Storage tank capacity meter dynamic calibration method and system based on multi-source data fusion

Abstract

The invention discloses a storage tank capacity meter dynamic calibration method and system based on multi-source data fusion. The method specifically comprises the steps that data are collected, oil gun refueling data and liquid level instrument data are obtained, standardized packaging processing is conducted on the oil gun refueling data and the liquid level instrument data, data format differences are eliminated, and a standardized data packet is formed; preprocessing the data in the standardized data packet, the preprocessing comprising synchronizing the timestamps of the oil gun sales data and the liquid level instrument data, and constructing a corresponding data unit; combining two adjacent data units which are overlapped in time; extracting a pure consumption interval and verifying abnormal data; if the storage tank has the initial tank capacity meter, calibrating the initial tank capacity meter of the storage tank by adopting an FGIR four-dimensional calibration algorithm; and if the initial capacity tank table does not exist, constructing the initial capacity tank table for the oil tank based on a geometric constraint strategy, and then calibrating the initial capacity tank table of the oil tank by adopting an FGIR four-dimensional calibration algorithm.

Description

Technical Field

[0001] This invention belongs to the field of automated metering and calibration technology for petroleum storage and transportation, and particularly relates to a dynamic calibration method and system for tank capacity tables based on multi-source data fusion. Background Technology

[0002] Tank capacity gauges are fundamental for accurate inventory measurement in gas stations, and their accuracy directly affects inventory accounting, loss control, and inventory balance management. Currently, common tank capacity gauge calibration methods mainly include the following three:

[0003] 1. Static manual calibration: This method involves injecting oil into the storage tank in stages, measuring the liquid level and recording the corresponding volume, and then plotting a complete tank volume curve. This method is time-consuming, requires shutting down the storage tank, disrupts normal operations, and cannot reflect volume changes caused by temperature variations or tank deformation during actual operation.

[0004] 2. Level gauge calibration alone: ​​Only the level gauge sensor is calibrated, without fully considering the dynamic volume changes of the storage tank due to temperature, pressure and structural deformation during actual use. As a result, the calibration results deviate from the actual volume.

[0005] 3. Periodic offline correction: The tank capacity table is adjusted uniformly based on historical inventory and transaction data. However, this method lacks real-time capability and the correction process is easily affected by abnormal data, which can lead to a decrease in accuracy.

[0006] The above method also has the following shortcomings:

[0007] • The calibration process requires tank shutdown, which affects the normal operation of the gas station and is costly;

[0008] • Inability to perform online real-time calibration, making it difficult to dynamically capture volume changes;

[0009] • The effects of dynamic factors such as temperature, pressure, and elastic deformation of the tank were not systematically considered;

[0010] • The lack of reliable data cleaning and outlier removal mechanisms makes the correction results prone to distortion. Summary of the Invention

[0011] Purpose of the invention: In order to solve the problems existing in the prior art, the present invention provides a method and system for dynamic calibration of tank capacity tables based on multi-source data fusion.

[0012] Technical solution: This invention provides a dynamic calibration method for tank capacity tables based on multi-source data fusion, specifically as follows:

[0013] Data acquisition: Acquires data from the fuel nozzle and level gauge, with all acquired data accompanied by timestamps accurate to the second.

[0014] The data from the fuel nozzle and the level gauge are standardized and packaged to eliminate data format differences and form a standardized data package.

[0015] The data in the standardized data packet is preprocessed, including synchronizing the timestamps of the oil gun sales data and the liquid level gauge data to construct corresponding data units; merging two adjacent data units that overlap in time; extracting pure consumption intervals; and verifying abnormal data.

[0016] If the tank has an initial capacity table, the FGIR four-dimensional calibration algorithm is used to calibrate the initial capacity table of the tank; if there is no initial capacity table, an initial capacity table is constructed for the tank based on a geometric constraint strategy, and then the FGIR four-dimensional calibration algorithm is used to calibrate the initial capacity table of the tank.

[0017] Furthermore, the refueling data includes the gas station code, tank number, nozzle number, refueling volume, and nozzle hanging time; the level gauge data includes: the oil level in the tank, basic data reflecting the remaining oil in the tank in real time, tank temperature, and water level at the bottom of the tank.

[0018] Furthermore, the timestamps of the fuel nozzle sales data and the level gauge data are synchronized, specifically as follows:

[0019] Based on the current refueling record, the time when the nozzle was hung up is used as the base time;

[0020] A binary search algorithm is used to find the liquid level record L with the closest timestamp in the continuous monitoring data sequence of the liquid level instrument;

[0021] Based on the preset forward and backward search time windows, and with the timestamp corresponding to record L as the center, the search timestamp is located at... All level gauge data within the range;

[0022] against All level gauge data within the interval are first processed, and invalid data with abnormal formats are removed. The remaining valid data are then sorted according to the difference between the acquisition time and the nozzle hanging time. The sorted data is then bound to the current refueling record to form a data association unit, and calculations are performed. The change in liquid level gauge height within the interval is recorded and compared with the nominal oil volume of the corresponding refueling record. The error is calculated, and if the error exceeds a preset error threshold, the verification process is stopped and an alarm is triggered. If the difference between the initial liquid level gauge height of the later data unit and the ending liquid level gauge height of the previous data unit is greater than a preset difference threshold, the verification process is stopped and an alarm is triggered. All refueling records are traversed sequentially to complete the construction of all data association units.

[0023] Furthermore, the extraction of the pure consumption interval is specifically as follows:

[0024] A set is formed based on all data units. The height at which the liquid level begins to change in each data unit is recorded as... ;

[0025] Compare two adjacent data points If the absolute value of the difference is less than or equal to the threshold, then the two points are considered not to constitute a trend change. If neither point t nor point T constitutes a trend change, then points t to T are considered a unified node. If there is an increasing trend after point T, then within the range of points t to T, [the following is considered]. The largest value is used as the value corresponding to the t-th point to the T-th point; if the value decreases after the T-th point, then within the range of the t-th point to the T-th point, select... The minimum value is taken as the value corresponding to the point from the t-th point to the T-th point;

[0026] Then record the height of the level gauge and the time corresponding to the peak and trough;

[0027] Periodic alternating verification: If two or more consecutive peaks or two or more consecutive troughs are detected, it is judged as a logic error, the verification process is interrupted, and an alarm is triggered.

[0028] The interval between the correct peak and trough is taken as the pure consumption interval.

[0029] Furthermore, the FGIR four-dimensional calibration algorithm is used to calibrate the initial tank capacity table of the oil tank, specifically as follows:

[0030] Step 1: Call the actual capacity calculation algorithm, starting from the lowest point of the pure consumption interval, and in reverse refueling time sequence, accumulate the refueling amount of each data unit record forward in sequence, and calculate the actual volume at the beginning of the previous record for each record.

[0031] Step 2: Call the calibration point acquisition method, combine the initial liquid level height recorded in each data unit with its corresponding calculated actual volume to form a calibration reference point, and perform temperature compensation on the actual volume according to the following formula, and then sort them according to the liquid level height.

[0032] ;

[0033] in, To calculate the actual volume, The volume at a standard temperature of 20°C This refers to the coefficient of volumetric expansion of the oil product.

[0034] Step 3: Periodically revise the initial tank capacity table using piecewise linear interpolation and linear fitting.

[0035] Step 4: Select the final tank table as the calibrated tank table from the multiple candidate tank tables obtained by piecewise linear interpolation, or from the multiple candidate tank tables obtained by linear fitting, or from both the multiple candidate tank tables obtained by piecewise linear interpolation and the multiple candidate tank tables obtained by linear fitting.

[0036] Furthermore, piecewise linear interpolation is used to revise the initial tank capacity table, specifically as follows:

[0037] If the initial scale h in the tank gauge is < If the volume value corresponding to that height point in the tank table remains unchanged;

[0038] If h > maxHeight, calculate the difference delta between the actual volume at height minHeight and the volume at height minHeight in the initial tank table, and increase the volume corresponding to the height in the initial tank table that is higher than minHeight by delta; minHeightwei is the lowest liquid level in the pure consumption range, and maxHeight is the highest liquid level in the pure consumption range;

[0039] If minHeight≤h≤maxHeight, then the highest height in the data cell corresponding to h is recorded. and minimum height , Corresponding volume as well as Corresponding volume Then, calculate the actual volume at height h using the following formula:

[0040] .

[0041] Furthermore, step 4 specifically involves:

[0042] Step 4.1: Select the evaluation parameters, including the target gas station, oil tank, evaluation period, and the daily shift end time of the gas station. The evaluation period will be automatically divided into multiple continuous sub-periods based on the shift end day.

[0043] Step 4.2: For any candidate tank capacity table, iterate through the end-of-shift period and interpolate the liquid level height value in each level gauge record within that period to obtain the corresponding theoretical volume value. ;

[0044] Step 4.3: For each refueling record within the shift's closing day, locate the level gauge record closest to its start and end times. and Calculate the theoretical volume change And compare it with the actual amount of fuel dispensed by the fuel dispenser in the refueling record. By comparison, the error per transaction is obtained. :

[0045] ;

[0046] ;

[0047] Step 4.4: Take the first and last level gauge records for each shift's closing date. and Calculate the total theoretical change for the day. ; and all actual refueling volume within the day of the class's conclusion. Comparison, the overall error of the day's work is obtained. :

[0048] ;

[0049] ;

[0050] Step 4.5: After iterating through all shift closing dates, calculate the sum of the absolute values ​​of the shift closing date errors (SADE) and the sum of squared residuals of individual errors (SSE):

[0051] ;

[0052] ;

[0053] Step 4.6: Select the candidate version with the smallest SADE value. If there are multiple candidate versions with equal SADE values ​​or the difference between SADE values ​​is less than the set threshold, these multiple candidate versions are set as set A, and the version with the smallest SSE is selected from set A.

[0054] Furthermore, the initial tank capacity table for the oil tank is constructed based on the geometric constraint strategy, specifically using the following formula:

[0055] ;

[0056] Where L refers to the transverse length of the central cylinder of the horizontal oil tank; R refers to the radius of the end caps at both ends of the horizontal oil tank. The markings indicate the height of the fuel level inside the tank; among which... , Indicates the height in the initial tank table. The corresponding volume.

[0057] A system for dynamic calibration of tank capacity tables based on multi-source data fusion includes...

[0058] The liquid level gauge data interface module is used to collect liquid level gauge data, and this module is compatible with multiple brands.

[0059] The refueling transaction data interface module is used to collect refueling data from fuel nozzles; this module has flexible multi-source data access capabilities.

[0060] The data upload module is used to standardize and encapsulate the data collected by the liquid level gauge data interface module and the refueling transaction data interface module into standardized data packets and upload them to the cloud processing center. At the same time, it monitors the data upload progress and success rate in real time, and promptly sends a warning to the alarm module through remote alarm information when abnormal situations such as transmission failure occur.

[0061] A cloud-based processing center is used for data preprocessing;

[0062] The FGIR four-dimensional calibration algorithm module is used to calibrate the initial tank table based on preprocessed data;

[0063] The alarm module is used to alert the system to any abnormal situations that occur during the real-time detection and calibration process.

[0064] Furthermore, both the liquid level gauge data interface module and the refueling transaction data interface module have built-in local cache units. When the network is temporarily interrupted or the cloud service is temporarily unavailable, the liquid level gauge data interface module and the refueling transaction data interface module will store the corresponding data in the local cache unit. When the network is restored or the cloud service is normal, the data upload module will automatically start the breakpoint resume function and upload the cached data in the order of data generation time.

[0065] The data upload module uses the HTTPS network protocol;

[0066] The abnormal situations include: oil tank leakage, insufficient number of valid data points for calibration, and single calibration error exceeding a preset threshold.

[0067] Beneficial effects:

[0068] 1. High efficiency and economy, no downtime required: The entire calibration process of this invention is completed online automatically, without the need to stop the tank, clean the tank, or rely on third-party calibration services, which greatly shortens the calibration time and effectively reduces labor and economic costs.

[0069] 2. Safe and environmentally friendly, with a closed operation: During the calibration process, the storage tank remains fully sealed, eliminating the safety hazards and oil and gas volatilization pollution caused by traditional tank opening operations, and meeting the requirements of green environmental protection and safe production.

[0070] 3. Adaptive and continuous optimization mechanism: Based on the residual feedback closed-loop structure, the system has the ability to continuously learn and optimize parameters, and can dynamically adapt to the slow deformation of the tank caused by temperature, pressure or foundation settlement.

[0071] 4. Wide compatibility and easy to promote: The system architecture is open, compatible with mainstream level gauges and station-level management systems, and the interface is standardized, making it easy to quickly deploy and expand the application in existing gas station facilities.

[0072] 5. High-precision calibration capability: Through the comprehensive application of the FGIR four-dimensional calibration algorithm, the system has both overall trend fitting and local fine-tuning functions, which significantly improves the accuracy of the tank capacity table and keeps the oil metering loss rate stably controlled within the national standard (±3‰). Attached Figure Description

[0073] Figure 1 This is a diagram of the overall architecture of the present invention.

[0074] Figure 2 This is a flowchart of data preprocessing and effective period extraction provided in an embodiment of the present invention.

[0075] Figure 3 This is a tank capacity comparison chart of liquid level and tank capacity provided in an embodiment of the present invention.

[0076] Figure 4 This is a cutoff pure consumption cycle diagram provided in an embodiment of the present invention.

[0077] Figure 5 This is a schematic side view of the parameters of a horizontal oil tank provided in an embodiment of the present invention. Detailed Implementation

[0078] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0079] This invention provides a method and system for dynamic calibration of tank volume tables based on multi-source data fusion, enabling high-precision automated calibration of tank volume tables. This solution aims to address the problems of existing calibration technologies, such as reliance on manual labor, high cost, low efficiency, and disruption to normal gas station operations. By employing data-driven analysis, it integrates high-precision metering data from fuel dispensers with time-series monitoring data from level gauges to achieve continuous adaptive correction of the tank volume tables, significantly improving calibration accuracy and reliability while ensuring uninterrupted operation.

[0080] like Figure 1 As shown, the system adopts a cloud-based collaborative distributed architecture, which mainly includes the following three layers:

[0081] Data acquisition layer (station end)

[0082] As the core data entry point for monitoring the condition of oil tanks at gas stations, this module is compatible with multiple brands and can be seamlessly adapted to level gauges from major global brands. At the data communication protocol level, the module adopts a dual-protocol architecture design: on the one hand, it follows the internationally recognized vestibule IFS protocol (International Gas Station Equipment Forum Standard Protocol) to ensure standardized data interaction with level gauge equipment that conforms to industry standards, guaranteeing the universality and compatibility of data formats; on the other hand, it deeply analyzes and integrates the proprietary protocols of various manufacturers, optimizing the data parsing algorithm for the protocol characteristics of different brands of level gauges ("interfacing with the existing standard interfaces of each level gauge manufacturer" is the core principle, verified by multiple brands on-site, ensuring interface compatibility and data transmission stability, laying a solid foundation for level data acquisition. While "optimizing the data parsing algorithm for protocol characteristics" is based on the existing technical framework, improvements through "dynamic adaptation, intelligent verification, and performance enhancement" can overcome the limitations of new protocol adaptation efficiency, complex scenario parsing reliability, and multi-device concurrent speed, improving compatibility, parsing accuracy, and real-time performance, providing technical support for subsequent new brand integration and high-requirement scenario implementation, achieving a seamless connection between current stable operation and future upgrades), avoiding data loss or parsing errors caused by protocol differences. During the data acquisition phase, the module captures core status parameters of the oil tank at a high frequency of once per minute. These parameters include: the oil level (mm) inside the tank, providing real-time data on the remaining oil volume (L); the tank temperature (°C), providing crucial information for oil volume conversion and mass measurement; and the water level at the bottom of the tank (mm). All collected data is accompanied by a timestamp accurate to the second, ensuring the time traceability of each data point and providing a precise time reference for subsequent inventory tracking and anomaly analysis.

[0083] Gasoline transaction data interface module:

[0084] This module serves as a crucial hub for gas station transaction data collection, possessing flexible multi-source data access capabilities. It can directly acquire transaction data from the gas station's core management system (including the BOS gas station back-end management system and POS cashier system) or the local control unit of the fuel dispenser, covering key information throughout the entire process from transaction initiation to completion. This completely streamlines the information chain from "refueling operation to data recording," avoiding errors and inefficiencies caused by manual recording. The transaction data collected by the module is comprehensive and accurate, with core components including: tank number, clearly identifying the fuel tank corresponding to each transaction, providing a basis for accurately matching tank inventory consumption with the corresponding transaction; fuel nozzle number, locating the specific refueling operation equipment, facilitating subsequent equipment management work such as fuel nozzle troubleshooting and refueling volume calibration; refueling volume (L), recording the actual fuel supply for each transaction, which is core data for transaction settlement and inventory accounting; and nozzle hanging time, accurately recording the time node when the refueling operation is completed, which, combined with a timestamp, enables a precise correspondence between "transaction time" and "operation completion time." In addition, the module can be expanded to collect additional fields such as transaction amount, payment method, and license plate information according to actual needs, to meet the diverse data needs of gas station refined management and customer profile analysis, and provide richer data dimensions for gas station operation decisions.

[0085] Data upload module:

[0086] As a "data bridge" connecting the local data collection terminal of the gas station with the cloud processing center, this module undertakes the key responsibilities of preliminary data processing, secure caching and reliable transmission, ensuring that the massive amounts of data collected locally can be transmitted to the cloud efficiently, securely and without loss, providing stable data input for subsequent cloud data storage, analysis and application.

[0087] In the data preprocessing stage, the module first standardizes and encapsulates the collected raw data from the level gauge and the raw data from refueling transactions: according to the data format specifications, it performs unified format conversion and field mapping on data from different sources and in different formats to eliminate data format differences and form standardized data packets, which facilitates data parsing and unified storage in the cloud processing center; at the same time, it performs preliminary legality verification on the data, filters out abnormal data caused by temporary equipment failures or signal interference (such as liquid level heights exceeding reasonable ranges, negative refueling volumes, etc.), and marks the data with abnormal verification for subsequent manual review and anomaly tracing.

[0088] Regarding the data caching mechanism, each module has an independent built-in local cache unit that adopts a "cache first, upload later" strategy: when the network is temporarily interrupted or the cloud service is temporarily unavailable, the module uses a preset local storage trigger logic to store the data to be uploaded into the local cache space, completely solving the problem of data loss caused by network fluctuations; after the network is restored or the cloud service is normal, the module automatically starts the breakpoint resume function and uploads the cached data in the order of data generation time, ensuring the continuity and integrity of data transmission and achieving a reliable guarantee of "no data loss when the network is down and automatic transmission when the network is up".

[0089] In the data transmission stage, the module adopts the highly reliable network protocol (HTTPS) to ensure the security of data during transmission and prevent data from being stolen or tampered with. In addition, the module also has a transmission status monitoring function, which can monitor the data upload progress and success rate in real time. When abnormal situations such as transmission failure occur, it will issue an early warning in a timely manner through remote alarm information, so that operation and maintenance personnel can intervene in time to ensure the continuous smooth transmission of data.

[0090] Data preprocessing workflow (cloud-based)

[0091] like Figure 2 As shown, the data preprocessing workflow includes the following steps:

[0092] Obtain fuel nozzle refueling data:

[0093] Search for data by "Gas Station Code (locating the specific gas station), Tank Number (corresponding tank), and Refueling Time (accurate to the second)" to obtain information such as nozzle number, refueling volume (accurate to 0.01L), and nozzle hanging time. Export to Excel / CSV is supported. Refueling data is sorted in reverse chronological order.

[0094] Obtain liquid level gauge data:

[0095] The query criteria are consistent with the refueling data (gas station code, tank number, start and end time), retrieving the tank's liquid level (mm), tank temperature (°C), total oil volume (L), and water level (mm), with a second-level timestamp. Excel / CSV export is supported. The liquid level gauge data is sorted in reverse chronological order.

[0096] Query the existing tank capacity table:

[0097] The tank capacity table corresponds to the specified gas station code and tank number. You can view different heights (mm) and a table of "height corresponding to oil volume" (L). The height gradient supports 1mm, 5mm and 10mm, with 1mm having the highest accuracy.

[0098] Preprocessing (time alignment) of data from oil nozzles and level gauges

[0099] Ensure that the timestamps of the fuel nozzle sales data and the level gauge data (tank level, tank temperature, tank oil volume, and level gauge data collection time) are synchronized. Use the same UTC / Asia / Shanghai time.

[0100] The collected liquid level data is processed in two steps: first, invalid information such as abnormal format (e.g., null values, incorrect timestamps) is removed; then, the remaining valid data is sorted by the difference between the collection time and the gun hanging time, and only the set of data with the smallest difference (i.e. the liquid level value closest to the time of gun hanging) is retained, so as to accurately anchor the liquid level status at the end of refueling.

[0101] Based on the time corresponding to the liquid level status at the end of refueling, trace back... Level gauge data for 1 minute and T_back minutes (both durations are configurable parameters that can be flexibly adjusted according to scenarios such as vehicle queuing time and fuel nozzle turnover efficiency during peak hours at gas stations) ensures that the data collection range fully covers the entire refueling process from lifting the nozzle to hanging it up, avoiding missing key level change nodes.

[0102] Data association and fuel volume verification: The liquid level data of the time window is uniquely bound to the corresponding refueling record through the tank number and nozzle number. Then, from the traced liquid level data, the stable initial liquid level before refueling is selected (such as the value of liquid level without fluctuation at 3 consecutive collection points). Combined with the tank capacity table parameters, the "initial liquid level volume" and "liquid level volume at the moment of nozzle hanging" are calculated respectively. The difference between the two is the change in tank capacity table. At the same time, the "nominal fuel volume" (the amount of fuel refueling displayed by the fuel dispenser) in the refueling record is extracted, and the error is calculated by "nominal fuel volume - change in tank capacity table".

[0103] Oil order consolidation:

[0104] If the last liquid level data of the previous refueling transaction is collected later than the first liquid level data of the subsequent refueling transaction, it means that the liquid level records of the two refueling transactions overlap. These two refueling transactions can be merged to summarize the total refueling volume, total time range, and total liquid level meter changes.

[0105] Abnormal data verification:

[0106] The data is checked for anomalies. If the volume of the first record is greater than that of the last record, it indicates that oil may be unloading or returning to the tank, and this record is deemed invalid. For the merged oil order, the starting and ending liquid level gauge heights and volumes are accurately assigned. Finally, abnormal data such as oil unloading and temperature fluctuations are filtered out.

[0107] Marking peaks and troughs:

[0108] In the processed liquid level data, find the "peak" (the point higher than the data before and after, which may be the stable oil volume before refueling) and the "trough" (the point lower than the data before and after, which may be the lower limit of the oil volume after refueling). The change from peak to trough is the stable refueling data of the oil tank.

[0109] The implementation steps of the peak and trough marking algorithm are as follows:

[0110] S1: Data Preparation and Initialization: Obtain a set of oilRecords consisting of multiple refueling records, where each refueling record includes the height at which the liquid level began to change. The Boolean fields high (peak) and low (trough) are used to indicate whether it is a peak or a trough, and the initial values ​​of high (peak) and low (trough) are both false.

[0111] S2: Data Preprocessing and Invalid Fluctuation Filtering: Traverse the oilRecords set and smooth consecutive data points. Specifically, set a liquid level change threshold HIGH_LOW_THRESHOLD (its value can be set according to the characteristics of the specific oil tank and the measurement accuracy). Compare the difference between the startH values ​​of two adjacent data points:

[0112] If the absolute value of the difference is less than or equal to the threshold HIGH_LOW_THRESHOLD, then the liquid level fluctuation between the two data points is considered to be measurement noise or invalid fluctuation, and does not constitute a trend change.

[0113] A group of consecutive data points with a difference less than or equal to HIGH_LOW_THRESHOLD is considered a "unified node". For this group of consecutive points, the representative value of the unified node is determined based on the previously known and explicit trend direction:

[0114] If the trend is increasing before reaching this point, the maximum value of startH in this set of points is taken as the representative value of this unified node, and compared with subsequent data. If the trend is decreasing before reaching this set of points, the minimum value of startH in this set of points is taken as the representative value of this unified node. This step effectively eliminates the interference of minor fluctuations, highlights the main liquid level change trend, and lays the foundation for accurate identification of extreme points in the future.

[0115] S3: Extreme Point Determination and Marking. Based on the valid data sequence formed after preprocessing in step S2 (which may contain original points and unified nodes), peaks and troughs are determined:

[0116] Peak determination: For a candidate data point i, if its startH value is greater than the corresponding values ​​of its previous valid node i-1 and its next valid node i+1, then the point is determined to be a peak, and its high (peak) field is marked as true.

[0117] Valley determination: For a candidate data point i, if its startH value is less than the corresponding values ​​of its previous valid node i-1 and its next valid node i+1, then the point is determined to be a valley, and its low (valley) field is marked as true.

[0118] S4: Periodic Alternation Verification. A valid liquid level change cycle must consist of alternating peaks and troughs. During the marking process, the system checks the order of peak and trough occurrences in real time.

[0119] If two consecutive peaks are detected (i.e., the previous extreme point is a peak and the current extreme point is also a peak), or two consecutive troughs are detected (i.e., the previous extreme point is a trough and the current extreme point is also a trough), a logical error is identified, and the processing flow is interrupted. This mechanism ensures the strictness of periodic identification and the validity of the data, preventing erroneous correction results due to data quality issues.

[0120] S5: Output Results. After traversing and processing the entire oilRecords set, the output is a data set with high (peaks) and low (troughs) correctly marked. Each marked peak and trough point together defines a complete cycle of the tank level change. These cycle data will serve as key input parameters for subsequent real-time correction of the tank capacity curve.

[0121] Capture valid refueling records for tank calibration:

[0122] Extract valid refueling records for subsequent tank calibration operations. Select pure consumption cycle data between two refueling operations (using the time period from peak to trough as the pure consumption cycle data) to avoid interference from inventory changes.

[0123] Select the refueling record between two unloading operations—this period only involves refueling consumption, without any interference from unloading. Delete data containing unloading or abnormal liquid levels, and generate a data list for tank calibration. The list includes refueling time, refueling volume (L), error (the difference between the volume change of fuel in the tank and the amount of fuel added by the nozzle) (L), the error between different values, start liquid level time, end liquid level time, start liquid level height (mm), and end liquid level height (mm).

[0124] Refueling record threshold verification:

[0125] If the error of a single refueling record exceeds the set range, the correction process should be immediately interrupted, and the abnormal error record should be returned. During the correction process, the merged fuel volume must be compared with the system's preset threshold; if it does not meet the threshold standard, the correction operation should be interrupted, and abnormal merged data should be returned synchronously. During correction, if the change in liquid level gauge volume since the last refueling is detected to be greater than a specific value (i.e., there are cases where there is no refueling record but the liquid level changes abnormally), the correction process must be interrupted, and abnormal liquid level gauge data should be returned.

[0126] Preliminary residual assessment:

[0127] Using the effective data from the tank calibration, calculate "Residual = Liquid level change - Actual refueling amount", and then calculate the relative residual (Residual ÷ Refueling amount × 100%). The relative residual can be used to initially assess the difference between the current liquid level change and the actual refueling amount.

[0128] Core algorithm layer;

[0129] Temperature compensation pretreatment: According to the national standard GB / T1885, the oil volume expansion coefficient (… This method converts the volume (Vt) measured by the level gauge at the actual temperature to the volume (V20) at the standard temperature of 20℃, eliminating the influence of temperature changes on volume calculation. The formula is: .

[0130] FGIR four-dimensional calibration algorithm core module:

[0131] This module, as the core of this invention, sequentially performs calculations and optimizations in four dimensions: Fitting, Geometric Constraints, Interpolation, and Residual Feedback. The specific steps are as follows:

[0132] Fitting: A multinomial regression model is used to fit the overall trend of the cleaned dataset, with the average liquid level height as the x-axis and the amount of oil added per unit height as the y-axis. This step aims to establish a baseline model for the current tank capacity error and to preliminarily identify and characterize systematic deviation patterns.

[0133] Geometric Constraints: The physical geometry of the horizontal oil tank is introduced as a strong constraint. This model accurately describes the mathematical relationship between the tank volume (including the central cylindrical body and the elliptical end caps) and the liquid level height. The calculation formula is as follows: ;

[0134] in, This refers to the total volume of fuel in the tank, measured in liters (L).

[0135] L refers to the transverse length of the central cylinder of the horizontal oil tank, in decimeters (dm).

[0136] R refers to the radius of the end caps at both ends of the horizontal oil tank, and the unit is "decimeter (dm)";

[0137] The markings indicate the height of the fuel level inside the tank; among which... The unit is "decimeter (dm)".

[0138] By constraining the fitting results to the physically feasible solution space, deviations from the actual tank shape are avoided. Furthermore, if the tank lacks an initial tank capacity table, it can be automatically generated based on its physical geometric parameters.

[0139] Interpolation: A piecewise linear interpolation method is used for local fine-tuning. The liquid level range is divided into multiple intervals. Within each interval, based on the error between the actual total refueling volume of the effective calibration units (Q_fuel nozzle) and the theoretical volume change monitored by the level gauge (Q_liquid level), a volume correction factor per millimeter is calculated. The calculation formula is as follows:

[0140] ;

[0141] Where V_mm is the oil volume per millimeter, in units of "liters per millimeter (L / mm)";

[0142] Q refers to the total volume of fuel actually dispensed by the fuel nozzle during the fuel filling process, expressed in liters (L).

[0143] h_start refers to the initial height of the fuel level in the tank before refueling begins, in millimeters (mm).

[0144] h_end refers to the final height of the fuel level in the tank after refueling, expressed in millimeters (mm).

[0145] The volume value corresponding to each millimeter of height within the interval is linearly recalibrated to generate a high-precision corrected tank capacity table.

[0146] Residual Feedback: The formula for calculating the residual after using the newly generated tank capacity table is as follows:

[0147] ;

[0148] ΔQ, measured in liters (L), is the fuel metering deviation. It measures the difference between the fuel nozzle output and the change in fuel volume within the tank during refueling and is a core indicator for judging metering accuracy. A value close to 0 indicates high accuracy of the tank capacity meter; if If the deviation from 0 is large, adjustments and iterations are needed to reduce the deviation.

[0149] The initial volume of fuel in the tank before refueling begins, expressed in liters (L).

[0150] This refers to the final volume of fuel in the tank after refueling, measured in liters (L).

[0151] Application and Evaluation Layer:

[0152] Calibration result output module: Used to generate and output a high-precision tank capacity table after dynamic calibration. The table uses the liquid level height (mm) as an index and maps it to the corresponding standard volume value (L), forming a liquid level-volume conversion basis that can be directly used by the level gauge.

[0153] Accuracy assessment module: The calibration effect is comprehensively evaluated by a multi-index fusion system. Specifically, it includes: calculating the sum of squared residuals (SSE) and mean absolute error (MAE) after calibration, and continuously monitoring the weekly oil loss rate in the actual operating environment to ensure that it is stably controlled within ±3‰, thereby verifying the reliability of the calibration results in terms of statistical accuracy and business consistency.

[0154] The early warning and reporting module detects anomalies during the calibration process in real time, including but not limited to single calibration errors exceeding a preset threshold, insufficient valid data points, or potential leaks (to determine if a leak exists in an oil tank, the temporal correlation between liquid level changes and refueling operations can be compared: if the oil tank level gauge records liquid level changes within a certain time period, but no corresponding refueling operation record is found during the same period, a leak should be suspected or preliminarily judged). It also automatically generates a structured calibration report, fully recording the calibration process, evaluation results, and anomaly handling information.

[0155] An embodiment of the present invention:

[0156] S1: Data Acquisition

[0157] The calibration system receives calibration commands initiated by users or calibration tasks triggered by the system at regular intervals. These commands or tasks include parameters such as the gas station code, tank number, calibration start time, and end time. Based on these parameters, the system calls the data service interface to obtain all refueling transaction records, continuous monitoring data from the level gauge, and currently used tank capacity table data for the corresponding tank within a specified time period. This data forms the basis for subsequent calibration calculations.

[0158] S2: Data Preprocessing

[0159] Time-series correlation matching of refueling records and level gauge data

[0160] Iterate through each refueling record and perform the following algorithm operation:

[0161] a. For the current refueling record, the time of nozzle hanging up is used as the baseline time. .

[0162] b. A binary search algorithm is used to quickly locate the level gauge record L whose timestamp is closest to T_ref in the continuous monitoring data sequence of the level gauge. This algorithm significantly improves the search efficiency in massive time-series data.

[0163] c. Based on the preset forward search time window T_front and backward search time window T_back (both durations are configurable parameters of the system and can be flexibly adjusted according to scenarios such as vehicle queuing time and fuel nozzle turnover efficiency during peak hours at gas stations, such as T_front being 3-5 minutes and T_back being 1-2 minutes).

[0164] d. Associate and bind all level gauge data found within this time period with the refueling record to form a data association unit. This unit completely includes the entire process of level change caused by this refueling action. Calculate the "initial level volume" and "level volume at the moment of nozzle hanging" using the tank capacity table parameters; the difference between the two is the tank capacity table change. Simultaneously, extract the "nominal fuel quantity" (the amount of fuel refueled as displayed by the fuel dispenser) from the refueling record, and calculate the error using "nominal fuel quantity - tank capacity table change".

[0165] Detection and merging of overlapping data association units

[0166] Iterate through all data association units generated in the above steps and detect the temporal overlap relationship between adjacent units:

[0167] a. Determine whether there is time overlap between two adjacent data association units (denoted as unit A and unit B). The specific judgment condition is: whether the timestamp of the last data in the liquid level gauge data associated with unit A is later than the timestamp of the first data in the liquid level gauge data associated with unit B.

[0168] b. If the above overlap exists, a merge operation is triggered. Cell A and Cell B are merged into a new composite data cell. The attributes of this composite data cell are generated according to the following rules:

[0169] Combined refueling volume: The refueling volume of the new unit is the sum of the refueling volumes of the original unit A and unit B.

[0170] Merging level gauge data sequences: The level gauge data sequence associated with the new unit is the union of the level gauge data sequences of the original unit A and unit B, and is reordered according to the timestamp.

[0171] c. This merging process is performed iteratively. It checks whether the merged new unit overlaps with its subsequent units until there is no longer any temporal overlap between all adjacent data units. This step effectively resolves issues caused by continuous refueling or level gauge recording.

[0172] The results of merging temporal correlation matching and overlapping data correlation units are shown in Table 1 below:

[0173] Table 1

[0174]

[0175] Intelligent truncation of pure consumption intervals

[0176] Based on the merged final data units, the system further identifies and extracts the effective pure consumption range for tank capacity calibration:

[0177] a. Analyze the liquid level gauge data sequence within each data unit and observe the trend of liquid level height change over time.

[0178] b. The system identifies and extracts only continuous data intervals where the liquid level shows a continuously decreasing trend. This interval represents the simple refueling and consumption process of the oil tank, without interference from other events (such as unloading).

[0179] c. The final extracted continuous data interval showing a continuously decreasing liquid level height is used as the effective pure consumption interval (i.e., the area between the peaks and troughs), that is, the complete data segment between two oil inflow events, and serves as the valid input for subsequent calibration calculations. The pure consumption interval extracted by this invention is as follows: Figure 4 As shown.

[0180] S3: Algorithm Strategy Selection and Execution. If an initial tank capacity table exists, skip geometric constraints. The system employs a parallel fitting and interpolation strategy. The initial height and volume correspondence within the tank is as follows: Figure 3 As shown. The core process of interpolation is as follows: 1. Reverse volume calculation: Call the actual volume calculation algorithm. Starting from the record with the lowest liquid level (trough), and using the height of its ending liquid level and the corresponding volume as a reference, the refueling volume of each record is accumulated backward in the reverse refueling sequence, calculating the actual volume at the beginning of the previous record for each record, and performing temperature compensation on the actual volume. During this process, it is possible to select whether to propagate the error to subsequent records through configuration parameters.

[0181] 2. Collect calibration points: Call the calibration point acquisition method, combine the starting liquid level height of each record (refueling record, liquid level gauge height) with its corresponding calculated actual volume to form calibration reference points, sort them by liquid level height, and record the global minimum height minHeight (trough) and the maximum height maxHeight (peak) simultaneously.

[0182] 3. Piecewise linear interpolation and volume revision: Traverse each height scale point h of the original tank volume table

[0183] If h < minHeight, the volume value at this height point remains unchanged;

[0184] If h > maxHeight, calculate the difference delta between the actual volume at height minHeigh and the volume at height minHeigh in the initial tank volume table, and increase the volumes corresponding to the heights higher than minHeigh in the initial tank volume table by delta;

[0185] If h is between minHeight and maxHeight, call the linear interpolation algorithm: First, find the data unit corresponding to h through binary search, and record the highest height and the lowest height , the corresponding volume and the corresponding volume , and then calculate the actual volume at height h according to the following formula:

[0186] ;

[0187] The execution of the fitting process includes the following steps: The first two steps are the same as those described in the above interpolation strategy. In the third step, the system calls the polynomial regression algorithm, uses the liquid level heights of all calibration points as the independent variable X and the actual volume as the dependent variable Y to perform Nth-order polynomial fitting (where the order N is a configurable parameter, such as y = β0 + β1X + β2X² + … + β n Xⁿ), so as to obtain a continuous and smooth volume-height characteristic curve function V = f(h). The system recalculates the new volume values corresponding to each height point according to this function.

[0188] 4. Optimal Tank Capacity Table Output: The optimal tank capacity table output is achieved through the aforementioned residual evaluation strategy, which employs a multi-version comprehensive evaluation algorithm at its core. This algorithm evaluates the overall deviation while fully considering the shift management cycle in the daily operation of the gas station, thereby selecting the tank capacity table version with the highest accuracy and optimal stability. The specific process is as follows: The residual method is used to select between the volume value output by piecewise linear interpolation and the fitted volume value.

[0189] Data preparation and time period division: Based on the evaluation parameters selected by the user, including the target gas station, oil tank, evaluation period (recommended to be the past month) and the daily shift end time of the gas station, the system automatically divides the evaluation period into multiple continuous sub-periods with "shift end day" as the unit.

[0190] Multiple version data loading: The system loads multiple candidate tank capacity table versions (Candidate_Version_1, Candidate_Version_2, ..., Candidate_Version_M) generated for the same tank during the calibration process. These versions can be generated based on different calibration algorithms (such as interpolation or fitting methods) or different data periods.

[0191] Iterative evaluation calculation: For each candidate tank capacity table version, the system performs the following evaluation steps:

[0192] Daily capacity backtracking calculation: Iterate through the end-of-shift period, and using the current candidate tank capacity table version, interpolate the liquid level height value h in each level gauge record within that period to obtain the corresponding theoretical volume value. ;

[0193] a. Single-entry error calculation: For each refueling record within the shift's closing day, locate the level gauge record closest to its start and end times, and calculate the theoretical volume change. And the actual amount of fuel dispensed is measured by the fuel dispenser. By comparison, the error per transaction is obtained. ;

[0194] b. Take the first and last level gauge records for each shift's closing date and calculate the theoretical total change for that day. The overall error of the shift's closing date is obtained by comparing it with the total actual refueling volume ΣQ_sale within the shift's closing date. ;

[0195] d. Summary Evaluation Indicators: After traversing all shift completion dates, calculate the two core indicators for the current candidate tank capacity table version:

[0196] Sum of absolute values ​​of shift-end errors (SADE): This is used to assess the cumulative deviation of the tank capacity table over multiple days of operation; the smaller the value, the better the long-term stability.

[0197] Sum of Squared Residuals (SSE) of Individual Errors: The value is used to measure the accuracy of the tank capacity table at the level of a single transaction; the smaller the value, the higher the local accuracy.

[0198] Optimal Version Selection: The system selects the optimal version for the final tank capacity table based on the following rules:

[0199] Primary rule: Select the candidate version with the lowest SADE value to prioritize the stability and accuracy of the tank capacity table at the macro level during long-term operation;

[0200] Secondary rule: If multiple candidate versions have the same SADE value or the difference is less than a set threshold, the version with the smallest SSE is selected, thus giving preference to the version with higher local accuracy under the condition of comparable macroscopic stability.

[0201] Output results: The system will use the final selected optimal tank capacity table version as the official calibration result, persistently store it in the database, and provide call support for downstream businesses such as level gauge systems and inventory management systems.

[0202] This method innovatively introduces "shift end date" as the basic evaluation unit, making the evaluation process consistent with the actual operation and management cycle of the gas station. Through the comprehensive evaluation of SADE and SSE two-level indicators, it takes into account both long-term stability and local accuracy, thereby significantly improving the overall accuracy and reliability of inventory management.

[0203] Example 2:

[0204] For newly built oil tanks without an initial tank capacity table or specific application scenarios, the system first executes a geometric constraint strategy. The system provides a parameter input interface to receive tank structure parameters submitted by the user, such as... Figure 5 As shown, this includes tank length L, radius R, elliptical head height e, etc. Using preset intervals (e.g., 10 mm) as steps, the system automatically calculates the theoretical volume corresponding to each liquid level height h, and generates a complete theoretical tank capacity table as the initial tank capacity table. After this, the system continues to execute steps S1, S2, S3, and S4 as described in Example 1 to achieve further dynamic calibration and optimization of the tank capacity table.

[0205] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any equivalent substitutions or modifications made based on the technical content disclosed in the present invention should be included within the scope of protection defined by the claims of the present invention.

Claims

1. A method for dynamic calibration of tank capacity tables based on multi-source data fusion, characterized in that, Specifically: Data acquisition: Acquires data from the fuel nozzle and level gauge, with all acquired data accompanied by timestamps accurate to the second. The data from the fuel nozzle and the level gauge are standardized and packaged to eliminate data format differences and form a standardized data package. The data in the standardized data packet is preprocessed, including synchronizing the timestamps of the oil gun sales data and the liquid level gauge data to construct corresponding data units; merging two adjacent data units that overlap in time; extracting pure consumption intervals; and verifying abnormal data. If the tank has an initial capacity table, the FGIR four-dimensional calibration algorithm is used to calibrate the initial capacity table of the tank; if there is no initial capacity table, an initial capacity table is constructed for the tank based on a geometric constraint strategy, and then the FGIR four-dimensional calibration algorithm is used to calibrate the initial capacity table of the tank.

2. The method for dynamic calibration of tank capacity tables based on multi-source data fusion according to claim 1, characterized in that, The fuel nozzle refueling data includes the gas station code, fuel tank number, fuel nozzle number, refueling volume, and nozzle hanging time. The level gauge data includes: the oil level in the tank, basic data reflecting the remaining oil volume in the tank in real time, the tank temperature, and the water level at the bottom of the tank.

3. The method for dynamic calibration of tank capacity tables based on multi-source data fusion according to claim 1, characterized in that, Synchronize the timestamps of the fuel nozzle sales data with the liquid level gauge data, specifically as follows: Based on the current refueling record, the time when the nozzle was hung up was used as the baseline time. ; A binary search algorithm is used to search for the timestamp and... in the continuous monitoring data sequence of the level gauge. The closest level gauge record L; Based on the preset forward lookup time window Backward lookup time window To record the timestamp corresponding to L Centered on, find the timestamp located at All level gauge data within the range; against All level gauge data within the interval are first processed, and invalid data with abnormal formats are removed. The remaining valid data are then sorted according to the difference between the acquisition time and the nozzle hanging time. The sorted data is then bound to the current refueling record to form a data association unit, and calculations are performed. The change in liquid level gauge height within the interval is recorded and compared with the nominal oil volume of the corresponding refueling record. The error is calculated, and if the error exceeds a preset error threshold, the verification process is stopped and an alarm is triggered. If the difference between the initial liquid level gauge height of the later data unit and the ending liquid level gauge height of the previous data unit is greater than a preset difference threshold, the verification process is stopped and an alarm is triggered. All refueling records are traversed sequentially to complete the construction of all data association units.

4. The method for dynamic calibration of tank capacity tables based on multi-source data fusion according to claim 1, characterized in that, The extraction of pure consumption intervals is as follows: A set is formed based on all data units. The height at which the liquid level begins to change in each data unit is recorded as... ; Compare two adjacent data points If the absolute value of the difference is less than or equal to the threshold, then the two points are considered not to constitute a trend change. If neither point t nor point T constitutes a trend change, then points t to T are considered a unified node. If there is an increasing trend after point T, then within the range of points t to T, [the following is considered]. The largest value is used as the value corresponding to the t-th point to the T-th point; if the value decreases after the T-th point, then within the range of the t-th point to the T-th point, select... The minimum value is taken as the value corresponding to the point from the t-th point to the T-th point; Then record the height of the level gauge and the time corresponding to the peak and trough; Periodic alternating verification: If two or more consecutive peaks or two or more consecutive troughs are detected, it is judged as a logic error, the verification process is interrupted, and an alarm is triggered. The interval between the correct peak and trough is taken as the pure consumption interval.

5. The method for dynamic calibration of tank capacity tables based on multi-source data fusion according to claim 1, characterized in that, The initial tank capacity table of the oil tank is calibrated using the FGIR four-dimensional calibration algorithm as follows: Step 1: Call the actual capacity calculation algorithm, starting from the lowest point of the pure consumption interval, and in reverse refueling time sequence, accumulate the refueling amount of each data unit record forward in sequence, and calculate the actual volume at the beginning of the previous record for each record. Step 2: Call the calibration point acquisition method, combine the initial liquid level height recorded in each data unit with its corresponding calculated actual volume to form a calibration reference point, and perform temperature compensation on the actual volume according to the following formula, and then sort them according to the liquid level height. ； in, To calculate the actual volume, The volume at a standard temperature of 20°C This refers to the coefficient of volumetric expansion of the oil product. Step 3: Periodically revise the initial tank capacity table using piecewise linear interpolation and linear fitting. Step 4: Select the final tank table as the calibrated tank table from the multiple candidate tank tables obtained by piecewise linear interpolation, or from the multiple candidate tank tables obtained by linear fitting, or from both the multiple candidate tank tables obtained by piecewise linear interpolation and the multiple candidate tank tables obtained by linear fitting.

6. The method for dynamic calibration of tank capacity tables based on multi-source data fusion according to claim 5, characterized in that, The initial tank capacity table is revised using piecewise linear interpolation, specifically as follows: If the initial scale h in the tank gauge is < If the volume value corresponding to that height point in the tank table remains unchanged; If h > maxHeight, calculate the difference delta between the actual volume at height minHeight and the volume at height minHeight in the initial tank table, and increase the volume corresponding to the height in the initial tank table that is higher than minHeight by delta; minHeightwei is the lowest liquid level in the pure consumption range, and maxHeight is the highest liquid level in the pure consumption range; If minHeight≤h≤maxHeight, then the highest height in the data cell corresponding to h is recorded. and minimum height , Corresponding volume as well as Corresponding volume Then, calculate the actual volume at height h using the following formula: 。 7. The method for dynamic calibration of tank capacity tables based on multi-source data fusion according to claim 5, characterized in that, Step 4 specifically involves: Step 4.1: Select the evaluation parameters, including the target gas station, oil tank, evaluation period, and the daily shift end time of the gas station. The evaluation period will be automatically divided into multiple continuous sub-periods based on the shift end day. Step 4.2: For any candidate tank capacity table, iterate through the end-of-shift period and interpolate the liquid level height value in each level gauge record within that period to obtain the corresponding theoretical volume value. ; Step 4.3: For each refueling record within the shift's closing day, locate the level gauge record closest to its start and end times. and Calculate the theoretical volume change And compare it with the actual amount of fuel dispensed by the fuel dispenser in the refueling record. By comparison, the error per transaction is obtained. : ； ； Step 4.4: Take the first and last level gauge records for each shift's closing date. and Calculate the total theoretical change for the day. ; and all actual refueling volume within the day of the class's conclusion. Comparison, the overall error of the day's work is obtained. : ； ； Step 4.5: After iterating through all shift closing dates, calculate the sum of the absolute values ​​of the shift closing date errors (SADE) and the sum of squared residuals of individual errors (SSE): ； ； Step 4.6: Select the candidate version with the smallest SADE value. If there are multiple candidate versions with equal SADE values ​​or the difference between SADE values ​​is less than the set threshold, these multiple candidate versions are set as set A, and the version with the smallest SSE is selected from set A.

8. The method for dynamic calibration of tank capacity tables based on multi-source data fusion according to claim 1, characterized in that, The geometric constraint-based strategy for constructing the initial tank capacity table for the oil tank uses the following formula: ； Where L refers to the transverse length of the central cylinder of the horizontal oil tank; R refers to the radius of the end caps at both ends of the horizontal oil tank. The markings indicate the height of the fuel level inside the tank; among which... , Indicates the height in the initial tank table. The corresponding volume.

9. A system applied to the dynamic calibration method for tank capacity tables based on multi-source data fusion as described in claim 1, characterized in that, include The liquid level gauge data interface module is used to collect liquid level gauge data, and this module is compatible with multiple brands. The refueling transaction data interface module is used to collect refueling data from fuel nozzles; this module has flexible multi-source data access capabilities. The data upload module is used to standardize and encapsulate the data collected by the liquid level gauge data interface module and the refueling transaction data interface module into standardized data packets and upload them to the cloud processing center. At the same time, it monitors the data upload progress and success rate in real time, and promptly sends a warning to the alarm module through remote alarm information when abnormal situations such as transmission failure occur. A cloud-based processing center is used for data preprocessing; The FGIR four-dimensional calibration algorithm module is used to calibrate the initial tank table based on preprocessed data; The alarm module is used to alert the system to any abnormal situations that occur during the real-time detection and calibration process.

10. A dynamic calibration system for tank capacity tables based on multi-source data fusion according to claim 9, characterized in that, The liquid level gauge data interface module and the refueling transaction data interface module have built-in local cache units. When the network is temporarily interrupted or the cloud service is temporarily unavailable, the liquid level gauge data interface module and the refueling transaction data interface module will store the corresponding data in the local cache unit. When the network is restored or the cloud service is normal, the data upload module will automatically start the breakpoint resume function and upload the cached data in the order of data generation time. The data upload module uses the HTTPS network protocol; The abnormal situations include: oil tank leakage, insufficient number of valid data points for calibration, and single calibration error exceeding a preset threshold.

Images