A device and method for detecting in-situ soil organic pollution while drilling

Abstract

A device and method for detecting organic soil pollution while drilling belong to the technical field of soil pollution detection. The air inlet pipe, middle section and air outlet pipe of the detection device of the present invention are arranged in sequence from front to back and fixedly connected. The signal acquisition circuit board I is fixedly connected to the front of the through hole II of the middle section, and the flow equalizing plate is fixedly connected to the back of the through hole II of the middle section; the signal acquisition circuit board II is fixedly connected to the front of the through hole IV in the guide section II of the air outlet pipe; the detection device of the present invention is directly mounted on the drill bit to avoid the loss of trace hot volatile gas during transportation; the distribution characteristics of olfactory cells in the nasal cavity of mammals with sensitive sense of smell and the characteristics of the nasal concha structure in terms of diversion and efficiency enhancement can be imitated to improve the sensitivity and accuracy of the detection device; the detection method of the present invention has the functions of unknown category detection and rapid self-learning and updating. It not only has very low requirements for training samples, but also can effectively identify unknown categories and increase the speed of model updating, thereby achieving more accurate detection and identification of organic soil pollution.

Description

Technical Field

[0001] The present invention belongs to the technical field of soil pollution detection, and in particular relates to a device and method for detecting in-situ soil organic pollution while drilling. Background Art

[0002] Organic pollution in soil not only degrades soil quality but also further pollutes the ecological environment through infiltration and precipitation, even endangering human health. It disrupts soil structure, hinders or inhibits soil microbial activity, and affects biological metabolism. All of these factors reduce soil organic matter and impact soil quality. Rapid and accurate detection of organic pollution in soil can effectively understand the soil environment, trace pollution sources, and implement scientific remediation. Currently, detection technologies for organic soil pollution primarily include chemical, chromatographic, and biological methods. However, none of these methods achieve true in situ detection and require soil sampling. The timeliness, accuracy, and standardization of on-site sampling significantly impact the determination of pollutant types, concentrations, and spatial distribution. Furthermore, soil is a highly heterogeneous medium, resulting in sampling errors far exceeding the analytical errors of standard traditional testing procedures. Furthermore, these methods require specialized personnel, making them infeasible for large-scale rapid pollution detection.

[0003] The membrane interface probe (MIP) is an in-situ tool that can thermally extract VOCs from the ground and transport the vapor to various detectors on the ground through Teflon tubing. It has the advantages of in-situ accuracy and environmental protection. This technology uses portable gas chromatographs, mass spectrometers and other gas detection equipment in conjunction with direct-push drilling equipment to achieve real-time monitoring of almost continuous distribution information of organic pollutants. However, in addition to the shortcomings of high cost and large size of these devices, they have a common disadvantage: the soil vapor sample will flow through a 30.5m long Teflon tube when transported before MIP detection. The vapor sample is diluted and partially absorbed by the Teflon tube, resulting in a high detection limit and affecting the detection accuracy. Therefore, inserting the detector into the drill bit for real-time soil organic pollution detection can improve detection accuracy.

[0004] Compared to the aforementioned detectors, metal oxide semiconductor-based artificial olfaction systems offer significant advantages, including small size, no need for sample pretreatment, low cost, high speed, and the ability to perform in-situ detection. They have been successfully used in laboratories to detect volatile gases such as gasoline contamination in soil and groundwater. Furthermore, the gas detection chamber is optimizable, enabling sampling while drilling to lower detection limits.

[0005] At the same time, existing data modeling and analysis strategies for artificial olfactory systems can only detect the types of pollution found in the training data, leaving unknown pollutants unidentified during actual testing. During actual testing, even when new organic pollutant types are detected, the model will still predict them as those found in the training data. This misprediction can reduce remediation efficiency and even lead to secondary soil contamination, exacerbating already contaminated soil. Summary of the Invention

[0006] The purpose of the present invention is to provide a high-precision, low-cost, high-reliability, while-drilling in-situ soil organic pollution detection device and detection method that can quickly and efficiently perform large-scale detection to address the problems existing in existing soil pollution detection technologies.

[0007] The in-situ soil organic pollution detection device while drilling of the present invention is composed of an air inlet pipe A, a signal acquisition circuit board IB, a middle section C, a flow equalizing plate D, a signal acquisition circuit board II E and an air outlet pipe F, wherein the air inlet pipe A is composed of an air inlet 1, a transition section I2 and a guide section I3, and the air inlet 1, the transition section I2 and the guide section I3 are arranged in sequence from front to back and are smoothly connected; the signal acquisition circuit board IB is composed of a substrate I4, a sensor group I6 and a resistor group I7, the substrate I4 is disc-shaped, and 11 guide ports of the guide port group I5 are evenly distributed on its circumference; a through hole I8 is provided near the lower end of the substrate I4; the four sensors of the sensor group I6 and the four resistors of the resistor group I7 are evenly spaced and fixed to the front of the substrate I4; the middle section C is composed of The circular tube 10 is composed of a pair of protrusions Ⅰ11 and a pair of protrusions Ⅱ13. The front end of the circular tube 10 is provided with a boss Ⅰ9 and the rear end is provided with a boss Ⅱ12; the lower end of the inner circle of the circular tube 10 is provided with a through hole Ⅱ14; the two protrusions of the protrusion pair Ⅰ11 and the two protrusions of the protrusion pair Ⅱ13 are located in the same circular cross-section in the middle of the circular tube 10, and the two protrusions of the protrusion pair Ⅰ11 are symmetrically fixed to the left and right sides of the center line aa of the upper end of the circular tube 10; the two protrusions of the protrusion pair Ⅱ13 are symmetrically fixed to the left and right ends of the circular tube 10; the flow equalizing plate D is composed of a gas collecting section 15, a drainage section 16, and a drainage plate group 17. The gas collecting section 15 is annular, and the upper end of the outer circle of the gas collecting section 15 is provided with a wiring port Ⅰ18; the drainage section 16 is a hollow frustum; the 11 drainage plate structures of the drainage plate group 17 The same, circular and evenly distributed, its outer end is fixed to the inner ring of the ring of the gas collecting section 15, and its back is fixed to the front of the drainage section 16; the rear end of the drainage plate group 17 is the air outlet 19; the signal acquisition circuit board ⅡE is composed of a substrate Ⅱ20, a sensor group Ⅱ22, a resistor group Ⅱ24, a power socket 26 and a data acquisition and transmission module 27. The substrate Ⅱ20 is disc-shaped, and the 10 guide ports of the guide port group Ⅱ21 are evenly distributed on its circumference; a wiring port Ⅱ23 is provided near the upper end of the substrate Ⅱ20; a through hole Ⅲ25 is provided near the lower end of the substrate Ⅱ20; the four sensors of the sensor group Ⅱ22 and the four resistors of the resistor group Ⅱ24 are evenly spaced and fixed to the front of the substrate Ⅱ20; the power socket 26 and the data acquisition and transmission module 27 are fixed to the substrate Ⅱ20; the outlet pipe F is composed of a guide section Ⅱ28, a transition section Ⅱ30 and an outlet Ⅱ32; the guide section Ⅱ28, the transition section Ⅱ30 and the outlet Ⅱ32 are arranged in sequence from front to back and are smoothly connected; two protrusions of protrusion pair Ⅲ29 are symmetrically provided on the left and right sides of the center line aa of the upper end of the same circular cross-section in the middle of the guide section Ⅱ28; two protrusions of protrusion pair IV33 are symmetrically provided on the left and right ends of the circular cross-section; a through hole IV34 is provided near the lower end of the center line aa of the circular cross-section; an outlet port 31 is provided in the middle position of the upper part of the transition section Ⅱ30; the air inlet pipe A, the middle section C and the air outlet pipe F are arranged in sequence from front to back and are fixedly connected; the signal acquisition circuit board IB is fixedly connected to the front of the through hole Ⅱ14 of the middle section C; the current equalizing plate D is fixedly connected to the back of the through hole Ⅱ14 of the middle section C;The signal acquisition circuit board ⅡE is fixed in front of the through hole Ⅳ34 of the guide section Ⅱ28 of the outlet pipe F.

[0008] The thickness d1 of the air inlet 1, transition section I2, and guide section I3 in the air inlet pipe A, and the guide section II28, transition section II30, and air outlet II32 in the air outlet pipe F are all 1-1.5 mm; the inner diameter D1 of the air inlet 1 and the air outlet II32 are both 3.2-4.5 mm, and the length L1 is both 5-7 mm; the inner diameter D2 of the guide section I3 and the guide section II28 are both 35-40 mm, and the length L3 is both 12-16 mm; the length L2 of the transition section I2 and the transition section II30 are both 8-10 mm; the diameter D3 of the substrate I4 in the signal acquisition circuit board IB and the substrate II20 in the signal acquisition circuit board IIE are both 35-40, and the thickness L4 is both 1-1.3 mm; The inner diameter D4 of the circular tube 10 in the middle section C is 35-40mm, the length L5 is 15.9-19.4mm, and the thickness d2 is 1-1.5mm; the outer diameter D5 of the gas collecting section 15 in the flow equalizing plate D is 35-40mm, the width L6 is 2.8-3.2mm, and the thickness d3 is 0.6-0.8mm; the front end outer diameter and thickness of the drainage section 16 are the same as the outer diameter and thickness of the gas collecting section 15, and the height L7 is 3.8-4.2mm; the length L9 of each drainage plate in the drainage plate group 17 is 10-11mm, the thickness L10 is 0.9-1mm, and the inner end height L8 is 3.3-3.5mm; the diameter D6 of the air outlet 19 is 3.1-3.3mm.

[0009] The detection method based on the in-situ soil organic pollution detection device while drilling of the present invention comprises the following steps:

[0010] 1) Establish a soil pollution odor signal feature rule library. Use the in-situ soil organic pollution detection device while drilling to collect odor sample data of different types of common / untreated pollution in the laboratory. Perform feature extraction on the odor sample data. The feature extractor uses wavelet transform. The extraction formula is:

[0011]

[0012] Then use the single-class support vector machine to learn the feature rules. The formula is:

[0013]

[0014] st(ω·K(x i ))≥ρ-ξ i ,ξ i ≥0

[0015] Where: K(x i ) represents the kernel function, ω is a weight vector, and its dimension is the same as the feature space K(x i) are the same, ∈ξ i is a non-negative relaxation factor, n is the number of samples; p∈(0,1) is a penalty term used to control the balance between the maximum distance from the origin and the number of support vectors in the sample; the important parameters that determine the boundary are ω, ρ, which need to be optimized; the Hinge loss function is used instead of ξ i , we get the unconstrained objective function; as shown in the formula:

[0016]

[0017] So that h(x) = ω T K(x i )-ρ, the decision function is expressed as formula:

[0018]

[0019] Where: h(x) is the characteristic rule learned from different types of odor samples; the characteristic rules of different types of pollution odor samples are stored in the rule base, which is located at the signal processing end;

[0020] 2) In-situ soil organic contamination detection application: an in-situ soil organic contamination detection device (hereinafter referred to as a detection device 39) is integrated and fixedly connected to the MIP probe 35, and is used in conjunction with the flow guide tube 36, the separation chamber 37, and the separation membrane 38. The separation chamber 37 is separated from the outside world by the separation membrane 38, the flow guide tube 36 is located at the front end of the separation chamber 37, and the detection device 39 is located at the upper end of the separation chamber 37. The MIP probe 35 enters the soil and heats the drill bit. The hot volatile gases from the soil are transported through the separation membrane 38 and the carrier gas in the flow guide tube 36 to the detection device 39, where they collect information on the hot volatile gases of the soil organic contamination and transmit it to the signal processing end.

[0021] 3) The signal processing end extracts features and establishes rules for the collected soil pollution odor sample data, using the same establishment method as step 1); then the collected soil odor sample rules are compared with the existing rules in the rule library in the logic gate. If the signal matches the rule successfully, the output signal is 1, indicating that the type of soil organic pollution has been detected; if the signal matches the rule successfully, the output signal is 0, indicating that the type of soil organic pollution is unknown and further testing using standard methods is required to determine the type of pollution. The rules are then expanded to the rule library, and the rule library can be continuously expanded.

[0022] The principle and working process of the present invention are as follows:

[0023] The in-situ soil organic pollution detection device while drilling of the present invention is fixed in a direct-push soil pollution detection drill bit, ensuring that the volatile gas signals of soil organic pollution are collected without loss as much as possible.

[0024] The present invention imitates the nasal cavity diversion structure characteristics of mammals with a keen sense of smell and designs the air inlet pipe A into a "small mouth and large cavity" shape. When the hot volatile gas of soil organic pollution enters the device with the carrier gas, this shape and structure can effectively reduce the gas flow rate and increase the contact time between the target gas and the metal oxide semiconductor gas sensor, thereby improving the sensitivity and response of the device.

[0025] The signal acquisition circuit board IB and signal acquisition circuit board IIE of the present invention imitate the distribution strategy of mammalian olfactory cells to improve the anti-interference ability to environmental background gases. The sensor group I array of the signal acquisition circuit board IB is arranged with a type that has specific identification of the target gas, and the sensor group II array of the signal acquisition circuit board IIE is arranged with a type that responds sensitively to environmental background gases, so as to assist in the detection of the target gas and improve the detection accuracy of the device.

[0026] The present invention simulates the uniform airflow effect of the nasal concha structure inside the nasal cavity of mammals with a keen sense of smell, and sets a flow equalizer D behind the signal acquisition circuit board IB. The upper edge of the signal acquisition circuit board IB is provided with a guide port. After the target gas passes through the front-end signal acquisition circuit board, it enters the flow equalizer D through the guide port group I, evenly mixing the carrier gas and target gas, and then impacting the signal acquisition circuit board IIE. The evenly mixed airflow disperses on the signal acquisition circuit board IIE and reacts with the sensor group II respectively. The setting of the flow equalizer D not only enables the sensor array to obtain more accurate soil organic pollution volatile gas information, avoiding the negative impact of uneven target gas concentration obtained by the sensor, but also reduces the speed of the target gas passing through the sensor surface and increases the pressure on the sensor surface, thereby making the sensor detection more accurate and reliable.

[0027] The detection method for the in-situ soil organic contamination detection device while drilling (WDT) of the present invention mimics a fundamental human ability to adapt to the environment. The feature rule base established corresponds to humans' initial category learning, the rule base corresponds to the brain's memory storage area, and the model application module corresponds to the human process of identifying learned and unlearned categories, as well as the human process of learning unknown categories. The proposed method features unknown category detection and rapid self-learning and updating. The core of this strategy is to learn feature rules for different categories using a single-class support vector machine, an unsupervised approach. The main advantages of this method are as follows: 1. Model training requires very few training samples. Furthermore, training is limited to learning rules, resulting in rapid training speed. 2. Category recognition is primarily based on learned rules, using logic gates. This results in fast recognition speed and is effective for identifying unknown categories. 3. Model updates require only learning new sample feature rules and adding them to the existing rule base, without requiring retraining on previous samples. This significantly improves model update speed and reduces the computational resources required for updates.

[0028] The beneficial effects of the present invention are that the detection device is directly mounted on the drill bit, thereby avoiding the loss of trace hot volatile gases during transportation. By imitating the distribution characteristics of olfactory cells in the nasal cavity of mammals with a keen sense of smell and the characteristics of the nasal concha structure that enhances the diversion efficiency, the sensitivity and accuracy of the detection device are improved. The detection method is proposed based on the process of imitating humans' learning to identify unknown substances. It has the functions of unknown category detection and rapid self-learning and updating. Compared with traditional recognition methods, it not only has very few requirements for training samples, but also can effectively identify unknown categories and increase the speed of model updating. It can achieve more accurate detection and identification of soil organic pollution. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 This is a structural diagram of the in-situ soil organic pollution detection device while drilling;

[0030] Figure 2 This is a cross-sectional view of the in-situ soil organic pollution detection device while drilling;

[0031] Figure 3 is a cross-sectional view of the intake pipe A;

[0032] Figure 4 This is the front view of the signal acquisition circuit board IB;

[0033] Figure 5 This is the left view of the signal acquisition circuit board IB;

[0034] Figure 6 This is the rear view of the signal acquisition circuit board IB;

[0035] Figure 7 It is a cross-sectional view of the middle section C;

[0036] Figure 8 It is the front view of the middle section C;

[0037] Figure 9 is a cross-sectional view of the current equalizing plate D;

[0038] Figure 10 It is the front view of the current equalizing plate D;

[0039] Figure 11 This is the front view of the signal acquisition circuit board IIE;

[0040] Figure 12 This is the left view of the signal acquisition circuit board IIE;

[0041] Figure 13 This is the rear view of the signal acquisition circuit board IIE;

[0042] Figure 14 is a cross-sectional view of the exhaust pipe F;

[0043] Figure 15 It is a front view of the exhaust pipe F;

[0044] Figure 16 Schematic diagram of the in-situ soil organic pollution detection device while drilling based on the artificial olfactory system and its bionic learning cognitive strategy;

[0045] Figure 17 This is a schematic diagram of the cognitive strategy framework for bionic learning;

[0046] Figure 18 Schematic diagram of the partition of data set A;

[0047] Where: A. Intake pipe B. Signal acquisition circuit board I C. Middle section D. Flow equalizing plate E. Signal acquisition circuit board II

[0048] F. Exhaust pipe 1. Air inlet 2. Transition section I 3. Diversion section I 4. Base plate I 5. Diversion port assembly I 6. Sensor assembly I 7. Resistor assembly I 8. Through hole I 9. Boss I 10. Round tube 11. Boss pair I 12. Boss II 13. Boss pair II

[0049] 14. Through hole II 15. Gas collecting section 16. Drainage section 17. Drainage plate group 18. Wiring port I 19. Gas outlet I 20. Substrate II 21. Diversion port group II 22. Sensor group II 23. Wiring port II 24. Resistor group II 25. Through hole III 26. Power socket 27. Data acquisition and transmission module 28. Diversion section II 29. Pair of protrusions III 30. Transition section II 31. Wire outlet 32. Gas outlet II 33. Pair of protrusions IV 34. Through hole IV 35. MIP probe 36. Diversion tube 37. Separation chamber 38. Separation membrane 39. Detection device DETAILED DESCRIPTION

[0050] The present invention will be described below with reference to the accompanying drawings.

[0051] like Figures 1 to 3 As shown, the in-situ soil organic pollution detection device while drilling of the present invention consists of an air inlet pipe A, a signal acquisition circuit board IB, a middle section C, a flow equalizing plate D, a signal acquisition circuit board IIE and an air outlet pipe F, wherein the air inlet pipe A consists of an air inlet 1, a transition section I2 and a guide section I3, and the air inlet 1, transition section I2 and guide section I3 are arranged in sequence from front to back and are smoothly connected.

[0052] like Figures 4 to 6 As shown, the signal acquisition circuit board IB is composed of a substrate I4, a sensor group I6 and a resistor group I7. The substrate I4 is disc-shaped, and the 11 guide ports of the guide port group I5 are evenly distributed on its circumference; a through hole I8 is provided near the lower end of the substrate I4; the four sensors of the sensor group I6 and the four resistors of the resistor group I7 are evenly spaced and fixed to the front of the substrate I4.

[0053] like Figure 7 and Figure 8 As shown, the middle section C is composed of a circular tube 10, a protrusion pair I11 and a protrusion pair II13. The front end of the circular tube 10 is provided with a boss I9 and the rear end is provided with a boss II12; the lower end of the inner circle of the circular tube 10 is provided with a through hole II14; the two protrusions of the protrusion pair I11 and the two protrusions of the protrusion pair II13 are located in the same circular cross-section in the middle of the circular tube 10, and the two protrusions of the protrusion pair I11 are symmetrically fixed to the left and right sides of the center line aa of the upper end of the circular tube 10; the two protrusions of the protrusion pair II13 are symmetrically fixed to the left and right ends of the circular tube 10.

[0054] like Figure 9 and Figure 10 As shown, the flow balancing plate D is composed of a gas collecting section 15, a drainage section 16, and a drainage plate group 17. The gas collecting section 15 is a circular ring, and a wiring port Ⅰ18 is provided at the upper end of the outer ring of the gas collecting section 15; the drainage section 16 is a hollow frustum; the 11 drainage plates of the drainage plate group 17 have the same structure and are evenly distributed in a circle, with their outer ends fixed to the inner ring of the circular ring of the gas collecting section 15, and their rear ends fixed to the front of the drainage section 16; the rear end of the drainage plate group 17 is the gas outlet 19.

[0055] like Figures 11 to 13 As shown, the signal acquisition circuit board ⅡE consists of a substrate Ⅱ20, a sensor group Ⅱ22, a resistor group Ⅱ24, a power socket 26 and a data acquisition and transmission module 27. The substrate Ⅱ20 is disc-shaped, and the 10 guide ports of the guide port group Ⅱ21 are evenly distributed on its circumference; a wiring port Ⅱ23 is provided near the upper end of the substrate Ⅱ20; a through hole Ⅲ25 is provided near the lower end of the substrate Ⅱ20; the four sensors of the sensor group Ⅱ22 and the four resistors of the resistor group Ⅱ24 are evenly spaced and fixed to the front of the substrate Ⅱ20; the power socket 26 and the data acquisition and transmission module 27 are fixed to the back of the substrate Ⅱ20.

[0056] like Figure 14 and Figure 15 As shown, the outlet pipe F is composed of a guide section II 28, a transition section II 30 and an air outlet II 32; the guide section II 28, the transition section II 30 and the air outlet II 32 are arranged in sequence from front to back and are smoothly connected; two protrusions of a protrusion pair III 29 are symmetrically provided on the left and right sides of the center line aa of the upper end of the same circular cross-section in the middle of the guide section II 28; two protrusions of a protrusion pair IV 33 are symmetrically provided on the left and right ends of the circular cross-section; a through hole IV 34 is provided near the lower end of the center line aa of the circular cross-section; an outlet 31 is provided at the upper middle position of the transition section II 30; the air inlet pipe A, the middle section C and the air outlet pipe F are arranged in sequence from front to back and are fixedly connected; the signal acquisition circuit board IB is fixedly connected to the front of the through hole II 14 of the middle section C; the flow equalizing plate D is fixedly connected to the back of the through hole II 14 of the middle section C; the signal acquisition circuit board II E is fixedly connected to the front of the through hole IV 34 of the guide section II 28 of the outlet pipe F.

[0057] like Figures 1 to 3As shown, in the described in-situ soil organic pollution detection device while drilling, the thickness d1 of the air inlet 1, transition section I2, and guide section I3 in the air inlet pipe A, and the guide section II28, transition section II30, and air outlet II32 in the air outlet pipe F are all 1-1.5 mm; the inner diameter D1 of the air inlet 1 and the air outlet II32 are both 3.2-4.5 mm, and the length L1 is both 5-7 mm; the inner diameter D2 of the guide section I3 and the guide section II28 are both 35-40 mm, and the length L3 is both 12-16 mm; the length L2 of the transition section I2 and the transition section II30 are both 8-10 mm.

[0058] like Figure 4 and Figure 5 As shown, the diameter D3 of the substrate I4 in the signal acquisition circuit board IB and the substrate II20 in the signal acquisition circuit board IIE are both 35-40 mm, and the thickness L4 are both 1-1.3 mm.

[0059] like Figure 7 As shown, the inner diameter D4 of the circular tube 10 in the middle section C is 35-40 mm, the length L5 is 15.9-19.4 mm, and the thickness d2 is 1-1.5 mm.

[0060] like Figure 9 and Figure 10 As shown, the outer diameter D5 of the gas collecting section 15 in the flow equalizing plate D is 35-40 mm, the width L6 is 2.8-3.2 mm, and the thickness d3 is 0.6-0.8 mm; the front end outer diameter and thickness of the drainage section 16 are the same as the outer diameter and thickness of the gas collecting section 15, and the height L7 is 3.8-4.2 mm; the length L9 of each drainage plate in the drainage plate group 17 is 10-11 mm, the thickness L10 is 0.9-1 mm, and the inner end height L8 is 3.3-3.5 mm.

[0061] like Figure 14 As shown, the diameter D6 of the air outlet 19 is 3.1-3.3 mm.

[0062] The detection method based on the in-situ soil organic pollution detection device while drilling of the present invention comprises the following steps:

[0063] 1) Establish a soil pollution odor signal feature rule library. Use the in-situ soil organic pollution detection device while drilling to collect odor sample data of different types of common / untreated pollution in the laboratory. Perform feature extraction on the odor sample data. The feature extractor uses wavelet transform. The extraction formula is:

[0064]

[0065] Then use the single-class support vector machine to learn the feature rules. The formula is:

[0066]

[0067] st(ω·K(x i ))≥ρ-ξ i ,ξ i ≥0

[0068] Where K(x i ) represents the kernel function, ω is a weight vector, and its dimension is the same as the feature space K(x i ) are the same, ξ i is a non-negative relaxation factor, n is the number of samples; p∈(0,1) is a penalty term used to control the balance between the maximum distance from the origin and the number of support vectors in the sample; the important parameters that determine the boundary are ω, ρ, which need to be optimized; the Hinge loss function is used instead of ξ i , we get the unconstrained objective function; as shown in the formula:

[0069]

[0070] So that h(x) = ω T K(x i )-ρ, the decision function can be expressed as formula:

[0071]

[0072] h(x) is the characteristic rule learned from different types of odor samples; the characteristic rules of different types of pollution odor samples are stored in the rule base, which is located at the signal processing end;

[0073] 2) If Figure 16 As shown, in the application of in-situ soil organic pollution detection, the in-situ soil organic pollution detection device while drilling (hereinafter referred to as the detection device 39) is integrated and fixedly connected to the MIP probe 35, and is used in conjunction with the guide tube 36, the separation chamber 37 and the separation membrane 38, wherein the separation chamber 37 is separated from the outside by the separation membrane 38, the guide tube 36 is located at the front end of the separation chamber 37, and the detection device 39 is located at the upper end of the separation chamber 37; the MIP probe 35 enters the soil and heats the drill bit, and the hot volatile gas of the soil is transported to the detection device 39 through the separation membrane 38 and the carrier gas in the guide tube 36 to collect the hot volatile gas information of the soil organic pollution and transmit it to the signal processing end;

[0074] 3) The signal processing end extracts features and establishes rules for the collected soil contamination odor sample data, using the same establishment method as step 1). The collected soil odor sample rules are then compared with the existing rules in the rule base in a logic gate. If the signal matches the rule, the output signal is 1, indicating that the type of soil organic pollution has been detected. If the signal fails to match the rule, the output signal is 0, indicating that the type of soil organic pollution is unknown and further testing using standard methods is required to determine the type of pollution. The rules are then expanded to the rule base, which can be continuously expanded.

[0075] Example verification:

[0076] like Figure 17 As shown, the detection method of the in-situ soil organic pollution detection device while drilling of the present invention mainly consists of three modules, namely model training, rule base and model application. The method of the invention is described and verified in conjunction with the examples below.

[0077] Examples of identifying different types of soil pesticide residues:

[0078] In order to verify the effectiveness and superiority of the detection method of the invented in-situ soil organic pollution detection device while drilling, a soil pesticide residue classification test experiment was carried out.

[0079] To obtain a sufficient number of soil samples contaminated with a variety of pesticide types and brands, the soil to be tested was artificially prepared. Leaching experiments were used to prepare the contaminated soil samples, which better reflect the actual diffusion of pesticides in soil. Three pesticide types were selected: chlorpyrifos, cyfluthrin, and mancozeb. The prepared soil was then sampled for odor information using the proposed in-situ soil pesticide residue detection device based on an artificial olfactory system. The sensor array included the following sensors: front-end signal acquisition circuit boards: MS1100, TGS2602, TGS2610, and MP901; and back-end signal acquisition circuit boards: TGS2620, MP-5, MP-702, and MP-4. 160 samples were collected for each pesticide, and 40 samples were collected for healthy soil. A total of 520 odor samples were obtained.

[0080] Model training and rule base establishment: 1. First, feature extraction is performed on 520 odor samples, using wavelet transform technology as a feature extractor to extract the features of the odor data;

[0081] 2. Using a single-class support vector machine for feature rule learning, the single-class support vector machine constructs a separating hyperplane, where the target is the maximum distance between the target data sample point and the distant point. If a new data sample is within the boundary, it is considered to belong to the target data; otherwise, it is considered not to belong to the target data. In high-dimensional space, data is usually linearly separable, so a kernel function is introduced to map the data to high-dimensional space. Based on the above description, the process of finding the separating hyperplane is defined as a quadratic programming problem, as shown in the formula:

[0082]

[0083] st(ω·K(x i ))≥ρ-ξ i ,ξ i ≥0

[0084] Where K(x i ) represents the kernel function, ω is a weight vector, and its dimension is the same as the feature space K(x i ) are the same, ξ i is a non-negative relaxation factor, n is the number of samples. p∈(0,1) is a penalty term that controls the balance between the maximum distance from the origin and the number of support vectors in the sample. The important parameters that determine the margin are ω and ρ, which need to be optimized. The Hinge loss function is used instead of ξ. i , we can get the unconstrained objective function. As shown in the formula:

[0085]

[0086] So that h(x) = ω T K(x i )-ρ, then the decision function can be expressed as formula:

[0087]

[0088] Then h(x) is the feature rule learned from different types of odor samples.

[0089] The 520 odor samples were divided into three datasets. In each dataset, two pesticides and healthy soil were present as known classes in both the training set and the test set, and one pesticide was present only in the test set as an unknown pesticide. In each dataset, 80% of the samples were used as training sets and 20% of the samples were used as test sets. The samples of the training set and the test set were selected by stratified random sampling. It should be noted that in order to test the system's ability to detect unknown pesticides, each test set contains all the unknown pesticide samples. The division of dataset A is as follows: Figure 18As shown, mancozeb is used as an unknown pesticide; the division of datasets B and C is the same as that of dataset A, with chlorpyrifos as an unknown pesticide in dataset B and cyfluthrin as an unknown pesticide in dataset C.

[0090] The three data sets A, B and C were analyzed and identified by using artificial neural network (ANN), random forest (RF), support vector machine (SVM), naive Bayes (NB) and the BLCS detection method of the in-situ soil organic pollution detection device while drilling. The results are shown in the table:

[0091] Table 1. Classification results of dataset A

[0092]

[0093] Table 2. Classification results of dataset B

[0094]

[0095] Table 3. Classification results of dataset C

[0096]

[0097] Note: Overall accuracy represents the overall recognition ability; known species accuracy represents the ability to recognize known soil types; unknown species accuracy represents the ability to detect unknown pesticide types, where "-" indicates no corresponding function; time cost is the total time from parameter optimization to recognition.

[0098] As shown in the table, the detection method using the in-situ soil organic contamination detection device while drilling (WDT) achieved the highest recognition rate of 95.26% for soil pesticide detection on the three datasets. The main reason for the extremely low overall recognition rates of the other classifiers is that the ANN, SVM, RF, and NB classifiers lack the ability to detect unknown classes. The larger number of unknown class samples significantly reduces their overall recognition rates. If we ignore the detection of unknown classes and only consider the recognition of known pesticides, the ANN, SVM, and RF classifiers outperform the WDT method. This is primarily due to their simplified operation, which increases recognition speed. However, for known pesticides, the invented method still achieved a recognition rate of over 80%, and outperformed NB for datasets A and B. Compared to other classifiers that cannot detect unknown classes, the WDT method exhibits excellent performance. The WDT method achieved recognition rates of 98.75%, 89.38%, and 95% for unknown pesticides on the three datasets, respectively. On the other hand, its operation time is much better than other classifiers. The fastest time is 0.01 seconds, and the slowest time is only 0.02 seconds. The fastest operation time of other classifiers takes 0.15 seconds.

Claims

1. A device for detecting organic pollution in soil while drilling, characterized in that: The invention is composed of an air intake pipe (A), a signal acquisition circuit board I (B), a middle section (C), a flow equalizing plate (D), a signal acquisition circuit board II (E) and an air outlet pipe (F), wherein the air intake pipe (A) is composed of an air inlet (1), a transition section I (2) and a flow guide section I (3), and the air inlet (1), the transition section I (2) and the flow guide section I (3) are arranged in sequence from front to back and are smoothly connected; the signal acquisition circuit board I (B) is composed of a substrate I (4), a sensor group I (6) and a resistor group I (7), and the substrate I (4) is disc-shaped, and the flow guide group I (7) is evenly distributed on its circumference. Ⅰ (5); a through hole Ⅰ (8) is provided near the lower end of the substrate Ⅰ (4); the four sensors of the sensor group Ⅰ (6) and the four resistors of the resistor group Ⅰ (7) are evenly spaced and fixed to the front of the substrate Ⅰ (4); the middle section (C) is composed of a circular tube (10), a protrusion pair Ⅰ (11) and a protrusion pair Ⅱ (13), the front end of the circular tube (10) is provided with a boss Ⅰ (9), and the rear end is provided with a boss Ⅱ (12); a through hole Ⅱ (14) is provided at the lower end of the inner circle of the circular tube (10); the two protrusions of the protrusion pair Ⅰ (11) and the two protrusions of the protrusion pair Ⅱ (13) are arranged at the same position. The same circular cross section is formed in the middle of the circular tube (10), and the two protrusions of the protrusion pair I (11) are symmetrically fixed to the left and right sides of the center line aa of the upper end of the circular tube (10); the two protrusions of the protrusion pair II (13) are symmetrically fixed to the left and right ends of the circular tube (10); the flow balancing plate (D) is composed of a gas collecting section (15), a drainage section (16), and a drainage plate group (17), the gas collecting section (15) is annular, and a wiring port I (18) is provided at the upper end of the outer ring of the gas collecting section (15); the drainage section (16) is a hollow truncated cone; the 11 drainage plates of the drainage plate group (17) have the same structure, The signal acquisition circuit board II (E) is composed of a substrate II (20), a sensor group II (22), a resistor group II (24), a power socket (26) and a data acquisition and transmission module (27). The substrate II (20) is in the shape of a disk, and 10 guide ports of the guide port group II (21) are evenly distributed on its circumference. A wiring port II (23) is provided near the upper end of the substrate II (20). (20) is provided with a through hole III (25); the four sensors of the sensor group II (22) and the four resistors of the resistor group II (24) are evenly spaced and fixed to the front of the substrate II (20); the power socket (26) and the data acquisition and transmission module (27) are fixed to the back of the substrate II (20); the outlet pipe (F) is composed of a guide section II (28), a transition section II (30) and an outlet II (32); the guide section II (28), the transition section II (30) and the outlet II (32) are arranged in sequence from front to back and are smoothly connected; the same circular cross-section in the middle of the guide section II (28) is symmetrically provided on the left and right sides of the center line aa of the upper end Two protrusions of the protrusion pair III (29); two protrusions of the protrusion pair IV (33) are symmetrically provided at the left and right ends of the circular cross section; a through hole IV (34) is provided near the lower end of the center line aa of the circular cross section; an outlet (31) is provided at the middle position of the upper part of the transition section II (30); the air inlet pipe (A), the middle section (C) and the air outlet pipe (F) are arranged in sequence from front to back and fixed; the signal acquisition circuit board I (B) is fixed in front of the through hole II (14) of the middle section (C); the flow equalizing plate (D) is fixed behind the through hole II (14) of the middle section (C); the signal acquisition circuit board II (E) is fixed in front of the through hole IV (34) of the guide section II (28) of the air outlet pipe (F).

2. The in-situ soil organic pollution detection device while drilling according to claim 1, characterized in that: The thickness d1 of the air inlet (1), transition section I (2), guide section I (3) in the air inlet pipe (A), and the guide section II (28), transition section II (30) and air outlet II (32) in the air outlet pipe (F) are all 1-1.5 mm; the inner diameter D1 of the air inlet (1) and the air outlet II (32) are both 3.2-4.5 mm, and the length L1 are both 5-7 mm; the inner diameter D2 of the guide section I (3) and the guide section II (28) are both 35-40 mm, and the length L3 are both 12-16 mm; the length L2 of the transition section I (2) and the transition section II (30) are both 8-10 mm; the diameter D3 of the substrate I (4) in the signal acquisition circuit board I (B) and the substrate II (20) in the signal acquisition circuit board II (E) are both 35-40, and the thickness L4 are both 1-1.3mm; the inner diameter D4 of the circular tube (10) in the middle section (C) is 35-40mm, the length L5 is 15.9-19.4mm, and the thickness d2 is 1-1.5mm; the outer diameter D5 of the gas collecting section (15) in the flow equalizing plate (D) is 35-40mm, the width L6 is 2.8-3.2mm, and the thickness d3 is 0.6-0.8mm; the front end outer diameter and thickness of the drainage section (16) are the same as the outer diameter and thickness of the gas collecting section (15), and the height L7 is 3.8-4.2mm; the length L9 of each drainage plate in the drainage plate group (17) is 10-11mm, the thickness L10 is 0.9-1mm, and the inner end height L8 is 3.3-3.5mm; the diameter D6 of the air outlet (19) is 3.1-3.3mm.

3. A detection method based on the in-situ soil organic pollution detection device while drilling according to claim 1, characterized in that: The following steps are involved: 1) Establishing a soil pollution odor signal feature rule library, using the in-situ soil organic pollution detection device while drilling as described in claim 1 to collect odor sample data of common / untreated different types of pollution in the laboratory, and performing feature extraction on the odor sample data. The feature extractor uses wavelet transform, and the extraction formula is: Then use the single-class support vector machine to learn the feature rules. The formula is: st(ω·K(x i ))≥r-ξ i ,x i ≥0 Where: K(x i ) represents the kernel function, ω is a weight vector, and its dimension is the same as the feature space K(x i ) are the same, ξ i is a non-negative relaxation factor, n is the number of samples; p∈(0,1) is a penalty term used to control the balance between the maximum distance from the origin and the number of support vectors in the sample; the important parameters that determine the boundary are ω, ρ, which need to be optimized; the Hinge loss function is used instead of ξ i , we get the unconstrained objective function; as shown in the formula: So that h(x) = ω T K(x i )-ρ, the decision function is expressed as formula: Where: h(x) is the characteristic rule learned from different types of odor samples; the characteristic rules of different types of pollution odor samples are stored in the rule base, which is located at the signal processing end; 2) In-situ soil organic pollution detection application, the in-situ soil organic pollution detection device while drilling according to claim 1 - in short: the detection device (39) is integrated and fixed in the MIP probe (35), and used in conjunction with the guide tube (36), the separation chamber (37) and the separation membrane (38), wherein the separation chamber (37) is separated from the outside by the separation membrane (38), the guide tube (36) is located at the front end of the separation chamber (37), and the detection device (39) is located at the upper end of the separation chamber (37); the MIP probe (35) enters the soil and heats the drill bit, and the hot volatile gas of the soil is transported to the detection device (39) through the separation membrane (38) and the carrier gas in the guide tube (36) to collect the hot volatile gas information of the soil organic pollution to the signal processing end; 3) The signal processing end extracts features and establishes rules for the collected soil pollution odor sample data, using the same establishment method as step 1); then the collected soil odor sample rules are compared with the existing rules in the rule library in the logic gate. If the signal matches the rule successfully, the output signal is 1, indicating that the type of soil organic pollution has been detected; if the signal matches the rule successfully, the output signal is 0, indicating that the type of soil organic pollution is unknown and further testing using standard methods is required to determine the type of pollution. The rules are then expanded to the rule library, and the rule library can be continuously expanded.

Images