Multi-channel stable flow meter calibration test method and system

Abstract

The invention discloses a multi-channel stable flow meter calibration test method and system, and the method comprises the steps: obtaining flow data of a multi-channel flow meter under different working conditions to construct a data set, and carrying out the suspicious value detection and marking of the flow data of each channel through a Grubbs test model; performing principal component analysis on the processed data, and determining a linear combination relationship between principal components and original flow data variables; on the basis of the analysis result, a calibration model is built in combination with flow meter structure parameters, measurement error parameters and working condition parameters, the model is used for calibrating measurement data, and then the stability and consistency of the calibrated data are evaluated. The system comprises a data acquisition unit, a suspicious value detection unit, a principal component analysis unit, a calibration model construction unit, a calibration execution unit and an evaluation unit which are sequentially connected and cooperatively operate. The calibration accuracy and stability of the multi-channel flow meter can be effectively improved, and the calibration requirements under complex working conditions can be met.

Description

Technical Field

[0001] This invention relates to the field of multi-channel stable flow meter calibration, and more particularly to a calibration test method and system for multi-channel stable flow meters. Background Technology

[0002] In industrial process control, energy monitoring, and fluid pipeline transportation, multi-channel flow meters play a crucial role in simultaneously monitoring the flow rates of multiple branches. The stability of their measurement results directly affects the economic efficiency and safety of the system operation. With the increasing diversity of fluid media and the expanding range of dynamic operating conditions in application scenarios, the measurement data of each channel of a multi-channel flow meter are easily affected by factors such as differences in pipeline resistance, sensor response delays, and media disturbances, exhibiting complex correlations and fluctuations. Traditional methods based on independent calibration of a single channel are no longer sufficient to meet the high-precision measurement requirements in multi-parameter coupled environments. Therefore, there is an urgent need for technical solutions that can collaboratively process multi-channel data and achieve dynamic calibration.

[0003] Existing technologies for calibrating multi-channel flow meters have two significant limitations. Firstly, the abnormal data identification mechanism is inadequate. It only filters and removes measurements exceeding a fixed threshold within a fixed range, failing to establish targeted verification rules based on the statistical distribution characteristics of the flow data from each channel. This results in a large amount of valid data being mistakenly deleted or abnormal data remaining when there is medium pulsation or momentary sensor drift, compromising the integrity of the calibration data foundation. Secondly, the depth of multi-channel data correlation analysis is insufficient. It only corrects for deviations in the measurements of each channel by simply comparing them, without incorporating the flow meter's structural and operating parameters into a multivariate statistical analysis framework. This fails to isolate the influence of cross-interference factors on the measurement results of each channel, resulting in significant inconsistencies between channels in the calibrated flow data. Summary of the Invention

[0004] In order to overcome the shortcomings and deficiencies of the existing technology, the present invention provides a calibration and testing method and system for a multi-channel stable flow meter.

[0005] The technical solution adopted in this invention is a calibration and testing method for a multi-channel stable flow meter, comprising the following steps:

[0006] Step S1: Obtain the flow data of the multi-channel flow meter under different operating conditions and construct a flow dataset, which includes the flow measurement values ​​of different channels under various combinations of operating conditions;

[0007] Step S2: For the constructed traffic dataset, use the Grubbs test model to detect outliers in the traffic data of each channel. Based on the statistical judgment rules of the Grubbs test, identify and mark outlier data points that deviate from the mean by more than a set threshold.

[0008] Step S3: Perform principal component analysis on the traffic data of different channels after outlier processing, determine the coefficient vector of each principal component, and construct a linear combination relationship between the principal components and the original traffic data variables; specifically, first calculate the covariance matrix of traffic data of different channels, then solve the eigenvalues ​​and eigenvectors of the covariance matrix, sort the main eigenvectors according to the size of the eigenvalues, and determine the principal components.

[0009] Step S4: Based on the principal component analysis results, construct a flow meter calibration model. Using the principal components as independent variables, and combining the structural parameters, measurement error parameters, and operating parameters of different channel flow meters, establish a calibration model through a preset linear or nonlinear combination method to calibrate the flow meter measurement values.

[0010] Step S5: Use the calibration model to calibrate the measurement data of the multi-channel flow meter. Based on the current operating conditions of different channels, substitute the data into the calibration model to calculate the calibrated flow value.

[0011] Step S6: Evaluate the calibrated flow data, compare the stability and consistency indicators of the flow data before and after calibration, and determine whether the calibration effect meets the preset standard.

[0012] Further, in step S2, the formula for calculating the statistic Z used to identify outliers in the Grubbs test model is as follows: ,in, Indicates the The unit time flow rate measurement value of each channel. This represents the average of all flow measurements for that channel. This represents the standard deviation of the flow measurement values ​​for that channel; for the flow data of each channel, when the calculated... The value is greater than the critical value determined based on the significance level and sample size. At that time, the corresponding The outlier was identified as an anomaly. After marking the outlier, the outlier was further verified based on the historical fluctuation pattern of the channel flow data and the flow change trend at adjacent measurement times.

[0013] Furthermore, in step S3, when determining the principal component coefficient vector during principal component analysis, the constructed optimization model is as follows: The constraints are ,in, The first covariance matrix is ​​the first... 1 eigenvalue, A matrix composed of eigenvectors. The identity matrix is ​​used; the optimization model is used to obtain the combination of eigenvectors that maximizes the variance contribution of the principal components, thereby determining the principal component coefficient vector; simultaneously, considering the correlation differences between the traffic data of different channels, a channel correlation weight matrix is ​​introduced when calculating the covariance matrix. The covariance calculation between different channels is weighted and adjusted.

[0014] Further, in step S4, the constructed flow meter calibration model is as follows: ,in, The calibrated flow rate value. For the first Principal components, For the first The coefficients corresponding to each principal component For the first Each flow meter structure parameter, For the first The coefficients corresponding to each structural parameter For the first Each operating condition parameter For the first The coefficients corresponding to each working condition parameter; Number of principal components The number of parameters in the flow table structure. This represents the number of operating parameters.

[0015] Furthermore, in step S5, when calculating the calibrated flow value using the flow data from different channels and substituting it into the calibration model, the measurement error parameters for each channel are modeled separately, taking into account the differences in the characteristics of measurement errors from different channels, thus constructing a channel measurement error correction model. ,in, This is the measurement error correction value. The standard deviation of the flow rate measurement for this channel. This represents the average value of the flow rate measurements for that channel. For measuring time.

[0016] Furthermore, in step 56, when evaluating the stability of the calibrated flow data, a stability evaluation index is constructed. ,in, For the first The flow rate value after calibration at each time point To evaluate the number of measurements within a given time period and to assess the consistency of calibrated flow data, a consistency evaluation index is constructed. in, For the first Under the first working condition The calibrated flow rate values ​​for each channel For the first The average flow rate of each channel after calibration under all operating conditions. For the number of working conditions, The number of channels is used; based on the comparison results of stability evaluation indicators and consistency evaluation indicators with preset thresholds, it is determined whether the calibration effect meets the requirements.

[0017] Furthermore, step S3, principal component analysis, specifically includes the following sub-steps:

[0018] Step S3-1: Perform zero-mean processing on the flow data of different channels. Subtract the mean of the flow data of each channel from the flow data of that channel to make the mean of the processed data zero, thereby eliminating the DC component of the data, which is used for the calculation of the covariance matrix and the extraction of principal components.

[0019] Step S3-2: Construct the covariance matrix of traffic data from different channels. By calculating the covariance between traffic data from different channels, the degree of linear correlation between the data from different channels can be reflected.

[0020] Step S3-3: Perform eigenvalue decomposition on the covariance matrix to solve for the eigenvalues ​​and corresponding eigenvectors of the covariance matrix. The magnitude of the eigenvalues ​​reflects the degree of contribution of the corresponding principal components to the data variance.

[0021] Step S3-4: Sort the eigenvectors according to the size of the eigenvalues, select the first few eigenvectors with larger eigenvalues, and the matrix formed by these eigenvectors is the coefficient matrix for determining the principal components, which is used to construct the linear combination relationship between the principal components and the original flow data variables.

[0022] Furthermore, step S4, constructing the flow meter calibration model, specifically includes the following sub-steps:

[0023] Step S4-1: Analyze the structural parameters of flow meters in different channels, including pipe inner diameter, sensor type and installation location parameters, and determine the influence of structural parameters on flow measurement.

[0024] Step S4-2: Investigate the effects of different operating parameters, including fluid temperature, pressure, and viscosity, on flow measurement, and obtain the relationship between operating parameters and flow measurement error through experiments or theoretical analysis;

[0025] Step S4-3: Combine the principal components, flow meter structural parameters, and operating parameters in a reasonable way to determine the functional form of the calibration model. Through training and fitting a large amount of known flow data, the coefficients in the model are initially determined.

[0026] Step S4-4: Validate and optimize the initially constructed calibration model. Test the model using traffic data that was not used in training, and adjust the model coefficients based on the test results to improve the model's accuracy and adaptability.

[0027] Furthermore, step S5, calibrating the multi-channel flow meter measurement data using a calibration model, specifically includes the following sub-steps:

[0028] Step S5-1: Collect real-time operating information of flow meters in different channels, including fluid temperature, pressure, and flow range parameters;

[0029] Step S5-2: Update the operating parameters in the calibration model based on the current operating condition information so that the model can adapt to the flow calibration requirements under different operating conditions.

[0030] Step S5-3: After the flow data measured from different channels are preprocessed with zero mean, the updated operating parameters are substituted into the calibration model for calculation to obtain the preliminary calibrated flow value.

[0031] Step S5-4: Post-process the flow rate value after preliminary calibration, including truncation or rounding based on the accuracy class and measurement range of the flow meter, to obtain the final calibrated flow rate value output.

[0032] A multi-channel stable flow meter calibration and testing system, the system comprising:

[0033] The data acquisition unit is used to acquire the flow data of the multi-channel flow meter under different operating conditions and transmit the data to the data processing unit.

[0034] An outlier detection unit, connected to the data processing unit, performs outlier detection and labeling on different channel traffic data transmitted from the data processing unit based on the Grubbs test model.

[0035] The principal component analysis unit, connected to the data processing unit, performs principal component analysis on the flow data after outlier processing, determines the principal component coefficient vector, and transmits it to the calibration model construction unit.

[0036] The calibration model construction unit is connected to the principal component analysis unit, the flow meter parameter acquisition unit, and the operating condition parameter acquisition unit. It constructs the flow meter calibration model based on the principal components, flow meter structural parameters, and operating condition parameters.

[0037] The calibration execution unit is connected to the calibration model construction unit and the data acquisition unit. It uses the calibration model to calibrate the acquired flow meter measurement data and outputs the calibrated flow data.

[0038] The evaluation unit, connected to the calibration execution unit, evaluates the stability and consistency of the calibrated flow data and feeds the evaluation results back to the system control unit for adjusting the system's operating parameters.

[0039] Beneficial Effects: This invention proposes a calibration and testing method and system for multi-channel stable flow meters. In terms of outlier handling, it utilizes the Grubbs test model combined with the statistical characteristics of flow data from each channel to detect outliers, avoiding the problems of false deletion of valid data or retention of outliers caused by fixed threshold screening. Dynamic judgment rules improve the integrity of the data foundation. In multi-channel data correlation analysis, principal component analysis is used to uncover potential correlations in the flow data of each channel. A calibration model is constructed by combining flow meter structural parameters and operating parameters, overcoming the limitation of simple linear fitting in removing cross-interference. The linear combination relationship between principal components and original parameters enhances the consistency of data between channels. Simultaneously, the step-by-step detailed principal component analysis, calibration model construction, and calibration execution process achieves full-process collaborative optimization from data acquisition to result evaluation, solving the problem that traditional single-channel calibration is difficult to adapt to multi-parameter coupling scenarios. Through the orderly connection and collaborative operation of each unit, the accuracy and stability of multi-channel flow meter calibration under complex operating conditions are significantly improved, providing a systematic technical guarantee for the accuracy of multi-channel flow measurement. Attached Figure Description

[0040] Figure 1 This is a flowchart of the method of the present invention;

[0041] Figure 2 This is a diagram showing the system unit composition of the present invention. Detailed Implementation

[0042] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0043] like Figure 1 As shown, a calibration test method for a multi-channel stable flow meter includes the following steps:

[0044] Step S1: Obtain the flow data of the multi-channel flow meter under different operating conditions and construct a flow dataset, which includes the flow measurement values ​​of different channels under various combinations of operating conditions;

[0045] Specifically, step S1 is the foundation of the entire calibration test method, and its quality directly affects the effectiveness of all subsequent analyses and calibrations. The technical parameters involved include the flow measurement values ​​for each channel, which are closely related to the flow meter's measurement range and accuracy class. The settings for different operating conditions need to consider factors such as fluid temperature, pressure, and flow rate, as these factors directly affect the flow meter's measurement results. The constructed flow dataset needs to comprehensively reflect the performance of the multi-channel flow meter under various possible operating conditions, providing sufficient and representative data support for subsequent outlier detection, principal component analysis, and the construction of the calibration model. By acquiring a sufficient quantity and diversity of flow data, it can be ensured that subsequent steps can fully explore the potential patterns and characteristics in the data, thereby improving the accuracy and reliability of the calibration test.

[0046] In the specific implementation process, the first step is to determine the total number of channels in the multi-channel flow meter and clarify the measurement range and accuracy level of each channel. Then, based on the potential scenarios in the actual application, multiple operating condition combinations are set, each including specific temperature, pressure, and flow rate ranges. For example, the temperature can be set at multiple gradients between -20℃ and 100℃, the pressure can be set at different levels between 0.1MPa and 10MPa, and the flow rate can cover 10% to 100% of the flow meter's measurement range. Under each operating condition, each channel is measured multiple times consecutively. The number of measurements is determined based on the flow meter's response characteristics and data stability requirements, typically no less than 50 times. The measured flow values ​​from each channel are then categorized and organized according to channel number and operating condition combination to form a structured flow dataset. During data acquisition, it is crucial to ensure the normal operation of the measuring equipment to avoid data distortion due to equipment failure or external interference. Simultaneously, the specific operating parameters for each measurement are recorded to ensure the dataset clearly corresponds to different operating conditions, providing an accurate and complete data foundation for subsequent steps.

[0047] Step S2: For the constructed traffic dataset, use the Grubbs test model to detect outliers in the traffic data of each channel. Based on the statistical judgment rules of the Grubbs test, identify and mark outlier data points that deviate from the mean by more than a set threshold.

[0048] Specifically, step S2 primarily aims to remove outliers from the dataset, ensuring data reliability and providing a clean data source for subsequent principal component analysis and calibration model construction. The technical parameters involved include the flow measurement value, mean, standard deviation, and the Grubbs test statistic and set threshold for each channel. Outlier detection avoids interference from abnormal data in subsequent analysis results, ensuring that the extracted features and constructed model accurately reflect the actual measurement characteristics of the flow meter. This is crucial for improving the accuracy and stability of the calibration model, as the presence of outliers can distort the principal component analysis results, thereby affecting the accuracy of the calibration model.

[0049] In practice, the constructed traffic dataset is first split by channel, resulting in independent traffic data sequences for each channel. For each channel's traffic data sequence, the mean and standard deviation are calculated. The mean reflects the average measurement level of the channel under specific operating conditions, while the standard deviation reflects the dispersion of the measurement values. Then, according to the statistical calculation method of the Grubbs test model, a statistic is calculated for each data point. This statistic measures the deviation of the data point from the mean. Next, based on the number of measurement data and the set significance level, a threshold for outlier detection is determined. The statistic for each data point is compared with the threshold; when the statistic exceeds the threshold, the data point is marked as an outlier. During the marking process, the statistical judgment rules of the Grubbs test must be strictly followed to ensure the objectivity and consistency of the judgment results. Marked outliers are not deleted, but only marked, so that they can be further processed in subsequent steps in conjunction with other analysis results. In this way, while preserving the original data information, obvious data interference from outliers can be eliminated for subsequent analysis.

[0050] Step S3: Perform principal component analysis on the traffic data of different channels after outlier processing, determine the coefficient vector of each principal component, and construct a linear combination relationship between the principal components and the original traffic data variables; specifically, first calculate the covariance matrix of traffic data of different channels, then solve the eigenvalues ​​and eigenvectors of the covariance matrix, sort the main eigenvectors according to the size of the eigenvalues, and determine the principal components.

[0051] Specifically, step S3 uses principal component analysis (PCA) to reduce dimensionality, extracting principal components that reflect the main information of the multi-channel flow data, thus reducing the dimensionality and complexity of the data. The technical parameters involved include the covariance matrix, eigenvalues, eigenvectors, and the coefficient vectors of the principal components. PCA effectively eliminates the correlation between flow data from different channels, transforming multiple correlated flow variables into a few uncorrelated principal components. These principal components can centrally reflect most of the information in the original data. This not only simplifies the subsequent model calibration process but also avoids model instability caused by multicollinearity of dependent variables, improving the model's generalization ability.

[0052] During implementation, the traffic data for each channel, after outlier processing, is first organized to ensure a consistent and complete data format. Then, the covariance matrix between the traffic data of each channel is calculated. Each element in the covariance matrix represents the covariance between two channels, reflecting the degree of linear correlation between them. The calculation of the covariance matrix needs to be based on all valid data points to ensure that the results accurately reflect the relationship between the data of each channel. Next, eigenvalue decomposition is performed on the covariance matrix to obtain all eigenvalues ​​and corresponding eigenvectors. Eigenvalues ​​represent the information energy carried by the corresponding eigenvector; the larger the eigenvalue, the stronger the explanatory power of the principal component represented by the eigenvector for the original data. The eigenvectors are sorted according to the magnitude of the eigenvalues, and the first few eigenvectors with larger eigenvalues ​​are selected as principal eigenvectors. Typically, the first few eigenvectors with a cumulative contribution rate of eigenvalues ​​reaching a certain proportion (e.g., above 85%) are selected to ensure that the principal components can contain most of the information from the original data. Finally, based on the selected principal eigenvectors, the coefficient vectors of each principal component are determined, and a linear combination relationship between the principal components and the original flow data variables is constructed. Each principal component is a linear combination of the original flow data variables, and the combination coefficients are determined by the elements of the eigenvectors.

[0053] Step S4: Based on the principal component analysis results, construct a flow meter calibration model. Using the principal components as independent variables, and combining the structural parameters, measurement error parameters, and operating parameters of different channel flow meters, establish a calibration model through a preset linear or nonlinear combination method to calibrate the flow meter measurement values.

[0054] Specifically, step S4 aims to establish a mathematical model that accurately describes the relationship between flow meter measurements and various influencing factors, providing a basis for flow meter calibration. The technical parameters involved include principal components, flow meter structural parameters (such as pipe inner diameter, sensor installation location, channel length, etc.), measurement error parameters (such as systematic error, random error, etc.), and operating condition parameters (such as fluid temperature, pressure, density, etc.). By incorporating these parameters into the calibration model, the influence of various factors on flow measurement can be comprehensively considered, improving calibration accuracy. The quality of the calibration model directly determines the accuracy of the final calibration result; therefore, a reasonable combination method is needed to construct the model to ensure that it accurately fits the measurement data and has good generalization ability.

[0055] In practice, the first step is to organize the principal components obtained from principal component analysis and use them as the independent variables of the calibration model. Simultaneously, the structural parameters, measurement error parameters, and operating parameters under different conditions of each flow meter channel are collected. These parameters need to be obtained through actual measurements or by consulting equipment manuals to ensure their accuracy. Then, based on the characteristics of the principal components, structural parameters, measurement error parameters, and operating parameters, an appropriate linear or nonlinear combination method is selected to construct the calibration model. For example, multiple linear regression can be used to establish a linear relationship between the principal components and each parameter and the flow measurement value; alternatively, nonlinear combinations such as multinomial regression and exponential functions can be used to better fit complex nonlinear relationships, depending on the actual situation. During model construction, the coefficients of the model need to be determined. These coefficients are obtained by fitting known measurement data. During the fitting process, flow data after outlier processing and principal component analysis is used. This data is substituted into the model, and the coefficients are solved by minimizing the difference between the model's predicted values ​​and the actual measured values. When determining the coefficients, an appropriate optimization algorithm is required to ensure that the model accurately reflects the relationship between each parameter and the flow measurement value. The completed calibration model needs to be able to receive principal components, structural parameters, measurement error parameters, and operating condition parameters as inputs, and output calibrated flow values, thereby achieving effective calibration of the flow meter measurement values.

[0056] Step S5: Use the calibration model to calibrate the measurement data of the multi-channel flow meter. Based on the current operating conditions of different channels, substitute the data into the calibration model to calculate the calibrated flow value.

[0057] Specifically, step S5 applies the constructed calibration model to the actual measurement data. Its purpose is to eliminate the influence of various factors on flow measurement through model calculation, thereby obtaining a more accurate flow value. The technical parameters involved include real-time measurement data from the multi-channel flow meter, current operating condition information (such as temperature, pressure, and flow rate), and various parameters of the calibration model. By acquiring operating condition information in real time and substituting it into the calibration model, it is ensured that the calibration process can adapt to different operating conditions, improving the timeliness and accuracy of the calibration results. This step serves as a bridge connecting the calibration model with practical applications; its implementation effect directly relates to the correction effect of the flow meter's measurement values, and is of great significance for improving the measurement accuracy of multi-channel flow meters in actual operation.

[0058] In practice, the first step is to collect real-time measurement data from each channel of the multi-channel flow meter, including instantaneous flow rate and cumulative flow rate. Simultaneously, sensors acquire real-time operating condition information for each channel, such as fluid temperature, pressure, and density, ensuring the accuracy and real-time nature of this information. The collected measurement data and operating condition information are then formatted and preprocessed according to the calibration model's requirements, making them directly input into the model. Next, the preprocessed principal components (recalculated based on real-time measurement data), structural parameters, measurement error parameters, and current operating condition parameters are substituted into the constructed calibration model. The calibration model calculates based on the input parameters, obtaining the calibrated flow rate values ​​for each channel through internal linear or nonlinear combination operations. During the calculation process, it is crucial to ensure the accuracy of all model parameters, and the calculation process must strictly adhere to the model's mathematical expressions. After obtaining the calibrated flow rate values, a rationality check is performed to ensure they are within the flow meter's measurement range and conform to actual operating conditions. If any anomalies are found, the input parameters and model calculation process must be re-checked until a reasonable calibration result is obtained.

[0059] Step S6: Evaluate the calibrated flow data, compare the stability and consistency indicators of the flow data before and after calibration, and determine whether the calibration effect meets the preset standard.

[0060] Specifically, step S6 aims to verify the validity of the calibration results, ensuring that the calibrated flow data meets the accuracy and reliability requirements of practical applications. The technical parameters involved include flow data before and after calibration, stability indicators (such as data fluctuation amplitude and standard deviation), consistency indicators (such as deviations and relative errors between channel data), and preset evaluation criteria. Evaluation can identify problems and deficiencies in the calibration model, providing a basis for model optimization and improvement. If the calibration effect does not meet the preset criteria, it is necessary to return to the previous steps to re-process the data, build the model, or perform the calibration process again until the requirements are met. This step is a crucial step in ensuring the effectiveness of the entire calibration test method, guaranteeing that the final output calibration data has high quality.

[0061] In practice, the flow data before and after calibration is first collected and categorized by channel and operating condition for comparative analysis. Then, stability indices of the flow data before and after calibration are calculated, such as the standard deviation and range of the flow data over a certain period. By comparing the changes in these indices, the improvement effect of calibration on data stability is evaluated. Simultaneously, consistency indices of the flow data between channels are calculated, such as the deviation of each channel's data from the average value and the relative error, to analyze whether the consistency between the channel data has improved after calibration. The calculated stability and consistency indices are compared with preset evaluation standards, which need to be determined based on the application scenario and accuracy requirements of the flow meter, such as setting the maximum allowable value of the standard deviation and the range of the relative error. If all indices meet the preset standards, the calibration effect is considered qualified; if some indices do not meet the requirements, the reasons need to be analyzed, which may be due to unreasonable parameter settings in the calibration model or errors in the data acquisition process. Based on the analysis results, the corresponding steps are returned for adjustment and optimization, such as rebuilding the calibration model and re-acquiring data, until the calibration effect meets the preset standards.

[0062] Preferably, in step S2, the formula for calculating the statistic Z used to identify outliers in the Grubbs test model is: ,in, Indicates the The unit time flow rate measurement value of each channel. This represents the average of all flow measurements for that channel. This represents the standard deviation of the flow measurement values ​​for that channel; for the flow data of each channel, when the calculated... The value is greater than the critical value determined based on the significance level and sample size. At that time, the corresponding The outlier was identified as an anomaly. After marking the outlier, the outlier was further verified by analyzing the historical fluctuation patterns of the channel flow data and the flow change trends at adjacent measurement times, in order to improve the accuracy of the outlier identification.

[0063] Specifically, in step S2, when applying the Grubbs test model, the technical parameters involved include the flow measurement values, mean, standard deviation, and critical values ​​determined based on the sample size and significance level for each channel. The significance of this step lies in improving the accuracy of suspicious value detection by constructing reasonable statistical judgment rules, avoiding misjudgments or missed judgments caused by simple threshold judgments. In implementation, the flow dataset is first split by channel, and the mean and standard deviation of the data sequence for each channel are calculated. The mean reflects the average measurement level of the channel under specific operating conditions, and the standard deviation reflects the degree of data dispersion. Subsequently, according to the statistical calculation method of the Grubbs test, the statistic is calculated point by point to measure the deviation of the data point from the mean. Then, the critical value is determined by combining the sample size and significance level. The statistic of each data point is compared with the critical value, and suspicious values ​​exceeding the threshold are marked. Simultaneously, a secondary verification is performed by combining the historical fluctuation patterns of the channel flow data and the flow change trends of adjacent measurement times to ensure the objectivity and consistency of suspicious value judgment, providing a more reliable foundation for subsequent data processing.

[0064] Preferably, in step S3, when determining the principal component coefficient vector during principal component analysis, the constructed optimization model is as follows: The constraints are ,in, The first covariance matrix is ​​the first... 1 eigenvalue, A matrix composed of eigenvectors. The identity matrix is ​​used; the optimization model is used to obtain the combination of eigenvectors that maximizes the variance contribution of the principal components, thereby determining the principal component coefficient vector; simultaneously, considering the correlation differences between the traffic data of different channels, a channel correlation weight matrix is ​​introduced when calculating the covariance matrix. We perform weighted adjustments on the covariance calculations between different channels to more accurately reflect the intrinsic relationships between the data from different channels.

[0065] Specifically, the rationality of determining principal component coefficient vectors in principal component analysis is improved by constructing an optimization model. The technical parameters involved include the eigenvalues ​​of the covariance matrix, the matrix composed of eigenvectors, and the identity matrix. Its significance lies in ensuring that the selected principal components can reflect the variance information of the original data to the greatest extent by maximizing the sum of eigenvalues ​​and constraining the orthogonality of the eigenvector matrix, thereby enhancing the principal components' ability to extract correlation features from multi-channel traffic data. In implementation, the covariance matrix is ​​first calculated for the traffic data after handling suspicious values. The elements of this matrix reflect the degree of linear correlation between data from each channel. Next, the eigenvalues ​​and eigenvectors of the covariance matrix are solved, with the magnitude of the eigenvalues ​​reflecting the information contribution of the corresponding principal components. Then, an optimization model is introduced. Under the constraint of ensuring the orthogonality of the eigenvector matrix, the combination of eigenvectors that maximizes the sum of eigenvalues ​​is selected as the principal eigenvector. Simultaneously, considering the differences in correlation between data from different channels, a channel correlation weight matrix is ​​introduced to weight and adjust the covariance calculation, making the covariance matrix more accurately reflect the actual relationship between channels, thereby determining a better principal component coefficient vector and improving the effectiveness of principal component analysis.

[0066] Preferably, in step S4, the constructed flow meter calibration model is as follows: ,in, The calibrated flow rate value. For the first Principal components, For the first The coefficients corresponding to each principal component For the first Each flow meter structure parameter, For the first The coefficients corresponding to each structural parameter For the first Each operating condition parameter For the first The coefficients corresponding to each working condition parameter; Number of principal components The number of parameters in the flow table structure. The number of operating parameters; and in determining the model coefficients At that time, the least squares method combined with cross-validation was used to improve the fitting accuracy and generalization ability of the calibration model for flow data under different operating conditions.

[0067] Specifically, the flow meter calibration model constructed in step S4 integrates principal components, flow meter structural parameters, measurement error parameters, and operating condition parameters. The technical parameters involved include principal components, structural parameters (such as pipe inner diameter and sensor installation location), measurement error parameters (such as systematic and random errors), operating condition parameters (such as temperature and pressure), and the coefficients corresponding to each parameter in the model. The significance of this model lies in comprehensively considering multiple influencing factors, establishing a more comprehensive relationship between flow measurement values ​​and calibration values, and improving calibration accuracy. During implementation, the principal components obtained from principal component analysis are first organized as independent variables. Simultaneously, structural parameters, measurement error parameters, and operating condition parameters for each channel are collected. These parameters need to be obtained through actual measurements or equipment manuals to ensure accuracy. Then, based on the parameter characteristics, a combination of methods such as multiple linear regression is selected to construct the model. The model coefficients are determined using the least squares method combined with cross-validation to ensure that the model can accurately fit the flow data under different operating conditions. The constructed model receives principal components and various parameter inputs and outputs the calibrated flow value, achieving effective correction of the flow meter measurement value and enhancing the model's adaptability to complex operating conditions.

[0068] Preferably, in step S5, when calculating the calibrated flow value using the calibration model for flow data from different channels, the measurement error parameters of each channel are modeled separately, taking into account the differences in measurement error characteristics of different channels, to construct a channel measurement error correction model. ,in, This is the measurement error correction value. The standard deviation of the flow rate measurement for this channel. This represents the average value of the flow rate measurements for that channel. The measurement time is used; the input parameters of the calibration model are adjusted using this error correction model to further improve the accuracy of the calibrated flow rate value.

[0069] Specifically, in step S5, a channel measurement error correction model is constructed to address the differences in measurement error characteristics among each channel. The technical parameters involved include the standard deviation, mean, measurement time, and calculated measurement error correction value for each channel's flow measurement. The significance of this step lies in improving the accuracy of the calibrated flow value by individually modeling and correcting the measurement error of each channel, thus compensating for calibration deviations caused by differences in channel characteristics. In implementation, the characteristics of the measurement error for each channel are first analyzed, including the source of the error and its fluctuation patterns. Based on these characteristics, an error correction model is constructed. In the model, the standard deviation reflects the degree of data dispersion, the mean reflects the average measurement level, and the measurement time is used to capture the trend of error change over time. When calculating the calibration value using the calibration model, the standard deviation, mean, and measurement time of each channel are substituted into the error correction model to obtain the corresponding error correction value. This correction value is used to adjust the input parameters of the calibration model before calibration calculation. In this way, the calibration process more closely reflects the actual error situation of each channel, reducing measurement deviations caused by insufficient error correction and improving the accuracy of the final calibration result.

[0070] Preferably, in step 56, when evaluating the stability of the calibrated flow data, a stability evaluation index is constructed. ,in, For the first The flow rate value after calibration at each time point To evaluate the number of measurements within a given time period and to assess the consistency of calibrated flow data, a consistency evaluation index is constructed. in, For the first Under the first working condition The calibrated flow rate values ​​for each channel For the first The average flow rate of each channel after calibration under all operating conditions. For the number of working conditions, The number of channels is used; based on the comparison results of stability evaluation indicators and consistency evaluation indicators with preset thresholds, it is determined whether the calibration effect meets the requirements.

[0071] Specifically, in step S6, stability assessment indicators and consistency assessment indicators are constructed. The technical parameters involved include the calibrated flow rate value, the number of measurements within the assessment period, the calibration values ​​of each channel under different operating conditions, the number of operating conditions, and the number of channels. The significance lies in objectively evaluating the calibration effect through quantitative indicators, providing a clear basis for judging whether the calibration meets the requirements, and avoiding the uncertainty of subjective judgment. During implementation, calibrated flow rate data is first collected and categorized by time series, operating condition, and channel. When calculating the stability assessment indicator, a certain assessment period is selected, and the sum of the squares of the differences between calibration values ​​at adjacent times within that period is calculated, then divided by the number of measurements minus one. This indicator reflects the fluctuation of the calibrated flow rate value over time. When calculating the consistency assessment indicator, the mean of the calibration value for each channel under all operating conditions is first calculated, then the sum of the absolute values ​​of the calibration value for each channel under each operating condition and the mean is calculated, and finally divided by the product of the number of operating conditions and the number of channels. This indicator reflects the degree of uniformity of the calibration values ​​for each channel under different operating conditions. The two indicators are compared with preset thresholds, and the comparison results are used to determine whether the calibration effect meets the requirements, providing a quantitative reference for optimizing the calibration process.

[0072] Preferably, step S3, principal component analysis, specifically includes the following sub-steps:

[0073] Step S3-1: Perform zero-mean processing on the flow data of different channels. Subtract the mean of the flow data of each channel from the flow data of that channel to make the mean of the processed data zero, so as to eliminate the DC component of the data and facilitate the subsequent calculation of the covariance matrix and extraction of principal components.

[0074] Step S3-2: Construct the covariance matrix of traffic data from different channels. By calculating the covariance between traffic data from different channels, the degree of linear correlation between the data from different channels is reflected, providing a basis for subsequent solution of eigenvalues ​​and eigenvectors.

[0075] Step S3-3: Perform eigenvalue decomposition on the covariance matrix to solve for the eigenvalues ​​and corresponding eigenvectors of the covariance matrix. The magnitude of the eigenvalues ​​reflects the degree of contribution of the corresponding principal components to the data variance.

[0076] Step S3-4: Sort the eigenvectors according to the size of the eigenvalues, select the first few eigenvectors with larger eigenvalues, and the matrix formed by these eigenvectors is the coefficient matrix for determining the principal components, which is used to construct the linear combination relationship between the principal components and the original flow data variables.

[0077] Preferably, step S4, constructing the flow meter calibration model, specifically includes the following sub-steps:

[0078] Step S4-1: Analyze the structural parameters of flow meters in different channels, including pipe inner diameter, sensor type and installation location parameters, and determine the influence of these structural parameters on flow measurement to provide a basis for incorporating them into the calibration model;

[0079] Step S4-2: Investigate the effects of different operating parameters, including fluid temperature, pressure, and viscosity, on flow measurement. Obtain the relationship between operating parameters and flow measurement error through experiments or theoretical analysis so that compensation can be made in the calibration model.

[0080] Step S4-3: Combine the principal components, flow meter structural parameters, and operating parameters in a reasonable way to determine the functional form of the calibration model. Through training and fitting a large amount of known flow data, the coefficients in the model are initially determined.

[0081] Step S4-4: Validate and optimize the initially constructed calibration model. Test the model using traffic data that was not used in training, and adjust the model coefficients based on the test results to improve the model's accuracy and adaptability.

[0082] Preferably, step S5, calibrating the multi-channel flow meter measurement data using a calibration model, specifically includes the following sub-steps:

[0083] Step S5-1: Collect real-time operating information of flow meters in different channels, including fluid temperature, pressure, and flow range parameters, to ensure the accuracy of the input parameters for the calibration model;

[0084] Step S5-2: Update the operating parameters in the calibration model based on the current operating condition information so that the model can adapt to the flow calibration requirements under different operating conditions.

[0085] Step S5-3: After the flow data measured from different channels are preprocessed with zero mean, the updated operating parameters are substituted into the calibration model for calculation to obtain the preliminary calibrated flow value.

[0086] Step S5-4: Post-process the flow rate value after preliminary calibration, including truncation or rounding based on the accuracy class and measurement range of the flow meter, to obtain the final calibrated flow rate value output.

[0087] like Figure 2 As shown, a multi-channel stable flow meter calibration and testing system includes:

[0088] The data acquisition unit is used to acquire the flow data of the multi-channel flow meter under different operating conditions and transmit the data to the data processing unit.

[0089] An outlier detection unit, connected to the data processing unit, performs outlier detection and labeling on different channel traffic data transmitted from the data processing unit based on the Grubbs test model.

[0090] The principal component analysis unit, connected to the data processing unit, performs principal component analysis on the flow data after outlier processing, determines the principal component coefficient vector, and transmits it to the calibration model construction unit.

[0091] The calibration model construction unit is connected to the principal component analysis unit, the flow meter parameter acquisition unit, and the operating condition parameter acquisition unit. It constructs the flow meter calibration model based on the principal components, flow meter structural parameters, and operating condition parameters.

[0092] The calibration execution unit is connected to the calibration model construction unit and the data acquisition unit. It uses the calibration model to calibrate the acquired flow meter measurement data and outputs the calibrated flow data.

[0093] The evaluation unit, connected to the calibration execution unit, evaluates the stability and consistency of the calibrated flow data and feeds the evaluation results back to the system control unit for adjusting the system's operating parameters.

[0094] A calibration and testing method and system for multi-channel stable flow meters is proposed. By introducing the Grubbs test model, dynamic judgment rules are formulated based on the statistical distribution characteristics of flow data in each channel, replacing the traditional fixed threshold screening method. During the testing process, by combining the dispersion and distribution patterns of flow data in each channel, outliers deviating from the normal range are accurately identified and marked. This avoids the accidental deletion of valid data due to simple threshold limitations, and also prevents the retention of unidentified outliers. Therefore, it overcomes the problem of impaired data integrity caused by inadequate outlier processing in previous technologies, providing more reliable raw data support for subsequent calibration processes.

[0095] In terms of multi-channel data correlation analysis and calibration model construction, this method and system deeply mine the potential correlations of flow data in each channel through principal component analysis, incorporating flow meter structural parameters and operating parameters into a multivariate statistical analysis framework. The constructed calibration model no longer relies on simple linear fitting, but effectively isolates the influence of cross-interference factors on the measurement results of each channel through the linear combination relationship between principal components and original parameters. This approach overcomes the limitations of insufficient depth in multi-channel data correlation analysis in traditional techniques, enhances the consistency of data between channels, enables the calibration model to better adapt to multi-parameter coupling scenarios under complex operating conditions, and significantly improves the reliability of calibration results.

[0096] The method and system's end-to-end collaborative optimization design is another outstanding advantage. The step-by-step detailed principal component analysis, calibration model construction, and calibration execution process achieve a seamless connection from data acquisition to result evaluation. Each unit operates collaboratively through clear connections, ensuring close coordination at every stage and forming an organic whole. This design solves the problem of traditional single-channel calibration's inability to adapt to multi-parameter coupling scenarios, improving the accuracy and stability of multi-channel flow meter calibration under complex operating conditions. Through a systematic technical solution, it comprehensively overcomes the shortcomings of background technologies, providing a strong guarantee for the accuracy of multi-channel flow measurement.

[0097] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," "link," and "fix" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0098] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various equivalent changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A calibration and testing method for a multi-channel stable flow meter, characterized in that, Includes the following steps: Step S1: Obtain the flow data of the multi-channel flow meter under different operating conditions and construct a flow dataset, which includes the flow measurement values ​​of different channels under various combinations of operating conditions; Step S2: For the constructed traffic dataset, use the Grubbs test model to detect outliers in the traffic data of each channel. Based on the statistical judgment rules of the Grubbs test, identify and mark outlier data points that deviate from the mean by more than a set threshold. Step S3: Perform principal component analysis on the traffic data of different channels after outlier processing, determine the coefficient vector of each principal component, and construct a linear combination relationship between the principal components and the original traffic data variables; specifically, first calculate the covariance matrix of traffic data of different channels, then solve the eigenvalues ​​and eigenvectors of the covariance matrix, sort the main eigenvectors according to the size of the eigenvalues, and determine the principal components. Step S4: Based on the principal component analysis results, construct a flow meter calibration model. Using the principal components as independent variables, and combining the structural parameters, measurement error parameters, and operating parameters of different channel flow meters, establish a calibration model through a preset linear or nonlinear combination method to calibrate the flow meter measurement values. Step S5: Use the calibration model to calibrate the measurement data of the multi-channel flow meter. Based on the current operating conditions of different channels, substitute the data into the calibration model to calculate the calibrated flow value. Step S6: Evaluate the calibrated flow data, compare the stability and consistency indicators of the flow data before and after calibration, and determine whether the calibration effect meets the preset standard.

2. The calibration and testing method for a multi-channel stable flow meter according to claim 1, characterized in that, In step S2, the formula for calculating the statistic Z used to identify outliers in the Grubbs test model is as follows: ,in, Indicates the The unit time flow rate measurement value of each channel. This represents the average of all flow measurements for that channel. This represents the standard deviation of the flow measurement values ​​for that channel; for the flow data of each channel, when the calculated... The value is greater than the critical value determined based on the significance level and sample size. At that time, the corresponding The outlier was identified as an anomaly. After marking the outlier, the outlier was further verified based on the historical fluctuation pattern of the channel flow data and the flow change trend at adjacent measurement times.

3. The calibration and testing method for a multi-channel stable flow meter according to claim 1, characterized in that, In step S3, when determining the principal component coefficient vector during principal component analysis, the constructed optimization model is as follows: The constraints are ,in, The first covariance matrix is ​​the first... 1 eigenvalue, A matrix composed of eigenvectors. The identity matrix is ​​used; the optimization model is used to obtain the combination of eigenvectors that maximizes the variance contribution of the principal components, thereby determining the principal component coefficient vector; simultaneously, considering the correlation differences between the traffic data of different channels, a channel correlation weight matrix is ​​introduced when calculating the covariance matrix. The covariance calculation between different channels is weighted and adjusted.

4. The calibration and testing method for a multi-channel stable flow meter according to claim 1, characterized in that, In step S4, the constructed flow meter calibration model is as follows: ,in, The calibrated flow rate value. For the first Principal components, For the first The coefficients corresponding to each principal component For the first Each flow meter structure parameter, For the first The coefficients corresponding to each structural parameter For the first Each operating condition parameter For the first The coefficients corresponding to each working condition parameter; Number of principal components The number of parameters in the flow table structure. This represents the number of operating parameters.

5. The calibration and testing method for a multi-channel stable flow meter according to claim 1, characterized in that, In step S5, when calculating the calibrated flow rate value by substituting the flow rate data from different channels into the calibration model, the measurement error parameters of each channel are modeled separately based on the differences in the characteristics of measurement errors of different channels, thus constructing a channel measurement error correction model. ,in, This is the measurement error correction value. The standard deviation of the flow rate measurement for this channel. This represents the average value of the flow rate measurements for that channel. For measuring time.

6. The calibration and testing method for a multi-channel stable flow meter according to claim 1, characterized in that, In step 56, when evaluating the stability of the calibrated flow data, a stability evaluation index is constructed. ,in, For the first The flow rate value after calibration at each time point To evaluate the number of measurements within a given time period and to assess the consistency of calibrated flow data, a consistency evaluation index is constructed. in, For the first Under the first working condition The calibrated flow rate values ​​for each channel For the first The average flow rate of each channel after calibration under all operating conditions. For the number of working conditions, The number of channels is used; based on the comparison results of stability evaluation indicators and consistency evaluation indicators with preset thresholds, it is determined whether the calibration effect meets the requirements.

7. The calibration and testing method for a multi-channel stable flow meter according to claim 1, characterized in that, Step S3, principal component analysis, specifically includes the following sub-steps: Step S3-1: Perform zero-mean processing on the flow data of different channels. Subtract the mean of the flow data of each channel from the flow data of that channel to make the mean of the processed data zero, thereby eliminating the DC component of the data, which is used for the calculation of the covariance matrix and the extraction of principal components. Step S3-2: Construct the covariance matrix of traffic data from different channels. By calculating the covariance between traffic data from different channels, the degree of linear correlation between the data from different channels can be reflected. Step S3-3: Perform eigenvalue decomposition on the covariance matrix to solve for the eigenvalues ​​and corresponding eigenvectors of the covariance matrix. The magnitude of the eigenvalues ​​reflects the degree of contribution of the corresponding principal components to the data variance. Step S3-4: Sort the eigenvectors according to the size of the eigenvalues, select the first few eigenvectors with larger eigenvalues, and the matrix formed by these eigenvectors is the coefficient matrix for determining the principal components, which is used to construct the linear combination relationship between the principal components and the original flow data variables.

8. The calibration and testing method for a multi-channel stable flow meter according to claim 1, characterized in that, Step S4, constructing the flow meter calibration model, specifically includes the following sub-steps: Step S4-1: Analyze the structural parameters of flow meters in different channels, including pipe inner diameter, sensor type and installation location parameters, and determine the influence of structural parameters on flow measurement. Step S4-2: Investigate the effects of different operating parameters, including fluid temperature, pressure, and viscosity, on flow measurement, and obtain the relationship between operating parameters and flow measurement error through experiments or theoretical analysis; Step S4-3: Combine the principal components, flow meter structural parameters, and operating parameters in a reasonable way to determine the functional form of the calibration model. Through training and fitting a large amount of known flow data, the coefficients in the model are initially determined. Step S4-4: Validate and optimize the initially constructed calibration model. Test the model using traffic data that was not used in training, and adjust the model coefficients based on the test results to improve the model's accuracy and adaptability.

9. The calibration and testing method for a multi-channel stable flow meter according to claim 1, characterized in that, Step S5, which involves calibrating the multi-channel flow meter measurement data using a calibration model, specifically includes the following sub-steps: Step S5-1: Collect real-time operating information of flow meters in different channels, including fluid temperature, pressure, and flow range parameters; Step S5-2: Update the operating parameters in the calibration model based on the current operating condition information so that the model can adapt to the flow calibration requirements under different operating conditions. Step S5-3: After the flow data measured from different channels are preprocessed with zero mean, the updated operating parameters are substituted into the calibration model for calculation to obtain the preliminary calibrated flow value. Step S5-4: Post-process the flow rate value after preliminary calibration, including truncation or rounding based on the accuracy class and measurement range of the flow meter, to obtain the final calibrated flow rate value output.

10. A calibration and testing system for a multi-channel stable flow meter, characterized in that, include: The data acquisition unit is used to acquire the flow data of the multi-channel flow meter under different operating conditions and transmit the data to the data processing unit. An outlier detection unit, connected to the data processing unit, performs outlier detection and labeling on different channel traffic data transmitted from the data processing unit based on the Grubbs test model. The principal component analysis unit, connected to the data processing unit, performs principal component analysis on the flow data after outlier processing, determines the principal component coefficient vector, and transmits it to the calibration model construction unit. The calibration model construction unit is connected to the principal component analysis unit, the flow meter parameter acquisition unit, and the operating condition parameter acquisition unit. It constructs the flow meter calibration model based on the principal components, flow meter structural parameters, and operating condition parameters. The calibration execution unit is connected to the calibration model construction unit and the data acquisition unit. It uses the calibration model to calibrate the acquired flow meter measurement data and outputs the calibrated flow data. The evaluation unit, connected to the calibration execution unit, evaluates the stability and consistency of the calibrated flow data and feeds the evaluation results back to the system control unit for adjusting the system's operating parameters.

Images