Vegetable pesticide content detection process

Abstract

The invention relates to a vegetable pesticide content detection process, which belongs to the technical field of vegetable pesticide detection and comprises a multi-source data acquisition module, a risk pre-judgment module, a sample preprocessing module, a rapid screening module, a high-precision detection and multi-pesticide identification module and a report generation module. The multi-source data acquisition module comprises a planting data acquisition unit, an Internet of Things environment data acquisition unit, a meteorological data acquisition unit and an unstructured data acquisition unit. According to the method, the risk probability is pre-judged through multi-source data modeling, so that accurate detection can be triggered only for high-risk samples, resource waste caused by high-cost analysis of all the samples is avoided, the detection period can be effectively shortened, and the detection accuracy is improved by combining the enzyme inhibition rate method for rapid preliminary screening and the near infrared spectrum-chromatography-mass spectrum for accurate confirmation. The timeliness of rapid screening is reserved, and the defect that a traditional single spectrum technology cannot identify specific pesticide types is overcome, so that the detection accuracy can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of vegetable pesticide detection, and particularly to a detection process for the pesticide content in vegetables. Background Art

[0002] In order to avoid the influence of pests and diseases on vegetables during the growth process, pesticides are generally sprayed at specific stages. Although this can ensure the growth of vegetables, excessive pesticide residues may cause acute poisoning or chronic diseases such as cancer and nervous system damage, especially posing significant hazards to sensitive populations such as children and pregnant women. Through detection, products exceeding the standard can be intercepted from entering the market, reducing health risks.

[0003] The traditional methods for detecting pesticide residues in vegetables have the following problems: 1. Relying on single detection data (such as laboratory chromatographic analysis) and not combining dynamic information such as planting environment and pesticide application records, resulting in insufficient risk prediction; 2. The rapid screening method (such as the enzyme inhibition method) has low sensitivity and cannot identify specific pesticide types, while the chromatographic method is accurate but time-consuming, expensive in equipment, and has low efficiency and high cost; 3. The processing method such as the Soxhlet extraction method is time-consuming and the purification steps are cumbersome, affecting the detection efficiency. To further shorten the detection cycle, improve the detection efficiency, and reduce the detection cost while ensuring the detection efficiency, a detection process for the pesticide content in vegetables is proposed. By modeling with multi-source data to predict the risk probability, precise detection can be triggered only for high-risk samples, avoiding the waste of resources in the high-cost analysis of all samples required by traditional chromatographic-mass spectrometry techniques such as GC-MS and LC-MS / MS. Moreover, the detection cycle can be effectively shortened, integrating multi-dimensional information such as planting environment, pesticide application records, and meteorological data, dynamically predicting the risk probability through the Stacking ensemble learning model. Compared with single-spectrum or chemical detection methods, it can detect hidden pesticides (such as systemic agents) in detail, thereby effectively reducing the false alarm rate. By conducting phased detection, the usage frequency of detection equipment can be reduced, thus effectively reducing the detection cost. Summary of the Invention

[0004] The present invention provides a detection process for the pesticide content in vegetables, which solves the problems raised in the above-mentioned background art. By modeling with multi-source data to predict the risk probability, precise detection only for high-risk samples can effectively avoid unnecessary detection work, thereby shortening the detection cycle. It can identify various pesticide types, thus effectively improving the detection accuracy. By fusing multi-source data to enhance the prediction ability, the false alarm rate can be effectively reduced.

[0005] The solution of the present invention to solve the above technical problems is as follows: a vegetable pesticide content detection process, including a multi-source data acquisition module, a risk prediction module, a sample preprocessing module, a rapid screening module, a high-precision detection and multi-pesticide identification module, and a report generation module, wherein the multi-source data acquisition module includes a planting data acquisition unit, an Internet of Things environment data acquisition unit, a meteorological data acquisition unit, and an unstructured data acquisition unit;

[0006] The risk prediction module includes a risk prediction model and a probability output unit;

[0007] The sample pre-processing module includes a chopping device and an ultrasonic extractor;

[0008] The rapid screening module includes a spectrophotometer;

[0009] The high-precision detection and multi-pesticide identification module includes a near-infrared spectrometer, a gas chromatography-mass spectrometer, and a liquid chromatography-tandem mass spectrometer;

[0010] The detection process includes the following steps:

[0011] S1: Data collection: Collect vegetable types, growth environment (temperature and humidity, soil pH), pesticide application records (pesticide type, dosage, time) and meteorological data to build a pesticide residue risk prediction model. The risk prediction model uses the Stacking ensemble learning framework

[0012] S2: Risk prediction: The preliminary risk probability is calculated through the risk prediction model based on the collected data. The model output value P∈[0,1] indicates the risk probability of excessive pesticide residues in the current sample. If it is higher than the preset threshold of 30%, the next step of precise detection is triggered;

[0013] S3: Sample pre-processing module: The collected samples are shredded by a shredder, and the shredded size is 1 cm 3 , after being chopped, ultrasonic extraction is performed by an ultrasonic extractor to remove the interference of pigments and lipids;

[0014] S4: Rapid screening by enzyme inhibition rate method: react the extract with acetylcholinesterase and colorimetric agent (5,5'-dithiodinitrobenzoic acid), and measure the inhibition rate by spectrophotometer. If the inhibition rate is ≥50%, it is judged as a positive sample and enters the precise detection stage;

[0015] S5: High-precision detection: 1. Near-infrared spectroscopy preliminary screening: Scanning the positive samples by near-infrared spectroscopy and comparing with the database to lock the suspected pesticide categories; 2. Gas chromatography-mass spectrometry (GC-MS) screening: Used for detecting volatile pesticides (such as organophosphorus), the chromatographic column is DB-5MS, and the programmed temperature is from 50°C to 280°C; 3. Liquid chromatography-tandem mass spectrometry (LC-MS / MS) screening: For thermally unstable pesticides (such as carbamates), the mobile phase is acetonitrile-0.1% formic acid water, and the electrospray ionization source is in the positive ion mode;

[0016] S6: Report generation: Integrating the risk probability, rapid screening results and accurate detection data to generate a comprehensive report including the types of pesticides, content and risk levels.

[0017] Based on the above technical solutions, the present invention can also be improved as follows.

[0018] Further, the planting data unit collects data on vegetable types, growth cycles and pesticide application records, the Internet of Things environmental data collection unit collects data on soil pH value, heavy metal content, air temperature and humidity, and light intensity, the meteorological data collection unit collects data on rainfall, wind speed and extreme weather events, and the unstructured data collection unit collects farmer operation logs and farmland monitoring image data.

[0019] Further, the risk prediction model sets a dynamic threshold setting mechanism, the initial value of the basic threshold is set to 30%, determined based on the historical data ROC curve (corresponding to a recall rate of 85% and a false alarm rate <15%), and is adaptively adjusted: When the detection results of 10 consecutive batches of samples have an error with the model prediction >15%, the threshold is updated, and the new threshold is:

[0020]

[0021] Further, in the step S1, the Stacking ensemble learning framework of the risk prediction model combines XGBoost to handle the high-dimensional non-linear relationship of structured data, combines XGBoost to handle the high-dimensional non-linear relationship of structured data, combines the time series LSTM to analyze the pesticide diffusion trend when environmental data is missing, and uses logistic regression (LogisticRegression) to perform weighted fusion on the output results of the base models. The specific formula is:

[0022]

[0023] where σ is the Sigmoid function, M i is the output of the base model, and W i is the weight coefficient;

[0024] The model is trained using a historical detection database containing more than 100,000 samples, covering positive and negative samples of 50 kinds of vegetables and 200 kinds of pesticides.

[0025] Use Focal Loss to solve the problem of class imbalance:

[0026] Loss=-α(1 - p t ) γ log(p t )

[0027] where α = 0.25, γ = 2, and Pt is the predicted probability.

[0028] Furthermore, in the step S3, different parts of the vegetable such as the leaf tip and epidermis are randomly selected before chopping the sample, avoiding rinsing, and wiping off the surface impurities.

[0029] Furthermore, in the step S3, ultrasonic-assisted extraction uses a methanol-water (4:1) mixed solvent, ultrasonically extracts for 10 minutes at 40°C, and combines with solid-phase extraction column (C18 packing) for purification.

[0030] Furthermore, in the step S5, the database covers characteristic peaks of 300 kinds of pesticides such as organophosphorus and carbamate.

[0031] Furthermore, in the step S6, the report generation module realizes traceability of detection data through blockchain technology, and conducts blind sample tests and instrument calibrations regularly to ensure the reliability of the results.

[0032] The beneficial effects of the present invention are as follows: The present invention provides a detection process for the pesticide content in vegetables, having the following advantages:

[0033] 1. By modeling with multi-source data to predict the risk probability, it is possible to trigger precise detection only for high-risk samples, avoiding the waste of resources in the traditional chromatograph-mass spectrometry techniques such as GC-MS and LC-MS / MS that require high-cost analysis of all samples, and effectively shortening the detection cycle;

[0034] 2. Combining the enzyme inhibition rate method for rapid preliminary screening and the near-infrared spectroscopy-chromatograph mass spectrometry combination for accurate confirmation not only retains the timeliness of rapid screening but also solves the defect that the traditional single spectroscopy technique cannot identify specific pesticide types, thereby effectively improving the detection accuracy;

[0035] 3. Integrating multi-dimensional information such as the planting environment, pesticide application records, and meteorological data, and dynamically predicting the risk probability through the Stacking ensemble learning model, compared with the single spectroscopy or chemical detection method, it can conduct detailed detection on concealed pesticides (such as systemic agents), thereby effectively reducing the false alarm rate;

[0036] 4. By conducting phased detection, the frequency of using detection equipment can be reduced, thereby effectively reducing the detection cost.

[0037] The above description is only an overview of the technical solution of the present invention. In order to understand the technical means of the present invention more clearly and implement it in accordance with the content of the specification, the following takes the preferred embodiments of the present invention and combines with the accompanying drawings to elaborate in detail as follows. The specific implementation manners of the present invention are given in detail by the following embodiments and their accompanying drawings. Description of the Drawings

[0038] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation to the present invention. In the drawings:

[0039] Figure 1 It is a process flow chart of a vegetable pesticide content detection process provided by an embodiment of the present invention;

[0040] Figure 2 It is an architecture diagram of a detection model in a vegetable pesticide content detection process provided by an embodiment of the present invention. Detailed Description of the Preferred Embodiments

[0041] The following combines with the attached Figure 1-2 The principles and features of the present invention are described below. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention. In the following paragraphs, the present invention is described more specifically by way of example with reference to the accompanying drawings. According to the following description and claims, the advantages and features of the present invention will be clearer. It should be noted that the drawings are all in a very simplified form and use non-precise scales, only for the purpose of facilitating and clearly assisting in explaining the objectives of the embodiments of the present invention.

[0042] It should be noted that when a component is referred to as being "fixed to" another component, it can be directly on the other component or there may also be an intermediate component. When a component is considered to be "connected to" another component, it can be directly connected to the other component or there may be an intermediate component at the same time. When a component is considered to be "disposed on" another component, it can be directly disposed on the other component or there may be an intermediate component at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are only for the purpose of illustration.

[0043] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs. The terms used in the specification of the present invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0044] like Figure 1-2 As shown, the present invention provides a vegetable pesticide content detection process, including a multi-source data acquisition module, a risk prediction module, a sample preprocessing module, a rapid screening module, a high-precision detection and multi-pesticide identification module, and a report generation module, characterized in that: the multi-source data acquisition module includes a planting data acquisition unit, an Internet of Things environmental data acquisition unit, a meteorological data acquisition unit, and an unstructured data acquisition unit, the planting data unit collects vegetable types, growth cycles, and pesticide application record data, the Internet of Things environmental data acquisition unit collects soil pH value, heavy metal content, air temperature and humidity, and light intensity data, the meteorological data acquisition unit collects rainfall, wind speed, and extreme weather event data, and the unstructured data acquisition unit collects farmer operation logs and farmland monitoring image data;

[0045] The risk prediction module includes a risk prediction model and a probability output unit. The risk prediction model sets a dynamic threshold setting mechanism. The initial value of the basic threshold is set to 30%, which is determined based on the historical data ROC curve (corresponding to a recall rate of 85% and a false alarm rate of <15%). Adaptive adjustment: When the error between the detection results of 10 consecutive batches of samples and the model prediction is >15%, the threshold update is triggered. The new threshold is:

[0046]

[0047] The sample pretreatment module includes a chopping device and an ultrasonic extractor;

[0048] The rapid screening module includes a spectrophotometer;

[0049] The high-precision detection and multi-pesticide identification module includes a near-infrared spectrometer, a gas chromatography-mass spectrometer, and a liquid chromatography-tandem mass spectrometer.

[0050] The specific working principle and use method of the present invention are:

[0051] S1: Data collection: Collect vegetable types, growth environment (temperature and humidity, soil pH), pesticide application records (pesticide type, dosage, time) and meteorological data to build a pesticide residue risk prediction model. The risk prediction model uses the Stacking ensemble learning framework. The Stacking ensemble learning framework of the risk prediction model combines XGBoost to process the high-dimensional nonlinear relationship of structured data, combines XGBoost to process the high-dimensional nonlinear relationship of structured data, and combines time series LSTM to analyze the pesticide diffusion trend when environmental data is missing. Use logistic regression to perform weighted fusion on the output results of the base model. The specific formula is:

[0052]

[0053] where σ is the Sigmoid function, and M i is the output of the base model, and W i is the weight coefficient;

[0054] The model is trained using a historical detection database containing more than 100,000 samples, covering positive and negative samples of 50 kinds of vegetables and 200 kinds of pesticides;

[0055] Use Focal Loss to solve the class imbalance problem:

[0056] Loss = -α(1 - p t ) γ log(p t )

[0057] where α = 0.25, γ = 2, and Pt is the predicted probability;

[0058] S2: Risk prediction: Calculate the preliminary risk probability through the risk prediction model based on the collected data. The model output value P ∈ [0, 1], representing the risk probability that the current sample has excessive pesticide residues. If it is higher than the preset threshold of 30%, the next step of precise detection is triggered;

[0059] S3: Sample preprocessing module: The collected samples are shredded by a shredding device, and the shredding size is 1 cm 3 , and different parts of the vegetables such as leaf tips and epidermis are randomly selected before shredding to avoid washing, wipe off the surface impurities, and after shredding, ultrasonic-assisted extraction is carried out by an ultrasonic extractor. The ultrasonic-assisted extraction uses a methanol-water (4:1) mixed solvent, and ultrasonic extraction is carried out at 40°C for 10 minutes, combined with purification by a solid-phase extraction column (C18 packing) to remove pigment and lipid interference;

[0060] S4: Rapid screening by enzyme inhibition rate method: React the extract with acetylcholinesterase and a chromogenic reagent (5,5'-dithiobis(2-nitrobenzoic acid)), and measure the inhibition rate by a spectrophotometer. If the inhibition rate ≥ 50%, it is determined as a positive sample and enters the precise detection stage;

[0061] S5: High-precision detection: 1. Near-infrared spectroscopy preliminary screening: Near-infrared spectroscopy scanning is carried out on positive samples and compared with the database. The database covers the characteristic peaks of 300 kinds of pesticides such as organophosphorus and carbamate to lock the suspected pesticide categories; 2. Gas chromatography-mass spectrometry (GC-MS) screening: Used for the detection of volatile pesticides (such as organophosphorus), the chromatographic column is DB-5MS, and the programmed temperature is from 50°C to 280°C; 3. Liquid chromatography-tandem mass spectrometry (LC-MS / MS) screening: For thermally unstable pesticides (such as carbamate), the mobile phase is acetonitrile-0.1% formic acid water, and the electrospray ionization source is in the positive ion mode;

[0062] S6: Report generation: Integrate the risk probability, rapid screening results, and precise detection data to generate a comprehensive report containing the types of pesticides, their contents, and risk levels. The report generation module realizes the traceability of detection data through blockchain technology and conducts blind sample tests and instrument calibrations regularly to ensure the reliability of the results.

[0063] It should be noted that in this document, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. The content not described in detail in this specification belongs to the prior art well-known to those skilled in the art.

[0064] As described above, it is only the preferred embodiment of the present invention, and there is no restriction in any form on the present invention; any ordinary technician in the industry can smoothly implement the present invention according to what is shown in the accompanying drawings of the specification and the above description; however, any equivalent changes made by those skilled in the art within the scope of the technical solution of the present invention by using the technical content disclosed above, such as slight modifications, decorations, and evolutions, are all equivalent embodiments of the present invention; at the same time, any equivalent changes, modifications, and evolutions made to the above embodiments based on the essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. A detection process for the pesticide content of vegetables, comprising a multi-source data acquisition module, a risk prediction module, a sample pretreatment module, a rapid screening module, a high-precision detection and multi-pesticide identification module, and a report generation module, characterized in that: The multi-source data acquisition module includes a planting data acquisition unit, an Internet of Things environmental data acquisition unit, a meteorological data acquisition unit, and an unstructured data acquisition unit; The risk prediction module includes a risk prediction model and a probability output unit; The sample pretreatment module includes a chopping device and an ultrasonic extractor; The rapid screening module includes a spectrophotometer; The high-precision detection and multi-pesticide identification module includes a near-infrared spectroscopy analyzer, a gas chromatography-mass spectrometry (GC-MS) instrument, and a liquid chromatography-tandem mass spectrometry (LC-MS / MS) instrument; The detection process includes the following steps: S1: Data acquisition: Collect vegetable types, growth environment (temperature and humidity, soil pH value), pesticide application records (pesticide type, dosage, time), and meteorological data, and construct a pesticide residue risk prediction model. The risk prediction model uses a Stacking ensemble learning framework S2: Risk prediction: Calculate the preliminary risk probability through the risk prediction model based on the collected data. The model output value P ∈ [0, 1], indicating the risk probability that the current sample has excessive pesticide residues. If it is higher than the preset threshold of 30%, the next step of precise detection is triggered; S3: Sample preprocessing module: The collected samples are chopped by a chopping device to a size of 1 cm 3 , and after chopping, ultrasonic-assisted extraction is carried out by an ultrasonic extractor; S4: Rapid screening by enzyme inhibition rate method: React the extract with acetylcholinesterase and a chromogenic agent (5,5'-dithiobis(2-nitrobenzoic acid)), and measure the inhibition rate with a spectrophotometer. If the inhibition rate ≥ 50%, it is determined as a positive sample and enters the precise detection stage; S5: High-precision detection:

1. Preliminary screening by near-infrared spectroscopy: Perform near-infrared spectroscopy scanning on the positive sample and compare it with the database to lock the suspected pesticide category; 2. Gas chromatography-mass spectrometry (GC-MS) screening: Used for the detection of volatile pesticides, the chromatographic column is DB-5MS, and the temperature is programmed from 50°C to 280°C; 3. Liquid chromatography-tandem mass spectrometry (LC-MS / MS) screening: For thermally unstable pesticides, the mobile phase is acetonitrile-0.1% formic acid water, and the electrospray ionization source is in the positive ion mode; S6: Report generation: Integrate the risk probability, rapid screening results, and precise detection data to generate a comprehensive report including pesticide types, contents, and risk levels.

2. The vegetable pesticide content detection process according to claim 1, wherein, The planting data unit collects vegetable types, growth cycles, and pesticide application record data. The Internet of Things environmental data acquisition unit collects soil pH value, heavy metal content, air temperature and humidity, and light intensity data. The meteorological data acquisition unit collects rainfall, wind speed, and extreme weather event data. The unstructured data acquisition unit collects farmer operation logs and farmland monitoring image data.

3. The vegetable pesticide content detection process according to claim 1, characterized in that, The risk prediction model sets a dynamic threshold setting mechanism. The initial value of the basic threshold is set to 30%, which is determined based on the historical data ROC curve (corresponding to a recall rate of 85% and a false alarm rate < 15%). Adaptive adjustment: When the detection results of 10 consecutive batches of samples have an error > 15% compared with the model prediction, the threshold is updated. The new threshold is:

4. The vegetable pesticide content detection process according to claim 1, characterized in that, In the step S1, the Stacking ensemble learning framework of the risk prediction model combines XGBoost to handle the high-dimensional non-linear relationship of structured data, combines XGBoost to handle the high-dimensional non-linear relationship of structured data, combines the time series LSTM to analyze the pesticide diffusion trend when environmental data is missing, and uses Logistic Regression to perform weighted fusion on the output results of the base model. The specific formula is: where σ is the Sigmoid function, and M i is the output of the base model, and W i is the weight coefficient; Use a database to train the model; Use Focal Loss to solve the problem of class imbalance: Loss=-α(1 - p t ) γ log(p t ) where α = 0.25, γ = 2, and Pt is the predicted probability.

5. The vegetable pesticide content detection process according to claim 1, characterized in that, In the step S3, randomly select different parts of the vegetable, such as the leaf tip and epidermis, before chopping the sample, avoid rinsing, and wipe off the surface impurities.

6. The vegetable pesticide content detection process according to claim 1, wherein, In the step S3, ultrasonic-assisted extraction uses a methanol-water (4:1) mixed solvent, ultrasonically extracts for 10 minutes at 40 °C, and combines solid-phase extraction column (C18 packing) purification.

7. The vegetable pesticide content detection process according to claim 1, wherein, In the step S5, the database covers the characteristic peaks of 300 pesticides such as organophosphorus and carbamate.

8. The vegetable pesticide content detection process according to claim 1, characterized in that, In the step S6, the report generation module realizes the traceability of detection data through blockchain technology, and regularly conducts blind sample tests and instrument calibrations.

Images