Detection error analysis method and system based on sampling analysis

Abstract

The invention provides a detection error analysis method and system based on sampling analysis, and relates to the technical field of alcohol detection.The detection error analysis method comprises the steps that a plurality of alcohol detectors are calibrated, and a plurality of alcohol detectors to be detected are obtained; acquiring basic test sample data of the test sample; placing the plurality of test sub-samples in a plurality of to-be-detected alcohol detectors, and obtaining a starting instruction; after receiving a starting instruction, detecting the concentration change of the plurality of test sub-samples to obtain a plurality of concentration detection data sets; constructing a concentration detection space; carrying out point group analysis on the plurality of scatter points, and fitting a plurality of target scatter points to obtain a sample detection concentration change curve; performing error coefficient identification on the sample concentration change curve and the sample detection concentration change curve to generate an error coefficient; and carrying out detection data error fitting. According to the alcohol detector, the technical problem that the detection precision of the alcohol detector in the prior art is reduced can be solved, and the technical effect of improving the alcohol detection accuracy is achieved.

Description

Technical Field

[0001] This application relates to the field of alcohol detection technology, and in particular to a detection error analysis method and system based on sampling analysis. Background Technology

[0002] A breathalyzer is an instrument used to measure alcohol concentration. It is widely used in drunk driving detection, industrial production, medical and other fields to help monitor alcohol intake and prevent drunk driving accidents.

[0003] Currently, in existing alcohol testing processes, changes in environmental conditions such as temperature and humidity, coupled with the volatility of alcohol, further increase the error of the testing instrument, affecting the repeatability of the results and causing systematic biases in the measurement results, leading to inaccurate results. When precise control of alcohol concentration is required, such as in drunk driving detection, deviations in the test results can lead to misjudgments and create safety hazards.

[0004] In summary, existing technologies suffer from the problem that alcohol is volatile, but error fitting is often not performed during alcohol detection, leading to decreased detection accuracy and further creating safety hazards. Summary of the Invention

[0005] The purpose of this application is to provide a detection error analysis method and system based on sampling analysis, in order to solve the technical problem in the prior art that the detection accuracy is reduced due to the volatility of alcohol, but error fitting is mostly not performed in the alcohol detection process, which further causes safety hazards.

[0006] In view of the above problems, this application provides a detection error analysis method and system based on sampling analysis.

[0007] Firstly, this application provides a detection error analysis method based on sampling analysis. This method is implemented through a detection error analysis system based on sampling analysis, comprising: acquiring multiple alcohol detectors from the same batch; calibrating each alcohol detector using a standard alcohol solution to obtain multiple calibrated alcohol detectors to be tested; preparing test samples using a standard ethanol solution and acquiring basic test sample data, including the sample ethanol concentration and a concentration change curve; equally dividing the test samples according to the number of alcohol detectors, placing the divided sub-samples into multiple alcohol detectors to be tested, and obtaining a start command; and receiving a start command from the multiple alcohol detectors under preset experimental conditions. After the start command is given, the concentration changes of multiple test sub-samples within a preset detection window are detected according to the first detection cycle, resulting in multiple concentration detection data sets. Each concentration detection data set has a detection time identifier. A concentration detection space is constructed, which includes multiple scatter points, each corresponding to a concentration detection data set in the multiple concentration detection data sets. Point cluster analysis is performed on the multiple scatter points in the concentration detection space to determine the fitting boundary. Multiple target scatter points located within the fitting boundary are fitted to obtain the sample detection concentration change curve. An error recognition network layer is used to identify the error coefficients of the sample concentration change curve and the sample detection concentration change curve, generating error coefficients for multiple alcohol detectors. Based on the error coefficients, the detection data errors of multiple alcohol detectors are fitted.

[0008] Secondly, this application also provides a detection error analysis system based on sampling analysis, used to execute the detection error analysis method based on sampling analysis as described in the first aspect. The system includes: a module for acquiring multiple alcohol detectors from the same batch, calibrating each detector using a standard alcohol solution, and obtaining calibrated alcohol detectors; a module for acquiring basic test sample data, used to prepare test samples using a standard ethanol solution and acquire basic test sample data, including ethanol concentration and concentration change curves; a module for acquiring a start command, used to equally divide the test samples according to the number of alcohol detectors, place the divided sub-samples into multiple alcohol detectors, and acquire a start command; and a module for acquiring concentration detection data sets, used to obtain multiple alcohol detectors under preset experimental conditions. After receiving the start command, the detector detects the concentration changes of multiple test sub-samples within a preset detection window according to the first detection cycle, obtaining multiple concentration detection data sets. Each concentration detection data set has a detection time identifier. The concentration detection space construction module is used to construct the concentration detection space, which includes multiple scattered points, each corresponding to a concentration detection data set in the multiple concentration detection data sets. The sample detection concentration change curve acquisition module is used to perform point cluster analysis on multiple scattered points in the concentration detection space, determine the fitting boundary, and fit multiple target scattered points located within the fitting boundary to obtain the sample detection concentration change curve. The error coefficient generation module is used to identify the error coefficients of the sample concentration change curve and the sample detection concentration change curve using an error recognition network layer, generating error coefficients for multiple alcohol detectors. The error fitting module is used to fit the detection data error of multiple alcohol detectors based on the error coefficients.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages: By acquiring multiple alcohol detectors from the same batch, each detector was calibrated using a standard alcohol solution, resulting in multiple calibrated alcohol detectors to be tested. Test samples were prepared using the standard ethanol solution, and basic test sample data was obtained, including sample ethanol concentration and concentration change curves. Based on the number of alcohol detectors, the test samples were equally divided, and the resulting sub-samples were placed into multiple alcohol detectors to be tested, receiving a start command. Under preset experimental conditions, upon receiving the start command, each alcohol detector detected the concentration changes of multiple sub-samples within a preset detection window according to a first detection cycle, obtaining multiple concentration detection data sets. Each concentration detection data point is identified by a detection time. A concentration detection space is constructed, comprising multiple scattered points, each corresponding to a single concentration detection data point from multiple concentration detection data sets. Point cluster analysis is performed on the multiple scattered points in the concentration detection space to determine the fitting boundary. Multiple target scattered points located within the fitting boundary are then fitted to obtain the sample detection concentration change curve. An error identification network layer is used to identify the error coefficients of the sample concentration change curve and the sample detection concentration change curve, generating error coefficients for multiple alcohol detectors. Based on the error coefficients, the detection data errors of multiple alcohol detectors are fitted. In other words, by simultaneously detecting with multiple alcohol detectors, the technical goal of enriching error analysis data is achieved, thereby improving the accuracy of the analysis.

[0010] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0012] Figure 1 This is a flowchart illustrating the detection error analysis method based on sampling analysis proposed in this application. Figure 2 This is a schematic diagram of the detection error analysis system based on sampling analysis in this application.

[0013] Explanation of reference numerals in the attached figures: The module includes: Module 11 for obtaining alcohol detection instrument data, Module 12 for obtaining basic test sample data, Module 13 for obtaining start command data, Module 14 for obtaining concentration detection data set data, Module 15 for constructing concentration detection space data, Module 16 for obtaining sample detection concentration change curve data, Module 17 for generating error coefficient data, and Module 18 for error fitting data. Detailed Implementation

[0014] This application provides a detection error analysis method and system based on sampling analysis, solving the technical problem in existing technologies where the volatility of alcohol, coupled with the lack of error fitting in most alcohol detection processes, leads to decreased detection accuracy and further poses safety hazards. It achieves the technical goal of providing abundant error analysis data, thereby improving analytical accuracy.

[0015] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them.

[0016] Example 1, please refer to the appendix. Figure 1 This application provides a detection error analysis method based on sampling analysis, wherein the detection error analysis method based on sampling analysis is applied to a detection error analysis system based on sampling analysis, and the specific steps of the detection error analysis method based on sampling analysis are as follows: Step 1: Obtain multiple alcohol detectors from the same batch, and calibrate each alcohol detector using a standard alcohol solution to obtain multiple calibrated alcohol detectors to be tested.

[0017] Specifically, several alcohol testing devices are selected from the same production batch. Then, a series of standard alcohol solutions of known concentrations are prepared. Each alcohol testing device is sequentially exposed to the standard alcohol solutions, and the response data of each device to solutions of different concentrations is recorded. Each device is precisely calibrated to ensure that the measurement results match the actual concentration of the standard alcohol solutions.

[0018] Step 2: Prepare test samples using standard ethanol solution and obtain basic test sample data, including the sample ethanol concentration and the sample concentration change curve.

[0019] Specifically, a certain amount of standard ethanol solution with a known concentration is prepared. Then, multiple samples are taken from the standard solution as test samples. For each test sample, the initial ethanol concentration is recorded as the baseline test sample data. Measurements are performed on each test sample to observe the change in ethanol concentration over time, including testing each sample at different time points and recording the ethanol concentration at each measurement. An ethanol concentration change curve is then plotted for each test sample, reflecting the trend of ethanol concentration change over time in the sample under experimental conditions. The concentration change curve of the test sample under natural evaporation conditions is also included.

[0020] Step 3: Divide the test samples equally according to the number of alcohol detectors, and place the divided sub-samples into multiple alcohol detectors to be tested, and obtain the start command.

[0021] Specifically, based on the number of available breathalyzers, the prepared test samples are evenly distributed. Each test sample is divided into several equal portions, ensuring that each breathalyzer device receives one sub-sample. The ethanol concentration of each sub-sample is ensured to be consistent, guaranteeing the accuracy and comparability of the test results. Subsequently, the sub-samples are placed into their respective breathalyzers. Then, the device is ready to receive the activation command and begin detecting the alcohol concentration.

[0022] Step 4: Under the preset experimental conditions, after receiving the start command, multiple alcohol detectors will detect the concentration changes of multiple test sub-samples within the preset detection window according to the first detection cycle, and obtain multiple concentration detection data sets. Each concentration detection data set has a detection time identifier.

[0023] Specifically, with experimental conditions pre-set and ready, when multiple alcohol detectors receive a start command, they begin testing their respective sub-samples according to a preset first testing cycle. The testing is conducted within a preset detection window to monitor changes in the ethanol concentration of the sub-samples during this time period. As the testing progresses, each alcohol detector records the concentration change data of the sub-samples, and each data point is marked with a detection time marker for subsequent analysis and comparison. The dataset constitutes a detailed record of the concentration change of each sub-sample within the preset detection window. Data analysis is used to assess the performance and accuracy of each alcohol detector and its behavior in detecting changes in ethanol concentration.

[0024] Step 5: Construct a concentration detection space, which includes multiple scatter points, each corresponding to a concentration detection data point from multiple concentration detection datasets.

[0025] Specifically, to analyze and visualize multiple concentration detection datasets, a concentration detection space is constructed. The concentration detection space is a mathematical or graphical representation that maps each concentration detection data point as a scatter plot in the space. The position of each scatter plot in the space corresponds to a specific concentration detection data point, typically determined based on the detection time marker and the corresponding ethanol concentration. In the concentration detection space, the horizontal axis can represent time, the vertical axis can represent ethanol concentration, and each scatter plot represents the concentration measurement at a specific time point. Therefore, as time changes, the distribution of multiple scatter plots in the space will form a trajectory, reflecting the concentration change of the test subsample over time.

[0026] Step 6: Perform point cluster analysis on multiple scatter points in the concentration detection space to determine the fitting boundary, and fit multiple target scatter points located within the fitting boundary to obtain the sample detection concentration change curve.

[0027] Specifically, within the constructed concentration detection space, point cluster analysis is performed on multiple scatter points to identify patterns or trends in their distribution, thereby determining a fitting boundary. The fitting boundary is set based on the distribution characteristics of the scatter points, including those representing concentration detection data while excluding outliers or noise. Once the fitting boundary is determined, the scatter points within it are considered target scatter points and used in the fitting process to obtain the distribution of the scatter points, thus yielding a sample detection concentration change curve.

[0028] Step 7: Use the error recognition network layer to identify the error coefficients of the sample concentration change curve and the sample detection concentration change curve, and generate the error coefficients of multiple alcohol detectors.

[0029] Specifically, an error identification network layer is used to evaluate and correct the measurement errors of alcohol detectors. This layer is an algorithmic component based on machine learning or deep learning techniques to identify and quantify errors in the data. The sample concentration change curve is used as a reference standard, representing the true or ideal change in the sample's ethanol concentration. The sample concentration change curve measured by the alcohol detector is then compared to the reference curve. The error identification network layer analyzes the differences between the two curves and identifies the error generated by each detector during the measurement process. An error coefficient is generated for each alcohol detector through the network layer. The error coefficient reflects the degree of deviation of the detector from the true value when measuring the sample ethanol concentration. With the error coefficient, the alcohol detector can be calibrated to improve its measurement accuracy and reliability.

[0030] Step 8: Fit the error of the detection data from multiple alcohol detectors based on the error coefficient.

[0031] Specifically, for each alcohol detector, the measured concentration data is combined with the corresponding error coefficient, including mathematical transformations of the data, such as multiplying by a correction factor or adding a correction value, to make the detector's measurement data closer to the true ethanol concentration value.

[0032] The sampling analysis-based detection error analysis method, when applied to a sampling analysis-based detection error analysis system, can achieve the technical goal of enriching error analysis data and improve the accuracy of the analysis.

[0033] Furthermore, this application also includes: A coordinate system for concentration detection space is constructed with detection time as the first coordinate axis and concentration detection data as the second coordinate axis. The detection times of multiple concentration detection data sets are normalized to obtain multiple standard concentration detection data sets. The multiple standard concentration detection data sets are sequentially input into the coordinate system to obtain multiple sets of coordinate points. The multiple sets of coordinate points are used as multiple scattered points, and the concentration detection space is generated based on the multiple scattered points and the coordinate system.

[0034] Specifically, to construct the concentration detection space, a coordinate system is defined, where the first coordinate axis represents the detection time and the second coordinate axis represents the concentration detection data. The coordinate system provides a framework for representing and analyzing the concentration detection data in space.

[0035] Next, the detection times in multiple concentration detection datasets are normalized. Normalization converts time data into a dimensionless form, typically to eliminate the influence of different time scales, allowing the data to be compared and analyzed on a uniform scale. After processing, multiple standard concentration detection datasets are obtained.

[0036] Then, the standard concentration detection data sets are sequentially input into the coordinate system. Each data point in each data set corresponds to a coordinate point in the coordinate system, which together constitute multiple scattered points in the concentration detection space.

[0037] Finally, by using scatter plots and a coordinate system, a concentration detection space is generated, which not only shows the change of concentration over time, but also allows for visualization analysis of the data, such as observing trends in concentration changes, detecting outliers, or evaluating performance differences between different detectors.

[0038] This provides a basis for obtaining concentration detection data and for further error analysis and fitting.

[0039] Furthermore, this application also includes: The sample concentration change curve is mapped onto the concentration detection space to obtain a first mapping curve; a straight line is fitted to the first mapping curve to generate a first fitted straight line, wherein the first fitted straight line has a first slope; using the first fitted straight line as the starting point, point cluster density is identified in the concentration detection space according to a preset first moving step size to determine the first point cluster analysis straight line; using the first point cluster analysis straight line as the starting point, point cluster density is identified in the concentration detection space according to a preset first angle step size to determine the target point cluster analysis straight line; a scatter distribution analysis is performed based on the target point cluster analysis straight line and the concentration detection space to determine the fitting boundary.

[0040] Specifically, the sample concentration change curve is mapped onto the concentration detection space to form the first mapping curve. The first mapping curve visually displays the trend of sample ethanol concentration over time in space.

[0041] Next, a straight line is fitted to the first mapping curve to generate the first fitted straight line. This process finds the best-fitting line to simplify the curve's variation and make it easier to analyze and understand. The first fitted straight line will have a first slope, representing the rate of concentration change.

[0042] Next, using the first fitted straight line as the starting point, and according to the preset first moving step size, the density of point clusters is identified in the concentration detection space. This includes moving along the first fitted straight line in the space with a certain step size and calculating the scatter density at each position, thereby determining a first point cluster analysis straight line.

[0043] Then, using the first point cluster analysis line as the starting point for rotation, and following a preset first angular step size, the point cluster density is identified in the concentration detection space. By rotating and calculating the scatter point density at different angles, a target point cluster analysis line is determined. Finally, based on the target point cluster analysis line and the concentration detection space, scatter point distribution analysis is performed to determine the fitting boundary. The fitting boundary is the boundary line that includes the dense region in the scatter point distribution, helping to define the scatter points that will be used in the final fitting process, thereby generating a sample detection concentration change curve, and thus effectively analyzing and interpreting the concentration detection data, improving the accuracy and reliability of data processing.

[0044] Furthermore, this application also includes: The number of multiple scattered points within a first tolerance bandwidth in the concentration detection space that are statistically related to the first fitted straight line is counted to obtain the first fitted scattered point quantity. The first left-side analytical scattered point quantity located in the left region of the concentration detection space and the first right-side analytical scattered point quantity located in the right region of the concentration detection space are counted, and the magnitudes of the first left-side analytical scattered point quantity and the first right-side analytical scattered point quantity are compared to determine the first moving direction. Based on the first moving direction, the first fitted straight line is moved in the concentration detection space according to the first moving step size to obtain the first moving fitted straight line, wherein the first moving fitted straight line includes the first moving fitted scattered point quantity. The first fitted straight line is iteratively updated based on the first moving fitted scattered point quantity. After multiple moves, until a preset condition is met, the moving fitted straight line corresponding to the maximum value of the fitted scattered point quantity during the move is taken as the first point group analytical straight line.

[0045] Specifically, in the concentration detection space, the analysis of the first fitted line involves counting the number of multiple scatter points, ensuring that the distance between the scatter points and the first fitted line is within a preset first tolerance bandwidth. This is called the first fitted scatter point quantity, representing the number of scatter points distributed near the first fitted line.

[0046] Next, the number of scatter points located to the left and right of the first fitted line in the concentration detection space is counted, i.e., the first left-side analysis scatter point quantity and the first right-side analysis scatter point quantity. Comparing these two quantities determines the first movement direction. If the left-side scatter point quantity is greater than the right-side quantity, the movement direction leans to the left; otherwise, it leans to the right. Based on the determined first movement direction, the first fitted line is moved within the concentration detection space according to a preset first movement step size. After each movement, a new first moving fitted line and the corresponding first moving fitted scatter point quantity are obtained.

[0047] Then, based on the first moving fitted scatter point quantity, the first fitted straight line is iteratively updated, including multiple moves and re-counting of the scatter point quantity, until a preset condition is met. The preset condition may be reaching the maximum fitted scatter point quantity, or reaching other predetermined stopping criteria.

[0048] Finally, after the iteration process is completed, the moving fitted line corresponding to the maximum value of the fitted scatter points during the movement is taken as the first point group analysis line, which reflects the main trend of the concentration detection data and is the most representative line in the scatter distribution.

[0049] By identifying and utilizing key information in concentration detection data, further data analysis and error correction can be performed.

[0050] Furthermore, this application also includes: The first point group analysis line is rotated according to a first angle step to obtain a first rotated analysis line, wherein the first rotated analysis line includes a first rotated scatter point quantity and the first point group analysis line includes a first point group fitted scatter point quantity. When the first point group fitted scatter point quantity is less than or equal to the first rotated scatter point quantity, the first rotated analysis line is updated to a stage point group analysis line. The stage point group analysis line is rotated according to a first angle step to obtain a second rotated analysis line, wherein the second rotated analysis line includes a second rotated scatter point quantity. The magnitudes of the second rotated scatter point quantity and the first rotated scatter point quantity are compared. Based on the comparison result, the stage point group analysis line is updated and iterated using the second rotated analysis line until the preset number of iterations is met. The stage point group analysis line corresponding to the maximum value of the rotated scatter point quantity during the iteration process is taken as the target point group analysis line.

[0051] Specifically, in the concentration detection space, the processing of the first point group analysis line involves rotating it according to a preset first angle step size. After each rotation, a new first rotated analysis line and a corresponding first rotated scatter point quantity are obtained, representing the number of scatter points distributed on the rotated line.

[0052] Simultaneously, compare the first point group fitted scatter value on the first point group analysis line with the first rotated scatter value on the first rotated analysis line. If the first point group fitted scatter value is less than or equal to the first rotated scatter value, it indicates that the rotated line better fits the scatter distribution, therefore the first rotated analysis line is updated to the stage point group analysis line.

[0053] Next, the new stage point group analysis line is rotated again according to the first angle step size to obtain the second rotated analysis line and the corresponding second rotated scatter points. The magnitudes of the second rotated scatter points and the first rotated scatter points are compared to determine whether to continue using the second rotated analysis line to update the stage point group analysis line.

[0054] Next, this process is repeated continuously, with each rotation and comparison updating the stage point group analysis line until the preset number of iterations is reached. During the iteration process, the stage point group analysis line corresponding to the maximum value of the scatter points after each rotation is recorded.

[0055] Finally, when the number of iterations meets the preset conditions, the stage point group analysis line corresponding to the maximum value of the rotation scatter point during the iteration process is determined as the final target point group analysis line, which represents the optimal fit of the concentration detection data. It is used to determine the fitting boundary and analyze the scatter point distribution, thereby improving the accuracy and reliability of data processing.

[0056] Furthermore, this application also includes: Based on the first extended bandwidth in the concentration detection space, the target point group analysis line is translated upwards and downwards respectively to determine the first-stage fitting upper boundary and the first-stage fitting lower boundary; the number of scattered points in the concentration detection space located between the first-stage fitting upper boundary and the first-stage fitting lower boundary is counted as the first fitting boundary scattered point quantity; based on the first extended bandwidth, the first-stage fitting upper boundary and the first-stage fitting lower boundary are translated upwards and downwards respectively to determine the second-stage fitting upper boundary and the second-stage fitting lower boundary, and the second fitting boundary scattered point quantity is determined; it is determined whether the difference between the first fitting boundary scattered point quantity and the second fitting boundary scattered point quantity is less than a preset difference. If so, the second-stage fitting upper boundary and the second-stage fitting lower boundary are translated to determine the third-stage fitting upper boundary and the third-stage fitting lower boundary, and scattered point distribution analysis is performed in the concentration detection space based on the third-stage fitting upper boundary and the third-stage fitting lower boundary; if not, the translation is stopped, and the second-stage fitting upper boundary and the second-stage fitting lower boundary are used as the fitting boundary.

[0057] Specifically, in the concentration detection space, the processing of the target point group analysis line first involves a translation operation according to a preset first expanded bandwidth. The target point group analysis line is translated upwards and downwards respectively to determine the upper boundary and lower boundary of the first-stage fitting.

[0058] Then, the number of scatter points located between the upper and lower boundaries of the first-stage fitting is counted, and this number is taken as the first fitting boundary scatter point count. The statistic represents the number of scatter points within the initial fitting boundary.

[0059] Next, based on the first extended bandwidth, the upper and lower bounds of the first-stage fitting are shifted upward and downward to determine the upper and lower bounds of the second-stage fitting, and the corresponding scatter points of the second fitting boundary are counted.

[0060] Then, the difference between the scatter points of the first and second fitting boundaries is compared with a preset difference. If the difference is less than the preset difference, it indicates that the change in the fitting boundary is not significant, and the translation operation can continue. The upper and lower boundaries of the second-stage fitting are translated to determine the upper and lower boundaries of the third-stage fitting, and scatter point distribution analysis is performed within the concentration detection space.

[0061] Next, if the difference is greater than or equal to the preset difference, it indicates a significant change in the fitting boundary. At this point, the translation operation is stopped, and the upper and lower boundaries of the second-stage fitting are used as the final fitting boundaries. The region of scatter point distribution is obtained for subsequent data analysis and error fitting, thereby determining the fitting boundary and improving the accuracy and reliability of data processing.

[0062] Furthermore, this application also includes: An error identification network layer is constructed, which includes a first feature extraction branch, a second feature extraction branch, and an error coefficient identification fully connected layer. The sample concentration change curve is input into the first feature extraction branch to obtain the sample concentration change feature. The sample detection concentration change curve is input into the second feature extraction branch to obtain the sample detection concentration change feature. The error coefficient identification fully connected layer is used to identify the error coefficients of the sample concentration change feature and the sample detection concentration change feature, and outputs the error coefficients.

[0063] Specifically, the main steps in constructing the error identification network layer include a first feature extraction branch, a second feature extraction branch, and a fully connected layer for error coefficient identification. First, the sample concentration change curve is input into the first feature extraction branch. This first feature extraction branch is a neural network component used to extract important features from the sample concentration change curve. It captures the inherent characteristics and trends of the curve, which is crucial for subsequent error identification.

[0064] Simultaneously, the sample concentration change curve is input into the second feature extraction branch. This branch is used by the neural network component to extract features. Key features are extracted from the sample concentration change curve, reflecting the characteristics and changes in the instrument's measurement results.

[0065] Finally, the fully connected layer is used to process the sample concentration change features and sample detection concentration change features obtained from the two feature extraction branches. The fully connected layer can learn the complex relationships between features and identify the error coefficient by analyzing the differences and correlations between the two sets of features. The error coefficient is the output of the network layer and quantifies the deviation between the detector's measurement results and the true values.

[0066] The error identification network layer can learn and identify error coefficients from concentration change data, providing an important basis for subsequent error correction and data analysis. The design and implementation of the network layer are key steps in improving the accuracy and reliability of alcohol detectors.

[0067] In summary, the detection error analysis method based on sampling analysis provided in this application has the following technical advantages: By acquiring multiple alcohol detectors from the same batch, each detector was calibrated using a standard alcohol solution, resulting in multiple calibrated alcohol detectors to be tested. Test samples were prepared using the standard ethanol solution, and basic test sample data was obtained, including sample ethanol concentration and concentration change curves. Based on the number of alcohol detectors, the test samples were equally divided, and the resulting sub-samples were placed into multiple alcohol detectors to be tested, receiving a start command. Under preset experimental conditions, upon receiving the start command, each alcohol detector detected the concentration changes of multiple sub-samples within a preset detection window according to a first detection cycle, obtaining multiple concentration detection data sets. Each concentration detection data point is identified by a detection time. A concentration detection space is constructed, comprising multiple scattered points, each corresponding to a single concentration detection data point from multiple concentration detection data sets. Point cluster analysis is performed on the multiple scattered points in the concentration detection space to determine the fitting boundary. Multiple target scattered points located within the fitting boundary are then fitted to obtain the sample detection concentration change curve. An error identification network layer is used to identify the error coefficients of the sample concentration change curve and the sample detection concentration change curve, generating error coefficients for multiple alcohol detectors. Based on the error coefficients, the detection data errors of multiple alcohol detectors are fitted. In other words, by simultaneously detecting with multiple alcohol detectors, the technical goal of enriching error analysis data is achieved, thereby improving the accuracy of the analysis.

[0068] Example 2: Based on the same inventive concept as the sampling analysis-based detection error analysis method in the previous examples, this application also provides a sampling analysis-based detection error analysis system. Please refer to the appendix. Figure 2 The detection error analysis system based on sampling analysis includes: The alcohol detector acquisition module 11 is used to acquire multiple alcohol detectors from the same batch, calibrate them using a standard alcohol solution, and obtain multiple calibrated alcohol detectors for testing. The basic test sample data acquisition module 12 is used to prepare test samples using a standard ethanol solution and acquire basic test sample data, including the sample ethanol concentration and concentration change curve. The start command acquisition module 13 is used to equally divide the test samples according to the number of alcohol detectors, place the divided sub-samples into multiple alcohol detectors for testing, and acquire a start command. The concentration detection data set acquisition module 14 is used to, under preset experimental conditions, after receiving the start command, perform multiple sub-samples on multiple alcohol detectors for testing within a preset detection window according to the first detection cycle. The system detects changes in concentration, obtaining multiple concentration detection data sets, each with a detection time identifier. A concentration detection space construction module 15 constructs a concentration detection space, comprising multiple scattered points, each corresponding to a single concentration detection data set. A sample detection concentration change curve acquisition module 16 performs point cluster analysis on the multiple scattered points in the concentration detection space, determines the fitting boundary, and fits multiple target scattered points within the fitting boundary to obtain the sample detection concentration change curve. An error coefficient generation module 17 uses an error recognition network layer to identify the error coefficients of the sample concentration change curve and the sample detection concentration change curve, generating error coefficients for multiple alcohol detectors. An error fitting module 18 fits the detection data errors of multiple alcohol detectors based on the error coefficients.

[0069] Furthermore, the detection error analysis system based on sampling analysis is also used for: A coordinate system for concentration detection space is constructed with detection time as the first coordinate axis and concentration detection data as the second coordinate axis. The detection times of multiple concentration detection data sets are normalized to obtain multiple standard concentration detection data sets. The multiple standard concentration detection data sets are sequentially input into the coordinate system to obtain multiple sets of coordinate points. The multiple sets of coordinate points are used as multiple scattered points, and the concentration detection space is generated based on the multiple scattered points and the coordinate system.

[0070] Furthermore, the detection error analysis system based on sampling analysis is also used for: The sample concentration change curve is mapped onto the concentration detection space to obtain a first mapping curve; a straight line is fitted to the first mapping curve to generate a first fitted straight line, wherein the first fitted straight line has a first slope; using the first fitted straight line as the starting point, point cluster density is identified in the concentration detection space according to a preset first moving step size to determine the first point cluster analysis straight line; using the first point cluster analysis straight line as the starting point, point cluster density is identified in the concentration detection space according to a preset first angle step size to determine the target point cluster analysis straight line; a scatter distribution analysis is performed based on the target point cluster analysis straight line and the concentration detection space to determine the fitting boundary.

[0071] Furthermore, the detection error analysis system based on sampling analysis is also used for: The number of multiple scattered points within a first tolerance bandwidth in the concentration detection space that are statistically related to the first fitted straight line is counted to obtain the first fitted scattered point quantity. The first left-side analytical scattered point quantity located in the left region of the concentration detection space and the first right-side analytical scattered point quantity located in the right region of the concentration detection space are counted, and the magnitudes of the first left-side analytical scattered point quantity and the first right-side analytical scattered point quantity are compared to determine the first moving direction. Based on the first moving direction, the first fitted straight line is moved in the concentration detection space according to the first moving step size to obtain the first moving fitted straight line, wherein the first moving fitted straight line includes the first moving fitted scattered point quantity. The first fitted straight line is iteratively updated based on the first moving fitted scattered point quantity. After multiple moves, until a preset condition is met, the moving fitted straight line corresponding to the maximum value of the fitted scattered point quantity during the move is taken as the first point group analytical straight line.

[0072] Furthermore, the detection error analysis system based on sampling analysis is also used for: The first point group analysis line is rotated according to a first angle step to obtain a first rotated analysis line, wherein the first rotated analysis line includes a first rotated scatter point quantity and the first point group analysis line includes a first point group fitted scatter point quantity. When the first point group fitted scatter point quantity is less than or equal to the first rotated scatter point quantity, the first rotated analysis line is updated to a stage point group analysis line. The stage point group analysis line is rotated according to a first angle step to obtain a second rotated analysis line, wherein the second rotated analysis line includes a second rotated scatter point quantity. The magnitudes of the second rotated scatter point quantity and the first rotated scatter point quantity are compared. Based on the comparison result, the stage point group analysis line is updated and iterated using the second rotated analysis line until the preset number of iterations is met. The stage point group analysis line corresponding to the maximum value of the rotated scatter point quantity during the iteration process is taken as the target point group analysis line.

[0073] Furthermore, the detection error analysis system based on sampling analysis is also used for: Based on the first extended bandwidth in the concentration detection space, the target point group analysis line is translated upwards and downwards respectively to determine the first-stage fitting upper boundary and the first-stage fitting lower boundary; the number of scattered points in the concentration detection space located between the first-stage fitting upper boundary and the first-stage fitting lower boundary is counted as the first fitting boundary scattered point quantity; based on the first extended bandwidth, the first-stage fitting upper boundary and the first-stage fitting lower boundary are translated upwards and downwards respectively to determine the second-stage fitting upper boundary and the second-stage fitting lower boundary, and the second fitting boundary scattered point quantity is determined; it is determined whether the difference between the first fitting boundary scattered point quantity and the second fitting boundary scattered point quantity is less than a preset difference. If so, the second-stage fitting upper boundary and the second-stage fitting lower boundary are translated to determine the third-stage fitting upper boundary and the third-stage fitting lower boundary, and scattered point distribution analysis is performed in the concentration detection space based on the third-stage fitting upper boundary and the third-stage fitting lower boundary; if not, the translation is stopped, and the second-stage fitting upper boundary and the second-stage fitting lower boundary are used as the fitting boundary.

[0074] Furthermore, the detection error analysis system based on sampling analysis is also used for: An error identification network layer is constructed, which includes a first feature extraction branch, a second feature extraction branch, and an error coefficient identification fully connected layer. The sample concentration change curve is input into the first feature extraction branch to obtain the sample concentration change feature. The sample detection concentration change curve is input into the second feature extraction branch to obtain the sample detection concentration change feature. The error coefficient identification fully connected layer is used to identify the error coefficients of the sample concentration change feature and the sample detection concentration change feature, and outputs the error coefficients.

[0075] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. The sampling analysis-based detection error analysis method and specific examples in Embodiment 1 are also applicable to the sampling analysis-based detection error analysis system of this embodiment. Through the foregoing detailed description of the sampling analysis-based detection error analysis method, those skilled in the art can clearly understand the sampling analysis-based detection error analysis system of this embodiment. Therefore, for the sake of brevity, it will not be described in detail here. As for the system disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and relevant parts can be referred to in the method section.

[0076] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0077] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application also intends to include such modifications and variations.

Claims

1. A detection error analysis method based on sampling analysis, characterized in that, include: Multiple alcohol detectors from the same batch were obtained, and each alcohol detector was calibrated using a standard alcohol solution to obtain multiple calibrated alcohol detectors to be tested. Test samples were prepared using a standard ethanol solution, and basic test sample data were obtained, including the ethanol concentration of the sample and the concentration change curve of the sample. Based on the number of multiple alcohol testing devices, the test samples are divided equally, and the divided sub-samples are placed in multiple alcohol testing devices to be tested, and a start command is obtained. Under preset experimental conditions, after receiving the start command, multiple alcohol detectors to be tested will detect the concentration changes of multiple test sub-samples within the preset detection window according to the first detection cycle, and obtain multiple concentration detection data sets, wherein each concentration detection data has a detection time identifier. Concentration detection space is constructed, which includes multiple scattered points, each of which corresponds to a concentration detection data in multiple concentration detection data sets; In the concentration detection space, point cluster analysis is performed on multiple scatter points to determine the fitting boundary. Then, multiple target scatter points located within the fitting boundary are fitted to obtain the sample detection concentration change curve. Error coefficients of sample concentration change curves and sample detection concentration change curves are identified using an error recognition network layer, generating error coefficients for multiple alcohol detectors. Error fitting of detection data from multiple alcohol testing devices was performed based on the error coefficient.

2. The detection error analysis method based on sampling analysis as described in claim 1, characterized in that, include: A coordinate system for concentration detection space is constructed with detection time as the first coordinate axis and concentration detection data as the second coordinate axis; The detection times of multiple concentration detection datasets are normalized to obtain multiple standard concentration detection datasets after processing. Multiple standard concentration detection data sets are sequentially input into a coordinate system to obtain multiple sets of coordinate points; Multiple sets of coordinate points are treated as multiple scattered points, and a concentration detection space is generated based on the multiple scattered points and the coordinate system.

3. The detection error analysis method based on sampling analysis as described in claim 1, characterized in that, include: The sample concentration change curve is mapped to the concentration detection space to obtain the first mapping curve; A straight line is fitted to the first mapping curve to generate a first fitted straight line, wherein the first fitted straight line has a first slope; Using the first fitted straight line as the starting point, point cluster density is identified in the concentration detection space according to the preset first moving step size to determine the first point cluster analysis straight line; Using the first point group analysis line as the starting point of rotation, point group density identification is performed in the concentration detection space according to the preset first angle step size to determine the target point group analysis line; Based on the target point group analysis line and the concentration detection space, scatter point distribution analysis is performed to determine the fitting boundary.

4. The detection error analysis method based on sampling analysis as described in claim 3, characterized in that, Using the first fitted straight line as the starting point, point cluster density is identified in the concentration detection space according to a preset first moving step size to determine the first point cluster analysis straight line, including: The number of multiple scatter points in the concentration detection space that are distanced to the first fitted line within the first tolerance bandwidth is counted to obtain the first fitted scatter point quantity; The first left-side analytical scatter point quantity located to the left of the first fitted line in the statistical concentration detection space and the first right-side analytical scatter point quantity located to the right of the first fitted line in the concentration detection space are compared to determine the first direction of movement. Based on the first moving direction, the first fitted straight line is moved in the concentration detection space according to the first moving step size to obtain the first moving fitted straight line, wherein the first moving fitted straight line includes the first moving fitted scatter point quantity; The first fitted line is iteratively updated based on the first moving fitted scatter point quantity. After multiple moves, until the preset conditions are met, the moving fitted line corresponding to the maximum value of the fitted scatter point quantity during the move is taken as the first point group analysis line.

5. The detection error analysis method based on sampling analysis as described in claim 4, characterized in that, include: The first point group analysis line is rotated according to the first angle step size to obtain the first rotated analysis line, wherein the first rotated analysis line includes the first rotated scatter point quantity, and the first point group analysis line includes the first point group fitted scatter point quantity. When the number of scatter points fitted by the first point group is less than or equal to the number of scatter points rotated by the first rotation, the first rotation analysis line is updated to the stage point group analysis line. The stage point group analysis line is rotated according to the first angle step size to obtain the second rotated analysis line, wherein the second rotated analysis line includes the second rotated scatter point quantity; Compare the magnitudes of the second rotating scatter point quantity and the first rotating scatter point quantity. Based on the comparison results, use the second rotating analysis line to update and iterate the stage point group analysis line until the preset number of iterations is met. The stage point group analysis line corresponding to the maximum value of the rotating scatter point quantity during the iteration process is taken as the target point group analysis line.

6. The detection error analysis method based on sampling analysis as described in claim 3, characterized in that, include: Based on the first extended bandwidth in the concentration detection space, the target point group analysis line is translated upwards and downwards respectively to determine the upper boundary and lower boundary of the first stage fitting; The number of scatter points in the concentration detection space located between the upper boundary and the lower boundary of the first-stage fitting is used as the number of scatter points on the first fitting boundary. Based on the first extended bandwidth, the upper and lower boundaries of the first-stage fitting are translated upward and downward respectively to determine the upper and lower boundaries of the second-stage fitting, and the scatter points of the second fitting boundary are determined. Determine whether the difference between the scatter points of the first fitting boundary and the scatter points of the second fitting boundary is less than a preset difference. If so, translate the upper boundary of the second stage fitting and the lower boundary of the second stage fitting to determine the upper boundary of the third stage fitting and the lower boundary of the third stage fitting, and perform scatter point distribution analysis in the concentration detection space based on the upper boundary of the third stage fitting and the lower boundary of the third stage fitting. If not, then stop the translation and use the upper boundary of the second-stage fitting and the lower boundary of the second-stage fitting as the fitting boundaries.

7. The detection error analysis method based on sampling analysis as described in claim 6, characterized in that, include: An error recognition network layer is constructed, which includes a first feature extraction branch, a second feature extraction branch, and a fully connected layer for error coefficient recognition. Input the sample concentration change curve into the first feature extraction branch to obtain the sample concentration change features; Input the sample detection concentration change curve into the second feature extraction branch to obtain the sample detection concentration change features; Error coefficients are used to identify the error coefficients of fully connected layers in terms of sample concentration change characteristics and sample detection concentration change characteristics, and the error coefficients are output.

8. A detection error analysis system based on sampling analysis, characterized in that, The steps for implementing the detection error analysis method based on sampling analysis according to any one of claims 1 to 7 include: The module for acquiring alcohol detectors to be tested is used to acquire multiple alcohol detectors from the same batch, calibrate the multiple alcohol detectors using a standard alcohol solution, and obtain multiple calibrated alcohol detectors to be tested. The basic test sample data acquisition module is used to prepare test samples using standard ethanol solution and acquire basic test sample data, which includes the sample ethanol concentration and the sample concentration change curve. The start command acquisition module is used to divide the test samples equally according to the number of multiple alcohol detectors, and place the divided test sub-samples into multiple alcohol detectors to be tested, and acquire the start command. The concentration detection data set acquisition module is used to detect the concentration changes of multiple test sub-samples within a preset detection window according to the first detection cycle after multiple alcohol detectors receive the start command under preset experimental conditions, and obtain multiple concentration detection data sets. Each concentration detection data set has a detection time identifier. The concentration detection space construction module is used to construct the concentration detection space, which includes multiple scattered points, each of which corresponds to a concentration detection data in multiple concentration detection data sets; The sample detection concentration change curve acquisition module is used to perform point cluster analysis on multiple scatter points in the concentration detection space, determine the fitting boundary, and fit multiple target scatter points located within the fitting boundary to obtain the sample detection concentration change curve. The error coefficient generation module is used to identify the error coefficients of the sample concentration change curve and the sample detection concentration change curve using the error recognition network layer, and generate the error coefficients of multiple alcohol detectors. The error fitting module is used to fit the error of detection data from multiple alcohol detectors based on the error coefficients.

Images