of the theoretical hydraulic value, the following steps are performed: Adaptive filtering processing based on the working stage of the frame body to eliminate measurement noise caused by environmental temperature and humidity; Integration of the correlation features of hydraulic pressure, pressure fluctuation and wave peak sequence by a spatiotemporal feature fusion algorithm; Output of the theoretical hydraulic value including the normal working condition fluctuation interval, and dynamic updating of the theoretical hydraulic value according to the aging state of the frame body.

3. The highway structure hydraulic monitoring control system based on Internet of Things according to claim 1, characterized in that, The adaptive control strategy module specifically comprises: Obtaining the stress difference values of each node in the stress difference matrix, and determining the initial control parameters according to the relationship between the stress difference values and the first stress difference threshold and the second stress difference threshold; When the stress difference value is less than or equal to the first stress difference threshold, the preset control parameter is used as the initial control parameter; When the stress difference value is greater than or equal to the second stress difference threshold, the preset emergency control parameter is used as the initial control parameter; When the stress difference value is greater than the first stress difference threshold and less than the second stress difference threshold, an average value of the first stress difference threshold and the second stress difference threshold is obtained, an adjustment coefficient is determined according to a ratio of the stress difference value to the average value, and a preset control parameter adjusted according to the adjustment coefficient is taken as an initial control parameter; The first stress difference threshold is less than the second stress difference threshold.

4. The highway structure hydraulic monitoring control system based on Internet of Things according to claim 1, characterized in that, The inspection path planning module is further configured to plan an inspection path according to the importance levels and the abnormality levels of the rack nodes, and the inspection path planning module comprises: The importance level of the rack node is taken as a first selection element, the abnormality level of the rack node is taken as a second selection element, and a distance between the inspection starting point and the rack node is taken as a third selection element; The inspection starts from the starting point, if there are multiple rack nodes with equal importance levels around the starting point, a rack node with a higher abnormality level is preferentially selected to start the inspection; If the importance level and the abnormality level of the rack node are equal, a rack node closest to the starting point is selected to start the inspection; When the inspection reaches a rack node, if the rack node is connected to only one to-be-inspected node, the to-be-inspected node is directly inspected; If the rack node is connected to two or more to-be-inspected nodes, a to-be-inspected node with a higher importance level is preferentially inspected, when the importance levels of multiple to-be-inspected nodes are equal, a to-be-inspected node with a higher abnormality level is preferentially inspected, and when the importance level and the abnormality level are equal, a to-be-inspected node is randomly selected for inspection; If the rack node is not connected to other to-be-inspected nodes and there are un-inspected rack nodes, a rack node closest to the un-inspected rack node is preferentially inspected; When all the rack nodes are inspected, the starting point is returned to; The inspection path is obtained according to the above rules.

5. The highway structure hydraulic monitoring control system based on Internet of Things according to claim 1, characterized in that, The data transmission priority module is further configured to determine a priority of data transmission, and the data transmission priority module comprises: The data acquisition node is identified, the type and the abnormality degree of the hydraulic data collected by the node are analyzed, the acquisition node containing abnormal data is screened, the range of the data node requiring priority transmission is determined, and an abnormal node list is obtained; Based on the abnormal node list, the data distribution in the affected node is analyzed, the data correlation degree and the interaction frequency between nodes are calculated, the influence range and the priority order of data transmission are determined, and a data transmission priority index is generated; The data correlation degree and the interaction frequency between nodes are calculated, including obtaining the data flow and the request frequency of the node in the data dimension, calculating the data interaction frequency according to the data flow and the request frequency between nodes, and determining the influence range and the priority order of data transmission according to the data interaction frequency.

6. The Internet of Things based highway pylon hydraulic monitoring control system according to claim 1, wherein, The feedback adjustment module is further configured to detect the adjusted hydraulic state data of the rack node by Kalman filtering, and feed back the processed hydraulic parameters to the hydraulic dynamic monitoring module and the adaptive control strategy module to adjust the control strategy, and the feedback adjustment module comprises: The hydraulic state data of the adjusted key node of the rack is collected; The collected hydraulic state data is processed by Kalman filtering to obtain processed hydraulic parameter data; The processed hydraulic parameter data is fed back to the hydraulic dynamic monitoring module and the adaptive control strategy module to identify the change trend and the periodic fluctuation of the hydraulic parameter, and obtain a hydraulic parameter range. The hydraulic parameter range and the expected range are compared, and the control strategy is adjusted according to the comparison result.

7. The IoT based highway pylon hydraulic monitoring control system as claimed in claim 1, wherein, The safety response module is further included for judging the abnormal rack body node according to the abnormal positioning result, adjusting the monitoring parameter of the node, distributing the control resource proportion, updating the monitoring control mode, and generating the safety control measure.

8. The Internet of Things based highway pylon hydraulic monitoring control system as claimed in claim 1, wherein, The execution control module is further included for converting the control parameter generated by the adaptive control strategy module into an execution instruction, specifically including: receiving the control parameter configured by the adaptive control strategy module, and dividing the control area level according to the control parameter type; matching the hydraulic execution mechanism corresponding to each control area level to generate an execution instruction chain containing the pressure regulation amplitude, response time sequence and action period; downloading the execution instruction chain to the regional hydraulic execution terminal through the Internet of Things communication protocol to drive the execution mechanism to perform the pressure regulation operation according to the instruction chain parameter; wherein the execution control module interacts with the abnormal area positioning module to obtain the update data of the abnormal probability distribution graph in real time and dynamically adjust the action period of the execution instruction chain.

9. The Internet of Things based highway pylon hydraulic monitoring control system as claimed in claim 1, wherein, The comprehensive feedback module is further included for calculating the rack health index according to the hydraulic monitoring data and the control strategy execution result, adjusting the monitoring frequency and the control strength of the whole system according to the health index, and forming a closed-loop control.

Images