Building wall hollowing detection method and system

Abstract

The invention relates to the technical field of building structure detection, and discloses a building wall hollowing detection method and system. The method comprises the following steps: firstly, performing thermal excitation treatment on a building wall surface, automatically selecting a natural heat source, an artificial heat source or a dual-mode heat source combining the natural heat source and the artificial heat source to perform heat loading on the wall surface based on real-time environmental conditions, and adaptively adjusting the power of the artificial heat source according to a wall surface material and an environmental temperature. And in the heat loading and subsequent cooling process, dynamic infrared thermal image sequence data of the wall surface are collected. And performing thermal gradient three-dimensional reconstruction based on the thermal image sequence data to obtain temperature gradient distribution of the surface and the deep part of the wall body. According to the temperature gradient distribution, the apparent temperature difference delta T, the edge gradient G and the depth attenuation rate beta are calculated, and a multi-dimensional hollowing criterion is established. And identifying a hollowing area by using the criterion, and marking and outputting an identification result. According to the invention, the automation degree, the accuracy and the depth identification capability of building wall hollowing detection can be effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of building detection and structural nondestructive testing, and particularly relates to a building wall hollow detection method and system based on infrared thermal imaging and thermal gradient analysis. BACKGROUND

[0002] Infrared thermal imaging detection is a detection method based on the principle of thermal conductivity difference. When there is a hollow in the building wall, an air cavity is formed between the mortar layer and the base layer, and the thermal conductivity of air is about 0.026 W / m·K, which is much lower than that of dense wall materials such as concrete, which is about 1.4 W / m·K. Due to the obvious difference in thermal conductivity, the heat transfer efficiency of the hollow area is poor. Under external thermal excitation conditions, such as being exposed to sunlight or artificial heating, the surface of the hollow area in the heating process warms up quickly, showing a high-temperature hot spot, and in the cooling process, the surface of the hollow area cools down quickly, thus forming a low-temperature cold spot.

[0003] In the prior art, after natural sunlight is used to irradiate the outer wall of a building, an infrared thermal imaging device is used to scan the surface of the wall, and a detection personnel manually observes the temperature distribution difference on the thermal image to find possible hollow areas. In order to further confirm the detection result, the staff will also manually knock on the suspicious area to determine whether the wall has a hollow phenomenon.

[0004] The prior art has a high false detection rate, because it fails to effectively exclude the interference of internal pipelines, uneven insulation layers and other factors, which have similar features to the hollow area on the thermal image, and a single temperature difference threshold cannot accurately distinguish between the real hollow and the abnormal area caused by the thermal capacity difference. At the same time, the existing method lacks the ability to detect the depth of the hollow, and can only identify the surface temperature difference change, and does not establish a correlation model between the temperature difference gradient and the depth of the hollow. In addition, the method has a deficiency in the timeliness of the operation, and is heavily dependent on specific natural climate conditions, such as being able to detect only in a short time window after sunset, and being unable to work in rainy weather. SUMMARY

[0005] Therefore, the purpose of the embodiments of the present application is to provide a building wall hollow detection method and system to solve at least one of the above technical problems.

[0006] To achieve the above purpose, in a first aspect, the present application provides a building wall hollow detection method, which comprises the following steps: performing thermal excitation treatment on the building wall; based on real-time environmental conditions, automatically selecting and controlling a double-mode heat source to load heat on the wall, the double-mode heat source including a natural heat source and an artificial heat source, and the power of the artificial heat source being adaptively adjusted according to the wall material and the environmental temperature; acquire a dynamic infrared thermal image sequence data of the wall surface during the heat loading process and the subsequent cooling process; based on the dynamic infrared thermal image sequence data, perform a thermal gradient three-dimensional reconstruction on the wall to obtain a temperature gradient distribution of the wall surface and deep part thereof; according to the temperature gradient distribution, calculate an apparent temperature difference ΔT, an edge gradient G and a depth attenuation rate β, and establish a multi-dimensional hollow criterion; based on the multi-dimensional hollow criterion, identify a hollow region, and mark and output the identified hollow region.

[0007] In a second aspect, a building wall hollow detection system is provided, and the system comprises: a thermal excitation device for performing thermal excitation treatment on a building wall to establish an initial temperature gradient before detection; a heat source control device for automatically selecting a natural heat source mode, an artificial heat source mode or a dual-mode heat loading scheme combining natural and artificial heat sources based on real-time collected environmental parameters, and adaptively adjusting the power of the artificial heat source according to the wall material and environmental conditions when the artificial heat source mode or the dual-mode scheme is selected; an infrared image acquisition device for acquiring a dynamic infrared thermal image sequence data of the wall surface during the heat loading process and the subsequent cooling process; a data processing device for performing a thermal gradient three-dimensional reconstruction on the wall based on the dynamic infrared thermal image sequence data to obtain a temperature gradient distribution of the wall surface and deep part thereof, calculating an apparent temperature difference ΔT, an edge gradient G and a depth attenuation rate β according to the temperature gradient distribution, establishing a multi-dimensional hollow criterion, identifying a hollow region based on the multi-dimensional hollow criterion, and outputting a detection result with a hollow region mark.

[0008] The above technical solution has the following beneficial effects: By means of heat excitation treatment before detection and combining with real-time environmental condition adaptive selection of dual-mode heat source for heat loading, the dependence on natural climate condition of the traditional method can be effectively overcome, and stable temperature field can be formed in various environments. Secondly, by means of continuous acquisition of dynamic infrared thermal image sequence data in the whole process of heat loading and cooling, and three-dimensional reconstruction of thermal gradient based on the data, not only the temperature information of the wall surface can be acquired, but also the temperature gradient distribution in the deep part can be deduced, and quantitative analysis on the depth of the hollowing can be realized. Thirdly, by means of calculation of multi-dimensional feature parameters such as apparent temperature difference, edge gradient and depth attenuation rate by using the temperature gradient distribution, and establishment of multi-dimensional hollowing criterion, the hollowing recognition is no longer dependent on single temperature difference threshold, and the misjudgment rate caused by interference factors such as pipeline and thermal insulation layer is reduced. Finally, by means of automatic recognition and marking of the hollowing area based on the multi-dimensional criterion, the automation degree of the detection process and the objectivity of the detection result are improved, the manual intervention and error are reduced, and the positioning accuracy and detection efficiency of the hollowing are improved. BRIEF DESCRIPTION OF DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, below the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative effort on the basis of these drawings.

[0010] Figure 1 is the overall flowchart of the building wall hollowing detection method of the embodiment of the present application; Figure 2 is the specific flowchart of step S10 of the embodiment of the present application; Figure 3 is the specific flowchart of step S20 of the embodiment of the present application; Figure 4 is the specific flowchart of step S30 of the embodiment of the present application; Figure 5 is the specific flowchart of step S40 of the embodiment of the present application; Figure 6 is the specific flowchart of step S43 of the embodiment of the present application; Figure 7 is the specific flowchart of step S50 of the embodiment of the present application; Figure 8 is the specific flowchart of step S60 of the embodiment of the present application; Figure 9 is the functional block diagram of the building wall hollowing detection system of the embodiment of the present application; Figure 10 is the functional block diagram of an electronic device of the embodiment of the present application. DETAILED DESCRIPTION

[0011] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0012] Embodiment one The terms used in the present embodiment are defined as follows: The thermal gradient refers to the rate of temperature change per unit distance, which is used to describe the change of temperature along the spatial position in the building wall detection process. By analyzing the thermal gradient, the corresponding relationship between the surface temperature distribution difference and the abnormal structure of the wall can be revealed, thereby providing a quantitative basis for identifying defects such as hollows.

[0013] The equivalent diameter refers to the diameter size obtained by converting the detected hollow area into a circle with the same area. This parameter can reflect the actual range of the hollow area in a unified scale, which is helpful for comparing and statistically analyzing hollow defects of different shapes and sizes.

[0014] The depth attenuation rate refers to the rate of exponential attenuation of the temperature signal with the increase of the wall depth, which is used to represent the characteristics of energy attenuation in the process of heat transfer from the surface to the interior. By analyzing the depth attenuation rate, a quantitative relationship between the surface temperature change and the depth of the hollow defect can be established, thereby realizing the calculation and evaluation of the depth of the hollow.

[0015] The apparent temperature difference ΔT refers to the temperature difference between the suspected hollow area and the adjacent normal wall area obtained by analyzing the thermal image data of the wall during infrared thermal imaging detection. This parameter reflects the difference in heat conduction performance caused by the local structural difference of the wall, thereby leading to the difference in surface temperature distribution, and is a direct feature for judging the existence of the hollow.

[0016] The edge gradient G refers to the rate of change of temperature with spatial position along the edge direction of the hollow area in the thermal image. The edge gradient reflects the degree of temperature mutation at the boundary of the region. Due to the difference in heat conduction performance between the hollow area and the normal wall, the temperature distribution changes significantly at the boundary, so the edge gradient G can be used to identify the outline of the hollow and locate the boundary.

[0017] The depth attenuation rate β refers to the rate of exponential attenuation of the temperature signal with the depth when the temperature signal propagates from the surface to the interior of the wall. By fitting the time series infrared data after thermal excitation, the value of β can be obtained, and by using the relationship between β and the depth of the defect, the depth range of the hollow in the wall can be calculated, thereby realizing the quantitative detection of the depth of the hollow.

[0018] As Figure 1 shown, the embodiment provides a building wall hollow detection method, the method comprises the following steps: S10: heat excitation treatment is carried out on the building wall.

[0019] In this step, when the building wall is subjected to heat excitation treatment, a portable infrared heating lamp or a hot air blower can be used as a heat excitation device. Before operation, it is necessary to confirm that there are no obvious obstacles on the wall, and the heating equipment is kept at a distance of 0.5-1.5 meters from the wall. After starting the equipment, the surface temperature of the wall is uniformly increased by 3-8℃ on the basis of the initial temperature. For example, for a concrete wall, an infrared heating lamp with a power of 1500W can be selected, and the heating is continued for 10-15 minutes. During this period, the temperature of each area of the wall is monitored in real time by an infrared thermometer to ensure that the temperature difference does not exceed ±1℃, so as to establish a stable initial temperature field and provide a uniform reference for subsequent heat loading.

[0020] S20: based on real-time environmental conditions, automatically select and control the double-mode heat source to load heat on the wall, the double-mode heat source includes natural heat source and artificial heat source, the power of the artificial heat source is self-adaptively adjusted according to the wall material and environmental temperature.

[0021] In this step, when the double-mode heat source is selected based on real-time environmental conditions, the environmental temperature (accuracy ±0.5℃), relative humidity (accuracy ±3%) and wind speed (accuracy ±0.1m / s) are first collected in real time by an environmental monitoring module integrated with temperature sensor, humidity sensor and wind speed sensor. When the environmental temperature is higher than 25℃ and the light intensity is ≥5000lux, the natural heat source mode is automatically selected, and the sunlight is used to load heat on the wall. When the environmental temperature is lower than 15℃ or the light intensity is <2000lux, the artificial heat source mode is switched to, and a hot air blower with adjustable power (power range 500-2000W) is used. If the wall is made of ceramic tiles, the power of the hot air blower is set to 1200W at an environmental temperature of 20℃; if it is a painted wall, the power is adjusted to 800W under the same environment, and the PID control algorithm is used to stabilize the wall temperature change rate at 0.5-1℃ / min.

[0022] S30: during the heat loading process and the subsequent cooling process, dynamic infrared thermal image sequence data of the wall surface is collected.

[0023] Specifically, when collecting dynamic infrared thermal image sequence data, an infrared thermal imager with a resolution of 640x512 pixels (temperature measurement range -20℃ to 150℃, accuracy ±2%) is installed on a tripod and kept 2-3 meters away from the wall surface to ensure that the detection area is completely included in the field of view. During the heat loading stage (lasting 20 minutes), one frame of thermal image is collected every 1 minute; after entering the natural cooling stage, one frame is collected every 2 minutes for the first 30 minutes, and then one frame is collected every 5 minutes, with a total collection time of 2 hours. During the collection process, the thermal imager is synchronized with the environmental monitoring module through a Bluetooth module, and each frame of image is attached with time stamp, environmental temperature, humidity and heat source power information, and stored in TIFF format to retain temperature data, while automatically rejecting blurred frames caused by slight device shaking (invalid frames are determined when the gradient amplitude is below the threshold value of 50 through image clarity evaluation index screening).

[0024] S40: Based on the dynamic infrared thermal image sequence data, the wall surface is reconstructed in three dimensions based on thermal gradient to obtain the temperature gradient distribution of the wall surface and its deep part.

[0025] In this step of three-dimensional thermal gradient reconstruction, the dynamic infrared thermal image sequence is first preprocessed: SIFT algorithm is used to register each frame of image to eliminate pixel offset caused by device displacement; 3x3 Gaussian filter is used to remove high-frequency noise; and temperature data is standardized to the range of 0-255 gray value. Then the temperature value of each pixel point in the whole sequence is extracted to form a temperature change curve with time as the horizontal axis, and the first derivative of the curve is calculated to obtain the temperature change rate (unit ℃ / s). Based on the basic equation of heat conduction (∂T / ∂t=α∇²T, where α is the thermal diffusion coefficient of the wall material, and the value of concrete is 1.2x10⁻ 6 m² / s), the least squares method is used to fit the temperature change rate sequence, and a three-dimensional thermal gradient matrix (x and y axes are two-dimensional coordinates of the wall surface, z axis is the depth direction, and step length is 0.5 cm) is constructed by MATLAB software to obtain the temperature gradient distribution from the surface to 5 cm depth, and the output is a three-dimensional visualization model.

[0026] S50: According to the temperature gradient distribution, apparent temperature difference ΔT, edge gradient G and depth attenuation rate β are calculated, and multi-dimensional hollow criterion is established.

[0027] The apparent temperature difference ΔT is the average temperature difference between the target detection area (a square with a side length of 0.3 m) and the reference area within a 10 cm range, for example, the target area temperature is 28°C and the reference area is 25°C, ΔT=3°C. The edge gradient G is obtained by calculating the temperature gradient amplitude of the edge pixels of the target area, and the Sobel operator is used to derive in x and y directions and take the root mean square, if the temperature at the edge drops from 28°C to 25°C, the distance is 0.05 m, then G=60°C / m. The depth attenuation rate β is obtained by exponential fitting (T=Ae^(-βt)+B) of the temperature attenuation curve at different depths (0.5 cm, 1 cm, 2 cm, 3 cm, 5 cm), and the average value of β values at different depths is taken, for example, β=0.02 / min is obtained by fitting. The multi-dimensional hollow criterion is set as: ΔT≥2°C, G≥50°C / m, β≤0.03 / min, all of which must be met.

[0028] S60: Identify the hollow area based on the multi-dimensional hollow criterion, and mark the identified hollow area for output.

[0029] In this step, the above criteria are applied to the three-dimensional temperature gradient model, and the regions that meet the conditions are extracted by the Python open source library OpenCV, and are marked as polygonal regions. For the identified hollow area, a red border is superimposed on the infrared thermal image, and the center coordinates (based on the left upper corner of the wall as the origin of the rectangular coordinate system), the area (for example, 0.09 m²) and the depth (for example, 1.5 cm) are marked. The final output result includes the original thermal image sequence, the three-dimensional temperature gradient graph and the marked detection report, which supports PDF format export, and is also stored as an editable CAD file to facilitate engineering application.

[0030] The building wall hollow detection technical scheme has the following advantages: first, the combination of thermal excitation and double-mode heat source control can establish stable initial conditions through uniform heating in the pretreatment stage, and can switch between natural and artificial heat sources and dynamically adjust the power according to environmental parameters (temperature, illumination, etc.), thereby greatly improving the environmental adaptability under different climates and wall materials; second, the dynamic infrared thermal image acquisition uses high-resolution equipment and a strict data screening mechanism to record multi-dimensional environmental parameters synchronously, thereby providing high-precision and high-integrity raw data support for subsequent analysis; third, the thermal gradient three-dimensional reconstruction process integrates image registration, noise filtering and mathematical modeling based on the heat conduction equation, thereby realizing accurate inversion of the temperature gradient from the surface to the deep part (up to 5 cm) and avoiding the limitation that the traditional detection can only reflect the surface state; fourth, the cooperative application of multi-dimensional hollow criteria (apparent temperature difference, edge gradient and depth attenuation rate) combined with contour extraction and quantitative marking can significantly reduce the misjudgment rate of single indicator judgment and improve the accuracy and positioning accuracy of hollow area recognition; finally, the diversified output forms (thermal image sequence, three-dimensional model, CAD file, etc.) take into account the data analysis and engineering application requirements, thereby forming a complete closed loop from data acquisition, processing to result application.

[0031] As shown in Figure 2 Step S10 is a preheating preparation process before detection, specifically including: S11: obtaining initial environmental conditions of the detection site by a portable meteorological detector and an infrared thermometer, the initial environmental conditions including environmental temperature, wall surface temperature, environmental humidity and wind speed of the air around the site, for evaluating whether the preheating condition before detection is met.

[0032] In specific implementation, the detector carries a portable meteorological detector (TESTO440) (which can measure environmental temperature, humidity and wind speed at the same time, the measurement ranges are-20-70℃, 0-100%RH and 0-20m / s respectively, and the accuracies are all ±2%) and an infrared thermometer (FLIRTG165) (temperature measurement range-30-500℃, accuracy ±1℃) to the detection site. First, the meteorological detector is placed at a ventilated position 1.5m away from the wall, and 3 groups of environmental data are continuously collected (each group is separated by 1 minute), and the average value is taken as the environmental temperature (for example, 23℃), the environmental humidity (for example, 65%) and the wind speed (for example, 1.2m / s); at the same time, the infrared thermometer is used to measure the surface temperature of the wall detection area in a grid shape (each 0.5m×0.5m is a measuring point), and the surface temperature of more than 20 measuring points is recorded (for example, 21-22℃). The collected data are compared with the preset threshold value: when the environmental temperature is 5-35℃, the humidity is ≤85%, the wind speed is ≤3m / s, and the temperature difference of each measuring point of the wall is ≤2℃, it is determined that the preheating condition is met.

[0033] S12: Before the detection starts, at least one of the following is selected: natural heat source preheating or manual temporary auxiliary heating using a portable heating device, to moderately preheat the detection area, so that the surface temperature distribution of the detection area tends to be uniform and an initial temperature gradient is formed; the portable heating device includes a hot air machine, an infrared heating lamp, or a hot air gun.

[0034] In a specific implementation, if the detection period is 10:00-15:00 on a sunny day, and the light intensity is greater than or equal to 4000 lux (measured by a light sensor), natural heat source preheating is selected: a sunshade is used to shield the non-detection area of the wall, and only the detection area (for example, a range of 2 m x 1 m) is allowed to receive sunlight, for 30 minutes, during which the infrared thermal imager is scanned every 5 minutes to ensure that the standard deviation of the surface temperature distribution is less than or equal to 1°C. If it is cloudy or the light is insufficient (light intensity <2000 lux), manual auxiliary heating is used: a Bosch GHG18-60 hot air gun with adjustable power (power 500-1800 W) is selected, and the hot air gun is moved at a 45° angle to the wall at a distance of 0.8 m, with an initial power of 1200 W, and the position is adjusted every 2 minutes, for 15 minutes; if the wall is made of gypsum, a 200W infrared heating lamp can be used to supplement heating, so that the average temperature of the detection area is increased by 5°C compared with the initial value, and the maximum temperature difference is controlled within ±1°C, forming an initial temperature gradient from the surface to the interior (the surface temperature is 2-3°C higher than the deep layer).

[0035] S13: Before preheating, the position planning of the detection area is completed using a building BIM model or a surveying and positioning system, and surface inspection is performed on the surface of the detection area through manual visual inspection or a camera-based image recognition algorithm, to ensure that the detection area is clean and unobstructed, so as to provide stable initial temperature conditions for subsequent automated heat loading and data collection.

[0036] In implementation, before preheating, the Autodesk Revit software is used to import the building BIM model, locate the coordinates of the detection area in the model (for example, 3-axis ~ 5-axis, elevation 2.5 ~ 3.5 m), and export the plan size of the area (for example, 3 m wide and 1.2 m high); at the same time, the boundary of the detection area is marked on site using the Trimble R10 surveying and positioning system (positioning accuracy ± 3 cm) and identified with chalk lines. During surface inspection, the detection personnel first visually check whether there are dust accumulation, paint peeling, attachments (such as posters, wires), etc.; then a Hikvision DS-2CD3T46WD-I3 camera (resolution 4 million pixels) is used to shoot high-definition images of the detection area, the Canny edge detection algorithm in the OpenCV library is used to identify the outline of the obstruction, and if an obstruction with a diameter > 5 cm or a stain with an area > 0.01 m² is found, the obstruction is cleaned with a dust-free cloth dipped in alcohol. Obstructions that cannot be removed (such as fixed pipelines) are marked in the BIM model, and the corresponding area is excluded during subsequent data processing to ensure that the surface cleanliness of the detection area is above 95% and there is no obvious obstruction.

[0037] The above technical solution can effectively avoid the interference of extreme environment on detection by accurately obtaining initial environmental parameters and evaluating preheating conditions; can establish a stable initial temperature gradient for subsequent heat loading by flexibly selecting natural or artificial preheating methods and controlling temperature uniformity; and can reduce the influence of obstructions on data acquisition by combining BIM model positioning and surface cleaning. The three work together to improve the environmental adaptability of detection and provide reliable initial conditions for subsequent automated processes, thereby reducing system error and improving the basic precision and stability of the hollow detection.

[0038] As shown in Figure 3 Step S20 is the formal heat loading process of the detection phase, which specifically includes: S21: After completing the preheating preparation of step S10, real-time collection of environmental parameters including environmental temperature, environmental humidity and wind speed, and wall surface temperature.

[0039] In this step, after completing the preheating preparation, the integrated environmental monitoring terminal (containing SHT30 temperature and humidity sensor, measurement range 0 ~ 100℃, 0 ~ 100%RH, accuracy ± 0.3℃, ± 2%RH; FS4000 wind speed sensor, range 0 ~ 10 m / s, accuracy ± 0.1 m / s) is started to collect environmental parameters in real time at a frequency of 10 times per second; at the same time, an online infrared thermal imager with a resolution of 320 x 240 (sampling frame rate 15 Hz) is started to continuously monitor the wall surface temperature, and the data is transmitted to the control terminal through Ethernet to generate real-time data streams of environmental parameters and wall surface temperature, wherein the wall surface temperature is the average value of 50 evenly distributed sampling points in the detection area.

[0040] S22: Based on the environmental parameters and the wall surface temperature, automatically select the natural heat source mode, the artificial heat source mode or the dual mode of natural heat source and artificial heat source according to the preset decision rule or decision model, and load heat on the wall.

[0041] In this step, the preset decision rule is as follows: when the environmental temperature is ≥28℃, the light intensity is ≥6000lux (monitored by BH1750 light sensor) and the wall surface temperature rising rate is ≥0.8℃ / min, automatically select the natural heat source mode, focus the sunlight on the detection area through the adjustable angle reflector; when the environmental temperature is ≤18℃ or the wind speed is ≥2.5m / s, switch to the artificial heat source mode, and enable the 2000W industrial infrared heating plate (covering 1.2 times the area of the detection area); when the environmental temperature is between 18~28℃ and the light intensity fluctuation is >2000lux, start the dual mode scheme, that is, the natural light is the main, supplemented by 800W hot air heater, and the two are controlled through the relay module linkage.

[0042] S23: When the artificial heat source mode or the dual mode scheme is used, the output power of the artificial heat source is dynamically adjusted through the power control algorithm to ensure that the wall surface temperature change is within the preset detection range.

[0043] In this step, the incremental PID algorithm is used to dynamically adjust the power of the artificial heat source. With the wall surface target temperature rising rate of 1.5℃ / min as the set value, the deviation of the current temperature change rate and the target value is calculated in real time: when the deviation is >0.3℃ / min, the power is increased by 10% steps (the highest is not more than 90% of the rated power); when the deviation is <-0.2℃ / min, the power is reduced by 15% steps (the lowest is not less than 30% of the rated power). For example, the initial power of the ceramic tile wall surface is set to 1200W when the environmental temperature is 22℃, if the temperature rises by 6℃ (rate 1.2℃ / min) in 5 minutes, the power is automatically adjusted to 1400W, ensuring that the temperature change is always stable in the range of 1.3~1.7℃ / min.

[0044] S24: In the heat loading process, the surface temperature change curve of the wall is monitored in real time, and when the temperature reaches the target change range, the subsequent dynamic infrared thermal image sequence data acquisition is started.

[0045] In this step, the terminal draws a wall surface temperature change curve in real time (horizontal axis is time, vertical axis is temperature), sets the target temperature change range to be the initial temperature after preheating +8~12℃. When the temperature values of 3 consecutive samplings (interval of 30 seconds) all fall within the range, the terminal triggers a signal to the infrared thermal image acquisition system, starts dynamic sequence acquisition, that is, in the heating stage, one frame of thermal image is collected every 30 seconds, in the cooling stage, one frame every 1 minute for the first hour and one frame every 5 minutes after 1 hour, and at the same time, the environmental parameters and heat source state at the starting time are marked in the collection log as the reference time point for subsequent data processing.

[0046] The above technical solution collects environmental and wall surface temperature data in real time at high frequency, provides accurate basis for heat source selection, automatically switches single / dual heat source mode according to environmental parameters, improves adaptability under different climates and light conditions, dynamically adjusts artificial heat source power with the aid of the PID algorithm, can stably control the wall surface temperature change rate, avoids temperature fluctuation interference, combines with the temperature standard trigger acquisition mechanism to ensure that the thermal image data is collected in the best response stage, and the four work together to effectively improve the stability of heat loading, environmental adaptability and effectiveness of subsequent data acquisition.

[0047] As shown in Figure 4 , step S30 specifically includes: S31: after completing the heat loading of step S20, starting the infrared thermal imager and monitoring the detection area on the building wall in real time.

[0048] In this embodiment, after completing the heat loading, an infrared thermal imager (resolution 640×512 pixels, thermal sensitivity <0.03℃, temperature measurement range -40~150℃) of FLIRT650sc type is started, which is fixed on an electric tripod, the lens is calibrated through a laser sight, the detection area (such as a 3m×2m wall) is completely included in the field of view, the lens is kept 3m away from the wall and perpendicular to the wall plane. The real-time monitoring mode of the thermal imager is turned on, the thermal image picture is transmitted to the control terminal through the matching software, the wall surface temperature distribution pseudo-color map is displayed in real time, the frame rate is set to 15fps, and the picture is ensured to be distortion-free and unobstructed.

[0049] S32: during the wall heating stage and the subsequent natural cooling stage, a plurality of infrared thermal image frames are continuously collected at a preset time interval, the plurality of infrared thermal image frames are arranged in time sequence to form dynamic infrared thermal image sequence data.

[0050] In this embodiment, during the wall heating stage (lasting 20 minutes), the thermal imager is set to automatically capture a frame of thermal image every 10 seconds; after entering the natural cooling stage, a frame is captured every 30 seconds for the first 30 minutes, a frame is captured every 2 minutes after 30 minutes, a frame is captured every 10 minutes after 2 hours, and a frame is captured every 10 minutes after 6 hours. The total number of frames collected is not less than 300 frames. Each frame of image is automatically named in the format of "YYYYMMDD_HHMMSS", stored in a 1TB industrial-grade solid state disk in chronological order, forming a dynamic infrared thermal image sequence dataset, and the time sequence correlation between frames is established through time stamp.

[0051] S33: During the acquisition process, the acquisition time, ambient temperature and heat loading information are recorded synchronously to provide a reference for the time and conditions for subsequent thermal gradient three-dimensional reconstruction.

[0052] In this embodiment, during the acquisition process, the thermal imager realizes data synchronization with the environment monitoring module (including DS18B20 temperature sensor, DHT22 humidity sensor, three-cup anemometer) through the RS485 interface. Each frame of thermal image is attached with metadata such as acquisition time (accurate to milliseconds), ambient temperature (±0.2℃), relative humidity (±2%), wind speed (±0.1m / s) and real-time power of artificial heat source (if enabled), and stored as an XML format auxiliary file associated with the corresponding thermal image through a unique identifier.

[0053] S34: The dynamic infrared thermal image sequence data collected is subjected to preliminary quality check using an image quality evaluation algorithm, and invalid frames caused by light interference, shaking or acquisition abnormalities are removed, and valid sequence data is retained for subsequent processing.

[0054] In this embodiment, a three-level quality check algorithm is used to screen the sequence data: first, the image sharpness value is calculated by Laplacian operator, and the blurred frames with sharpness <80 (threshold value determined by 100 standard images) are removed; second, the temperature standard deviation of single frame image is calculated, and when the standard deviation >5℃, it is determined as a light interference frame and removed; finally, the registration error of adjacent frames is calculated by SIFT feature matching, and if the error >3 pixels, it is determined as a device shaking frame. After screening, the proportion of valid frames is not less than 90%, and a quality check report is generated, marking the time point of invalid frames and the reason for removal, to ensure the reliability of the subsequent processing data.

[0055] As shown in Figure 5 , step S40 specifically includes: S41: Preprocessing the dynamic infrared thermal image sequence data continuously acquired and arranged in chronological order in step S30, the preprocessing including image registration, noise removal and temperature data standardization, to obtain the preprocessed dynamic infrared thermal image sequence data.

[0056] Specifically, when pre-processing the dynamic infrared thermal image sequence data, first, the ORB feature matching algorithm is used for image registration, that is, the first frame in the sequence is selected as the reference frame, and no less than 500 feature points are extracted from each subsequent frame of image, and the k- nearest neighbor matching algorithm is used to realize the alignment of the pixels between frames, and the registration error is controlled within 1 pixel; then 5x5 median filtering is used to remove salt and pepper noise, and then Gaussian filtering (σ = 1.2) is used to smooth the temperature field distribution; finally, temperature data standardization is performed, and each frame of temperature value is converted to the range of 0~1 according to the formula T_std = (T-T_min) / (T_max-T_min) (where T_min and T_max are the minimum and maximum temperatures in the whole sequence respectively), eliminating the influence of absolute temperature difference between different frames, and the pre-processed data is stored in the format of 16-bit grayscale image.

[0057] S42: Based on the pre-processed dynamic infrared thermal image sequence data, the temperature values of each pixel point at different time frames are extracted to form the temperature-time curve of each pixel point, and the change rate of the temperature-time curve is calculated to obtain the thermal gradient data reflecting the time sequence characteristics of the wall surface temperature.

[0058] Specifically, based on the pre-processed sequence data, the 640x512 pixel matrix of each frame of image is traversed through the OpenCV library of Python, the temperature value of each pixel point (x, y) in the whole sequence (300 frames) is extracted, and a time sequence curve with time as the horizontal axis (interval of 10 seconds to 10 minutes, converted to seconds according to the collection interval) and temperature as the vertical axis is formed. The central difference method is used to calculate the temperature change rate of the curve: for the temperature value T(t) of the t-th frame, the change rate v(t) = [T(t+1)-T(t-1)] / [2Δt] (Δt is the time interval between adjacent frames), and the forward / backward difference method is used to supplement the calculation of the first and last frames of the sequence, and finally a thermal gradient data matrix of 640x512x300 (the third dimension is time) is generated, reflecting the temperature change rate characteristics of each spatial point.

[0059] S43: Forming a thermal gradient stereoscopic matrix according to the thermal gradient data, and performing three-dimensional reconstruction on the thermal gradient stereoscopic matrix based on the heat conduction physical model by using a fitting algorithm to obtain the temperature gradient distribution of the wall from the surface to the deep part, and the fitting algorithm realizes the inversion of the deep temperature field by parameter fitting on the temperature-time curve.

[0060] Specifically, the thermal gradient data is arranged according to the spatial coordinates (x, y) and the time t, and a 1024*768*180 thermal gradient three-dimensional matrix is constructed (the x and y axes correspond to the 0.01 m*0.01 m pixel resolution of the wall surface, and the t axis is resampled at 60-second intervals to 180 time points). Based on the heat conduction equation ∂T / ∂t=λ / (ρc)∇²T (λ is the thermal conductivity of the wall material, the concrete is taken as 1.7 W / (m*K); ρ is the density, taken as 2400 kg / m³; c is the specific heat capacity, taken as 880 J / (kg*K)), a three-dimensional reconstruction is performed on the three-dimensional matrix by using the finite element analysis method: a layered model of the wall surface (surface layer, base layer, internal structure layer) is established by using ANSYS software, the thermal gradient data is taken as the boundary condition, the least square method is used to fit the temperature conduction coefficients of each layer, the temperature gradient distribution in the depth direction (0~10 cm, step 0.5 cm) is inversely calculated, and finally a three-dimensional visual temperature gradient cloud chart is output, which clearly shows the temperature change trend from the surface to the deep part.

[0061] The technical scheme described above effectively eliminates the data deviation caused by equipment jitter and environmental interference through image registration, noise filtering and standardization preprocessing, improves the consistency and reliability of the infrared thermal image sequence, accurately captures the subtle temperature dynamic characteristics of each region of the wall surface by extracting the temperature time curve of each pixel point and calculating the change rate, and provides high-resolution basic data for subsequent analysis. Combined with the three-dimensional reconstruction of the heat conduction physical model and the fitting algorithm, the limitations of traditional surface detection are avoided, the temperature gradient distribution in the deep part of the wall can be inversely calculated, and the accuracy and depth of the temperature field analysis are improved through the synergistic effect of the three, thereby providing scientific and comprehensive thermal characteristic basis for accurate identification of the hollow drum area.

[0062] As shown in Figure 6 , step S43 specifically includes: S431: based on the thermal gradient data obtained in step S42, arrange the thermal gradient data corresponding to different time points according to the pixel spatial coordinates, and construct a thermal gradient three-dimensional matrix in a three-dimensional coordinate system, wherein the three-dimensional coordinate system takes the two-dimensional plane position of the wall as the horizontal and vertical coordinates, and the derived dimension in the depth direction as the depth coordinate.

[0063] This step is based on the 640x512x300 thermal gradient data (x, y is the pixel coordinate, corresponding to the actual size of the wall 0.01m x 0.01m / pixel, time dimension 300 frames) obtained in step S42, a three-dimensional coordinate system is constructed: taking the lower left corner of the wall as the origin, x axis along the horizontal direction (0~6.4m), y axis along the vertical direction (0~5.12m), z axis is the depth derivation dimension (0~10cm, step 0.5cm). The thermal gradient data at different times is arranged according to (x, y, z) coordinates to form a 640x512x20 thermal gradient stereo matrix (z axis contains 20 depth layers), wherein each matrix element represents the temperature change rate (unit: ℃ / s) at a specific time for the corresponding position and depth, and the matrix data is stored in HDF5 format to preserve the three-dimensional structure information.

[0064] S432: Based on the thermal gradient stereo matrix, curve fitting is performed in the time dimension using the wall heat conduction equation to establish the mathematical relationship between the time change rate and the heat conduction depth, and the characteristic curve of the temperature change rate with the depth is obtained.

[0065] This step is for the thermal gradient stereo matrix, for each (x, y) plane position, the temperature change rate data of z axis (depth) and time t is extracted, based on the wall heat conduction equation ∂T / ∂t=α(∂²T / ∂x²+∂²T / ∂y²+∂²T / ∂z²) (α is the thermal diffusion coefficient of wall material, 0.7x10⁻ 6 m² / s for brick wall), curve fitting is performed in the time dimension using Levenberg-Marquardt (LM optimization algorithm) algorithm. For example, at x=2.5m, y=3.0m, the relationship between the temperature change rate and the depth is obtained by fitting: wherein v(z) represents the temperature change rate at the depth z position, v0 represents the initial temperature change rate at the surface of the wall, k is the attenuation coefficient of the temperature change rate along the depth direction, and c is a constant term. By adjusting the parameters, the fitting error is less than 5%, and finally the characteristic curve of the temperature change rate with the depth at each position is obtained, wherein the attenuation coefficient k reflects the conduction rate of heat in the wall.

[0066] S433: According to the characteristic curve, the temperature gradient inside the wall is three-dimensionally inverted to generate a temperature gradient distribution image from the surface of the wall to the deep part.

[0067] Based on the characteristic curve, a three-dimensional inversion algorithm (combined with finite element analysis software COMSOL Multiphysics) is used to construct the wall internal temperature gradient field model: taking the surface thermal gradient data as the boundary condition, setting the wall material properties (density, specific heat capacity, thermal conductivity), and iteratively calculating the temperature gradient distribution at different depths (0-10 cm). The three-dimensional temperature gradient field obtained by inversion is output in the form of visual slice images, and a 2D temperature gradient map (unit: ℃ / cm) is generated every 0.5 cm depth, and the gradient change from the surface to the deep part is intuitively displayed through the pseudo-color color map (red-yellow represents high gradient, blue-green represents low gradient), and the hollow area will show obvious gradient anomaly area due to the obstruction of heat conduction.

[0068] S434: Extract the gradient value at different depth positions and the change data at the corresponding time point from the temperature gradient distribution image to form the input data set for the calculation of apparent temperature difference ΔT, edge gradient G and depth attenuation rate β in subsequent step S50.

[0069] This step extracts the gradient value data of each depth layer (20 layers in total) from the three-dimensional temperature gradient distribution image at 0.5 cm intervals, and synchronously records the change amount at the corresponding time point (take one time node every 10 minutes, a total of 12 time points). Organize these data into a structured data set, including fields: spatial coordinates (x, y), depth z, time t, temperature gradient value G, temperature change rate v, and store as a CSV format file. For example, at x=2.5m, y=3.0m, z=2cm, record G=0.8℃ / cm at t=0min, G=0.6℃ / cm at t=10min, etc. Form a standardized input data set for the calculation of apparent temperature difference ΔT, edge gradient G and depth attenuation rate β.

[0070] This technical solution integrates space-time and depth data by constructing a three-dimensional thermal gradient stereoscopic matrix, establishes a time-depth correlation by combining a heat conduction equation, accurately presents the wall deep temperature gradient distribution through three-dimensional inversion, and finally forms a standardized data set, which not only improves the systematicness of the data structure and the accuracy of the model fitting, but also avoids the limitations of surface detection, provides reliable input for subsequent hollow criterion calculation, and effectively enhances the detection depth and accuracy.

[0071] As shown in Figure 7 , step S50 specifically includes: S51: Based on the temperature gradient distribution of the wall surface and deep part obtained in step S40, extract the temperature values of the target detection area and its surrounding reference area at the same time frame, calculate the temperature difference between the two, and obtain the apparent temperature difference ΔT.

[0072] Based on the temperature gradient distribution obtained in S40, in the 30th minute frame of the thermal image sequence (a typical time point in the cooling stage), a suspected hollow area target detection area (in units of 50x50 pixels, corresponding to an actual area of 0.5m x 0.5m) is selected, and a 10-pixel-wide annular area around it is designated as a reference area (to ensure that there is no hollow feature). The temperature values of all pixels in the two areas are extracted by the data processing software, and the difference between the average temperature of the target area (for example, 32.5°C) and the average temperature of the reference area (for example, 29.8°C) is calculated to obtain an apparent temperature difference ΔT = 2.7°C, and the depth layer corresponding to this value (the surface layer and 2cm, 5cm depth) is recorded.

[0073] S52: Calculate the spatial direction change rate of the temperature gradient distribution to determine the temperature change rate of the edge of the target detection area, and obtain the edge gradient G.

[0074] When calculating the spatial direction change rate of the temperature gradient distribution, a 3x3 Sobel operator is used to derive in the x-axis (horizontal) and y-axis (vertical) directions to obtain the gradient components of the edge of the target detection area. For example, at the left edge of the target area, the temperatures of the adjacent pixels in the horizontal direction drop from 32.5°C to 30.1°C, and the pixel spacing corresponds to an actual distance of 0.02m, so the x-direction gradient is (32.5-30.1) / 0.02=120°C / m; the vertical direction edge gradient is calculated in the same way. Take the root mean square of the gradient amplitude of all pixels on the edge as the edge gradient G, such as G=85°C / m, which reflects the temperature sudden change characteristics of the boundary between the hollow area and the normal area.

[0075] S53: Using the temperature change curve of each pixel point in the dynamic infrared thermal image sequence data over time, the temperature decay characteristics of different depth positions of the wall are fitted to obtain the depth decay rate β.

[0076] For the temperature change curve of each pixel point in the target area in the dynamic infrared thermal image sequence (covering the complete period of 6 hours from heating to cooling), data from the surface layer, 2cm, and 5cm depth layers are selected, and an exponential decay model (T(t) represents the temperature detected at cooling time t, T0 is the initial temperature difference, T∞ is the ambient temperature, t is the cooling time, and β is the decay rate of temperature decay over time) is used for nonlinear fitting. The β values for each depth are calculated by least squares iteration: β1=0.015 / min for the surface layer, β2=0.012 / min for the 2cm depth, and β3=0.008 / min for the 5cm depth. The average of the three is taken as the depth decay rate β=0.0117 / min, which represents the decay speed of heat in the wall.

[0077] S54: Establish a multi-dimensional hollow criterion for hollow detection by taking the apparent temperature difference ΔT, the edge gradient G and the depth attenuation rate β as multi-dimensional parameters.

[0078] Based on the experimental data of 50 groups of known hollow samples and 50 groups of normal wall samples, the threshold of the multi-dimensional hollow criterion is determined through ROC curve analysis: apparent temperature difference ΔT≥2℃ (the true positive rate reaches 92%), edge gradient G≥60℃ / m (the false positive rate is less than 5%), and depth attenuation rate β≤0.02 / min (the decay of the hollow area is slower due to air insulation). The three parameters are integrated into a logical judgment model: when ΔT, G and β meet the above threshold conditions at the same time, the corresponding detection area is identified as a hollow area, forming a comprehensive criterion system that takes into account surface features and internal heat conduction characteristics.

[0079] The technical scheme calculates the apparent temperature difference, the edge gradient and the depth attenuation rate, captures the hollow characteristics from the surface temperature difference, the boundary mutation and the internal heat decay in multiple dimensions, and establishes a comprehensive criterion that is more comprehensive than a single index, effectively improving the accuracy and reliability of hollow identification and reducing misjudgment and missed judgment.

[0080] As shown in Figure 8 , step S60 specifically includes: S61: Apply the apparent temperature difference ΔT, the edge gradient G, the depth attenuation rate β calculated in step S50 and the established multi-dimensional hollow criterion to the target detection area, and determine whether the apparent temperature difference ΔT is greater than a first preset threshold, whether the edge gradient G is greater than a second preset threshold, and whether the depth attenuation rate β meets a third preset threshold condition.

[0081] Specifically, the multi-dimensional hollow criterion established in S50 is applied to the target detection area, wherein the first preset threshold is set to apparent temperature difference ΔT≥2℃, the second preset threshold is edge gradient G≥60℃ / m, and the third preset threshold is depth attenuation rate β≤0.02 / min. The parameters of a suspected area are extracted from the processed data set: ΔT=2.7℃ (greater than 2℃), G=85℃ / m (greater than 60℃ / m), and β=0.0117 / min (less than 0.02 / min). The three parameters are compared with the corresponding thresholds one by one through the program, and the satisfaction status of each criterion is recorded (all are yes).

[0082] S62: When the above three criteria are met at the same time, the corresponding detection area is identified as a hollow area.

[0083] Specifically, based on the judgment result of step S61, when the apparent temperature difference ΔT, the edge gradient G, and the depth attenuation rate β of the target detection area all meet the respective preset threshold conditions, the system automatically triggers the hollow identification mechanism. For example, the three parameters of the above-mentioned suspected area all meet the standards, at this time the region (coordinate range x: 2.3~2.8m, y: 2.9~3.4m) is marked as a candidate hollow region, and its depth information (mainly concentrated in the depth of 2~5cm) is associated.

[0084] S63: Image marking processing is performed on the identified hollow area, and the marked detection result is output.

[0085] Specifically, image marking processing is performed on the identified hollow area, and an OpenCV library is used to draw a red dashed polygon frame (line width 3 pixels) on the original infrared thermal image, and fill the frame with a 50% transparent red mask; at the same time, the area (0.25m²), center coordinates (2.55m, 3.15m) and core parameters (ΔT=2.7℃, G=85℃ / m, β=0.0117 / min) outside the frame are labeled. The final output result includes a marked infrared thermal image sequence (key frame), a hollow area three-dimensional position map, and a structured detection report (PDF format), and the report is attached with the temperature change curve and gradient distribution slice map of the region, which is convenient for subsequent review and processing.

[0086] The technical scheme ensures the accuracy of hollow identification through multi-criterion simultaneous verification, combines clear marking and multi-form output, reduces false positives and false negatives, facilitates intuitive understanding and engineering application, and improves the reliability of the detection result.

[0087] In further embodiments, the following step is further included before step S60: S55: using a pre-established interference feature library to compare the dynamic infrared thermal image sequence data obtained in step S30, identifying interference thermal image features caused by pipeline, water seepage, and insulation layer defects, and excluding the interference thermal image features in the hollow determination process, thereby reducing the false positive rate of detection.

[0088] In a specific implementation, an interference feature library is established in advance, containing heat map feature parameters of 1000+ typical interference samples: pipeline interference is manifested as a linear high-temperature zone with a width of 2-5 cm (temperature difference of 3-5 ℃, continuously distributed along the horizontal / vertical direction), water seepage interference presents an irregular low-temperature zone (2-4 ℃ lower than the normal area, with a fuzzy edge and expanding over time), and insulation layer defects are manifested as a local temperature anomaly zone (temperature difference > 5 ℃ but without obvious edge gradient). The dynamic infrared thermal image sequence data of step S30 is input into the feature comparison system, and a feature matching algorithm based on a convolutional neural network (ResNet-18 model, trained with 500 labeled samples, with an identification accuracy of 96%) is used to compare the similarity of each region in the thermal image with the features in the library frame by frame. For example, if a region is detected to present a 2 cm wide, horizontally continuous 3.2 ℃ temperature difference zone, and the pipeline feature matching degree reaches 92%, it is marked as a pipeline interference area; for the identified interference area, its temperature parameters are automatically shielded in the subsequent hollow detection (for example, the ΔT, G, and β calculation of the area is excluded), or a correction coefficient is added to the criterion (for example, the edge gradient threshold of the pipeline area is increased by 20%).

[0089] In this embodiment, the data set used for algorithm verification covers three common types of building wall surfaces, namely, ceramic tile finished wall, paint finished wall, and stone finished wall. A total of 172 representative hollow samples are collected and labeled in the data set, each sample is verified by a real object in the field and standardized recorded, ensuring that the data is fully representative and reliable under various wall materials, different thicknesses, and diverse environmental conditions. The data set provides a comprehensive foundation for the training, verification, and performance evaluation of the algorithm.

[0090] In this embodiment, in order to improve the adaptability of the detection algorithm to complex actual scenes, an interference feature library containing various interference factors is established. The interference library pre-stores heat map feature templates caused by non-hollow anomalies such as typical pipeline distribution, water seepage traces, and insulation layer defects, and refines and classifies them through a large number of actual infrared thermal images. The algorithm can use the interference library to compare and identify the heat map features during operation, thereby effectively reducing the misjudgment rate caused by non-hollow factors and improving the accuracy and robustness of the detection results.

[0091] The advantages of this embodiment are: first, it can greatly reduce the false detection rate; second, it breaks through the limitations of the prior art that cannot achieve deep detection, and can realize quantitative detection of the depth of hollows in the range of 5-50 mm, with a correlation coefficient of 0.93 compared with a laser depth gauge, and high precision. Finally, the technical solution of this embodiment performs outstandingly in environmental adaptability, even in overcast conditions, the active detection mode can still ensure a high detection success rate.

[0092] Embodiment Two As Figure 9As shown, the embodiment provides a building wall hollow detection system, which comprises: A thermal excitation device for thermal excitation treatment of the building wall to establish an initial temperature gradient before detection; A heat source control device for automatically selecting a natural heat source mode, an artificial heat source mode or a dual-mode heat loading scheme combining natural and artificial heat sources based on real-time collected environmental parameters, and adaptively adjusting the power of the artificial heat source according to the wall material and environmental conditions when the artificial heat source mode or the dual-mode scheme is selected; An infrared image acquisition device for acquiring dynamic infrared thermal image sequence data of the wall surface during the heat loading process and the subsequent cooling process; A data processing device for three-dimensional reconstruction of the thermal gradient of the wall based on the dynamic infrared thermal image sequence data, obtaining the temperature gradient distribution of the wall surface and its deep part, calculating the apparent temperature difference ΔT, the edge gradient G and the depth attenuation rate β according to the temperature gradient distribution, and establishing a multi-dimensional hollow criterion; identifying the hollow area based on the multi-dimensional hollow criterion, and outputting the detection result with the hollow area marked.

[0093] In some embodiments, the thermal excitation device specifically comprises: An initial environmental condition acquisition module for acquiring the initial environmental conditions of the detection site by a portable meteorological detector and an infrared thermometer, including environmental temperature, wall surface temperature, environmental humidity and wind speed of the surrounding air, for evaluating whether the preheating condition before detection is met; A preheating execution module for at least one of natural heat source preheating in a non-controlled mode or artificial temporary auxiliary heating using a portable heating device to moderately preheat the detection area before detection starts, so that the surface temperature distribution of the detection area tends to be uniform and an initial temperature gradient is formed; the portable heating device includes a hot air blower, an infrared heating lamp or a hot air gun; A detection area preprocessing module for completing the position planning of the detection area by a building BIM model or a surveying and positioning system before preheating, and performing surface inspection on the surface of the detection area by manual visual inspection or camera-based image recognition algorithm, to ensure that the detection area is clean and unobstructed, so as to provide stable initial temperature conditions for subsequent automated heat loading and data acquisition.

[0094] In some embodiments, the heat source control device specifically comprises: A real-time parameter acquisition module for real-time acquisition of environmental parameters including environmental temperature, environmental humidity and wind speed, and wall surface temperature after completing preheating preparation; The loading scheme selection module is configured to automatically select a natural heat source mode, an artificial heat source mode, or a dual-mode loading scheme combining the natural heat source and the artificial heat source according to a preset decision rule or decision model based on the environmental parameter and the wall surface temperature, and load heat to the wall. The power dynamic adjustment module is configured to dynamically adjust the output power of the artificial heat source by a power control algorithm to ensure that the wall surface temperature change is within a preset detection range when the artificial heat source mode or the dual-mode scheme is adopted. The temperature monitoring starting module is configured to monitor a surface temperature change curve of the wall in real time during the heat loading process, and start subsequent dynamic infrared thermal image sequence data acquisition when the temperature reaches a target change range.

[0095] In some embodiments, the infrared image acquisition device specifically includes: The infrared thermal image monitoring starting module is configured to start the infrared thermal imager and monitor the detection area on the building wall in real time after the heat loading is completed. The sequence data forming module is configured to continuously acquire a plurality of infrared thermal image frames at a preset time interval during a wall heating stage and a subsequent natural cooling stage, arrange the plurality of infrared thermal image frames in time sequence, and form dynamic infrared thermal image sequence data. The information synchronous recording module is configured to synchronously record acquisition time, environmental temperature, and heat loading information during the acquisition process, so as to provide a time and condition reference for subsequent thermal gradient three-dimensional reconstruction. The quality inspection module is configured to preliminarily inspect the quality of the acquired dynamic infrared thermal image sequence data by using an image quality evaluation algorithm, eliminate invalid frames caused by light interference, shaking, or acquisition abnormalities, and retain valid sequence data for subsequent processing.

[0096] In some embodiments, the data processing device specifically includes: The data preprocessing module is configured to pre-process the dynamic infrared thermal image sequence data continuously acquired and arranged in time sequence, the pre-processing including image registration, noise filtering, and temperature data standardization, to obtain pre-processed dynamic infrared thermal image sequence data. The thermal gradient data acquisition module is configured to extract temperature values of each pixel point at different time frames based on the pre-processed dynamic infrared thermal image sequence data, form a temperature change curve of each pixel point with time, and calculate a change rate of the temperature change curve with time, to obtain thermal gradient data reflecting time sequence characteristics of the wall surface temperature. a three-dimensional reconstruction module configured to form a thermal gradient stereoscopic matrix according to the thermal gradient data, and perform three-dimensional reconstruction on the thermal gradient stereoscopic matrix based on a thermal conduction physical model by using a fitting algorithm to obtain a temperature gradient distribution of the wall from the surface to the deep part, the fitting algorithm being configured to realize inversion of the deep temperature field by parameter fitting on the temperature curve changing over time.

[0097] In some embodiments, the three-dimensional reconstruction module specifically comprises: a stereoscopic matrix construction unit configured to arrange the thermal gradient data corresponding to different time points according to pixel spatial coordinates based on the thermal gradient data, and construct a thermal gradient stereoscopic matrix in a three-dimensional coordinate system, the three-dimensional coordinate system taking a two-dimensional planar position of the wall as horizontal and vertical coordinates, and a derived dimension in the depth direction as a depth coordinate; a mathematical relationship establishment unit configured to perform curve fitting on the thermal gradient stereoscopic matrix in the time dimension by using a wall thermal conduction equation to establish a mathematical relationship between the time variation rate and the thermal conduction depth, and obtain a characteristic curve of the temperature variation rate changing with the depth; a temperature gradient inversion unit configured to perform three-dimensional inversion on the temperature gradient inside the wall according to the characteristic curve, and generate a temperature gradient distribution image from the surface to the deep part of the wall; an input data set formation unit configured to extract gradient values at different depth positions and change data at corresponding time points from the temperature gradient distribution image, and form an input data set for calculating an apparent temperature difference ΔT, an edge gradient G, and a depth attenuation rate β.

[0098] In some embodiments, the data processing device specifically comprises: an apparent temperature difference calculation module configured to extract temperature values of a target detection region and a peripheral reference region thereof in the same time frame based on the temperature gradient distribution of the surface and the deep part of the wall, calculate a temperature difference between the two, and obtain an apparent temperature difference ΔT; an edge gradient calculation module configured to calculate a change rate in the spatial direction of the temperature gradient distribution, determine a temperature change rate of the edge of the target detection region, and obtain an edge gradient G; a depth attenuation rate calculation module configured to fit temperature attenuation characteristics of different depth positions of the wall by using temperature curves changing over time of each pixel point in the dynamic infrared thermal image sequence data, and obtain a depth attenuation rate β; a criterion establishment module configured to take the apparent temperature difference ΔT, the edge gradient G, and the depth attenuation rate β as multi-dimensional parameters, and establish a multi-dimensional hollow criterion for hollow detection.

[0099] In some embodiments, the data processing device specifically comprises: A criterion judgment module is configured to apply the calculated apparent temperature difference ΔT, the edge gradient G, the depth attenuation rate β, and the established multi-dimensional hollow criterion to the target detection region, judge whether the apparent temperature difference ΔT is greater than a first preset threshold, whether the edge gradient G is greater than a second preset threshold, and whether the depth attenuation rate β satisfies a third preset threshold condition. A hollow region identification module is configured to identify the corresponding detection region as a hollow region when the above three criteria are met simultaneously. A marking output module is configured to perform image marking processing on the identified hollow region and output the marked detection result.

[0100] In some embodiments, the building wall hollow detection system further comprises an interference elimination device configured to compare the obtained dynamic infrared thermal image sequence data with a pre-established interference feature library before identifying the hollow region, identify interference thermal image features caused by non-hollow factors such as pipelines, water seepage, and insulation layer defects, and eliminate the interference thermal image features in the hollow determination process, thereby reducing the misjudgment rate of detection.

[0101] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above functional units and modules is exemplified, and in actual application, the above functions can be completed by different functional units or modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit or module in the embodiment can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The above integrated unit can be realized in the form of hardware or in the form of software functional unit. In addition, the specific names of each functional unit or module are only for easy distinction, and do not limit the protection scope of the present application. The specific working process of the units or modules in the above system can refer to the corresponding process in the foregoing method embodiments, which will not be repeated here.

[0102] Embodiment Three Referring to Figure 10 The embodiment of the present application also provides an electronic device, which comprises at least one processor and a memory in communication connection with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the building wall hollow detection method in the foregoing method embodiments.

[0103] The embodiment of the present application further provides a non-transitory computer readable storage medium storing computer instructions for causing the computer to execute the building wall hollow detection method in the foregoing method embodiment.

[0104] The embodiment of the present application further provides a computer program product comprising a computer program stored on a non-transitory computer readable storage medium, the computer program comprising program instructions which, when executed by a computer, cause the computer to execute the building wall hollow detection method in the foregoing method embodiment.

[0105] Reference is made below to Figure 10 which shows a structural schematic diagram of an electronic device 800 suitable for use to implement the embodiment of the present application. The electronic device 800 in the embodiment of the present application can include, but is not limited to, a mobile terminal such as a mobile phone, a notebook computer, a digital broadcast receiver, a PDA (Personal Digital Assistant), a PAD (Tablet Personal Computer), a PMP (Portable Multimedia Player), a vehicle terminal (e.g., a car navigation terminal), and the like, and a fixed terminal such as a digital TV, a desktop computer, and the like. Figure 8 The electronic device 800 shown is merely an example and should not impose any limitation on the function and use range of the embodiment of the present application.

[0106] As shown in Figure 10 , the electronic device 800 can include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 801 which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 802 or a program loaded from a storage device 808 into a random access memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the electronic device 800 are also stored. The processing device 801, the ROM 802, and the RAM 803 are connected to each other through a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0107] Generally, the following devices can be connected to the I / O interface 805: an input device 806 including, for example, a touch screen, a touch pad, a key, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, and the like; an output device 807 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, and the like; a storage device 808 including, for example, a magnetic tape, a hard disk, and the like; and a communication device 809. The communication device 809 can allow the electronic device 800 to communicate with other devices wirelessly or by wire to exchange data. Although the electronic device 800 with various devices is shown in the figure, it should be understood that it is not required to implement or have all the devices shown. More or less devices can be alternatively implemented or provided.

[0108] In particular, according to embodiments of the present disclosure, the processes described above with reference to the flowcharts can be implemented as a computer software program. For example, embodiments of the present disclosure include a computer program product comprising a computer program carried on a computer readable medium, the computer program containing program code for executing the methods illustrated by the flowcharts. In such embodiments, the computer program can be downloaded and installed from the network by the communication device 809, or installed from the storage device 808, or installed from the ROM 802. When the computer program is executed by the processing device 801, the above-mentioned functions defined in the methods of the embodiments of the present disclosure are executed.

[0109] It should be noted that the computer readable medium described above in the present disclosure can be a computer readable signal medium or a computer readable storage medium or any combination of the two. The computer readable storage medium may, for example, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or apparatus, or any combination of the above. More specific examples of the computer readable storage medium can include, but are not limited to, an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0110] The above description is merely a specific implementation of the present disclosure, but the scope of protection of the present disclosure is not limited thereto. Any changes or replacements easily conceivable by those skilled in the art within the technical scope of the present disclosure should be encompassed within the scope of protection of the present disclosure. Therefore, the scope of protection of the present disclosure should be subject to the scope of protection of the claims.

Claims

1. A method for detecting hollow areas in building walls, characterized in that, Includes the following steps: S10: Apply heat stimulation treatment to the building wall surface; S20: Based on real-time environmental conditions, automatically select and control a dual-mode heat source to apply heat to the wall surface. The dual-mode heat source includes a natural heat source and an artificial heat source. The power of the artificial heat source is adaptively adjusted according to the wall surface material and ambient temperature. S30: During the heat loading process and the subsequent cooling process, dynamic infrared thermal imaging sequence data of the wall surface are acquired; S40: Based on the dynamic infrared thermal imaging sequence data, perform three-dimensional reconstruction of the thermal gradient of the wall surface to obtain the temperature gradient distribution of the wall surface and its depth. S50: Based on the temperature gradient distribution, calculate the apparent temperature difference ΔT, edge gradient G, and depth attenuation rate β to establish a multi-dimensional hollow criterion; S60: Identify hollow areas based on the multi-dimensional hollow criterion, and mark and output the identified hollow areas.

2. The method for detecting hollow areas in building walls according to claim 1, characterized in that, Step S10 is the preheating preparation process before testing, which specifically includes: S11: Obtain the initial environmental conditions of the testing site using a portable meteorological detector and an infrared thermometer. The initial environmental conditions include ambient temperature, wall surface temperature, ambient humidity and wind speed of the surrounding air, in order to assess whether the preheating conditions are met before testing. S12: Before the detection begins, at least one of the following methods is selected: natural heat source preheating in an uncontrolled manner or artificial temporary auxiliary heating using a portable heating device, to moderately preheat the detection area so that the surface temperature distribution of the detection area tends to be uniform and an initial temperature gradient is formed. S13: Before preheating, use a building BIM model or surveying and positioning system to complete the location planning of the inspection area, and conduct a surface inspection of the inspection area by manual visual inspection or image recognition algorithm based on camera to ensure that the inspection area is clean and unobstructed.

3. The method for detecting hollow areas in building walls according to claim 1, characterized in that, Step S20 is the formal heat loading process during the detection phase, specifically including: S21: After completing the preheating preparation in step S10, collect environmental parameters, including ambient temperature, ambient humidity and wind speed, and wall surface temperature in real time. S22: Based on the environmental parameters and the wall surface temperature, automatically select the natural heat source mode, artificial heat source mode, or a dual-mode loading scheme combining natural and artificial heat sources according to preset decision rules or decision models to load heat onto the wall. S23: When using artificial heat source mode or dual-mode scheme, the output power of artificial heat source is dynamically adjusted through power control algorithm to ensure that the temperature change of wall surface is within the preset detection range; S24: Monitor the surface temperature change curve of the wall in real time during the heat loading process, and start subsequent dynamic infrared thermal imaging sequence data acquisition when the temperature reaches the target change range.

4. The method for detecting hollow areas in building walls according to claim 1, characterized in that, Step S30 specifically includes: S31: After completing the heat loading in step S20, start the infrared thermal imager and monitor the detection area on the building wall in real time. S32: During the wall heating stage and the subsequent natural cooling stage, multiple frames of infrared thermal images are continuously acquired at preset time intervals, and the multiple frames of infrared thermal images are arranged in chronological order to form dynamic infrared thermal image sequence data. S33: During the acquisition process, the acquisition time, ambient temperature, and heat loading information are recorded simultaneously to provide a reference for the time and conditions for subsequent three-dimensional reconstruction of the thermal gradient; S34: Use an image quality assessment algorithm to perform a preliminary quality check on the acquired dynamic infrared thermal image sequence data, remove invalid frames caused by light interference, jitter or acquisition abnormalities, and retain valid sequence data for subsequent processing.

5. The method for detecting hollow areas in building walls according to claim 1, characterized in that, Step S40 specifically includes: S41: Preprocess the dynamic infrared thermal image sequence data that was continuously acquired and arranged in chronological order in step S30 to obtain the preprocessed dynamic infrared thermal image sequence data. S42: Based on the preprocessed dynamic infrared thermal image sequence data, extract the temperature value of each pixel at different time frames to form the temperature change curve of each pixel over time, and calculate the rate of change of the temperature change curve over time to obtain thermal gradient data reflecting the temporal characteristics of the wall surface temperature. S43: A thermal gradient 3D matrix is ​​formed based on the thermal gradient data, and a fitting algorithm is used to reconstruct the thermal gradient 3D matrix in three dimensions based on the heat conduction physical model to obtain the temperature gradient distribution of the wall from the surface to the depth. The fitting algorithm realizes the inversion of the deep temperature field by fitting the parameters of the temperature change curve over time.

6. The method for detecting hollow areas in building walls according to claim 1, characterized in that, Step S43 specifically includes: S431: Based on the thermal gradient data obtained in step S42, the thermal gradient data corresponding to different times are arranged according to pixel space coordinates to construct a three-dimensional thermal gradient matrix in a three-dimensional coordinate system. The three-dimensional coordinate system uses the two-dimensional plane position of the wall as the horizontal and vertical coordinates and the derived dimension of the depth direction as the vertical coordinate. S432: Based on the thermal gradient three-dimensional matrix, curve fitting is performed using the wall heat conduction equation in the time dimension to establish a mathematical relationship between the rate of change of time and the depth of heat conduction, and to obtain the characteristic curve of the rate of change of temperature with depth. S433: Perform three-dimensional inversion of the temperature gradient inside the wall based on the characteristic curve to generate a temperature gradient distribution image from the wall surface to the depth. S434: Extract gradient values ​​at different depth locations and corresponding time point change data from the temperature gradient distribution image to form an input dataset for calculating the apparent temperature difference ΔT, edge gradient G, and depth decay rate β in the subsequent step S50.

7. The method for detecting hollow areas in building walls according to claim 1, characterized in that, Step S50 specifically includes: S51: Based on the temperature gradient distribution of the wall surface and depth obtained in step S40, extract the temperature values ​​of the target detection area and its surrounding reference area in the same time frame, calculate the temperature difference between the two, and obtain the apparent temperature difference ΔT. S52: Calculate the rate of change of the temperature gradient distribution in the spatial direction to determine the temperature change rate at the edge of the target detection area and obtain the edge gradient G; S53: Using the temperature change curve of each pixel in the dynamic infrared thermal image sequence data over time, the temperature attenuation characteristics at different depths of the wall are fitted to obtain the depth attenuation rate β. S54: Using the apparent temperature difference ΔT, edge gradient G, and depth attenuation rate β as multi-dimensional parameters, establish a multi-dimensional void criterion for void detection.

8. The method for detecting hollow areas in building walls according to claim 1, characterized in that, Step S60 specifically includes: S61: Apply the apparent temperature difference ΔT, edge gradient G, depth attenuation rate β and the established multi-dimensional hollow criterion obtained in step S50 to the target detection area to determine whether the apparent temperature difference ΔT is greater than the first preset threshold, whether the edge gradient G is greater than the second preset threshold, and whether the depth attenuation rate β meets the third preset threshold condition. S62: When all three criteria above are met, the corresponding detection area is identified as a hollow area; S63: Perform image labeling processing on the identified hollow areas and output the labeled detection results.

9. The method for detecting hollow areas in building walls according to claim 1, characterized in that, The following steps are included before step S60: S55: The dynamic infrared thermal image sequence data obtained in step S30 is compared using a pre-established interference feature library to identify interference thermal image features caused by non-hollow factors such as pipelines, water seepage, and insulation layer defects, and the interference thermal image features are removed during the hollow determination process.

10. A system for detecting hollow areas in building walls, characterized in that, The system includes: Thermal excitation equipment is used to thermally excite building walls to establish an initial temperature gradient before testing; The heat source control equipment is used to automatically select a natural heat source mode, an artificial heat source mode, or a dual-mode heat loading scheme that combines natural and artificial modes based on real-time collected environmental parameters, and to adaptively adjust the power of the artificial heat source according to the wall material and environmental conditions when selecting the artificial heat source mode or the dual-mode scheme. An infrared image acquisition device is used to acquire dynamic infrared thermal image sequence data of the wall surface during the heat loading process and the subsequent cooling process. The data processing device is used to perform three-dimensional reconstruction of the thermal gradient of the wall surface based on the dynamic infrared thermal image sequence data to obtain the temperature gradient distribution of the wall surface and its depth; calculate the apparent temperature difference ΔT, edge gradient G and depth attenuation rate β according to the temperature gradient distribution, and establish a multi-dimensional hollow criterion; identify hollow areas based on the multi-dimensional hollow criterion, and output the detection results with hollow area markings.

Images