A fully automatic trace element automatic sampling system

Abstract

The present invention discloses a fully automatic trace element automatic sampling system, which relates to the technical field of clinical diagnosis and detection, and solves the technical problems that it is difficult to analyze whether a sampling needle is added to a sample tube and it is difficult to make an initial warning; secondly, it is difficult to detect and analyze bubbles in the sample tube according to the judgment result; then, it is difficult to identify and classify the samples, and it is difficult to analyze the content and distribution of trace elements in the samples according to the analysis results; finally, it is difficult to comprehensively analyze the abnormal comprehensive index and make an early warning according to various data. The present invention realizes the analysis of whether a sampling needle is added to a sample tube according to video images, makes an initial early warning, and improves the detection efficiency through a data acquisition module, a data processing module, a data analysis module, and a comprehensive evaluation module; it realizes the comprehensive analysis of the bubble influence coefficient, the sample deviation value, the trace element variation coefficient, and the environmental pollution coefficient, and makes an early warning.

Description

Technical Field

[0001] The invention belongs to the technical field of clinical diagnosis and detection, and in particular is a full-automatic trace element automatic sampling system. Background Art

[0002] The automatic sample loading system is designed to reduce manual operation and human errors in the laboratory and improve the accuracy and efficiency of the experiment. With the advancement of science and technology, modern automatic sample loading systems have been fully automated, which can not only complete the automatic addition and distribution of samples, but also perform a variety of operations such as mixing, heating, cooling, and shaking. Trace elements are vital to the human body and are closely related to the etiology, diagnosis, treatment, prevention, and health care of clinical diseases. Traditional trace element detection methods have problems such as cumbersome operation and large human errors, and cannot meet the accuracy and efficiency requirements of modern clinical diagnosis.

[0003] The existing fully automatic trace element automatic sampling system has the following problems: first, it is difficult to analyze whether the sampling needle is added to the sample tube and it is difficult to issue an initial warning; second, it is difficult to detect and analyze the bubbles in the sample tube based on the judgment results; then, it is difficult to identify and classify the samples, and it is difficult to analyze the content and distribution of trace elements in the samples based on the analysis results; finally, it is difficult to comprehensively analyze the abnormal comprehensive index based on various data to issue a warning. Summary of the invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes a fully automatic trace element automatic sampling system for solving the following technical problems:

[0005] First, it is difficult to analyze whether the sample needle is inserted into the sample tube and to issue an initial warning. Second, it is difficult to detect and analyze the bubbles in the sample tube based on the judgment results. Then, it is difficult to identify and classify the samples, and to analyze the content and distribution of trace elements in the samples based on the analysis results. Finally, it is difficult to comprehensively analyze the abnormal comprehensive index based on various data to issue a warning.

[0006] To solve the above problems, the first aspect of the present invention provides a fully automatic trace element automatic sampling system, comprising the following modules:

[0007] Data acquisition module: real-time acquisition of sample data and video images during the sampling process;

[0008] Data processing module: cleans, denoises and standardizes the collected data;

[0009] De-noising and image enhancement of the collected video image data;

[0010] Data analysis module: Analyze whether the sample needle is added to the sample tube according to the collected and processed video images, and determine whether to issue an initial warning; analyze bubbles and calculate the bubble influence coefficient according to the analysis results; analyze and calculate the sample deviation value according to the collected and processed sample data; identify and classify samples according to the collected and processed video images, and analyze the content and distribution of trace elements in the samples according to the analysis results; analyze the coefficient of variation of trace elements in the samples according to the image recognition results and trace element detection results; analyze the pollutants in the samples and calculate the environmental pollution coefficient according to the analysis results;

[0011] Comprehensive evaluation module: Based on the bubble influence coefficient, sample deviation value, trace element variation coefficient and environmental pollution coefficient, the abnormal comprehensive index is calculated and analyzed, and a comprehensive evaluation and early warning are carried out based on the calculation results.

[0012] As a further solution of the present invention: real-time acquisition of sample data during the sample addition process includes the following steps:

[0013] In the fully automatic trace element automatic sampling system, multiple sensors are installed to monitor the sample data in real time during the sampling process; the sensors include but are not limited to volume sensors, speed sensors, acceleration sensors, position sensors and time sensors;

[0014] The sample data includes: sample volume, sample speed, acceleration of the sample needle, actual height of the sample needle to the target height of the container mouth, and actual time from the start of movement of the sample needle to the target height.

[0015] As a further solution of the present invention: according to the collected and processed video images, analyzing whether the sample adding needle is added to the sample tube and judging whether to issue an initial warning comprises the following steps:

[0016] According to the collected and processed video images of the sample addition process, the images are converted into grayscale images or binary images, and the features of the sample addition needle and the sample tube in the images are extracted using image segmentation or edge detection technology, including: the outline, position and direction of the sample addition needle and the sample tube;

[0017] By extracting the contour, a contour point set of the sample needle and the sample tube is obtained; using the contour point set, the pixel coordinates of the center of mass of the sample needle and the center of mass of the sample tube are obtained;

[0018] By analyzing the formula:

[0019]

[0020] Get the centroid coordinates of the contour points , where N represents the number of contour points, Represents the coordinates of the contour point set;

[0021] By substituting the above formula, we can get the pixel coordinates of the centroid of the sample needle: , pixel coordinates of the centroid of the sample tube ;

[0022] By analyzing the formula:

[0023] The distance D between the pixel coordinate point of the mass center of the sample injection needle and the pixel coordinate point of the mass center of the sample tube is obtained; wherein, represents the pixel coordinates of the centroid of the pipette, represents the pixel coordinates of the centroid of the sample tube;

[0024] By analyzing the formula:

[0025]

[0026] According to the above analytical formula, the inverse trigonometric function formula is calculated by analysis:

[0027]

[0028] Get the angle between the direction vector of the center of mass of the sample needle and the direction vector of the center of mass of the sample tube ;in,( , ) represents the direction vector of the center of mass of the sample injection needle, ( , ) represents the direction vector of the sample tube’s center of mass, represents the modulus of the direction vector of the center of mass of the sample needle, The modulus of the direction vector representing the centroid of the sample tube;

[0029] According to the distance between the pixel coordinate point of the centroid of the sample needle and the pixel coordinate point of the centroid of the sample tube and the angle between the direction vectors, a threshold is set to determine whether the sample needle is added into the sample tube;

[0030] when and When the sample is added, the sample needle is added into the sample tube, and the sample addition and detection are continued;

[0031] when or When the sample needle is not added into the sample tube, an initial warning is issued.

[0032] As a further solution of the present invention: analyzing the bubbles and calculating the bubble influence coefficient according to the analysis results, comprising the following steps:

[0033] When the sample needle is added to the sample tube, the pre-processed video image is used to detect bubbles using the deep learning model;

[0034] Collect and annotate bubble images in the sample tube, and input the annotated bubble images into the deep learning model for learning and training;

[0035] According to the video images collected and processed in real time by the fully automatic trace element automatic sampling system, whether there are bubbles in the video images of the sampling process;

[0036] When bubbles appear when the sample needle is inserted into the sample tube, the size and number of the bubbles are analyzed;

[0037] When no bubbles appear when the sample needle is inserted into the sample tube, continue to add samples for testing;

[0038] By analyzing the formula:

[0039] Get the pixel area A of the bubble, where S represents the total number of pixels occupied by the bubble in the image. represents the area occupied by the a-th pixel;

[0040] By analyzing the formula:

[0041] Get the bubble influence coefficient ,in represents the pixel area of ​​the bubble, Indicates the number of bubbles.

[0042] As a further solution of the present invention: analyzing and calculating the sample deviation value according to the collected and pre-processed sample data, the following steps are included:

[0043] By analyzing the formula:

[0044] After finishing, we get:

[0045] Solving this quadratic equation using the root-finding formula yields:

[0046] The time t required for the sample needle to reach the target height is obtained, where represents the initial velocity of the sample injection needle, a represents the acceleration of the sample injection needle, and h represents the target height of the sample injection needle to the container opening;

[0047] The sample volume is calculated by the formula:

[0048] Get the sample volume ,in, represents the flow rate of the sample, Indicates the time required for the sample needle to reach the target height;

[0049] The deviation calculation formula of the required time is analyzed:

[0050] The deviation of the sample size is calculated by the formula:

[0051] By analyzing the sample deviation value calculation formula:

[0052] Get the sample deviation value P, where Indicates the deviation value of the time required for the sample needle to reach the target height; Indicates the deviation value of the sample amount; Indicates the actual time from the start of the sample needle movement to the target height. Indicates the actual amount of sample added.

[0053] As a further solution of the present invention: according to the collected and processed video images, the samples are identified and classified, and the content and distribution of trace elements in the samples are analyzed according to the analysis results, including the following steps:

[0054] Pre-treat the samples, including grinding, digestion, dilution and volume adjustment, filtration and separation;

[0055] Annotate the collected pre-processed sample images and extract image features, including: color, texture and shape;

[0056] Use deep learning models to train and learn the extracted features;

[0057] The video images are collected in real time by the fully automatic trace element automatic sampling system, and the sample images are recognized by using the trained model; the samples are classified according to the recognition results, and the classification includes the name and content range of the samples;

[0058] According to the recognition results of the video images, the m samples are tested for trace elements; the content and distribution of trace elements in the i-th sample are determined using spectral analysis or mass spectrometry, where .

[0059] As a further solution of the present invention: analyzing the coefficient of variation of trace elements in the sample according to the recognition result of the video image and the detection result of trace elements, comprising the following steps:

[0060] By analyzing the mean value of the trace element content in the i-th sample, the formula is:

[0061]

[0062] By analyzing the standard deviation of the trace element content in the i-th sample, the formula is:

[0063]

[0064] Get the mean value of trace element content in the i-th sample and standard deviation ;in, represents the content of trace elements detected in the i-th sample, and m represents the total number of samples tested;

[0065] By analyzing the coefficient of variation of trace elements, the formula is calculated:

[0066] The coefficient of variation of trace elements β is obtained; where represents the mean value of the trace element content in the i-th sample, Represents the standard deviation of the trace element content in the i-th sample.

[0067] As a further solution of the present invention: analyzing the pollutants in the sample and calculating the environmental pollution coefficient according to the analysis results, comprising the following steps:

[0068] During the sample addition process, the sample is tested for pollutants; the environmental data of the sample is monitored in real time through a gas detector, a temperature sensor, a humidity sensor, and a light intensity sensor, and the collected environmental data is cleaned and normalized, and the environmental data includes: temperature, humidity, light intensity, and gas concentration;

[0069] After completing the sampling and pollutant detection, the environmental pollution coefficient is calculated by analyzing the formula:

[0070]

[0071] The environmental pollution coefficient R is obtained; where T represents temperature, S represents humidity, and L represents light intensity. Indicates gas concentration; represents the influence weight of temperature, represents the influence weight of humidity, Represents the influence weight of light intensity, Indicates the influence weight of gas concentration; is 0.35, is 0.3, is 0.15, is 0.2.

[0072] As a further solution of the present invention: comprehensively analyzing and calculating the abnormal comprehensive index according to the bubble influence coefficient, sample deviation value, trace element variation coefficient and environmental pollution coefficient, and making a comprehensive evaluation and early warning according to the calculation results, including the following steps:

[0073] By analyzing the abnormal comprehensive index calculation formula:

[0074]

[0075] Get the abnormal comprehensive index U; represents the bubble influence coefficient, Indicates the sample deviation value, represents the coefficient of variation of trace elements, Indicates the environmental pollution coefficient; represents the weight of the bubble influence coefficient, represents the weight of the sample deviation value, represents the weight of the coefficient of variation of trace elements, Indicates the weight of the environmental pollution coefficient;

[0076] According to the calculation results, when the abnormal comprehensive index When the abnormal comprehensive index Continue testing.

[0077] Compared with the prior art, the present invention has the following beneficial effects:

[0078] The present invention recognizes and classifies samples through video images, and automatically analyzes the content and distribution of trace elements in samples, as well as the coefficient of variation of trace elements, thereby achieving a high degree of automation and intelligence; the automated and intelligent data processing and analysis process greatly improves work efficiency and reduces labor costs; the present invention can issue an initial warning when the analysis sample addition needle is not added to the sample tube through the initial early warning function, thereby improving laboratory work efficiency, reducing human errors, and improving data accuracy and reliability;

[0079] The present invention calculates the abnormal comprehensive index by comprehensive analysis based on the bubble influence coefficient, sample deviation value, trace element variation coefficient and environmental pollution coefficient, and issues a comprehensive evaluation warning, which helps to timely discover potential problems and take preventive measures. Through real-time monitoring and early warning, problems in the experimental process can be discovered and solved in time to avoid possible losses and risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0081] Figure 1 It is a system module diagram of the present invention. DETAILED DESCRIPTION

[0082] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0083] See also Figure 1 The first aspect of the present invention provides a fully automatic trace element automatic sampling system, comprising the following modules:

[0084] Data acquisition module: real-time acquisition of sample data and video images during the sampling process;

[0085] Data processing module: cleans, denoises and standardizes the collected data;

[0086] De-noising and image enhancement of the collected video image data;

[0087] Data analysis module: Analyze whether the sample needle is added to the sample tube according to the collected and processed video images, and determine whether to issue an initial warning; analyze bubbles and calculate the bubble influence coefficient according to the analysis results; analyze and calculate the sample deviation value according to the collected and processed sample data; identify and classify samples according to the collected and processed video images, and analyze the content and distribution of trace elements in the samples according to the analysis results; analyze the coefficient of variation of trace elements in the samples according to the image recognition results and trace element detection results; analyze the pollutants in the samples and calculate the environmental pollution coefficient according to the analysis results;

[0088] Comprehensive evaluation module: Based on the bubble influence coefficient, sample deviation value, trace element variation coefficient and environmental pollution coefficient, the abnormal comprehensive index is calculated and analyzed, and a comprehensive evaluation and early warning are carried out based on the calculation results.

[0089] Specifically, the data acquisition module is used to collect sample data and video images in the process of adding samples in real time by using high-definition cameras and sensors. The data processing module is used to clean, denoise and standardize the collected data. For video image data, denoise and enhance the contrast and edge information of the image. The data analysis module is used to analyze whether the sample addition needle is added to the sample tube according to the collected and processed video images, determine whether to issue an initial warning, and analyze the bubbles and calculate the bubble influence coefficient when the sample addition needle is added to the sample tube. The sample deviation value is analyzed and calculated according to the collected and processed sample data, and the samples are identified and classified according to the collected and processed video images, and the content and distribution of trace elements in the samples are analyzed according to the analysis results. According to the image recognition results and the trace element detection results, the coefficient of variation of trace elements in the samples is analyzed; the pollutants of the samples are analyzed and the environmental pollution coefficient is calculated according to the analysis results. The comprehensive evaluation module is used to calculate the abnormal comprehensive index according to the comprehensive analysis of various data. The index reflects the overall abnormal situation in terms of bubbles, trace element content, variability and pollution degree. Comprehensive evaluation and warning are carried out according to the calculation results, and if the abnormal comprehensive index exceeds the set threshold, a warning signal is issued. By establishing accurate data analysis models and methods, key indicators such as bubble influence coefficient, sample deviation value, trace element variation coefficient and environmental pollution coefficient are accurately calculated and analyzed. Reasonable abnormal comprehensive index thresholds and early warning mechanisms are set to ensure that abnormal situations can be discovered and handled in a timely manner.

[0090] In one embodiment of the present invention, real-time acquisition of sample data during the sample addition process includes the following steps:

[0091] In the fully automatic trace element automatic sampling system, multiple sensors are installed to monitor the sample data in real time during the sampling process; the sensors include but are not limited to volume sensors, speed sensors, acceleration sensors, position sensors and time sensors;

[0092] The sample data includes: sample volume, sample speed, acceleration of the sample needle, actual height of the sample needle to the target height of the container mouth, and actual time from the start of movement of the sample needle to the target height.

[0093] Specifically, in the fully automatic trace element automatic sampling system, a variety of sensors are installed to monitor the sample data in real time during the sampling process. These sensors include but are not limited to volume sensors for measuring the sample volume, speed sensors for measuring the sample speed, acceleration sensors for measuring the acceleration of the sample needle, position sensors for measuring the target height from the sample needle to the container port, and time sensors for recording the time of the sample process.

[0094] In one embodiment of the present invention, analyzing whether the sample adding needle is added to the sample tube according to the collected and processed video image and determining whether to issue an initial warning includes the following steps:

[0095] According to the collected and processed video images of the sample addition process, the images are converted into grayscale images or binary images, and the features of the sample addition needle and the sample tube in the images are extracted using image segmentation or edge detection technology, including: the outline, position and direction of the sample addition needle and the sample tube;

[0096] By extracting the contour, a contour point set of the sample needle and the sample tube is obtained; using the contour point set, the pixel coordinates of the center of mass of the sample needle and the center of mass of the sample tube are obtained;

[0097] By analyzing the formula:

[0098]

[0099] Get the centroid coordinates of the contour points , where N represents the number of contour points, Represents the coordinates of the contour point set;

[0100] By substituting the above formula, we can get the pixel coordinates of the centroid of the sample needle: , pixel coordinates of the centroid of the sample tube ;

[0101] By analyzing the formula:

[0102] The distance D between the pixel coordinate point of the mass center of the sample injection needle and the pixel coordinate point of the mass center of the sample tube is obtained; wherein, represents the pixel coordinates of the centroid of the pipette, The pixel coordinates of the centroid of the sample tube are shown;

[0103] By analyzing the formula:

[0104]

[0105] According to the above analytical formula, the inverse trigonometric function formula is calculated by analysis:

[0106]

[0107] Get the angle between the direction vector of the center of mass of the sample needle and the direction vector of the center of mass of the sample tube ;in,( , ) represents the direction vector of the center of mass of the sample injection needle, ( , ) represents the direction vector of the sample tube’s center of mass, represents the modulus of the direction vector of the center of mass of the sample needle, The modulus of the direction vector representing the centroid of the sample tube;

[0108] According to the distance between the pixel coordinate point of the centroid of the sample needle and the pixel coordinate point of the centroid of the sample tube and the angle between the direction vectors, a threshold is set to determine whether the sample needle is added into the sample tube;

[0109] when and When the sample is added, the sample needle is added into the sample tube, and the sample addition and detection are continued;

[0110] when or When the sample needle is not added into the sample tube, an initial warning is issued.

[0111] Specifically, the video image of the sampling process is collected in real time from the camera of the fully automatic trace element automatic sampling system. The collected color video image is converted into a grayscale image, or further converted into a binary image, so as to facilitate subsequent image segmentation and feature extraction. The sampling needle and the sample tube in the image are segmented by using image segmentation techniques including threshold segmentation, region growing, edge detection, etc. The contour point set of the sampling needle and the sample tube is obtained by the contour extraction algorithm including Canny edge detection. According to the contour point set, the centroid coordinate point of the sampling needle and the sample tube is calculated by the formula. The pixel distance between the centroid of the sampling needle and the centroid of the sample tube is calculated by the Euclidean distance formula; according to the coordinates of the centroid of the sampling needle and the centroid of the sample tube, their respective direction vectors are calculated. The angle value between the two direction vectors is calculated by the inverse trigonometric function formula. The distance threshold is set to 15 mm, and the angle threshold is set to 5 degrees; the distance threshold and the angle threshold are dynamically adjusted according to the actual application experiment. When D is less than 15 mm and θ is less than 5 degrees, it is judged that the sampling needle has been added to the sample tube, and the sampling detection continues. When D is greater than or equal to 15 mm, or θ is greater than or equal to 5 degrees, it is determined that the sample injection needle has not been added to the sample tube, and an initial warning is issued.

[0112] In one embodiment of the present invention, analyzing bubbles and calculating bubble influence coefficients according to the analysis results include the following steps:

[0113] When the sample needle is added to the sample tube, the pre-processed video image is used to detect bubbles using the deep learning model;

[0114] Collect and annotate bubble images in the sample tube, and input the annotated bubble images into the deep learning model for learning and training;

[0115] According to the video images collected and processed in real time by the fully automatic trace element automatic sampling system, whether there are bubbles in the video images of the sampling process;

[0116] When bubbles appear when the sample needle is inserted into the sample tube, the size and number of the bubbles are analyzed;

[0117] When no bubbles appear when the sample needle is inserted into the sample tube, continue to add samples for testing;

[0118] By analyzing the formula:

[0119] Get the pixel area A of the bubble, where S represents the total number of pixels occupied by the bubble in the image. represents the area occupied by the ath pixel;

[0120] By analyzing the formula:

[0121] Get the bubble influence coefficient ,in represents the pixel area of ​​the bubbles, represents the number of bubbles.

[0122] Specifically, the bubble images in the sample tube are collected and labeled, and the labeled bubble images are input into the deep learning model for learning and training. Through training with a large number of samples, the model can accurately identify bubbles in video images. According to the video images collected and processed in real time by the fully automatic trace element automatic sampling system, the trained deep learning model is used to detect whether there are bubbles in the video images of the sampling process. Output the detection result information, including: bubble size and quantity information. When bubbles are detected, analyze the size and number of bubbles. The bubble influence coefficient is obtained by analyzing the pixel area and number of bubbles. The area occupied by the ath pixel can be determined according to the image resolution. In the subsequent sampling process, continue to monitor and analyze the video image in real time to ensure the stability and reliability of the sampling process.

[0123] In one embodiment of the present invention, the sample deviation value is calculated based on the collected and pre-processed sample data analysis, including the following steps:

[0124] By analyzing the formula:

[0125] After finishing, we get:

[0126] Solving this quadratic equation using the root-finding formula yields:

[0127] The time t required for the sample needle to reach the target height is obtained, where represents the initial velocity of the sample injection needle, a represents the acceleration of the sample injection needle, and h represents the target height of the sample injection needle to the container opening;

[0128] The sample volume is calculated by the formula:

[0129] Get the sample volume ,in, represents the flow rate of the sample, Indicates the time required for the sample needle to reach the target height;

[0130] The deviation calculation formula of the time required is analyzed:

[0131] The deviation of the sample size is calculated by the formula:

[0132] By analyzing the sample deviation value calculation formula:

[0133] Get the sample deviation value P, where Indicates the deviation value of the time required for the sample needle to reach the target height; Indicates the deviation value of the sample amount; Indicates the actual time from the start of the sample needle movement to the target height. Indicates the actual amount of sample added.

[0134] Specifically, based on the real-time collection of sample data during the sampling process using sensors, the movement of the sampling needle can be approximated as uniformly accelerated linear motion, and the time t required for the sampling needle to reach the target height is calculated by using the quadratic equation root formula, combined with the initial velocity, acceleration and target height of the sampling needle. Since time cannot be a negative value, the positive root is selected as the solution. Based on the flow rate of the sample and the time required for the sampling needle to reach the target height, the actual amount of sample added is obtained by the sample amount calculation formula. The actual required time is compared with the theoretical required time to obtain the time deviation value. The actual amount of sample added is compared with the theoretical sample amount to obtain the sample amount deviation value. Taking into account the time deviation and sample amount deviation, the sample deviation value P is obtained by the sample deviation value calculation formula. This value comprehensively reflects the overall accuracy of time and sample amount control during the sampling process.

[0135] In one embodiment of the present invention, the sample is identified and classified according to the collected and processed video images, and the content and distribution of trace elements in the sample are analyzed according to the analysis results, including the following steps:

[0136] Pre-treat the samples, including grinding, digestion, dilution and volume adjustment, filtration and separation;

[0137] Annotate the collected pre-processed sample images and extract image features, including: color, texture and shape;

[0138] Use deep learning models to train and learn the extracted features;

[0139] The video images are collected in real time by the fully automatic trace element automatic sampling system, and the sample images are recognized by using the trained model; the samples are classified according to the recognition results, and the classification includes the name and content range of the samples;

[0140] According to the recognition results of the video images, the m samples are tested for trace elements; the content and distribution of trace elements in the i-th sample are determined using spectral analysis or mass spectrometry, where .

[0141] Specifically, the sample to be tested is ground, digested, diluted and fixed to facilitate subsequent filtration and separation, so that the sample reaches a state suitable for spectral analysis or mass spectrometry analysis. The pretreated sample is imaged and manually annotated, including the name of the sample, content range and other information. Then, key features such as color, texture and shape are extracted from the annotated images. These features will be used for subsequent deep learning model training. The convolutional neural network of the deep learning model is trained and learned using the extracted feature data. By continuously optimizing the model parameters, it can accurately identify different types of samples and their content ranges. According to the real-time acquisition of video images by the fully automatic trace element automatic sampling system, the trained deep learning model is used to identify the samples in the video images. The identification results include the name of the sample and the approximate content range. According to the identification results of the video image, trace element detection is performed on specific types of samples. The content and distribution of trace elements in the sample are determined using spectral analysis or mass spectrometry.

[0142] In one embodiment of the present invention, analyzing the coefficient of variation of trace elements in a sample according to the recognition result of the video image and the detection result of the trace elements includes the following steps:

[0143] By analyzing the mean value of the trace element content in the i-th sample, the formula is:

[0144]

[0145] By analyzing the standard deviation of the trace element content in the i-th sample, the formula is:

[0146]

[0147] Get the mean value of trace element content in the i-th sample and standard deviation ;in, represents the content of trace elements detected in the i-th sample, and m represents the total number of samples tested;

[0148] By analyzing the coefficient of variation of trace elements, the formula is calculated:

[0149] The coefficient of variation of trace elements β is obtained; where represents the mean value of the trace element content in the i-th sample, Represents the standard deviation of the trace element content in the i-th sample.

[0150] Specifically, according to the recognition results of the video image, determine the type of sample to be tested. Use spectral analysis or mass spectrometry technology to detect trace elements in the i-th sample, and record the content of trace elements in each sample. Calculate the mean of the trace element content in the i-th sample by formula. Among them, represents the content of the trace element detected in the i-th sample, and m represents the total number of samples tested. This step provides the overall level of the trace element content of the sample. Calculate the standard deviation of the trace element content in the i-th sample by formula. The standard deviation reflects the degree of dispersion of the trace element content between samples, that is, the volatility of the content. Using the results of the mean and standard deviation, calculate the coefficient of variation β of the trace element by formula. The coefficient of variation is the ratio of the standard deviation to the mean, which is used to measure the relative volatility of the data. In trace element analysis, the coefficient of variation can reflect the relative degree of difference in the trace element content between samples.

[0151] In one embodiment of the present invention, analyzing the pollutants in the sample and calculating the environmental pollution coefficient according to the analysis results include the following steps:

[0152] During the sample addition process, the sample is tested for pollutants; the environmental data of the sample is monitored in real time through a gas detector, a temperature sensor, a humidity sensor, and a light intensity sensor, and the collected environmental data is cleaned and normalized, and the environmental data includes: temperature, humidity, light intensity, and gas concentration;

[0153] After completing the sampling and pollutant detection, the environmental pollution coefficient is calculated by analyzing the formula:

[0154]

[0155] The environmental pollution coefficient R is obtained; where T represents temperature, S represents humidity, and L represents light intensity. Indicates gas concentration; represents the influence weight of temperature, represents the influence weight of humidity, Represents the influence weight of light intensity, Indicates the influence weight of gas concentration; is 0.35, is 0.3, is 0.15, is 0.2.

[0156] Specifically, during the sampling process, the samples are tested for pollutants, and the environmental data of the collected samples are monitored in real time through gas detectors, temperature sensors, humidity sensors, and light intensity sensors. The environmental pollution coefficient is obtained by calculating the product of temperature, humidity, light intensity, gas concentration and the corresponding impact weights. The impact weight of temperature is 0.35, the impact weight of humidity is 0.3, the impact weight of light intensity is 0.15, and the impact weight of gas concentration is 0.2; the impact weights are dynamically adjusted according to the pollution impact of different data, where the data can come from laboratory research, field monitoring or historical records.

[0157] In one embodiment of the present invention, the abnormal comprehensive index is calculated based on the bubble influence coefficient, the sample deviation value, the trace element variation coefficient and the environmental pollution coefficient, and a comprehensive evaluation and early warning are performed based on the calculation results, including the following steps:

[0158] By analyzing the abnormal comprehensive index calculation formula:

[0159]

[0160] Get the abnormal comprehensive index U; represents the bubble influence coefficient, Indicates the sample deviation value, represents the coefficient of variation of trace elements, Indicates the environmental pollution coefficient; represents the weight of the bubble influence coefficient, represents the weight of the sample deviation value, represents the weight of the coefficient of variation of trace elements, Indicates the weight of the environmental pollution coefficient;

[0161] According to the calculation results, when the abnormal comprehensive index When the abnormal comprehensive index Continue testing.

[0162] Specifically, the abnormal comprehensive index is calculated by comprehensive analysis of the calculated bubble influence coefficient, sample deviation value, trace element variation coefficient and environmental pollution coefficient. Based on the results of data analysis and expert consultation, a weight is dynamically assigned to each parameter. These weights should reflect the relative importance of each parameter in the abnormal comprehensive index. Indicates that the weight of the bubble influence coefficient is 0.3, The weight of the sample deviation value is 0.3. The weight of the coefficient of variation of trace elements is 0.25. The weight of the environmental pollution coefficient is 0.15. Finally, the abnormal comprehensive index is calculated and multiplied by the percentage so that the final result is in percentage form, making the result clearer. By setting the warning threshold of the abnormal comprehensive index, when it exceeds this threshold, the system will issue a warning signal, prompting the need to re-sample and test. If it is lower than the warning threshold, continue with regular testing.

[0163] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A fully automatic trace element automatic sampling system, characterized in that: Includes the following modules: Data acquisition module: real-time acquisition of sample data and video images during the sampling process; Data processing module: cleans, denoises and standardizes the collected data; denoises and enhances the collected video image data; Data analysis module: Analyzes whether the sample needle is added to the sample tube based on the collected and processed video images, and analyzes whether to issue an initial warning; Analyze bubbles and calculate bubble influence coefficients based on initial warning analysis results; Calculate sample deviation values ​​based on collected and processed sample data; identify and classify samples based on collected and processed video images, and analyze the content and distribution of trace elements in samples based on the results of identification and classification analysis; analyze the coefficient of variation of trace elements in samples based on image recognition results and trace element detection results; analyze pollutants in samples and calculate environmental pollution coefficients based on the results of pollutant analysis of samples; Comprehensive evaluation module: Based on the bubble influence coefficient, sample deviation value, trace element variation coefficient and environmental pollution coefficient, the abnormal comprehensive index is calculated and analyzed, and a comprehensive evaluation and early warning are carried out according to the calculation results; Wherein, analyzing bubbles and calculating bubble influence coefficients according to the initial warning analysis results include the following steps: When the sample needle is added to the sample tube, the pre-processed video image is used to detect bubbles using the deep learning model; Collect and annotate bubble images in the sample tube, and input the annotated bubble images into the deep learning model for learning and training; According to the video images collected and processed in real time by the fully automatic trace element automatic sampling system, whether there are bubbles in the video images of the sampling process; When bubbles appear when the sample needle is inserted into the sample tube, the size and number of the bubbles are analyzed; When no bubbles appear when the sample needle is inserted into the sample tube, continue to add samples for testing; Formula for calculating the pixel area of ​​a bubble: Get the pixel area A of the bubble, where S represents the total number of pixels occupied by the bubble in the image. represents the area occupied by the ath pixel; Formula for calculating bubble influence coefficient: Get the bubble influence coefficient ,in represents the pixel area of ​​the bubble, Indicates the number of bubbles; The sample deviation value is calculated based on the collected and preprocessed sample data, including the following steps: Formula for calculating the target height from the sample needle to the container mouth: After finishing, we get: Solving this quadratic equation using the root-finding formula yields: The time t required for the sample needle to reach the target height is obtained, where represents the initial velocity of the sample injection needle, a represents the acceleration of the sample injection needle, and h represents the target height of the sample injection needle to the container opening; The sample volume is calculated by the formula: Get the sample volume ,in, represents the flow rate of the sample, Indicates the time required for the sample needle to reach the target height; The deviation calculation formula of the required time is analyzed: The deviation of the sample size is calculated by the formula: By analyzing the sample deviation value calculation formula: Get the sample deviation value P, where Indicates the deviation value of the time required for the sample needle to reach the target height; Indicates the deviation value of the sample amount; Indicates the actual time from the start of the sample needle movement to the target height. Indicates the actual amount of sample added; The method of analyzing the coefficient of variation of trace elements in the sample according to the recognition result of the video image and the detection result of trace elements includes the following steps: By analyzing the mean value of the trace element content in the i-th sample, the formula is: By analyzing the standard deviation of the trace element content in the i-th sample, the formula is: Get the mean value of trace element content in the i-th sample and standard deviation ;in, represents the content of trace elements detected in the i-th sample, and m represents the total number of samples tested; By analyzing the coefficient of variation of trace elements, the formula is calculated: The coefficient of variation of trace elements β is obtained; where represents the mean value of the trace element content in the i-th sample, represents the standard deviation of the trace element content in the i-th sample; The steps of analyzing the pollutants in the sample and calculating the environmental pollution coefficient according to the pollutant analysis results of the sample include: During the sample addition process, the sample is tested for pollutants; the environmental data of the sample is monitored in real time through a gas detector, a temperature sensor, a humidity sensor, and a light intensity sensor, and the collected environmental data is cleaned and normalized, and the environmental data includes: temperature, humidity, light intensity, and gas concentration; After completing the sampling and pollutant detection, the environmental pollution coefficient is calculated by analyzing the formula: The environmental pollution coefficient R is obtained; where T represents temperature, S represents humidity, and L represents light intensity. Indicates gas concentration; represents the influence weight of temperature, represents the influence weight of humidity, Represents the influence weight of light intensity, Indicates the influence weight of gas concentration; is 0.35, is 0.3, is 0.15, is 0.

2.

2. A fully automatic trace element automatic sampling system according to claim 1, characterized in that: Real-time collection of sample data during the sample addition process includes the following steps: In the fully automatic trace element automatic sampling system, multiple sensors are installed to monitor the sample data in real time during the sampling process; the sensors include but are not limited to volume sensors, speed sensors, acceleration sensors, position sensors and time sensors; The sample data includes: sample volume, sample speed, acceleration of the sample needle, target height from the sample needle to the container opening, and actual time from the start of movement of the sample needle to the target height.

3. A fully automatic trace element automatic sampling system according to claim 1, characterized in that: According to the collected and processed video images, analyzing whether the sample injection needle is added into the sample tube and analyzing whether an initial warning is issued include the following steps: According to the collected and processed video images of the sample addition process, the images are converted into grayscale images or binary images, and the features of the sample addition needle and the sample tube in the images are extracted using image segmentation or edge detection technology, including: the outline, position and direction of the sample addition needle and the sample tube; By extracting the contour, a contour point set of the sample needle and the sample tube is obtained; using the contour point set, the pixel coordinates of the center of mass of the sample needle and the center of mass of the sample tube are obtained; Formula for calculating the centroid coordinates of contour points: Get the centroid coordinates of the contour points , where N represents the number of contour points, Represents the coordinates of the contour point set; By substituting the centroid coordinate formula of the contour point, the pixel coordinates of the centroid of the sample needle are obtained. , pixel coordinates of the centroid of the sample tube ; The formula for calculating the distance between the pixel coordinate point of the centroid of the sample needle and the pixel coordinate point of the centroid of the sample tube is: The distance D between the pixel coordinate point of the mass center of the sample needle and the pixel coordinate point of the mass center of the sample tube is obtained; wherein, represents the pixel coordinates of the centroid of the pipette, represents the pixel coordinates of the centroid of the sample tube; The formula for calculating the cosine of the angle is: According to the above angle cosine value formula, the inverse trigonometric function formula is calculated by analysis: Get the angle between the direction vector of the center of mass of the sample needle and the direction vector of the center of mass of the sample tube ;in,( , ) represents the direction vector of the center of mass of the sample injection needle, ( , ) represents the direction vector of the sample tube’s center of mass, represents the modulus of the direction vector of the center of mass of the sample needle, The modulus of the direction vector representing the centroid of the sample tube; According to the distance between the pixel coordinate point of the centroid of the sample needle and the pixel coordinate point of the centroid of the sample tube and the angle between the direction vectors, a threshold is set to determine whether the sample needle is added into the sample tube; when and When the sample is added, the sample needle is added into the sample tube, and the sample addition and detection are continued; when or When the sample needle is not added into the sample tube, an initial warning is issued.

4. A fully automatic trace element automatic sampling system according to claim 1, characterized in that: According to the collected and processed video images, the samples are identified and classified, and the content and distribution of trace elements in the samples are analyzed according to the identification and classification analysis results, including the following steps: Pre-treat the samples, including grinding, digestion, dilution and volume adjustment, filtration and separation; Annotate the collected pre-processed sample images and extract image features, including: color, texture and shape; Use deep learning models to train and learn the extracted features; The video images are collected in real time by the fully automatic trace element automatic sampling system, and the sample images are recognized by using the trained model; the samples are classified according to the recognition results, and the classification includes the name and content range of the samples; According to the recognition results of the video images, the m samples are tested for trace elements; the content and distribution of trace elements in the i-th sample are determined using spectral analysis or mass spectrometry, where .

5. A fully automatic trace element automatic sampling system according to claim 1, characterized in that: The abnormal comprehensive index is calculated based on the bubble influence coefficient, sample deviation value, trace element variation coefficient and environmental pollution coefficient, and a comprehensive evaluation and early warning is carried out based on the calculation results, including the following steps: By analyzing the abnormal comprehensive index calculation formula: Get the abnormal comprehensive index U; represents the bubble influence coefficient, Indicates the sample deviation value, represents the coefficient of variation of trace elements, Indicates the environmental pollution coefficient; represents the weight of the bubble influence coefficient, represents the weight of the sample deviation value, represents the weight of the coefficient of variation of trace elements, Indicates the weight of the environmental pollution coefficient; According to the calculation results of the abnormal comprehensive index, when the abnormal comprehensive index When the abnormal comprehensive index Continue testing.

Images