A digital-twin-based pottery jar wine damage grading pre-warning method, device and system

Abstract

This application discloses a method, device, and system for graded early warning of wine loss in ceramic jars based on digital twins, belonging to the field of wine loss detection technology. The method includes: establishing a digital twin for the ceramic jar and generating a unique individual wine loss baseline based on multi-dimensional historical loss data of the jar; simultaneously receiving fixed high-density continuous liquid level data and portable discrete inspection liquid level data, and calculating the loss deviation by matching the corresponding data with the jar's equipment mounting status; comparing the deviation with a dynamic early warning threshold to generate a graded early warning label, and automatically issuing a portable retest task to obtain the retest liquid level; correcting the jar's liquid level baseline and marking the sensor status based on the retest data, and adaptively updating the early warning threshold. This technical solution can achieve accurate early warning, significantly reducing the overall hardware deployment cost of the warehouse, while also reducing false alarms caused by equipment drift, and improving the level of intelligence in wine loss management.

Description

Technical Field

[0001] This application belongs to the field of wine damage detection technology, specifically relating to a method, device and system for graded early warning of wine damage in ceramic jars based on digital twins. Background Technology

[0002] During the aging process of baijiu (Chinese liquor), the loss of liquor due to the microporous structure of the earthenware jars includes both normal evaporation and abnormal leakage. Large-scale liquor storage facilities often contain tens of thousands of earthenware jars. Traditional monitoring methods are limited by significant shortcomings in their technical architecture. For example, while fixed online monitoring equipment offers millimeter-level accuracy and continuous data acquisition capabilities, its high cost per point limits its coverage to only a small number of high-value jars, failing to achieve universal coverage across the entire storage facility. Portable inspection equipment, on the other hand, is cheaper and has a wider coverage area, but relies on manual operation of each jar, resulting in high data dispersion and inspection intervals often lasting several months, making it difficult to detect early signs of leakage. Furthermore, existing early warning mechanisms generally use a uniform threshold method, ignoring individual differences between jars from different manufacturers, batches, and ages, leading to high false alarm and false negative rates. How to achieve effective coverage of the entire storage facility while controlling hardware costs, and how to establish a precise anomaly detection mechanism based on the individual characteristics of each earthenware jar, has become an industry challenge hindering the intelligent management of baijiu storage and transportation. Summary of the Invention

[0003] This application provides a method, device, and system for graded early warning of wine damage in ceramic jars based on digital twins. The purpose is to solve the technical problems in existing monitoring of wine damage in ceramic jars, such as the high cost of fixed equipment that cannot achieve full coverage, the low frequency of portable equipment that makes it difficult to detect leakage in a timely manner, and the high false alarm and false negative rates caused by the uniform threshold judgment that ignores individual differences.

[0004] In a first aspect, embodiments of this application provide a method for graded early warning of wine spoilage in ceramic jars based on digital twins, the method comprising: Step 1, Construct a set of digital twins: Establish an independent digital twin for each earthenware jar in the wine storage cellar, and configure an individual wine loss baseline for each digital twin. The individual wine loss baseline is obtained by fitting historical loss data based on the manufacturer attributes, batch attributes, and years of use of the corresponding earthenware jar. Step 2, collect dual-modal liquid level data: receive the first liquid level time series data uploaded by the fixed liquid level gauge at a preset sampling frequency, and the second liquid level time series data uploaded by the portable liquid level gauge in response to the inspection task. The time resolution of the first liquid level time series data is higher than that of the second liquid level time series data. Step 3, calculate the real-time deviation: Based on the equipment mounting status of the target ceramic vat, call the corresponding first liquid level time series data or second liquid level time series data, substitute it into the individual wine loss baseline for calculation, and obtain the deviation between the measured value of the current wine loss status and the baseline. Step 4, generate warning identifier: compare the deviation with a preset dynamic warning threshold. When the deviation exceeds the dynamic warning threshold, generate a corresponding warning level identifier and write the warning level identifier into the digital twin of the target ceramic jar. Step 5, Trigger Retest Verification: In response to the generation of the warning level identifier, automatically generate retest prompt information, and dispatch a portable measuring terminal to conduct on-site retesting of the target ceramic vat to obtain retest liquid level data; Step 6, Automatic parameter update: Based on the comparison results between the re-measured liquid level data and the measured liquid level time series data, update the liquid level reference value or sensor health status identifier in the digital twin of the target ceramic tank, and adaptively adjust the dynamic early warning threshold according to the early warning and handling feedback results of all ceramic tanks in the warehouse.

[0005] Furthermore, in step 1, the fitting process for the individual alcohol impairment baseline includes: If the historical loss data of the target ceramic jar is insufficient, the mean monthly loss rate μ and standard deviation σ of the reference ceramic jar set with the same manufacturer attributes, batch attributes and initial alcohol content attributes are calculated in the first year of use as the initial parameters of the individual wine loss baseline. If the historical loss data of the target ceramic jar reaches the preset duration, an individualized loss baseline is obtained by fitting the liquid level time series data of the ceramic jar itself.

[0006] Furthermore, in step 1, for earthenware jars whose service life exceeds a preset age threshold, an aging coefficient is introduced into the individual wine damage baseline to compensate for the long-term drift trend caused by the aging of the earthenware jar's microporous structure.

[0007] Furthermore, the individual wine loss baseline also includes a temperature and humidity coupled correction function, which is used to normalize the wine loss rate according to changes in ambient temperature and humidity.

[0008] Furthermore, in step 3, when the target ceramic tank is equipped with a fixed liquid level timer, the first liquid level time series data is called up to calculate the liquid level change rate V(t) and the liquid level change acceleration A(t). When the target ceramic tank is not equipped with a fixed liquid level timer, the second liquid level time sequence data is called up, and the daily average loss rate is calculated based on the liquid level difference between two adjacent inspections.

[0009] Furthermore, in step 4, the warning level identifier includes at least: A blue warning indicator signifies that the acceleration of liquid level change changes from fluctuating around zero to a sustained positive value. A yellow warning sign indicates that the rate of change in liquid level or the average daily loss rate exceeds the individual wine loss baseline μ+3σ. An orange alert indicates that the rate of change in liquid level remains positive and shows no signs of slowing down during the nighttime cooling period. A red warning sign indicates a sudden drop in liquid level within a short time window.

[0010] Furthermore, in step 6, the retested liquid level data is collected by the portable measuring terminal after performing standardized calibration before each use; Accordingly, the method further includes: The deviation between the retested liquid level data and the first liquid level time series data is calculated. When the deviation exceeds the preset tolerance range, it is determined that the fixed liquid level gauge has drifted, and the sensor health status indicator is updated.

[0011] Furthermore, in step 6, the adaptive adjustment of the dynamic early warning threshold includes: The ceramic jars in the warehouse were grouped according to manufacturer, batch, and years of use, and the early warning hit rate and false alarm rate of each group were calculated. For groups whose warning hit rate is higher than the first threshold, their dynamic warning threshold is lowered to improve sensitivity; For groups with false alarm rates exceeding the second threshold, their dynamic warning thresholds are increased to reduce the false alarm rate. During periods of high summer temperatures, the dynamic warning thresholds for all groups are uniformly lowered by a preset percentage.

[0012] Secondly, embodiments of this application provide a digital twin-based early warning device for graded spoilage of ceramic wine jars. The device is deployed on a digital twin management platform and includes: The digital twin building block is used to construct a collection of digital twins and configure an individual alcohol damage baseline for each digital twin; The data acquisition module is used to receive first liquid level time series data uploaded by a fixed liquid level gauge and second liquid level time series data uploaded by a portable liquid level gauge. The time resolution of the first liquid level time series data is higher than that of the second liquid level time series data. The deviation calculation module is used to call the corresponding liquid level time series data and calculate the deviation from the individual wine loss baseline based on the equipment mounting status of the target ceramic vat. The early warning generation module is used to compare the deviation with the dynamic early warning threshold, generate an early warning level identifier, and write it into the digital twin; The retest scheduling module is used to generate retest prompt information in response to the warning level identifier and to schedule portable measurement terminals to perform on-site retests. The parameter update module is used to update the liquid level reference value or sensor health status indicator based on the retest results, and adaptively adjust the dynamic warning threshold.

[0013] Thirdly, embodiments of this application provide a digital twin-based early warning system for graded spoilage of ceramic wine jars, the system comprising: Fixed level gauges are deployed in some key ceramic vats in the wine storage cellar to collect first liquid level time-series data at a preset sampling frequency; A portable measuring terminal is used to perform on-site retesting of the target ceramic jar and collect retest liquid level data when a retest prompt is received; The digital twin management platform is communicatively connected to the fixed level gauge and the portable measuring terminal, and is configured to execute the digital twin-based method for classifying and warning of wine loss in ceramic jars as described above.

[0014] Fourthly, embodiments of this application provide a readable storage medium on which a program or instructions are stored, which, when executed by a processor, implement the steps of the method described above.

[0015] Fifthly, embodiments of this application provide a chip, the chip including a processor and a communication interface, the communication interface being coupled to the processor, the processor being used to run programs or instructions to implement the method described above.

[0016] The technical solution provided in this application, by constructing a digital twin of the ceramic jar and an individualized wine loss baseline, breaks through the limitations of the traditional one-size-fits-all threshold, realizing a precise early warning strategy with one standard per jar. By utilizing the collaborative processing of fixed high-frequency data and portable discrete data, it significantly reduces the hardware deployment cost of the entire warehouse while ensuring millimeter-level monitoring accuracy. At the same time, through retesting and adaptive updating of sensor health status, it effectively eliminates false alarms caused by equipment drift, realizes feedback dynamic optimization of threshold parameters, and enables the system to have the self-evolution capability of becoming more accurate with use, significantly improving the level of intelligence in wine loss management. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the method for graded early warning of wine loss in ceramic jars based on digital twins provided in Embodiment 1 of this application; Figure 2 This is a schematic diagram of the overall process provided in the embodiments of this application; Figure 3 Flowchart of the logic for graded early warning of fixed liquid level gauge ceramic tank; Figure 4 Flowchart for cross-validation of fixed and portable data; Figure 5 A closed-loop feedback diagram for group pattern learning and threshold adaptive optimization; Figure 6 This is a schematic diagram of the structure of the ceramic jar wine damage classification and early warning device based on digital twin provided in Embodiment 2 of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, specific embodiments of this application will be described in further detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely for explaining this application and not for limiting it. It should also be noted that, for ease of description, only the parts relevant to this application are shown in the drawings, not all of them. Before discussing exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe operations (or steps) as sequential processes, many of these operations can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but may also have additional steps not included in the drawings. The process can correspond to a method, function, procedure, subroutine, subroutine, etc.

[0019] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0020] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0021] The following detailed description, in conjunction with the accompanying drawings, of the digital twin-based method for classifying and warning of wine damage in ceramic jars provided in this application, through specific embodiments and application scenarios, will be provided in detail.

[0022] Example 1 Figure 1 This is a flowchart illustrating the method for graded early warning of wine spoilage in ceramic jars based on digital twins, provided in Embodiment 1 of this application. Figure 1 As shown, the method specifically includes: Step 1, Construct a set of digital twins: Establish an independent digital twin for each earthenware jar in the wine storage cellar, and configure an individual wine loss baseline for each digital twin. The individual wine loss baseline is obtained by fitting historical loss data based on the manufacturer attributes, batch attributes, and years of use of the corresponding earthenware jar. A wine storage cellar refers to a dedicated warehouse space used to store ceramic jars and carry out long-term aging of baijiu (Chinese liquor), such as an underground constant-temperature wine cellar or a sealed wine room on the ground.

[0023] Earthenware jars refer to ceramic containers with a natural microporous structure used for storing raw wine for long-term aging, such as 1000L unglazed earthenware jars and 500L glazed earthenware jars.

[0024] A digital twin is a digital data carrier that corresponds one-to-one with a single physical ceramic jar throughout its entire lifecycle. For example, it can be a structured database form that stores jar information, wine storage records, measurement data, and early warning records.

[0025] Individual wine loss baseline refers to the standard mathematical model of normal wine loss specific to a single earthenware jar, such as the daily, weekly or monthly loss rate range model under normal operating conditions for that earthenware jar.

[0026] Manufacturer attributes refer to the unique identification information of ceramic jar manufacturers, such as two types of manufacturer markings: A ceramic factory and B ceramic factory.

[0027] Batch attribute refers to the production batch number of the ceramic jar when it leaves the factory, such as the 2015 batch or the 2022 batch.

[0028] The service life dimension refers to the cumulative usage time of the earthenware jar from the day it was put into use to the present, such as 3 years of use, 10 years of use, etc.

[0029] Historical loss data refers to the loss rate sequence calculated from each liquid level measurement of the ceramic jar, such as monthly loss rate data for 12 consecutive months.

[0030] In this solution, the digital twin management platform can batch import or manually input basic information about earthenware jars, assigning a unique code to each physical jar to generate a dedicated digital file, thus completing a one-to-one mapping between physical jars and digital files. The platform can then retrieve multi-dimensional historical loss data for the corresponding jar, generate a dedicated loss model using a built-in fitting algorithm, and bind and store it within the corresponding digital twin. Next, it reads historical loss datasets based on three dimensions: manufacturer, batch, and years of use, calculates the standard loss range using regression analysis, and outputs all parameters of the individual wine loss baseline.

[0031] Step 2, collect dual-modal liquid level data: receive the first liquid level time series data uploaded by the fixed liquid level gauge at a preset sampling frequency, and the second liquid level time series data uploaded by the portable liquid level gauge in response to the inspection task. The time resolution of the first liquid level time series data is higher than that of the second liquid level time series data. Dual-modal liquid level data refers to liquid level datasets with two completely different acquisition modes and acquisition densities, such as time-series data acquired continuously by fixed equipment and time-series data acquired discretely by portable equipment.

[0032] Fixed level gauges refer to non-contact level sensors with millimeter-level accuracy that are fixedly installed above ceramic tanks. For example, they can be infrared level sensors with an accuracy of ±1mm, deployed only in 5% to 15% of the high-value ceramic tanks in the warehouse.

[0033] The preset sampling frequency refers to the automatic upload cycle of sensor data configured in advance by the platform, such as collecting liquid level and temperature data once per hour.

[0034] The first liquid level time series data refers to the continuous liquid level sequence pushed by the fixed liquid level gauge at regular intervals and marked with high-density timestamps.

[0035] Inspection tasks refer to periodic on-site measurement work orders automatically issued by the platform based on the risk level of the ceramic jar. For example, it could be a quarterly inspection work order for a normal ceramic jar or a monthly inspection work order for a ceramic jar with a high risk of leakage.

[0036] Portable level gauges refer to manual, mobile level detection devices with millimeter-level accuracy. For example, a handheld level detector with an integrated infrared thermal imaging module can be used to measure all ordinary ceramic tanks in the storage room.

[0037] The second liquid level time series data refers to the discrete liquid level data collected by the portable liquid level gauge according to the inspection work order interval, for example, only one liquid level measuring point is generated every three months.

[0038] Temporal resolution refers to the number of liquid level measurement points that can be acquired per unit of time, such as hourly high resolution and quarterly low resolution.

[0039] Specifically, the digital twin management platform can receive monitoring data pushed by the fixed liquid level gauge in real time through the wireless communication link. After the portable terminal completes the cylinder measurement, the platform synchronously receives the inspection liquid level data returned by the terminal. After the portable measuring terminal receives the inspection work order issued by the platform, the operator completes the ceramic tank liquid level detection on site, and the terminal automatically returns the liquid level time sequence data measured this time to the digital twin management platform.

[0040] Step 3, calculate the real-time deviation: Based on the equipment mounting status of the target ceramic vat, call the corresponding first liquid level time series data or second liquid level time series data, substitute it into the individual wine loss baseline for calculation, and obtain the deviation between the measured value of the current wine loss status and the baseline. Among them, the target earthenware jar refers to any earthenware jar used for wine storage that the platform is currently performing wine loss analysis and early warning judgment, such as the ten-year-old wine jar with the number T-2015-B-0012.

[0041] Equipment mounting status refers to the pre-marked hardware deployment identifier of the ceramic tank within the digital twin, such as two types of status markers: mounted fixed level gauge and not mounted fixed level gauge.

[0042] The measured value is the current actual wine loss rate of the ceramic jar, calculated from the liquid level time series data, for example, an average daily loss of 0.001%.

[0043] Deviation refers to the numerical difference between the current measured rate of wine loss and the baseline standard rate of wine loss for an individual. It is used to quantify the degree to which the loss deviates from the normal range. For example, the measured loss exceeds the baseline standard by 0.02%.

[0044] Specifically, the digital twin management platform can retrieve the corresponding first or second liquid level time-series data. It reads the device mounting status identifier stored within the digital twin of the target ceramic vat and automatically matches and retrieves the corresponding type of liquid level time-series dataset based on the identifier, eliminating the need for manual data filtering. The digital twin management platform then inputs the retrieved liquid level time-series data into a pre-stored individual wine loss baseline mathematical model. The program calculates the difference between the measured loss value and the baseline standard value, outputting the deviation value.

[0045] Step 4, generate warning identifier: compare the deviation with a preset dynamic warning threshold. When the deviation exceeds the dynamic warning threshold, generate a corresponding warning level identifier and write the warning level identifier into the digital twin of the target ceramic jar. The dynamic early warning threshold refers to the loss deviation from the critical value that is bound to the ceramic jar group and can be automatically adjusted according to the season and leakage risk, such as μ+2.5σ for high-risk jar group and μ+3σ for low-risk jar group.

[0046] Warning level identifiers are tiered labels defined by the platform to distinguish the severity of abnormal losses, such as blue warning codes, yellow warning codes, orange warning codes, and red warning codes.

[0047] The digital twin management platform can read the currently calculated deviation value and compare it with the dynamic warning threshold of the corresponding group of the ceramic jar. When the deviation value exceeds the threshold threshold, the platform program matches the corresponding severity level according to the degree of deviation and automatically generates a warning mark in the form of a code. Then, the generated warning level identifier, warning generation timestamp, and corresponding liquid level data are synchronously stored in the digital twin file of the target ceramic jar to permanently retain the warning record.

[0048] Step 5, Trigger Retest Verification: In response to the generation of the warning level identifier, automatically generate retest prompt information, and dispatch a portable measuring terminal to conduct on-site retesting of the target ceramic vat to obtain retest liquid level data; The retest prompt information refers to the on-site verification work order data automatically generated by the platform. It can be an electronic task order that includes the ceramic jar number, warning level, and on-site testing requirements.

[0049] A portable measuring terminal refers to a handheld detection device used by operators for on-site retesting of liquid levels. It can be a handheld terminal that integrates liquid level and infrared thermal imaging detection functions.

[0050] The retested liquid level data refers to the standard liquid level value obtained by taking the average of multiple measurements on site using a portable measuring terminal. It can be the average liquid level of three measurements, such as 127.1 cm.

[0051] In this solution, after the platform program detects a new warning level identifier in the digital twin, it automatically initiates the retesting and verification process. This includes capturing the target ceramic tank number, warning level, and tank location information, automatically assembling a standardized retesting task work order, and sending the retesting prompt information to the portable measuring terminal held by the corresponding inspection personnel. The on-site retesting task is then pushed to the operator who carries the portable measuring terminal to the target ceramic tank to complete the standardized retesting. The terminal collects the liquid level value and transmits it back to the platform for storage.

[0052] Step 6, Automatic parameter update: Based on the comparison results between the re-measured liquid level data and the measured liquid level time series data, update the liquid level reference value or sensor health status identifier in the digital twin of the target ceramic tank, and adaptively adjust the dynamic early warning threshold according to the early warning and handling feedback results of all ceramic tanks in the warehouse.

[0053] Measured liquid level time series data refers to the liquid level sequence continuously collected by a fixed liquid level gauge before the warning is triggered, such as continuous liquid level data for 72 hours before the warning.

[0054] The liquid level reference value refers to the standard liquid level reference value of the ceramic tank stored in the digital twin, such as the standard liquid level of 127.1 cm after portable calibration.

[0055] Sensor health status label refers to a label that marks the working status of a fixed liquid level gauge, and can be classified into three categories: normal, drift fault, and pending maintenance.

[0056] The early warning and response feedback results refer to the response records that are sent back to the platform by on-site personnel after completing retesting and maintenance, such as the response conclusions for leakage repair, sensor drift, and seal aging.

[0057] The digital twin management platform can extract the average value of the retested liquid level and the fixed synchronous liquid level during the early warning period, calculate the deviation between the two, and update the cylinder liquid level benchmark with the retested data if the deviation is within the tolerance; if the deviation exceeds the limit, the sensor health status label is changed to drift fault. The platform summarizes all early warning handling records in the entire database, automatically corrects the early warning threshold value of each group according to the cylinder group, and completes the threshold iterative update.

[0058] This technical solution uses computer-automated calculations to replace manual experience-based judgment by digitally archiving ceramic jars, collecting dual-modal liquid level data, calculating personalized losses, and providing tiered early warnings. It balances monitoring coverage with early warning accuracy and enables long-term autonomous iterative optimization of the system.

[0059] In one embodiment, optionally, in step 1, the fitting process of the individual alcohol impairment baseline includes: If the historical loss data of the target ceramic jar is insufficient, the mean monthly loss rate μ and standard deviation σ of the reference ceramic jar set with the same manufacturer attributes, batch attributes and initial alcohol content attributes are calculated in the first year of use as the initial parameters of the individual wine loss baseline. If the historical loss data of the target ceramic jar reaches the preset duration, an individualized loss baseline is obtained by fitting the liquid level time series data of the ceramic jar itself.

[0060] The initial alcohol content attribute refers to the initial alcohol content of the original liquor contained in the earthenware jar, such as 65-degree light aroma original liquor and 62-degree strong aroma original liquor.

[0061] The reference set of ceramic jars refers to all ceramic jars that are completely identical to the target ceramic jars in terms of manufacturer, batch, and alcohol content, and that have accumulated sufficient historical loss data. For example, the 128 ceramic jars in stock of 65-degree liquor from manufacturer B in 2015.

[0062] The average monthly wear rate μ refers to the arithmetic mean of the monthly wear rate of the reference ceramic jar in the first year of its use, such as 0.035% per month.

[0063] The standard deviation σ refers to the statistical measure of the dispersion of the monthly loss rate data of the reference ceramic jar, for example, it can be 0.008% per month.

[0064] The preset duration refers to the data accumulation period that the platform pre-sets to fit a specific baseline, such as 12 consecutive months of measurement data.

[0065] The digital twin management platform can extract the monthly loss rate of the ceramic jar in the first year of its use as a reference, calculate the mean μ and standard deviation σ through statistical calculations, and generate a unique loss baseline model by retrieving the liquid level time series data of the target ceramic jar for a preset period of time and using a regression algorithm.

[0066] This technical solution addresses the issue of newly commissioned ceramic jars lacking sufficient historical data by reusing the loss statistics of mature jars of the same type to generate an initial baseline. For ceramic jars with sufficient existing data, the solution uses its own data to fit a dedicated baseline, thus solving the problem of new jars lacking judgment standards and thus being unable to provide early warnings. This ensures that all ceramic jars in the entire warehouse can generate effective individual wine loss baselines.

[0067] In one embodiment, in step 1, for earthenware jars whose service life exceeds a preset age threshold, an aging coefficient is introduced into the individual wine damage baseline to compensate for the long-term drift trend caused by the aging of the earthenware jar's microporous structure.

[0068] The preset age threshold refers to the critical value set by the platform for determining the aging of the cylinder block, such as 5 years, 8 years, etc.

[0069] The aging factor is a correction parameter used to correct the wear drift of older cylinders. For example, it can be a correction factor of 1.02 for each additional year of use.

[0070] Microporous structure aging refers to the physical changes that occur after long-term use of ceramic jars, where the micropores in the jar wall become larger and the porosity increases, leading to a continuous increase in the rate of evaporation and leakage of the wine.

[0071] The long-term drift trend refers to the pattern of slow and continuous increase in the monthly wear rate of older cylinders as their service life increases.

[0072] This solution identifies earthenware jars whose age exceeds a threshold, embeds the aging coefficient into the individual wine loss baseline calculation formula, and participates in the loss value calculation. By amplifying the baseline loss standard through the aging coefficient, it matches the actual loss increase pattern of old jars and avoids frequent false alarms for old jars.

[0073] This technical solution adds a loss correction parameter to the ceramic jars that have been used for a long time to offset the loss drift caused by the aging of the micropores in the jar body, reduce the deviation between the baseline standard of the old jar and the actual loss, and further reduce the probability of false alarms of abnormal loss in the old jar.

[0074] In one embodiment, optionally, the individual wine loss baseline further includes a temperature and humidity coupled correction function for normalizing the wine loss rate according to changes in ambient temperature and humidity.

[0075] The temperature and humidity coupling correction function is a mathematical function that outputs a loss correction factor based on the input of ambient temperature and humidity values. For example, it can be a loss upward correction function for high temperature and high humidity environments.

[0076] Ambient temperature and humidity refer to the real-time temperature and humidity monitoring values ​​collected in the wine cellar area where the earthenware jars are located. For example, it could be 32℃ and 70% humidity in summer, and 10℃ and 50% humidity in winter.

[0077] Normalization refers to converting the measured loss rate under different temperature and humidity conditions to the equivalent loss value under standard temperature and humidity conditions, thereby eliminating environmental interference.

[0078] The digital twin management platform reads the synchronously collected temperature and humidity data in the warehouse, substitutes it into the temperature and humidity coupling correction function to calculate the correction coefficient, corrects the measured loss rate, and then compares it with the baseline.

[0079] This technical solution eliminates the interference of normal volatilization caused by seasonal temperature and humidity fluctuations, standardizes loss data under different environments, avoids normal volatilization in high-temperature summers from being judged as abnormal loss, and effectively reduces seasonal false alarms.

[0080] In one embodiment, optionally, in step 3, when the target ceramic tank is equipped with a fixed liquid level timer, the first liquid level time series data is called to calculate the liquid level change rate V(t) and the liquid level change acceleration A(t). When the target ceramic tank is not equipped with a fixed liquid level timer, the second liquid level time sequence data is called up, and the daily average loss rate is calculated based on the liquid level difference between two adjacent inspections.

[0081] The rate of change of liquid level V(t) refers to the magnitude of the drop in liquid level in the ceramic jar per unit time, which characterizes the speed of wine loss. For example, it can be a drop in liquid level of 0.02 mm per day.

[0082] The liquid level change acceleration A(t) refers to the rate of change of the liquid level, which characterizes whether the loss is stable, accelerating or slowing down. For example, it can be the loss rate continuously increasing by 0.001 mm / day.

[0083] The difference in liquid level between two adjacent inspections refers to the difference in liquid level values ​​obtained from two rounds of portable inspection measurements. For example, it could be a 9mm drop in liquid level over a three-month interval.

[0084] The average daily loss rate refers to the percentage of daily loss obtained by dividing the difference between two inspections of liquid level equally over each day. For example, it could be an average daily loss of 0.0004%.

[0085] The digital twin management platform performs differential operations on continuous liquid level time series data to solve for the liquid level change and rate change per unit time, and further reads two discrete inspection liquid level measurement points. Combining the number of days between the two inspections, the platform averages the daily loss level.

[0086] This technical solution distinguishes between two types of monitoring data sources and uses differentiated loss calculation indicators. Fixed continuous data can capture early signs of accelerated loss, while portable discrete data can quantify long-term average loss, adapting to the monitoring capabilities of ceramic jars deployed on different hardware.

[0087] In one embodiment, optionally, in step 4, the warning level identifier includes at least: A blue warning indicator signifies that the acceleration of liquid level change changes from fluctuating around zero to a sustained positive value. A yellow warning sign indicates that the rate of change in liquid level or the average daily loss rate exceeds the individual wine loss baseline μ+3σ. An orange alert indicates that the rate of change in liquid level remains positive and shows no signs of slowing down during the nighttime cooling period. A red warning sign indicates a sudden drop in liquid level within a short time window.

[0088] The blue warning indicator represents an abnormal loss trend at the level of concern (e.g., BLUE01). The yellow warning indicator represents excessive loss at the level of abnormality (e.g., YELLOW02). The orange warning indicator represents suspected leakage at the level of severity (e.g., ORANGE03). The red warning indicator represents confirmed leakage (e.g., RED04).

[0089] The nighttime cooling period refers to the fixed period during which the ambient temperature in the wine cellar continues to drop at night, from 10 PM to 6 AM the next day.

[0090] A short-term window refers to a short monitoring interval set by the platform, which can be a continuous 72-hour sliding monitoring window.

[0091] A step drop refers to a sudden and significant drop in liquid level within a short period of time, as opposed to slow and uniform evaporation, such as a drop of more than 5 mm in 72 hours.

[0092] Specifically, after identifying the corresponding data features, the system automatically matches the corresponding warning level identifier to complete the marking.

[0093] This technical solution sets up a four-level progressive early warning logic, which distinguishes the severity of abnormalities from multiple levels, such as accelerated loss trend, excessive loss, elimination of temperature interference, and sudden drop in liquid level, so as to realize early warning of leakage without waiting for a large amount of wine to be lost before alarming.

[0094] In one embodiment, optionally, in step 6, the retested liquid level data is collected by a portable measuring terminal after performing standardized calibration before each use; Accordingly, the method further includes: The deviation between the retested liquid level data and the first liquid level time series data is calculated. When the deviation exceeds the preset tolerance range, it is determined that the fixed liquid level gauge has drifted, and the sensor health status indicator is updated.

[0095] Standardized calibration refers to the uniform zeroing calibration process performed on portable terminals before each inspection. For example, a standard calibration gauge block can be used to correct the zero point of liquid level measurement.

[0096] The deviation value refers to the difference between the portable re-measured liquid level and the fixed synchronous liquid level at the same time, such as 0.1cm.

[0097] The preset tolerance range refers to the allowable difference range between the measurements of two types of equipment preset by the platform, such as ±0.5cm.

[0098] Sensor drift refers to the phenomenon where the zero point of a fixed level gauge slowly shifts after long-term installation and use, resulting in a persistent fixed deviation in the measured value.

[0099] After powering on, the portable measuring terminal automatically runs its built-in calibration program. Once calibration is complete, it collects liquid level data from the ceramic tank, calculates the deviation between the re-measured liquid level data and the first liquid level time-series data, and extracts the liquid level values ​​of the two types of equipment at the same time point, performing a difference calculation to obtain the deviation value. When it is determined that the fixed liquid level gauge has drifted, the platform compares the deviation value with the preset tolerance range. If the deviation exceeds the limit, it determines that the sensor has a drift fault, and the sensor status label within the ceramic tank's digital twin can be changed to a drift fault, simultaneously sending a maintenance reminder.

[0100] This technical solution uses the portable device after each calibration as a high-confidence measurement benchmark to cross-verify the accuracy of the fixed sensor, automatically identify sensor drift faults, distinguish between real leakage anomalies and equipment measurement false alarms, and reduce unnecessary manual troubleshooting.

[0101] In one embodiment, optionally, in step 6, the adaptive adjustment of the dynamic early warning threshold includes: The ceramic jars in the warehouse were grouped according to manufacturer, batch, and years of use, and the early warning hit rate and false alarm rate of each group were calculated. For groups whose warning hit rate is higher than the first threshold, their dynamic warning threshold is lowered to improve sensitivity; For groups with false alarm rates exceeding the second threshold, their dynamic warning thresholds are increased to reduce the false alarm rate. During periods of high summer temperatures, the dynamic warning thresholds for all groups are uniformly lowered by a preset percentage.

[0102] The warning hit rate refers to the proportion of cases within a group where a warning is triggered and confirmed as a real leakage anomaly on-site. For example, the warning hit rate for a certain cylinder group is 6.1%.

[0103] False alarm rate refers to the proportion of cases within a group where, after a warning is triggered, on-site verification determines the cause to be normal evaporation or sensor malfunction. For example, the false alarm rate for a certain cylinder group is 5%.

[0104] The first threshold is the critical hit rate value for determining a high risk of leakage in a group, such as 5%. The second threshold is the critical false alarm rate value for determining an excessive number of false alarms in a group, such as 4%.

[0105] The high-temperature period in summer refers to a fixed range of months in which the temperature in the wine cellar is consistently high, such as from June to September each year.

[0106] The preset percentage refers to a fixed percentage by which the warning threshold is uniformly lowered in the summer, such as 15%.

[0107] Specifically, it can read information on manufacturers, batches, and years of use from all digital twins of ceramic jars, automatically group similar jars, summarize historical warnings and on-site handling records for each group, and calculate the hit rate and false alarm rate for each group through statistical calculations. The platform narrows the warning judgment range for high leakage risk groups and widens the judgment range for high false alarm groups; and tightens the warning standards for the entire warehouse in the summer.

[0108] This technical solution autonomously learns the leakage patterns of different cylinder groups based on historical early warning data from the entire database, dynamically adjusts the early warning thresholds in a differentiated manner, and adapts to the high-temperature volatilization characteristics of summer. The longer the system runs, the more accurate the early warning identification becomes.

[0109] To enable those skilled in the art to better understand this solution, this application also provides a preferred embodiment.

[0110] This paper presents a digital twin-based method for graded early warning of wine loss in ceramic jars and a collaborative measurement method using fixed and portable level gauges. This method achieves effective coverage of all ceramic jars in the warehouse while controlling the overall hardware cost. It enables accurate anomaly detection and early warning of leakage through individualized loss baselines of ceramic jars, and achieves adaptive optimization of the warning threshold through group pattern learning.

[0111] Figure 2 This is a schematic diagram of the overall process provided in the embodiments of this application, such as... Figure 2 As shown, this embodiment provides a method for graded early warning of wine loss in ceramic jars and a coordinated measurement method using fixed-portable level gauges based on digital twins, including the following steps: S1: Digital Twin Initialization. An independent digital twin is created for each ceramic jar, recording its full lifecycle attribute data, including: basic attributes (identification code, manufacturer, batch number, manufacturing date, geometric dimensions, wall thickness distribution, design volume, location coordinates), historical usage data (type of wine contained, initial alcohol content, sealing / opening date, cumulative years of use, historical loss rate sequence), maintenance records (sealing material type and replacement date, leakage repair records), and measurement data archives (timestamps of each measurement, liquid level value, temperature value, alcohol content value, measurement method identifier).

[0112] S2: Fixed Monitoring Node Deployment. Fixed level gauges (accuracy ±1mm) will be deployed only on a subset of key ceramic vats, representing 5% to 15% of the total storage area. Priority will be given to high-value ceramic vats aged over 10 years and those with historically significant wear and tear. Selection criteria also include areas with large temperature and humidity fluctuations, and vats whose sealing materials or ceramic jars are nearing their recommended lifespan. The fixed level gauges will automatically collect liquid level L_f(t) and temperature T(t) at preset time intervals and upload them to the digital twin management platform, forming a high-temporal-resolution continuous liquid level sequence.

[0113] S3: Periodic Inspections Using Portable Equipment. Ceramic vats without fixed level gauges are inspected using portable level gauges (accuracy ±1mm). The inspection cycle is dynamically adjusted based on risk level: once per quarter for normal vats, once per month for vats with warnings triggered or high-risk vats, once per week for vats with confirmed minor leaks but not yet repaired, and for vats with severe leaks requiring immediate removal of the wine. Risk level classification is based on the position of the manufacturer-batch-year group in the risk matrix, the number of historical warnings, and the degree of deviation of the loss rate from the baseline. The portable equipment simultaneously collects liquid level L_p, temperature T_p, and alcohol content A_p, and integrates an infrared thermal imaging module to collect the temperature distribution of the vat wall. The data is uploaded to the digital twin management platform to form a discrete liquid level sequence.

[0114] S4: Individualized Loss Baseline Establishment and Dynamic Update. When historical data for newly commissioned ceramic jars is insufficient, digital twins of ceramic jars with accumulated sufficient data from the same manufacturer, batch, and alcohol content group are used to extract the monthly loss rate sequence for the first year of commissioning, and the ensemble mean μ and standard deviation σ are calculated as the initial baseline. After historical data accumulates for more than a preset period, individualized baselines are established based on their own liquid level sequences: for fixed ceramic jars, a fine baseline including diurnal fluctuation patterns and the loss rate-temperature and humidity mapping relationship is established; for ceramic jars relying solely on portable systems, a basic baseline is established based on discrete measurement points. The baseline is refitted every time new data is received, and for ceramic jars whose usage period exceeds a threshold, an aging coefficient is introduced to compensate for long-term drift trends.

[0115] S5: Multi-level Feature Extraction and Graded Early Warning. For ceramic tanks with fixed level gauges, the rate of change of level V(t) and acceleration A(t) are calculated using a continuous level sequence. A(t) is numerically differentiated after sliding window polynomial filtering to suppress noise. Graded early warning rules: Blue warning (attention level) – A(t) changes from fluctuating near zero to a sustained positive value exceeding a preset threshold, indicating an accelerating loss trend; Yellow warning (abnormal level) – V(t) exceeds the individualized baseline μ+3σ; Orange warning (severe level) – V(t) remains positive during nighttime cooling periods without significant slowdown, excluding temperature-driven factors; Red warning (leakage confirmation level) – L(t) drops sharply within a short window, exceeding the threshold. For ceramic tanks relying solely on portable inspections, the daily average loss rate is calculated based on the difference in level between two adjacent inspections, triggering a yellow warning when it exceeds the baseline threshold.

[0116] S6: Portable Triggered Verification and Cross-Validation. Upon triggering a yellow or higher alert, a verification task sheet is generated, assigning a portable device for verification. For deployed fixed ceramic tanks: the average portable measurement L_p is compared with the fixed reading L_f for the same period, calculating the deviation Δ. Although both devices have millimeter-level accuracy, the fixed device may exhibit systematic deviations due to long-term operational drift and changes in installation conditions; the portable device can be standardized and calibrated before each use, and its measurement system is independent. After multiple standardized measurements and averaging, it can serve as a high-confidence independent reference benchmark. If Δ is within the tolerance range, the alert is confirmed as valid, and the current verification data is used as the benchmark point, recording the drift Δ to the sensor drift time series; if Δ exceeds the limit, a verification work order is generated, updating the liquid level data based on the average of multiple portable remeasurements. When the drift series shows a monotonically increasing trend and approaches the maintenance threshold, a sensor maintenance alert is generated in advance.

[0117] S7: Anomaly Type Determination and Handling Recommendations. Based on verified data and the individualized loss baseline of the ceramic tank digital twin, anomaly type determination is performed: Accelerated liquid level drop without significant alcohol content change and uniform tank wall temperature indicate poor sealing and accelerated evaporation, generating a seal inspection and replacement recommendation. Infrared thermal imaging detects a localized low-temperature anomaly area on the tank wall, indicating suspected leakage, generating a leak point detection recommendation and marking the suspected area. Confirmation of fixed sensor drift and normal portable retest results indicate a false alarm, recording "sensor drift false alarm" in the digital twin for optimizing warning thresholds.

[0118] S8: Group Pattern Learning and Threshold Adaptive Optimization. It aggregates all early warning records and handling results from the entire database, establishing a tiered risk matrix based on manufacturer, batch, and year. For high-risk groups, the early warning trigger threshold is lowered to improve sensitivity; for low-risk groups, the threshold is raised to reduce false alarm rates. Optimization results are written back to the digital twins of each ceramic jar, providing a basis for dynamic adjustments to the S3 inspection cycle. During the high-temperature summer period, the trigger threshold for evaporation anomalies is automatically lowered, and the default value is restored in winter.

[0119] Figure 3 The flowchart for the graded early warning logic of a fixed liquid level gauge ceramic tank is as follows: Figure 3 As shown, the process begins with acquiring a continuous liquid level sequence L(t). The system automatically calculates the liquid level change rate V(t) and the liquid level change acceleration A(t). The first step determines whether A(t) is continuously positive and has a duration that meets the standard. If so, a blue warning (attention level) is output; otherwise, the process proceeds directly to the next step. Next, it compares whether V(t) exceeds the baseline threshold. If it does, a yellow warning (abnormal level) is triggered; otherwise, the judgment is canceled. Based on the yellow warning, it further checks whether V(t) remains positive without decay during the nighttime cooling period. If so, an orange warning (severe level) is generated. Finally, if the liquid level drops excessively in a short window, a red warning (leakage confirmed) is directly issued. Each level corresponds to a specific handling procedure. The entire process relies on multi-layered progressive judgment based on time-series data. It first captures the accelerating trend of loss and provides early warning, then eliminates temperature interference layer by layer, and finally identifies sudden leaks. This solves the problem of traditional alarms relying solely on the delayed liquid level drop, achieving gradient-based early warning.

[0120] Figure 4 A flowchart for cross-validation of fixed and portable data, such as Figure 4 As shown, the initial condition is a yellow or higher warning trigger. The system automatically generates a verification task sheet and sends it to a portable terminal. The operator takes multiple standardized measurements and averages them to obtain the portable liquid level Lp. The platform retrieves the fixed reading Lf from the same time period and calculates the difference Δ between the two. If Δ ≤ preset tolerance, the warning is considered valid, the drift amount Δ is recorded, and the drift trend is continuously monitored. If Δ exceeds the limit, an equipment verification work order is generated to correct the liquid level data using the portable remeasurement average. The process includes an additional drift trend judgment branch. When the drift amount continues to increase monotonically, the platform pushes a maintenance warning for the fixed sensor in advance. Relying on the calibrated portable device as a standard reference, the system automatically distinguishes between real cylinder leakage faults and false alarms caused by sensor drift, reducing unnecessary on-site repairs.

[0121] Figure 5 For group pattern learning and threshold adaptive optimization closed-loop feedback graph, such as Figure 5 As shown, the bottom layer consists of an independent digital twin for each ceramic tank. All tank warning records and on-site handling results are uniformly aggregated into a group pattern learning module. This module calculates leakage rate and false alarm rate according to three dimensions: manufacturer, batch, and service life, building a hierarchical risk matrix to classify tanks into high, medium, and low-risk groups. Warning thresholds are lowered for high-risk groups to improve sensitivity, while thresholds are raised for low-risk groups to reduce false alarms. In summer, the evaporation warning threshold is uniformly lowered. The optimized threshold parameters are batch-written back to all ceramic tank digital twins, simultaneously updating the dynamic inspection cycle of each tank. This feedback loop influences the initial liquid level acquisition and warning judgment stages, forming a complete closed loop of data acquisition, warning, handling, autonomous learning, and parameter iteration. The richer the system data, the more accurate the warning judgment.

[0122] Specific example: A deployment of 50,000 earthenware jars in a wine cellar; A wine cellar stores 50,000 earthenware jars from 3 manufacturers, spanning 5 production batches, with usage periods ranging from 1 to 15 years. Among them, there are approximately 400 high-value earthenware jars aged over 10 years, and approximately 200 earthenware jars with abnormal historical damage records.

[0123] Deployment plan: 600 key ceramic tanks will be equipped with fixed level gauges (approximately 12% of the total tank), with an accuracy of ±1mm, and data will be collected once per hour. The remaining 49,400 ceramic tanks will be inspected using portable level gauges, equipped with 3 portable level gauges (accuracy ±1mm), each of which will inspect approximately 600 tanks per day, with a baseline inspection cycle of one month.

[0124] Example of digital twin initialization: T-2015-B-0012 (Deployed Fixed Type): Manufacturer B, 2015 batch, 1000L capacity, glazed jar; sealed in 2016, 65% ABV light aroma type original liquor, aged 10 years; sealing gasket replaced in 2020.

[0125] T-2022-C-0308 (Portable Inspection Only): Manufacturer C, 2022 batch, 1000L capacity, unglazed jar; sealed in 2023, 62% ABV strong-aroma base liquor, 3-year aging; no maintenance records.

[0126] Initial baseline establishment for newly commissioned ceramic jars: T-2025-B-0500 belongs to the "Manufacturer B-2015 Batch-65 Degrees" group. Data from 128 ceramic jars in this group that have accumulated over a year of data was retrieved. The monthly loss rate sequence for the first year of commissioning was extracted, and μ=0.035% / month and σ=0.008% / month were calculated. This data was then written into the digital twin of this ceramic jar as the initial baseline.

[0127] Fixed monitoring and early warning example: Data from T-2015-B-0012 in August 2025 shows that: the monthly loss rate from August 1st to 8th was 0.045%, which is within the normal range for summer; starting from August 9th, the acceleration A(t) changed from -0.001% / h² to +0.002% / h² and lasted for more than 36 hours, triggering a blue warning; on August 15th, V(t) reached 0.072% / day, exceeding the individualized summer baseline of 0.048% + 3σ (σ = 0.007%) = 0.069%, triggering a yellow warning.

[0128] Cross-validation and anomaly assessment: After the yellow alert was triggered, the operator carried a portable level gauge and an infrared thermal imager to verify the readings. The average of three measurements taken with the portable level gauge was L_p = 127.1 cm, and the simultaneous reading with the fixed level gauge was L_f = 127.0 cm, with a deviation Δ = 0.1 cm, within the preset tolerance of ±0.5 cm, indicating the alert was reliable. The drift sequence [0.1, 0.0, 0.2, 0.1] cm did not show a monotonically increasing trend, indicating the sensor was functioning normally. Infrared thermal imaging revealed a low-temperature anomaly zone with a diameter of approximately 4 cm on the northwest side of the tank bottom, with a temperature about 0.4℃ lower than the surrounding area. Based on the combined assessment, a micro-leakage was identified, and a leak point investigation suggestion was generated. A micro-crack was confirmed on-site and repaired promptly.

[0129] Group Learning Optimization: After one year of system operation, the group learning module discovered that ceramic jars from Manufacturer B's 2015 batch, with a usage period exceeding 8 years, had a microleakage rate of 6.1%, significantly higher than the average of 2.8%. The system automatically lowered the yellow alert threshold for this group from μ+3σ to μ+2.5σ, and the orange alert threshold from μ+3.5σ to μ+3σ. In the following quarter, this group received 3 new early warnings, triggering an average of approximately 6 days in advance. During the summer (June-September), the system automatically lowered the trigger threshold for abnormal volatilization of all ceramic jars by 15%, reverting to the default value in winter.

[0130] The technical advantages of this solution are as follows: First, by using a collaborative architecture of "a small number of fixed monitoring cylinders for key cylinders + a large number of portable cylinders for general cylinders", effective coverage of the entire database can be achieved while controlling costs. The two have comparable accuracy and complementary acquisition modes.

[0131] Second, by using the higher-order characteristic quantity of acceleration A(t), the leakage warning is moved forward from "the liquid level has dropped significantly" to the stage of "the rate of loss has begun to accelerate".

[0132] Third, by using individualized digital twins and loss baselines, we can solve the problem of false alarms and missed alarms caused by the "one-size-fits-all" threshold, and achieve "one policy per cylinder".

[0133] Fourth, by combining portable verification with fixed cross-validation, it provides both independent reliability testing and the detection of long-term sensor drift trends. Portable verification, through standardized calibration each time, eliminates cumulative drift from long-term installation and can serve as a high-confidence independent reference benchmark.

[0134] Fifth, through group pattern learning and threshold adaptive optimization, the system has the ability to become more accurate with use, and the accuracy of early warnings continues to improve as data accumulates.

[0135] Example 2 Figure 6 This is a schematic diagram of the structure of the digital twin-based ceramic jar wine spoilage classification and early warning device provided in Embodiment 2 of this application. Figure 6As shown, the device is deployed on a digital twin management platform, and the device includes: Digital twin building module 601 is used to build a set of digital twins and configure an individual alcohol damage baseline for each digital twin; The data acquisition module 602 is used to receive first liquid level time series data uploaded by a fixed liquid level gauge and second liquid level time series data uploaded by a portable liquid level gauge, wherein the time resolution of the first liquid level time series data is higher than that of the second liquid level time series data. The deviation calculation module 603 is used to call the corresponding liquid level time series data and calculate the deviation from the individual wine loss baseline based on the equipment mounting status of the target ceramic vat. The early warning generation module 604 is used to compare the deviation with the dynamic early warning threshold, generate an early warning level identifier, and write it into the digital twin; The retest scheduling module 605 is used to generate retest prompt information in response to the warning level identifier and to schedule the portable measurement terminal to perform on-site retesting; The parameter update module 606 is used to update the liquid level reference value or sensor health status indicator based on the retest results, and adaptively adjust the dynamic warning threshold.

[0136] The apparatus provided in this embodiment has the same execution process and beneficial effects as the method embodiments described above, and will not be repeated here to avoid repetition.

[0137] Example 3 This embodiment also provides a digital twin-based early warning system for graded spoilage of ceramic wine jars, the system comprising: Fixed level gauges are deployed in some key ceramic vats in the wine storage cellar to collect first liquid level time-series data at a preset sampling frequency; A portable measuring terminal is used to perform on-site retesting of the target ceramic jar and collect retest liquid level data when a retest prompt is received; The digital twin management platform is communicatively connected to the fixed level gauge and the portable measuring terminal, and is configured to execute the digital twin-based grading and early warning method for wine loss in ceramic jars as described in the above embodiments.

[0138] The system provided in this embodiment has the same execution process and beneficial effects as the method embodiments described above, and will not be repeated here to avoid duplication.

[0139] Example 4 This application also provides a readable storage medium storing a program or instructions. When the program or instructions are executed by a processor, they implement the various processes of the above-described embodiment of the method for graded early warning of wine damage in ceramic jars based on digital twins, and achieve the same technical effect. To avoid repetition, these will not be described again here.

[0140] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0141] Example 5 This application also provides a chip, which includes a processor and a communication interface. The communication interface is coupled to the processor. The processor is used to run programs or instructions to implement the various processes of the above-described embodiments of the method for graded early warning of wine damage in ceramic jars based on digital twins, and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0142] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, device chip, chip system, or system-on-a-chip, etc.

[0143] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element. Furthermore, it should be noted that the scope of the methods and systems in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0144] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a computer software product. This computer software product is stored in a storage medium (such as ROM / RAM, disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0145] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms fall within the scope of protection of this application.

[0146] The above description is merely a preferred embodiment and the technical principles employed in this application. This application is not limited to the specific embodiments described herein. Various obvious changes, readjustments, and substitutions that can be made by those skilled in the art will not depart from the scope of protection of this application. Therefore, although this application has been described in detail through the above embodiments, this application is not limited to the above embodiments. Many other equivalent embodiments may be included without departing from the concept of this application, and the scope of this application is determined by the scope of the claims.

Claims

1. A method for graded early warning of wine spoilage in ceramic jars based on digital twins, characterized in that, The method includes: Step 1, Construct a set of digital twins: Establish an independent digital twin for each earthenware jar in the wine storage cellar, and configure an individual wine loss baseline for each digital twin. The individual wine loss baseline is obtained by fitting historical loss data based on the manufacturer attributes, batch attributes, and years of use of the corresponding earthenware jar. Step 2, collect dual-modal liquid level data: receive the first liquid level time series data uploaded by the fixed liquid level gauge at a preset sampling frequency, and the second liquid level time series data uploaded by the portable liquid level gauge in response to the inspection task. The time resolution of the first liquid level time series data is higher than that of the second liquid level time series data. Step 3, calculate the real-time deviation: Based on the equipment mounting status of the target ceramic vat, call the corresponding first liquid level time series data or second liquid level time series data, substitute it into the individual wine loss baseline for calculation, and obtain the deviation between the measured value of the current wine loss status and the baseline. Step 4, generate warning identifier: compare the deviation with a preset dynamic warning threshold. When the deviation exceeds the dynamic warning threshold, generate a corresponding warning level identifier and write the warning level identifier into the digital twin of the target ceramic jar. Step 5, Trigger Retest Verification: In response to the generation of the warning level identifier, automatically generate retest prompt information, and dispatch a portable measuring terminal to conduct on-site retesting of the target ceramic vat to obtain retest liquid level data; Step 6, Automatic parameter update: Based on the comparison results between the re-measured liquid level data and the measured liquid level time series data, update the liquid level reference value or sensor health status identifier in the digital twin of the target ceramic tank, and adaptively adjust the dynamic early warning threshold according to the early warning and handling feedback results of all ceramic tanks in the warehouse.

2. The method according to claim 1, characterized in that, In step 1, the fitting process for the individual alcohol damage baseline includes: If the historical loss data of the target ceramic jar is insufficient, the mean monthly loss rate μ and standard deviation σ of the reference ceramic jar set with the same manufacturer attributes, batch attributes and initial alcohol content attributes are calculated in the first year of use as the initial parameters of the individual wine loss baseline. If the historical loss data of the target ceramic jar reaches the preset duration, an individualized loss baseline is obtained by fitting the liquid level time series data of the ceramic jar itself.

3. The method according to claim 2, characterized in that, In step 1, for earthenware jars whose service life exceeds a preset age threshold, an aging coefficient is introduced into the individual wine damage baseline to compensate for the long-term drift trend caused by the aging of the microporous structure of the earthenware jar.

4. The method according to claim 2, characterized in that, The individual wine damage baseline also includes a temperature and humidity coupling correction function, which is used to normalize the wine damage rate according to changes in ambient temperature and humidity.

5. The method according to claim 1, characterized in that, In step 3, when the target ceramic tank is equipped with a fixed liquid level timer, the first liquid level time series data is called up to calculate the liquid level change rate V(t) and the liquid level change acceleration A(t). When the target ceramic tank is not equipped with a fixed liquid level timer, the second liquid level time sequence data is called up, and the daily average loss rate is calculated based on the liquid level difference between two adjacent inspections.

6. The method according to claim 5, characterized in that, In step 4, the warning level identifier includes at least: A blue warning indicator signifies that the acceleration of liquid level change changes from fluctuating around zero to a sustained positive value. A yellow warning sign indicates that the rate of change in liquid level or the average daily loss rate exceeds the individual wine loss baseline μ+3σ. An orange alert indicates that the rate of change in liquid level remains positive and shows no signs of slowing down during the nighttime cooling period. A red warning sign indicates a sudden drop in liquid level within a short time window.

7. The method according to claim 1, characterized in that, In step 6, the retested liquid level data is collected by the portable measuring terminal after performing standardized calibration before each use; Accordingly, the method further includes: The deviation between the retested liquid level data and the first liquid level time series data is calculated. When the deviation exceeds the preset tolerance range, it is determined that the fixed liquid level gauge has drifted, and the sensor health status indicator is updated.

8. The method according to claim 1, characterized in that, In step 6, the adaptive adjustment of the dynamic early warning threshold includes: The ceramic jars in the warehouse were grouped according to manufacturer, batch, and years of use, and the early warning hit rate and false alarm rate of each group were calculated. For groups whose warning hit rate is higher than the first threshold, their dynamic warning threshold is lowered to improve sensitivity; For groups with false alarm rates exceeding the second threshold, their dynamic warning thresholds are increased to reduce the false alarm rate. During periods of high summer temperatures, the dynamic warning thresholds for all groups are uniformly lowered by a preset percentage.

9. A digital twin-based early warning device for classifying and warning about wine spoilage in ceramic jars, characterized in that, Deployed on a digital twin management platform, including: The digital twin building block is used to construct a collection of digital twins and configure an individual alcohol damage baseline for each digital twin; The data acquisition module is used to receive first liquid level time-series data uploaded by a fixed liquid level gauge and second liquid level time-series data uploaded by a portable liquid level gauge. The time resolution of the first liquid level time-series data is higher than that of the second liquid level time-series data. The deviation calculation module is used to call the corresponding liquid level time series data and calculate the deviation from the individual wine loss baseline based on the equipment mounting status of the target ceramic vat. The early warning generation module is used to compare the deviation with the dynamic early warning threshold, generate an early warning level identifier, and write it into the digital twin; The retest scheduling module is used to generate retest prompt information in response to the warning level identifier and to schedule portable measurement terminals to perform on-site retests. The parameter update module is used to update the liquid level reference value or sensor health status indicator based on the retest results, and adaptively adjust the dynamic warning threshold.

10. A digital twin-based early warning system for graded spoilage of ceramic wine jars, characterized in that, include: Fixed level gauges are deployed in some key ceramic tanks in the wine storage cellar to collect first liquid level time-series data at a preset sampling frequency; A portable measuring terminal is used to perform on-site retesting of the target ceramic jar and collect retest liquid level data when a retest prompt is received; The digital twin management platform is communicatively connected to the fixed level gauge and the portable measuring terminal, and is configured to execute the digital twin-based method for graded early warning of wine loss in ceramic jars as described in any one of claims 1-8.

Images