Gate three-dimensional intelligent monitoring device and method based on multi-source heterogeneous signal fusion

Abstract

The application discloses a kind of gate three-dimensional intelligent monitoring device and method based on multi-source heterogeneous signal fusion, it is related to water conservancy engineering safety monitoring technical field, the device includes industrial computer, double acquisition card component, sensor module, modular power supply system, man-machine interaction module and three-dimensional intelligent analysis module;Sensor module is used to collect the strain signal, vibration signal, acoustic emission signal and inclination angle signal of the gate to be measured;Double acquisition card component includes first acquisition card and second acquisition card, for receiving strain signal and vibration signal and acoustic emission signal;Three-dimensional intelligent analysis module is used for three-dimensional dynamic modeling and intelligent fault diagnosis according to strain signal, vibration signal, acoustic emission signal and inclination angle signal;Man-machine interaction module is used for man-machine interaction and real-time display.The application can realize multi-source heterogeneous signal high-precision synchronous acquisition, fusion analysis and three-dimensional intelligent diagnosis, improve the accuracy and efficiency of gate monitoring and fault diagnosis.

Description

Technical Field

[0001] This application relates to the field of water conservancy project safety monitoring technology, and in particular to a three-dimensional intelligent monitoring device and method for gates based on multi-source heterogeneous signal fusion. Background Technology

[0002] The working gates of flood discharge systems are critical equipment in hydropower projects. However, accidents such as support arm fractures and structural deformation not only cause huge economic losses but also threaten the lives and property of downstream residents. Therefore, online safety monitoring of metal structure equipment is an important means to solve this problem. However, most existing monitoring systems adopt a distributed sensor combined with PLC architecture, resulting in severe system fragmentation. The independent operation of multiple devices leads to synchronization errors in the data, making it impossible to accurately correlate multi-source heterogeneous signals such as stress, vibration, and acoustic emission, and failing to achieve high-precision synchronous acquisition of multi-source heterogeneous signals. At the same time, existing monitoring systems lack fusion analysis and three-dimensional intelligent diagnostic capabilities, failing to construct a three-dimensional dynamic model of the gate's operating status and failing to achieve accurate intelligent fault diagnosis. The accuracy and efficiency of gate monitoring and fault diagnosis are low.

[0003] Therefore, there is an urgent need for a monitoring device and method that can achieve high-precision synchronous acquisition, fusion analysis, and three-dimensional intelligent diagnosis of multi-source heterogeneous signals to meet the demand for high-precision and high-efficiency online monitoring of gate metal structure equipment. Summary of the Invention

[0004] The purpose of this application is to provide a gate three-dimensional intelligent monitoring device and method based on multi-source heterogeneous signal fusion, which realizes high-precision synchronous acquisition, fusion analysis and three-dimensional intelligent diagnosis of multi-source heterogeneous signals, and can improve the accuracy and efficiency of gate monitoring and fault diagnosis.

[0005] To achieve the above objectives, this application provides the following solution.

[0006] In the first aspect, this application provides a gate three-dimensional intelligent monitoring device based on multi-source heterogeneous signal fusion, including an industrial control computer, a dual acquisition card assembly, a sensor module, a modular power supply system, a human-machine interaction module, and a three-dimensional intelligent analysis module.

[0007] The industrial control computer is connected to the dual data acquisition card assembly, the human-machine interaction module, the modular power supply system, and the three-dimensional intelligent analysis module. The industrial control computer is used to perform data reception, processing, storage, and coordinated control of each module. The modular power supply system is used to supply power to each module.

[0008] The sensor module is installed on the gate under test. The sensor module includes a strain sensor, a vibration sensor, an acoustic emission sensor, and a tilt sensor, which are used to collect the strain signal, vibration signal, acoustic emission signal, and tilt angle signal of the gate under test, respectively.

[0009] The dual acquisition card assembly includes a first acquisition card and a second acquisition card; the first acquisition card is connected to the strain sensor and the vibration sensor respectively, and the second acquisition card is connected to the acoustic emission sensor; the first acquisition card is used to receive the strain signal and the vibration signal and transmit them to the industrial control computer, and the second acquisition card is used to receive the acoustic emission signal and transmit it to the industrial control computer.

[0010] The tilt sensor is connected to the industrial control computer, and the tilt sensor is also used to transmit the tilt angle signal to the industrial control computer.

[0011] The three-dimensional intelligent analysis module is used to perform three-dimensional dynamic modeling and intelligent fault diagnosis based on the strain signal, the vibration signal, the acoustic emission signal and the tilt angle signal, so as to obtain the three-dimensional dynamic model and fault diagnosis results of the gate under test.

[0012] The human-computer interaction module is used to interact with the user and display the monitoring data of the gate under test, the three-dimensional dynamic model, and the fault diagnosis results in real time.

[0013] Optionally, the first acquisition card is a PCI interface acquisition card, adapted for the acquisition of 28 strain signals and 33 vibration signals; the second acquisition card is a PCIe interface acquisition card, adapted for the acquisition of 6 acoustic emission signals; the first acquisition card and the second acquisition card adopt clock synchronization technology, with a synchronization error ≤0.05ms.

[0014] Optionally, the strain sensor is a foil strain gauge with a sensitivity coefficient of 2.0±1%; the vibration sensor is a piezoelectric accelerometer with a measurement range of ±50g; the acoustic emission sensor is a resonant sensor with a center frequency of 150kHz; and the tilt sensor has a measurement accuracy of ±0.1°.

[0015] Optionally, the three-dimensional intelligent analysis module includes a preprocessing unit, a feature extraction unit, a three-dimensional dynamic modeling unit, a fault intelligent diagnosis unit, and a result output unit.

[0016] The preprocessing unit is used to standardize, differentiate denoising, and repair anomalies in the strain signal, vibration signal, acoustic emission signal, and tilt angle signal to obtain preprocessed data.

[0017] The feature extraction unit is used to extract time-domain, frequency-domain, and spatiotemporal correlation features from the preprocessed data and perform dimensionality reduction optimization to obtain core input features.

[0018] The three-dimensional dynamic modeling unit is used to construct the finite element basic model of the gate under test based on the core input features. Through data-driven methods, it realizes static calibration of the model, dynamic attitude update and real-time calculation of stress and deformation, and completes visualization rendering to obtain a three-dimensional dynamic model.

[0019] The intelligent fault diagnosis unit is used to input the core input features into the fault diagnosis model to identify the fault type, locate the fault, assess the severity and predict the trend of the gate under test, and obtain the fault diagnosis result.

[0020] The result output unit is used to output the three-dimensional dynamic model and the fault diagnosis results to the human-computer interaction module.

[0021] Optionally, during differential denoising, the preprocessing unit applies 50Hz power frequency notch filtering and db4 wavelet 3-level decomposition threshold denoising to the strain signal, 8kHz low-pass filtering and singular value decomposition denoising to the vibration signal, 100kHz-200kHz bandpass filtering and energy threshold denoising to the acoustic emission signal, and 5-point moving average filtering to the tilt angle signal.

[0022] Optionally, during three-dimensional dynamic modeling, the three-dimensional dynamic modeling unit performs static calibration based on strain data, adjusts boundary conditions to ensure strain deviation ≤5%; updates the model space attitude based on the Euler angle rotation formula of tilt data with an update delay ≤0.5s; calculates global stress and deformation based on vibration-strain fusion data, combined with Hooke's law and finite element algorithm; identifies defects based on acoustic emission data, refines the mesh in the defect area, and corrects material parameters.

[0023] Optionally, the fault intelligent diagnosis unit constructs a fault feature library containing stress concentration, microcracks, structural deformation, and sensor faults during fault intelligent diagnosis; adopts a hybrid model based on random forest and SVM as the fault diagnosis model, and uses cosine similarity matching to determine the fault; and evaluates the severity of the fault based on the evaluation function.

[0024] Secondly, this application proposes a gate three-dimensional intelligent monitoring method based on multi-source heterogeneous signal fusion. The gate three-dimensional intelligent monitoring method based on multi-source heterogeneous signal fusion is implemented based on the gate three-dimensional intelligent monitoring device based on multi-source heterogeneous signal fusion as described in any of the first aspects. The gate three-dimensional intelligent monitoring method based on multi-source heterogeneous signal fusion includes the following steps.

[0025] The strain signal, vibration signal, acoustic emission signal and tilt angle signal of the gate under test are collected.

[0026] Based on the strain signal, vibration signal, acoustic emission signal and tilt angle signal, three-dimensional dynamic modeling and intelligent fault diagnosis are performed to obtain the three-dimensional dynamic model and fault diagnosis results of the gate under test.

[0027] The system displays the monitoring data of the gate under test, the three-dimensional dynamic model, and the fault diagnosis results in real time.

[0028] Optionally, three-dimensional dynamic modeling and intelligent fault diagnosis are performed based on the strain signal, the vibration signal, the acoustic emission signal, and the tilt angle signal to obtain the three-dimensional dynamic model and fault diagnosis results of the gate under test, specifically including the following steps.

[0029] The strain signal, vibration signal, acoustic emission signal, and tilt angle signal are standardized, differentially denoised, and anomaly repaired to obtain preprocessed data.

[0030] The core input features are obtained by extracting time-domain, frequency-domain, and spatiotemporal correlation features from the preprocessed data and performing dimensionality reduction optimization.

[0031] Based on the core input features, a finite element model of the gate under test is constructed. Through data-driven methods, static calibration, dynamic attitude update, and real-time calculation of stress and deformation of the model are realized, and visualization rendering is completed to obtain a three-dimensional dynamic model.

[0032] The core input features are input into the fault diagnosis model to identify the fault type, locate the fault, assess the severity, and predict the trend of the gate under test, thereby obtaining the fault diagnosis result.

[0033] Optionally, the fault diagnosis result includes a fault severity level, which includes mild fault, moderate fault, and severe fault.

[0034] After performing three-dimensional dynamic modeling and intelligent fault diagnosis based on the strain signal, the vibration signal, the acoustic emission signal and the tilt angle signal to obtain the three-dimensional dynamic model and fault diagnosis results of the gate under test, the gate three-dimensional intelligent monitoring method based on multi-source heterogeneous signal fusion further includes: generating hierarchical operation and maintenance decision suggestions.

[0035] The tiered operation and maintenance decision-making recommendations include: continuous monitoring for minor faults, dedicated maintenance within a preset time for moderate faults, and immediate shutdown and maintenance for severe faults.

[0036] According to the specific embodiments provided in this application, this application has the following technical effects.

[0037] This application provides a three-dimensional intelligent monitoring device and method for gates based on multi-source heterogeneous signal fusion. The device includes an industrial control computer, a dual acquisition card assembly, a sensor module, a modular power supply system, a human-machine interface module, and a three-dimensional intelligent analysis module. It integrates four types of sensors: strain, vibration, acoustic emission, and tilt angle, enabling simultaneous acquisition of comprehensive data on gate stress, vibration characteristics, defect initiation, and attitude changes. The dual acquisition card assembly includes a first acquisition card and a second acquisition card, with clearly defined functions. The first acquisition card receives low-to-medium speed strain and vibration signals, while the second acquisition card receives high-frequency acoustic emission signals. The tilt angle signal is directly connected to the industrial control computer, thus enabling high-precision synchronous acquisition of multi-source heterogeneous signals, eliminating monitoring blind spots, ensuring efficient transmission of various data types, and avoiding the limitations of single-signal monitoring. Furthermore, this application utilizes the three-dimensional intelligent analysis module to perform three-dimensional dynamic modeling and intelligent fault diagnosis based on strain, vibration, acoustic emission, and tilt angle signals. It can construct a three-dimensional dynamic model of the gate under test based on the aforementioned multi-source heterogeneous signals, intuitively presenting key states such as stress distribution and structural deformation. At the same time, it can automatically complete the intelligent fault diagnosis of the gate under test based on the above-mentioned multi-source heterogeneous signals, which greatly reduces the dependence on the experience of operation and maintenance personnel and improves the accuracy and efficiency of gate monitoring and fault diagnosis. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0039] Figure 1 This is a schematic diagram of the overall structure of a gate three-dimensional intelligent monitoring device based on multi-source heterogeneous signal fusion, provided as an embodiment of this application.

[0040] Figure 2 This is a schematic diagram of the external structure of a chassis provided in an embodiment of this application.

[0041] Figure 3 This is a schematic diagram of the internal structure of a chassis provided in an embodiment of this application.

[0042] Figure 4 This is a schematic diagram of the internal wiring of a chassis provided in an embodiment of this application.

[0043] Figure 5 This is a flowchart illustrating a three-dimensional intelligent monitoring method for gates based on multi-source heterogeneous signal fusion, provided as an embodiment of this application.

[0044] Figure labeling: 1-Industrial computer; 2-Temperature and humidity sensor; 3-Control device; 4-220V double-layer terminal; 5-Guide rail socket; 6-2P circuit breaker; 7-6-position grounding bar; 8-24V power supply; 9-28V power supply; 10-Acoustic emission preamplifier; 11-Cable trough; 12-Air inlet; 13-Terminal block; 14-220V cable inlet; 15-Heater; 16-Sensor cable inlet; 17-100P linear board; 18-68P adapter board; 19-4520 serial port expansion module; 20-Strain sensor amplifier; 21-Adapter; 22-Exhaust fan; 23-Capacitive touch screen. Detailed Implementation

[0045] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0046] The purpose of this application is to provide a gate three-dimensional intelligent monitoring device and method based on multi-source heterogeneous signal fusion, which aims to achieve high-precision synchronous acquisition, fusion analysis and three-dimensional intelligent diagnosis of multi-source heterogeneous signals, improve the accuracy and efficiency of gate monitoring and fault diagnosis, and solve the problems of large synchronous error in multi-source heterogeneous signal acquisition and insufficient multi-source heterogeneous signal fusion analysis and three-dimensional intelligent diagnosis functions in existing monitoring systems.

[0047] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0048] like Figure 1 As shown in the figure, this application provides a gate three-dimensional intelligent monitoring device based on multi-source heterogeneous signal fusion. The device mainly includes an industrial control computer 1, a dual acquisition card assembly, a sensor module, a modular power supply system, a human-machine interaction module, and a three-dimensional intelligent analysis module.

[0049] The industrial control computer 1 is connected to the dual data acquisition card assembly, the human-machine interaction module, the modular power supply system, and the three-dimensional intelligent analysis module. The industrial control computer 1 is used to perform data reception, processing, storage, and coordinated control of each module. The modular power supply system is used to supply power to each module.

[0050] The sensor module is installed on the gate under test. The sensor module includes a strain sensor, a vibration sensor, an acoustic emission sensor, and a tilt sensor, which are used to collect the strain signal, vibration signal, acoustic emission signal, and tilt angle signal of the gate under test, respectively.

[0051] The dual acquisition card assembly includes a first acquisition card and a second acquisition card; the first acquisition card is connected to the strain sensor and the vibration sensor respectively, and the second acquisition card is connected to the acoustic emission sensor; the first acquisition card is used to receive the strain signal and the vibration signal and transmit them to the industrial control computer 1, and the second acquisition card is used to receive the acoustic emission signal and transmit it to the industrial control computer 1.

[0052] The tilt sensor is connected to the industrial computer 1, and the tilt sensor is also used to transmit the tilt angle signal to the industrial computer 1.

[0053] The three-dimensional intelligent analysis module is used to perform three-dimensional dynamic modeling and intelligent fault diagnosis based on the strain signal, the vibration signal, the acoustic emission signal and the tilt angle signal, so as to obtain the three-dimensional dynamic model and fault diagnosis results of the gate under test.

[0054] The human-machine interface module is used for user interaction and to display the monitoring data, 3D dynamic model, and fault diagnosis results of the gate under test in real time. The module allows for real-time display of monitoring data, the 3D dynamic model, and fault diagnosis results, supports data querying and export, and facilitates maintenance personnel's rapid understanding of the gate's operating status. This design provides quantitative basis for maintenance decisions, enabling a shift from passive maintenance to proactive early warning.

[0055] As an optional implementation, the gate 3D intelligent monitoring device based on multi-source heterogeneous signal fusion also includes a chassis, preferably a wall-mounted chassis, which integrates an industrial control computer, dual data acquisition card components, a modular power supply system, a human-machine interaction module, and a 3D intelligent analysis module. Multiple strain sensors, vibration sensors, and acoustic emission sensors are deployed on key structural parts of the gate body under test. The strain sensors cover all key stress areas of the gate, capturing stress distribution and changes; the vibration sensors cover different elevations and planar areas of the gate, capturing full-modal vibration characteristics; and the acoustic emission sensors focus on high-risk areas prone to crack initiation, capturing acoustic emission signals of defect propagation. Multiple tilt sensors are also present, all mounted on the rigid main beam of the gate, accurately measuring the overall opening and closing angle and attitude of the gate. The strain sensor and vibration sensor are connected to the first acquisition card inside the chassis via wired or wireless connection, the acoustic emission sensor is connected to the second acquisition card inside the chassis via wired or wireless connection, and the tilt sensor is connected to the industrial control computer 1 inside the chassis via wired or wireless connection, so as to realize high-precision synchronous acquisition of multi-source heterogeneous signals.

[0056] This application embodiment integrates various functional modules inside the chassis, resulting in a compact structure that is more suitable for installation conditions with high humidity and limited space in the gate chamber compared to traditional bulky monitoring systems. The modular power supply system has overload and short-circuit protection functions, which can prevent the spread of local faults and ensure long-term stable operation of the system.

[0057] As an optional implementation, the first acquisition card is a PCI interface acquisition card, which is compatible with the acquisition of 28 strain signals and 33 vibration signals; the second acquisition card is a PCIe interface acquisition card, which is compatible with the acquisition of 6 acoustic emission signals; the first acquisition card and the second acquisition card adopt clock synchronization technology, and the synchronization error is ≤0.05ms.

[0058] Most existing monitoring systems employ a distributed sensor architecture combined with a PLC. However, the PLC system typically has no more than 16 AI channels, which is insufficient to meet the monitoring requirements of devices such as arc gates that require 67 sensors (including 28 strain sensors, 33 vibration sensors, and 6 acoustic emission sensors). This application's embodiment addresses this by designing a dual-acquisition card assembly, comprising a first acquisition card and a second acquisition card. The first acquisition card receives low-to-medium speed strain and vibration signals, while the second acquisition card receives high-frequency acoustic emission signals. This clear division of labor satisfies the monitoring requirements of 67 sensors (including 28 strain sensors, 33 vibration sensors, and 6 acoustic emission sensors). Furthermore, the dual-acquisition card assembly utilizes high-precision clock synchronization technology, controlling the synchronization error of multi-source signals to ≤0.05ms, thus resolving the analytical bias problem caused by data asynchrony in traditional systems. This design can accurately capture the microscopic timing logic of "strain change - vibration response - acoustic emission signal," providing reliable data support for early fault identification.

[0059] As an optional implementation, the strain sensor is a foil strain gauge with a sensitivity coefficient of 2.0±1%; the vibration sensor is a piezoelectric accelerometer with a measurement range of ±50g; the acoustic emission sensor is a resonant sensor with a center frequency of 150kHz; and the tilt sensor has a measurement accuracy of ±0.1°.

[0060] As an optional implementation, the three-dimensional intelligent analysis module includes a preprocessing unit, a feature extraction unit, a three-dimensional dynamic modeling unit, a fault intelligent diagnosis unit, and a result output unit.

[0061] The preprocessing unit is used to standardize, differentiate denoising, and repair anomalies in the strain signal, vibration signal, acoustic emission signal, and tilt angle signal to obtain preprocessed data.

[0062] The feature extraction unit is used to extract time-domain, frequency-domain, and spatiotemporal correlation features from the preprocessed data and perform dimensionality reduction optimization to obtain core input features.

[0063] The three-dimensional dynamic modeling unit is used to construct the finite element basic model of the gate under test based on the core input features. Through data-driven methods, it realizes static calibration of the model, dynamic attitude update and real-time calculation of stress and deformation, and completes visualization rendering to obtain a three-dimensional dynamic model.

[0064] The intelligent fault diagnosis unit is used to input the core input features into the fault diagnosis model to identify the fault type, locate the fault, assess the severity and predict the trend of the gate under test, and obtain the fault diagnosis result.

[0065] The result output unit is used to output the three-dimensional dynamic model and the fault diagnosis results to the human-computer interaction module.

[0066] As an optional implementation, the preprocessing unit, during differential denoising, applies a 50Hz power frequency notch filter and a db4 wavelet 3-level decomposition threshold denoising to the strain signal, an 8kHz low-pass filter and singular value decomposition denoising to the vibration signal, a 100kHz-200kHz bandpass filter and energy threshold denoising to the acoustic emission signal, and a 5-point moving average filter to the tilt angle signal.

[0067] As an optional implementation, the three-dimensional dynamic modeling unit, during three-dimensional dynamic modeling, adjusts boundary conditions to ensure strain deviation ≤5% based on static calibration of strain data; updates the model space attitude based on Euler angle rotation formula of tilt data with an update delay ≤0.5s; calculates global stress and deformation based on vibration-strain fusion data, combined with Hooke's law and finite element algorithm; identifies defects based on acoustic emission data, refines the mesh in defect areas, and corrects material parameters.

[0068] As an optional implementation, the fault intelligent diagnosis unit constructs a fault feature library containing stress concentration, microcracks, structural deformation, and sensor faults during fault intelligent diagnosis; it adopts a hybrid model based on random forest and SVM as the fault diagnosis model, and uses cosine similarity matching to achieve fault determination; and it evaluates the severity of the fault based on the evaluation function.

[0069] To make the technical solution of this application clearer, the device structure and monitoring process of this application will be described in detail below with examples.

[0070] This application proposes a gate three-dimensional intelligent monitoring device based on multi-source heterogeneous signal fusion, which mainly consists of a wall-mounted chassis, an industrial control computer 1, a dual acquisition card assembly (first acquisition card and second acquisition card), a sensor module, a modular power supply system, a human-machine interaction module, and a three-dimensional intelligent analysis module.

[0071] Wall-mounted chassis: Serving as the mounting platform for the entire device, it is made of high-strength metal materials, offering excellent dustproof and waterproof performance, and capable of adapting to harsh environments such as gate chambers. The internal components are rationally arranged to achieve a compact design, reducing the overall size of the device.

[0072] Industrial PC 1: An Advantech ITA3650E industrial PC 1 is selected as the core control unit of the entire monitoring system, responsible for data reception, processing, storage, and control and coordination of various modules.

[0073] In this embodiment, the dual data acquisition card components work together as follows.

[0074] The first acquisition card is preferably a PCI-1747U acquisition card, mainly used to process 61 channels of medium- and low-speed signals, including 28 channels of strain signals and 33 channels of vibration signals. This first acquisition card has high sampling accuracy and stability, and can meet the acquisition requirements of medium- and low-speed signals.

[0075] Second acquisition card: A PCIe-1812 acquisition card is preferred, specifically designed for acquiring 6 channels of high-frequency acoustic emission signals. This second acquisition card employs advanced signal acquisition technology, featuring a high sampling rate and excellent anti-interference capabilities, enabling accurate acquisition of high-frequency acoustic emission signals.

[0076] The sensor module includes strain sensors, vibration sensors, acoustic emission sensors, and tilt sensors, which are used to collect strain, vibration, acoustic emission, and tilt angle signals of the gate, providing comprehensive raw data for gate operation status assessment and 3D modeling. Specifically, the strain sensor uses a high-precision foil strain gauge with a sensitivity coefficient of 2.0±1%; the vibration sensor is a piezoelectric accelerometer with a measurement range of ±50g; the acoustic emission sensor is a resonant sensor with a center frequency of 150kHz; and the tilt sensor has a measurement accuracy of ±0.1°, ensuring the reliability of data acquisition across all dimensions.

[0077] Modular power supply system: Configured with 24V or 28V power supplies according to the power requirements of different modules, and equipped with AC30-10530 DIN rail socket 5 to provide 220V AC power to industrial computer 1. Each power module integrates overload protection, short circuit protection, and overvoltage protection functions. When a power supply abnormality occurs in a module, the power supply to that module can be quickly cut off to prevent the fault from spreading and ensure the overall stable operation of the system.

[0078] Human-Machine Interaction Module: A capacitive touchscreen 23, such as a 15-inch capacitive touchscreen 23, is installed on the upper center of the front of the chassis, supporting multi-touch operation. The interface can display real-time gate operating status parameters (strain, vibration, acoustic emission, temperature and humidity, etc.), a 3D model visualization, fault alarm information, etc. Users can perform parameter settings, historical data queries, and fault record export operations through the capacitive touchscreen 23, making the interaction convenient and intuitive.

[0079] The 3D intelligent analysis module, integrated into the industrial control computer software system, constructs a 3D dynamic model of the arc-shaped gate based on multi-source monitoring data (strain, vibration, acoustic emission, tilt angle, etc.) and employs finite element analysis and data fusion algorithms. This module can calculate key parameters of the gate structure in real time, such as stress distribution, deformation, and vibration modes, and present them intuitively through color rendering and animation. Simultaneously, it establishes a fault diagnosis model, combining historical fault data with real-time monitoring data to automatically identify, locate, and assess the severity of early-stage gate faults, generating diagnostic reports to support operational and maintenance decisions.

[0080] In this embodiment, the three-dimensional intelligent analysis module serves as the core unit. Relying on multi-source heterogeneous synchronous data, it achieves intelligent analysis of the gate's operating status through a closed-loop process of "multi-source data preprocessing - multi-dimensional feature extraction - three-dimensional dynamic modeling - intelligent fault diagnosis - result output and interaction". Specifically, it includes the following steps.

[0081] S1. Multi-source data preprocessing: The heterogeneous raw data collected by the dual acquisition card components are standardized, denoised differently, and anomaly repaired to construct a time-consistent three-dimensional data matrix and obtain preprocessed data.

[0082] S2. Multi-dimensional feature extraction: Extract time-domain, frequency-domain, and spatiotemporal correlation features from preprocessed data, and obtain core input features through dimensionality reduction optimization.

[0083] S3. 3D Dynamic Modeling: Based on the gate CAD model, a finite element basic model is constructed. Through data-driven methods, static calibration, dynamic attitude update, and real-time stress / deformation calculation of the model are realized, and visualization rendering is completed to obtain a 3D dynamic model.

[0084] S4. Intelligent Fault Diagnosis: Based on an offline-trained fault feature library and a hybrid diagnostic model, it realizes fault type identification, location positioning, severity assessment and trend prediction, and obtains fault diagnosis results.

[0085] S5. Results Output and Interaction: Real-time visualization of analysis results, support for data export, and generation of targeted operation and maintenance decision suggestions.

[0086] In this embodiment of the application, the multi-source data preprocessing in step S1 specifically includes the following steps.

[0087] S11. Data format standardization: Receives 28 strain signals, 33 vibration signals, 6 acoustic emission signals and 1 tilt signal, converts them into 64-bit floating-point type through data parsing algorithm, and binds them to the timestamp (1μs precision) based on the system's high-precision clock synchronization signal to construct a three-dimensional data matrix of "time-physical quantity-sensor number".

[0088] S12. Targeted Denoising and Enhancement: For strain signals, a processing method of "50Hz power frequency notch filtering + db4 wavelet 3-level decomposition threshold denoising" is used to suppress noise to ≤5με; for vibration signals, a processing method of "8kHz low-pass filtering + singular value decomposition denoising" is used to retain the first 3 principal singular values, making the noise ≤0.01g; for acoustic emission signals, a processing method of "100kHz-200kHz bandpass filtering + energy threshold denoising" is used to remove signals with energy <30dB; for tilt angle signals, a processing method of 5-point moving average filtering is used to ensure that the angle accuracy is stable at ±0.1°.

[0089] S13. Data Anomaly Detection and Repair: Based on the 3σ criterion, outliers exceeding [μ-3σ, μ+3σ] are identified. Single-point anomalies are repaired by linear interpolation of 5 valid data points before and after the point. If more than 3 consecutive anomalies occur, an alarm is triggered and a backup channel is switched or a historical trend prediction value is used as a substitute.

[0090] In this embodiment of the application, the multi-dimensional feature extraction in step S2 specifically includes the following steps.

[0091] S21. Time-domain feature extraction: Extract 8 features such as peak value and valley value from strain signal; extract 8 features such as peak value and effective value from vibration signal; extract 6 features such as event count from acoustic emission signal; extract 4 features such as mean angle from tilt signal.

[0092] S22. Frequency domain feature extraction: Perform a 1024-point Fast Fourier Transform (FFT) on the vibration signal to extract five features, including the fundamental frequency and harmonics; perform power spectral density (PSD) analysis on the acoustic emission signal to extract three features, including the peak frequency; and normalize all frequency domain features to the [0, 1] interval using min-max normalization.

[0093] S23. Spatiotemporal correlation feature construction: Calculate the strain difference, strain gradient and phase difference of adjacent strain sensors to construct spatial correlation features; use cross-correlation analysis to calculate the time delay parameters (≤0.05ms) of "strain-vibration" and "vibration-acoustic emission" signals to construct temporal correlation features.

[0094] S24. Feature Dimension Reduction and Optimization: Construct a feature matrix from the original 42-dimensional feature vectors, calculate the eigenvalues ​​and eigenvectors of the covariance matrix using the PCA algorithm, and select the top 12 principal components with a cumulative contribution rate ≥ 95% as the final input features.

[0095] In this embodiment of the application, the three-dimensional dynamic modeling in step S3 specifically includes the following steps.

[0096] S31. Basic Model Construction: Import and simplify the STEP / IGES format gate CAD model based on steel parameters (elastic modulus 206 GPa, Poisson's ratio 0.3, density 7850 kg / m³). 3 Set the finite element mesh (50mm tetrahedral mesh for the main body and 10mm refined mesh for the sensor area) and complete the model initialization.

[0097] S32. Data-driven dynamic model update: Substitute static strain data into the model and adjust the boundary conditions using the least squares method to complete static calibration by ensuring the strain deviation is ≤5%; update the model's spatial attitude based on the Euler angle rotation formula by analyzing the tilt angle data (delay ≤0.5s); combine Hooke's law and finite element algorithm to calculate key parameters such as global stress distribution and deformation in real time at a step size of 1kHz.

[0098] In this embodiment of the application, the data-driven model dynamic update described in step S32 specifically includes the following steps.

[0099] S321. Dynamic fusion of multi-source data and 3D models, mainly including the following:

[0100] (1) Static calibration based on strain data: Establish a "sensor-mesh node" association table between the installation coordinates of 28 strain sensors and the mesh nodes of the finite element model, calculate the deviation Δε between the theoretical strain and the measured strain of the model under static conditions, construct the error objective function using the weighted least squares method, adjust the boundary constraint stiffness and local material parameters, and iterate until the strain deviation of all associated nodes is ≤5%.

[0101] (2) Dynamic attitude update based on tilt angle data: Analyze the gate opening angle measured in real time by the tilt angle sensor The spatial coordinates of the panel center are calculated by combining the center coordinates of the gate hinge seat. The angle data is converted into the rotation matrix of the model based on the Euler angle rotation formula, and the spatial attitude of the model is updated, as shown in the following formula.

[0102] .

[0103] in,( , , () represents the node coordinates of the model after static calibration; , , () represents the updated node coordinates; The gate opening angle is set; the attitude update delay is ≤0.5s to ensure synchronization with the actual opening and closing action of the gate.

[0104] (3) Real-time calculation of stress and deformation based on vibration-strain fusion data: Construct a "strain-vibration" data fusion matrix, use Kalman filtering algorithm to fuse data, combine Hooke's law to calculate local stress, use vibration signal main frequency and amplitude characteristics to identify vibration mode and correct dynamic load influence; based on the fused stress data, use the Newton-Raphson iterative algorithm finite element displacement solver to calculate the displacement vector of each node, obtain the global deformation, and refresh the model stress and deformation data at a sampling frequency of 1kHz.

[0105] (4) Defect correlation modeling based on acoustic emission data: extract the event count, peak amplitude and duration characteristic parameters of acoustic emission signal to identify the defect type and location; perform local mesh secondary refinement on the defect area, combine regional stress distribution data to correct material parameters to simulate damage, and characterize the degree of defect damage through color gradient rendering.

[0106] S322, Calculation of key parameters, specifically including the following:

[0107] (1) Calculation of stress at local measuring points: Select the corresponding calculation model according to the stress type of different areas of the gate. The formula for calculating the local normal stress in the uniaxial stress area is as follows.

[0108] .

[0109] in, This represents the local normal stress in the gate structure (unit: MPa). The elastic modulus of the gate steel (in GPa) is given in the embodiments of this application. The value is 206 GPa; The effective strain after pretreatment (in με).

[0110] The formulas for calculating the local normal stress, measured strain, and shear strain in a plane stress region are as follows.

[0111] .

[0112] .

[0113] .

[0114] in, , They are respectively , Local normal stress in the direction (unit: MPa); Shear stress (unit: MPa); , They are respectively , Measured strain in the direction (unit: με); Shear modulus (in GPa); The shear strain (in rad) is determined by actual measurement using a biaxial strain gauge or derived from vibration data. The Poisson's ratio of steel is given in the embodiments of this application. The value is 0.3.

[0115] The radial basis function interpolation method is adopted, and the local stress of the 28 strain sensors is used as the constraint condition to construct the interpolation function, which is expressed as follows.

[0116] .

[0117] in, Represents any grid node The stress value (in MPa); Represents grid nodes ; For shape parameters, embodiments of this application The value is 0.01; It is a natural exponential function. By satisfying the stress constraints at 28 measurement points, the local normal stress, shear stress, and principal stress of all mesh nodes in the finite element model are obtained, forming a continuous global stress field.

[0118] (2) Local displacement calculation: Taking the gate hinge seat as the zero displacement point, the linear displacement and angular displacement are calculated by strain integration. The formula for calculating the linear displacement is as follows.

[0119] .

[0120] in, for along the position Linear displacement in the direction (in mm); Initial displacement (boundary condition, at the gate hinge) =0, unit: mm); for The distribution function of directional strain along the length is obtained by interpolating the strain at the measuring points; The independent variable for integration is (unit: mm).

[0121] Angular displacement (applicable to the rotation posture of the gate) is expressed as follows.

[0122] .

[0123] in, for Angular displacement at a location (in rad); Initial angular displacement (at the gate hinge) =0 (unit: rad). The overall tilt angle measured by the tilt sensor is used to correct the local displacement calculation results, ensuring consistency with the overall gate attitude. for Measured strain at the location (in με); The measured strain at the gate hinge (unit: με).

[0124] (3) Global deformation calculation: based on finite element equilibrium equations Solve for the global node displacements, where, The structural stiffness matrix is ​​calculated from mesh information and material parameters. Let be the nodal displacement vector. This is the nodal load vector obtained from the global stress transformation. Finally, it is expressed using the formula... Calculate the maximum deformation, where, , , , These represent the maximum three-dimensional deformation of the gate structure (in mm). Maximum linear displacement in the direction (in mm) Maximum linear displacement in the direction (in mm) Maximum linear displacement in the direction (in mm).

[0125] (4) Vibration mode calculation: The 33 vibration acceleration signals were windowed using a Hanning window, converted to the frequency domain using a fast Fourier transform, and the power spectral density was calculated using the formula: ,in, Power spectral density (unit: g) 2 / Hz). This represents the frequency domain amplitude of the vibration signal. The sampling time of the vibration signal is in seconds. The dominant frequency is identified by the peak position of the power spectral density curve. The natural frequencies were screened using peak clustering, and the amplitude and phase of each sensor at the natural frequency were extracted. Based on the spatial coordinates of the sensors, the amplitude and phase were combined, and the relative vibration displacement of each node was calculated using a formula. ,in, This represents the relative vibration displacement (in mm). The amplitude of the vibration (in mm). For the complex exponent term of Euler's formula, The vibration phase is represented by rad. The vibration modes are obtained by normalizing the relative vibration displacement.

[0126] S33. Model Visualization and Rendering: Stress distribution is rendered using color gradient (0-500MPa, blue-red gradient), with flashing alerts for areas exceeding the threshold; deformation is demonstrated in animation with a 10x default magnification factor; sensor positions are highlighted and real-time information is displayed; multi-view switching is supported.

[0127] The intelligent fault diagnosis described in step S4 specifically includes the following steps.

[0128] S41. Fault Feature Database Construction: Data from four typical scenarios—stress concentration, micro-cracks, structural deformation, and sensor failure—were collected, and fault information was labeled to construct a sample feature database with more than 1000 samples. A hybrid model of "random forest + SVM" was used, and the top 8 key features were selected and input into the radial basis kernel function. The parameters (C=10) were optimized through 5-fold cross-validation. =0.1), which makes the offline training accuracy ≥92%.

[0129] S42. Real-time Fault Diagnosis: The matching degree between real-time dimensionality-reduced features and sample features is calculated using cosine similarity. If the matching degree is greater than or equal to 85%, a fault is determined; if it is greater than or equal to 70% but less than 85%, a suspected fault is indicated; and if it is less than 70%, normal is determined. Combining sensor coordinates and 3D model data, the fault area is located using an interpolation algorithm (deviation ≤ 100mm). The severity is graded based on the evaluation function S = 0.3 × w1 + 0.25 × w2 + 0.25 × w3 + 0.2 × w4 (where w1 is the stress exceeding threshold ratio weight, w2 is the vibration amplitude growth rate weight, w3 is the acoustic emission event count weight, and w4 is the feature similarity weight). The evaluation function value is 0-0.3 for mild, 0.3-0.7 for moderate, and 0.7-1.0 for severe. The diagnosis time is ≤ 3 seconds.

[0130] S43. Fault Trend Prediction: Based on the LSTM network, input the real-time feature sequence of the past 5 minutes (time step 30) to predict the fault trend in the next 1 hour. If it is predicted that the fault will develop into a moderate or above fault within 24 hours, an early warning will be issued.

[0131] In this embodiment of the application, the result output and interaction in step S5 specifically includes the following steps.

[0132] S51. Visualization Output: Real-time display of 3D dynamic models, multi-source signal curves, fault alarm pop-ups, and key parameter statistics on the capacitive touchscreen 23. Supports screenshots, animations, and data export in PNG, MP4, CSV, and PDF formats.

[0133] S52, Operation and Maintenance Decision Support: For minor faults, prompt "Continuous monitoring, routine maintenance and key checks"; for moderate faults, prompt "Special maintenance within 72 hours"; for severe faults, prompt "Immediate shutdown and maintenance"; output processing priorities based on the importance, severity and development trend of the fault location.

[0134] This application embodiment uses a wall-mounted chassis. Figure 2 The external structure of the chassis is shown. The wall-mounted chassis has dimensions of 1100mm (L) × 800mm (W) × 220mm (H), making it compact and saving significant installation space. The chassis is a standard gray color, with a single-opening door on the left side. A capacitive touchscreen 23 (e.g., a 15-inch capacitive touchscreen 23) is located on the front for easy operation and observation. A cable entry hole with a waterproof seal is located at the bottom of the chassis to prevent moisture ingress. Louvered air inlets 12 are located on the lower left and right side panels, while a grille-style air outlet is located at the top, equipped with two exhaust fans 22 for cooling, forming a "bottom-in, top-out" airflow to improve heat dissipation efficiency. The rear of the chassis features a power interface, sensor signal input interfaces (including interfaces for strain sensors, vibration sensors, acoustic emission sensors, and tilt sensors), a network interface (RJ45), and data output interfaces (USB, RS485) for convenient on-site cabling and equipment connection.

[0135] Figure 3The internal structure of the chassis is shown. Inside the chassis are an industrial computer 1, a temperature and humidity sensor 2, a control device 3, a 220V double-layer terminal block 4, a DIN rail socket 5, a 2P circuit breaker 6, a 6-position grounding bar 7, a 24V power supply 8, a 28V power supply 9, an acoustic emission preamplifier 10, a cable tray 11, an air inlet 12, a terminal block 13 (including double-layer and triple-layer terminal blocks), a 220V inlet 14, a heater 15, a sensor inlet 16, a 100P linear board 17, a 68P adapter board 18, a 4520 serial port expansion module 19, a strain sensor amplifier 20, an adapter 21, and an exhaust fan 22. The lower air intake 12 of the chassis forms a natural air intake channel. External AC power is connected via the 220V input port 14, and the circuit switching and overload protection are achieved through the 2P circuit breaker 6. The 220V AC power is then distributed and converted by the 220V double-layer terminal 4 and the DIN rail socket 5 to provide AC power for the entire machine. The 6-position grounding busbar 7 uniformly grounds the equipment to ensure power and signal safety. The 24V power supply 8 and 28V power supply 9 convert the AC power into the corresponding DC voltage, providing stable power to devices such as the strain sensor amplifier 20, the acoustic emission preamplifier 10, and the tilt sensor. The middle part of the chassis is the core area for signal processing and control. The industrial control computer 1 serves as the main control unit of the system, coordinating data acquisition, calculation and analysis, command issuance, and overall machine collaborative control. The control device 3 works with the industrial control computer 1 to complete local logic management and loop switching. External sensor signals are introduced through sensor inlet 16, and the wiring is organized and signal is transferred via terminal block 13; strain sensor amplifier 20 amplifies and conditions weak strain signals, and acoustic emission preamplifier 10 enhances acoustic emission signal strength and improves anti-interference capability; 100P linear board 17 and 68P adapter board 18 complete multi-channel signal expansion and wiring transfer, and 4520 serial port expansion module 19 expands serial port resources to meet the communication needs of serial port devices such as tilt sensors; adapter 21 adapts to different device interfaces and voltage specifications to ensure normal signal and power transmission; cable tray 11 organizes all cables to avoid interference and safety hazards caused by messy wiring. The upper part of the chassis and surrounding environmental and heat dissipation components include: temperature and humidity sensor 2 for real-time monitoring of internal temperature and humidity and timely feedback of environmental status; heater 15 for starting in low temperature and high humidity environments to prevent condensation and low temperature from affecting the operation of electronic components; top exhaust fan 22 and bottom air inlet 12 to form a circulating air duct to accelerate heat dissipation inside the chassis; the entire set of components works together to ensure long-term, stable and reliable operation of the monitoring device under the complex working conditions of the hydraulic gate chamber.

[0136] In this embodiment, the modular power supply system includes a 24V power supply 8, a 28V power supply 9, and an AC30-10530 model DIN rail socket 5. The 24V power supply 8 and 28V power supply 9 are fixed to the mounting bracket at the bottom of the chassis via DIN rails, resulting in a compact layout that facilitates installation and maintenance. The power cables are color-coded (e.g., blue for 24V, red for 28V, and black for 220V) to avoid wiring confusion.

[0137] This embodiment uses an Advantech ITA3650E industrial computer 1, which is connected to the side panel of the chassis via an L-shaped mounting bracket to ensure a secure installation. The PCI and PCIe slots of the industrial computer 1 face to the right for easy insertion of the data acquisition card. The right-side signal processing area houses a UKK3 terminal block 13, four JX series strain sensor amplifiers 20, and six acoustic emission preamplifiers 10. The terminal block 13 is fixed to the mounting column of the chassis with screws. The strain sensor amplifiers 20 and acoustic emission preamplifiers 10 are located to the left and right of the terminal block 13, respectively, and are connected to the terminal block 13 and the data acquisition card via cables to amplify and transfer signals. Simultaneously, this area also houses a 4520 serial port expansion module 19, which connects to the serial port of the industrial computer 1 via a DB9 serial cable to expand the number of serial ports on the industrial computer 1, meeting the communication needs of devices such as tilt sensors.

[0138] This embodiment uses an SHT30 temperature and humidity sensor 2 to ensure accurate monitoring of the temperature and humidity around the industrial computer 1. A heating plate is also fixed to the upper part of the rear panel of the chassis with screws, improving heat transfer efficiency through close contact with the rear panel. Two exhaust fans 22 for heat dissipation are symmetrically installed at the exhaust vents on the top of the chassis, with the fan airflow pointing upwards. The fan speed is adjusted by connecting to the PWM interface of the industrial computer 1 via a speed control module.

[0139] In this embodiment, cable fixing clips are installed on the inner side wall of the chassis to organize power cables and signal cables and avoid interference caused by messy cables; at the same time, a 100mm operating space is reserved between the acquisition card and the terminal block 13 to facilitate subsequent wiring inspection and maintenance.

[0140] Figure 4 The internal wiring of the chassis is shown. In this embodiment, the installation and wiring methods of each component of the device are as follows.

[0141] Industrial PC 1 Installation: First, secure the L-shaped mounting bracket to the left-side panel in the middle of the chassis using screws. Adjust the bracket position to ensure that the PCI and PCIe slots of Industrial PC 1 face right after installation. Then, place the Advantech ITA3650E Industrial PC 1 on the bracket, aligning the mounting holes of Industrial PC 1 with the bracket holes. Secure Industrial PC 1 to the bracket using M4 screws, ensuring it is secure and maintains a 50mm gap between it and the chassis side panel to allow for heat dissipation.

[0142] Data Acquisition Card Installation: Open the chassis cover of Industrial PC 1 and identify the positions of the PCI and PCIe slots; slowly insert the PCI-1747U data acquisition card into the PCI slot, ensuring that the gold fingers of the data acquisition card are in full contact with the slot, and then use M3 screws to fix the mounting plate of the data acquisition card to the chassis of Industrial PC 1; in the same way, insert the PCIe-1812 data acquisition card into the PCIe slot and fix it; after installation, check whether the data acquisition card is installed in place, and close the chassis cover of Industrial PC 1 after confirming that everything is correct.

[0143] Serial port expansion module connection: Fix the 4520 serial port expansion module 19 to the lower part of the terminal block 13 in the middle area via DIN rail. Use a shielded DB9 serial cable to connect the input serial port of the module to the COM1 port of the industrial computer 1. Secure both ends of the cable with screws to ensure good contact. The output serial port of the module is used to connect external devices such as tilt sensors. The interface is marked for easy wiring later.

[0144] Power module mounting: Install the 24V power supply 8, 28V power supply 9 and AC30-10530 model DIN rail socket 5 in sequence on the DIN rail at the bottom of the chassis, keeping a distance of 30mm between adjacent modules to ensure heat dissipation; use a multimeter to measure the input and output terminals of the power module, and after confirming that there is no short circuit, fix the DIN rail to the bracket at the bottom of the chassis.

[0145] Power cable selection and wiring: Cables are selected according to the power requirements of each module. The 24V cable powering the strain sensor amplifier 20 uses AWG18 shielded wire, the 28V cable powering the acoustic emission preamplifier 10 uses AWG20 shielded wire, and the 220V cable powering the industrial computer 1 uses RVV3×1.5mm² wire. 2 Standardized cables; power cables are routed along the cable clips on the side wall of the chassis to avoid crossing with signal cables and reduce electromagnetic interference; cable bends are rounded to prevent breakage.

[0146] Power Connection: Connect an external 220V AC power supply to the inlet hole at the bottom of the chassis via a waterproof cable, and connect it to the input terminal of the AC30-10530 model DIN rail socket 5. Wrap the wiring with insulating tape and heat shrink tubing to prevent short circuits. Connect the output terminal of DIN rail socket 5 to the power interface of industrial computer 1 via a 220V cable to power industrial computer 1. Connect the input terminal of 24V power supply 8 to DIN rail socket 5, and connect the output terminal to the power input terminal of four strain sensor amplifiers 20 via AWG18 shielded cable. Install a fuse (rated current 1A) at the power interface of each amplifier. Similarly, connect the input terminal of 28V power supply 9 to DIN rail socket 5, and connect the output terminal to the power input terminal of six acoustic emission preamplifiers 10 via AWG20 shielded cable. Install a fuse (rated current 0.5A) at the interface. After all power connections are completed, use a multimeter to measure the voltage of each circuit to confirm that the output voltage meets the requirements and that there are no reverse connections or short circuits.

[0147] Strain gauge channel wiring (taking channels 1-8 as an example): The strain gauge sensor is connected to a 1.6m long AWG24 shielded cable via a 5P waterproof connector. When connecting the connector, ensure the positioning pins are aligned to prevent reverse connection. The other end of the cable is connected to the CH1-CH8 signal input terminals of the strain gauge amplifier 20. Each sensor corresponds to one amplifier channel. When wiring, distinguish between the positive, negative, and ground wires. The signal output terminal of the strain gauge amplifier 20 is connected to terminals 1-8 of the UKK3 type terminal block 13 via an AWG22 shielded cable. The channel number is marked on the terminal block. Terminals 1-8 of the terminal block 13 are connected to the AI0-AI7 channel input terminals of the PCI-1747U data acquisition card via an AWG22 shielded cable. The corresponding channel number is also marked on the data acquisition card terminals. After all wiring is completed, use a multimeter to measure the continuity of the signal lines to confirm that there are no loose connections.

[0148] In this embodiment, the strain terminal block is installed in the signal processing area of ​​the chassis, powered by a 24V power supply 8 and connected to a 6-position grounding busbar 7 for grounding protection. The input end is connected to each strain sensor and the strain sensor amplifier 20, and the output end is connected to the first acquisition card through a line. The cables are uniformly stored in the cable tray 11. It is mainly used for centralized conversion and regularization of multiple strain signals, preliminary signal splitting and impedance matching to reduce line crosstalk, and at the same time provides wiring nodes for the strain acquisition circuit, which facilitates line maintenance, channel expansion and fault diagnosis, and ensures that the strain signal is stably and accurately transmitted to the acquisition card.

[0149] Vibration channel wiring: The vibration sensor is connected to a 2m long coaxial cable via a BNC connector. The other end of the cable is connected to terminals 9-41 of UKK3 type terminal block 13 (corresponding to 33 vibration signals). Terminal block 13 is connected to the AI8-AI40 channels of the PCI-1747U acquisition card via an AWG22 shielded cable. When wiring, ensure that the signal shielding layer is grounded to reduce interference.

[0150] Acoustic emission channel wiring: The acoustic emission sensor is directly connected to the input terminal of the acoustic emission preamplifier 10 via a BNC coaxial cable (capacitance ≤100pF / m). The output terminal of the acoustic emission preamplifier 10 is connected to the CH1-CH6 channels of the PCIe-1812 acquisition card via differential signal lines. The acquisition card adopts differential input mode to improve anti-interference capability. The cable shielding layer is grounded at one end (grounded only at the acquisition card end) to avoid forming ground loop current.

[0151] Tilt sensor wiring: The tilt sensor is connected to the output serial port of the 4520 serial port expansion module 19 via an RS485 bus, using an RVVSP2×0.5mm... 2 Use shielded twisted-pair cable, distinguish between signal lines A and B when wiring, and ground the shielding layer; connect the sensor power supply to a 24V power supply to ensure stable power supply.

[0152] In this embodiment, the tilt switch and the tilt sensor are installed together on the rigid main beam of the gate. The signal is connected to the chassis through the sensor inlet 16, and after being converted through the terminal block 13, it is connected to the control device 3, the 4520 serial port expansion module 19 and the industrial computer 1 respectively. It is powered by a 24V power supply 8 and grounded by a 6-position grounding busbar 7. As a posture safety auxiliary device, it can preset the tilt threshold. When the gate tilts beyond the limit, it triggers the contact action. On the one hand, it links the control device 3 to realize safety protection such as shutdown and locking and start the on-site audible and visual alarm. On the other hand, it transmits the switch signal to the industrial computer 1 to help distinguish the operating conditions and optimize fault judgment. It can also supplement the monitoring and protection capabilities when the tilt sensor or line fails. During the maintenance phase, it can also cooperate to complete the circuit switching, so as to comprehensively ensure the safety of gate operation.

[0153] In this embodiment, the preamplifier power supply separation signal unit is arranged in the signal processing area inside the chassis, powered by a 28V power supply 9. Its input is connected to the acoustic emission preamplifier 10, and its output is connected to the second acquisition card via a terminal block 13. The wiring is housed in a cable tray 11 and connected to a 6-position grounding busbar 7 for grounding protection. This device primarily achieves physical separation between the power supply circuit and the signal circuit. On one hand, it isolates the high-frequency acoustic emission signal from current noise and interference generated by the power supply circuit, ensuring the integrity and accuracy of signal transmission. On the other hand, it can independently control the power supply to the acoustic emission preamplifier 10, cutting off the power supply during equipment maintenance and troubleshooting without affecting signal link detection. It can also block ground loop currents between different modules, improving the anti-interference capability and electrical safety of the entire monitoring device and ensuring the long-term stable operation of the acoustic emission acquisition channel.

[0154] 3D Intelligent Analysis Module Deployment: Install the Windows 10 operating system and Advantech device drivers on Industrial PC 1 to ensure that the dual acquisition cards can be recognized by the system; install 3D intelligent analysis software developed based on C++ and Python, which integrates data acquisition, preprocessing, 3D modeling, fault diagnosis and other functional modules; configure acquisition parameters (sampling rate, range, filtering method, etc.) through the software, set the mesh accuracy and material parameters of 3D modeling (input the elastic modulus of steel, Poisson's ratio, etc. based on the gate design drawings); import the gate CAD model as the basic template for 3D modeling, and complete the linkage debugging of software and hardware.

[0155] In this embodiment of the application, the debugging process of the three-dimensional intelligent analysis module includes the following: Data Acquisition and Preprocessing Debugging: Start the system and acquire strain, vibration, and acoustic emission data of the gate in static state. Filter the data using software (50Hz power frequency filter for strain signals, low-pass filter for vibration signals, and band-pass filter for acoustic emission signals) and perform noise reduction. Compare the original sensor data with the preprocessed data to verify the effectiveness of the preprocessing algorithm and ensure that data noise is reduced to within acceptable limits (strain signal noise ≤ 5με, vibration signal noise ≤ 0.01g, acoustic emission signal noise ≤ 10dB). If noise exceeds the limit, adjust the filtering parameters or check the sensor wiring to eliminate interference factors.

[0156] 3D Modeling and Debugging: Based on the collected static strain data and combined with the gate CAD model, an initial 3D model of the gate is constructed using the finite element analysis module of the software; the installation positions of the strain sensors are marked in the model, and the measured strain values ​​are compared with the calculated strain values ​​of the model, with a deviation requirement of ≤5%; if the deviation is too large, the boundary conditions of the model (such as constraint positions and load distribution) are adjusted, and the calculation is repeated until the deviation meets the requirements; the gate opening and closing process is simulated, and the gate angle data collected by the tilt sensor is input. The software should update the attitude of the 3D model in real time and dynamically display the rotation process of the gate, with an attitude update delay of no more than 0.5s.

[0157] Fault diagnosis and debugging: By setting faults on the gate simulation test bench (such as adding weights to simulate local stress concentration, and knocking on the gate to generate acoustic emission signals simulating cracks), multi-source heterogeneous signals under fault conditions are collected; the fault data is input into the software fault diagnosis module to verify whether the module can accurately identify the fault type (stress concentration, crack initiation), locate the fault location (deviation ≤100mm), and assess the severity of the fault (matching the actual fault severity); if the diagnosis results are inaccurate, the fault feature extraction algorithm and diagnostic model parameters are optimized, and sample training data is increased until the diagnostic accuracy reaches more than 90%.

[0158] Based on the same inventive concept, this application also provides a method for three-dimensional intelligent gate monitoring based on multi-source heterogeneous signal fusion, corresponding to the aforementioned gate three-dimensional intelligent monitoring device based on multi-source heterogeneous signal fusion. The solution provided by this method is similar to the solution described in the aforementioned device. Therefore, the specific limitations in the embodiments of the gate three-dimensional intelligent monitoring method based on multi-source heterogeneous signal fusion provided below can be found in the limitations of the gate three-dimensional intelligent monitoring device based on multi-source heterogeneous signal fusion described above, and will not be repeated here.

[0159] In one exemplary embodiment, such as Figure 5 As shown, a three-dimensional intelligent monitoring method for gates based on multi-source heterogeneous signal fusion is provided, which specifically includes the following steps.

[0160] A1: Collect strain signals, vibration signals, acoustic emission signals, and tilt angle signals of the gate under test.

[0161] A2: Based on the strain signal, the vibration signal, the acoustic emission signal and the tilt angle signal, perform three-dimensional dynamic modeling and intelligent fault diagnosis to obtain the three-dimensional dynamic model and fault diagnosis results of the gate under test.

[0162] A3: Real-time display of the monitoring data of the gate under test, the three-dimensional dynamic model, and the fault diagnosis results.

[0163] As an optional implementation, step A2 involves performing three-dimensional dynamic modeling and intelligent fault diagnosis based on the strain signal, the vibration signal, the acoustic emission signal, and the tilt angle signal to obtain the three-dimensional dynamic model and fault diagnosis results of the gate under test. This specifically includes the following steps.

[0164] A21: Standardize, differentiate denoising, and anomaly repair processing are performed on the strain signal, vibration signal, acoustic emission signal, and tilt angle signal to obtain preprocessed data.

[0165] A22: Extract time-domain, frequency-domain, and spatiotemporal correlation features from the preprocessed data and perform dimensionality reduction optimization to obtain core input features.

[0166] A23: Based on the core input features, a finite element model of the gate to be tested is constructed. Through data-driven methods, static calibration, dynamic attitude update, and real-time calculation of stress and deformation of the model are realized, and visualization rendering is completed to obtain a three-dimensional dynamic model.

[0167] A24: Input the core input features into the fault diagnosis model to identify the fault type, locate the fault, assess the severity and predict the trend of the gate under test, and obtain the fault diagnosis result.

[0168] As an optional implementation, the fault diagnosis result includes a fault severity level, which includes mild fault, moderate fault, and severe fault.

[0169] After performing three-dimensional dynamic modeling and intelligent fault diagnosis based on the strain signal, the vibration signal, the acoustic emission signal and the tilt angle signal to obtain the three-dimensional dynamic model and fault diagnosis results of the gate under test, the gate three-dimensional intelligent monitoring method based on multi-source heterogeneous signal fusion further includes: generating hierarchical operation and maintenance decision suggestions.

[0170] The tiered operation and maintenance decision-making recommendations include: continuous monitoring for minor faults, dedicated maintenance within a preset time for moderate faults, and immediate shutdown and maintenance for severe faults.

[0171] The present application provides a gate three-dimensional intelligent monitoring device and method based on multi-source heterogeneous signal fusion, which has the following advantages.

[0172] (1) High-precision synchronous acquisition of multi-source heterogeneous signals was achieved, laying the foundation for accurate correlation analysis.

[0173] To address the challenges of poor signal synchronization and ineffective data correlation in existing monitoring systems due to the use of distributed, independent acquisition cards, this application employs a collaborative architecture combining PCI-1747U and PCIe-1812 dual acquisition cards, integrating high-precision clock synchronization technology to control the synchronization error of multiple signals within 0.05ms. This improvement enables the device to accurately capture the microscopic temporal logic between "strain change - vibration response - acoustic emission signal" during gate operation, thereby accurately identifying fault evolution patterns such as "stress concentration preceding vibration abrupt change" and "acoustic emission signal preceding structural deformation." This provides key technical support for overcoming the core obstacle of multi-source heterogeneous signal fusion analysis and creates conditions for achieving early fault warning.

[0174] (2) It realizes intelligent and accurate diagnosis and visual location of faults, which greatly improves the efficiency of operation and maintenance.

[0175] To verify the effectiveness and feasibility of the technical solution of this application, this application uses stress concentration, micro-cracks and mixed faults as fault types, and conducts experimental comparisons between the traditional method of PLC combined with manual labor and the method of this application in terms of diagnostic accuracy, response time and positioning deviation. Table 1 shows the comparison results of diagnostic accuracy, Table 2 shows the comparison results of response time and Table 3 shows the comparison results of positioning deviation.

[0176] Table 1. Comparison of Diagnostic Accuracy

[0177] Table 2 Response Time Comparison Results

[0178] Table 3 Comparison Results of Positioning Deviation

[0179] As can be clearly seen from Tables 1, 2 and 3, the technical solution of this application not only shortens the response time, but also improves the diagnostic accuracy and reduces the positioning deviation.

[0180] To address the shortcomings of traditional systems, such as reliance on human experience, low diagnostic efficiency, and ambiguous localization, this application leverages high-quality, multi-source heterogeneous signals obtained through high-precision synchronous acquisition. Through a built-in 3D intelligent analysis and machine learning module, it achieves automated and precise fault diagnosis. Specifically, this is manifested in the following ways.

[0181] 1) Accurate and rapid diagnosis: It can automatically identify fault types such as stress concentration and micro cracks, improve the diagnostic accuracy to over 90%, and shorten the response time from tens of minutes in traditional methods to seconds (e.g., within 3 seconds).

[0182] 2) Intuitive and visual positioning: The fault area can be calculated and highlighted in real time in the dynamic three-dimensional model of the gate, realizing the visual positioning of the fault, and the positioning deviation is no more than 100mm.

[0183] 3) Operation and maintenance decision support: It not only issues alarms, but also provides specific quantitative operation and maintenance suggestions such as "stress exceeding threshold ratio and recommended maintenance time limit", thereby significantly reducing the reliance on the experience of operation and maintenance personnel and improving the safety of gate operation and the foresight of maintenance work.

[0184] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0185] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A three-dimensional intelligent monitoring device for gates based on multi-source heterogeneous signal fusion, characterized in that, It includes an industrial control computer, dual data acquisition card assembly, sensor module, modular power supply system, human-machine interaction module, and 3D intelligent analysis module; The industrial control computer is connected to the dual data acquisition card assembly, the human-machine interaction module, the modular power supply system, and the three-dimensional intelligent analysis module. The industrial control computer is used to perform data reception, processing, storage, and coordinated control of each module. The modular power supply system is used to supply power to each module. The sensor module is installed on the gate under test. The sensor module includes a strain sensor, a vibration sensor, an acoustic emission sensor and a tilt sensor, which are used to collect the strain signal, vibration signal, acoustic emission signal and tilt angle signal of the gate under test, respectively. The dual acquisition card assembly includes a first acquisition card and a second acquisition card; the first acquisition card is connected to the strain sensor and the vibration sensor respectively, and the second acquisition card is connected to the acoustic emission sensor; the first acquisition card is used to receive the strain signal and the vibration signal and transmit them to the industrial control computer, and the second acquisition card is used to receive the acoustic emission signal and transmit it to the industrial control computer; The tilt sensor is connected to the industrial control computer, and the tilt sensor is also used to transmit the tilt angle signal to the industrial control computer; The three-dimensional intelligent analysis module is used to perform three-dimensional dynamic modeling and intelligent fault diagnosis based on the strain signal, the vibration signal, the acoustic emission signal and the tilt angle signal, so as to obtain the three-dimensional dynamic model and fault diagnosis results of the gate under test; The human-computer interaction module is used to interact with the user and display the monitoring data of the gate under test, the three-dimensional dynamic model, and the fault diagnosis results in real time.

2. The gate three-dimensional intelligent monitoring device based on multi-source heterogeneous signal fusion according to claim 1, characterized in that, The first acquisition card is a PCI interface acquisition card, which is compatible with the acquisition of 28 strain signals and 33 vibration signals; the second acquisition card is a PCIe interface acquisition card, which is compatible with the acquisition of 6 acoustic emission signals; the first acquisition card and the second acquisition card adopt clock synchronization technology, and the synchronization error is ≤0.05ms.

3. The gate three-dimensional intelligent monitoring device based on multi-source heterogeneous signal fusion according to claim 1, characterized in that, The strain sensor is a foil strain gauge with a sensitivity coefficient of 2.0±1%; the vibration sensor is a piezoelectric accelerometer with a measurement range of ±50g; the acoustic emission sensor is a resonant sensor with a center frequency of 150kHz; and the tilt sensor has a measurement accuracy of ±0.1°.

4. The gate three-dimensional intelligent monitoring device based on multi-source heterogeneous signal fusion according to claim 1, characterized in that, The three-dimensional intelligent analysis module includes a preprocessing unit, a feature extraction unit, a three-dimensional dynamic modeling unit, a fault intelligent diagnosis unit, and a result output unit; The preprocessing unit is used to standardize, differentiate denoising, and repair anomalies in the strain signal, vibration signal, acoustic emission signal, and tilt angle signal to obtain preprocessed data. The feature extraction unit is used to extract time-domain, frequency-domain, and spatiotemporal correlation features from the preprocessed data and perform dimensionality reduction and optimization to obtain core input features; The three-dimensional dynamic modeling unit is used to construct the finite element basic model of the gate under test based on the core input features. It realizes static calibration, dynamic attitude update and real-time calculation of stress and deformation of the model through data-driven process, and completes visualization rendering to obtain a three-dimensional dynamic model. The intelligent fault diagnosis unit is used to input the core input features into the fault diagnosis model to identify the fault type, locate the fault, assess the severity and predict the trend of the gate under test, and obtain the fault diagnosis result. The result output unit is used to output the three-dimensional dynamic model and the fault diagnosis results to the human-computer interaction module.

5. The gate three-dimensional intelligent monitoring device based on multi-source heterogeneous signal fusion according to claim 4, characterized in that, During differential denoising, the preprocessing unit employs 50Hz power frequency notch filtering and db4 wavelet 3-level decomposition threshold denoising for the strain signal, 8kHz low-pass filtering and singular value decomposition denoising for the vibration signal, 100kHz-200kHz bandpass filtering and energy threshold denoising for the acoustic emission signal, and 5-point moving average filtering for the tilt angle signal.

6. The gate three-dimensional intelligent monitoring device based on multi-source heterogeneous signal fusion according to claim 4, characterized in that, During 3D dynamic modeling, the three-dimensional dynamic modeling unit performs static calibration based on strain data, adjusts boundary conditions to ensure strain deviation is ≤5%, updates the model space attitude based on Euler angle rotation formula based on tilt data with an update delay of ≤0.5s, calculates global stress and deformation based on vibration-strain fusion data, combined with Hooke's law and finite element algorithm, and identifies defects based on acoustic emission data, refines the mesh in defect areas, and corrects material parameters.

7. The gate three-dimensional intelligent monitoring device based on multi-source heterogeneous signal fusion according to claim 4, characterized in that, The intelligent fault diagnosis unit constructs a fault feature library containing stress concentration, microcracks, structural deformation, and sensor faults during intelligent fault diagnosis; it adopts a hybrid model based on random forest and SVM as the fault diagnosis model, and uses cosine similarity matching to determine the fault; and it evaluates the severity of the fault based on the evaluation function.

8. A three-dimensional intelligent monitoring method for gates based on multi-source heterogeneous signal fusion, characterized in that, The gate three-dimensional intelligent monitoring method based on multi-source heterogeneous signal fusion is implemented based on the gate three-dimensional intelligent monitoring device based on multi-source heterogeneous signal fusion as described in any one of claims 1-7. The gate three-dimensional intelligent monitoring method based on multi-source heterogeneous signal fusion includes: The strain signal, vibration signal, acoustic emission signal, and tilt angle signal of the gate under test are collected. Based on the strain signal, vibration signal, acoustic emission signal and tilt angle signal, three-dimensional dynamic modeling and intelligent fault diagnosis are performed to obtain the three-dimensional dynamic model and fault diagnosis results of the gate under test. The system displays the monitoring data of the gate under test, the three-dimensional dynamic model, and the fault diagnosis results in real time.

9. The gate three-dimensional intelligent monitoring method based on multi-source heterogeneous signal fusion according to claim 8, characterized in that, Based on the strain signal, vibration signal, acoustic emission signal, and tilt angle signal, three-dimensional dynamic modeling and intelligent fault diagnosis are performed to obtain the three-dimensional dynamic model and fault diagnosis results of the gate under test, specifically including: The strain signal, vibration signal, acoustic emission signal, and tilt angle signal are standardized, differentially denoised, and anomaly repaired to obtain preprocessed data. The time-domain, frequency-domain, and spatiotemporal correlation features are extracted from the preprocessed data and then optimized by dimensionality reduction to obtain the core input features. Based on the core input features, a finite element basic model of the gate under test is constructed. Through data-driven methods, static calibration, dynamic attitude update, and real-time calculation of stress and deformation of the model are realized, and visualization rendering is completed to obtain a three-dimensional dynamic model. The core input features are input into the fault diagnosis model to identify the fault type, locate the fault, assess the severity, and predict the trend of the gate under test, thereby obtaining the fault diagnosis result.

10. The gate three-dimensional intelligent monitoring method based on multi-source heterogeneous signal fusion according to claim 8, characterized in that, The fault diagnosis results include the fault severity level, which includes mild fault, moderate fault and severe fault. After performing three-dimensional dynamic modeling and intelligent fault diagnosis based on the strain signal, the vibration signal, the acoustic emission signal and the tilt angle signal to obtain the three-dimensional dynamic model and fault diagnosis results of the gate under test, the gate three-dimensional intelligent monitoring method based on multi-source heterogeneous signal fusion further includes: generating hierarchical operation and maintenance decision suggestions. The tiered operation and maintenance decision-making recommendations include: continuous monitoring for minor faults, dedicated maintenance within a preset time for moderate faults, and immediate shutdown and maintenance for severe faults.

Images