Foundation solidification monitoring method based on infrared imaging technology

Abstract

The invention relates to the technical field of foundation detection methods, and particularly discloses a foundation solidification monitoring method based on an infrared imaging technology, which comprises the following steps of: 1) monitoring a soft foundation added with a chemical curing agent in real time by using infrared imaging equipment to obtain real-time infrared imaging data; 2) establishing a large-scale database covering various foundation conditions and chemical curing agent types, and constructing a deep learning framework training prediction model; 3) outputting and predicting the optimal dosage of the chemical curing agent required by the soft foundation under different conditions; 4) giving a chemical curing agent putting strategy; and (5) a remote monitoring and operating platform is constructed, the condition of the soft foundation with the chemical curing agent is obtained, remote control can be carried out, and an alarm can be automatically given out when an abnormal condition occurs. By integrating advanced real-time data analysis, a self-adaptive adjustment mechanism and a deep learning algorithm, real-time and accurate monitoring of the soft foundation solidification process is realized.

Description

Technical Field

[0001] The invention relates to the technical field of foundation detection methods, and in particular to a foundation solidification monitoring method based on infrared imaging technology. Background Art

[0002] The solidification of soft foundation is a key step to improve the stability and bearing capacity of the foundation. Especially under complex geological conditions, it is particularly important to ensure the effectiveness of the solidification process. However, existing monitoring methods mainly rely on physical sampling and laboratory analysis, which are delayed, localized and have limited data, and are difficult to meet the real-time monitoring needs of large-scale projects. At the same time, traditional monitoring technologies are inefficient in the use of resources, often resulting in excessive or insufficient application of curing agents.

[0003] With the development of infrared imaging technology, it has become possible to monitor the foundation solidification process using temperature changes. However, current technology still lacks intelligent data analysis systems, automated adjustment mechanisms, and complete solutions that integrate remote monitoring. Therefore, there is an urgent need for an innovative technology that can monitor the soft foundation solidification process in real time, comprehensively, and accurately to fill the gap in existing technology. Summary of the invention

[0004] The purpose of the present invention is to provide a foundation solidification monitoring method based on infrared imaging technology, which realizes real-time and accurate monitoring of the soft foundation solidification process by integrating advanced real-time data analysis, adaptive adjustment mechanism and deep learning algorithm.

[0005] The present invention is implemented by the following technical solution: A foundation solidification monitoring method based on infrared imaging technology comprises the following steps:

[0006] (1) Based on the temperature change generated during the reaction between the chemical curing agent and the soft foundation, an infrared imaging device is used to monitor the soft foundation to which the chemical curing agent is added in real time to obtain real-time infrared imaging data;

[0007] (2) Establish a large-scale database covering various foundation conditions and chemical curing agent types, and build a deep learning framework to train the prediction model based on this database;

[0008] (3) The infrared imaging data are analyzed and processed by training the prediction model through the constructed deep learning framework, the diffusion range and depth of the chemical curing agent in the soft foundation, and the impact on the foundation hardness are evaluated, and the optimal amount of chemical curing agent required for the soft foundation under different conditions is predicted;

[0009] (4) Based on the predicted optimal dosage of chemical curing agent, a strategy for the placement of chemical curing agent is given; during the process of placing chemical curing agent, the collected infrared imaging data will change, so the optimal dosage of chemical curing agent output by the constructed deep learning framework training will change dynamically, and then the corresponding chemical curing agent placement strategy is adaptively and dynamically generated, that is, for overheated areas, the amount of curing agent is reduced; for cold spot areas, the amount of curing agent is increased or the placement time is extended to ensure the heat balance and uniformity of the soft foundation curing process to adapt to possible new situations or environmental changes;

[0010] (5) Build a remote monitoring and operation platform to obtain the temperature change curve, hardness distribution diagram and curing agent placement status of the soft foundation with chemical curing agent, and remotely control the chemical curing agent placement equipment manually or automatically, and automatically issue an alarm when abnormal conditions in the soft foundation are detected.

[0011] In order to better implement the method of the present invention, further, the infrared imaging device in step (1) is an infrared camera, the resolution of the infrared camera is 640×480, the temperature measurement range is -20°C to 150°C, and the measured infrared band is 3μm to 5μm or 8μm to 14μm.

[0012] In order to better implement the method of the present invention, further, the specific process of the step (1) of collecting infrared imaging data by using an infrared camera is as follows:

[0013] (1.1) Use a standard black source to perform preliminary calibration of the infrared camera;

[0014] (1.2) Fix the infrared camera on the drone through a bracket, use the drone to test fly at the monitoring site, and perform the second calibration of the infrared camera;

[0015] (1.3) A monitoring matrix is ​​formed by using multiple drones equipped with infrared cameras, so that the infrared cameras can form infrared imaging data collection with overlapping viewing angles;

[0016] (1.4) Infrared imaging data is collected on a soft foundation with a chemical curing agent added using an acquisition frequency of 1 second / frame to 5 seconds / frame, and then transmitted to an edge computing device via a high-speed data link for data storage, wherein the high-speed data link includes Ethernet and optical fiber.

[0017] In order to better implement the method of the present invention, further, in the step (1.3), a monitoring matrix is ​​formed by using multiple drones equipped with infrared cameras so that the infrared cameras can form infrared imaging data acquisition with overlapping viewing angles. The specific process is as follows:

[0018] (1.3.1) Choose a drone with long endurance to ensure the flight stability and image quality of the drone;

[0019] (1.3.2) Develop the flight path of the drone based on the terrain characteristics of the solidification area and the construction plan; the flight path must cover all key areas and ensure the drone's flight altitude and viewing angle to obtain the best imaging effect;

[0020] (1.3.3) The drone is equipped with automatic navigation function, combining GPS and inertial navigation system to achieve accurate positioning and path tracking; at the same time, it uses laser radar or visual sensors to achieve real-time obstacle avoidance to ensure flight safety;

[0021] (1.3.4) During the flight of the UAV, the ground temperature data is continuously collected and the collected infrared image data is transmitted to the ground station in real time through the wireless communication link. The data should be synchronously marked with timestamps to ensure the consistency of subsequent processing;

[0022] (1.3.5) The basic computing unit on the drone can perform basic denoising and compression on the preliminary data, reducing the transmission load and improving data processing efficiency;

[0023] (1.3.6) After the drone flight is completed, the system generates a temperature distribution map covering the entire curing area and identifies areas of abnormal temperature or uneven curing;

[0024] (1.3.7) For the identified abnormal areas, the drone performs low-altitude high-resolution scanning to obtain more detailed temperature data and solidification conditions for further analysis and processing;

[0025] (1.3.8) Integrate the data collected by the drone with the data from the ground infrared imaging equipment to generate a global curing assessment report, providing more comprehensive and accurate curing status information;

[0026] (1.3.9) All data collected by the drones are uploaded to the constructed remote monitoring and operation platform for storage. The operator can monitor the flight status and data collection progress of the drone in real time through the remote monitoring and operation platform, and adjust the flight parameters or mission plan as needed.

[0027] In order to better implement the method of the present invention, further, in the step (1.4), the infrared imaging data acquisition uses hotspot recognition technology to dynamically track the soft foundation to which the chemical curing agent is added, and the specific process is as follows:

[0028] (1.4.1) By calculating the temperature gradient of pixels in the infrared image, the area with sharp temperature changes can be identified and the hot spot location can be preliminarily determined;

[0029] (1.4.2) Using the initially identified hotspot area as a seed, the region growing algorithm is used to expand the detection range and accurately locate the boundaries of the hotspot area;

[0030] (1.4.3) Use multi-scale analysis technology to identify hot spots at different scales, ensuring that small hot spots are not ignored while taking into account the accurate identification of large-scale hot spots

[0031] (1.4.4) After identifying the hotspot, the changes of the hotspot are tracked through the analysis of continuous frames, including: using the Kalman filter to predict and update the hotspot position, filtering out noise interference, and ensuring the smoothness of the tracking path; using the template matching algorithm to find matching points in the hotspot area in continuous infrared image frames, and tracking the movement, changes and splitting of the hotspot area; when the hotspot splits or merges, the system automatically updates the number and position of the hotspot to ensure the real-time and accuracy of the hotspot data.

[0032] In order to better implement the method of the present invention, further, in step (2), a large-scale database covering various foundation conditions and chemical curing agent types is established, and the specific process of constructing a deep learning framework training prediction model based on the database is as follows:

[0033] (2.1) Collect temperature and hardness data of different foundation materials and different types of curing agents through field monitoring and laboratory simulation;

[0034] (2.2) Accurately label the data, including the curing agent diffusion range, temperature change curve and actual hardness test results, to provide high-quality labeled data for model training;

[0035] (2.3) using data enhancement techniques to expand the training data set and improve the generalization ability of the model to different environments and conditions, wherein the data enhancement techniques include at least one of random rotation, scaling, and adding noise;

[0036] (2.4) Using a deep learning framework to train a prediction model, the deep learning framework is TensorFlow or PyTorch;

[0037] (2.5) The trained model is deployed on the edge computing device to process the collected infrared data in real time.

[0038] In order to better implement the method of the present invention, further, in step (3), the fused data is analyzed and processed by training the prediction model using the constructed deep learning framework, and the specific process of evaluating the diffusion range and depth of the chemical curing agent in the soft foundation and the influence on the foundation hardness is as follows:

[0039] (3.1) Preprocessing the collected infrared imaging data;

[0040] (3.2) Determine the diffusion range of the chemical curing agent in the soft foundation based on the preprocessed infrared imaging data;

[0041] (3.3) Analyze the temperature variation trend of soft foundation within the diffusion range of chemical curing agent;

[0042] (3.4) Infer the hardness change of the soft foundation based on the temperature change trend of the soft foundation.

[0043] In order to better implement the method of the present invention, further, in the step (4), based on the predicted optimal dosage of the chemical curing agent, a strategy for the placement of the chemical curing agent is provided, and the strategy for the placement of the chemical curing agent specifically includes the amount of the chemical curing agent, the rate of placement and the location of placement, and the specific process is as follows:

[0044] (4.1) Use PID controller to adjust the curing agent feeding rate according to hardness and temperature feedback to ensure a smooth feeding process and avoid sudden changes;

[0045] (4.2) Adaptive algorithms based on machine learning optimize the delivery strategy through historical data learning and automatically adjust the delivery volume and delivery speed according to the current foundation conditions;

[0046] (4.3) In view of the uneven diffusion, a multi-point delivery system is designed to control the delivery of curing agent in each area to achieve uniform curing in the entire area;

[0047] (4.4) Through a closed-loop control system, real-time feedback of temperature and hardness changes is applied to release control.

[0048] In order to better implement the method of the present invention, further, the remote monitoring and operation platform constructed in step (5) includes a front-end display layer, a business logic layer and a data storage layer;

[0049] The front-end display layer provides users with an intuitive monitoring dashboard through a web interface or mobile application, showing temperature change curves, hardness distribution diagrams, and curing agent delivery conditions. Users can view detailed information or historical data by clicking on a specific area;

[0050] The business logic layer adopts a microservice architecture to process core business logic; Docker containers are used to deploy various services to ensure the scalability and stability of the platform;

[0051] The data storage layer uses a distributed database to store a large amount of real-time data and historical records to ensure high data availability and fast query performance.

[0052] In order to better implement the method of the present invention, it further includes physical sampling of the soft foundation with the chemical curing agent placed therein to obtain the actual parameters of the soft foundation to verify and calibrate the infrared imaging data and fuse it with the infrared imaging data. The specific process is as follows:

[0053] (a) Arrange sampling points according to the type of foundation and the diffusion range of the curing agent. Each sampling point should cover a different foundation area to ensure the representativeness of the data;

[0054] (b) Collect soil samples using a geological sampler and determine the hardness, moisture content and density of the samples through laboratory analysis;

[0055] (c) spatially registering the physical sampling data with the infrared imaging data, and extending the sampling point data to the entire infrared image area by interpolation method;

[0056] (d) Using a weighted fusion algorithm, the infrared temperature data is combined with the sampled hardness data to generate a comprehensive evaluation model;

[0057] (e) Dynamically update the infrared imaging data based on the physical sampling data through the generated comprehensive evaluation model.

[0058] The main working principles of this technical solution are as follows:

[0059] 1. Deployment and data collection of infrared imaging equipment

[0060] 1.1 Selection and configuration of infrared imaging equipment: The core of infrared imaging equipment is a high-resolution infrared camera. The selection of the camera should be based on the thermal properties of the foundation material and the requirements of the construction environment, including the following configurations:

[0061] Resolution: Use an infrared camera with a resolution of 640×480 or higher to ensure that subtle temperature changes on the foundation surface can be captured.

[0062] Temperature measurement range: The camera’s temperature measurement range should cover the expected temperature changes during the foundation curing process (usually -20°C to 150°C).

[0063] Detection band: Select the appropriate infrared band (such as 3 microns to 5 microns or 8 microns to 14 microns) according to the transmittance and reflectivity of the foundation material to ensure the best imaging effect.

[0064] Environmental adaptability: The equipment must be dustproof and waterproof (IP65 or above) to adapt to harsh outdoor environmental conditions.

[0065] 1.2 Sensor calibration and installation: Before installing the equipment, perform a comprehensive sensor calibration to ensure data consistency between the temperature sensor and the infrared camera. The calibration process includes:

[0066] Blackbody calibration: Use a standard blackbody source to calibrate the camera’s temperature to ensure the accuracy of the device’s measurements.

[0067] On-site calibration: Perform secondary calibration at the actual construction site to adjust measurement errors caused by environmental factors.

[0068] When installing the equipment, the camera should be fixed on a stable bracket to ensure that the camera's field of view can cover the entire curing area and avoid blurry images caused by camera shaking. In order to expand the monitoring range, multiple cameras can be arranged to form a monitoring matrix, and overlapping perspectives can be used to compensate for the limitations of a single camera's perspective.

[0069] 1.3 Data acquisition frequency and storage: The data acquisition frequency is determined by the dynamics of the curing process. It is usually recommended to set it to 1 second / frame to 5 seconds / frame to ensure the timeliness of the data. All collected infrared imaging data should be transmitted to the edge computing device via a high-speed data link (such as Ethernet or optical fiber) and stored and backed up locally to prevent data loss.

[0070] 2. Real-time data analysis and processing

[0071] 2.1 Data preprocessing: In order to improve the accuracy of data analysis, a series of preprocessing operations are required before data processing:

[0072] Image denoising: Use median filtering or Gaussian filtering algorithm to denoise infrared images and eliminate the impact of environmental noise on image quality.

[0073] Temperature correction: Based on the data from the ambient temperature sensor, the temperature value in the infrared image is corrected to ensure that the reflected temperature changes are accurate.

[0074] Image registration: In the case of multi-view imaging, image registration is performed, using feature matching algorithms (such as SIFT or SURF) to ensure that images from different viewpoints can be accurately superimposed.

[0075] 2.2 Detection of curing agent diffusion range: The detection of diffusion range is based on image segmentation technology. The specific steps are as follows:

[0076] Region growing algorithm: Taking the initial hot spot area as the seed point, the detection area is gradually expanded by analyzing the temperature gradient of adjacent pixels to identify the boundary of curing agent diffusion.

[0077] Threshold segmentation: Set the temperature threshold based on experience, segment the infrared image, and extract the possible solidification area. Then use morphological operations (such as dilation and erosion) to optimize the segmentation results and remove noise and small pseudo areas.

[0078] Edge detection: The Canny edge detection algorithm is used to further accurately locate the actual boundary of the curing agent diffusion and generate a contour map of the contact between the foundation surface and the curing agent.

[0079] 2.3 Temperature change trend analysis: The analysis of temperature change is achieved through time series analysis and prediction model:

[0080] Sliding Window Analysis: The temperature change over time is analyzed by sliding window technique, where the window size is determined based on the typical time constant of the curing agent reaction.

[0081] ARIMA model prediction: Input temperature data into the ARIMA (autoregressive integrated moving average) model to predict future temperature change trends and identify turning points and anomalies in temperature changes.

[0082] Temperature gradient calculation: Perform gradient calculation on the temperature field in the infrared image to generate a temperature gradient distribution map, and further analyze the temperature diffusion direction and rate.

[0083] 2.4 Inference of foundation hardness change: The inference of hardness change is based on the relationship model between temperature change and foundation material properties, including:

[0084] Temperature-hardness mapping model: The mapping relationship between temperature and foundation hardness is established through experiments, and the hardness distribution of the current foundation is inferred in real time based on infrared data.

[0085] Local hardness assessment: Apply the temperature-hardness model to various areas of the infrared image to generate a hardness distribution map and determine the rate of change of hardness through differential analysis.

[0086] Regional hardness assessment: Interpolate the hardness data of multiple monitoring points to generate a hardness contour map of the entire curing area, providing a basis for construction quality control.

[0087] 3. Adaptive adjustment mechanism

[0088] 3.1 Delivery parameter adjustment strategy: According to the real-time analysis results, adjust the delivery amount and rate of the curing agent to optimize the curing effect:

[0089] PID control: Utilizes a proportional-integral-derivative (PID) controller to adjust the curing agent delivery rate based on hardness and temperature feedback to ensure a smooth delivery process and avoid sudden changes.

[0090] Adaptive delivery algorithm: Based on machine learning, the adaptive algorithm optimizes the delivery strategy through historical data learning, and automatically adjusts the delivery volume and delivery speed according to the current foundation conditions.

[0091] Multi-point delivery control: In view of the uneven diffusion situation, a multi-point delivery system is designed to control the delivery of curing agent in each area to achieve uniform curing in the whole area.

[0092] 3.2 Real-time feedback and adjustment: Through a closed-loop control system, real-time feedback of temperature and hardness changes is applied to delivery control. The specific steps include:

[0093] Instant feedback: The analysis results are transmitted to the controller to generate adjustment instructions and adjust the parameters of the delivery equipment.

[0094] · Delivery rate control: Real-time adjustment of the pump operating frequency to control the flow of the curing agent to ensure that it adapts to the absorption characteristics of the foundation material.

[0095] Automatic correction: During the curing process, the system continuously monitors the delivery effect and corrects the delivery parameters in real time to avoid over- or under-delivery of curing agent.

[0096] 4. Deep learning algorithm optimization

[0097] 4.1 Dataset construction and preprocessing: Before model training, a large-scale dataset covering a variety of foundation conditions and curing agent types needs to be established:

[0098] Data collection: Through field monitoring and laboratory simulation, collect temperature and hardness data of different foundation materials (such as sand, clay, gravel, etc.) and different types of curing agents.

[0099] Data annotation: Accurately annotate the data, including the curing agent diffusion range, temperature change curve and actual hardness test results, to provide high-quality annotated data for model training.

[0100] Data augmentation: Use data augmentation techniques (such as random rotation, scaling, and adding noise) to expand the training dataset and improve the model's generalization ability to different environments and conditions.

[0101] 4.2 Model training and optimization: Use a deep learning framework (such as TensorFlow or PyTorch) to train the prediction model. The specific steps are as follows:

[0102] Convolutional Neural Network (CNN) Architecture: Design a CNN architecture suitable for infrared image feature extraction, including multiple convolutional layers, pooling layers, and fully connected layers to extract spatial features in the image.

[0103] Model optimization: Prevent model overfitting and improve the model’s predictive performance through hyperparameter tuning (such as learning rate, batch size, network depth) and regularization (such as L2 regularization, Dropout).

[0104] Model validation and testing: Verify the accuracy of the model on an independent test set, evaluate the generalization ability of the model through cross-validation, and further optimize the model architecture based on the validation results.

[0105] 4.3 Real-time model application: The trained model is deployed on the edge computing device to process the collected infrared data in real time:

[0106] Real-time reasoning: New data is fed into the trained model to generate real-time strategic recommendations for curing agent delivery, including delivery amount, delivery rate, and delivery location.

[0107] Adaptive learning: During the curing process, model parameters are continuously updated and the model is adjusted based on new data to adapt to new situations or environmental changes that may arise.

[0108] 5. Remote monitoring and operation platform

[0109] 5.1 Platform architecture and design: The remote monitoring platform is designed based on cloud computing architecture and is divided into front-end display layer, business logic layer and data storage layer:

[0110] Front-end display layer: Provide users with intuitive monitoring dashboards through web interfaces or mobile applications to display real-time data, historical trends, and system status. The front-end is built using HTML5, JavaScript, and visualization libraries (such as D3.js) to ensure a good user experience.

[0111] Business logic layer: Adopt microservice architecture to process core business logic such as data analysis, equipment control, and alarm management. Use Docker containers to deploy each service to ensure the scalability and stability of the platform.

[0112] Data storage layer: Use distributed databases (such as Cassandra or HBase) to store large amounts of real-time data and historical records to ensure high data availability and fast query performance.

[0113] 5.2 Real-time data display and interaction: The main functions of the platform are real-time data display and user interaction:

[0114] Temperature and hardness real-time chart: Displays temperature change curve, hardness distribution and curing agent dosage. Users can view detailed information or historical data by clicking on a specific area.

[0115] ·3D visualization: WebGL technology is used to display the 3D model of the solidification area. Users can rotate and zoom the view to intuitively understand the solidification status inside the foundation.

[0116] Multi-user collaboration: Supports multiple users to log in and operate at the same time. The platform provides permission management function to ensure data security and operational independence.

[0117] 5.3 Remote control and automated operation: The platform supports remote control of curing equipment and provides automated operation options:

[0118] Manual control mode: Users can manually adjust the curing agent dosage parameters through the platform interface and view the adjustment effect in real time. The system provides parameter suggestions to assist users in making decisions.

[0119] Automated operation mode: In automated mode, the system automatically adjusts delivery parameters based on real-time analysis results and preset strategies. Users only need to monitor key indicators, reducing human intervention.

[0120] Alarm management: When the system detects an abnormal situation (such as abnormal temperature, uneven hardness distribution, etc.), it will automatically issue an alarm, and the user can choose to suspend the automated operation or intervene in manual control.

[0121] 5.4 Data storage and playback: The platform provides data storage and playback functions to support the analysis and evaluation of historical data:

[0122] Data storage: All monitoring data, operation logs and system status are securely stored in the platform’s distributed database and can be accessed by users at any time.

[0123] Data playback: Users can select monitoring data from a specific time period for playback to view historical changes in the curing process. The playback function supports acceleration, pause, and frame-by-frame viewing for detailed analysis.

[0124] 6. Fusion of infrared imaging and physical sampling data

[0125] 6.1 Physical sampling strategy and execution: The purpose of physical sampling is to obtain the actual hardness, moisture content and other parameters of the foundation to verify and calibrate the infrared imaging data:

[0126] Sampling point layout: According to the foundation type and the diffusion range of the curing agent, the sampling points should be arranged reasonably. Each sampling point should cover different foundation areas (such as surface, middle and deep layers) to ensure the representativeness of the data.

[0127] Sampling tools and methods: Use a geological sampler (such as a handheld drill) to collect soil samples, and determine the physical parameters of the samples such as hardness, moisture content and density through laboratory analysis.

[0128] 6.2 Data fusion and model calibration: The steps for fusing physical sampling data with infrared imaging data are as follows:

[0129] Data mapping: spatially align the physical sampling data with the infrared imaging data, and extend the sampling point data to the entire infrared image area through interpolation methods.

[0130] Multimodal fusion: A weighted fusion algorithm is used to combine infrared temperature data with sampled hardness data to generate a comprehensive evaluation model. This model is used to calibrate the temperature-hardness mapping relationship and improve the accuracy of infrared imaging inference.

[0131] Model update: Dynamically update the parameters of the infrared imaging model based on the physical sampling results, so that the model can better reflect the changes in the actual foundation.

[0132] 6.3 Comprehensive evaluation report generation: The results of data fusion are used to generate a comprehensive foundation solidification status evaluation report:

[0133] Report content: The assessment report includes the distribution of foundation hardness, diffusion range of curing agent, changes in moisture content, potential problem areas, etc., and provides visual charts and data analysis results.

[0134] Report output: Reports can be exported in PDF or other standard formats for engineers to make detailed evaluations and decision-making references. At the same time, reports can be directly shared with relevant team members through the remote monitoring platform.

[0135] 7. Dynamic hotspot tracking algorithm

[0136] 7.1 Hotspot Identification Technology: Hotspot identification is the key to dynamic tracking. The specific implementation includes:

[0137] Temperature gradient analysis: By calculating the temperature gradient of pixels in infrared images, areas with rapid temperature changes are identified and hot spots are preliminarily determined.

[0138] Region growing algorithm: Using the initially identified hotspot area as a seed, the region growing algorithm is used to expand the detection range and accurately locate the boundaries of the hotspot area.

[0139] Multi-scale analysis: Multi-scale analysis technology is used to identify hot spots at different scales, ensuring that small hot spots are not ignored while taking into account the accurate identification of large-scale hot spots.

[0140] 7.2 Hotspot tracking and updating: After identifying the hotspot, the system tracks the changes of the hotspot through the analysis of consecutive frames:

[0141] Kalman filter: Use Kalman filter to predict and update the hotspot location, filter out noise interference, and ensure the smoothness of the tracking path.

[0142] Region matching: Through the template matching algorithm, the matching points of the hot spot area are found in the continuous infrared image frames, and the movement, change and splitting of the hot spot area are tracked.

[0143] Hotspot update and merger: When a hotspot is split or merged, the system automatically updates the number and location of hotspots to ensure the real-time and accuracy of hotspot data.

[0144] 7.3 Heat balance and control: According to the results of hot spot tracking, the system adjusts the curing agent delivery strategy in real time:

[0145] Thermal Mapping: Generates real-time thermal profiles to analyze thermal uniformity across the curing area and identify areas of overheating or cold spots.

[0146] Adjustment of delivery strategy: For overheated areas, reduce the amount of curing agent delivered; for cold spots, increase the amount delivered or extend the delivery time to ensure uniformity of overall heat distribution.

[0147] Adaptive adjustment: According to the dynamic changes of hot spots, the system continuously adjusts the delivery strategy to ensure the heat balance and uniformity of the curing process.

[0148] 8. 3D infrared imaging technology

[0149] 8.1 Multi-view imaging settings: In order to obtain the three-dimensional temperature distribution of the curing area, infrared cameras need to be arranged at different positions:

[0150] Camera layout: Select multiple suitable camera positions according to the size and shape of the curing area to ensure multi-view imaging covering the entire area.

[0151] Synchronous control: All cameras should shoot synchronously to ensure the time consistency of data from different perspectives. Use a unified trigger system or timestamp to mark the synchronized data.

[0152] 8.2 Three-dimensional reconstruction algorithm: Use multi-view infrared image data to construct a three-dimensional temperature model of the curing area:

[0153] Stereo matching: Use stereo matching algorithms (such as SIFT matching based on feature points) to identify corresponding points between different perspectives and calculate depth information.

[0154] Volume rendering: The depth information is combined with the temperature data through the volume rendering algorithm to generate a three-dimensional temperature distribution map. Volume rendering can be achieved using techniques such as ray casting and slice stacking.

[0155] Temperature field analysis: Analyze the distribution of the three-dimensional temperature field, identify the abnormal temperature areas or unevenly solidified parts, and provide a basis for subsequent processing.

[0156] 8.3 3D model optimization and application: The 3D temperature model can be further optimized and applied to practical operations:

[0157] Noise filtering and correction: Through spatial filtering and morphological processing, the noise points in the model are eliminated and possible errors are corrected.

[0158] 3D thermal analysis: Use 3D models to conduct more in-depth thermal analysis, such as calculating parameters such as temperature gradient and heat flux density, and evaluating the diffusion efficiency of the curing agent.

[0159] Guide construction adjustments: Based on the 3D temperature distribution, adjust the curing agent delivery strategy, especially for cold spots or overheated areas identified in the model.

[0160] 9. Integration of infrared imaging data with geographic information system (GIS)

[0161] 9.1 Data Mapping and Coordinate Conversion: The first step in integrating infrared imaging data with GIS data is to perform coordinate conversion and data mapping:

[0162] Coordinate conversion: Convert pixel coordinates in infrared images to actual geographic coordinates, using geo-registration techniques (such as affine transformation) to ensure coordinate consistency.

[0163] Data mapping: Associate the converted temperature data with the geographic information layer in GIS (such as terrain and soil type) to generate a temperature-geographic information map.

[0164] 9.2GIS spatial analysis and optimization: Use the spatial analysis function of GIS to further optimize the solidification process:

[0165] · Terrain analysis: Analyze the slope, orientation and other characteristics of the curing area based on terrain data, and adjust the curing agent delivery strategy, such as increasing the delivery amount for slopes to ensure uniform coverage.

[0166] Soil type analysis: Combine soil type data to optimize the selection and placement of curing agents. Different soil types may require different curing agent formulas or dosages.

[0167] Spatial thermal map generation: Use GIS to generate a spatial thermal map of the curing area to intuitively display the spatial differences in temperature distribution and provide a basis for subsequent construction adjustments.

[0168] 9.3 Integration of GIS and remote monitoring platform: Integrate GIS data into the remote monitoring platform to enhance the comprehensive management capabilities of the system:

[0169] Real-time geographic display: The geographical information of the curing area is displayed in real time on the monitoring platform, including temperature distribution, terrain characteristics and soil type, etc. Users can intuitively view the status of each area on the map.

[0170] Regional management and planning: Regional management is carried out through GIS data, such as defining key monitoring areas, adjusting the curing agent release plan, and optimizing resource allocation.

[0171] Data sharing and analysis: Supports the export and sharing of GIS data, facilitating data exchange and comprehensive analysis with other systems (such as environmental monitoring systems).

[0172] 10. Intelligent early warning mechanism

[0173] 10.1 Design and training of anomaly detection model: The core of the intelligent early warning mechanism is the anomaly detection model. Its design and training steps include:

[0174] Abnormality definition and annotation: Based on actual experience and historical data, define abnormal conditions that may occur during the curing process, such as excessive temperature, insufficient hardness, etc. Detailed annotation is performed for the model training data to ensure that the model can identify these abnormalities.

[0175] Model selection: According to the characteristics of the anomaly detection task, choose a suitable model architecture, such as support vector machine (SVM), isolation forest, or autoencoder.

[0176] Model training and validation: Use labeled data to train the anomaly detection model and test the model’s performance on an independent validation set. Adjust model parameters through cross-validation to ensure high accuracy and low false positive rate of the model.

[0177] 10.2 Real-time monitoring and early warning signal generation: The anomaly detection model deployed in the system analyzes infrared imaging data in real time to identify potential problems:

[0178] Real-time detection: Input temperature, hardness and curing agent dosage data into the anomaly detection model to analyze and identify anomalies in real time. The model generates an early warning signal immediately when an anomaly is detected.

[0179] Warning level classification: According to the severity of the abnormality, the warning level is divided into low-level, medium-level and high-level warnings. Different levels of warnings correspond to different response strategies and processing priorities.

[0180] Multi-channel warning notification: Warning signals are transmitted to operators through multiple channels, such as real-time pop-up notifications on the platform, mobile phone text messages or emails, to ensure timely response.

[0181] 10.3 Automatic response strategy and manual intervention: After the system issues an early warning signal, it can automatically or manually take corresponding response measures:

[0182] Automatic response: In the automatic mode, the system automatically adjusts the curing agent delivery parameters according to the preset strategy, such as reducing the delivery amount or adjusting the delivery position to alleviate or solve abnormal problems.

[0183] Manual intervention: For high-level warnings or complex situations, operators can intervene manually through the platform, such as pausing automated operations, switching to manual control mode, or contacting on-site technicians for processing.

[0184] Response log recording: All warning response measures will be recorded in the system log, including warning content, response strategy and final results, to facilitate subsequent analysis and optimization.

[0185] 11. Combination of infrared imaging and drone technology

[0186] 11.1 UAV-mounted equipment and flight control: UAVs are equipped with infrared imaging equipment to achieve high-altitude and large-scale ground-based monitoring:

[0187] Equipment: Select lightweight infrared cameras and temperature sensors suitable for drones to ensure the flight stability and image quality of the drone. The equipment should have a long endurance to meet the needs of long-term flight.

[0188] Flight path planning: Develop the flight path of the drone based on the terrain characteristics of the solidification area and the construction plan. The flight path needs to cover all key areas and ensure the appropriate flight altitude and viewing angle to obtain the best imaging effect.

[0189] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0190] (1) The present invention provides a method for monitoring the curing of soft foundations using chemical curing agents, which uses high-performance edge computing devices to process infrared imaging data in real time. The system can accurately analyze the diffusion range and depth of the chemical curing agent and its real-time impact on the foundation hardness. Through filtering algorithms and pattern recognition technology, noise and interference data can be effectively eliminated;

[0191] (2) The present invention introduces an adaptive adjustment mechanism based on the feedback control principle, which can adjust the curing agent delivery parameters in real time to adapt to the physical properties and chemical reaction rates of different foundations, and dynamically optimize the curing process through predictive control algorithms, such as model-based predictive control;

[0192] (3) The present invention uses a large amount of infrared imaging data collected on site to train a deep neural network model to optimize the curing agent delivery strategy. The model can predict the optimal amount of curing agent under different conditions, reduce waste and improve the curing effect;

[0193] (4) The present invention provides a remote monitoring platform based on cloud computing, where operators can access the monitoring data of the curing process in real time through mobile devices or computers and make remote adjustments. The platform supports multi-user collaboration and data synchronization to ensure efficient communication and decision-making of the project team;

[0194] (5) In order to improve the accuracy of the assessment report, the present invention combines infrared imaging data with traditional physical sampling data, and generates a more comprehensive foundation solidification status assessment through multimodal data fusion technology to ensure the reliability and operability of the report;

[0195] (6) The present invention adopts a dynamic hot spot tracking algorithm, which can identify and track the hot spot area of ​​the chemical curing reaction in real time in infrared imaging data, and uses multi-scale segmentation and region growing algorithms to accurately determine the hot spot location and guide the distribution of the curing agent;

[0196] (7) The present invention adopts a three-dimensional reconstruction technology based on multi-view imaging, which can generate a three-dimensional temperature distribution map inside the foundation during the curing process; this technology provides a three-dimensional view of the distribution of the curing agent inside the foundation through multi-view data fusion and volume rendering algorithm, providing support for subsequent foundation strength assessment;

[0197] (8) The present invention innovatively integrates infrared imaging data with geographic information systems, optimizes the management and resource allocation of curing areas based on precise positioning and data analysis of geographic information. Through geospatial analysis, the planning efficiency of the curing process is further improved;

[0198] (9) The present invention develops an intelligent early warning mechanism based on infrared imaging data, which predicts potential curing problems, such as uneven curing or insufficient chemical reaction, through machine learning algorithms, and issues warning signals, allowing operators to adjust the curing process in time to avoid potential quality problems;

[0199] (10) The present invention combines drone technology and uses drones equipped with infrared imaging equipment to achieve large-scale, high-precision ground-based monitoring. The drone system is equipped with automatic navigation and mission planning functions, which is particularly suitable for inaccessible areas in the wild, improving monitoring flexibility and range. BRIEF DESCRIPTION OF THE DRAWINGS

[0200] Figure 1 This is a flow chart of foundation solidification monitoring based on infrared imaging technology of the present invention. DETAILED DESCRIPTION

[0201] In order to make the purpose, process conditions and advantages of the present invention more clearly understood, the present invention is further described in detail in conjunction with the following embodiments, but the embodiments of the present invention are not limited thereto. Without departing from the above-mentioned technical ideas of the present invention, various substitutions and changes are made according to the common technical knowledge and customary means in the field, which should all be included in the scope of the present invention. The specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0202] Embodiment 1:

[0203] This embodiment provides a foundation solidification monitoring method based on infrared imaging technology, comprising the following steps:

[0204] (1) Based on the temperature change generated during the reaction between the chemical curing agent and the soft foundation, an infrared imaging device is used to monitor the soft foundation to which the chemical curing agent is added in real time to obtain real-time infrared imaging data;

[0205] (2) Establish a large-scale database covering various foundation conditions and chemical curing agent types, and build a deep learning framework to train the prediction model based on this database;

[0206] (3) The infrared imaging data are analyzed and processed by training the prediction model through the constructed deep learning framework, the diffusion range and depth of the chemical curing agent in the soft foundation, and the impact on the foundation hardness are evaluated, and the optimal amount of chemical curing agent required for the soft foundation under different conditions is predicted;

[0207] (4) Based on the predicted optimal dosage of chemical curing agent, a strategy for the placement of chemical curing agent is given; during the process of placing chemical curing agent, the collected infrared imaging data will change, so the optimal dosage of chemical curing agent output by the constructed deep learning framework training will change dynamically, and then the corresponding chemical curing agent placement strategy is adaptively and dynamically generated, that is, for overheated areas, the amount of curing agent is reduced; for cold spot areas, the amount of curing agent is increased or the placement time is extended to ensure the heat balance and uniformity of the soft foundation curing process to adapt to possible new situations or environmental changes;

[0208] (5) Build a remote monitoring and operation platform to obtain the temperature change curve, hardness distribution diagram and curing agent placement status of the soft foundation with chemical curing agent, and remotely control the chemical curing agent placement equipment manually or automatically, and automatically issue an alarm when abnormal conditions in the soft foundation are detected.

[0209] Embodiment 2:

[0210] This embodiment further adds a physical sampling process based on the above embodiment, including the following steps:

[0211] (1) Based on the temperature change generated during the reaction between the chemical curing agent and the soft foundation, an infrared imaging device is used to monitor the soft foundation to which the chemical curing agent is added in real time to obtain real-time infrared imaging data;

[0212] (2) Physical sampling of the soft foundation with chemical curing agent to obtain the actual parameters of the soft foundation to verify and calibrate the infrared imaging data and fuse it with the infrared imaging data;

[0213] (3) Establish a large-scale database covering various foundation conditions and chemical curing agent types, and build a deep learning framework to train the prediction model based on this database;

[0214] (4) The fused data is analyzed and processed by training the prediction model through the constructed deep learning framework, the diffusion range and depth of the chemical curing agent in the soft foundation, and the impact on the foundation hardness are evaluated, and the optimal amount of chemical curing agent required for the soft foundation under different conditions is predicted;

[0215] (5) Based on the predicted optimal dosage of chemical curing agent, a strategy for the placement of chemical curing agent is given; during the process of placing chemical curing agent, the collected infrared imaging data will change, so the optimal dosage of chemical curing agent output by the constructed deep learning framework training will change dynamically, and then the corresponding chemical curing agent placement strategy is adaptively and dynamically generated, that is, for overheated areas, the amount of curing agent is reduced; for cold spot areas, the amount of curing agent is increased or the placement time is extended to ensure the heat balance and uniformity of the soft foundation curing process to adapt to possible new situations or environmental changes;

[0216] (6) Build a remote monitoring and operation platform to obtain the temperature change curve, hardness distribution diagram and curing agent placement status of the soft foundation with chemical curing agent, and remotely control the chemical curing agent placement equipment manually or automatically, and automatically issue an alarm when abnormal conditions in the soft foundation are detected.

[0217] Embodiment 3:

[0218] Based on the above embodiment, this embodiment further defines the infrared imaging device in step (1) as an infrared camera, the resolution of the infrared camera is 640×480, the temperature measurement range is -20°C to 150°C, and the measured infrared band is 3 microns to 5 microns or 8 microns to 14 microns. The rest of this embodiment is the same as the above embodiment and will not be repeated here.

[0219] Embodiment 4:

[0220] Based on the above embodiment, this embodiment further defines the specific process of step (1) collecting infrared imaging data through an infrared camera as follows:

[0221] (1.1) Use a standard black source to perform preliminary calibration of the infrared camera;

[0222] (1.2) Fix the infrared camera on the drone through a bracket, use the drone to test fly at the monitoring site, and perform the second calibration of the infrared camera;

[0223] (1.3) A monitoring matrix is ​​formed by multiple drones equipped with infrared cameras, so that the infrared cameras can form infrared imaging data collection with overlapping viewing angles; the specific process is as follows:

[0224] (1.3.1) Choose a drone with long endurance to ensure the flight stability and image quality of the drone;

[0225] (1.3.2) Develop the flight path of the drone based on the terrain characteristics of the solidification area and the construction plan; the flight path must cover all key areas and ensure the drone's flight altitude and viewing angle to obtain the best imaging effect;

[0226] (1.3.3) The drone is equipped with automatic navigation function, combining GPS and inertial navigation system to achieve accurate positioning and path tracking; at the same time, it uses laser radar or visual sensors to achieve real-time obstacle avoidance to ensure flight safety;

[0227] (1.3.4) During the flight of the UAV, the ground temperature data is continuously collected and the collected infrared image data is transmitted to the ground station in real time through the wireless communication link. The data should be synchronously marked with timestamps to ensure the consistency of subsequent processing;

[0228] (1.3.5) The basic computing unit on the drone can perform basic denoising and compression on the preliminary data, reducing the transmission load and improving data processing efficiency;

[0229] (1.3.6) After the drone flight is completed, the system generates a temperature distribution map covering the entire curing area and identifies areas of abnormal temperature or uneven curing;

[0230] (1.3.7) For the identified abnormal areas, the drone performs low-altitude high-resolution scanning to obtain more detailed temperature data and solidification conditions for further analysis and processing;

[0231] (1.3.8) Integrate the data collected by the drone with the data from the ground infrared imaging equipment to generate a global curing assessment report, providing more comprehensive and accurate curing status information;

[0232] (1.3.9) All data collected by the drones are uploaded to the constructed remote monitoring and operation platform for storage. The operator can monitor the flight status and data collection progress of the drones in real time through the remote monitoring and operation platform, and adjust the flight parameters or mission plan as needed;

[0233] (1.4) Infrared imaging data is collected from the soft foundation with the chemical curing agent added at a collection frequency of 1 second / frame to 5 seconds / frame, and then transmitted to the edge computing device via a high-speed data link for data storage. The high-speed data link includes Ethernet and optical fiber. The specific process is as follows:

[0234] (1.4.1) By calculating the temperature gradient of pixels in the infrared image, the area with sharp temperature changes can be identified and the hot spot location can be preliminarily determined;

[0235] (1.4.2) Using the initially identified hotspot area as a seed, the region growing algorithm is used to expand the detection range and accurately locate the boundaries of the hotspot area;

[0236] (1.4.3) Use multi-scale analysis technology to identify hotspots at different scales, ensuring that small hotspots are not ignored while taking into account the accurate identification of large-scale hotspots;

[0237] (1.4.4) After identifying the hotspot, the changes of the hotspot are tracked through the analysis of continuous frames, specifically including: using the Kalman filter to predict and update the hotspot position, filtering out noise interference, and ensuring the smoothness of the tracking path; using the template matching algorithm to find matching points in the hotspot area in continuous infrared image frames, and tracking the movement, change and splitting of the hotspot area; when the hotspot splits or merges, the system automatically updates the number and position of the hotspot to ensure the real-time and accuracy of the hotspot data. The rest of this embodiment is the same as the above embodiment and will not be repeated here.

[0238] Embodiment 5:

[0239] Based on the above embodiment, this embodiment further defines the specific process of performing physical sampling on the soft foundation with the chemical curing agent in step (2) to obtain the actual parameters of the soft foundation to verify and calibrate the infrared imaging data, and to fuse the infrared imaging data as follows:

[0240] (2.1) Arrange sampling points according to the foundation type and the diffusion range of the curing agent. Each sampling point should cover a different foundation area to ensure the representativeness of the data;

[0241] (2.2) Collect soil samples using a geological sampler and determine the hardness, moisture content and density of the samples through laboratory analysis;

[0242] (2.3) spatially aligning the physical sampling data with the infrared imaging data, and extending the sampling point data to the entire infrared image area through interpolation methods;

[0243] (2.4) Using a weighted fusion algorithm, the infrared temperature data is combined with the sampled hardness data to generate a comprehensive evaluation model;

[0244] (2.5) Dynamically update the infrared imaging data based on the generated comprehensive evaluation model and physical sampling data. The other parts of this embodiment are the same as those of the above embodiment and will not be described again.

[0245] Embodiment 6:

[0246] Based on the above embodiment, this embodiment further defines the step (3) of establishing a large-scale database covering various foundation conditions and chemical curing agent types, and the specific process of constructing a deep learning framework training prediction model based on the database is as follows:

[0247] (3.1) Collect temperature and hardness data of different foundation materials and different curing agent types through field monitoring and laboratory simulation;

[0248] (3.2) Accurately label the data, including the curing agent diffusion range, temperature change curve and actual hardness test results, to provide high-quality labeled data for model training;

[0249] (3.3) Using data enhancement technology to expand the training data set and improve the generalization ability of the model to different environments and conditions, the data enhancement technology includes at least one of random rotation, scaling, and adding noise;

[0250] (3.4) Using a deep learning framework to train the prediction model, the deep learning framework is TensorFlow or PyTorch; including:

[0251] (3.4.1) Design a CNN architecture suitable for infrared image feature extraction, including multiple convolutional layers, pooling layers, and fully connected layers to extract spatial features in the image;

[0252] (3.4.2) Preventing model overfitting and improving the prediction performance of the model by hyperparameter tuning and regularization; the hyperparameters include at least one of learning rate, batch size, and network depth; the regularization includes L2 regularization or Dropout;

[0253] (3.4.3) Verify the accuracy of the model on an independent test set, evaluate the generalization ability of the model through cross-validation, and further optimize the model architecture based on the verification results;

[0254] (3.5) The trained model is deployed on the edge computing device to process the collected infrared data in real time. The rest of this embodiment is the same as the above embodiment and will not be described here.

[0255] Embodiment 7:

[0256] Based on the above embodiment, this embodiment further defines the specific process of analyzing and processing the fused data by training the prediction model through the constructed deep learning framework in step (4) to evaluate the diffusion range and depth of the chemical curing agent in the soft foundation, and the influence on the foundation hardness as follows:

[0257] (4.1) Preprocessing of the collected fusion data, including:

[0258] (4.1.1) Use median filtering or Gaussian filtering algorithm to denoise the fused data image to eliminate the impact of environmental noise on image quality;

[0259] (4.1.2) In the case of multi-view imaging, image registration is performed, and a feature matching algorithm is used to ensure that images from different viewpoints can be accurately superimposed; the feature matching algorithm is a SIFT method or a SURF method;

[0260] (4.2) Determine the diffusion range of the chemical curing agent in the soft foundation based on the preprocessed fusion data. The specific process includes:

[0261] (4.2.1) Taking the initial hot spot area as the seed point, the detection area is gradually expanded by analyzing the temperature gradient of adjacent pixels to identify the boundary of curing agent diffusion;

[0262] (4.2.2) Set the temperature threshold based on experience, segment the infrared image, and extract the possible solidification area; then use morphological operations to optimize the segmentation results and remove noise and small pseudo areas;

[0263] (4.2.3) Using the Canny edge detection algorithm, the actual boundary of the curing agent diffusion is further accurately located to generate a contour map of the contact between the foundation surface and the curing agent;

[0264] (4.3) Analyze the temperature variation trend of the soft foundation within the diffusion range of the chemical curing agent, including:

[0265] (4.3.1) Analyze the temperature change over time by sliding window technique, where the window size is determined based on the typical time constant of the curing agent reaction;

[0266] (4.3.2) Input the temperature data into the autoregressive integrated moving average model to predict the future temperature change trend and identify the turning points and anomalies of temperature change;

[0267] (4.3.3) Calculate the gradient of the temperature field in the infrared image, generate a temperature gradient distribution map, and further analyze the temperature diffusion direction and rate;

[0268] (4.4) Infer the hardness change of the soft foundation based on the temperature change trend of the soft foundation, including:

[0269] (4.4.1) Establish the mapping relationship between temperature and foundation hardness through experiments, establish a temperature-hardness model, and infer the current foundation hardness distribution in real time based on infrared data;

[0270] (4.4.2) Apply the temperature-hardness model to each area of ​​the infrared image to generate a hardness distribution map, and determine the rate of change of hardness through differential analysis;

[0271] (4.4.3) Interpolate the hardness data of multiple monitoring points to generate a hardness contour map of the entire curing area to provide a basis for construction quality control. The rest of this embodiment is the same as the above embodiment and will not be repeated here.

[0272] Embodiment 8:

[0273] On the basis of the above-mentioned embodiment, this embodiment further defines the optimal dosage of the chemical curing agent in step (5) based on the prediction, and provides a strategy for the placement of the chemical curing agent. The strategy for the placement of the chemical curing agent specifically includes the amount, rate and position of the chemical curing agent. The specific process is as follows:

[0274] (5.1) Use PID controller to adjust the curing agent feeding rate according to hardness and temperature feedback to ensure a smooth feeding process and avoid sudden changes;

[0275] (5.2) Adaptive algorithms based on machine learning optimize the delivery strategy through historical data learning and automatically adjust the delivery volume and delivery speed according to the current foundation conditions;

[0276] (5.3) In view of the uneven diffusion, a multi-point delivery system is designed to control the delivery of curing agent in each area to achieve uniform curing in the entire area;

[0277] (5.4) Through a closed-loop control system, real-time feedback of temperature and hardness changes is applied to delivery control, including:

[0278] (5.4.1) Transmit the analysis results to the controller, generate adjustment instructions, and adjust the parameters of the delivery equipment;

[0279] (5.4.2) Adjust the operating frequency of the pump in real time to control the flow rate of the curing agent to ensure that it adapts to the absorption characteristics of the foundation material;

[0280] (5.4.3) During the curing process, the system continuously monitors the delivery effect and corrects the delivery parameters in real time to avoid over- or under-delivery of the curing agent. The rest of this embodiment is the same as the above embodiment and will not be described again.

[0281] Embodiment 9:

[0282] Based on the above embodiment, this embodiment further defines that the remote monitoring and operation platform constructed in step (6) includes a front-end display layer, a business logic layer and a data storage layer;

[0283] The front-end display layer provides users with an intuitive monitoring dashboard through a Web interface or mobile application, showing the temperature change curve, hardness distribution diagram and curing agent delivery status. Users can view detailed information or historical data by clicking on a specific area. The front-end display layer is built using HTML5, JavaScript and a visualization library. WebGL technology is used to display a three-dimensional model of the curing area. Users can rotate and zoom the view to intuitively understand the curing status inside the foundation. It supports multiple users to log in and operate at the same time, and the platform provides permission management functions to ensure data security and independence of operations.

[0284] The business logic layer adopts a microservice architecture to process core business logic; uses Docker containers to deploy various services to ensure the scalability and stability of the platform; the business logic layer controls the drone route to select multiple suitable infrared camera positions according to the size and shape of the curing area to ensure multi-view imaging covering the entire area, and controls all infrared cameras to shoot synchronously to ensure the time consistency of data from different perspectives; uses a unified trigger system or timestamp to mark synchronized data; uses multi-view infrared image data to build a three-dimensional temperature model of the curing area, which specifically includes using a stereo matching algorithm to identify corresponding points between different perspectives and calculate depth information. The matching algorithm is the SIFT matching algorithm based on feature points. The depth information is combined with the temperature data through the volume rendering algorithm to generate a three-dimensional temperature distribution map. The volume rendering is achieved using ray casting technology or slice stacking technology. The distribution of the three-dimensional temperature field is analyzed to identify the temperature abnormality area or the unevenly solidified part, providing a basis for subsequent processing. Then, through spatial filtering and morphological processing, the noise points in the model are eliminated and possible errors are corrected. The three-dimensional model is used for more in-depth thermal analysis, such as calculating the temperature gradient and heat flux density, and evaluating the diffusion efficiency of the curing agent. Based on the three-dimensional temperature distribution, the curing agent delivery strategy is adjusted, especially for the cold spots or overheated areas identified in the model.

[0285] The business logic layer also has a geographic information system module, which integrates the fusion data with the geographic information system. The specific contents include coordinate conversion and data mapping. The coordinate conversion is to convert the pixel coordinates in the infrared image into actual geographic coordinates, and use geographic registration technology to ensure the consistency of the coordinates. The data mapping is to associate the converted temperature data with the geographic information layer in the geographic information system to generate a temperature-geographic information mapping diagram; the spatial analysis function of the geographic information system is used to analyze the geographical characteristics of the curing area according to the terrain data, and adjust the curing agent delivery strategy; combined with soil type data, the selection and delivery method of the curing agent are optimized. Different soil types may require different curing agent formulas or delivery amounts; the spatial heat map of the curing area is generated by the geographic information system to intuitively display the spatial differences in temperature distribution and provide a basis for subsequent construction adjustments; and the geographic information of the curing area, including temperature distribution, terrain characteristics and soil types, is displayed in real time on the monitoring platform, and users can intuitively view the status of each area on the map; regional management is carried out through geographic information system data, such as demarcating key monitoring areas, adjusting the curing agent delivery plan, optimizing resource allocation, supporting the export and sharing of geographic information system data, and facilitating data exchange and comprehensive analysis with other systems;

[0286] The business logic layer also has an intelligent alarm module, which includes an anomaly detection model. The intelligent alarm module defines the abnormal situations that may occur in the curing process based on actual experience and historical data, and annotates the training data of the anomaly detection model in detail to ensure that the model can recognize these anomalies; according to the characteristics of the anomaly detection task, the anomaly detection model selects a suitable model architecture, which is at least one of a support vector machine, an isolation forest or an autoencoder; the anomaly detection model is trained using annotated data, and the performance of the model is tested on an independent validation set; the parameters of the anomaly detection model are adjusted through cross-validation to ensure the high accuracy and low false alarm rate of the anomaly detection model; the anomaly detection model analyzes infrared imaging data in real time to identify potential problems, specifically including inputting temperature, hardness and curing agent delivery data into the anomaly detection model, analyzing and identifying anomalies in real time, and the anomaly detection model immediately generates an anomaly detection error when an anomaly is detected. The system generates an early warning signal; the early warning level is divided according to the severity of the abnormality, and different levels of early warnings correspond to different response strategies and processing priorities; the early warning signal is transmitted to the operator through multiple channels, including real-time pop-up notifications, mobile phone text messages or emails on the remote monitoring and operation platform to ensure timely response; after the early warning signal is issued, corresponding response measures can be taken automatically or manually. The automatic response is that in the automatic mode, the system automatically adjusts the curing agent delivery parameters according to the preset strategy, such as reducing the delivery amount or adjusting the delivery position to alleviate or solve the abnormal problem; the manual intervention is for high-level early warnings or complex situations. The operator can perform manual intervention through the platform, such as suspending automatic operation, switching to manual control mode or contacting on-site technicians for processing; all early warning response measures will be recorded in the system log, including the early warning content, response strategy and final result, to facilitate subsequent analysis and optimization;

[0287] The data storage layer uses a distributed database to store a large amount of real-time data and historical records to ensure high data availability and fast query performance; the distributed database is Cassandra or Hbase; the data storage layer provides data storage and data playback functions to support the analysis and evaluation of historical data. The data storage is that all monitoring data, operation logs and system status are securely stored in the distributed database of the platform, and users can access them at any time. The data playback is that users can select monitoring data of a specific time period for playback to view the historical changes of the curing process. The playback function supports acceleration, pause and frame-by-frame viewing for detailed analysis. The other parts of this embodiment are the same as the above embodiment and will not be repeated here.

[0288] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A foundation solidification monitoring method based on infrared imaging technology, characterized in that: The following steps are involved: (1) Based on the temperature change generated during the reaction between the chemical curing agent and the soft foundation, an infrared imaging device is used to monitor the soft foundation to which the chemical curing agent is added in real time to obtain real-time infrared imaging data; (2) Establish a large-scale database covering various foundation conditions and chemical curing agent types, and build a deep learning framework to train the prediction model based on this database; (3) The infrared imaging data are analyzed and processed by training the prediction model through the constructed deep learning framework, the diffusion range and depth of the chemical curing agent in the soft foundation, and the impact on the foundation hardness are evaluated, and the optimal amount of chemical curing agent required for the soft foundation under different conditions is predicted; (4) Based on the predicted optimal dosage of chemical curing agent, a strategy for the placement of chemical curing agent is given; during the process of placing chemical curing agent, the collected infrared imaging data will change, so the optimal dosage of chemical curing agent output by the constructed deep learning framework training will change dynamically, and then the corresponding chemical curing agent placement strategy is adaptively and dynamically generated, that is, for overheated areas, the amount of curing agent is reduced; for cold spot areas, the amount of curing agent is increased or the placement time is extended to ensure the heat balance and uniformity of the soft foundation curing process to adapt to possible new situations or environmental changes; (5) Build a remote monitoring and operation platform to obtain the temperature change curve, hardness distribution diagram and curing agent placement status of the soft foundation with chemical curing agent, and remotely control the chemical curing agent placement equipment manually or automatically, and automatically issue an alarm when abnormal conditions in the soft foundation are detected.

2. The method for monitoring foundation solidification based on infrared imaging technology according to claim 1, characterized in that: The infrared imaging device in step (1) is an infrared camera, the resolution of the infrared camera is 640×480, the temperature measurement range is -20°C to 150°C, and the measured infrared band is 3 microns to 5 microns or 8 microns to 14 microns.

3. The method for monitoring foundation solidification based on infrared imaging technology according to claim 2, characterized in that: The specific process of step (1) collecting infrared imaging data by using an infrared camera is as follows: (1.1) Use a standard black source to perform preliminary calibration of the infrared camera; (1.2) Fix the infrared camera on the drone through a bracket, use the drone to test fly at the monitoring site, and perform the second calibration of the infrared camera; (1.3) A monitoring matrix is ​​formed by using multiple drones equipped with infrared cameras, so that the infrared cameras can form infrared imaging data collection with overlapping viewing angles; (1.4) Infrared imaging data is collected on a soft foundation with a chemical curing agent added using an acquisition frequency of 1 second / frame to 5 seconds / frame, and then transmitted to an edge computing device via a high-speed data link for data storage, wherein the high-speed data link includes Ethernet and optical fiber.

4. The method for monitoring foundation solidification based on infrared imaging technology according to claim 3 is characterized in that: In the step (1.3), a monitoring matrix is ​​formed by using multiple drones equipped with infrared cameras so that the infrared cameras can form infrared imaging data acquisition with overlapping viewing angles. The specific process is as follows: (1.3.1) Choose a drone with long endurance to ensure the flight stability and image quality of the drone; (1.3.2) Develop the flight path of the drone based on the terrain characteristics of the solidification area and the construction plan; the flight path must cover all key areas and ensure the drone's flight altitude and viewing angle to obtain the best imaging effect; (1.3.3) The drone is equipped with automatic navigation function, combining GPS and inertial navigation system to achieve accurate positioning and path tracking; at the same time, it uses laser radar or visual sensors to achieve real-time obstacle avoidance to ensure flight safety; (1.3.4) During the flight of the UAV, the ground temperature data is continuously collected and the collected infrared image data is transmitted to the ground station in real time through the wireless communication link. The data should be synchronously marked with timestamps to ensure the consistency of subsequent processing; (1.3.5) The basic computing unit on the drone can perform basic denoising and compression on the preliminary data, reducing the transmission load and improving data processing efficiency; (1.3.6) After the drone flight is completed, the system generates a temperature distribution map covering the entire curing area and identifies areas of abnormal temperature or uneven curing; (1.3.7) For the identified abnormal areas, the drone performs low-altitude high-resolution scanning to obtain more detailed temperature data and solidification conditions for further analysis and processing; (1.3.8) Integrate the data collected by the drone with the data from the ground infrared imaging equipment to generate a global curing assessment report, providing more comprehensive and accurate curing status information; (1.3.9) All data collected by the drones are uploaded to the constructed remote monitoring and operation platform for storage. The operator can monitor the flight status and data collection progress of the drone in real time through the remote monitoring and operation platform, and adjust the flight parameters or mission plan as needed.

5. The method for monitoring foundation solidification based on infrared imaging technology according to claim 4, characterized in that: In the step (1.4), the infrared imaging data acquisition uses hotspot recognition technology to dynamically track the soft foundation to which the chemical curing agent is added, and the specific process is as follows: (1.4.1) By calculating the temperature gradient of pixels in the infrared image, the area with sharp temperature changes can be identified and the hot spot location can be preliminarily determined; (1.4.2) Using the initially identified hotspot area as a seed, the region growing algorithm is used to expand the detection range and accurately locate the boundaries of the hotspot area; (1.4.3) Use multi-scale analysis technology to identify hotspots at different scales, ensuring that small hotspots are not ignored while taking into account the accurate identification of large-scale hotspots; (1.4.4) After identifying the hotspot, the changes of the hotspot are tracked through the analysis of continuous frames, including: using the Kalman filter to predict and update the hotspot position, filtering out noise interference, and ensuring the smoothness of the tracking path; using the template matching algorithm to find matching points in the hotspot area in continuous infrared image frames, and tracking the movement, changes and splitting of the hotspot area; when the hotspot splits or merges, the system automatically updates the number and position of the hotspot to ensure the real-time and accuracy of the hotspot data.

6. The method for monitoring foundation solidification based on infrared imaging technology according to claim 5, characterized in that: In step (2), a large-scale database covering various foundation conditions and chemical curing agent types is established. The specific process of constructing a deep learning framework training prediction model based on the database is as follows: (2.1) Collect temperature and hardness data of different foundation materials and different types of curing agents through field monitoring and laboratory simulation; (2.2) Accurately label the data, including the curing agent diffusion range, temperature change curve and actual hardness test results, to provide high-quality labeled data for model training; (2.3) using data enhancement techniques to expand the training data set and improve the generalization ability of the model to different environments and conditions, wherein the data enhancement techniques include at least one of random rotation, scaling, and adding noise; (2.4) Using a deep learning framework to train a prediction model, the deep learning framework is TensorFlow or PyTorch; (2.5) The trained model is deployed on the edge computing device to process the collected infrared data in real time.

7. The method for monitoring foundation solidification based on infrared imaging technology according to claim 6, characterized in that: In step (3), the fused data is analyzed and processed by training the prediction model using the constructed deep learning framework, and the specific process of evaluating the diffusion range and depth of the chemical curing agent in the soft foundation and its influence on the foundation hardness is as follows: (3.1) Preprocessing the collected infrared imaging data; (3.2) Determine the diffusion range of the chemical curing agent in the soft foundation based on the preprocessed infrared imaging data; (3.3) Analyze the temperature variation trend of soft foundation within the diffusion range of chemical curing agent; (3.4) Infer the hardness change of the soft foundation based on the temperature change trend of the soft foundation.

8. The method for monitoring foundation solidification based on infrared imaging technology according to claim 7, characterized in that: In the step (4), based on the predicted optimal dosage of the chemical curing agent, a strategy for the placement of the chemical curing agent is provided. The strategy for the placement of the chemical curing agent specifically includes the amount, rate and position of the chemical curing agent to be placed. The specific process is as follows: (4.1) Use PID controller to adjust the curing agent feeding rate according to hardness and temperature feedback to ensure a smooth feeding process and avoid sudden changes; (4.2) Adaptive algorithms based on machine learning optimize the delivery strategy through historical data learning and automatically adjust the delivery volume and delivery speed according to the current foundation conditions; (4.3) In view of the uneven diffusion, a multi-point delivery system is designed to control the delivery of curing agent in each area to achieve uniform curing in the entire area; (4.4) Through a closed-loop control system, real-time feedback of temperature and hardness changes is applied to release control.

9. The method for monitoring foundation solidification based on infrared imaging technology according to claim 8, characterized in that: The remote monitoring and operation platform constructed in step (5) includes a front-end display layer, a business logic layer, and a data storage layer; The front-end display layer provides users with an intuitive monitoring dashboard through a web interface or mobile application, showing temperature change curves, hardness distribution diagrams, and curing agent delivery conditions. Users can view detailed information or historical data by clicking on a specific area; The business logic layer adopts a microservice architecture to process core business logic; Docker containers are used to deploy various services to ensure the scalability and stability of the platform; The data storage layer uses a distributed database to store a large amount of real-time data and historical records to ensure high data availability and fast query performance.

10. A foundation solidification monitoring method based on infrared imaging technology according to claims 1 to 9, characterized in that: It also includes physical sampling of the soft foundation with chemical curing agent, obtaining the actual parameters of the soft foundation, verifying and calibrating the infrared imaging data, and fusing it with the infrared imaging data. The specific process is as follows: (a) Arrange sampling points according to the type of foundation and the diffusion range of the curing agent. Each sampling point should cover a different foundation area to ensure the representativeness of the data; (b) Collect soil samples using a geological sampler and determine the hardness, moisture content and density of the samples through laboratory analysis; (c) spatially registering the physical sampling data with the infrared imaging data, and extending the sampling point data to the entire infrared image area by interpolation method; (d) Using a weighted fusion algorithm, the infrared temperature data is combined with the sampled hardness data to generate a comprehensive evaluation model; (e) Dynamically update the infrared imaging data based on the physical sampling data through the generated comprehensive evaluation model.

Images