Indoor physical simulation test and method for electromagnetic wave amplitude and spectrum under multiple factors

Abstract

The application provides an indoor physical simulation test and method for electromagnetic wave amplitude and spectrum under multiple factors, comprising a total test mechanism and a collection and analysis mechanism; the total test mechanism is used for simulating a pollution scene with different medium lithology, water content and LNAPL saturation in a region above underground water below the ground; the collection and analysis mechanism comprises a geological radar and an auxiliary measuring device; the geological radar is used for nondestructive detection of the medium in the total test mechanism, collection of original radar echo signals, and the auxiliary measuring device is used for synchronous acquisition of lithology characteristics and water content state parameters of the medium; the whole method takes nondestructive detection of the geological radar as a core, is combined with a matched rapid in-situ detection device, and compared with a traditional drilling sampling-laboratory analysis, the method shows great advantages of high efficiency, nondestruction and repeatability in a test stage and a final application stage.

Description

Technical Field

[0001] This invention relates to the field of oil leak testing technology, specifically to an indoor physical simulation test and method for electromagnetic wave amplitude and spectrum under multiple factors. Background Technology

[0002] Ground-penetrating radar (GPR) is a geophysical method that uses the reflection of high-frequency electromagnetic waves to detect targets and geological phenomena. In the early 20th century, Hulsenbeck first used electromagnetic signals for metal detection. In 1910, Leimback and Lowy formally proposed the concept of GPR, applying radar principles to ground exploration, by using dipole antennas buried in boreholes to detect areas with relatively high electrical conductivity underground, and obtained a patent. In 1926, Hulsenbeck first proposed the idea of ​​using pulsed electromagnetic waves to detect underground targets and successfully used this technology to study underground lithological structures. His theory that electromagnetic waves are reflected at the interface of media with different dielectric constants has become a fundamental theory in the field of GPR research. By the early 21st century, radar technology had been successfully applied in industrial and civilian fields.

[0003] The pollution of soil and groundwater by industrial organic compound (LNAPL) leaks is receiving increasing attention. Once organic pollutants enter the underground environment, they mostly contaminate soil and groundwater in the form of non-aqueous liquids. Natural non-aqueous liquids (LNAPLs) are immiscible fluids, classified as light or heavy LNAPLs based on their density relative to water. Improper use, leakage, and disposal of LNAPLs from petroleum fuels and coal tar can cause serious soil and groundwater pollution. Oil leaks from gas stations or underground oil storage facilities can allow LNAPLs to enter the vadose zone, becoming a persistent source of soil and groundwater pollution in some areas, seriously threatening environmental and public health safety. LNAPL pollution has become a global environmental problem.

[0004] Currently, the detection of soil and groundwater pollution caused by LNAPL mainly relies on traditional on-site drilling and sampling methods, followed by laboratory processing and analysis. The biggest limitation of traditional detection methods is their inability to quickly identify LNAPL pollution saturation and extent at a regional scale. Due to the cumbersome pretreatment process, complex detection methods, long detection cycle, high cost, invasive sampling process, and the potential for secondary pollution, it is questionable to use traditional methods to determine LNAPL pollution saturation and extent in emergency pollution or large-scale detection situations. Although there are currently methods that utilize the two-way travel time and three-instantaneous properties of ground-penetrating radar electromagnetic waves in the medium to indirectly detect LNAPL pollution, this still needs to be combined with traditional drilling methods and has certain limitations. Summary of the Invention

[0005] The purpose of this invention is to provide an indoor physical simulation test and method for electromagnetic wave amplitude and spectrum under multiple factors. It aims to solve the problems of existing technologies that mainly rely on traditional on-site drilling and sampling methods, which are cumbersome to operate and prone to secondary pollution. Although ground-penetrating radar is also used, it still needs to be combined with drilling methods, which has limitations.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: the indoor physical simulation test and method for electromagnetic wave amplitude and spectrum under multiple factors, including a main test mechanism and a data acquisition and analysis mechanism; The overall test facility is used to simulate pollution scenarios with different media lithology, water content and LNAPL saturation in areas below ground and above groundwater. The data acquisition and analysis unit includes a ground-penetrating radar and auxiliary measuring equipment. The ground-penetrating radar is used to perform non-destructive testing on the medium within the main test unit and to acquire raw radar echo signals. The auxiliary measuring equipment is used to simultaneously acquire the lithological characteristics and water content parameters of the medium. The original radar echo signal, lithological characteristics, and water content state parameters, together with the preset LNAPL saturation, constitute the calibration dataset used to build the machine learning model.

[0007] Preferably, the overall test apparatus includes a first test component, a second test component, and a third test component. The first test component is used to simulate homogeneous media with different lithologies, water contents, and LNAPL saturation under controlled conditions. The second test component is used to simulate the dynamic leakage and migration process of LNAPL in the medium. The third test component is used to simulate and monitor LNAPL pollution in undisturbed or remolded soil filled with different media in an outdoor environment to obtain a validation dataset under semi-natural conditions.

[0008] Preferably, the first test component includes a main container, a first filling medium, and a first contaminant; The first filling medium uses soil with different particle size distribution to simulate different lithological characteristics. The first main container is a sealed container made of a material with low dielectric response to electromagnetic waves to contain the first filling medium. The first pollutant is LNAPL, which is uniformly distributed in the first filling medium to simulate a static and homogeneous pollution distribution state. The data acquisition and analysis mechanism performs data measurement and analysis on the first experimental component.

[0009] Preferably, the second test assembly includes a sand tank, a leakage simulation component, a second filling medium, a water level control component, and a second contaminant; The water level control component is installed on the inner bottom wall of the sand tank, and the leakage simulation component is fixedly installed on the inner wall of the sand tank and is higher than the water level control component. The sand tank is a sealed container made of a material with low dielectric response to electromagnetic waves, used to contain the second filling medium. The second contaminant is LNAPL. The leakage simulation component is used to controllably inject LNAPL into the second filling medium to simulate the infiltration, lateral distribution and dynamic action of the second contaminant on the water surface of the second filling medium. The data acquisition and analysis unit performs data measurement and analysis on the second test component.

[0010] Preferably, the leakage simulation component includes a pipeline, a liquid storage container, a supply pump, and a control system; The pipeline is installed inside the sand tank and above the water level control component. The input end of the supply pump is connected to the liquid storage container, and the output end of the supply pump is connected to the pipeline through a connecting pipe. The supply pump is also connected to the control system signal.

[0011] Preferably, the water level control component includes an overflow system and a water supply system; The water supply system is installed on the inner bottom wall of the sand trough, and the overflow system is installed on the inner side wall of the sand trough, so that the water supply system supplies water from the bottom of the sand trough, and the water rises to wet the second filling medium until the water level reaches the overflow system, and the excess water is discharged from the overflow system. The overflow system is equipped with a filter to prevent the discharge of the second filling medium.

[0012] Preferably, the third test assembly includes four iron drums, four third filling media, a third contaminant, and an integrated sensor; All four iron drums are made of large cylindrical iron sand troughs, and all four iron drums are buried in the test field in the outdoor environment, so that the internal medium can experience the real day and night temperature changes and the influence of rainfall and evaporation. The four types of third filling media are medium sand, silt, silty clay and undisturbed soil, and the medium sand, silt, silty clay and undisturbed soil are respectively filled into the corresponding iron buckets; The integrated sensor is pre-embedded in one of four third filling media and is located in the center of the iron barrel. The third pollutant is LNAPL. After the third filling medium in each iron drum is naturally compacted, an equal amount of LNAPL is added to each iron drum to simulate a known amount of pollution event. The data acquisition and analysis unit performs data measurement and analysis on the third experimental component.

[0013] Preferably, the ground-penetrating radar includes a radar main unit, an antenna, and a power supply system; The power supply system is electrically connected to the radar host and the antenna respectively to provide power to the radar host and the antenna. The antenna is buried in the first filling medium, the second filling medium and the third filling medium. The radar host transmits high-frequency, short-pulse electromagnetic waves into the ground through the antenna. The electromagnetic waves are reflected at the interfaces of different media. Some of the electromagnetic wave energy is reflected back to the ground surface and captured by the receiving antenna. The receiving antenna receives the electromagnetic wave signals reflected back from each underground layer and transmits them back to the radar host for processing.

[0014] Preferably, the auxiliary measuring equipment includes a gas chromatography-mass spectrometry system and a soil pollution detector; The gas chromatography-mass spectrometry instrument is used for the measurement of the first and second packing media, and the soil pollution detector is used for the measurement of the third packing media.

[0015] Preferably, the specific operation is as follows: S1. The first filling medium is uniformly filled into the main container and its lithological characteristics are recorded. The soil moisture content is increased in a 2% gradient. After each moisture content gradient is reached and the moisture is evenly distributed, the soil column is scanned with ground-penetrating radar to collect the original radar data. The first pollutant is added. After each gradient is added and the pollutant is evenly mixed, the ground-penetrating radar is scanned again to take samples from multiple points in the main container. The samples are immediately sealed and sent to the laboratory. The accurate LNAPL saturation is determined by gas chromatography-mass spectrometry and used as the true value. S2, the second filling medium is filled into the sand tank, the water level control component is activated to stabilize the groundwater level at the preset height, before leakage, the entire sand tank is scanned by ground-penetrating radar to obtain the background radar profile, the leakage simulation component is activated to inject the second filling medium into the sand tank at a set constant rate, and ground-penetrating radar scans are repeated at fixed time intervals throughout the leakage process and the settling period after the leakage stops to obtain a series of time-slice radar data. After the test, the second filling medium in the sand tank is systematically destructively gridded sampled, and the LNAPL content of each sample is analyzed by gas chromatography-mass spectrometry to draw the final spatial distribution map of the pollutants. S3. Four iron drums, each containing one of the four third filling media, were buried in the test site. Integrated sensors were pre-buried and allowed to settle and stabilize naturally. An equal amount of the third pollutant was added to each iron drum. Ground-penetrating radar was used to scan each iron drum, and the moisture content data returned by the integrated sensors was recorded. A soil pollution detector was used to perform in-situ detection at preset fixed measuring points inside the iron drums, and the daily variation data of LNAPL saturation was recorded. Weather station data was recorded simultaneously to help interpret the data changes. S4 uses Radar 7 and Geogiga professional software to preprocess the raw radar data. From the processed radar data, the preset electromagnetic wave amplitude attributes are quantitatively extracted. The radar signal features extracted from each detection are precisely matched with the real values ​​obtained by synchronous measurement in the scene to form a structured data record. S5 summarizes, cleans, and standardizes the structured data records generated by all the first, second, and third experimental components to construct the final calibration dataset.

[0016] The beneficial effects are as follows: 1. The entire method, centered on non-destructive testing with ground-penetrating radar (GPR) and combined with supporting rapid in-situ detection equipment, demonstrates significant advantages in efficiency, non-destructiveness, and repeatability compared to traditional borehole sampling and laboratory analysis in both the experimental and final application stages. In the experimental stage, the same physical model can be continuously and dynamically monitored without damaging its structure; in the application stage, the trained model, combined with GPR field scanning, enables large-scale, rapid surveys and detailed evaluations of contaminated sites, greatly improving the efficiency and scope of pollution investigations and avoiding the risks of secondary pollution and high costs associated with invasive sampling.

[0017] 2. Through a three-dimensional system design consisting of the first, second, and third experimental components, it can systematically and controllably generate a standardized, high-precision calibration dataset covering the four-dimensional relationship of "lithology-water content-LNAPL saturation-Ground radar response". This fundamentally solves the core problems of traditional methods being unable to simulate dynamic processes, unclear "input-output" relationships in field measured data, and the scarcity and low quality of training data for machine learning models. It provides a solid data foundation for the accurate identification of LNAPL pollution under complex heterogeneous and multiphase conditions. Attached Figure Description

[0018] Figure 1 This is a partial cross-sectional structural schematic diagram of the first experimental component of the present invention; Figure 2 This is a partial cross-sectional structural schematic diagram of the second test component of the present invention; Figure 3 This is a partial cross-sectional structural schematic diagram of the third test component of the present invention; Figure 4 This is a top view of the third experimental component of the present invention.

[0019] In the diagram: 1. First test component; 101. Main container; 102. First filling medium; 103. First contaminant; 2. Second test component; 201. Sand tank; 202. Second filling medium; 203. Second contaminant; 3. Third test component; 301. Iron drum; 302. Third filling medium; 303. Third contaminant; 304. Integrated sensor; 4. Ground-penetrating radar; 401. Radar host; 402. Antenna; 6. Pipeline; 7. Liquid storage container; 8. Supply pump; 9. Overflow system; 10. Water supply system. Detailed Implementation

[0020] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0021] An indoor physical simulation test device for electromagnetic wave amplitude and spectrum under multiple factors includes a main test mechanism and a data acquisition and analysis mechanism. The main test mechanism enables indoor physical simulation tests, while the data acquisition and analysis mechanism can measure and analyze the electromagnetic wave amplitude and spectrum under multiple factors. Compared with traditional measurement methods, this device does not require drilling and has high repeatability, which greatly improves its applicability.

[0022] like Figures 1-4 As shown, the overall test facility is used to simulate pollution scenarios with different media lithology, water content, and LNAPL saturation in areas below ground level and above groundwater.

[0023] Specifically, the overall test facility includes a first test component 1, a second test component 2, and a third test component 3. The first test component 1 is used to simulate homogeneous media with different lithologies, water contents, and LNAPL saturation under controlled conditions. The second test component 2 is used to simulate the dynamic leakage and migration process of LNAPL in the medium. The third test component 3 is used to simulate and monitor LNAPL pollution in undisturbed or remolded soil filled with different media in an outdoor environment to obtain a validation dataset under semi-natural conditions. The overall test facility is the carrier of the physical scenario and consists of three test components with clearly defined functions and progressive order.

[0024] like Figure 1 As shown, the first experimental component 1 includes a main container 101, a first filling medium 102, and a first contaminant 103. Under ideal laboratory conditions, a precise quantitative relationship is established between the four factors: lithology, water content, LNAPL saturation, and radar signal.

[0025] Specifically, the first filling medium 102 uses soil with different particle size distributions to simulate different lithological characteristics. The first main container 101 is a sealed container made of a material with low dielectric response to electromagnetic waves. It is made of transparent plexiglass with a thickness of 10mm and is processed into a cylindrical container with an inner diameter of 300mm and a height of 1000mm. The dielectric constant of the plexiglass is stable (ε≈3.6) and much lower than that of soil and water. Its reflection and absorption of high-frequency electromagnetic waves are negligible, ensuring that the radar signal mainly reflects the information of the internal medium. The container is equipped with a drainage / sampling valve at the bottom and a lid on the top to prevent water evaporation. It is used to contain the first filling medium 102. The first pollutant 103 is LNAPL. The LNAPL is uniformly distributed in the first filling medium 102 to simulate a static and homogeneous pollution distribution state. The data acquisition and analysis mechanism performs data measurement and analysis on the first test component 1.

[0026] In this embodiment, the first filling medium 102 is selected from typical soil along the Yellow River. After being air-dried and sieved through a 2mm sieve, pure sand (>80% sand particles), silt (>80% silt particles), silty clay (clay content >20%), and sand-silt-clay mixed soils of different proportions are artificially prepared. The particle size distribution of all media is completed by the Henan Provincial Rock and Mineral Testing Center, which has national qualifications, in accordance with the "Standard for Geotechnical Testing Methods" to ensure the accuracy of lithological characteristic parameters. The first pollutant 103 uses diesel oil as the representative pollutant of LNAPL, which has stable physicochemical properties and is representative.

[0027] like Figure 2 As shown, the second test component 2 includes a sand tank 201, a leakage simulation component, a second filling medium 202, a water level control component, and a second contaminant 203. This component is used to simulate the dynamic process of LNAPL leakage, migration, and interaction with groundwater.

[0028] Specifically, the water level control component is installed on the inner bottom wall of the sand tank 201, and the leakage simulation component is fixedly installed on the inner wall of the sand tank 201 with the leakage simulation component higher than the water level control component. The sand tank 201 is a sealed container made of a material with low dielectric response to electromagnetic waves and is made of 20mm thick plexiglass. Its internal dimensions are 2000mm (length) × 000mm (width) × 800mm (height). One of its long sides is the observation surface, which is convenient for parallel arrangement with the data acquisition and analysis mechanism. It is used to contain the second filling medium 202. The second pollutant 203 is LNAPL. The leakage simulation component is used to controllably inject LNAPL into the second filling medium 202 to simulate the infiltration, lateral distribution and dynamic action of the second pollutant 203 in the second filling medium 202 and the water table. The data acquisition and analysis mechanism performs data measurement and analysis on the second test component 2.

[0029] In this embodiment, the second filling medium 202 is filled with uniform medium sand (particle size 0.25-0.5mm) to simplify the transport model and highlight the dynamic process. The second pollutant 203 is made of the same material as the first pollutant 103 and uses diesel oil.

[0030] The leakage simulation component includes a pipe 6, a liquid storage container 7, a supply pump 8, and a control system. In this embodiment, the pipe 6 is a PVC pipe with an inner diameter of 10mm, horizontally buried in the middle of the sand tank 201, 400mm from the bottom of the tank (i.e., 100mm above the simulated water level). The pipe 6 has 1mm diameter leakage holes evenly drilled along its length. The pipe 6 is installed inside the sand tank 201 and above the water level control component. The input end of the supply pump 8 is connected to the liquid storage container 7, and the output end of the supply pump 8 is connected to the pipe 6 via a connecting pipe. An injection pump is used as the supply pump 8, with a flow rate accuracy of ±0.5%. The liquid storage container 7 is a sealed glass bottle filled with diesel fuel. The injection pump is connected to the pipe 6 via a Teflon tube. The supply pump 8 is connected to the control system via a signal connection. Using the injection pump's built-in programmable controller, the single injection volume, flow rate, and interval time can be set to simulate different scenarios such as instantaneous leakage and continuous seepage.

[0031] The water level control components include an overflow system 9 and a water supply system 10. The water supply system 10 is installed on the inner bottom wall of the sand tank 201, and the overflow system 9 is installed on the inner side wall of the sand tank 201. A circular overflow port with an inner diameter of 20 mm is opened at a height of 300 mm from the bottom on one side panel of the sand tank 201. An adjustable transparent overflow pipe is connected to the overflow port. A stainless steel filter screen with a pore size of 0.075 mm (200 mesh) is tightly attached to the inside of the overflow port to effectively prevent the loss of fine particulate media. The water supply system 10 supplies water from the bottom of the sand tank 201. The water rises and wets the second filling medium 202 until the water level reaches the overflow system 9. Excess water is discharged from the overflow system 9. A filter screen is installed in the overflow system 9 to prevent the second filling medium 202 from being discharged. The water supply system 10 uses a Marshall bottle as a constant head water supply device and slowly supplies water to the bottom of the sand tank 201 through a perforated pipe arranged at the bottom.

[0032] like Figure 3 As shown, the third experimental component 3 includes four iron barrels 301, four third filling media 302, a third pollutant 303, and an integrated sensor 304. This component is set up in the soil of the experimental area to connect indoor experiments with real field environments.

[0033] Specifically, each of the four iron drums 301 uses a large columnar iron sand trough. Four custom-made galvanized steel drums, 800mm in diameter and 1200mm in height, with a wall thickness of 2mm, are buried underground, with 200mm protruding above ground to reduce the influence of side wall temperature. All four iron drums 301 are placed in an outdoor test site, allowing their internal media to experience real-world day-night temperature changes and the effects of rainfall and evaporation. The four third-stage filling media 302 are medium sand, silt, silty clay, and undisturbed soil, respectively. These are filled into their corresponding iron drums 301, maintaining their original structure and density. Integrated sensors 304 are pre-embedded in the four types of media. The third filling medium 302 is filled with an integrated sensor 304 located in the center of the iron drum 301. In this embodiment, an integrated sensor 304 is buried at depths of 20cm, 50cm, and 80cm on the vertical cross-section of the center of each iron drum 301. The integrated sensor 304 is a TDR (Time Domain Reflectometry) soil moisture sensor, which records the volumetric water content in real time through a data acquisition device. The third pollutant 303 is LNAPL. After the third filling medium 302 in each iron drum 301 is naturally compacted, an equal amount of LNAPL is added to each iron drum 301 to simulate a known pollution event. The data acquisition and analysis mechanism performs data measurement and analysis on the third experimental component 3.

[0034] In this embodiment, the third pollutant 303 is made of the same material as the second pollutant 203 and the first pollutant 103, and uses diesel fuel. like Figures 1-4 As shown, the data acquisition and analysis unit includes a ground-penetrating radar 4 and auxiliary measuring equipment. The ground-penetrating radar 4 is used to perform non-destructive testing on the medium within the main test unit and to acquire raw radar echo signals. The auxiliary measuring equipment is used to simultaneously acquire the lithological characteristics and water content parameters of the medium.

[0035] Specifically, the ground-penetrating radar 4 includes a radar host 401, an antenna 402, and a power supply system. In this embodiment, the radar host 401 is an SIR-4000 type. The power supply system of the ground-penetrating radar host is electrically connected to both the radar host 401 and the antenna 402 to provide power to both. In this embodiment, the power supply system uses a dedicated rechargeable lithium battery for the radar host 401. The antenna 402 is embedded in the first filling medium 102, the second filling medium 202, and the third filling medium 302. The radar host 401 transmits high-frequency, short-pulse electromagnetic waves into the ground through the antenna 402. The electromagnetic waves are distributed in different media. When the interface of the material is reflected, some of the electromagnetic wave energy is reflected back to the surface and captured by the receiving antenna 402. The receiving antenna 402 receives the electromagnetic wave signals reflected from various underground layers and transmits them back to the radar host 401 for processing. The antenna 402 is equipped with a 1.5GHz high-frequency shielded antenna (for high-resolution detection of the first test component 1), a 900MHz center frequency antenna (for dynamic process monitoring of the second test component 2), and a 400MHz low-frequency antenna (for deeper detection of the third test component 3) according to the component size and detection depth. The antenna 402 is synchronized with the radar host 401 through the measuring wheel to achieve ranging and positioning.

[0036] The auxiliary measurement equipment includes a gas chromatography-mass spectrometry (GC-MS) instrument and a soil pollution detector. The GC-MS instrument is used for the measurement of the first packed medium 102 and the second packed medium 202. In this embodiment, the GC-MS instrument is an Agilent 7890B-5977B GC-MS, which is used to perform precise quantitative analysis of diesel content (LNAPL saturation) on soil samples collected from the first test component 1 and the second test component 2. The soil pollution detector is used for the measurement of the third packed medium 302. In this embodiment, the soil pollution detector is an Ecoprobe 5 on-site soil pollution detector, which is based on the principle of laser-induced fluorescence and can quickly obtain the relative concentration of total petroleum hydrocarbons in the soil within 30 seconds. After calibration with GC-MS data, it can be used as a rapid, in-situ measurement tool for long-term monitoring of the third test component 3.

[0037] The original radar echo signal, lithological characteristics, and water content state parameters, together with the preset LNAPL saturation, constitute the calibration dataset used to build the machine learning model.

[0038] Working principle: The experimental method of the indoor physical simulation test device for electromagnetic wave amplitude and spectrum under multiple factors is as follows: S1, the first filling medium 102 (5 kg of dry medium of a specific grade (such as silt) is filled into the main container 101 in layers of 10 cm each, and gently tapped and compacted to the predetermined bulk density) is evenly filled into the main container 101 and its lithological characteristics are recorded. Initial state (water content ≈ 0%): use a 1.5 GHz antenna to scan vertically along the outer wall of the container and record the radar profile a1. Take a 50g soil sample from the sampling valve of the container, seal it and label it b1, and set it aside. Increase the soil moisture content in a 2% gradient. After each moisture content gradient is reached and the moisture is evenly distributed (calculate and add water equivalent to 2% mass moisture content, spray evenly with a sprayer and stir, let stand for 24 hours to allow the moisture to equalize), repeat the above operation. Use a ground-penetrating radar 4 to scan the soil column and collect raw radar data. Take a sample b2, and repeat the above operation until the moisture content increases to 20%. A total of 11 sets of data pairs (a1, b1) to (a10, b10) are obtained. Immediately send soil samples b0-n10 into an oven and dry them at 105℃ to constant weight. Calculate the measured mass moisture content θ_w of each sample. Select the state with a moisture content of 10% (corresponding to sample b5) as the starting point, calculate and add the first contaminant 103 equivalent to 2% mass oil content, and add it to each sample. A gradient was added, and after the contaminants were mixed evenly, the mixture was allowed to stand for 12 hours. A ground-penetrating radar 4 was used to scan the surface, obtaining profile a5-1. Multiple samples were taken from the main container 101 to obtain b5-1. The above operation was repeated, adding the first contaminant 103 sequentially until the oil content reached 10%, resulting in 5 sets of data pairs (a5-1, b5-1) to (a5-5, b5-5). These samples were immediately sealed and sent to the laboratory. The precise LNAPL saturation was determined using gas chromatography-mass spectrometry (GC-MS) and used as the true value. Samples S5-1 to S5-5 were immediately sealed in brown sample bottles, stored at low temperature, and sent to the laboratory within 24 hours. The precise diesel mass fraction was determined using GC-MS and converted to volume saturation S_o. The medium was then changed to another particle size distribution (such as pure sand), and all the above steps were repeated until all planned lithological types were covered. S2, fill the sand tank 201 with the second filling medium 202 to a height of 700mm, activate the water level control component to allow the water to rise slowly from the bottom until the water level stabilizes at the overflow port height (300mm), shut off the water supply, and allow the system to stabilize for 24 hours. Before the leak, use a ground-penetrating radar 4 to perform a comprehensive scan of the entire sand tank to obtain the background radar profile (scan along the center line of the long side of the sand tank with a 900MHz antenna to obtain the background radar profile c0). Activate the leak simulation component and continuously inject diesel fuel at a constant rate of 5 mL / min, for a total injection of 1000mL (simulating a small-scale leak of 5L). Inject the second filling medium 202 at a set constant rate. Activate the injection pump and the continuous acquisition mode of the ground-penetrating radar 4. Set the radar to automatically scan every 5 minutes along a fixed survey line. Repeat the ground-penetrating radar 4 scan at fixed time intervals throughout the leak process and during the settling period after the leak stops to obtain a series of time-slice radar data (after the leak begins, the radar automatically records time-series profiles c1, c2). ...cn, and at the same time, a high-definition camera was used to record the macroscopic migration morphology of the pollution plume through the transparent sidewall. After the injection was completed, monitoring continued for 12 hours to record the redistribution process of pollutants in a static state. After the experiment, the water in the sand tank was drained, and the second filling medium 202 in the sand tank 201 was systematically (with a 10cm×10cm grid) destructively sampled. The three-dimensional coordinates of each sample were recorded, and the LNAPL content of each sample was analyzed by gas chromatography-mass spectrometry. The final spatial distribution map of the pollutants was drawn for accurate spatial registration and comparison with the radar profile cn at the final moment. S3. Four iron drums 301, each containing one of the four third filling media 302, were buried in the test site and allowed to undergo at least two complete rainfall-drying cycles (2 weeks) to ensure that the media and sensor readings were sufficiently stable. The initial moisture content background value was recorded. The integrated sensor 304 was pre-buried and allowed to settle naturally. 2.0L of the third pollutant 303 was added to each iron drum 301 at one time. Every morning, the ground-penetrating radar 4 was used to scan each iron drum 301 in a cross pattern. The moisture content data returned by the integrated sensor 304 was recorded. Using a soil pollution detector, the probe was inserted to a depth of 30cm at the center of the iron drum 301 and at four points 20cm away from the center. The fluorescence intensity value (RFU) was read and recorded. This value can be converted into relative saturation after calibration with gas chromatography-mass spectrometry. The daily variation data of LNAPL saturation was recorded. Weather station data was recorded simultaneously to help interpret the data changes. This process lasted for 30 days. S4. All raw radar data are preprocessed using Radar 7 and Geogiga professional software (zero-point correction, background removal, bandpass filtering, energy attenuation gain). For the dynamic data of the second experimental component 2, time slice analysis is also required. From the processed radar data, the preset electromagnetic wave amplitude attributes are quantitatively extracted. The radar signal features extracted from each detection (for each effective radar detection (corresponding to a specific scenario), multiple attributes of the waveform or time window are extracted within the target medium depth range) are precisely matched with the real values ​​obtained by synchronous measurement in the scenario to form a structured data record (create a total database, each record contains the following fields: scenario ID, input features, output label, and metadata). S5. Summarize, clean, and standardize all structured data records generated by the first experimental component 1, the second experimental component 2, and the third experimental component 3 to construct the final calibration dataset (the data generated by the first experimental component 1 and the second experimental component 2 are used as the training set (70%) to train the model; the data generated by the third experimental component 3 are used as an independent validation set (30%) to evaluate the generalization ability and robustness of the model). S6, the standardized, multi-dimensional calibration dataset generated by this experimental method is specifically used to train and validate a supervised learning machine learning model. This model can learn the complex mapping relationship between input features and output targets (the random forest regression algorithm is selected as the machine learning model, and the Python Scikit-learn library is used to train the model with "input features" as X and "output labels" S_o as y, and the model hyperparameters are optimized through grid search cross-validation). The trained model can be applied to real-world contaminated sites, and by inputting the corresponding parameters measured in the field, it can achieve rapid, non-destructive, qualitative and quantitative evaluation of soil LNAPL pollution (the validation set is used to evaluate model performance. Three core indicators are calculated: the coefficient of determination (R²) between predicted and measured values, target > 0.85; the root mean square error (RMSE), target < 5% saturation; and the Nash efficiency coefficient (NSE), target > 0.80. Only models that simultaneously meet the above indicators are considered qualified). The trained model is integrated into an interactive software interface.

[0039] The embodiments of the present invention described above do not constitute a limitation on the scope of protection of the present invention. The basic concept of the present invention is that, by using ground-penetrating radar non-destructive testing as the core, combined with supporting rapid in-situ detection equipment, this method demonstrates significant advantages in terms of high efficiency, non-destructive nature, and repeatability compared to traditional borehole sampling-laboratory analysis in both the experimental and final application stages. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. An indoor physical simulation experimental device for electromagnetic wave amplitude and spectrum under multiple factors, characterized in that, Including the main testing facility and the data collection and analysis facility; The overall test facility is used to simulate pollution scenarios with different media lithology, water content and LNAPL saturation in areas below ground and above groundwater. The acquisition and analysis unit includes a ground-penetrating radar (4) and an auxiliary measuring device (5). The ground-penetrating radar (4) is used to perform non-destructive testing on the medium in the main test unit and acquire the original radar echo signal. The auxiliary measuring device (5) is used to simultaneously acquire the lithological characteristics and water content state parameters of the medium. The original radar echo signal, lithological characteristics, and water content state parameters, together with the preset LNAPL saturation, constitute the calibration dataset used to build the machine learning model.

2. The indoor physical simulation experimental device for electromagnetic wave amplitude and spectrum under multiple factors as described in claim 1, characterized in that, The overall test facility includes a first test component (1), a second test component (2), and a third test component (3). The first test component (1) is used to simulate homogeneous media with different lithologies, water content, and LNAPL saturation under controlled conditions. The second test component (2) is used to simulate the dynamic leakage and migration process of LNAPL in the medium. The third test component (3) is used to simulate and monitor LNAPL pollution in undisturbed or remolded soil filled with different media in an outdoor environment to obtain a verification dataset under semi-natural conditions.

3. The indoor physical simulation experimental device for electromagnetic wave amplitude and spectrum under multiple factors as described in claim 2, characterized in that, The first test component (1) includes a main container (101), a first filling medium (102), and a first contaminant (103). The first filling medium (102) uses soil with different particle size distribution to simulate different lithological characteristics. The first main container (101) is a sealed container made of a material with low dielectric response to electromagnetic waves to contain the first filling medium (102). The first pollutant (103) is LNAPL. The LNAPL is uniformly distributed in the first filling medium (102) to simulate a static and homogeneous pollution distribution state. The data acquisition and analysis unit performs data measurement and analysis on the first test component (1).

4. The indoor physical simulation experimental device for electromagnetic wave amplitude and spectrum under multiple factors as described in claim 3, characterized in that, The second test component (2) includes a sand tank (201), a leakage simulation component, a second filling medium (202), a water level control component, and a second pollutant (203); The water level control component is installed on the inner bottom wall of the sand tank (201), and the leakage simulation component is fixedly installed on the inner wall of the sand tank (201) and the leakage simulation component is higher than the water level control component. The sand tank (201) is a sealed container made of a material with low dielectric response to electromagnetic waves, used to contain the second filling medium (202). The second pollutant (203) is LNAPL. The leakage simulation component is used to controllably inject LNAPL into the second filling medium (202) to simulate the infiltration, lateral distribution and dynamic action of the second pollutant (203) in the second filling medium (202) on the water level control component. The data acquisition and analysis unit performs data measurement and analysis on the second test component (2).

5. The indoor physical simulation test device for electromagnetic wave amplitude and spectrum under multiple factors according to claim 4, characterized in that, The leakage simulation components include a pipe (6), a liquid storage container (7), a supply pump (8), and a control system; The pipe (6) is installed in the sand tank (201) and located above the water level control component. The input end of the supply pump (8) is connected to the liquid storage container (7). The output end of the supply pump (8) is connected to the pipe (6) through a connecting pipe. The supply pump (8) is connected to the control system signal.

6. The indoor physical simulation experimental device for electromagnetic wave amplitude and spectrum under multiple factors according to claim 5, characterized in that, The water level control components include an overflow system (9) and a water supply system (10). The water supply system (10) is installed on the inner bottom wall of the sand trough (201), and the overflow system (9) is installed on the inner side wall of the sand trough (201). The water supply system (10) supplies water from the bottom of the sand trough (201), and the water flows upward to wet the second filling medium (202) until the water level reaches the overflow system (9). Excess water is discharged from the overflow system (9). The overflow system (9) is equipped with a filter screen to prevent the discharge of the second filling medium (202).

7. The indoor physical simulation test device for electromagnetic wave amplitude and spectrum under multiple factors according to claim 4, characterized in that, The third test component (3) includes four iron barrels (301), four third filling media (302), a third contaminant (303), and an integrated sensor (304). All four iron barrels (301) are made of large columnar iron sand troughs, and all four iron barrels (301) are buried in the test field in the outdoor environment, so that the internal medium can experience the real day and night temperature changes and the influence of rainfall and evaporation environment. The four types of third filling media (302) are medium sand, silt, silty clay and undisturbed soil, respectively, and the medium sand, silt, silty clay and undisturbed soil are respectively filled into the corresponding iron buckets (301); The integrated sensor (304) is pre-embedded in four third filling media (302), and the integrated sensor (304) is located in the center of the iron barrel (301); The third pollutant (303) is LNAPL. After the third filling medium (302) in each iron drum (301) is naturally compacted, an equal amount of LNAPL is added to each iron drum (301) to simulate a known amount of pollution event. The data acquisition and analysis unit performs data measurement and analysis on the third test component (3).

8. The indoor physical simulation test device for electromagnetic wave amplitude and spectrum under multiple factors according to claim 7, characterized in that, The ground-penetrating radar (4) includes a radar host (401), an antenna (402), and a power supply system; The power supply system is electrically connected to the radar host (401) and the antenna (402) respectively, so as to provide power to the radar host (401) and the antenna (402). The antenna (402) is buried in the first filling medium (102), the second filling medium (202) and the third filling medium (302). The radar host (401) transmits high-frequency, short-pulse electromagnetic waves into the ground through the antenna (402). The electromagnetic waves are reflected at the interface of different media, and some of the electromagnetic wave energy is reflected back to the ground surface and captured by the receiving antenna (402). The receiving antenna (402) receives the electromagnetic wave signals reflected back from each underground layer and transmits them back to the radar host (401) for processing.

9. The indoor physical simulation experimental device for electromagnetic wave amplitude and spectrum under multiple factors as described in claim 8, characterized in that, The auxiliary measuring equipment (5) includes a gas chromatograph-mass spectrometer (501) and a soil pollution detector (502). The gas chromatograph-mass spectrometer (501) is used for measuring the first packing medium (102) and the second packing medium (202), and the soil pollution detector (502) is used for measuring the third packing medium (302).

10. The test method of the indoor physical simulation test device for electromagnetic wave amplitude and spectrum under multiple factors according to claim 9, characterized in that, The specific steps are as follows: S1, the first filling medium (102) is uniformly filled into the main container (101), and its lithological characteristics are recorded. The soil moisture content was increased in a 2% gradient. After each moisture content gradient was reached and the moisture was evenly distributed, the soil column was scanned with ground-penetrating radar (4) to collect the original radar data. The first pollutant (103) was added. After each gradient was added and the pollutant was evenly mixed, the ground-penetrating radar (4) was used to scan again and samples were taken from multiple points in the main container (101). The samples were immediately sealed and sent to the laboratory. The accurate LNAPL saturation was determined by gas chromatography-mass spectrometry (501) as the true value. S2, fill the second filling medium (202) into the sand tank (201), start the water level control component to stabilize the groundwater level at the preset height, before leakage, use ground radar (4) to scan the entire sand tank to obtain the background radar profile, start the leakage simulation component to inject the second filling medium (202) at a set constant rate, and repeat the ground radar (4) scan at fixed time intervals throughout the leakage process and the static period after the leakage stops to obtain a series of time slice radar data. After the experiment, systematically destructively gridded sample the second filling medium (202) in the sand tank (201), analyze the LNAPL content of each sample with gas chromatography-mass spectrometry (501), and draw the final spatial distribution map of pollutants. S3, four iron drums (301) containing four different third filling media (302) are buried in the test site, integrated sensors (304) are pre-buried, and they are allowed to settle and stabilize naturally. An equal amount of third pollutant (303) is added to each iron drum (301). Each iron drum (301) is scanned with ground-penetrating radar (4), and the moisture content data returned by the integrated sensor (304) is recorded. Soil pollution detector (502) is used to perform in-situ detection at the fixed measuring points preset in the iron drum (301), and the daily variation data of LNAPL saturation is recorded. Weather station data is recorded simultaneously to help interpret the data changes. S4 uses Radan 7 and Geogiga professional software to preprocess the raw radar data. From the processed radar data, the preset electromagnetic wave amplitude attributes are quantitatively extracted. The radar signal features extracted from each detection are precisely matched with the real values ​​obtained by synchronous measurement in the scene to form a structured data record. S5. Summarize, clean and standardize the structured data records generated by all the first experimental component (1), the second experimental component (2) and the third experimental component (3) to construct the final calibration dataset.

Images