A storage tank metering data processing method, computer equipment and storage medium

Abstract

The application discloses a kind of storage tank metering data processing method, computer equipment and storage medium, it is related to metrological modeling technical field, including, acquisition multi-physical field operating parameter data, carry out filtering, interpolation and abnormal rejection, obtain multidimensional original data set;Based on the data set, establish digital twin model and calculate initial parameter set;Collect real-time observation data, execute residual optimization inversion to obtain correction parameter set;Uncertainty analysis and confidence weighted fusion are carried out using correction parameter set, to obtain fused metrological parameter set;Multi-objective cost function is constructed and updated weight using exponential gradient evolution algorithm, to obtain feedforward control intensity;Combined with correction parameter set and feedforward control intensity, cross-tank robust aggregation and gate determination are carried out, to obtain new digital twin model parameters.The application realizes the adaptive fusion of multi-source metrological information and model closed-loop evolution by confidence weighted fusion and exponential gradient evolution weight optimization mechanism and executes cross-tank robust aggregation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of metering modeling, and in particular to a tank metering data processing method, a computer device and a storage medium. BACKGROUND

[0002] With the acceleration of digital transformation of petrochemical, chemical and energy industries, tank metering technology gradually develops from traditional static liquid level measurement to multi-source sensing, dynamic modeling and intelligent calibration. In recent years, the use of liquid level, temperature, pressure and structural strain and other multi-physical field parameters to realize accurate estimation of tank volume and mass has become a research hotspot. In order to improve the measurement accuracy, the industry has tried to introduce Internet of Things sensor networks, thermal compensation models and cloud data analysis platforms to realize online monitoring and metering correction of tank state.

[0003] The existing tank metering method mainly uses a single tank and a single sensing signal as the core of the data processing mode, lacks system modeling of the coupling relationship of liquid level, density, temperature and structural strain and other multi-physical fields, and is difficult to reflect the real volume change of the tank under complex thermal and mechanical action. At the same time, the global consistency constraint of the metering results across the tanks has not been established, and it does not have the ability of self-evolution optimization, and cannot realize dynamic calibration and collaborative correction among multiple tanks. SUMMARY

[0004] In view of the above existing problems, the present application is proposed.

[0005] Therefore, the present application provides a tank metering data processing method to solve the problems of not establishing a multi-physical field coupling model and being unable to realize dynamic calibration and collaborative correction among multiple tanks in the prior art.

[0006] To solve the above technical problems, the present application provides the following technical solutions:

[0007] In a first aspect, the present application provides a tank metering data processing method, which comprises collecting multi-physical field operating parameter data, performing noise filtering, time interpolation and abnormality rejection, obtaining a multi-physical field state vector, using a spatial inverse distance weighted algorithm to correct and fuse the multi-physical field state vector in space and time, and obtaining a multi-dimensional original data set;

[0008] Based on the multi-dimensional original data set, a digital twin model is established, an initial parameter set is calculated, historical calibration data is read, a least squares method is used to fit the parameters of the digital twin model, the error between the output of the digital twin model and the historical calibration data is minimized, and the initial parameter set is obtained;

[0009] Real-time observation data is collected, the initial parameter set and the real-time observation data are input into the digital twin model, the residual error is calculated, proximal optimization inversion is performed, and a corrected parameter set is obtained;

[0010] The uncertainty propagation analysis is performed on the multi-physical field operation parameter data by the correction parameter set, the volume and mass uncertainties are calculated, the uncertainty is weighted and fused by a confidence weighted fusion algorithm, the fused volume and fused mass are calculated, the comprehensive uncertainty evaluation is performed, and the fused metrological parameter set is obtained;

[0011] According to the fused metrological parameter set, a multi-objective cost function is constructed, weight updating is performed through an exponential gradient evolution algorithm, and feedforward control strength is obtained.

[0012] Based on the correction parameter set and the feedforward control strength, cross-tank robust aggregation is performed to obtain global correction parameters, and the new digital twin model parameters are obtained through consistency and uncertainty gating.

[0013] As a preferred scheme of the storage tank metrological data processing method, the multi-physical field operation parameter data is collected, noise filtering, time interpolation and abnormality rejection are performed, the multi-physical field state vector is obtained, the spatial inverse distance weighted algorithm is used to perform space-time correction and fusion on the multi-physical field state vector, and the multi-dimensional original data set is obtained.

[0014] A distributed sensor network is established, and the multi-physical field operation parameter data is collected through a master clock synchronization mechanism.

[0015] Noise filtering, time interpolation and abnormality rejection are performed on the multi-physical field operation parameter data to obtain a trusted data stream, and the multi-physical field state vector is calculated through the trusted data stream.

[0016] The spatial inverse distance weighted algorithm is used to perform space-time correction and fusion on the multi-physical field state vector to obtain a multi-dimensional original data set.

[0017] As a preferred scheme of the storage tank metrological data processing method, the digital twin model is established based on the multi-dimensional original data set, the initial parameter set is calculated, the historical calibration data is read, the least squares method is used to fit the digital twin model parameters, the error between the digital twin model output and the historical calibration data is minimized, and the initial parameter set is obtained.

[0018] Based on the multi-dimensional original data set, a liquid level-volume sub-model, a thermal coupling sub-model and a structure strain sub-model are established to obtain a digital twin model.

[0019] The historical calibration data is read, the historical calibration data is input into the digital twin model, and the error between the output of the digital twin model and the calibration data is calculated.

[0020] The parameters of the digital twin model are fitted using the least squares method. By minimizing the error between the output of the digital twin model and the calibration data, the initial values ​​of the Bessel shape factor, thermal expansion coefficient, pressure compressibility coefficient, tank deformation influence coefficient, and reference density are calculated to obtain the initial parameter set.

[0021] As a preferred embodiment of the method for processing tank metering data according to the present invention, the steps of collecting real-time observation data, inputting the initial parameter set and the real-time observation data into a digital twin model, calculating residuals, performing near-end optimization inversion, and obtaining a correction parameter set are as follows.

[0022] Collect real-time observation data, input the real-time observation data and the initial parameter set into the digital twin model, and calculate the predicted volume and predicted quality;

[0023] The predicted volume and mass are compared with the measured volume and mass values ​​to obtain the residuals of volume and mass. When the residuals exceed the dynamic threshold range, a residual-driven proximal optimization inversion is performed to update the parameters of the digital twin model.

[0024] After multiple iterations, when the mean square value of the residuals of volume and mass is lower than the preset convergence threshold, the set of correction parameters is obtained.

[0025] As a preferred embodiment of the method for processing tank metering data according to the present invention, the steps include: performing uncertainty propagation analysis on multi-physics operating parameter data using a calibration parameter set, calculating the uncertainty of volume and mass, employing a confidence-weighted fusion algorithm to weight and fuse the uncertainties, calculating the fused volume and fused mass, performing a comprehensive uncertainty assessment, and obtaining the fused metering parameter set. The specific steps are as follows:

[0026] The set of calibration parameters is applied to the digital twin model, and a sensitivity mapping relationship is established by applying small-amplitude perturbations to the multiphysics operating parameter data.

[0027] Based on the sensitivity mapping relationship, the measurement variance, temperature drift, and noise power of the multiphysics operating parameter data are weighted and accumulated to calculate the uncertainty of volume and mass.

[0028] A confidence-weighted fusion algorithm is used to weight and fuse uncertainties, calculate the fusion volume and fusion quality, perform comprehensive uncertainty assessment, and obtain the fusion measurement parameter set.

[0029] As a preferred embodiment of the method for processing tank metering data according to the present invention, the steps of constructing a multi-objective cost function based on the fused metering parameter set, updating weights through an exponential gradient evolution algorithm, and obtaining the feedforward control strength are as follows:

[0030] Based on the fusion of measurement parameter sets, accuracy, stability and computational efficiency indicators are extracted, a multi-objective cost function is constructed, and the weights of accuracy, stability and computational efficiency indicators are updated using the exponential gradient evolution algorithm to obtain the weight distribution.

[0031] Based on the weight distribution, the current operating conditions are comprehensively evaluated to determine the preference for accuracy, stability, and efficiency, and the feedforward control strength is adjusted accordingly.

[0032] As a preferred embodiment of the method for processing tank metering data according to the present invention, the step of performing robust aggregation across tanks based on the correction parameter set and feedforward control strength to obtain global correction parameters includes the following specific steps.

[0033] Based on the correction parameter set, the parameter increment of the current tank is calculated, and weighted geometric median-Huber robust aggregation is performed on the parameter increments from multiple tanks to obtain the global correction parameters.

[0034] In a preferred embodiment of the method for processing tank metering data according to the present invention, the step of obtaining new digital twin model parameters by determining consistency and uncertainty gating methods includes the following specific steps.

[0035] The global correction parameters are determined by using consistency and uncertainty gating methods, and full update, partial update or freeze is performed. The operation mode of the digital twin model is adjusted based on the feedforward control strength to obtain new digital twin model parameters.

[0036] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the method for processing tank metering data as described in the first aspect of the present invention.

[0037] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the method for processing tank metering data as described in the first aspect of the present invention.

[0038] The beneficial effects of this invention are as follows: By constructing a multi-submodel digital twin system that integrates Bessel fitting, thermo-coupling, and structural compensation, and by optimizing historical calibration data based on least squares, collaborative modeling of tank geometry, thermal properties, and elastic response, and acquisition of initial parameter sets are achieved; by constructing a digital twin model that includes liquid level-volume, thermo-coupling, and structural strain, and by performing parameter self-calibration based on residual-driven execution, dynamic updating and adaptive convergence of model parameters are achieved; by introducing confidence-weighted fusion and exponential gradient evolution weight optimization mechanisms and performing robust aggregation across tanks, adaptive fusion of multi-source metering information and closed-loop evolution of the model are achieved. Attached Figure Description

[0039] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Fig. 1 This is a flowchart illustrating the method for processing metering data from storage tanks.

[0041] Fig. 2 A flowchart for modeling and calibrating digital twins.

[0042] Fig. 3 A flowchart for uncertainty propagation and control weight updates.

[0043] Fig. 4 This is a flowchart for parameter aggregation and model closed-loop update. Detailed Implementation

[0044] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0045] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0046] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0047] Reference Figs. 1-4 As one embodiment of the present invention, this embodiment provides a method for processing tank metering data, including the following steps:

[0048] S1. Collect multi-physics operating parameter data of the storage tank, perform noise filtering, time interpolation and anomaly removal to obtain a multi-dimensional raw dataset.

[0049] A distributed sensor network was established to collect multi-physics information from the storage tank. The sensor network included level sensors, temperature sensors, pressure sensors, and strain gauges, which were installed at the liquid area, tank wall, and supporting structure of the tank, respectively. This network simultaneously acquired level, temperature, pressure, and structural strain signals, obtaining multi-channel signals. To ensure the time consistency of the multi-channel signals, a network time synchronization strategy was adopted, using the tank top level sensor as the master clock. The clocks of each node were uniformly corrected to ensure that the data sampling time error between different sensors did not exceed 1 millisecond, thus obtaining a multi-dimensional raw data vector.

[0050] To eliminate random noise in sensor signals, a weighted average of several sampled values ​​before and after the current sensor sampling point is performed within a preset time window to reduce high-frequency fluctuations. A cubic spline interpolation algorithm is used to interpolate the low-frequency signal, ensuring all signals are resampled within the same time interval, thus obtaining a data sequence with a uniform sampling step size on the time axis. At each moment, four types of signals—liquid level, temperature, pressure, and strain—are aggregated to form the multiphysics state vector of the storage tank at that moment. To effectively identify abnormal fluctuations caused by temperature abrupt changes, bubble interference, or structural impacts, and significantly improve the overall data reliability, a multivariate anomaly identification method using Mahalanobis distance is employed. The Mahalanobis distance is calculated as follows:

[0051] ;

[0052] in, Indicates time The squared Mahalanobis distance of the corresponding multidimensional observation data, Indicates time Multiphysics State Vector This represents the mean of the multiphysics state vector. This represents the inverse of the covariance matrix.

[0053] It should be noted that, It is obtained by statistically averaging reliable sampling data within a set historical time window, reflecting the typical physical parameter levels of the storage tank under stable operating conditions.

[0054] When the squared Mahalanobis distance exceeds the preset recognition threshold, the multiphysics state vector at the current moment is determined to be abnormal and marked as an invalid sample. For continuous abnormal segments, a time neighborhood interpolation algorithm is used to repair them in order to maintain temporal continuity and obtain a reliable data stream that eliminates abnormal interference and maintains temporal consistency.

[0055] It should be noted that the preset confidence identification threshold is constructed by building a residual distribution from historical samples with high confidence, calculating the squared Mahalanobis distance of each sample, and referring to the chi-square distribution critical value under the corresponding degrees of freedom (e.g., 95% confidence level) as the initial identification threshold. Subsequently, during real-time operation, the variation amplitude of Mahalanobis distance and its stability index are dynamically calculated in combination with the changing trend of the residual vector. When abnormal fluctuations are frequent, the threshold is appropriately lowered to improve sensitivity, and raised during stable operation to suppress false alarms.

[0056] Perform spatiotemporal correction processing on trusted data streams.

[0057] Furthermore, an algorithm based on inverse spatial distance weighting (IDW) is employed to weight the data of each node according to its distance from the center of the tank, achieving spatial fusion at the same time to obtain a multidimensional original dataset representing the overall operating status of the storage tank. The expression is as follows:

[0058] ;

[0059] in, Represents the original multidimensional dataset. Indicates the first Each sensing node at time Collected multiphysics state vectors. Indicates the first The spatial distance from each sensor node to the geometric center of the storage tank Indicates the first Spatial weight coefficients for each sensor node.

[0060] S2. Establish a digital twin model based on the multidimensional original dataset and calculate the initial parameter set.

[0061] Based on the multidimensional original dataset, a digital twin model is established. The digital twin model includes three sub-models: a liquid level-volume sub-model, a thermo-coupling sub-model, and a structural strain sub-model. To explicitly express the nonlinear relationship between the tank liquid level and volume, a geometric fitting model based on piecewise Bessel functions is adopted to establish the liquid level-volume sub-model, with the expression:

[0062] ;

[0063] in, Indicates at time From liquid level height The corresponding storage tank geometric volume, Represents pi (π). This indicates the reference radius at the bottom of the storage tank. Indicates the first Each Bessel shape coefficient Indicates the first Bessel basis functions This represents the normalized variable along the height of the tank. This represents the total number of orders of the Bessel basis functions. Represents the differential symbol.

[0064] It should be noted that the Bessel basis function is a smooth basis function that controls the continuous change of the cross-sectional radius of the tank with height. It achieves the gradual change of shape from the bottom of the tank, the tank wall to the top of the tank by changing the weight of the control points of the Bessel curve, and can keep the curve smooth and without sharp corners when splicing segments.

[0065] To characterize the combined effects of temperature and pressure on liquid density and volume, a thermo-mechanical coupling sub-model is established, with the following expression:

[0066] ;

[0067] in, Indicates at time The density of the liquid, This represents the reference density of a liquid under standard conditions and is a common density coefficient for liquids. This represents the coefficient of thermal expansion of a liquid by volume; it is a common coefficient for liquids. This represents the pressure compressibility coefficient of a liquid, and is a common coefficient for liquids. Indicates at time The average temperature of the medium inside the storage tank, Indicates the reference temperature. Indicates at time The absolute pressure value inside the storage tank. This indicates the reference pressure.

[0068] It should be noted that, , and The national or industry standard value is preferred, and it is usually taken as... , .

[0069] To account for the elastic deformation of the tank under thermal expansion and internal pressure, a structural strain correction term is introduced to compensate for the liquid level geometric volume, establishing a structural strain sub-model, the expression of which is:

[0070] ;

[0071] in, Indicates at time The volume correction amount caused by the deformation of the tank structure is used to correct... deviation, This represents the coefficient of influence of tank deformation. Indicates time The average strain value of the storage tank structure.

[0072] It should be noted that, It is obtained by spatially weighted averaging of local strain values ​​collected in real time by a distributed strain sensor array; It simulates elastic deformation under typical thermal expansion and internal pressure conditions, and calculates the average strain. The resulting volume change ratio was obtained through linear regression.

[0073] Historical calibration data, including liquid level-volume calibration curves, temperature-density correction tables, and tank strain response data, are read. For each set of historical observation data, the residuals between the model predictions and the measured calibration values ​​are calculated. The sum of squares of the residuals between the predictions and the measured calibration values ​​is constructed as the objective function. Based on the objective function, the parameters are solved using the conventional least squares method. The Bessel shape factor, volumetric thermal expansion coefficient, pressure compressibility coefficient, tank deformation influence coefficient, and reference density are used as variables to be optimized. The parameters are iteratively updated to minimize the sum of squares of the residuals. The parameter combination that minimizes the residuals is output as the initial parameter set of the digital twin model.

[0074] S3. Collect real-time observation data, input the initial parameter set and real-time observation data into the digital twin model, calculate the residuals, perform near-end optimization inversion, and obtain the correction parameter set.

[0075] Within each sampling period (e.g., 1 second), the liquid level, temperature, pressure, and strain signals within the tank are read to obtain a real-time input vector. This real-time input vector, along with the initial parameter set, is input into the digital twin model to calculate theoretical prediction values, including a predicted volume value reflecting the geometric volume corrected for liquid level and strain, and a predicted mass value reflecting the medium mass corrected for temperature and pressure. Simultaneously, measured volume and mass values ​​are acquired, independently measured by the liquid level sensor and mass flow meter, respectively. The measured volume and mass values ​​are compared with the theoretical prediction values ​​output by the digital twin model. By calculating the deviation between the measured values ​​and the theoretical prediction values, the residuals for volume and mass are obtained. When the residuals for volume and mass exceed a dynamic threshold range, a self-calibration process for the digital twin model is performed.

[0076] It should be noted that the dynamic threshold is calculated by taking the mean and standard deviation of the volume residual and the mass residual within the most recent sliding time window (e.g., the most recent 24 hours or 1000 sampling periods), using the historical mean as the baseline of the residual center, and using several times the standard deviation as the upper and lower limits of the deviation, thus forming the dynamic threshold range.

[0077] To enable the digital twin model to automatically correct its parameters during operation, a residual-driven proximal optimization inversion algorithm is adopted. This algorithm uses the sensitivity matrix output by the digital twin model to correct the parameters, achieving adaptive convergence of the digital twin model from initial parameters to corrected parameters. The expression is as follows:

[0078] ;

[0079] in, Indicates at time The set of correction parameters Represents the initial parameter set. This indicates the convergent learning step size. The Jacobian matrix representing the digital twin model, Represents the residual weighting matrix. This represents the residual vector.

[0080] It should be noted that, The step size is determined by a combination of offline calibration and online adaptive adjustment. The initial step size is calculated based on the average sensitivity of the residual to the parameter. During online operation, the step size is dynamically corrected according to the trend of residual change. When the residual continues to decrease, the step size is gradually reduced to improve stability. When the residual increases or oscillates, the step size is automatically increased to improve the response speed. The sensitivity matrix of the digital twin model output to the parameters is obtained by the finite difference method. Its elements are the first-order partial derivatives of the digital twin model output (volume, mass) with respect to each parameter in the parameter set. The initial weights are determined by statistically analyzing the historical residual variance of each output channel, and the current weights are updated in real time using a sliding window. This allows the weights to dynamically reflect the confidence level of different output channels and dynamically reflect the differences in confidence levels among the output channels.

[0081] To ensure the physical rationality of the digital twin model parameters and prevent divergence, a proximal constraint mapping operation is performed on the parameters, projecting the calibration parameter set onto a preset physical constraint set, and mapping the thermal expansion coefficient to 0 to... The pressure compressibility coefficient is mapped to 0 to a range that covers the thermal expansion properties of common liquids. The pressure response range of industrial liquids is covered, and the tank deformation influence coefficient is mapped to between 0 and 2 to reflect the elastic deformation ratio of different tank structures. The reference density is kept in the range of 500 to 1500 kg / m³ to cover the standard density range of various liquid media. If these ranges are exceeded, the calculation results of the digital twin model will not match the physical laws, causing volume prediction deviation or parameter update divergence.

[0082] When the mean square value of the residuals of the digital twin model is lower than the preset convergence threshold after several consecutive iterations, it indicates that the parameters of the digital twin model have converged and the calibration process is stable. The calibration parameter set is then saved into the digital twin model.

[0083] It should be noted that the preset convergence threshold is adaptively determined based on sensor noise statistics and real-time residual fluctuations.

[0084] S4. Perform uncertainty propagation analysis on the multi-physics operating parameter data through the calibration parameter set, and obtain the fused metrology parameter set through the confidence weighted fusion algorithm.

[0085] The liquid level, density, and temperature data output from each sensor are input into a digital twin model calibrated using a set of calibration parameters. Using the calibration parameter set as a reference, very small perturbations are applied to the liquid level, density, and temperature respectively. The difference in the output volume and mass of the digital twin model before and after the perturbation is calculated, and the sensitivity value is expressed as the ratio of the difference to the perturbation amount. Through this method of sequential perturbation and comparison, the sensitivity of each input variable is obtained, and these sensitivity values ​​are compiled into a sensitivity mapping matrix. Based on the sensitivity mapping matrix, the measurement variance, temperature drift, and noise power of each input variable are weighted and accumulated, making the contribution of input signals with high sensitivity and large errors more significant to the output uncertainty, thereby quantitatively calculating the uncertainty of volume and mass.

[0086] Since the liquid level method, density method, and mass method estimate the condition of the storage tank from geometric, physical property, and mechanical perspectives, respectively, they are different methods.

[0087] Each method has different accuracy and noise characteristics. To improve the overall measurement accuracy, a confidence-weighted fusion algorithm is adopted. Weights are automatically assigned according to the uncertainty of each estimation method, fusing multiple independent results into a globally reliable result. The expression is as follows:

[0088] ;

[0089] ;

[0090] in, Indicates at time The fused estimate of volume or mass obtained through confidence-weighted fusion. Indicates the first Each independent measurement channel (e.g., level method, density method, or mass method) at time... The obtained volume or mass estimate, Indicates the first, Each independent measurement channel at time The confidence weight coefficients, Indicates the first Each independent measurement channel at time The uncertainty variance.

[0091] A comprehensive uncertainty assessment is performed on the fusion estimate of volume or mass. The comprehensive uncertainty of the fusion output is calculated by statistically analyzing the variance of uncertainty of each measurement channel and accumulating it according to the weight square ratio. A confidence interval is calculated based on the comprehensive uncertainty to measure the credibility of the fusion result. When the width of the confidence interval is lower than the preset confidence threshold, the current fusion result is determined to be credible. The fusion volume, fusion mass and comprehensive uncertainty are encapsulated into a fusion measurement parameter set.

[0092] It should be noted that the preset confidence threshold is based on a large number of calibration samples with known real volume or mass. The distribution of the confidence interval width of the corresponding fusion estimate is statistically analyzed, and the quantile value that can cover more than 95% of normal samples is selected as the initial confidence threshold. During the operation phase, the distribution of the confidence interval width of recent fusion results is continuously monitored. If a significant convergence trend appears within a certain time window (for example, most widths are significantly reduced), it is considered that the measurement quality has improved, and the confidence threshold is automatically tightened. Conversely, if the fluctuation intensifies, the threshold is appropriately relaxed to avoid misjudgment.

[0093] S5. Based on the fusion econometric parameter set, construct a multi-objective cost function, update the weights using the exponential gradient evolution algorithm, and obtain the feedforward control strength.

[0094] Within each sampling period, based on the changes in fusion volume and fusion quality, accuracy, stability, and computational efficiency indicators are extracted. The accuracy indicator is obtained by comparing the ratio of the comprehensive uncertainty to the allowable upper limit. The stability indicator is obtained by calculating the short-term variance of the fluctuation range of fusion volume and quality within the sliding window. The computational efficiency indicator is obtained by measuring the ratio of the computation time in this period to the target time limit, reflecting the efficiency of computational resource utilization.

[0095] It should be noted that the allowable upper limit is an initial relative uncertainty limit set by industry metrology standards, and is corrected based on the tank geometry and sensor resolution parameters. The allowable upper limit is dynamically updated using exponential smoothing based on the statistical mean of the comprehensive uncertainty in recent fusion metrology processes. The target time limit is initially set to 60%–80% of the sampling period, and is dynamically updated using exponential smoothing based on the statistical results of calculation time consumption over a recent period. If the target time limit is set to less than 50% of the sampling period, the fusion task may be interrupted or the results delayed due to calculation timeouts when data volume fluctuates or the computational load is high. If the target time limit is set too high, exceeding 90% of the sampling period, there will be no buffer margin in the event of sudden delays, affecting the stability and continuity of the fusion output, and may even conflict with the next sampling period.

[0096] Accuracy, stability, and computational efficiency are considered as three sub-objectives in a multi-objective optimization. A comprehensive cost function is constructed using linear weighting to balance the conflicting relationships between the different objectives. To ensure that the weights can be automatically adjusted according to changes in operating conditions, an exponential gradient evolution algorithm is used for updating, maintaining non-negativity and normalization while guaranteeing an exponentially sensitive response to changes in the indicators. The expression is as follows:

[0097] ;

[0098] in, Indicates at time At that time, the first The weighting coefficient of each objective. Indicates at time The updated version Each target weight Indicates the first The goal is at any time The normalized value, This represents the step size for weight learning.

[0099] It should be noted that, The initial values ​​are determined through offline simulation calibration, and during operation, they are automatically adjusted based on the ratio of the normalized value to the weight change rate.

[0100] Based on the weight distribution updated by the exponential gradient evolution algorithm in the previous cycle, the current preference for prioritizing accuracy, stability, or efficiency is comprehensively evaluated. If the accuracy and stability weights are high under the current operating conditions, it indicates that the digital twin model needs to maintain strong measurement reliability and data smoothness. Therefore, the feedforward control strength needs to be automatically increased, increasing the data sampling frequency, widening the filtering window, and improving the accuracy level to enhance the suppression of small fluctuations and noise disturbances. If the proportion of the efficiency weight increases, it indicates that the current operating conditions require a higher response speed. In this case, the feedforward control strength is reduced, and the data sampling frequency or calculation window is appropriately decreased to reduce the computational load and improve real-time performance. Through this adaptive adjustment method based on weight ratios, the feedforward control strength can automatically achieve the optimal trade-off between stability and efficiency under different operating conditions, ensuring both the accuracy of the tank metering results and maintaining real-time response capability.

[0101] S6. Based on the correction parameter set and feedforward control strength, perform robust aggregation across storage tanks to obtain global correction parameters. Then, fuse these parameters through consistency and uncertainty gating to obtain new digital twin model parameters.

[0102] The difference between the calibration parameter set and the parameters from the previous stable operating cycle is used as the parameter increment. A reliability weight is calculated based on the overall uncertainty level of each tank; the lower the uncertainty, the higher the weight. The reliability weights are then obtained, and a weighted geometric median-Huber robust aggregation is performed on the parameter increments from multiple tanks and the reliability weights to resist abnormal sites and sudden changes in operating conditions. The expression is as follows:

[0103] ;

[0104] in, Represents the global correction parameter vector. Denotes the variable that minimizes the objective function. That is, to find the optimal global correction vector. Indicates to the source The data from each storage tank site were aggregated and summed. Indicates the first The credibility weighting coefficient of each storage tank. This represents the Huber loss function. Indicates the first The local parameter increment vector of each storage tank.

[0105] Compare the direction of change of each component of the global correction parameter vector with the direction of change during the last parameter calibration. If the two directions of change are consistent, it means that the global correction is consistent with the convergence trend of the digital twin model. The current correction is considered to be physically reasonable and is updated directly. If the directions of change are opposite, it is considered inconsistent. The correction magnitude of the components is reduced or temporarily suspended to prevent the introduction of reverse offset.

[0106] By combining the overall uncertainty and the aggregate uncertainty, uncertainty gating is performed on each parameter component. When the root mean square of the overall uncertainty is lower than the allowable upper limit, the accuracy is considered high. When the aggregate uncertainty does not exceed 1.5 times the long-term average, the aggregation result is considered stable, and the update weight of the current component is set to 1, and a full update is performed. If either the overall uncertainty or the aggregate uncertainty fails to meet the limit, the update weight is reduced to 0.5, and only a partial update is performed. If both the overall uncertainty and the aggregate uncertainty exceed the limit, the current component is frozen and awaits re-aggregation in the next cycle.

[0107] It should be noted that the aggregation uncertainty is obtained by statistically analyzing the variance of the aggregation residuals of multiple tank parameters.

[0108] Adaptive adjustment is performed on the digital twin model based on the feedforward control strength.

[0109] Furthermore, when the feedforward control strength is high, it indicates a greater emphasis on accuracy and stability, and the digital twin model prioritizes accuracy maintenance. In this mode, recalibration is initiated. By shortening the data sampling interval, increasing the data filtering order, and reducing the parameter update step size, the digital twin model re-estimates key parameters at a higher confidence level, thereby eliminating residual errors and improving its stability. When the feedforward control strength is at a medium level, normal operation is maintained, with only minor recalibration performed within a preset time window. Without interrupting operation, real-time monitoring data within this short time window is used to perform small-step local corrections on key parameters with large deviations, quickly eliminating minor drifts and maintaining the accuracy of the digital twin model. When the feedforward control strength is low, it indicates a higher weighting of computational efficiency in the previous cycle. In this mode, a fast operation is automatically switched, reducing the sampling frequency, relaxing filtering constraints, and maintaining existing parameters to improve computational response speed. Through an adaptive control mechanism based on feedforward control strength, the balance between accuracy and efficiency is dynamically adjusted according to actual operating conditions in each operating cycle, achieving continuous self-calibration and stable operation of the digital twin model. In the next sampling period, the global correction parameter vector is set as the new initial parameter set, realizing continuous iterative updates of parameters between periods, thus forming an adaptive and evolving closed-loop optimization mechanism.

[0110] This embodiment also provides a computer device applicable to the method for processing tank metering data, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the method for processing tank metering data as proposed in the above embodiment.

[0111] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0112] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the method for processing tank metering data as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0113] In summary, this invention achieves collaborative modeling and initial parameter set acquisition of tank geometry, thermal properties, and elastic response by: constructing a multi-submodel digital twin system integrating Bessel fitting, thermodynamic coupling, and structural compensation, and optimizing historical calibration data based on least squares; dynamically updating and adaptively converging model parameters by constructing a digital twin model including level-volume, thermodynamic coupling, and structural strain, and performing parameter self-calibration based on residual-driven execution; and adaptively fusing multi-source metering information and achieving closed-loop evolution of the model by introducing confidence-weighted fusion and exponential gradient evolution weight optimization mechanisms and performing robust cross-tank aggregation.

[0114] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for processing storage tank metering data, characterized in that: include, Collect multiphysics field operation parameter data, perform noise filtering, time interpolation and anomaly removal, obtain multiphysics field state vectors, and use spatial inverse distance weighting algorithm to perform spatiotemporal correction and fusion of multiphysics field state vectors to obtain multidimensional raw datasets; A digital twin model is established based on a multidimensional original dataset. The initial parameter set is calculated, historical calibration data is read, and the parameters of the digital twin model are fitted using the least squares method. The initial parameter set is obtained by minimizing the error between the output of the digital twin model and the historical calibration data. Collect real-time observation data, input the initial parameter set and real-time observation data into the digital twin model, calculate the residuals, perform near-end optimization inversion, and obtain the correction parameter set; Uncertainty propagation analysis is performed on the multiphysics operating parameter data using a set of calibration parameters to calculate the uncertainty of volume and mass. A confidence-weighted fusion algorithm is then used to weight and fuse the uncertainties to calculate the fused volume and fused mass. A comprehensive uncertainty assessment is then performed to obtain the fused metrology parameter set. Based on the fusion econometric parameter set, a multi-objective cost function is constructed, and the weights are updated through the exponential gradient evolution algorithm to obtain the feedforward control strength. Based on the correction parameter set and feedforward control strength, robust aggregation across storage tanks is performed to obtain global correction parameters. These parameters are then determined through consistency and uncertainty gating to obtain new digital twin model parameters. The set of calibration parameters is applied to the digital twin model, and a sensitivity mapping relationship is established by applying small-amplitude perturbations to the multiphysics operating parameter data. Based on the sensitivity mapping relationship, the measurement variance, temperature drift, and noise power of the multiphysics operating parameter data are weighted and accumulated to calculate the uncertainty of volume and mass. A confidence-weighted fusion algorithm is used to weight and fuse uncertainties, calculate the fusion volume and fusion quality, perform comprehensive uncertainty assessment, and obtain the fusion measurement parameter set. Based on the fusion of measurement parameter sets, accuracy, stability and computational efficiency indicators are extracted, a multi-objective cost function is constructed, and the weights of accuracy, stability and computational efficiency indicators are updated using the exponential gradient evolution algorithm to obtain the weight distribution. Based on the weight distribution, the current operating conditions are comprehensively evaluated for their preference for accuracy, stability, and efficiency, and the feedforward control strength is adjusted accordingly. Based on the correction parameter set, the parameter increment of the current tank is calculated, and weighted geometric median-Huber robust aggregation is performed on the parameter increments from multiple tanks to obtain the global correction parameters.

2. The method for processing tank metering data as described in claim 1, characterized in that: The process involves collecting multiphysics operating parameter data, performing noise filtering, time interpolation, and anomaly removal to obtain multiphysics state vectors. A spatial inverse distance weighting algorithm is then used to perform spatiotemporal correction and fusion of the multiphysics state vectors to obtain a multidimensional raw dataset. The specific steps are as follows: Establish a distributed sensor network and collect multi-physics field operation parameter data through a master clock synchronization mechanism; Noise filtering, time interpolation, and anomaly removal are performed on the multiphysics operating parameter data to obtain a reliable data stream, and the multiphysics state vector is calculated through the reliable data stream. A spatial inverse distance weighting algorithm is used to perform spatiotemporal correction and fusion of multiphysics state vectors to obtain a multidimensional original dataset.

3. The method for processing tank metering data as described in claim 2, characterized in that: The process involves establishing a digital twin model based on a multidimensional original dataset, calculating an initial parameter set, reading historical calibration data, fitting the digital twin model parameters using the least squares method, and obtaining the initial parameter set by minimizing the error between the digital twin model output and the historical calibration data. The specific steps are as follows: Based on the multidimensional original dataset, a liquid level-volume sub-model, a thermo-coupling sub-model, and a structural strain sub-model are established to obtain a digital twin model; Read historical calibration data, input the historical calibration data into the digital twin model, and calculate the error between the output of the digital twin model and the calibration data; The parameters of the digital twin model are fitted using the least squares method. By minimizing the error between the output of the digital twin model and the calibration data, the initial values ​​of the Bessel shape factor, thermal expansion coefficient, pressure compressibility coefficient, tank deformation influence coefficient, and reference density are calculated to obtain the initial parameter set.

4. The method for processing tank metering data as described in claim 3, characterized in that: The process involves collecting real-time observation data, inputting the initial parameter set and the real-time observation data into a digital twin model, calculating residuals, performing near-end optimization inversion, and obtaining a correction parameter set. The specific steps are as follows: Collect real-time observation data, input the real-time observation data and the initial parameter set into the digital twin model, and calculate the predicted volume and predicted quality; The predicted volume and mass are compared with the measured volume and mass values ​​to obtain the residuals of volume and mass. When the residuals exceed the dynamic threshold range, a residual-driven proximal optimization inversion is performed to update the parameters of the digital twin model. After multiple iterations, when the mean square value of the residuals of volume and mass is lower than the preset convergence threshold, the set of correction parameters is obtained.

5. The method for processing tank metering data as described in claim 4, characterized in that: The process of obtaining new digital twin model parameters through consistency and uncertainty gating involves the following steps: The global correction parameters are determined by using consistency and uncertainty gating methods, and full update, partial update or freeze is performed. The operation mode of the digital twin model is adjusted based on the feedforward control strength to obtain new digital twin model parameters.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the method for processing tank metering data according to any one of claims 1 to 5.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the method for processing tank metering data according to any one of claims 1 to 5.

Images