Method for detecting the seal of a solid beverage can and can

Abstract

This invention relates to a method for detecting the seal of a solid beverage can and the can itself, comprising: acquiring initial pressure data from a detection chamber and determining whether contamination exists by using a preset threshold; if the initial pressure data is lower than the preset threshold, determining the location of contamination and activating a cleaning mechanism to acquire sealed chamber data after cleaning; acquiring a pressure change curve after gas injection based on the sealed chamber data after cleaning, and determining the pressure drop pattern caused by background noise through an analysis algorithm; acquiring time series data for the compensated pressure change curve, determining whether periodic fluctuations exist through a transformation algorithm, and identifying the signal components caused by actual leakage; acquiring subsequent pressure change curves through the optimized detection process data, repeating the compensation process, and determining the final seal integrity result.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a method for detecting the seal of a solid beverage can and the can itself. Background Technology

[0002] In the production of high-end solid beverages, the integrity of the can seal directly determines whether the product can resist moisture and oxygen corrosion, serving as a core defense for ensuring its long-term quality and safety. Therefore, achieving 100% reliable fully automated online inspection is crucial. However, existing inspection methods reveal insurmountable shortcomings when dealing with such powder products. Traditional methods, such as offline water inspection, can damage the product and are inefficient, while online visual inspection can only judge the appearance of rolled edges and cannot detect internal defects.

[0003] More importantly, directly applying the pressure decay method for online detection presents two unique, interconnected challenges. The first challenge stems from the inherent physical properties of the powder product: during filling, fine powder inevitably disperses and adheres to the sealed area of ​​the can opening. These micro-contaminants, invisible to the naked eye, can disrupt the initial seal of the detection chamber when the detection head is pressed, causing the instrument to misjudge a leak and generate numerous false alarms. The second challenge is more deeply related to the microstructure of the powder material: even if the can is completely sealed, numerous voids exist between the loose powder particles inside. When detection gas is injected into the test chamber and pressurized, the gas seeps into these particle gaps and is compressed and absorbed. This process generates a slow pressure drop signal that closely resembles a real leak. This "background noise" signal, caused by the material's inherent characteristics, is extremely strong and often completely drowns out the subtle pressure changes caused by genuine micro-leaks, making it impossible for the detection system to accurately distinguish between them.

[0004] Therefore, designing a detection method that can effectively resist the interference of powder contamination at the can opening on the sealing detection process and avoid false alarms, while accurately separating the background changes caused by the powder's own structure from the complex pressure decay signal, and thus extracting the tiny signal that uniquely reflects the actual leak, has become the key issue for achieving high-speed and reliable full inspection of the sealing performance of solid beverage cans. Summary of the Invention

[0005] This invention provides a method for detecting the seal of a solid beverage can and the can itself, mainly comprising:

[0006] Initial pressure data is acquired from the detection chamber, and a preset threshold is used to determine if contamination is present. If the initial pressure data is below the preset threshold, the contamination location is identified and a cleaning mechanism is activated to acquire cleaned chamber sealing data. Based on the cleaned chamber sealing data, a pressure change curve after gas injection is obtained, and an analysis algorithm is used to determine the pressure drop pattern caused by background noise. The pressure drop pattern is used to determine whether gas absorption dominates the signal. If gas absorption dominates the signal, a compensation value is obtained from a preset model to generate a compensated pressure change curve. For the compensated pressure change curve, time series data is acquired, and a transformation algorithm is used to determine if periodic fluctuations exist, identifying the signal components caused by actual leakage. Peak amplitude data is extracted from the signal components and compared with a preset leakage threshold to determine the seal integrity. If the peak amplitude data exceeds the preset leakage threshold, it is marked as a leakage defect, and a defect classification result is obtained. Based on the defect classification result, continuous sequence data is acquired, and it is determined whether the continuous defects exceed a preset number threshold. If they do, the detection parameters are adjusted to generate optimized detection process data. Using the optimized detection process data, subsequent pressure change curves are acquired, the compensation process is repeated, and the final seal integrity result is determined. Furthermore, the step of acquiring initial pressure data from the detection chamber and determining whether there is a contamination impact by using a preset threshold includes: collecting initial pressure data of the detection chamber through a sensor and comparing the initial pressure data with a preset threshold; if the initial pressure data is lower than the preset threshold, it is determined that there is a contamination impact, and the deviation value of the initial pressure data is recorded; based on the deviation value, a contamination impact assessment report is generated to determine the preliminary range of the contamination impact; based on the preliminary range and in conjunction with historical data comparison, the severity of the contamination impact is determined; based on the severity, subsequent processing procedures are activated to obtain quantitative indicators of the contamination impact; based on the quantitative indicators, it is determined whether further detection is needed, and initial detection results are generated; based on the initial detection results, the status record of the detection chamber is updated to provide data support for subsequent cleaning mechanisms; based on the status record, a detection log is generated to save the comparison results of the initial pressure data; based on the detection log, a tracking record of the contamination impact is constructed to ensure the continuity of subsequent steps.Furthermore, if the initial pressure data is lower than a preset threshold, the process of determining the contamination location and activating the cleaning mechanism to obtain cleaned cavity sealing data includes: processing the surface data of the detected cavity using an image recognition algorithm to locate the contamination attachment location; generating a contamination distribution map for the contamination attachment location to determine the boundary of the contamination area; adjusting the operating parameters of the cleaning mechanism and activating the cleaning device based on the contamination distribution map; performing targeted cleaning of the contamination area using the cleaning device and obtaining real-time feedback data of the cleaning process; judging the cleaning progress based on the feedback data and determining whether the cleaning has reached a preset standard; stopping the cleaning mechanism and obtaining cleaned cavity sealing data if the cleaning has reached the preset standard; comparing the cleaned cavity sealing data with the initial pressure data to evaluate the cleaning effect; generating a cleaning effect report based on the cleaning effect and recording the trend of the sealing data change; and updating the cavity status record based on the trend of change to provide a reference for subsequent testing. Furthermore, the step of obtaining the pressure change curve after gas injection based on the cleaned cavity sealing data and determining the pressure drop mode caused by background noise through analysis algorithms includes: collecting the cleaned cavity sealing data through multi-point sensors and extracting the gas injection time series; obtaining the initial pressure value after injection based on the time series and adjusting the data using a temperature compensation correction method; constructing a pressure change curve using the adjusted data and generating a smoothed curve sequence; calculating the difference between adjacent points using a differential analysis algorithm for the smoothed curve sequence and obtaining a difference sequence; calculating the average and standard deviation of the difference based on the difference sequence to determine the background noise impact threshold; if the background noise impact threshold exceeds a preset range, separating the interference components through a noise filtering mechanism; obtaining the pure downward trend through the separated data and determining the pressure drop mode; performing pattern matching verification for the pressure drop mode and calculating the similarity with a preset noise mode; generating a background noise impact report based on the similarity and recording the characteristics of the pressure drop mode.Furthermore, the step of determining whether gas absorption is the dominant signal through the pressure drop mode, and if the gas absorption is the dominant signal, obtaining a compensation value from a preset model to generate a compensated pressure change curve, includes: collecting background noise data through a sensor, analyzing the pressure drop mode, and obtaining a pressure drop judgment result; determining whether gas absorption is the dominant signal based on the judgment result, and generating a dominant status assessment; if the gas absorption is the dominant signal, extracting a preset absorption compensation value from the particle gap model; adjusting the corresponding part of the pressure change curve using the absorption compensation value to generate a compensated curve; performing real-time calibration processing on the compensated curve using preset calibration parameters; adjusting the curve deviation based on the calibration processing to obtain a calibrated pressure change curve; generating a compensation effect report based on the calibrated curve, recording the data comparison before and after adjustment; updating the status record of the pressure change curve based on the comparison result; and generating a compensation process log based on the status record to provide data support for subsequent analysis. Furthermore, the step of acquiring time-series data from the compensated pressure change curve and determining whether periodic fluctuations exist through a transformation algorithm to identify the signal components caused by the actual leak includes: extracting time-series data from the compensated pressure change curve and processing the data using a Fourier transform algorithm; generating a frequency spectrum from the processed data and extracting peak frequency components; determining whether periodic fluctuations exist based on the peak frequency components and identifying potential signal patterns; applying noise filtering to the potential signal patterns to obtain filtered signals; determining the amplitude of minute fluctuations from the filtered signals and comparing it with a preset threshold; if the amplitude of minute fluctuations exceeds the preset threshold, it is determined to be a leak-related component; extracting the signal components caused by the actual leak based on the leak-related components and generating a signal analysis report; recording the quantification results of the fluctuation amplitude in the signal analysis report; and updating the feature records of the signal components based on the quantification results to provide a basis for subsequent detection.Furthermore, the step of extracting peak amplitude data from the signal components, comparing it with a preset leakage threshold to determine the seal integrity, and marking it as a leakage defect if the peak amplitude data exceeds the preset leakage threshold, and obtaining the defect classification result, includes: obtaining initial signal data from the signal components and removing high-frequency noise through low-pass filtering; obtaining clean signal components from the processed data and extracting peak amplitude data; determining the amplitude value using a peak detection quantization method for the peak amplitude data; comparing the amplitude value with a preset leakage threshold to determine if there is a potential leak; marking the location of the leakage defect if the amplitude value exceeds the preset leakage threshold; performing multi-point sampling verification of the signal based on the leakage defect location to obtain the defect classification result; generating a defect analysis report based on the defect classification result and recording the comparison result of the peak amplitude data; updating the recorded data of the defect location based on the comparison result; and generating a defect tracking log based on the recorded data to support subsequent processes. Furthermore, the step of obtaining continuous sequence data based on the defect classification results, determining whether continuous defects exceed a preset quantity threshold, and adjusting detection parameters to generate optimized detection process data, includes: extracting continuous sequence data from the defect classification results and counting the number of consecutive tank defects; comparing the number of consecutive defects with a preset quantity threshold to obtain a defect quantity statistics result; if the defect quantity statistics result exceeds the preset quantity threshold, adjusting the detection parameters of the online full inspection; determining a parameter adjustment mechanism based on the adjusted parameters and processing the sequence data; using defect pattern recognition to subdivide defect types using the processed data; generating process optimization results based on the subdivided defect types and recording the parameter adjustment effect; integrating the defect classification results with the process optimization results to form a parameter feedback loop; determining the continuous defect exceedance situation through the feedback loop and generating optimized detection process data; and updating the operation record of the detection system based on the detection process data. Furthermore, the step of obtaining subsequent pressure change curves through the optimized detection process data, repeating the compensation process, and determining the final seal integrity result includes: acquiring subsequent pressure change curves of the tanks through the optimized detection process data; obtaining preliminary pressure distribution characteristics based on the pressure change curves and recording the distribution data; repeating the compensation process for the distribution data and obtaining adjustment parameters from temperature influencing factors; generating a compensated curve using the adjustment parameters and recording the compensated data characteristics; judging the comparison differences between multiple tanks based on the compensated curves and obtaining the seal deviation value; if the seal deviation value exceeds a preset threshold, repeating the compensation process and adjusting the curve data; determining the final seal integrity result and generating a detection report using the adjusted curve data; recording the quantitative indicators of the seal integrity result in the detection report; and updating the status record of the detection system based on the quantitative indicators to provide a reference for subsequent detection.Furthermore, the step of obtaining subsequent pressure change curves through the optimized detection process data, repeating the compensation process, and determining the final seal integrity result includes: extracting subsequent tank detection data from the optimized detection process data; constructing subsequent pressure change curves based on the detection data and obtaining curve characteristic parameters; repeating the compensation process for the curve characteristic parameters and adjusting temperature influencing factors; generating a compensated pressure curve using the adjusted data and recording the curve change trend; analyzing the sealing differences between multiple tanks based on the change trend and determining the deviation range; if the deviation range exceeds a preset threshold, repeating the compensation process to optimize the curve data; generating the final seal integrity result using the optimized curve data and recording the detection conclusion; updating the detection system's operation log based on the detection conclusion; and generating reference data for subsequent detections based on the operation log to ensure process continuity.

[0007] A can body, comprising a can body, wherein the can body is optimized for testing using a sealing detection method for solid beverage cans.

[0008] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0009] This invention discloses a method that proposes an integrated solution to address the interconnected business scenario problems in tank seal integrity detection, including powder contamination, background noise interference, gas absorption effects, and the identification of minute leak signals. The method determines contamination by comparing initial pressure data with a threshold, locates the contamination using image recognition, and activates a cleaning mechanism. Subsequently, it analyzes the pressure change curve based on post-cleaning data, uses differential analysis to eliminate background noise, compensates for gas absorption effects using a particle gap model, and finally uses Fourier transform to extract minute leak signals from periodic fluctuations. These signals are then compared with a leak threshold to determine seal integrity. Simultaneously, detection parameters are optimized for continuous defects to ensure subsequent detection accuracy. This invention significantly improves the accuracy and reliability of seal detection through multi-algorithm fusion and dynamic parameter adjustment, achieving end-to-end optimization from contamination treatment to leak determination. Attached Figure Description

[0010] Figure 1 This is a flowchart of a method for detecting the seal of a solid beverage can according to the present invention. Detailed Embodiments

[0011] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0012] like Figure 1 This embodiment of a method for detecting the seal of a solid beverage can may specifically include:

[0013] Step S101: Obtain initial pressure data from the detection chamber, determine whether powder contamination exists by using a preset threshold, and if the initial sealing value of the detection chamber is lower than the threshold, use an image recognition algorithm to determine the contamination attachment location and activate the cleaning mechanism to obtain the sealed data of the cleaned chamber.

[0014] Initial pressure data is acquired from the detection chamber. A preset threshold is used to determine the presence of powder contamination. If the initial pressure data is below the threshold, an image recognition algorithm is used to determine the contamination location. A cleaning mechanism is activated at the contamination location to obtain a cleaned chamber. Sealing data is acquired from the cleaned chamber, and the cleaning effect is determined by comparing the sealing data with the initial pressure data. If the sealing data is above the threshold, the sealing data is updated to obtain a chamber linkage monitoring record.

[0015] In one implementation, initial pressure data can be obtained from the detection chamber using a pressure sensor installed within the chamber.

[0016] Specifically, the sensor monitors pressure changes inside the cavity in real time. For example, in powder handling equipment such as a pharmaceutical mixing chamber, the initial pressure data reflects the cavity's sealing condition. If powder contamination exists in the cavity, it may lead to a leaky seal, causing the pressure value to drop. In this way, the accuracy of data acquisition is ensured, providing a basis for subsequent judgment. Furthermore, determining the presence of powder contamination involves comparing the acquired initial pressure data with a preset sealing threshold.

[0017] For example, the preset threshold can be set according to the cavity design standards. In industrial powder conveying cavities, the threshold is 95% of the standard atmospheric pressure. If the initial pressure is lower than this threshold, powder contamination is determined to exist. This judgment process is simple and efficient, enabling rapid identification of potential problems and preventing contamination from affecting equipment operation.

[0018] Preferably, if the initial sealing value of the detection cavity is lower than a threshold, an image recognition algorithm is used to determine the location of contamination. This algorithm first captures images using a camera built into the cavity, and then applies edge detection and color analysis techniques to identify the contaminated areas.

[0019] Specifically, the image recognition process includes preprocessing the image to remove noise, followed by extracting features such as the texture and color differences of powder particles. For example, in a food powder cavity, the algorithm can distinguish between a normal surface and adhered dust. This algorithm accurately locates contamination sites, such as cavity walls or corners, ensuring targeted cleaning. This step is the core innovation because it combines visual analysis with automated judgment, improving the accuracy and efficiency of contamination detection and reducing the risks associated with manual intervention in practical applications.

[0020] In one possible implementation, activating the cleaning mechanism to obtain the cleaned cavity seal data can be achieved by activating the built-in jet device or a vibration cleaner.

[0021] For example, based on the location determined by image recognition, the system controls the injection of high-pressure gas or cleaning agent to specifically remove powder contaminants. After cleaning, the cavity sealing data is obtained again via a pressure sensor to verify the cleaning effect. This mechanism is particularly suitable for powder storage cavities, as it can restore sealing performance.

[0022] It should be noted that image recognition algorithms can be implemented using template matching methods to further improve positioning accuracy.

[0023] Specifically, this method uses a pre-stored clean cavity image as a template, compares it with the current image, and calculates the similarity to identify contaminated areas.

[0024] For example, in a chemical powder reaction chamber, if the similarity is less than 80%, it is marked as a contaminated location. This method is simple, reliable, and supports real-time processing.

[0025] For example, in another embodiment, the cleaning mechanism can incorporate ultrasonic vibration to assist in the removal of stubborn powder. Upon activation, the system applies directional vibration waves based on location data to loosen the deposits, which are then removed by a vacuum cleaner. Post-cleaning seal data shows that the pressure has returned to above a threshold, indicating that the contamination has been effectively removed.

[0026] Understandably, this technical solution is applicable to various powder chamber scenarios, such as handling pharmaceutical powder contamination in pharmaceutical equipment or detecting flour residues in food processing chambers. These implementation methods demonstrate the solution's versatility, ensuring flexible application within the same field. Furthermore, to enhance robustness, the threshold can be dynamically adjusted based on historical data.

[0027] For example, if the cavity is used frequently, the threshold can be slightly lowered to accommodate normal wear and tear. This adjustment does not change the core process but improves the adaptability of the judgment.

[0028] In one embodiment, the entire process is integrated into the control system, forming a closed loop from data acquisition to cleaning activation.

[0029] For example, image capture is triggered when the initial pressure is below a threshold, and cleaning is immediately activated after the algorithm outputs the location. Data feedback confirms the cleaning effect. This integration into the powder mixing chamber can significantly improve maintenance efficiency and reduce downtime.

[0030] Specifically, the cleaned cavity sealing data can be used not only for verification but also stored as a log to support subsequent analysis.

[0031] For example, in a continuous production environment, this data helps optimize threshold settings to enable preventative maintenance.

[0032] Step S102: Based on the sealed data of the cleaned cavity, obtain the pressure change curve after gas injection, and use the differential analysis algorithm to determine the pressure drop mode caused by background noise.

[0033] By fusing multiple sensors, a gas injection time series is extracted from the cleaned cavity sealing data to obtain the initial pressure value after injection. Based on the initial pressure value, a temperature compensation correction method is used to apply a temperature offset adjustment to the time series to construct a pressure change curve, determining a smoothed curve sequence. For the smoothed curve sequence, a differential analysis algorithm is used to calculate the difference between adjacent points, and the ratio of the average difference to the standard deviation is used to determine the background noise impact threshold. If the background noise impact threshold exceeds a preset range, a low-pass filter is applied to the difference sequence through a noise filtering mechanism to separate the interfering components and obtain a pure decreasing trend. Based on the pure decreasing trend, pattern matching verification is performed, and the similarity between the trend sequence and a preset noise pattern sequence is calculated to determine the pressure decrease pattern caused by background noise.

[0034] In one implementation, the cavity is first cleaned to obtain sealing data. This data includes the initial pressure value inside the cavity, temperature parameters, and the geometry of the sealing interface. This data is acquired using dedicated sensors to ensure it is free from contaminant interference.

[0035] Specifically, the cleaning process involves using high-pressure air to purge the interior of the chamber and applying vacuum suction to remove residual particles, thereby providing reliable baseline data for subsequent pressure analysis. Based on the cleaning data, pressure change curves after gas injection are then obtained.

[0036] For example, in the field of seal testing, an inert gas, such as nitrogen, is injected into a cavity at a predetermined volume, for example, 0.5 liters of gas per cubic meter of cavity. After injection, a precision pressure sensor is used to monitor the internal pressure of the cavity in real time, collecting data points once per second to form a time-series curve. This curve reflects the possible downward trend of the pressure after it rises from the initial value to the peak.

[0037] It should be noted that curve generation relies on data smoothing processes, such as using moving average filtering, to reduce random fluctuations.

[0038] Preferably, based on the acquired curve, a differential analysis algorithm is introduced to determine the pressure drop pattern caused by background noise. Differential analysis is a signal processing method that works by calculating the difference between adjacent data points to highlight the rate of change, thereby distinguishing between slow noise effects and rapid leakage drops.

[0039] Specifically, the algorithm first performs first-order difference calculation on the pressure change curve, that is, for the time series P(t), calculates ΔP(t) = P(t+1) - P(t) to obtain the difference sequence. Then, a threshold filter is applied to the difference sequence, with the threshold set based on historical noise data, for example, a threshold of 0.01 Pa / s, and changes below this value are considered background noise.

[0040] In one possible implementation, the pressure drop pattern caused by background noise is characterized by a gradual, low-amplitude decrease, unlike the steep drop of a sudden leak.

[0041] For example, in a sealed container testing scenario, if the curve shows a pressure drop of 0.05 Pa per minute and the differential value remains stable within a small range, the algorithm identifies it as a noise pattern rather than a seal failure. This pattern identification helps improve detection accuracy and avoid misjudgments. Furthermore, the implementation process of the differential analysis algorithm is explained in detail. The algorithm begins with the preprocessing stage of the curve data, including normalization, scaling the pressure values ​​to the range of 0 to 1 to unify the influence of different cavity sizes. Next, differential calculations are performed, not limited to first-order but also extended to second-order differentials, i.e., Δ²P(t) = ΔP(t+1) - ΔP(t), to capture acceleration changes, thereby more accurately identifying the randomness of noise.

[0042] It should be noted that noise patterns typically originate from environmental factors such as temperature fluctuations or sensor drift. The decreases caused by these factors follow a normal distribution. The algorithm models this by using the mean and variance of the statistical difference sequence. For example, when the mean is close to zero and the variance is less than 0.001, it is identified as a background noise pattern. The innovation of this process lies in combining statistical analysis, which improves the sensitivity to weak signals. In seal detection, it can effectively isolate noise and ensure that only true leaks are reported.

[0043] For example, in a practical application, during a sealing test of an industrial gas storage tank, the pressure curve collected after gas injection showed an initial pressure of 100 kPa, which dropped to 99.8 kPa within 10 minutes. Applying a difference algorithm, the mean of the difference sequence was calculated to be -0.0002 kPa / s, with a variance of 0.00005, which meets the noise mode threshold; therefore, no leakage was determined. This example demonstrates the practicality of the algorithm in time series analysis.

[0044] Understandably, this algorithm can be further optimized by introducing an adaptive threshold, adjusting it based on real-time environmental data. For example, the threshold could be increased in high-temperature environments to compensate for the effects of thermal expansion. This makes the determination of the pressure drop pattern more robust and supports applications in different sealing scenarios. In another embodiment, multi-sensor data augmentation analysis is incorporated. After gas injection, not only are pressure curves acquired, but temperature and humidity curves are also recorded simultaneously. The differential algorithm is extended to multi-dimensional differential, calculating the cross-difference between pressure and temperature to quantify the noise contribution.

[0045] For example, if the rate of pressure change caused by a rise in temperature matches the observed decrease, the algorithm classifies it as a noise pattern rather than a sealing problem. This method performs well in complex environments and reduces false alarm rates.

[0046] Specifically, the results show that, through this differential analysis, the accuracy rate in sealing inspection tasks can reach over 95%, which reduces the impact of noise interference and provides more reliable pattern recognition compared to traditional threshold methods.

[0047] Preferably, the entire process is integrated into an automated system, forming a closed-loop process from cleaning data acquisition to pattern determination, ensuring versatility within the same sealing testing field, such as application to cavity testing of different specifications, without introducing irrelevant scenarios.

[0048] Step S103: Determine whether gas absorption is the dominant signal by the pressure drop pattern caused by background noise. If gas absorption is the dominant signal, obtain the preset absorption compensation value from the particle gap model and subtract the value to obtain the compensated pressure change curve.

[0049] Background noise data is collected by sensors to determine the pressure drop pattern caused by the background noise, thus obtaining a pressure drop judgment result. Based on this pressure drop judgment result, it is analyzed whether gas absorption is the dominant signal to determine the dominant gas absorption status. If gas absorption is the dominant signal, a preset absorption compensation value is obtained from the particle gap model, which is constructed using preset particle distribution data and gap simulation parameters. The corresponding portion of the pressure change curve is subtracted from the absorption compensation value to generate a compensated pressure change curve. Real-time calibration is performed using the compensated pressure change curve, adjusting the curve deviation using preset calibration parameters to obtain the final compensated curve.

[0050] In one implementation, the dominance of the gas absorption signal is determined by monitoring the pressure drop pattern caused by background noise.

[0051] Specifically, background noise usually refers to random sound waves or vibrations in the environment. These disturbances can cause pressure fluctuations in gas flow systems, forming specific drop patterns.

[0052] For example, in industrial pipeline gas detection scenarios, pressure sensor data is first collected, and the frequency and amplitude changes of the noise signal are analyzed. If the pressure drop curve shows attenuation characteristics related to gas molecule absorption, such as gradual attenuation rather than abrupt change, then gas absorption is determined to be the dominant factor. This judgment is based on the physical principle that gas absorption consumes energy through intermolecular collisions, leading to a gradual weakening of the pressure signal. Furthermore, if gas absorption is determined to be the dominant signal, a preset absorption compensation value is obtained from the particle gap model. The particle gap model is a framework for simulating the influence of gaps between particles on gas flow. In this model, particles are considered porous structures, and the gaps allow gas to permeate and be absorbed.

[0053] It should be noted that the model calculates the compensation value using a pre-established database.

[0054] For example, the pressure loss caused by absorption can be estimated based on the ratio of particle density to interstitial volume. The process involves inputting current particle parameters, such as average particle size and distribution density, and then querying the corresponding compensation value from the model library. This value represents the quantified impact of the absorption effect on pressure, thus ensuring the accuracy of subsequent curves.

[0055] Preferably, after obtaining the compensation value, the pressure change curve after compensation is obtained by subtracting the compensation value.

[0056] In one possible implementation, this subtraction operation is applied to the data points of the original pressure curve.

[0057] For example, subtracting the compensation value from the pressure value at each time point creates a new curve. This compensation helps eliminate absorption bias and improves detection accuracy.

[0058] For example, in ambient air quality monitoring, if gas absorption is dominant, the compensated curve can more accurately reflect the level of particulate pollution, rather than being amplified by noise.

[0059] For example, the construction process of the particle gap model needs to take into account the particle arrangement in detail. The particle gap model assumes that the particles form a loose packing, and the gap volume is determined by the particle diameter and the packing factor.

[0060] Specifically, the model first calculates the gap ratio, which is the proportion of gap volume to the total volume, and then estimates the absorption rate based on the gas flow rate. For example, in filtration system applications, if the particle gap ratio is high, the absorption compensation value will be increased accordingly to compensate for the pressure drop caused by more gas being captured. The preset values ​​of this model are obtained through experimental calibration to ensure applicability at different particle concentrations.

[0061] In one embodiment, the detailed process of determining whether gas absorption is dominant includes comparing the slope of a pressure drop pattern. A pressure drop pattern caused by background noise manifests as a gradual change in the curve slope; if the slope exceeds a threshold and matches the absorption spectrum, dominance is confirmed.

[0062] Specifically, this judgment can be achieved through signal processing algorithms, such as Fourier transform to extract the noise spectrum and then comparing it with the gas absorption spectrum. A high match indicates absorption dominance, triggering a compensation step. This method is particularly useful in industrial gas purification scenarios, enabling real-time curve adjustments and avoiding misjudgments of pollution levels. Furthermore, the compensated pressure change curve can be used for subsequent analysis.

[0063] For example, in flue gas emission monitoring within the same field, the compensation curve shows a more stable downward trend, which helps in assessing filtration efficiency.

[0064] It should be noted that this compensation does not change the essence of the original data, but rather corrects the deviation through subtraction to achieve a more accurate dynamic description of pressure.

[0065] Understandably, in another implementation, the acquisition of the preset absorption compensation value can be optimized to dynamic querying. The particle gap model incorporates multiple sub-models corresponding to different gas types, such as carbon dioxide or nitrogen. For example, for high-concentration gases, the model adjusts the gap parameters to generate a specific compensation value, which is then subtracted to obtain the curve. This flexibility enhances the versatility of the technical solution in the field of gas detection.

[0066] Specifically, the entire process logically follows a sequence from noise acquisition to judgment and compensation, ensuring the reliability of the output curve. In particle-dense environments, this method effectively distinguishes between absorption and noise effects, providing accurate data to support decision-making.

[0067] Step S104: For the compensated pressure change curve, obtain time series data, and use Fourier transform algorithm to determine whether there are periodic fluctuations, thereby determining the small signal components caused by the actual leakage.

[0068] For the compensated pressure change curve, time-series data is acquired and processed using a Fourier transform algorithm to obtain a frequency spectrum. Peak frequency components are extracted from the frequency spectrum to determine if periodic fluctuations exist. If such periodic fluctuations are present, a potential signal pattern is identified. Noise filtering is applied to the potential signal pattern to obtain a filtered signal, and the amplitude of minor fluctuations is determined. The filtered signal is compared with a preset threshold; if the amplitude exceeds the preset threshold, it is determined to be a leak-related component. From the leak-related components, the minute signal components caused by actual leaks are extracted.

[0069] In one implementation, the compensated pressure change curve is processed by first acquiring time series data.

[0070] Specifically, the compensated pressure change curve is a continuous curve obtained by correcting the original pressure data for factors such as temperature and environmental noise. Based on this, the curve is converted into a time series format.

[0071] For example, pressure values ​​are sampled at fixed time intervals to form a sequence consisting of time points and their corresponding pressure values. This sequence captures the dynamic changes in pressure within the system for subsequent analysis. Furthermore, the time series data is processed using a Fourier transform algorithm. The Fourier transform is a mathematical method that converts a time-domain signal into a frequency-domain signal, revealing hidden periodic patterns by calculating the frequency components of the sequence.

[0072] For example, in a pipeline leak detection scenario, the time series is input into a Fourier transform formula to obtain an amplitude spectrum and a phase spectrum, where the amplitude spectrum shows the energy distribution at different frequencies. If periodic fluctuations exist, these fluctuations will appear as obvious peaks in the spectrum.

[0073] It should be noted that determining whether periodic fluctuations exist involves analyzing the peak characteristics in the spectrum.

[0074] Specifically, a frequency threshold is set; for example, for a typical piping system, a low-frequency range below a certain hertz is selected as the focus. By comparing the peak amplitude with the average noise level, if the peak exceeds a preset multiple, it is determined that periodic fluctuations exist. This judgment helps to distinguish between periodic signals caused by environmental interference and non-periodic minute changes caused by actual leakage.

[0075] In one possible implementation, to determine the minute signal components caused by the actual leakage, the non-periodic components are further extracted.

[0076] Preferably, after removing the detected periodic components, filtering techniques, such as low-pass filters, are applied to the remaining signal to isolate minute leakage signals.

[0077] For example, in natural gas pipeline monitoring, leaks may cause slight irregular fluctuations in the pressure curve. By reconstructing the signal through inverse Fourier transform, these components can be highlighted, thereby achieving accurate detection.

[0078] Understandably, this method is applicable to multiple scenarios within the same field. For example, in an oil pipeline system, time-series data can be collected from multiple sensors, and Fourier transform processes each sequence in parallel to determine the overall periodicity. In another embodiment, when detecting water pipe networks, the transform window size is adjusted to accommodate different pipe diameters, enhancing sensitivity to minute signals.

[0079] Specifically, the implementation of the Fourier transform includes the computational steps of the Discrete Fourier Transform. First, the time series is weighted using a window function to reduce edge effects, and then the Fast Fourier Transform algorithm is executed to obtain the spectral data. Based on this, periodic fluctuations are identified by searching for harmonic peaks in the spectrum.

[0080] For example, if the peak value corresponds to a known pumping cycle, it is marked as interference; otherwise, it is considered a potential leak signal. This detailed procedure ensures the reliability of the judgment and, in practical applications, can effectively reduce the false alarm rate and improve the accuracy of leak detection.

[0081] For example, in industrial pipeline maintenance scenarios, this technology can be integrated into a monitoring system to acquire pressure curve data in real time and apply Fourier transform to quickly identify the components of even minor leaks, thereby triggering alarms promptly and preventing greater losses. Furthermore, in another implementation, multiple sets of time-series data are combined for joint analysis.

[0082] Preferably, sequences are extracted from the compensated pressure curves, Fourier transforms are applied, and the periodicity consistency of each spectrum is compared to confirm the global nature of the actual leakage signal. This method expands the applicability of the technology and performs excellently in complex pipe networks.

[0083] Step S105: Obtain peak amplitude data from the tiny signal components caused by actual leakage, and determine the seal integrity by comparing it with a preset leakage threshold. If the peak amplitude data exceeds the threshold, it is marked as a leakage defect to obtain the defect classification result.

[0084] Initial signal data is obtained from the minute signal components caused by actual leakage. High-frequency noise interference is removed through low-pass filtering to obtain clean signal components. Peak amplitude data is extracted from these clean signal components, and peak detection quantization is used to determine the amplitude values. A preset leakage threshold is obtained, and the amplitude values ​​are compared with the threshold. If the amplitude value exceeds the threshold, it is determined to be a potential leakage. Based on the potential leakage, the leakage defect location is marked, and the defect classification result is obtained from multi-point signal sampling verification.

[0085] In one implementation, the seal integrity testing process first analyzes the minute signal components caused by actual leaks. These minute signal components typically originate from subtle changes in the seal structure under pressure or vibration; for example, in pipeline seal testing, the presence of a tiny crack will generate specific acoustic or vibration signals. After these signals are acquired by sensors, peak amplitude data needs to be extracted from them.

[0086] Specifically, signal acquisition devices, such as acoustic sensors, are installed near sealed areas to monitor subtle fluctuations in the environment in real time. The process of acquiring peak amplitude data includes filtering the acquired signal to remove noise interference, and then identifying the point of maximum amplitude in the signal.

[0087] For example, in a sealed container inspection scenario, after the signal is converted into digital form, an amplitude calculation method is used to determine the peak value, i.e., the highest point value of the signal waveform. This extraction method ensures the accuracy of the data and provides a basis for subsequent judgment. Further, the acquired peak amplitude data is compared with a preset leakage threshold to determine the seal integrity. The preset leakage threshold is a reference value determined based on historical data and experiments. For example, in the field of industrial pipeline sealing, the threshold may be set to a specific amplitude, such as 0.5 units, representing the upper limit of the signal under normal sealing conditions. The comparison process involves directly comparing the extracted peak amplitude with the threshold. If the peak amplitude exceeds the threshold, it indicates a potential leakage risk. This judgment logic relies on the principle of signal processing, namely that a real leak amplifies tiny signal components, causing an abnormal increase in amplitude.

[0088] In one possible implementation, the comparison can be achieved through a software module that takes peak data as input and outputs the comparison result, thereby enabling automated detection.

[0089] It should be noted that if the peak amplitude data exceeds a threshold, it is marked as a leakage defect, resulting in a defect classification result. This marking process involves classifying signal points exceeding the threshold into defect categories, such as "minor leakage" or "serious leakage," depending on the degree to which the amplitude exceeds the threshold.

[0090] Specifically, in a seal inspection system, the classification results can be further divided into multiple levels. For example, a defect exceeding a threshold by less than 10% is marked as a minor defect, while a defect exceeding that threshold is classified as a serious defect. This classification aids in subsequent maintenance decisions. For instance, in an oil pipeline sealing scenario, a minor defect may only require localized repair, while a serious defect necessitates complete replacement. In this way, the defect classification results not only provide information on leak detection but also support risk assessment.

[0091] For example, in a specific seal inspection embodiment, the integrity assessment of a high-pressure vessel seal is considered. First, minute signal components, originating from vibrations caused by pressure changes within the vessel, are acquired. Peak amplitude data is obtained through time-domain analysis, such as calculating the maximum value in the signal sequence. This is then compared to a threshold; if the amplitude exceeds a preset value, such as 1.0 unit, it is marked as a leakage defect. Results show that this method effectively identifies early signs of leakage in practical testing, improving the timeliness of seal maintenance.

[0092] Preferably, in another embodiment, for the detection of vacuum sealing equipment, the minute signal components may include thermal radiation or air pressure fluctuation signals. When obtaining the peak amplitude, frequency domain analysis can be used as an aid, for example, by identifying the amplitude peak at the dominant frequency through Fourier transform. Then, a threshold comparison is performed, and if the threshold is exceeded, the defect classification result is marked as "vacuum leakage". This approach expands the application of the technical solution within the same field, demonstrating its adaptability to different sealing types.

[0093] In one embodiment, the generation of defect classification results can also incorporate a multi-threshold mechanism.

[0094] For example, a primary threshold and a high threshold can be set, with peak values ​​between the two marked as "potential defects" and those exceeding the high threshold as "confirmed leaks." This hierarchical classification is particularly useful in pharmaceutical container seal testing, as it can differentiate between different risk levels, thereby optimizing resource allocation.

[0095] Understandably, the overall process described above ensures continuity from signal extraction to classification. In practice, the system can integrate a display module to output the defect classification results for user review.

[0096] For example, in a laboratory sealing test, once the peak amplitude exceeds a threshold, the system automatically generates a report, marking the specific defect location. Furthermore, to enhance detection accuracy, peak amplitude data acquisition may include a calibration step.

[0097] Specifically, before signal acquisition, the sensor is calibrated to match the specific sealed environment, such as adjusting sensitivity to capture minute components. This calibration helps reduce false alarms and improves reliability in industrial applications.

[0098] In one embodiment, considering a continuous monitoring scenario, peak amplitude data is updated in real time and compared with a threshold. If the threshold is exceeded multiple times, it is cumulatively marked as a persistent leakage defect. This dynamic classification result supports long-term seal integrity management; for example, in chemical pipelines, continuous monitoring can detect progressive defects early.

[0099] Step S106: Based on the defect classification results, obtain the continuous tank sequence data for online full inspection, determine whether the number of continuous defects in the sequence exceeds the preset number, and if it exceeds the preset number, adjust the detection parameters to obtain optimized detection process data.

[0100] Continuous sequence data is obtained from the defect classification results. It is determined that the number of consecutive defects in the tank exceeds a preset threshold, resulting in a defect count. If the defect count exceeds the preset threshold, the detection parameters for the online full inspection process are adjusted, and a parameter adjustment mechanism is determined. The results of the sequence data are processed through this parameter adjustment mechanism, and defect pattern recognition is used to further subdivide defect types based on defect shape and location, resulting in a process optimization result. The process optimization result is then fused with the defect classification result to form a parameter feedback loop, which determines if consecutive defects exceed the threshold, resulting in optimized inspection process data.

[0101] In one implementation, obtaining continuous tank sequence data for online full inspection based on defect classification results requires understanding that online full inspection refers to a real-time, comprehensive defect detection process for each tank on the tank production line. This involves capturing images of the tank surface using optical sensors or imaging equipment and applying classification algorithms to identify defect types such as scratches, dents, or contamination. The defect classification result can be a sequence of defect labels for each tank, categorizing defects as minor, moderate, or severe. Continuous tank sequence data refers to the chronologically ordered data stream of multiple tanks on the production line, such as defect records from the first to the tenth tank, forming a time-series sequence for subsequent analysis. This acquisition method ensures data continuity and real-time performance, providing a basis for judging continuous defects. Furthermore, determining whether continuous defects in the sequence exceed a preset number can be achieved by traversing the sequence data.

[0102] Specifically, the preset quantity can be set according to production standards. For example, setting it to 3 means that no more than 3 consecutive tanks are allowed to have the same type of defect.

[0103] For example, in a sequence, if the first three tanks are all classified as having the same scratch defect, the count is 3; if the fourth tank is defect-free, the count is reset.

[0104] It should be noted that this judgment process involves a sliding window mechanism, which checks the defect labels of adjacent tanks one by one. If the consecutive defect count exceeds a preset value, such as 5, an alarm is triggered or subsequent adjustments are made. This mechanism helps to identify potential systemic problems on the production line early, such as the occurrence of consecutive defects caused by equipment wear.

[0105] Preferably, if the judgment result shows that the number of consecutive defects exceeds the preset number, the detection parameters are adjusted to obtain optimized detection process data.

[0106] For example, in a tank inspection system, inspection parameters include image resolution, defect threshold, or the sensitivity of the classification algorithm. The adjustment process can involve gradually increasing the resolution, for example, from a standard 800x600 pixels to 1024x768 pixels, to improve defect recognition accuracy; or decreasing the defect threshold, for example, from 0.5 to 0.3, to make the system more sensitive to minor defects. Through these adjustments, the resulting optimized inspection process data can be an updated parameter configuration file used to reconfigure the inspection equipment, ensuring a reduced defect rate in subsequent tank sequences.

[0107] In one possible implementation, consider scenarios with different can types, such as a metal beverage can production line, where the sequence data might include defect records for 100 consecutive cans. When judging consecutive defects, if four consecutive cans have dents exceeding a preset number of three, parameters are adjusted, such as increasing lighting intensity, to improve image capture quality. This scenario demonstrates the application of the technology on a high-speed production line, enabling dynamic optimization of the inspection process.

[0108] Specifically, in another embodiment, for online full inspection of plastic cans, the determination of consecutive defects can be based on historical data, such as analyzing sequences from the past hour. If the number of consecutive defects exceeds a preset limit, such as 6, the parameters of the classification algorithm are adjusted, for example, by modifying the decision boundary of the support vector machine to adapt to specific defect patterns. This adjustment yields optimized inspection process data, which helps reduce misjudgments and improve overall production efficiency.

[0109] Understandably, in the above process, the steps of acquiring sequence data and making judgments are closely linked. For example, the sequence data is directly input into the judgment module to ensure smooth logic. When further adjusting parameters, a feedback loop is generated based on the judgment result. For example, the sequence is re-verified after automatically updating the parameters to confirm the optimization effect.

[0110] For example, in a specific scenario, when a production line is inspecting glass jars, if seven consecutive jars in the sequence show contamination defects exceeding the preset value of 5, the detection parameters are adjusted, such as increasing the sensor sensitivity from the standard value of 80% to 95%, thereby obtaining optimized process data containing the new parameters for subsequent batch inspections.

[0111] It should be noted that the versatility of this technical solution is reflected in various tank materials, such as aluminum alloy tanks. The process of judging continuous defects is the same, but the adjustment parameters can be tailored to the material characteristics. For example, the angle of the light source can be adjusted to suit the reflectivity of aluminum alloys.

[0112] In one embodiment, the entire process forms a closed loop, starting with defect classification results and ending with optimized data output. For example, parameter adjustments can be implemented via a software interface, making it suitable for continuous production environments. Furthermore, this method is effective in responding promptly to continuous defects and improving detection accuracy. For instance, in practical applications, it reduces batch defect problems caused by fixed parameters.

[0113] Step S107: Obtain the pressure change curve of the tank through the optimized detection process data, and repeat the compensation process to determine the final seal integrity result.

[0114] By optimizing the detection process data, the pressure change curve of the subsequent tank is obtained, yielding preliminary pressure distribution characteristics. For this pressure change curve, the compensation process is repeated, and adjustment parameters are obtained from temperature-related factors to determine the compensated curve. Based on the compensated curve, the differences between multiple tanks are compared to obtain the sealing deviation value. If the sealing deviation value exceeds a preset threshold, the compensation process is repeated to obtain the final sealing integrity result.

[0115] In one implementation, the pressure change curve of the tank is obtained through optimized detection process data. First, the initial detection data needs to be processed.

[0116] Specifically, the detection process data comes from pressure sensors inside the tank. These sensors collect pressure values ​​in real time and use filtering algorithms to remove noise interference.

[0117] For example, a moving average filtering method can be used to smooth the collected pressure sequence, thereby obtaining more accurate optimized data. Based on this, the subsequent pressure change curve of the tank is fitted using this optimized data. The curve generation is based on a multinomial regression model, with time as the independent variable and pressure value as the dependent variable, to calculate the functional expression of the curve. This method ensures the continuity and predictability of the curve, effectively reflecting the pressure change over time in the field of tank seal inspection. Furthermore, the repeated compensation process to determine the final seal integrity result is based on iterative adjustments of the aforementioned pressure change curve.

[0118] In one possible implementation, the compensation process involves deviation analysis of the curve data. First, the difference between the actual measured curve and the ideal sealing model curve is identified, for example, by quantifying the deviation through calculating the root mean square error. Then, a compensation algorithm, such as introducing a temperature compensation factor, is repeatedly applied to correct the curve. The curve parameters are updated after each iteration until the deviation falls below a preset threshold.

[0119] It should be noted that this repeated compensation can be cycled 3 to 5 times, depending on the size and material type of the tank. For example, in the testing of large oil storage tanks, the compensation process will take into account external environmental factors such as atmospheric pressure, thereby gradually approaching the actual sealing state.

[0120] For example, in an industrial application of tank seal integrity testing, assuming an operation is being performed on a chemical storage tank, initial pressure data is first collected and optimized to generate a preliminary curve. Then, a compensation phase begins. The first compensation may adjust for sensor accuracy errors, calculating a corrected curve. The second compensation focuses on pressure fluctuations caused by material expansion, repeatedly calculating using an expansion coefficient model, ultimately outputting the seal integrity result, such as intact, leak-free, or requiring repair. This multi-round compensation enhances the reliability of the test.

[0121] Preferably, in another embodiment, the optimized detection process for acquiring pressure change curves can combine multi-sensor fusion, such as integrating data from internal pressure sensors and external vibration sensors, and generating a comprehensive curve through a weighted average algorithm. The specific workflow of this fusion process includes data synchronization, alignment, and weight allocation; for example, assigning higher weights to pressure data to highlight seal-related changes, thereby more accurately determining results in repeated compensation. In the field of tank inspection, this method is applicable to high-pressure gas tanks, ensuring that the compensation process covers various pressure fluctuation scenarios.

[0122] Understandably, the principle of the repeated compensation process lies in iteratively optimizing the error model, for example, by gradually minimizing the curve deviation through the gradient descent method, without involving complex numerical calculations.

[0123] Specifically, in one embodiment, for the sealing test of food storage tanks, after obtaining the initial curve, the first round of compensation analysis analyzes the effect of temperature on pressure and adjusts the slope of the curve; subsequent rounds compensate for humidity factors, repeating the process until the curve stabilizes, and the results show that the sealing integrity meets the standards. This process can bring higher detection accuracy and reduce the risk of misjudgment in business operations.

[0124] In one embodiment, the entire process, from data optimization to compensation for repetitions, can be integrated into an automated testing system, for example, using an embedded controller to process data in real time. When acquiring a curve, the system first verifies data integrity and then generates the curve; during compensation, the system iteratively executes correction steps and finally outputs an integrity assessment report. This integration demonstrates the versatility of the technical solution in the field of tank inspection. Furthermore, to support widespread application, in laboratory scenarios for tank seal testing, optimized data can be used to generate virtual pressure curves through simulation software, and then repeated compensation can be performed to verify the algorithm's effectiveness.

[0125] For example, by simulating curves for different levels of leakage and performing multiple compensations to determine the result, this approach expands the flexibility of the solution without exceeding the scope of the domain.

[0126] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for detecting the seal of a solid beverage can, characterized in that, include: Initial pressure data is obtained from the detection chamber, and a preset threshold is used to determine whether there is any contamination. If the initial pressure data is lower than a preset threshold, the contamination location is determined and the cleaning mechanism is activated to obtain the cleaned cavity sealing data. Based on the cleaned cavity sealing data, the pressure change curve after gas injection is obtained, and the pressure drop pattern caused by background noise is determined by an analysis algorithm. The pressure drop pattern is used to determine whether gas absorption is the dominant signal. If gas absorption is the dominant signal, a compensation value is obtained from a preset model to generate a compensated pressure change curve. For the compensated pressure change curve, time series data is obtained, and a transformation algorithm is used to determine whether there are periodic fluctuations to identify the signal components caused by the actual leakage. Peak amplitude data is extracted from the signal components, and the seal integrity is determined by comparing it with a preset leakage threshold. If the peak amplitude data exceeds the preset leakage threshold, it is marked as a leakage defect, and the defect classification result is obtained. Based on the defect classification results, continuous sequence data is obtained, and it is determined whether the continuous defects exceed the preset number threshold. If they do, the detection parameters are adjusted to generate optimized detection process data. Using the optimized detection process data, the subsequent pressure change curve is obtained, the compensation process is repeated, and the final seal integrity result is determined.

2. The method for detecting the seal of a solid beverage can as described in claim 1, characterized in that, The step of acquiring initial pressure data from the detection chamber and determining whether there is a contamination impact by using a preset threshold includes: collecting initial pressure data of the detection chamber through a sensor and comparing the initial pressure data with a preset threshold; if the initial pressure data is lower than the preset threshold, it is determined that there is a contamination impact, and the deviation value of the initial pressure data is recorded; based on the deviation value, a contamination impact assessment report is generated to determine the preliminary range of the contamination impact; based on the preliminary range and in conjunction with historical data comparison, the severity of the contamination impact is determined; based on the severity, subsequent processing procedures are activated to obtain quantitative indicators of the contamination impact; based on the quantitative indicators, it is determined whether further detection is needed, and initial detection results are generated; based on the initial detection results, the status record of the detection chamber is updated to provide data support for subsequent cleaning mechanisms; based on the status record, a detection log is generated to save the comparison results of the initial pressure data; based on the detection log, a tracking record of the contamination impact is constructed to ensure the continuity of subsequent steps.

3. The method for detecting the seal of a solid beverage can as described in claim 1, characterized in that, If the initial pressure data is lower than a preset threshold, the contamination location is determined and the cleaning mechanism is activated to obtain the cleaned cavity sealing data. This includes: processing the surface data of the detected cavity using an image recognition algorithm to locate the contamination attachment location; generating a contamination distribution map for the contamination attachment location to determine the boundary of the contamination area; adjusting the operating parameters of the cleaning mechanism and activating the cleaning device based on the contamination distribution map; performing targeted cleaning of the contamination area using the cleaning device and obtaining real-time feedback data of the cleaning process; judging the cleaning progress based on the feedback data and determining whether the cleaning has reached a preset standard; if the cleaning has reached the preset standard, stopping the cleaning mechanism and obtaining the cleaned cavity sealing data; comparing the cleaned cavity sealing data with the initial pressure data to evaluate the cleaning effect; generating a cleaning effect report based on the cleaning effect and recording the trend of the sealing data change; updating the cavity status record based on the trend of change to provide a reference for subsequent testing.

4. The method for detecting the seal of a solid beverage can as described in claim 1, characterized in that, The step of obtaining the pressure change curve after gas injection based on the cleaned cavity sealing data and determining the pressure drop pattern caused by background noise through analysis algorithms includes: collecting the cleaned cavity sealing data through multi-point sensors and extracting the gas injection time series; obtaining the initial pressure value after injection based on the time series and adjusting the data using a temperature compensation correction method; constructing a pressure change curve using the adjusted data and generating a smoothed curve sequence; calculating the difference between adjacent points using a differential analysis algorithm for the smoothed curve sequence and obtaining a difference sequence; calculating the average and standard deviation of the difference based on the difference sequence to determine the background noise impact threshold; if the background noise impact threshold exceeds a preset range, separating the interfering components through a noise filtering mechanism; obtaining the pure downward trend through the separated data and determining the pressure drop pattern; performing pattern matching verification on the pressure drop pattern and calculating the similarity with a preset noise pattern; generating a background noise impact report based on the similarity and recording the characteristics of the pressure drop pattern.

5. The method for detecting the seal of a solid beverage can as described in claim 1, characterized in that, The process of determining whether gas absorption is the dominant signal based on the pressure drop pattern, and if so, obtaining a compensation value from a preset model to generate a compensated pressure change curve, includes: collecting background noise data through a sensor, analyzing the pressure drop pattern, and obtaining a pressure drop judgment result; determining whether gas absorption is the dominant signal based on the judgment result and generating a dominant status assessment; if gas absorption is the dominant signal, extracting a preset absorption compensation value from the particle gap model; adjusting the corresponding part of the pressure change curve using the absorption compensation value to generate a compensated curve; performing real-time calibration on the compensated curve using preset calibration parameters; adjusting the curve deviation based on the calibration process to obtain a calibrated pressure change curve; generating a compensation effect report based on the calibrated curve and recording the data comparison before and after adjustment; updating the status record of the pressure change curve based on the comparison result; and generating a compensation process log based on the status record to provide data support for subsequent analysis.

6. The method for detecting the seal of a solid beverage can as described in claim 1, characterized in that, The process of acquiring time-series data from the compensated pressure change curve and determining the presence of periodic fluctuations using a transformation algorithm to identify the signal components caused by the actual leak includes: extracting time-series data from the compensated pressure change curve and processing the data using a Fourier transform algorithm; generating a frequency spectrum from the processed data and extracting peak frequency components; determining the presence of periodic fluctuations based on the peak frequency components and identifying potential signal patterns; applying noise filtering to the potential signal patterns to obtain filtered signals; determining the amplitude of minute fluctuations from the filtered signals and comparing them with a preset threshold; if the amplitude of minute fluctuations exceeds the preset threshold, it is determined to be a leak-related component; extracting the signal components caused by the actual leak based on the leak-related components and generating a signal analysis report; recording the quantification results of the fluctuation amplitude in the signal analysis report; and updating the feature records of the signal components based on the quantification results to provide a basis for subsequent detection.

7. The method for detecting the seal of a solid beverage can as described in claim 1, characterized in that, The process of extracting peak amplitude data from the signal components and comparing it with a preset leakage threshold to determine the seal integrity, and marking a leakage defect if the peak amplitude data exceeds the preset leakage threshold, and obtaining a defect classification result, includes: obtaining initial signal data from the signal components and removing high-frequency noise through low-pass filtering; obtaining clean signal components from the processed data and extracting peak amplitude data; determining the amplitude value using a peak detection quantization method for the peak amplitude data; comparing the amplitude value with a preset leakage threshold to determine if there is a potential leak; marking the location of the leakage defect if the amplitude value exceeds the preset leakage threshold; performing multi-point sampling verification of the signal based on the leakage defect location to obtain a defect classification result; generating a defect analysis report based on the defect classification result and recording the comparison results of the peak amplitude data; updating the recorded data of the defect location based on the comparison results; and generating a defect tracking log based on the recorded data to support subsequent processes.

8. The method for detecting the seal of a solid beverage can as described in claim 1, characterized in that, The process of obtaining continuous sequence data based on the defect classification results, determining whether continuous defects exceed a preset quantity threshold, and adjusting detection parameters to generate optimized detection process data includes: extracting continuous sequence data from the defect classification results and counting the number of consecutive tank defects; comparing the number of consecutive defects with a preset quantity threshold to obtain a defect quantity statistics result; adjusting the detection parameters of the online full inspection if the defect quantity statistics result exceeds the preset quantity threshold; determining a parameter adjustment mechanism based on the adjusted parameters and processing the sequence data; using defect pattern recognition to subdivide defect types based on the processed data; generating process optimization results based on the subdivided defect types and recording the parameter adjustment effect; integrating the defect classification results with the process optimization results to form a parameter feedback loop; determining the continuous defect exceedance situation through the feedback loop and generating optimized detection process data; and updating the operation record of the detection system based on the detection process data.

9. The method for detecting the seal of a solid beverage can as described in claim 1, characterized in that, The process of obtaining subsequent pressure change curves through the optimized detection process data, repeating the compensation process, and determining the final seal integrity result includes: acquiring subsequent pressure change curves of the tanks through the optimized detection process data; obtaining preliminary pressure distribution characteristics based on the pressure change curves and recording the distribution data; repeating the compensation process for the distribution data and obtaining adjustment parameters from temperature influencing factors; generating a compensated curve using the adjustment parameters and recording the compensated data characteristics; determining the comparison differences between multiple tanks based on the compensated curves and obtaining the seal deviation value; if the seal deviation value exceeds a preset threshold, repeating the compensation process and adjusting the curve data; determining the final seal integrity result and generating a detection report using the adjusted curve data; recording the quantitative indicators of the seal integrity result in the detection report; and updating the status record of the detection system based on the quantitative indicators to provide a reference for subsequent detection.

10. A tank body, characterized in that, The container includes a can body, which is tested and optimized using a sealing detection method for solid beverage cans according to any one of claims 1-9.

Images