Waterproof engineering leakage intelligent positioning diagnosis method and system

Abstract

The present application relates to the field of engineering detection technology, in particular to a waterproof engineering leakage intelligent positioning diagnosis method and system, comprising: acquiring an infrared thermal imaging image sequence of a waterproof area to be detected, collecting environmental parameter information at the corresponding time, and binding the infrared thermal imaging image and the environmental parameter information at the same time as a current detection data group; performing heat conduction analysis on the infrared thermal imaging image sequence in the current detection data group to generate initial temperature field distribution data and temperature gradient change information corresponding to the initial temperature field distribution data, and correcting the initial temperature field distribution data using the environmental parameter information to obtain modified temperature field distribution data. The present application can more accurately identify and locate the leakage situation of the waterproof area by combining the infrared thermal imaging image sequence and the environmental parameter information, improve the detection accuracy, and use heat conduction analysis and modified temperature field distribution data to reflect the temperature change of the waterproof area in real time.

Description

Technical Field

[0001] This invention relates to the field of engineering testing technology, specifically to an intelligent positioning and diagnostic method and system for leakage in waterproofing projects. Background Technology

[0002] Currently, traditional methods rely heavily on manual inspection or simple visual observation, which are easily affected by subjective factors, resulting in inaccurate test results and the possibility of missing hidden leaks. Moreover, many traditional detection methods require a long time to conduct a comprehensive inspection and cannot reflect the status of the waterproof area in real time. This makes it impossible to take measures quickly when leaks occur, which may lead to more serious damage.

[0003] Furthermore, traditional methods often lack effective data integration and analysis tools, making it difficult to comprehensively consider temperature distribution, leakage location, and historical data. This results in insufficient scientific rigor and rationality in model training, thus affecting the reliability of detection. Moreover, traditional methods often fail to effectively predict potential leakage problems, leading to a lack of foresight in maintenance work. Instead, they can only be dealt with passively after problems occur, increasing maintenance costs. Additionally, traditional detection methods are highly sensitive to environmental interference, making it difficult to effectively eliminate noise and anomalies, which can easily lead to misjudgments and missed detections, reducing the effectiveness of detection. Summary of the Invention

[0004] To achieve the above objectives, the present invention provides the following technical solution: an intelligent positioning and diagnosis method for leakage in waterproofing projects, comprising: Acquire the infrared thermal imaging image sequence of the waterproof area to be detected, collect the environmental parameter information at the corresponding time, and bind the infrared thermal imaging image and environmental parameter information at the same time into the current detection data group; Thermal conduction analysis is performed on the infrared thermal imaging image sequence in the current detection data set to generate initial temperature field distribution data and temperature gradient change information corresponding to the initial temperature field distribution data. The thermal conduction coefficient of the initial temperature field distribution data is corrected using environmental parameter information to obtain corrected temperature field distribution data. Based on the corrected temperature field distribution data and temperature gradient change information, combined with historical infrared image information, the thermal anomaly trend of the waterproofing project is predicted to obtain the trend data to be diagnosed. The current thermal radiation characteristics, historical thermal radiation characteristics and external heat exchange constraints are used as constraints on the trend data to be diagnosed. The initial temperature field distribution data and temperature gradient change information are optimized with weights to obtain the thermal field data to be optimized. The temperature change rate of each local thermal anomaly region in the thermal field data to be optimized is used for posterior estimation, and the weight value to be detected is assigned. The weight value to be detected is then used to detect large noise anomalies, and local thermal anomaly regions that do not meet the preset temperature change threshold are removed to obtain the target thermal anomaly region and the corresponding target weight value. The thermal field data to be optimized, all target thermal anomaly areas and target weight values ​​are input into the pre-established intelligent leakage location and diagnosis model. Based on the target weight values, all target thermal anomaly areas are spliced ​​together globally according to the thermal field data to be optimized to obtain the target leakage location data and global thermal distribution information of the waterproof area to be detected. Based on the target leakage location data, global heat distribution information, and corrected temperature field distribution data, a total loss function is established. The leakage intelligent location and diagnosis model is iteratively trained with the goal of minimizing the total loss function until the leakage location output by the model converges, thus obtaining the trained leakage intelligent location and diagnosis model.

[0005] Preferably, the initial temperature field distribution data is corrected for thermal conductivity using environmental parameter information to obtain corrected temperature field distribution data, including: Based on the infrared image information at the current time, the infrared image information at the historical time, and the environmental parameter information, calculate the thermal convection error information and thermal radiation error information of the waterproof area to be detected between the current time and the historical time. Based on the ambient temperature information, the first correction calculation is performed on the thermal convection error information and thermal radiation information to obtain the change in surface heat flux density of the waterproof area to be tested between the current time and historical time. Based on the environmental humidity information, a second correction calculation is performed on the thermal convection error information and thermal radiation information to obtain the change in internal thermal diffusion of the waterproof area to be tested between the current time and historical time. Based on thermal radiation information, the thermal conductivity coefficient information at different depths is determined. Based on the thermal conductivity coefficient information at different depths, a third correction calculation is performed on the thermal radiation information and thermal radiation error information to obtain the change in thermal conductivity of the waterproof area to be tested between the current time and historical time. The changes in surface heat flux density, internal heat diffusion, and heat conduction are used as the changes in the thermal field. The changes in the thermal field are then superimposed with the initial temperature field distribution data from the previous moment to obtain the corrected temperature field distribution data.

[0006] Preferably, generating temperature gradient change information corresponding to the initial temperature field distribution data includes: Count the actual temperature extraction count of all pixels, and define the maximum temperature acquisition count corresponding to all pixels within the time window between the current time and historical time. The temperature information corresponding to all pixels is projected into the prior heat conduction model of the environmental parameter sensor to obtain the predicted temperature information, and the temperature difference error information between the predicted temperature information and the current temperature information is calculated. Based on the temperature difference error information, the maximum number of temperature samples collected, and the actual number of temperature samples extracted, temperature gradient change information is generated.

[0007] Preferably, based on the target leakage location data, global heat distribution information, and corrected temperature field distribution data, a total loss function is established, including: A first loss function is established based on the corrected temperature field distribution data and the thermal field data to be optimized, which is used to constrain the consistency of the thermal field distribution. A second loss function is established based on the target leakage location data and global heat distribution information to constrain the positioning accuracy of the leakage coordinates; A third loss function is established based on the target leakage location data and the corrected temperature field distribution data to constrain the accuracy of temperature distribution prediction. Weight coefficients are set for the first loss function, the second loss function, and the third loss function, and a total loss function is established based on the three loss functions and their weight coefficients.

[0008] Preferably, a second loss function is established based on the target leakage location data and global heat distribution information, including: According to the pre-set engineering structure layer, region growth calculation is performed on each target leakage location data to obtain the actual leakage coordinate value of the target leakage location data and the predicted leakage coordinate value of the global heat distribution information. For each target leakage location data, determine its corresponding historical leakage location data in the previous frame of infrared thermal imaging image. Based on the actual leakage coordinate values ​​and the actual coordinate values ​​of the historical leakage location data, calculate the actual displacement, actual diffusion velocity, and actual diffusion acceleration of the target leakage location data relative to the previous frame. For each predicted leakage coordinate value of global heat distribution information, determine its corresponding predicted historical coordinate value in the previous frame of infrared thermal imaging image. Based on the predicted leakage coordinate value and the predicted historical coordinate value, calculate the predicted displacement, predicted diffusion velocity and predicted diffusion acceleration of global heat distribution information relative to the previous frame. A second loss function is established using the actual displacement, actual diffusion velocity, actual diffusion acceleration, and predicted displacement, predicted diffusion velocity, and predicted diffusion acceleration.

[0009] Preferably, based on the actual leakage coordinates and the actual coordinates of historical leakage location data, the actual displacement, actual diffusion velocity, and actual diffusion acceleration of the target leakage location data relative to the previous frame are calculated, including: Subtract the true coordinates of historical leakage locations from the true coordinates of the target leakage location data to obtain the true displacement. Obtain the inter-frame time interval, remove the real bits by the inter-frame time interval, and obtain the real diffusion speed; The true diffusion acceleration is calculated using the actual diffusion velocity and the inter-frame time interval.

[0010] Preferably, based on the predicted leakage coordinates and predicted historical coordinates, the predicted displacement, predicted diffusion velocity, and predicted diffusion acceleration of the global heat distribution information relative to the previous frame are calculated, including: The predicted displacement is obtained by subtracting the predicted historical coordinates from the predicted leakage coordinates of the global heat distribution information. Obtain the inter-frame time interval, remove the prediction bits by the inter-frame time interval, and obtain the prediction spread rate; The predicted diffusion acceleration is calculated using the predicted diffusion velocity and the inter-frame time interval.

[0011] Preferably, a third loss function is established based on the target leakage location data and the corrected temperature field distribution data, including: Determine the true temperature distribution value of the data for each target leakage location and the predicted temperature distribution value of the corrected temperature field distribution data; For each target leakage location data, determine its corresponding historical temperature distribution data in the previous frame of infrared thermal imaging image. Based on the real temperature distribution value and the real temperature value of the historical temperature distribution data, calculate the real temperature difference, real temperature difference rate of change and real temperature difference acceleration of the target leakage location data relative to the previous frame. For each corrected temperature field distribution data, the predicted temperature distribution value is determined, and its corresponding predicted historical temperature value in the previous frame of infrared thermal imaging is determined. Based on the predicted temperature distribution value and the predicted historical temperature value, the predicted temperature difference, the rate of change of the predicted temperature difference, and the acceleration of the change of the predicted temperature difference relative to the previous frame are calculated. A third loss function is established using the actual temperature difference, the actual temperature difference rate of change, the actual temperature difference acceleration, the predicted temperature difference, the predicted temperature difference rate of change, and the predicted temperature difference acceleration.

[0012] Preferably, the detection of large noise anomalies is performed on the weight value to be detected, and local thermal anomaly regions that do not meet the preset temperature change threshold are removed, including: Construct the heat transfer matrix for the current and historical times based on environmental parameter information. Based on the heat transfer matrix and the current thermal coordinate system of the infrared thermal imaging acquisition device at the current time, estimate the historical thermal coordinate system of the infrared thermal imaging acquisition device at the historical time, and determine the thermal coordinate transformation value between the historical thermal coordinate system and the current thermal coordinate system. If any local thermal anomaly region is a steady-state heat source region, the historical heat source location information of the steady-state heat source region in the historical thermal coordinate system is determined based on the thermal imaging model and thermal coordinate transformation value of the infrared thermal imaging acquisition device. Determine the current heat source location information of the steady-state heat source region in the current thermal coordinate system, determine whether the temperature difference interval between the current heat source location information and the historical heat source location information is greater than the preset temperature change threshold, and identify all steady-state heat source regions that do not meet the preset temperature change threshold as environmental interference points and remove them.

[0013] A smart system for locating and diagnosing leaks in waterproofing projects, applicable to the aforementioned smart method for locating and diagnosing leaks in waterproofing projects, including: The image acquisition unit is used to acquire the infrared thermal imaging image sequence of the waterproof area to be detected, acquire the environmental parameter information at the corresponding time, and bind the infrared thermal imaging image and environmental parameter information at the same time into the current detection data group. The conduction correction unit is used to perform thermal conduction analysis on the infrared thermal imaging image sequence in the current detection data group, generate initial temperature field distribution data and temperature gradient change information corresponding to the initial temperature field distribution data, and use environmental parameter information to correct the thermal conduction coefficient of the initial temperature field distribution data to obtain corrected temperature field distribution data. The thermal field optimization unit is used to predict the thermal anomaly trend of the waterproofing project based on the corrected temperature field distribution data and temperature gradient change information, combined with historical infrared image information, to obtain the trend data to be diagnosed. The current thermal radiation characteristics, historical thermal radiation characteristics and external heat exchange constraints are used as constraints on the trend data to be diagnosed. The initial temperature field distribution data and temperature gradient change information are used to perform weighted local thermal anomaly optimization to obtain the thermal field data to be optimized. Anomaly detection unit is used to perform posterior estimation using the temperature change rate of each local thermal anomaly region in the thermal field data to be optimized, assign a weight value to be detected, perform large noise anomaly detection on the weight value to be detected, remove local thermal anomaly regions that do not meet the preset temperature change threshold, and obtain the target thermal anomaly region and the corresponding target weight value. The thermal field splicing unit is used to input the thermal field data to be optimized, all target thermal anomaly areas and target weight values ​​into the pre-established intelligent leakage location and diagnosis model. Based on the target weight values, all target thermal anomaly areas are spliced ​​globally according to the thermal field data to be optimized to obtain the target leakage location data and global thermal distribution information of the waterproof area to be detected. The location diagnosis unit is used to establish a total loss function based on the target leakage location data, global heat distribution information, and corrected temperature field distribution data. The leakage intelligent location diagnosis model is iteratively trained with the goal of minimizing the total loss function until the leakage location output by the model converges, thus obtaining the trained leakage intelligent location diagnosis model.

[0014] Compared with the prior art, the beneficial effects of the present invention are: (1) By combining infrared thermal imaging image sequences and environmental parameter information, this invention can more accurately identify and locate the leakage in waterproof areas, improving detection accuracy. Moreover, by using heat conduction analysis and correcting temperature field distribution data, it can reflect the temperature changes in waterproof areas in real time, providing a dynamic basis for leakage detection and enhancing response speed. Furthermore, by establishing the total loss function and its sub-functions, it effectively integrates information on temperature distribution, leakage location, and historical data, making model training more scientific and reasonable, thereby improving the accuracy and reliability of the model. (2) This invention can predict the thermal anomaly trend of waterproofing projects by using infrared image information from historical moments, provide early warning of potential leakage problems, reduce maintenance costs, and provide a scientific basis for subsequent maintenance and repair work by analyzing target leakage location data and global thermal distribution information, helping decision-makers to formulate more reasonable maintenance plans. Furthermore, by adopting a high-noise anomaly point detection method, environmental interference points can be effectively eliminated to ensure the validity of detection data, thereby reducing the possibility of misjudgment and omission. Attached Figure Description

[0015] Figure 1 This is a schematic flowchart of the overall method in one embodiment of the present invention; Figure 2 This is a schematic diagram of the overall system architecture in one embodiment of the present invention.

[0016] In the diagram: 1. Image acquisition unit; 2. Conduction correction unit; 3. Thermal field optimization unit; 4. Anomaly detection unit; 5. Thermal field stitching unit; 6. Positioning and diagnosis unit. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example 1, please refer to Figure 1 This invention provides a technical solution: an intelligent positioning and diagnosis method for leakage in waterproofing projects, comprising: S1. Obtain the infrared thermal imaging image sequence of the waterproof area to be detected, collect the environmental parameter information at the corresponding time, and bind the infrared thermal imaging image and environmental parameter information at the same time into the current detection data group. S2. Perform heat conduction analysis on the infrared thermal imaging image sequence in the current detection data group to generate initial temperature field distribution data and temperature gradient change information corresponding to the initial temperature field distribution data. Use environmental parameter information to correct the heat conduction coefficient of the initial temperature field distribution data to obtain corrected temperature field distribution data. S3. Based on the corrected temperature field distribution data and temperature gradient change information, combined with historical infrared image information, predict the thermal anomaly trend of the waterproofing project to obtain the trend data to be diagnosed. Use the current thermal radiation characteristics, historical thermal radiation characteristics and external heat exchange constraints as constraints on the trend data to be diagnosed. Perform weighted local thermal anomaly optimization on the initial temperature field distribution data and temperature gradient change information to obtain the thermal field data to be optimized. S4. Use the temperature change rate of each local thermal anomaly region in the thermal field data to be optimized to perform posterior estimation, assign the weight value to be detected, perform large noise anomaly detection on the weight value to be detected, remove local thermal anomaly regions that do not meet the preset temperature change threshold, and obtain the target thermal anomaly region and the corresponding target weight value. S5. Input the thermal field data to be optimized, all target thermal anomaly areas and target weight values ​​into the pre-established intelligent leakage location diagnosis model. Based on the target weight values, stitch all target thermal anomaly areas together according to the thermal field data to be optimized to obtain the target leakage location data and global thermal distribution information of the waterproof area to be detected. S6. Based on the target leakage location data, global heat distribution information, and corrected temperature field distribution data, establish a total loss function. With the goal of minimizing the total loss function, iteratively train the leakage intelligent location diagnosis model until the leakage location output by the model converges, thus obtaining the trained leakage intelligent location diagnosis model.

[0019] It should be noted that the thermal imaging image sequence of the waterproof area to be detected is obtained through infrared thermal imaging technology; at the same time, the corresponding environmental parameters (such as temperature, humidity, etc.) are recorded; these images and environmental parameters will be combined into a current detection data set for subsequent analysis. Thermal conduction analysis is performed on the infrared thermal imaging images in the current detection data set; this process generates initial temperature field distribution data and extracts the related temperature gradient change information; then, the initial temperature field distribution data is corrected using the collected environmental parameters to improve the accuracy of the data and obtain the corrected temperature field distribution. By combining infrared image information from historical moments, and based on corrected temperature field distribution data and temperature gradient change information, trend prediction is made for potential thermal anomalies in waterproofing projects. The result of this step is called the trend data to be diagnosed, and current and historical thermal radiation characteristics as well as external heat exchange conditions are used as constraints. Based on the trend data to be diagnosed, the initial temperature field distribution data and temperature gradient change information are optimized to identify local thermal anomaly regions. The temperature change rate of each local thermal anomaly region is used for posterior estimation and then weighted. In this process, large noise anomalies are also detected, and those regions that have not reached the preset temperature change threshold are eliminated, thereby determining the target thermal anomaly region and its weight value. The thermal field data to be optimized, all target thermal anomaly areas and their weight values ​​are input into the pre-established intelligent leakage location and diagnosis model. In this model, all thermal anomaly areas are globally stitched together according to the target weight values ​​to obtain the target leakage location and global thermal distribution information of the waterproof area to be detected. Based on the target leakage location data, global thermal distribution information, and corrected temperature field distribution data, a total loss function is established. The leakage intelligent location and diagnosis model is iteratively trained by minimizing this function. This process continues until the leakage location results output by the model converge, thus completing the model training.

[0020] In an optional embodiment, the initial temperature field distribution data is corrected for thermal conductivity using environmental parameter information to obtain corrected temperature field distribution data, including: Based on the infrared image information at the current time, the infrared image information at the historical time, and the environmental parameter information, calculate the thermal convection error information and thermal radiation error information of the waterproof area to be detected between the current time and the historical time. Based on the ambient temperature information, the first correction calculation is performed on the thermal convection error information and thermal radiation information to obtain the change in surface heat flux density of the waterproof area to be tested between the current time and historical time. Based on the environmental humidity information, a second correction calculation is performed on the thermal convection error information and thermal radiation information to obtain the change in internal thermal diffusion of the waterproof area to be tested between the current time and historical time. Based on thermal radiation information, the thermal conductivity coefficient information at different depths is determined. Based on the thermal conductivity coefficient information at different depths, a third correction calculation is performed on the thermal radiation information and thermal radiation error information to obtain the change in thermal conductivity of the waterproof area to be tested between the current time and historical time. The changes in surface heat flux density, internal heat diffusion, and heat conduction are used as the changes in the thermal field. The changes in the thermal field are then superimposed with the initial temperature field distribution data from the previous moment to obtain the corrected temperature field distribution data.

[0021] It should be noted that, based on the infrared image information and environmental parameters at the current and historical times, the thermal convection error and thermal radiation error of the waterproof area to be detected between these two times are calculated; these errors reflect the deviation of thermal characteristics caused by environmental changes. The ambient temperature information is used to correct the previously calculated thermal convection and thermal radiation errors; this process yields the change in surface heat flux density of the area under test between the current time and historical time, which helps to better understand the flow of surface heat. The environmental humidity information is used to further correct the thermal convection error and thermal radiation error; this will produce the amount of change in thermal diffusion inside the area to be tested, reflecting the diffusion characteristics of heat inside the material due to humidity changes. Based on thermal radiation information, the thermal conductivity coefficients at different depths are determined. These coefficients can help assess the efficiency of heat conduction in materials. Subsequently, these thermal conductivity coefficients are used to correct the thermal radiation information and thermal radiation errors, and the change in thermal conductivity of the area under test between the current time and historical time is calculated. The changes in surface heat flux density, internal heat diffusion, and heat conduction obtained through the above steps are combined to form a comprehensive thermal field change. Then, this thermal field change is added to the initial temperature field distribution data from the previous moment to obtain new corrected temperature field distribution data.

[0022] In an optional embodiment, generating temperature gradient change information corresponding to the initial temperature field distribution data includes: Count the actual temperature extraction count of all pixels, and define the maximum temperature acquisition count corresponding to all pixels within the time window between the current time and historical time. The temperature information corresponding to all pixels is projected into the prior heat conduction model of the environmental parameter sensor to obtain the predicted temperature information, and the temperature difference error information between the predicted temperature information and the current temperature information is calculated. Based on the temperature difference error information, the maximum number of temperature samples collected, and the actual number of temperature samples extracted, temperature gradient change information is generated.

[0023] It should be noted that the actual temperature data extracted from all pixels between the current time and historical time points is statistically analyzed. This statistical process defines a time window to determine the number of temperature samples collected for each pixel within this time period and identifies the corresponding maximum number of temperature samples collected. This is to ensure that subsequent analysis has a sufficient data foundation. The temperature information of all pixels is projected into the prior heat conduction model of the environmental parameter sensor; this model uses environmental parameters to predict the temperature information that each pixel should have under ideal conditions; through this projection, a set of predicted temperature data can be obtained. Calculate the temperature difference between the predicted temperature information and the actual temperature information at the current moment; this temperature difference error reflects the deviation between the actual measurement and the model prediction, and is an important indicator for evaluating the heat transfer effect; By combining temperature difference error information, the maximum number of temperature samples, and the actual number of temperature samples extracted, temperature gradient change information is generated. This information represents the temperature change trend between different pixels, which helps to identify thermal anomalies or potential problems in specific areas.

[0024] In an optional embodiment, a total loss function is established based on target leakage location data, global heat distribution information, and corrected temperature field distribution data, including: A first loss function is established based on the corrected temperature field distribution data and the thermal field data to be optimized, which is used to constrain the consistency of the thermal field distribution. A second loss function is established based on the target leakage location data and global heat distribution information to constrain the positioning accuracy of the leakage coordinates; A third loss function is established based on the target leakage location data and the corrected temperature field distribution data to constrain the accuracy of temperature distribution prediction. Weight coefficients are set for the first loss function, the second loss function, and the third loss function, and a total loss function is established based on the three loss functions and their weight coefficients.

[0025] It should be noted that a first loss function is established based on the relationship between the corrected temperature field distribution data and the thermal field data to be optimized. This function is used to evaluate the consistency of the thermal field distribution, ensuring that the corrected temperature field matches the expected thermal field data, thus providing a basic constraint. This helps to reduce errors caused by inconsistent model predictions. Based on the target leakage location data and global heat distribution information, a second loss function is established. The main function of this function is to constrain the positioning accuracy of the leakage coordinates. This means that it will focus on how to accurately determine the location of potential leakage points and ensure that the model can effectively identify these key areas. A third loss function is established using the target leakage location data and the corrected temperature field distribution data. This function is used to constrain the accuracy of temperature distribution prediction, ensuring that the temperature changes predicted by the model match the actual measured data, thereby improving the overall reliability of temperature prediction. To balance the influence of these three loss functions in the optimization process, a weight coefficient is set for each loss function. These weight coefficients reflect the importance of each loss function in the total loss and can be adjusted according to specific needs to achieve the best optimization effect. The three loss functions and their corresponding weighting coefficients are combined to construct the total loss function. This total loss function comprehensively considers thermal field consistency, leakage location accuracy, and temperature prediction accuracy, providing a comprehensive evaluation standard for the entire optimization process.

[0026] In an optional embodiment, a second loss function is established based on target leakage location data and global heat distribution information, including: According to the pre-set engineering structure layer, region growth calculation is performed on each target leakage location data to obtain the actual leakage coordinate value of the target leakage location data and the predicted leakage coordinate value of the global heat distribution information. For each target leakage location data, determine its corresponding historical leakage location data in the previous frame of infrared thermal imaging image. Based on the actual leakage coordinate values ​​and the actual coordinate values ​​of the historical leakage location data, calculate the actual displacement, actual diffusion velocity, and actual diffusion acceleration of the target leakage location data relative to the previous frame. For each predicted leakage coordinate value of global heat distribution information, determine its corresponding predicted historical coordinate value in the previous frame of infrared thermal imaging image. Based on the predicted leakage coordinate value and the predicted historical coordinate value, calculate the predicted displacement, predicted diffusion velocity and predicted diffusion acceleration of global heat distribution information relative to the previous frame. A second loss function is established using the actual displacement, actual diffusion velocity, actual diffusion acceleration, and predicted displacement, predicted diffusion velocity, and predicted diffusion acceleration.

[0027] It should be noted that, based on the pre-defined engineering structure layer, a region growing calculation is performed on the data of each target leakage location. The purpose of this process is to extract the true leakage coordinate values ​​from the initial leakage location data. At the same time, combined with global heat distribution information, the corresponding leakage coordinate values ​​are predicted. These two sets of coordinate values ​​will be used for subsequent error calculations. For each target leakage location data, determine its corresponding historical leakage location data in the previous frame of infrared thermal imaging image; this step is to obtain the development of the target leakage point in the time series, and then calculate its change with the current frame; Based on the actual leakage coordinates and historical leakage location data from the previous frame, the actual displacement, actual diffusion velocity, and actual diffusion acceleration of the target leakage location relative to the previous frame are calculated. These parameters help to understand how the leakage point changes over time, including its movement speed and diffusion trend. For each predicted leakage coordinate value of global thermal distribution information, determine its corresponding predicted historical coordinate value in the previous frame of infrared thermal imaging image; this can evaluate the accuracy of the model in predicting the leakage location. By utilizing the relationship between predicted leakage coordinates and predicted historical coordinates, the predicted displacement, predicted diffusion velocity, and predicted diffusion acceleration of global heat distribution information relative to the previous frame are calculated; these prediction parameters will be used to compare the difference between the actual and the predicted values. A second loss function is established by combining the actual displacement, actual diffusion acceleration, and predicted displacement, predicted diffusion velocity, and predicted diffusion acceleration. This loss function aims to quantify the difference between the actual leakage dynamics and the prediction model, thereby optimizing the positioning accuracy of the leakage coordinates.

[0028] In an optional embodiment, based on the actual leakage coordinates and the actual coordinates of historical leakage location data, the actual displacement, actual diffusion velocity, and actual diffusion acceleration of the target leakage location data relative to the previous frame are calculated, including: Subtract the true coordinates of historical leakage locations from the true coordinates of the target leakage location data to obtain the true displacement. Obtain the inter-frame time interval, remove the real bits by the inter-frame time interval, and obtain the real diffusion speed; The true diffusion acceleration is calculated using the actual diffusion velocity and the inter-frame time interval.

[0029] It should be noted that the actual leakage coordinates corresponding to the target leakage location data in the current frame are compared with the actual coordinates of the historical leakage location data in the previous frame. By subtracting the historical leakage coordinates, the actual displacement of the target leakage location relative to the previous frame can be obtained. This displacement value reflects the actual distance the leakage point moves between the two frames. Determine the time interval between two frames; this time interval is crucial for analyzing dynamic changes because it provides the basis for subsequent velocity and acceleration calculations. By using the obtained true displacement and inter-frame time interval, the true bit is removed by the time interval to obtain the true diffusion velocity; this velocity represents the movement rate of the seepage point during this time period and is a quantitative description of its dynamic behavior. By using the previously calculated true diffusion velocity and inter-frame time interval, the true diffusion acceleration is further derived; this acceleration reflects the change in the movement rate of the leak point, which helps to understand its diffusion trend and potential risks.

[0030] In an optional embodiment, based on the predicted leakage coordinates and predicted historical coordinates, the predicted displacement, predicted diffusion velocity, and predicted diffusion acceleration of the global heat distribution information relative to the previous frame are calculated, including: The predicted displacement is obtained by subtracting the predicted historical coordinates from the predicted leakage coordinates of the global heat distribution information. Obtain the inter-frame time interval, remove the prediction bits by the inter-frame time interval, and obtain the prediction spread rate; The predicted diffusion acceleration is calculated using the predicted diffusion velocity and the inter-frame time interval.

[0031] It should be noted that the predicted leakage coordinates in the global heat distribution information are compared with the corresponding predicted historical coordinates; by subtracting them, the predicted displacement can be obtained; this displacement reflects the model's prediction of the movement of the leakage location between the current frame and the previous frame. Determine the time interval between the current frame and the previous frame; this is a fundamental parameter for calculating velocity and acceleration, as it describes the length of time it takes for observed dynamic changes to occur. Using the calculated predicted displacement and the previously obtained inter-frame time interval, the predicted bit is removed by the time interval to obtain the predicted diffusion velocity; this velocity represents the rate at which the predicted leak point moves during this time period, indicating the model's understanding of the dynamic behavior of the leak. Based on the predicted diffusion rate and inter-frame time interval, the predicted diffusion acceleration is further calculated; this acceleration quantifies how fast the leakage point rate changes, revealing its diffusion trend and the evolution of potential risks.

[0032] In an optional embodiment, a third loss function is established based on the target leakage location data and the corrected temperature field distribution data, including: Determine the true temperature distribution value of the data for each target leakage location and the predicted temperature distribution value of the corrected temperature field distribution data; For each target leakage location data, determine its corresponding historical temperature distribution data in the previous frame of infrared thermal imaging image. Based on the real temperature distribution value and the real temperature value of the historical temperature distribution data, calculate the real temperature difference, real temperature difference rate of change and real temperature difference acceleration of the target leakage location data relative to the previous frame. For each corrected temperature field distribution data, the predicted temperature distribution value is determined, and its corresponding predicted historical temperature value in the previous frame of infrared thermal imaging is determined. Based on the predicted temperature distribution value and the predicted historical temperature value, the predicted temperature difference, the rate of change of the predicted temperature difference, and the acceleration of the change of the predicted temperature difference relative to the previous frame are calculated. A third loss function is established using the actual temperature difference, the actual temperature difference rate of change, the actual temperature difference acceleration, the predicted temperature difference, the predicted temperature difference rate of change, and the predicted temperature difference acceleration.

[0033] It should be noted that for each target leakage location, the corresponding actual temperature distribution value is obtained, and the predicted temperature distribution value is extracted from the corrected temperature field distribution data; these two values ​​represent the actual observed temperature and the model-based prediction result, respectively. For each target leakage location data, it is also necessary to find the corresponding historical temperature distribution data in the previous frame of infrared thermal imaging image; this step is to compare the difference between the current temperature state and the previous state. Based on the actual temperature distribution value and historical temperature distribution data, the actual temperature difference of the target leakage location data relative to the previous frame is calculated; further, the actual temperature difference information is used to calculate the rate of change and acceleration of the actual temperature difference; these indicators can reflect the changing trend and dynamic characteristics of temperature in the time series. For each predicted temperature distribution value of the corrected temperature field distribution data, it is necessary to determine its corresponding predicted historical temperature value in the previous frame, and then calculate the predicted temperature difference, the rate of change of the predicted temperature difference, and the acceleration of change; these prediction indicators demonstrate the model's understanding and prediction ability of temperature changes. The actual temperature difference, actual temperature difference rate of change, and actual temperature difference acceleration obtained above are combined with the predicted temperature difference, predicted temperature difference rate of change, and predicted temperature difference acceleration to form a third loss function. The role of this loss function is to quantify the difference between the model's prediction results and the actual observations, thereby optimizing the model and improving its prediction accuracy for leakage location and temperature changes.

[0034] In an optional embodiment, the detection of large noise anomalies in the weight value to be detected, and the elimination of local thermal anomaly regions that do not meet the preset temperature change threshold, includes: Construct the heat transfer matrix for the current and historical times based on environmental parameter information. Based on the heat transfer matrix and the current thermal coordinate system of the infrared thermal imaging acquisition device at the current time, estimate the historical thermal coordinate system of the infrared thermal imaging acquisition device at the historical time, and determine the thermal coordinate transformation value between the historical thermal coordinate system and the current thermal coordinate system. If any local thermal anomaly region is a steady-state heat source region, the historical heat source location information of the steady-state heat source region in the historical thermal coordinate system is determined based on the thermal imaging model and thermal coordinate transformation value of the infrared thermal imaging acquisition device. Determine the current heat source location information of the steady-state heat source region in the current thermal coordinate system, determine whether the temperature difference interval between the current heat source location information and the historical heat source location information is greater than the preset temperature change threshold, and identify all steady-state heat source regions that do not meet the preset temperature change threshold as environmental interference points and remove them.

[0035] It should be noted that, based on environmental parameter information, a heat transfer matrix is ​​constructed for the current moment and historical moments; this matrix is ​​a tool used to describe how heat is transferred and changes between different points in time. Using the heat transfer matrix and the current thermal coordinate system of the infrared thermal imaging acquisition device at the current moment, the thermal coordinate system of the infrared thermal imaging acquisition device at historical moments is estimated; this step provides the necessary coordinate reference for subsequent comparisons, enabling temperature data at different time points to be correlated. By comparing the current thermal coordinate system with the historical thermal coordinate system, the thermal coordinate transformation values ​​between the two are determined. These transformation values ​​will be used to transform the location of historical heat sources to the current coordinate system, thereby enabling effective comparison. If a local thermal anomaly area is identified as a steady-state heat source area, the heat source characteristics of this area are further analyzed; based on the thermal imaging model of the infrared thermal imaging acquisition device and the previously calculated thermal coordinate transformation values, the position of the steady-state heat source area in the historical thermal coordinate system is determined. Determine the location of the steady-state heat source region in the current thermal coordinate system; next, compare the temperature difference between the current heat source location information and the historical heat source location information. Check whether the temperature difference between the current heat source location information and the historical heat source location information is greater than the preset temperature change threshold; if the temperature difference does not reach this threshold, these steady-state heat source areas are regarded as environmental disturbance points and are removed; this process ensures that only those heat source areas with significant changes are retained, while potential environmental disturbances or noise are excluded.

[0036] Example 2, please refer to Figure 2 This invention provides a technical solution: an intelligent positioning and diagnostic system for leaks in waterproofing projects, applicable to the aforementioned intelligent positioning and diagnostic method for leaks in waterproofing projects, comprising: Image acquisition unit 1 is used to acquire infrared thermal imaging image sequence of the waterproof area to be detected, acquire environmental parameter information at the corresponding time, and bind the infrared thermal imaging image and environmental parameter information at the same time into the current detection data group; The conduction correction unit 2 is used to perform thermal conduction analysis on the infrared thermal imaging image sequence in the current detection data group, generate initial temperature field distribution data and temperature gradient change information corresponding to the initial temperature field distribution data, and use environmental parameter information to correct the thermal conduction coefficient of the initial temperature field distribution data to obtain corrected temperature field distribution data. The thermal field optimization unit 3 is used to predict the thermal anomaly trend of the waterproofing project based on the corrected temperature field distribution data and temperature gradient change information, combined with historical infrared image information, to obtain the trend data to be diagnosed. The current thermal radiation characteristics, historical thermal radiation characteristics and external heat exchange constraints are used as constraints on the trend data to be diagnosed. The initial temperature field distribution data and temperature gradient change information are used to perform weighted local thermal anomaly optimization to obtain the thermal field data to be optimized. Anomaly detection unit 4 is used to perform posterior estimation using the temperature change rate of each local thermal anomaly region in the thermal field data to be optimized, assign a weight value to be detected, perform large noise anomaly detection on the weight value to be detected, remove local thermal anomaly regions that do not meet the preset temperature change threshold, and obtain the target thermal anomaly region and the corresponding target weight value. The thermal field splicing unit 5 is used to input the thermal field data to be optimized, all target thermal anomaly areas and target weight values ​​into the pre-established intelligent leakage location diagnosis model. Based on the target weight values, all target thermal anomaly areas are spliced ​​globally according to the thermal field data to be optimized to obtain the target leakage location data and global thermal distribution information of the waterproof area to be detected. The location diagnosis unit 6 is used to establish a total loss function based on the target leakage location data, global heat distribution information, and corrected temperature field distribution data. The leakage intelligent location diagnosis model is iteratively trained with the goal of minimizing the total loss function until the leakage location output by the model converges, thus obtaining the trained leakage intelligent location diagnosis model.

[0037] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. A smart method for locating and diagnosing leaks in waterproofing projects, characterized in that, include: Acquire the infrared thermal imaging image sequence of the waterproof area to be detected, collect the environmental parameter information at the corresponding time, and bind the infrared thermal imaging image and environmental parameter information at the same time into the current detection data group; Thermal conduction analysis is performed on the infrared thermal imaging image sequence in the current detection data set to generate initial temperature field distribution data and temperature gradient change information corresponding to the initial temperature field distribution data. The thermal conduction coefficient of the initial temperature field distribution data is corrected using environmental parameter information to obtain corrected temperature field distribution data. Based on the corrected temperature field distribution data and temperature gradient change information, combined with historical infrared image information, the thermal anomaly trend of the waterproofing project is predicted to obtain the trend data to be diagnosed. The current thermal radiation characteristics, historical thermal radiation characteristics and external heat exchange constraints are used as constraints on the trend data to be diagnosed. The initial temperature field distribution data and temperature gradient change information are optimized with weights to obtain the thermal field data to be optimized. The temperature change rate of each local thermal anomaly region in the thermal field data to be optimized is used for posterior estimation, and the weight value to be detected is assigned. The weight value to be detected is then used to detect large noise anomalies, and local thermal anomaly regions that do not meet the preset temperature change threshold are removed to obtain the target thermal anomaly region and the corresponding target weight value. The thermal field data to be optimized, all target thermal anomaly areas and target weight values ​​are input into the pre-established intelligent leakage location and diagnosis model. Based on the target weight values, all target thermal anomaly areas are spliced ​​together globally according to the thermal field data to be optimized to obtain the target leakage location data and global thermal distribution information of the waterproof area to be detected. Based on the target leakage location data, global heat distribution information, and corrected temperature field distribution data, a total loss function is established. The leakage intelligent location and diagnosis model is iteratively trained with the goal of minimizing the total loss function until the leakage location output by the model converges, thus obtaining the trained leakage intelligent location and diagnosis model.

2. The intelligent positioning and diagnosis method for waterproofing project leakage according to claim 1, characterized in that, The initial temperature field distribution data is corrected for thermal conductivity using environmental parameter information to obtain corrected temperature field distribution data, including: Based on the infrared image information at the current time, the infrared image information at the historical time, and the environmental parameter information, calculate the thermal convection error information and thermal radiation error information of the waterproof area to be detected between the current time and the historical time. Based on the ambient temperature information, the first correction calculation is performed on the thermal convection error information and thermal radiation information to obtain the change in surface heat flux density of the waterproof area to be tested between the current time and historical time. Based on the environmental humidity information, a second correction calculation is performed on the thermal convection error information and thermal radiation information to obtain the change in internal thermal diffusion of the waterproof area to be tested between the current time and historical time. Based on thermal radiation information, the thermal conductivity coefficient information at different depths is determined. Based on the thermal conductivity coefficient information at different depths, a third correction calculation is performed on the thermal radiation information and thermal radiation error information to obtain the change in thermal conductivity of the waterproof area to be tested between the current time and historical time. The changes in surface heat flux density, internal heat diffusion, and heat conduction are used as the changes in the thermal field. The changes in the thermal field are then superimposed with the initial temperature field distribution data from the previous moment to obtain the corrected temperature field distribution data.

3. The intelligent positioning and diagnosis method for waterproofing project leakage according to claim 2, characterized in that, Generate temperature gradient change information corresponding to the initial temperature field distribution data, including: Count the actual temperature extraction count of all pixels, and define the maximum temperature acquisition count corresponding to all pixels within the time window between the current time and historical time. The temperature information corresponding to all pixels is projected into the prior heat conduction model of the environmental parameter sensor to obtain the predicted temperature information, and the temperature difference error information between the predicted temperature information and the current temperature information is calculated. Based on the temperature difference error information, the maximum number of temperature samples collected, and the actual number of temperature samples extracted, temperature gradient change information is generated.

4. The intelligent positioning and diagnosis method for waterproofing project leakage according to claim 3, characterized in that, Based on the target leakage location data, global heat distribution information, and corrected temperature field distribution data, a total loss function is established, including: A first loss function is established based on the corrected temperature field distribution data and the thermal field data to be optimized, which is used to constrain the consistency of the thermal field distribution. A second loss function is established based on the target leakage location data and global heat distribution information to constrain the positioning accuracy of the leakage coordinates; A third loss function is established based on the target leakage location data and the corrected temperature field distribution data to constrain the accuracy of temperature distribution prediction. Weight coefficients are set for the first loss function, the second loss function, and the third loss function, and a total loss function is established based on the three loss functions and their weight coefficients.

5. The intelligent positioning and diagnosis method for waterproofing project leakage according to claim 4, characterized in that, A second loss function is established based on the target leakage location data and global heat distribution information, including: According to the pre-set engineering structure layer, region growth calculation is performed on each target leakage location data to obtain the actual leakage coordinate value of the target leakage location data and the predicted leakage coordinate value of the global heat distribution information. For each target leakage location data, determine its corresponding historical leakage location data in the previous frame of infrared thermal imaging image. Based on the actual leakage coordinate values ​​and the actual coordinate values ​​of the historical leakage location data, calculate the actual displacement, actual diffusion velocity, and actual diffusion acceleration of the target leakage location data relative to the previous frame. For each predicted leakage coordinate value of global heat distribution information, determine its corresponding predicted historical coordinate value in the previous frame of infrared thermal imaging image. Based on the predicted leakage coordinate value and the predicted historical coordinate value, calculate the predicted displacement, predicted diffusion velocity and predicted diffusion acceleration of global heat distribution information relative to the previous frame. A second loss function is established using the actual displacement, actual diffusion velocity, actual diffusion acceleration, and predicted displacement, predicted diffusion velocity, and predicted diffusion acceleration.

6. The intelligent positioning and diagnosis method for waterproofing project leakage according to claim 5, characterized in that, Based on the actual leakage coordinates and the actual coordinates of historical leakage location data, the actual displacement, actual diffusion velocity, and actual diffusion acceleration of the target leakage location relative to the previous frame are calculated, including: Subtract the true coordinates of historical leakage locations from the true coordinates of the target leakage location data to obtain the true displacement. Obtain the inter-frame time interval, remove the real bits by the inter-frame time interval, and obtain the real diffusion speed; The true diffusion acceleration is calculated using the actual diffusion velocity and the inter-frame time interval.

7. The intelligent positioning and diagnosis method for waterproofing project leakage according to claim 6, characterized in that, Based on the predicted leakage coordinates and predicted historical coordinates, the predicted displacement, predicted diffusion velocity, and predicted diffusion acceleration of the global heat distribution information relative to the previous frame are calculated, including: The predicted displacement is obtained by subtracting the predicted historical coordinates from the predicted leakage coordinates of the global heat distribution information. Obtain the inter-frame time interval, remove the prediction bits by the inter-frame time interval, and obtain the prediction spread rate; The predicted diffusion acceleration is calculated using the predicted diffusion velocity and the inter-frame time interval.

8. The intelligent positioning and diagnosis method for waterproofing project leakage according to claim 7, characterized in that, A third loss function is established based on the target leakage location data and the corrected temperature field distribution data, including: Determine the true temperature distribution value of the data for each target leakage location and the predicted temperature distribution value of the corrected temperature field distribution data; For each target leakage location data, determine its corresponding historical temperature distribution data in the previous frame of infrared thermal imaging image. Based on the real temperature distribution value and the real temperature value of the historical temperature distribution data, calculate the real temperature difference, real temperature difference rate of change and real temperature difference acceleration of the target leakage location data relative to the previous frame. For each corrected temperature field distribution data, the predicted temperature distribution value is determined, and its corresponding predicted historical temperature value in the previous frame of infrared thermal imaging is determined. Based on the predicted temperature distribution value and the predicted historical temperature value, the predicted temperature difference, the rate of change of the predicted temperature difference, and the acceleration of the change of the predicted temperature difference relative to the previous frame are calculated. A third loss function is established using the actual temperature difference, the actual temperature difference rate of change, the actual temperature difference acceleration, the predicted temperature difference, the predicted temperature difference rate of change, and the predicted temperature difference acceleration.

9. The intelligent positioning and diagnosis method for waterproofing project leakage according to claim 8, characterized in that, Large noise anomaly detection is performed on the weight values ​​to be detected, and local thermal anomaly regions that do not meet the preset temperature change threshold are removed, including: Construct the heat transfer matrix for the current and historical times based on environmental parameter information. Based on the heat transfer matrix and the current thermal coordinate system of the infrared thermal imaging acquisition device at the current time, estimate the historical thermal coordinate system of the infrared thermal imaging acquisition device at the historical time, and determine the thermal coordinate transformation value between the historical thermal coordinate system and the current thermal coordinate system. If any local thermal anomaly region is a steady-state heat source region, the historical heat source location information of the steady-state heat source region in the historical thermal coordinate system is determined based on the thermal imaging model and thermal coordinate transformation value of the infrared thermal imaging acquisition device. Determine the current heat source location information of the steady-state heat source region in the current thermal coordinate system, determine whether the temperature difference interval between the current heat source location information and the historical heat source location information is greater than the preset temperature change threshold, and identify all steady-state heat source regions that do not meet the preset temperature change threshold as environmental interference points and remove them.

10. A smart positioning and diagnostic system for waterproofing project leakage, applicable to the smart positioning and diagnostic method for waterproofing project leakage as described in any one of claims 1-9, characterized in that, include: The image acquisition unit is used to acquire the infrared thermal imaging image sequence of the waterproof area to be detected, acquire the environmental parameter information at the corresponding time, and bind the infrared thermal imaging image and environmental parameter information at the same time into the current detection data group. The conduction correction unit is used to perform thermal conduction analysis on the infrared thermal imaging image sequence in the current detection data group, generate initial temperature field distribution data and temperature gradient change information corresponding to the initial temperature field distribution data, and use environmental parameter information to correct the thermal conduction coefficient of the initial temperature field distribution data to obtain corrected temperature field distribution data. The thermal field optimization unit is used to predict the thermal anomaly trend of the waterproofing project based on the corrected temperature field distribution data and temperature gradient change information, combined with historical infrared image information, to obtain the trend data to be diagnosed. The current thermal radiation characteristics, historical thermal radiation characteristics and external heat exchange constraints are used as constraints on the trend data to be diagnosed. The initial temperature field distribution data and temperature gradient change information are used to perform weighted local thermal anomaly optimization to obtain the thermal field data to be optimized. Anomaly detection unit is used to perform posterior estimation using the temperature change rate of each local thermal anomaly region in the thermal field data to be optimized, assign a weight value to be detected, perform large noise anomaly detection on the weight value to be detected, remove local thermal anomaly regions that do not meet the preset temperature change threshold, and obtain the target thermal anomaly region and the corresponding target weight value. The thermal field splicing unit is used to input the thermal field data to be optimized, all target thermal anomaly areas and target weight values ​​into the pre-established intelligent leakage location and diagnosis model. Based on the target weight values, all target thermal anomaly areas are spliced ​​globally according to the thermal field data to be optimized to obtain the target leakage location data and global thermal distribution information of the waterproof area to be detected. The location diagnosis unit is used to establish a total loss function based on the target leakage location data, global heat distribution information, and corrected temperature field distribution data. The leakage intelligent location diagnosis model is iteratively trained with the goal of minimizing the total loss function until the leakage location output by the model converges, thus obtaining the trained leakage intelligent location diagnosis model.

Images