CN118427577B - Pesticide residue detection method and device for agricultural products - Google Patents
Pesticide residue detection method and device for agricultural products Download PDFInfo
- Publication number
- CN118427577B CN118427577B CN202410889725.4A CN202410889725A CN118427577B CN 118427577 B CN118427577 B CN 118427577B CN 202410889725 A CN202410889725 A CN 202410889725A CN 118427577 B CN118427577 B CN 118427577B
- Authority
- CN
- China
- Prior art keywords
- detection
- online
- analysis
- result
- historical
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 541
- 239000000447 pesticide residue Substances 0.000 title claims abstract description 53
- 238000004458 analytical method Methods 0.000 claims abstract description 172
- 238000005070 sampling Methods 0.000 claims abstract description 67
- 239000000575 pesticide Substances 0.000 claims abstract description 57
- 238000011897 real-time detection Methods 0.000 claims abstract description 41
- 238000000034 method Methods 0.000 claims abstract description 33
- 238000003062 neural network model Methods 0.000 claims abstract description 26
- 230000003213 activating effect Effects 0.000 claims abstract description 21
- 238000012795 verification Methods 0.000 claims description 18
- 230000008859 change Effects 0.000 claims description 14
- 230000001360 synchronised effect Effects 0.000 claims description 14
- 238000012549 training Methods 0.000 claims description 14
- 238000011144 upstream manufacturing Methods 0.000 claims description 12
- 238000000605 extraction Methods 0.000 claims description 11
- 230000003044 adaptive effect Effects 0.000 claims description 8
- 230000004927 fusion Effects 0.000 claims description 8
- 238000006243 chemical reaction Methods 0.000 claims description 7
- 238000010276 construction Methods 0.000 claims description 7
- 238000005457 optimization Methods 0.000 claims description 6
- 238000000513 principal component analysis Methods 0.000 claims description 6
- 238000009826 distribution Methods 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 3
- 238000012544 monitoring process Methods 0.000 abstract description 10
- 230000000694 effects Effects 0.000 abstract description 5
- 230000008569 process Effects 0.000 description 6
- 238000012545 processing Methods 0.000 description 4
- 230000008602 contraction Effects 0.000 description 3
- 230000007613 environmental effect Effects 0.000 description 3
- 230000003993 interaction Effects 0.000 description 3
- 238000001228 spectrum Methods 0.000 description 3
- 238000010183 spectrum analysis Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 239000003905 agrochemical Substances 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000004817 gas chromatography Methods 0.000 description 2
- 238000004811 liquid chromatography Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 239000002728 pyrethroid Substances 0.000 description 2
- 238000003908 quality control method Methods 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- KXDHJXZQYSOELW-UHFFFAOYSA-M Carbamate Chemical compound NC([O-])=O KXDHJXZQYSOELW-UHFFFAOYSA-M 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 239000000152 carbamate pesticide Substances 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 230000001066 destructive effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000003018 immunoassay Methods 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 244000005700 microbiome Species 0.000 description 1
- 238000000491 multivariate analysis Methods 0.000 description 1
- 238000009659 non-destructive testing Methods 0.000 description 1
- 239000003987 organophosphate pesticide Substances 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 238000010223 real-time analysis Methods 0.000 description 1
- 230000002787 reinforcement Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
- Y02A40/10—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in agriculture
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a pesticide residue detection method and a device for agricultural products, and relates to the technical field of online detection, wherein the method comprises the following steps: the method comprises the steps of interacting a target scene, obtaining a detection index set and carrying out online detection feasibility analysis; acquiring a historical pesticide detection record according to a feasibility analysis result; based on historical pesticide detection records and a neural network model, an online analysis channel is constructed, and the online analysis channel comprises a plurality of self-adaptive analysis channels with different complexity; activating an online detection device and a sampling detection device, and synchronously detecting and collecting a target scene by combining a feasibility analysis result to obtain a real-time detection data set, wherein the real-time detection data set comprises continuous detection data and interval detection data; based on the online analysis channel, analyzing the real-time detection data set, obtaining interval detection results and continuous detection results, fitting the results, and obtaining residual detection results. Thereby achieving the technical effects of optimizing the monitoring cost and improving the monitoring comprehensiveness.
Description
Technical Field
The invention relates to the technical field of online detection, in particular to a pesticide residue detection method and device for agricultural products.
Background
The detection of pesticide residues in agricultural products is an important link in food safety control, and the pesticide residues not only affect the quality of the agricultural products, but also form a potential threat to the health of consumers. The existing agricultural product pesticide residue detection comprises destructive sampling detection based on methods such as Gas Chromatography (GC), liquid Chromatography (LC), immunoassay and the like, and has high detection precision and sensitivity, but the equipment is large in size, high in cost and complex in operation. The method also comprises a nondestructive testing method based on spectrum analysis, and has high monitoring efficiency, but low application broad spectrum and large influence by environmental factors. The existing pesticide residue detection method has the technical problems of high detection cost and incomplete detection.
Disclosure of Invention
The invention provides a pesticide residue detection method and device for agricultural products, which are used for solving the technical problems of high detection cost and incomplete detection in the prior art, realizing the technical effects of optimizing the monitoring cost and improving the comprehensiveness of monitoring.
In a first aspect, the present invention provides a pesticide residue detection method for agricultural products, wherein the method comprises:
The method comprises the steps of interacting a target scene, obtaining a detection index set, and carrying out online detection feasibility analysis on the detection index set; acquiring a historical pesticide detection record according to a feasibility analysis result, wherein the historical pesticide detection record comprises a historical online detection record and a historical sampling detection record; based on the historical pesticide detection record, combining a neural network model to construct an online analysis channel, wherein the online analysis channel comprises a plurality of self-adaptive analysis channels with different complexity; activating an online detection device and a sampling detection device, and carrying out synchronous detection acquisition of the target scene by combining a feasibility analysis result to obtain a real-time detection data set, wherein the real-time detection data set comprises continuous detection data and interval detection data; and analyzing the real-time detection data set based on the online analysis channel, obtaining an interval detection result and a continuous detection result, fitting the interval detection result and the continuous detection result, and obtaining a residual detection result.
In a second aspect, the present invention also provides a pesticide residue detection device for agricultural products, wherein the device comprises:
the index acquisition and analysis module is used for acquiring a detection index set according to the interaction target scene and carrying out online detection feasibility analysis on the detection index set.
The detection record acquisition module is used for acquiring historical pesticide detection records according to feasibility analysis results, wherein the historical pesticide detection records comprise historical online detection records and historical sampling detection records.
The analysis channel construction module is used for constructing an online analysis channel based on the historical pesticide detection record and combining a neural network model, wherein the online analysis channel comprises a plurality of self-adaptive analysis channels with different complexity.
The synchronous detection acquisition module is used for activating the online detection equipment and the sampling detection equipment, and carrying out synchronous detection acquisition of the target scene by combining a feasibility analysis result to acquire a real-time detection data set, wherein the real-time detection data set comprises continuous detection data and interval detection data.
And the analysis output module is used for analyzing the real-time detection data set based on the online analysis channel, obtaining an interval detection result and a continuous detection result, fitting the interval detection result and the continuous detection result and obtaining a residual detection result.
The invention discloses a pesticide residue detection method and device for agricultural products, comprising the following steps: the method comprises the steps of interacting a target scene, obtaining a detection index set, and carrying out online detection feasibility analysis on the detection index set; according to the feasibility analysis result, acquiring a historical pesticide detection record, wherein the historical pesticide detection record comprises an online detection record and a sampling detection record; based on historical pesticide detection records, an online analysis channel is built by combining a neural network model, and the online analysis channel comprises a plurality of self-adaptive analysis channels with different complexity; activating an online detection device and a sampling detection device, and synchronously detecting and acquiring a target scene by combining a feasibility analysis result to acquire a real-time detection data set, wherein the real-time detection data set comprises continuous detection data and interval detection data; based on the online analysis channel, analyzing the real-time detection data set, obtaining an interval detection result and a continuous detection result, fitting the interval detection result and the continuous detection result, and obtaining a residual detection result. The pesticide residue detection method and device for agricultural products solve the technical problems of high detection cost and incomplete detection, and achieve the technical effects of optimizing the monitoring cost and improving the comprehensiveness of monitoring.
Drawings
FIG. 1 is a schematic flow chart of a method for detecting pesticide residues for agricultural products according to the present invention;
fig. 2 is a schematic structural view of the agricultural chemical residue detecting apparatus for agricultural products according to the present invention.
Reference numerals illustrate: the system comprises an index acquisition and analysis module 11, a detection record acquisition module 12, an analysis channel construction module 13, a synchronous detection acquisition module 14 and an analysis output module 15.
Detailed Description
The technical scheme provided by the embodiment of the invention aims to solve the technical problems of high detection cost and incomplete detection in the prior art, and adopts the following overall thought:
Firstly, interacting with a target scene to obtain a series of detection indexes, and carrying out feasibility analysis of online detection on the indexes. Based on the results of the feasibility analysis, relevant historical pesticide detection records are collected, and the records comprise data of online detection and sampling detection.
And then, constructing an online analysis channel by utilizing the historical pesticide detection records and combining a neural network model, wherein the channel comprises a plurality of self-adaptive analysis channels with different complexity so as to adapt to different detection requirements.
And then, activating the online detection equipment and the sampling detection equipment, and synchronously detecting and collecting data of the target scene according to the result of the feasibility analysis, thereby obtaining a real-time detection data set. This data set includes continuous detection data and interval detection data.
And finally, analyzing the real-time detection data set through an online analysis channel to obtain an interval detection result and a continuous detection result. Fitting the two detection results to obtain a more accurate pesticide residue detection result.
The foregoing aspects will be better understood by reference to the following detailed description of the invention taken in conjunction with the accompanying drawings and detailed description. It should be apparent that the described embodiments are only some embodiments of the present invention and not all embodiments of the present invention, and it should be understood that the present invention is not limited by the exemplary embodiments used only to explain the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to fall within the scope of the invention. It should be noted that, for convenience of description, only some, but not all of the drawings related to the present invention are shown.
Example 1
Fig. 1 is a schematic flow chart of a method for detecting pesticide residues for agricultural products according to the present invention, wherein the method comprises:
And (3) interacting the target scene, acquiring a detection index set, and carrying out online detection feasibility analysis on the detection index set.
Optionally, first, a target scene is defined, where the target scene refers to an agricultural product quality control scene or link that needs to be detected by pesticide residues, and the agricultural product quality control scene includes an agricultural product type, a detection environment, a detection device and a detection standard. The types of agricultural products to be detected, detection targets and the like corresponding to different scenes are different, and the target scene can be a quality detection link of a certain agricultural product processing production line or a quality detection workshop of a certain agricultural product storage unit and the like.
Alternatively, by interacting with the target scene, a set of detection indicators may be obtained, the set of detection indicators characterizing the detection target of the target scene. Specifically, the detection index set comprises the types of pesticides to be detected and the control limit of the residual level, and common pesticide types comprise organophosphorus pesticides, carbamate pesticides, pyrethroid pesticides and the like.
Exemplary, the target scene is interacted through a man-machine interaction interface (such as a touch screen, a keyboard, a mouse and the like), and target scene information is obtained; then, extracting related detection indexes such as pesticide residue, chemical components, microorganism indexes and the like from the target scene information, and storing the detection indexes as a detection index set; and finally, carrying out online detection feasibility analysis on the detection index set by using methods such as statistical analysis, machine learning and the like so as to evaluate whether the detection index set is suitable for online detection. Through the process, the accuracy and the practicability of the detection method are guaranteed, and a basis is provided for subsequent pesticide residue detection.
In some embodiments, interacting with a target scene, obtaining a detection index set, and performing online detection feasibility analysis on the detection index set, including:
And interacting the target scene, extracting a plurality of residual detection indexes based on a pesticide residue detection task book of the target scene, and storing the residual detection indexes as the detection index set.
Traversing the detection index set, acquiring a plurality of multi-dimensional discrimination feature sets corresponding to a plurality of residual detection indexes based on big data, and carrying out online detection feasibility analysis based on the multi-dimensional discrimination feature sets, wherein the multi-dimensional discrimination feature sets comprise equipment adaptability, accuracy accessibility and detection instantaneity.
Optionally, the pesticide residue detection task book includes key information of pesticide residue detection such as pesticide type, detection product, detection standard and requirement that need to detect, and based on the detection task book, a plurality of residual detection indexes are extracted, and exemplary includes: pesticide type (such as organophosphorus pesticide, carbamate pesticide, pyrethroid pesticide, etc.), maximum allowable concentration (MRL), etc.
Further, each detection index in the detection index sets is analyzed, and a multi-dimensional discrimination feature set corresponding to a plurality of residual detection indexes is obtained based on a big data technology. Specifically, the multi-dimensional discriminating characteristic set includes: the device adaptability is used for judging whether the detection device is suitable for the extracted detection index, including detection items, detection range and stability of the device; accuracy accessibility, is used for judging whether the detection equipment can reach the required accuracy requirement, including detection accuracy, detection Limit (LOD), error range, repeatability and consistency; the real-time detection is used for judging whether the detection process can meet the requirements of real-time detection, and comprises detection speed, data processing capacity and response time.
Optionally, if the multi-dimensional distinguishing feature set meets the requirement of the pesticide residue detection task book, the residue detection index corresponding to the multi-dimensional distinguishing feature set is considered to be available for online detection in the target scene, and the residue detection index corresponding to the multi-dimensional distinguishing feature set is marked as a feasible index.
Through the steps, the reliability and the effectiveness of the detection process are ensured by comprehensively evaluating the applicability, the precision and the instantaneity of the detection equipment corresponding to the residual detection indexes in the on-line detection of the pesticide residues. The acquired feasibility analysis result is favorable for selecting a proper residual detection index to carry out subsequent online detection, so that the detection efficiency and accuracy are improved.
And acquiring a historical pesticide detection record according to the feasibility analysis result, wherein the historical pesticide detection record comprises a historical online detection record and a historical sampling detection record.
Optionally, the feasibility analysis result includes a plurality of residual detection indicators marked as feasible indicators, which are considered to be available for subsequent online detection.
Specifically, the online detection record refers to a pesticide residue detection record obtained by an online detection system in the history detection record, and the detection data in the online detection record is continuous data, in other words, the online detection record has a higher sampling frequency, so that the change trend of the pesticide residue concentration can be better reflected. The sampling detection record refers to pesticide residue detection data obtained through periodic or aperiodic sampling detection, the sampling detection record is discrete data, and the detection frequency is relatively low, but the detection accuracy and the confidence are higher.
Illustratively, relevant online detection records are extracted from an online detection system database, ensuring data integrity and accuracy. The online detection record contains information such as detection time, detection place, pesticide type, detection concentration, detection equipment, equipment state and calibration condition.
The relevant sample test records are illustratively extracted from a database of a test laboratory or test facility to ensure accuracy and reliability of the data. The sampling detection record contains information such as detection time, sampling place, pesticide type, detection concentration, sample number, detection method and equipment, laboratory report, analysis result and the like.
In some embodiments, obtaining a historical pesticide detection record based on the feasibility analysis results comprises:
And extracting feasible detection indexes based on the feasibility analysis result to perform principal component analysis, and obtaining a key online detection index set.
And accessing a pesticide detection database by taking the key online detection index set as index constraint, and calling a historical online detection record and a historical sampling detection record.
Extracting the original detection data of the historical online detection record and the detection result data of the historical sampling detection record, and storing the detection result data into the historical pesticide detection record.
Alternatively, a Principal Component Analysis (PCA) or other multivariate analysis method is used to extract a plurality of key detection indicators from the on-line detection data and output as a set of key on-line detection indicators. On one hand, the online detection index which can most reflect the pesticide residue level can be identified through PCA and other methods, on the other hand, the online detection method is beneficial to improving the subsequent detection efficiency and accuracy, on the other hand, the historical online detection record and the historical sampling detection record which are called are reduced, and the online analysis channel construction efficiency is improved.
Further, the historical online detection record and the sampling detection record in the pesticide detection database are accessed by using the extracted key online detection index set as an index. And extracting original detection data from the historical online detection record, extracting detection result data from the historical sampling detection record, and storing the detection result data into the historical pesticide detection record. The original detection data are samples of input data of the online detection analysis, and the detection result data are samples of output results of the online detection analysis.
For example, for online detection based on spectral analysis, the raw detection data is raw spectral signal data collected by a spectral analysis device. These data, including raw readings of the spectrum and measured parameters, are typically stored in a multi-dimensional form. Specifically, the method includes spectrum intensity, excitation wavelength range, emission wavelength range, time stamp, environmental conditions (environmental parameters such as temperature and humidity during detection, etc.), equipment status, calibration information, etc. The detection result data is an analysis result output after the original spectrum data is processed. The pesticide residue detection information which is shown as easy to understand specifically includes: pesticide residue concentration, the concentration value of a particular pesticide in a sample, typically in mg/kg or ppm (parts per million); detection precision, accuracy and precision of detection results are generally represented by confidence intervals or error ranges; limit of detection (LOD), the lowest detectable pesticide residue concentration; sample information, description of the sample being tested, such as sample type, source, etc.; the time stamp of the detection result, the time and date of the analysis result generation, and the like.
By extracting and analyzing the key on-line detection index set, the historical on-line detection record and the sampling detection record can be effectively utilized, and the comprehensive and accurate historical pesticide detection record can be constructed. The construction and training of the subsequent online analysis channels provides reliable data support.
And constructing an online analysis channel based on the historical pesticide detection record and combining a neural network model, wherein the online analysis channel comprises a plurality of self-adaptive analysis channels with different complexity.
Optionally, an online analysis channel is constructed according to the neural network model. The online analysis channel comprises a plurality of adaptive analysis channels, each channel having a different complexity and computational resource requirement. The low-complexity channel has the rapid and real-time analysis capability through the feature extraction of a larger scale, and is suitable for primary screening and real-time monitoring. The medium complexity channel is suitable for routine detection by balancing computational resources and accuracy. The high-complexity channel is configured with the feature extraction parameters with smaller scale, so that high-precision analysis is realized, and the method is suitable for detailed detection and result verification.
In some embodiments, an online analysis channel is constructed based on the historical pesticide detection record in combination with a neural network model, wherein the online analysis channel comprises a plurality of adaptive analysis channels of different complexity, including:
Defining a multi-scale structure, configuring corresponding self-adaptive analysis channels based on the multi-scale structure, and generating a plurality of self-adaptive analysis channels.
And constructing an upstream network layer of the neural network model, wherein the upstream network layer comprises an input layer and a low-level feature extraction layer.
And constructing a downstream network layer of the neural network model, wherein the downstream network layer comprises a cascade fusion layer, a dimension conversion layer, a full connection layer and an output layer.
And the upstream network layer, the plurality of self-adaptive analysis channels and the downstream network layer are sequentially connected, the supervision training of the neural network model is performed based on the historical pesticide detection record, and a training result is output as the online analysis channel.
Optionally, a multi-scale structure is defined based on the Inception structure, and then a corresponding adaptive analysis channel is configured based on the multi-scale structure, where the adaptive analysis channel includes, for example, 1 max pooling layer and multiple parallel convolution layers, where the multiple parallel convolution layers are configured with convolution kernels of different sizes, so as to extract multi-scale depth features.
Optionally, the upstream network layer refers to a preamble network layer of the adaptive analysis channel, and includes an input layer for receiving the preprocessed raw detection data, and a low-level feature extraction layer responsible for extracting basic features, where the low-level feature extraction layer performs low-level feature extraction by using 64 convolution kernels with a size of 3×3, as an example.
Optionally, the downstream network layer refers to a subsequent network layer of the adaptive analysis channel, and comprises a cascade fusion layer, a dimension conversion layer, a full-connection layer and an output layer, wherein the cascade fusion layer carries out cascade fusion on the characteristics from different analysis channels so as to comprehensively consider the characteristics under different scales; the dimension conversion layer converts the fused features into a form suitable for the processing of the full connection layer (namely, the fused multidimensional features are unidimensionalized); the full connection layer is used for further extraction and combination of the features; the output layer is used for outputting the results of pesticide residue concentration, detection precision and the like.
Further, the output end of the low-level characteristic extraction layer in the upstream network layer is connected with the input end of the self-adaptive analysis channel, and the input end of the cascade fusion layer in the downstream network layer is connected with the output end of the self-adaptive analysis channel, so that a complete neural network model is generated.
Optionally, the generated neural network model is supervised and trained by using the historical pesticide detection record so as to optimize network parameters. After training, the training result is output as an online analysis channel for detecting pesticide residues in practical application. The original detection data in the historical pesticide detection record is an input sample for supervising the training process and is used for providing the characteristics required by model training; the detection result data is a target label and is used for providing a true value required by model training.
Specifically, the preprocessed original detection data is used as input layer data of a neural network model, the detection result data is used as a target label of the neural network model, the multi-scale structural configuration self-adaptive analysis channel is utilized to extract multi-level features of the original detection data, the features are comprehensively analyzed through a cascade fusion layer, a dimension conversion layer and a full connection layer, network parameters are optimized based on a back propagation algorithm and a loss function, and a predicted value output by the model is enabled to be as close to the detection result data as possible.
And activating an online detection device and a sampling detection device, and carrying out synchronous detection acquisition of the target scene by combining a feasibility analysis result to obtain a real-time detection data set, wherein the real-time detection data set comprises continuous detection data and interval detection data.
In some embodiments, activating an online detection device and a sampling detection device, and performing synchronous detection acquisition of the target scene according to a feasibility analysis result to obtain a real-time detection data set, including:
and performing equipment matching based on the feasibility analysis result, and activating the online detection equipment according to the equipment matching result.
And activating the sampling detection device based on the difference set between the device matching result and the detection device set of the target scene.
And the on-line detection equipment and the sampling detection equipment synchronously detect and collect according to the preset on-line sampling granularity parameter and output a real-time detection data set.
Specifically, a real-time detection data set is acquired, and firstly, an online detection device and a sampling detection device which are most suitable for a target scene are selected according to a feasibility analysis result. The matching of the equipment ensures that the detection equipment is matched with the requirements of the target scene, and is beneficial to improving the detection accuracy. And then activating the selected on-line detection equipment according to the equipment matching result, ensuring that the on-line detection equipment is in a working state, and carrying out necessary calibration and configuration.
Further, according to the difference set of the device matching result and the detection device set of the target scene, determining the detection requirement of the target scene which is not covered by the online detection device, activating the sampling detection device corresponding to the difference set of the detection device set of the target scene, and preparing for corresponding sampling, so as to ensure that all the detection requirements can be covered.
Further, the on-line detection equipment and the sampling detection equipment synchronously detect and collect according to the preset on-line sampling granularity parameters, and combine the data collected by the on-line detection equipment and the sampling detection equipment to form a real-time detection data set, so that the real-time performance and the integrity of the data are ensured, and the follow-up analysis and processing are facilitated. The online sampling granularity parameters comprise online granularity parameters and sampling granularity parameters, and the detection frequencies of the online detection equipment and the sampling detection equipment are specified.
And analyzing the real-time detection data set based on the online analysis channel, obtaining an interval detection result and a continuous detection result, fitting the interval detection result and the continuous detection result, and obtaining a residual detection result.
Optionally, the on-line analysis channel takes the continuous detection result in the real-time detection data set as input, analyzes and obtains the continuous detection result, the continuous detection result is characterized as a time sequence of concentration values, the sampling frequency is higher, and the interval detection result is a plurality of concentration values output by the sampling detection device.
In some embodiments, analyzing the real-time detection dataset based on the online analysis channel, obtaining an interval detection result and a continuous detection result, and fitting the interval detection result and the continuous detection result, obtaining a residual detection result, comprising:
inputting the continuous detection data to the online analysis channel, obtaining an online concentration sequence, performing curve fitting on the online concentration sequence, obtaining a residual concentration fluctuation curve, and outputting the continuous detection result.
And carrying out scattered point distribution based on the interval detection result to obtain a fitting reference space.
And performing expansion and contraction transformation on the residual concentration fluctuation curve of the continuous detection result, fitting the residual concentration fluctuation curve to the fitting reference space, and obtaining an approximate concentration change curve.
And outputting a residual detection result based on the approximate concentration change curve and the continuous detection result.
Optionally, the online analysis channel processes the continuous detection data and outputs an online concentration sequence, wherein the online concentration sequence comprises a change trend of pesticide residue concentration. And then, performing curve fitting on the online concentration sequence to find a mathematical model which best describes the data change trend so as to more accurately describe the change of the concentration along with time and obtain a residual concentration fluctuation curve. Finally, the residual concentration fluctuation curve is output as a result of continuous detection.
Optionally, the pesticide residue results detected at intervals are subjected to scattered point distribution, and a fitting reference space is obtained. The fitting reference space is used for representing actual values of pesticide residue concentration at different time points, which cannot be obtained through online detection. The fit reference space is a two-dimensional or three-dimensional space, and each point represents a group of pesticide residue concentration values corresponding to a sampling node in the interval detection result.
Further, the residual concentration fluctuation curve of the continuous detection result is subjected to expansion and contraction conversion. Including translation, rotation, scaling, etc., so that the transformed residual concentration fluctuation curve can match the fit reference space as much as possible. Specifically, the fitting process is optimized by minimizing the distances between a plurality of scattered points in the fitting reference space and the residual concentration fluctuation curve, so that fitting is realized, and an approximate concentration change curve is obtained.
Through the steps, the obtained approximate concentration change curve can better describe the concentration continuous change condition of the detection indexes which cannot be detected on line in the detection index set by utilizing the existing interval detection result and the continuous detection result, and an effective alternative scheme is provided for the indexes which cannot be detected on line, so that the concentration change trend of the detection indexes can be continuously and dynamically reflected. The steps of the data-based driving method can be suitable for various pesticide residue conditions, and have good universality and adaptability.
Further, after the residual detection result is obtained, the method further includes:
And based on the verification sampling interval, carrying out supplementary detection by the sampling detection equipment to obtain verification detection data.
And calling the residual detection result, and performing deviation analysis according to the verification detection data.
And if the deviation analysis result does not meet the preset detection error space, carrying out feedback optimization of the online analysis channel based on the verification detection data.
Specifically, first, according to a preset verification sampling interval, a sampling detection device is used for carrying out supplementary detection on a target scene, and a verification detection data set is obtained. The verification detection data is a data base for evaluating the accuracy of the detection result of the online analysis channel. And then, calling a residual detection result, and performing deviation analysis according to the newly acquired verification detection data, wherein the deviation analysis comprises the steps of calculating the difference between a predicted value and an actual value, calculating the statistical characteristics (such as average value, standard deviation and the like) of the difference, and the like.
Optionally, if the result of the deviation analysis does not meet the preset detection error space, it may be considered that there is a larger deviation in the online analysis channel, and feedback optimization needs to be performed on the online analysis channel, including performing reinforcement training on the online analysis channel by using verification detection data, and adjusting parameters of the model to improve accuracy and reliability of the model.
Through the steps, the detection and feedback optimization are carried out regularly, so that the online analysis channel is continuously optimized and improved, the pesticide residue detection can be accurately and reliably carried out, and the accuracy and the reliability of the detection result are improved.
In summary, the pesticide residue detection method for agricultural products provided by the invention has the following technical effects:
Acquiring a detection index set through an interactive target scene, and carrying out online detection feasibility analysis on the detection index set; acquiring a historical pesticide detection record according to a feasibility analysis result; based on historical pesticide detection records, an online analysis channel is built by combining a neural network model, and the online analysis channel comprises a plurality of self-adaptive analysis channels with different complexity; activating an online detection device and a sampling detection device, and synchronously detecting and acquiring a target scene by combining a feasibility analysis result to acquire a real-time detection data set, wherein the real-time detection data set comprises continuous detection data and interval detection data; based on the online analysis channel, analyzing the real-time detection data set, obtaining an interval detection result and a continuous detection result, fitting the interval detection result and the continuous detection result, and obtaining a residual detection result. The method and the device for detecting the pesticide residue of the agricultural product solve the technical problems of high detection cost and incomplete detection. Therefore, the technical effects of optimizing the monitoring cost and improving the comprehensiveness of monitoring are achieved.
Example two
Fig. 2 is a schematic structural view of the agricultural chemical residue detecting apparatus for agricultural products according to the present invention. For example, the flow chart of the pesticide residue detection method for agricultural products of the present invention in fig. 1 may be implemented by the structure shown in fig. 2.
Based on the same concept as the pesticide residue detection method for agricultural products in the embodiment, the pesticide residue detection device for agricultural products further provided by the invention comprises:
The index acquisition and analysis module 11 is used for acquiring a detection index set according to the interaction target scene and carrying out online detection feasibility analysis on the detection index set.
The detection record obtaining module 12 is configured to obtain a historical pesticide detection record according to the feasibility analysis result, where the historical pesticide detection record includes a historical online detection record and a historical sampling detection record.
And the analysis channel construction module 13 is used for constructing an online analysis channel based on the historical pesticide detection record and combining a neural network model, wherein the online analysis channel comprises a plurality of adaptive analysis channels with different complexity.
And the synchronous detection acquisition module 14 is used for activating the online detection equipment and the sampling detection equipment, and carrying out synchronous detection acquisition of the target scene by combining the feasibility analysis result to acquire a real-time detection data set, wherein the real-time detection data set comprises continuous detection data and interval detection data.
And the analysis output module 15 is configured to analyze the real-time detection dataset based on the online analysis channel, obtain an interval detection result and a continuous detection result, and fit the interval detection result and the continuous detection result to obtain a residual detection result.
Wherein, the index acquisition analysis module 11 includes:
The residual detection index extraction unit is used for interacting the target scene, extracting a plurality of residual detection indexes based on a pesticide residual detection task book of the target scene, and storing the plurality of residual detection indexes as the detection index set.
The feasibility analysis unit is used for traversing the detection index set, acquiring a plurality of multi-dimensional discrimination feature sets corresponding to the residual detection indexes based on big data, and carrying out online detection feasibility analysis based on the multi-dimensional discrimination feature sets, wherein the multi-dimensional discrimination feature sets comprise equipment adaptability, accuracy accessibility and detection instantaneity.
Further, the detection record acquisition module 12 includes:
And the key analysis unit is used for extracting feasible detection indexes to perform principal component analysis based on the feasibility analysis result to obtain a key online detection index set.
And the record calling unit is used for accessing the pesticide detection database by taking the key online detection index set as index constraint and calling the historical online detection record and the historical sampling detection record.
And the record cleaning unit is used for extracting the original detection data of the historical online detection record and the detection result data of the historical sampling detection record and storing the original detection data and the detection result data into the historical pesticide detection record.
Further, the analysis channel construction module 13 includes:
The multi-scale configuration unit is used for defining a multi-scale structure and configuring corresponding self-adaptive analysis channels based on the multi-scale structure to generate a plurality of self-adaptive analysis channels.
And the upstream network layer unit is used for constructing an upstream network layer of the neural network model, wherein the upstream network layer comprises an input layer and a low-level feature extraction layer.
And the downstream network layer unit is used for constructing a downstream network layer of the neural network model, wherein the downstream network layer comprises a cascade fusion layer, a dimension conversion layer, a full connection layer and an output layer.
The connection training unit is used for sequentially connecting the upstream network layer, the plurality of parallel self-adaptive analysis channels and the downstream network layer, performing supervision training of the neural network model based on the historical pesticide detection record and outputting a training result as the online analysis channel.
Further, the synchronous detection and acquisition module 14 includes:
and the online detection equipment matching activation unit is used for carrying out equipment matching based on the feasibility analysis result and activating the online detection equipment according to the equipment matching result.
And the sampling detection device matching and activating unit is used for activating the sampling detection device based on the difference set between the device matching result and the detection device set of the target scene.
And the real-time synchronous detection unit is used for synchronously detecting and acquiring the on-line detection equipment and the sampling detection equipment according to the preset on-line sampling granularity parameter and outputting a real-time detection data set.
Further, the parsing output module 15 includes:
And the continuous analysis unit is used for inputting the continuous detection data into the online analysis channel, obtaining an online concentration sequence, performing curve fitting on the online concentration sequence, obtaining a residual concentration fluctuation curve, and outputting the residual concentration fluctuation curve as the continuous detection result.
And the interval mapping unit is used for carrying out scattered point distribution based on the interval detection result to obtain a fitting reference space.
And the approximate fitting unit is used for performing expansion and contraction transformation on the residual concentration fluctuation curve of the continuous detection result, fitting the residual concentration fluctuation curve to the fitting reference space and obtaining an approximate concentration change curve.
And the detection output unit is used for outputting a residual detection result based on the approximate concentration change curve and the continuous detection result.
In some embodiments, the apparatus further comprises:
And the supplementary detection unit is used for carrying out supplementary detection through the sampling detection equipment based on the verification sampling interval to acquire verification detection data.
And the deviation analysis unit is used for calling the residual detection result and carrying out deviation analysis according to the verification detection data.
And the feedback optimization unit is used for carrying out feedback optimization on the online analysis channel based on the verification detection data if the deviation analysis result does not meet the preset detection error space.
It should be understood that the embodiments mentioned in this specification focus on the differences from other embodiments, and the specific embodiment in the first embodiment is equally applicable to the pesticide residue detecting device for agricultural products described in the second embodiment, and is not further developed herein for brevity of the specification.
It is to be understood that both the foregoing description and the embodiments of the present invention enable one skilled in the art to utilize the present invention. While the invention is not limited to the embodiments described above, it should be understood that: modifications of the technical solutions described in the foregoing embodiments or equivalent substitutions of some technical features thereof may be still performed by those skilled in the art; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention, and are intended to be included in the scope of the present invention.
Claims (5)
1. A pesticide residue detection method for agricultural products, characterized in that the method comprises:
the method comprises the steps of interacting a target scene, obtaining a detection index set, and carrying out online detection feasibility analysis on the detection index set;
acquiring a historical pesticide detection record according to a feasibility analysis result, wherein the historical pesticide detection record comprises a historical online detection record and a historical sampling detection record;
Based on the historical pesticide detection record, combining a neural network model to construct an online analysis channel, wherein the online analysis channel comprises a plurality of self-adaptive analysis channels with different complexity;
activating an online detection device and a sampling detection device, and carrying out synchronous detection acquisition of the target scene by combining a feasibility analysis result to obtain a real-time detection data set, wherein the real-time detection data set comprises continuous detection data and interval detection data;
Analyzing the real-time detection data set based on the online analysis channel, obtaining an interval detection result and a continuous detection result, fitting the interval detection result and the continuous detection result, and obtaining a residual detection result;
the method for analyzing the detection target scene comprises the steps of acquiring a detection index set, and carrying out online detection feasibility analysis on the detection index set, wherein the method comprises the following steps:
The target scene is interacted, a plurality of residual detection indexes are extracted based on a pesticide residue detection task book of the target scene, and the residual detection indexes are stored as the detection index set;
Traversing the detection index set, acquiring a plurality of multi-dimensional discrimination feature sets corresponding to a plurality of residual detection indexes based on big data, and carrying out online detection feasibility analysis based on the plurality of multi-dimensional discrimination feature sets, wherein the multi-dimensional discrimination feature sets comprise equipment adaptability, accuracy accessibility and detection instantaneity;
Based on the historical pesticide detection record, an online analysis channel is constructed by combining a neural network model, wherein the online analysis channel comprises a plurality of self-adaptive analysis channels with different complexity, and the online analysis channel comprises:
Defining a multi-scale structure, configuring corresponding self-adaptive analysis channels based on the multi-scale structure, and generating a plurality of self-adaptive analysis channels;
An upstream network layer of the neural network model is constructed, wherein the upstream network layer comprises an input layer and a low-level feature extraction layer;
constructing a downstream network layer of the neural network model, wherein the downstream network layer comprises a cascade fusion layer, a dimension conversion layer, a full connection layer and an output layer;
The upstream network layer, the plurality of adaptive analysis channels and the downstream network layer are sequentially connected, supervision training of the neural network model is performed based on the historical pesticide detection record, and a training result is output as the online analysis channel;
Based on the online analysis channel, analyzing the real-time detection data set, obtaining an interval detection result and a continuous detection result, fitting the interval detection result and the continuous detection result, and obtaining a residual detection result, including:
inputting the continuous detection data to the online analysis channel, obtaining an online concentration sequence, performing curve fitting on the online concentration sequence, obtaining a residual concentration fluctuation curve, and outputting the residual concentration fluctuation curve as the continuous detection result;
Carrying out scattered point distribution based on the interval detection result to obtain a fitting reference space;
performing expansion transformation on the residual concentration fluctuation curve of the continuous detection result, fitting the residual concentration fluctuation curve to the fitting reference space, and obtaining an approximate concentration change curve;
and outputting a residual detection result based on the approximate concentration change curve and the continuous detection result.
2. The pesticide residue detection method for agricultural products of claim 1, wherein acquiring a historical pesticide detection record based on the feasibility analysis result comprises:
Based on the feasibility analysis result, extracting feasible detection indexes to perform principal component analysis, and obtaining a key online detection index set;
accessing a pesticide detection database by taking the key online detection index set as index constraint, and calling a historical online detection record and a historical sampling detection record;
extracting the original detection data of the historical online detection record and the detection result data of the historical sampling detection record, and storing the detection result data into the historical pesticide detection record.
3. The pesticide residue detection method for agricultural products of claim 1, wherein activating an on-line detection device and a sampling detection device, and performing synchronous detection acquisition of the target scene in combination with a feasibility analysis result, to obtain a real-time detection data set, comprises:
performing equipment matching based on the feasibility analysis result, and activating the online detection equipment according to the equipment matching result;
activating the sampling detection device based on a difference set between the device matching result and the detection device set of the target scene;
and the on-line detection equipment and the sampling detection equipment synchronously detect and collect according to the preset on-line sampling granularity parameter and output a real-time detection data set.
4. The pesticide residue detection method for agricultural products according to claim 1, wherein after obtaining the residue detection result, the method further comprises:
Based on the verification sampling interval, carrying out supplementary detection by the sampling detection equipment to obtain verification detection data;
Calling the residual detection result, and performing deviation analysis according to the verification detection data;
and if the deviation analysis result does not meet the preset detection error space, carrying out feedback optimization of the online analysis channel based on the verification detection data.
5. A pesticide residue detecting apparatus for agricultural products, wherein the apparatus is for performing the pesticide residue detecting method for agricultural products as set forth in any one of claims 1 to 4, the apparatus comprising:
the index acquisition and analysis module is used for acquiring a detection index set and carrying out online detection feasibility analysis on the detection index set;
The detection record acquisition module is used for acquiring historical pesticide detection records according to feasibility analysis results, wherein the historical pesticide detection records comprise historical online detection records and historical sampling detection records;
The analysis channel construction module is used for constructing an online analysis channel based on the historical pesticide detection record and combining a neural network model, wherein the online analysis channel comprises a plurality of self-adaptive analysis channels with different complexity;
The synchronous detection acquisition module is used for activating the online detection equipment and the sampling detection equipment, and carrying out synchronous detection acquisition of the target scene by combining a feasibility analysis result to acquire a real-time detection data set, wherein the real-time detection data set comprises continuous detection data and interval detection data;
The analysis output module is used for analyzing the real-time detection data set based on the online analysis channel, obtaining an interval detection result and a continuous detection result, fitting the interval detection result and the continuous detection result and obtaining a residual detection result;
the method for analyzing the detection target scene comprises the steps of acquiring a detection index set, and carrying out online detection feasibility analysis on the detection index set, wherein the method comprises the following steps:
The target scene is interacted, a plurality of residual detection indexes are extracted based on a pesticide residue detection task book of the target scene, and the residual detection indexes are stored as the detection index set;
Traversing the detection index set, acquiring a plurality of multi-dimensional discrimination feature sets corresponding to a plurality of residual detection indexes based on big data, and carrying out online detection feasibility analysis based on the multi-dimensional discrimination feature sets, wherein the multi-dimensional discrimination feature sets comprise equipment adaptability, accuracy accessibility and detection instantaneity.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410889725.4A CN118427577B (en) | 2024-07-04 | 2024-07-04 | Pesticide residue detection method and device for agricultural products |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410889725.4A CN118427577B (en) | 2024-07-04 | 2024-07-04 | Pesticide residue detection method and device for agricultural products |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118427577A CN118427577A (en) | 2024-08-02 |
| CN118427577B true CN118427577B (en) | 2024-09-20 |
Family
ID=92321887
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410889725.4A Active CN118427577B (en) | 2024-07-04 | 2024-07-04 | Pesticide residue detection method and device for agricultural products |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118427577B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103630509A (en) * | 2013-12-03 | 2014-03-12 | 江苏大学 | On-line pesticide concentration detection device and method |
| CN107103571A (en) * | 2017-04-17 | 2017-08-29 | 中国检验检疫科学研究院 | Residues of pesticides detecting data platform and detecting report automatic generation method based on high resolution mass spectrum, internet and data science |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107180076B (en) * | 2017-04-18 | 2018-08-24 | 中国检验检疫科学研究院 | Pesticide residue visual method based on high resolution mass spectrum+internet+geography information |
| CN109633106B (en) * | 2018-12-27 | 2021-06-08 | 广州安食通信息科技有限公司 | Online pesticide residue rapid detection method and system and storage medium |
| CN216117310U (en) * | 2021-09-07 | 2022-03-22 | 福州大学 | Multi-spectral detection system for rapidly detecting pesticide residues of fruits and vegetables in real time |
| CN117787510B (en) * | 2024-02-28 | 2024-05-03 | 青岛小蜂生物科技有限公司 | Optimization method of pesticide residue monitoring process based on time sequence predictive analysis |
-
2024
- 2024-07-04 CN CN202410889725.4A patent/CN118427577B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103630509A (en) * | 2013-12-03 | 2014-03-12 | 江苏大学 | On-line pesticide concentration detection device and method |
| CN107103571A (en) * | 2017-04-17 | 2017-08-29 | 中国检验检疫科学研究院 | Residues of pesticides detecting data platform and detecting report automatic generation method based on high resolution mass spectrum, internet and data science |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118427577A (en) | 2024-08-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20070061144A1 (en) | Batch statistics process model method and system | |
| CN117196066A (en) | Intelligent operation and maintenance information analysis model | |
| CN117368651B (en) | Comprehensive analysis system and method for faults of power distribution network | |
| CN114723287A (en) | Quantitative statistical method for risk formation based on enterprise characteristics and operation behaviors | |
| CN116985183B (en) | Quality monitoring and management method and system for near infrared spectrum analyzer | |
| CN117111544B (en) | Automatic-adaptation building internet of things monitoring method and system | |
| CN113537642A (en) | Product quality prediction method, device, electronic equipment and storage medium | |
| CN116187861A (en) | Isotope-based water quality traceability monitoring method and related device | |
| CN111160393A (en) | Data-driven modularized modeling method for carrier rocket health assessment model | |
| CN113642209B (en) | Structure implantation fault response data acquisition and evaluation method based on digital twinning | |
| CN118211882A (en) | Product quality management system and method based on big data | |
| CN117131364B (en) | Rolling bearing process detection integration method and system | |
| CN117664518B (en) | Method and system for optical calibration by using stable light source | |
| CN118427577B (en) | Pesticide residue detection method and device for agricultural products | |
| CN113283512A (en) | Data anomaly detection method, device, equipment and storage medium | |
| Tolas et al. | Periodicity detection algorithm and applications on IoT data | |
| CN115659271A (en) | Sensor abnormality detection method, model training method, system, device, and medium | |
| CN114116831B (en) | Big data mining processing method and device | |
| JP6641056B1 (en) | Device abnormality diagnosis method and device abnormality diagnosis system | |
| JP6633403B2 (en) | Analysis target determination apparatus and analysis target determination method | |
| CN118566173B (en) | Noise suppression effect inspection method and device | |
| Yadav et al. | Hybrid model for software fault prediction | |
| CN118211061B (en) | Multi-index fusion and service perception acquisition system operation monitoring method and system | |
| Galatro et al. | Data Analytics for Process Engineers: Prediction, Control and Optimization | |
| CN115169423B (en) | Stamping signal processing method, device, equipment and readable storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |