A leakage detection method and system for transformer oil tank production

Abstract

The present application relates to the field of measurement and testing, and relates to a kind of leakage detection method and system for transformer oil tank production, comprising: the oil tank to be measured is divided into multiple test areas;The pressure drop average rate of oil tank cavity in target time period, the pressure change rate corresponding to temperature change of target test area and the volume change amount are used to determine ideal net pressure decay rate;Ideal net pressure decay rate and measured net pressure decay rate are used to determine pressure decay deviation;The acoustic activity degree of acoustic signal in acoustic event and the background noise average energy of multiple test areas in the acoustic event are used to determine acoustic activity weight;Acoustic activity weight, pressure decay deviation are used to determine regional leakage comprehensive index, and then the leakage detection condition of oil tank is determined using regional leakage comprehensive index;The technical scheme of the present application effectively improves the accuracy and reliability of leakage detection, effectively eliminates environmental temperature interference and dynamically adapts to background noise change.

Description

Technical Field

[0001] This invention relates to the field of measurement and testing technology, specifically to a leakage detection method and system for transformer tank production. Background Technology

[0002] Leak detection is a crucial step in ensuring product quality during transformer tank production. While the traditional negative pressure wave attenuation method is suitable for rapid online full inspection, its core indicator, pressure attenuation rate, is easily affected by ambient temperature fluctuations. Temperature changes cause gas expansion and contraction, resulting in significant pressure changes, making it difficult to accurately identify pressure losses caused by actual leaks. For example, a rise in temperature causes gas expansion and increased pressure, potentially masking minor leaks; a drop in temperature causes pressure decrease, which can easily be misjudged as a leak. Furthermore, existing acoustic signal processing technologies use fixed energy thresholds to identify ultrasonic events, which cannot adapt to changing background noise. In noisy industrial environments, the signal-to-noise ratio is low, often misjudging noise as a leak or missing real signals. When introducing auxiliary features such as acoustics, the system cannot dynamically adjust weights based on real-time changes in signal quality. For example, the reliability of acoustic features decreases when background noise suddenly increases, while fixed weights introduce interference, reducing decision reliability. These problems severely affect the accuracy and reliability of transformer tank leak detection, necessitating a new method that effectively addresses these technical issues.

[0003] To address the aforementioned issues, existing technologies still need improvement. Summary of the Invention

[0004] In order to solve the existing technical problems, the purpose of this invention is to provide a leakage detection method and system for transformer tank production, which can effectively improve the accuracy and reliability of leakage detection, effectively eliminate environmental temperature interference, and dynamically adapt to changes in background noise.

[0005] To address the aforementioned technical problems, this application adopts the following technical solution: a leakage detection method for transformer tank production, comprising: dividing the tank under test into multiple test areas; determining the ideal net pressure attenuation rate by utilizing the average rate of pressure drop in the tank cavity, the pressure change rate corresponding to temperature changes, and the volume change in the target test area within a target time period; determining the pressure attenuation deviation by utilizing the ideal net pressure attenuation rate and the measured net pressure attenuation rate; determining the acoustic activity weight by utilizing the acoustic activity of the acoustic signal in the acoustic event and the average background noise energy of the multiple test areas in the acoustic event; determining the regional leakage comprehensive index by utilizing the acoustic activity weight and the pressure attenuation deviation, and then determining the leakage detection status of the tank by utilizing the regional leakage comprehensive index.

[0006] In one embodiment of the present invention, determining the ideal net pressure decay rate by utilizing the average rate of pressure drop, the rate of pressure change corresponding to temperature change, and the volume change of the fuel tank cavity in the target test area within a target time period includes: acquiring the initial pressure and initial ambient temperature of each test area at the start of the pressure holding period, and acquiring the end pressure and end ambient temperature corresponding to the end of the pressure holding period; determining the average rate of pressure drop in the fuel tank cavity of each test area within the target time period using the initial pressure, the end pressure, the start time of the pressure holding period, and the end time of the pressure holding period; determining the ideal pressure value using the initial pressure, the initial ambient temperature, and the change in ambient temperature; and determining the ideal net pressure decay rate using the ideal pressure value, the start time of the pressure holding period, the end time of the pressure holding period, and the volume change of the fuel tank cavity within the pressure holding period.

[0007] In one embodiment of the present invention, determining the ideal net pressure decay rate using the ideal pressure value, the pressure holding start time, the pressure holding end time, and the volume change of the tank cavity during the pressure holding period includes: obtaining the time difference between the pressure holding start time and the pressure holding end time; determining the pressure change rate corresponding to the temperature change using the ratio between the time difference and the ideal pressure value; obtaining the initial volume of the tank cavity; determining the volume change using the initial volume; and determining the corrected ideal pressure value using the volume change; and determining the ideal net pressure decay rate using the corrected ideal pressure value, the average pressure drop rate, and the pressure change rate.

[0008] In one embodiment of the present invention, determining the pressure decay deviation using the ideal net pressure decay rate and the measured net pressure decay rate includes: obtaining the average rate of pressure drop in the tank cavity of each test area within a target time period, and obtaining the standard deviation of the ratio corresponding to the average rate of pressure drop in all historical testing processes; and using the standard deviation, the ideal net pressure decay rate, and the measured net pressure decay rate to determine the pressure decay deviation.

[0009] In one embodiment of the present invention, determining the acoustic activity weight by utilizing the acoustic activity of the acoustic signal in an acoustic event and the average background noise energy of multiple test areas in the acoustic event includes: acquiring the energy of the acoustic signal collected by a background noise sampling window of a target size, and determining the average background noise energy by using the sum of squares of the signals corresponding to each frame in the background noise sampling window and the mean of the sum of squares of all frames in the background noise sampling window; acquiring the acoustic activity; and determining the acoustic activity weight by using the acoustic activity and the average background noise energy.

[0010] In one embodiment of the present invention, obtaining the acoustic activity degree includes: obtaining the smoothness of the ideal pressure curve and the smoothness of the measured pressure curve; determining the curve smoothness deviation using the smoothness of the ideal pressure curve and the smoothness of the measured pressure curve; and normalizing the curve smoothness deviation to obtain the acoustic activity degree.

[0011] In one embodiment of the present invention, the method further includes: acquiring multiple signal segments identified as acoustic events, and acquiring the energy integral of the acoustic events during the holding period to determine the measured average energy of the acoustic events corresponding to each signal segment; and determining the acoustic activity weight using the measured average energy of the acoustic events, the acoustic activity level, and the average energy of the background noise.

[0012] In one embodiment of the present invention, the step of determining a comprehensive regional leakage index using the acoustic activity weight and the pressure attenuation deviation, and then determining the leakage detection status of the fuel tank using the comprehensive regional leakage index, includes: obtaining a preset minimum value; determining a pressure deviation weight using the pressure attenuation deviation; determining the comprehensive regional leakage index using the pressure attenuation deviation, the pressure deviation weight, the acoustic activity weight, and the preset minimum value; and determining the leakage detection status of the fuel tank using the comprehensive regional leakage index, wherein, in response to the comprehensive regional leakage index of the target test area being greater than a first threshold, it is determined that the target test area with the comprehensive regional leakage index greater than the first threshold has a leakage; or, in response to at least one of the multiple comprehensive regional leakage indices of the target test area being less than the first threshold, it is determined that the target test area has interference.

[0013] In one embodiment of the present invention, determining the regional leakage comprehensive index using the pressure attenuation deviation, the pressure deviation weight, the acoustic activity weight, and the preset minimum value includes: determining a first value using the pressure attenuation deviation and the pressure deviation weight; determining a second value using the acoustic activity weight and the acoustic activity level; determining a third value using the pressure deviation weight, the acoustic activity weight, and the preset minimum value; and determining the regional leakage comprehensive index using the first value, the second value, and the third value.

[0014] To address the aforementioned technical problems, another technical solution adopted in this application is to provide a leakage detection system for transformer tank production, comprising: a division module for dividing the tank under test into multiple test areas; a first determination module for obtaining an ideal net pressure attenuation rate by utilizing the average rate of pressure drop in the tank cavity, the pressure change rate corresponding to temperature change, and the volume change in the target test area within a target time period; a second determination module for determining a pressure attenuation deviation by utilizing the ideal net pressure attenuation rate and the measured net pressure attenuation rate; a third determination module for determining an acoustic activity weight by utilizing the acoustic activity of the acoustic signal in an acoustic event and the average background noise energy of multiple test areas in the acoustic event; and a fourth determination module for determining a comprehensive leakage index for the area by utilizing the acoustic activity weight and the pressure attenuation deviation, and then using the comprehensive leakage index for the area to determine the leakage detection status of the tank.

[0015] The beneficial effects of the present invention are as follows: The leakage detection method and system provided for transformer tank production effectively eliminate the interference of temperature changes on the pressure attenuation rate by combining the pressure attenuation deviation degree and the dynamic acoustic activity weight to calculate the regional leakage comprehensive index, and adaptively adjust the acoustic signal weight, thereby effectively improving the accuracy and reliability of leakage detection, effectively eliminating environmental temperature interference, and dynamically adapting to changes in background noise. Attached Figure Description

[0016] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating the leakage detection method for transformer tank production provided by the present invention.

[0018] Figure 2 This is a schematic diagram of the leakage detection system for transformer tank production provided by the present invention. Detailed Implementation

[0019] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a leakage detection method and system for transformer tank production based on the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0021] In existing technologies, transformer tank leak detection commonly employs the negative pressure wave attenuation method, which determines leaks by monitoring the pressure attenuation rate. However, this method has significant drawbacks: ambient temperature fluctuations cause gas expansion and contraction, leading to confusion between pressure changes and actual leak signals, resulting in misjudgments. For example, gas expansion during temperature increases may mask minor leaks, while pressure decreases during temperature drops can be misinterpreted as leaks. Furthermore, acoustic signal processing relies on fixed energy thresholds to identify ultrasonic events, which cannot adapt to varying background noise in industrial environments, resulting in low signal-to-noise ratios and high false detection rates. When introducing acoustic auxiliary features, existing technologies use a fixed-weight fusion strategy, assigning high weights to acoustic features even during sudden noise increases, which in turn introduces interference.

[0022] To address the aforementioned issues, a leak detection method capable of removing temperature interference and dynamically adjusting the weights of multiple features needs to be designed. First, to address the temperature interference problem, a physical model incorporating temperature compensation needs to be established. Theoretical pressure changes are calculated using the ideal gas law to separate the actual leak signal. Second, to address the low signal-to-noise ratio of the acoustic signal, a dynamic background noise assessment mechanism needs to be introduced, adjusting the acoustic event judgment threshold based on real-time noise levels. Finally, a dynamic fusion model of pressure and acoustic features needs to be constructed, adjusting the weight allocation in real-time based on signal quality to improve decision reliability.

[0023] This application proposes a leakage detection method and system for transformer tank production, which has the advantages of improving leakage detection accuracy, effectively eliminating environmental temperature interference, and dynamically adapting to changes in background noise, thereby effectively improving detection precision.

[0024] The following description, in conjunction with the accompanying drawings, details a specific scheme for a leakage detection method for transformer tank production provided by the present invention.

[0025] Please see Figure 1 The diagram illustrates a flow chart of the leakage detection method for transformer tank production provided by the present invention.

[0026] like Figure 1 As shown, the leakage detection method for transformer tank production includes the following steps:

[0027] S10. Divide the fuel tank under test into multiple test areas.

[0028] The test area division refers to dividing the test object into several independent test units according to the structural characteristics of the oil tank. This can be achieved by using a grid division algorithm based on a CAD model. Each area is equipped with an independent sensor group to facilitate the location of the leak.

[0029] S20. Determine the ideal net pressure decay rate by using the average rate of pressure drop in the tank cavity, the rate of pressure change corresponding to temperature change, and the volume change in the target test area within the target time period.

[0030] Among them, the ideal net pressure decay rate refers to the theoretical pressure decay rate after excluding the influence of temperature and volume changes. It can be obtained by calculating the initial pressure, temperature change and volume change through the ideal gas law, and is used to eliminate the interference of environmental factors on the pressure signal.

[0031] S30. Determine the pressure decay deviation using the ideal net pressure decay rate and the measured net pressure decay rate.

[0032] Among them, the pressure attenuation deviation refers to the degree of deviation between the measured net pressure attenuation rate and the ideal value. It is calculated by standard deviation normalization and reflects the degree of pressure signal abnormality.

[0033] S40. Determine the acoustic activity weights by using the acoustic activity degree of the acoustic signal in the acoustic event and the average energy of the background noise in multiple test areas in the acoustic event.

[0034] Among them, the acoustic activity weight refers to the weight value determined by the ratio of the effective energy of the acoustic event to the background noise. Specifically, it can be determined by calculating the functional relationship between the acoustic activity degree and the average energy of the background noise, and is used to evaluate the reliability of the acoustic signal.

[0035] Specifically, in response to the presence of an acoustic event, it is necessary to calculate the acoustic activity weight, that is, to obtain the measured average energy of the acoustic event and the average energy of the background noise, and then to determine the acoustic activity weight by using the ratio of the measured average energy of the acoustic event to the average energy of the background noise.

[0036] S50. Using acoustic activity weights and pressure attenuation deviation, determine the regional leakage comprehensive index, and then use the regional leakage comprehensive index to determine the leakage detection status of the oil tank.

[0037] Among them, the regional leakage comprehensive index is a quantitative indicator that integrates pressure deviation and acoustic weight, and is generated using a dynamic weighting algorithm, serving as the core basis for leakage judgment.

[0038] Specifically, during the pressure holding test, pressure, temperature, and acoustic signals from each test area are simultaneously collected. A temperature compensation model is constructed based on the ideal gas law to calculate the theoretical pressure change curve, and the net pressure attenuation rate is obtained by comparing it with the measured pressure curve. For the acoustic signals, an adaptive background noise detection window is set, and the mean and standard deviation of noise energy are calculated in real time, dynamically adjusting the event judgment threshold. The pressure attenuation deviation is obtained by calculating the standardized difference between the measured value and the ideal value, reflecting the degree of pressure anomaly. The acoustic activity weight is dynamically adjusted based on the ratio of event energy to background noise, automatically reducing the acoustic feature weight when noise suddenly increases. Finally, the pressure deviation and acoustic weight are proportionally integrated to generate a comprehensive regional leakage index. Leakage is determined when the index exceeds a set threshold; for example, if the pressure deviation is greater than or equal to the acoustic weight, the final comprehensive regional leakage index ranges from 0 to 1.

[0039] Compared to existing technologies, traditional methods use a single pressure attenuation rate index and neglect temperature compensation. This solution effectively eliminates thermodynamic interference by constructing an ideal pressure model, improving the realism of pressure signal representation. Existing acoustic detection methods use fixed thresholds, leading to high false positive rates. This solution introduces a dynamic background noise evaluation mechanism to improve the accuracy of acoustic event recognition. Existing multi-feature fusion methods use static weight allocation. This solution adjusts weights in real time based on signal quality, automatically reducing the decision weights of unreliable features in the event of noise interference, thus enhancing the system's anti-interference capability.

[0040] Through the above technical solutions, this application effectively solves the problem of temperature fluctuations interfering with pressure signals, and achieves accurate extraction of real leakage signals through an ideal pressure model. Dynamically adjusted acoustic activity weights significantly improve the signal-to-noise ratio of acoustic event detection, avoiding false alarms caused by fixed thresholds. The dynamic fusion mechanism of pressure and acoustic features can optimize the decision model according to changes in operating conditions, maintaining high detection accuracy in complex industrial environments and reducing false alarm and false negative rates.

[0041] In some embodiments, during the intelligent partitioning and sealing process, the system first calls a high-precision CAD model of the tank to be tested. This model must include the geometric dimensions, interface locations, and sealing surface distribution of each cavity in the tank. Based on the model information, the system automatically uses a grid partitioning algorithm to plan N independent test areas. The number of partitions N needs to be dynamically adjusted according to the tank volume and structure, with the volume of a single area not exceeding 5L. Subsequently, the system drives a distributed negative pressure fixture to perform synchronous sealing actions, and each test area is equipped with an independent negative pressure channel, a high-precision pressure sensor, and an ultrasonic microphone.

[0042] In some embodiments, a synchronous triggering and acquisition phase may also be included; specifically, the detection system uses a high-precision clock module as a unified time reference to synchronously trigger the data acquisition channels of N test areas; during the vacuuming phase (vacuuming time set to 5-8s) and the pressure holding phase (pressure holding time set to 10-15s), three types of core data are acquired in parallel, including the real-time pressure sequence of each area. Acoustic signal sequences for each region were acquired at 100ms intervals. During the data acquisition process, a 30-50kHz bandpass filter was used in real time to filter low-frequency interference signals; ambient temperature and fuel tank temperature sequences were also analyzed. Temperature data is collected by platinum resistance temperature sensors distributed on the outer wall of the oil tank and in the test environment, with the sampling interval consistent with the pressure sequence; where i represents any test area; and t represents any sampling time point.

[0043] The collected raw pressure data was processed using a wavelet denoising algorithm, selecting the db4 wavelet basis function and setting the decomposition level to 3 to eliminate minor pressure fluctuations caused by environmental vibrations. A Gaussian filtering algorithm was used for the acoustic signal to reduce environmental acoustic interference. After preprocessing, all data were tagged with a unified timestamp (timestamp accuracy consistent with the system clock reference) and region labels (such as "Region 1 - Top Cavity" and "Region 2 - Bottom Interface"). Finally, the data was integrated to form a spatiotemporally correlated multimodal data package containing pressure, acoustic, and temperature data, which is convenient for subsequent steps to call and parse.

[0044] In some embodiments, the ideal net pressure decay rate is determined by utilizing the average rate of pressure drop in the tank cavity of the target test area within a target time period, the rate of pressure change corresponding to temperature change, and the volume change, including the following operations:

[0045] Obtain the initial pressure and initial ambient temperature of each test area at the start of the pressure holding process, and obtain the ending pressure and ending ambient temperature at the end of the pressure holding process.

[0046] By using the initial pressure, the final pressure, the start time of pressure holding, and the end time of pressure holding, the average rate of pressure drop in the tank cavity of each test area within the target time period is determined.

[0047] The ideal pressure value is determined using the initial pressure, initial ambient temperature, and change in ambient temperature.

[0048] The ideal net pressure decay rate is determined by using the ideal pressure value, the start time of pressure holding, the end time of pressure holding, and the volume change of the oil tank cavity during the pressure holding period.

[0049] The initial pressure refers to the gas pressure value in the test area at the beginning of the pressure holding phase. This pressure can be acquired in real-time using a high-precision pressure sensor and is used to establish a benchmark for pressure changes. The final pressure is the gas pressure value in the test area at the end of the pressure holding phase. The initial ambient temperature refers to the ambient temperature at the beginning of the pressure holding phase, which can be measured using a platinum resistance temperature sensor and serves as an input parameter for thermodynamic compensation. The average rate of pressure drop refers to the amount of pressure change per unit time, calculated by dividing the pressure difference between the beginning and end of the pressure holding phase by the time difference, and is used to characterize the overall trend of pressure change. The ideal pressure value is the theoretical pressure value caused solely by temperature changes under the assumption of no leakage. It is derived from the ideal gas law using the initial pressure and the change in ambient temperature, and is used to eliminate the interference of thermodynamic effects on pressure decay. The volume change refers to the volume change of the tank cavity caused by thermal expansion or contraction due to temperature changes, calculated using the initial volume and the material's coefficient of thermal expansion, and is used to correct the volume parameters in the ideal gas law.

[0050] Specifically, at the beginning of the pressure holding phase, the initial pressure and ambient temperature of each test area are recorded using pressure and temperature sensors. After the pressure holding phase ends, the pressure and temperature data at the end are collected again. Based on the difference between the initial and final pressures, and combined with the duration of the pressure holding phase, the average rate of pressure drop is calculated. Simultaneously, based on the ideal gas law, the initial pressure and temperature changes are substituted into the formula to derive the theoretical ideal pressure value under leak-free conditions. Further considering the minute volume changes in the tank cavity caused by temperature variations, the volume change is calculated using the initial volume and the coefficient of thermal expansion to correct the ideal pressure value. Finally, combining the corrected ideal pressure value, the average rate of pressure drop, and time parameters, the ideal net pressure decay rate, excluding temperature interference, is calculated.

[0051] Through the above technical solution, this application can accurately eliminate the interference of ambient temperature fluctuations and material thermal deformation on pressure decay measurement, effectively improving the accuracy of leak detection. For example, under high-temperature conditions, the system automatically compensates for the pressure rise caused by gas expansion, avoiding misjudging normal thermodynamic changes as leaks; under low-temperature conditions, it corrects the effect of volume contraction on pressure decay, preventing the missed detection of minor leaks. This method achieves refined modeling of pressure decay behavior through multi-parameter collaborative calculation, providing a reliable data foundation for subsequent leak determination.

[0052] In some embodiments, the ideal net pressure decay rate is determined using the ideal pressure value, the start time of pressure holding, the end time of pressure holding, and the volume change of the tank cavity during the pressure holding period, including the following operations:

[0053] Obtain the time difference between the start and end of the pressure holding period, and use the ratio between the time difference and the ideal pressure value to determine the pressure change rate corresponding to the temperature change.

[0054] Obtain the initial volume of the oil tank cavity, use the initial volume to determine the volume change, and use the volume change to determine the corrected ideal pressure value.

[0055] The ideal net pressure decay rate is determined by using the corrected ideal pressure value, the average rate of pressure drop, and the rate of pressure change.

[0056] The time difference refers to the time interval between the start and end of the pressure holding period. This is achieved by synchronously acquiring timestamps and calculating the difference using a high-precision clock module, quantifying the duration of pressure changes. The volume change refers to the volume change of the tank cavity due to temperature variations during the pressure holding process. This is calculated by multiplying the initial volume by the material's thermal expansion coefficient and the temperature change, used to correct the ideal pressure value and eliminate the influence of thermal deformation on the pressure calculation. The corrected ideal pressure value is the ideal pressure value recalculated after considering the volume change. This is obtained by adjusting the ratio of the initial pressure value to the volume change, more accurately reflecting the theoretical pressure change trend under leak-free conditions.

[0057] Specifically, during the pressure holding phase, timestamps are recorded synchronously and time differences are calculated. The pressure change rate caused by temperature changes is derived using the ideal gas law. The volume change of the tank cavity due to thermal expansion or contraction is calculated using the initial volume and temperature change, thus correcting the ideal pressure value. The corrected ideal pressure value is compared with the measured pressure drop rate to eliminate the dual interference of temperature and volume changes, ultimately calculating the ideal net pressure decay rate caused only by leakage. For example, when the thermal expansion coefficient of the tank material is known, the volume change can be accurately calculated as the initial volume multiplied by the temperature change and the material coefficient, ensuring that the corrected ideal pressure value accurately reflects the pressure change under leak-free conditions.

[0058] For example, pressure signal analysis typically only calculates the macroscopic attenuation rate, but this value is easily affected by thermodynamic effects caused by fluctuations in ambient temperature, and cannot isolate the actual pressure loss caused by leakage, leading to misjudgment. At the same time, acoustic signal processing generally uses a fixed energy threshold to identify ultrasonic leakage events, which cannot adapt to changing ambient background noise. In noisy workshop environments, it is easy to misjudge noise as leakage or miss the real signal, resulting in a low signal-to-noise ratio.

[0059] Specifically, in negative pressure testing, the core observation is the change in pressure over time. Ideally, a leak would cause a pressure drop. However, temperature changes cause gases to expand and contract, resulting in significant pressure changes. When the temperature rises, the gas expands, causing the pressure inside the sealed cavity to rise. This may mask a pressure drop caused by a tiny leak. When the temperature drops, the gas contracts, causing the pressure inside the sealed cavity to drop. This may amplify or disguise a leak. Therefore, the directly observed pressure decay is a mixed signal, containing both the actual leak signal and thermodynamic noise. If this is not distinguished, it can lead to misjudgment.

[0060] Select the start and end times of the pressure holding process, and obtain the time difference between them; obtain the measured pressure values ​​at the corresponding start and end times; obtain the difference between the measured pressure values ​​at the start and end times, and calculate the ratio of this difference to the time difference, denoted as [missing value]. This value represents the average rate of pressure drop in the measured cavity during the detection period. The value can be positive or negative. When the value is positive, it indicates that the pressure is dropping continuously and rapidly. When the value is negative, the pressure is rising instead of falling, which is usually not a sign of a traditional leak.

[0061] Next, according to the ideal gas law (PV = nRT), with constant volume V and number of moles n, pressure P is directly proportional to thermodynamic temperature T. Assuming the gas in the tank is approximately an ideal gas, since volume V, number of moles n, and constant R are all constant, the equation can be simplified to... ,in Since is a constant, this equation indicates that, under leak-free conditions, the pressure of a gas is proportional to its thermodynamic temperature.

[0062] Based on the above operations, the initial pressure and initial ambient temperature were obtained at the start of the pressure holding process. At the end of the pressure holding period, the ambient temperature changed. Therefore, the current temperature is .

[0063] Based on the aforementioned relationship between pressure and temperature, under leak-free conditions, the relationship between initial pressure and initial ambient temperature is explained. The ratio should be equal to the pressure at the end time and the temperature at the end time. The ratio; by transforming the equation, the theoretical ideal pressure value at the end time can be obtained. :

[0064] ;

[0065] in, This is the initial pressure.

[0066] Therefore, the pressure change caused purely by temperature change It is the difference between the theoretical ideal pressure and the initial pressure; that is... This value represents the pressure change caused by temperature variation.

[0067] This yields the ratio between the pressure change and the time difference between the start and end times of the pressure holding period, thus providing the rate of pressure change caused by temperature variation. ;

[0068] This allows us to obtain the average rate of pressure drop in the measured cavity. Rate of pressure change caused by temperature change The difference between them is denoted as the net pressure attenuation rate. It represents the true rate of pressure drop in the test area due to leakage after excluding the interference of ambient temperature; the value includes positive and negative values, where a positive value indicates that the rate of pressure drop exceeds the extent that temperature contraction can explain, and the probability of leakage is greater; this feature is used as the first feature.

[0069] Furthermore, since pressure decay may be accompanied by slight fluctuations when leakage occurs, the reciprocal of the standard deviation of the pressure sequence during the pressure holding phase is calculated and denoted as... , which represents the static stability of the sequence; the larger the value, the closer the current closed environment is to the typical characteristics of an ideal sealed state; conversely, the smaller the value, the greater the probability of intermittent leakage or minor unstable leakage; this feature is used as the second feature.

[0070] In addition, the pressure decay threshold on which the leak is determined is usually based on historical data or fixed empirical values ​​set by standards. This ignores the structural differences between individual oil tanks under test (such as small changes in volume) and the instantaneous fluctuations of the initial test conditions (such as pressure and temperature). Using static thresholds to measure these dynamic variables uniformly leads to low accuracy in judging micro-leakage in critical conditions and a high risk of false alarms.

[0071] First, the detection system calls up the digital twin model of the fuel tank to be inspected. This model is an extension of the fuel tank's 3D CAD model, and includes geometric dimensions (such as the volume of each area). In addition to wall thickness and sealing surface dimensions, material property parameters also need to be integrated, including the elastic modulus, Poisson's ratio, and coefficient of thermal expansion of the fuel tank shell material. .

[0072] During the model parameter input phase, the detection system automatically extracts initial state parameters from the aforementioned generated multimodal data package, including the initial pressure and initial internal temperature of each test area. Simultaneously, the system collects atmospheric pressure and relative humidity of the test environment through environmental sensors and uses these as boundary conditions for the model. Based on these parameters, the system constructs a physical model centered on the ideal gas law (PV=nRT) and the first law of thermodynamics. The model assumes that each test area of ​​the fuel tank is an independent, sealed cavity, and that the mass of the gas inside the cavity remains constant under leak-free conditions, and that the gas is an ideal gas. It also considers the influence of the thermoelastic deformation of the fuel tank shell material on the cavity volume, calculating the volume change caused by temperature changes using the coefficient of thermal expansion. :

[0073] ,

[0074] And incorporate it into the correction terms of the ideal gas law.

[0075] The predicted calculation time range is consistent with the pressure holding stage in step one, during which the temperature change... Extract directly from the temperature sequence acquired in step one; correct for ideal pressure values. The formula, calculated using the modified ideal gas law, is as follows:

[0076] .

[0077] in, As the initial pressure, The initial internal temperature is not 0; The change in temperature This indicates minute changes in fuel tank volume due to temperature. For the volume of the test area, if much smaller Then the volume correction term It can be approximated as 1.

[0078] Through the above calculations, the system can generate the ideal pressure values ​​for each test area i at all time points during the pressure holding phase, forming a complete ideal pressure sequence. .

[0079] Ideal net pressure decay rate The calculation method and step two Completely identical, except the analysis object is replaced with an ideal pressure sequence. It should be noted that in the ideal state of "absolute zero leakage," the gas mass is constant, and pressure changes are caused only by changes in temperature and volume. The theoretical value should be close to 0.

[0080] Through the above technical solution, this application can more accurately separate the combined effects of temperature fluctuations and tank deformation on pressure decay, accurately identify the pressure change characteristics caused by actual leakage, avoid calculation errors caused by ignoring volume changes, and improve the reliability of leakage detection.

[0081] In some embodiments, the pressure decay deviation is determined using the ideal net pressure decay rate and the measured net pressure decay rate, including the following operations:

[0082] Obtain the average rate of pressure drop in the tank cavity for each test area within the target time period, and obtain the standard deviation of the ratio corresponding to the average rate of pressure drop in all historical test processes.

[0083] The pressure decay deviation is determined by using the standard deviation, the ideal net pressure decay rate, and the measured net pressure decay rate.

[0084] The standard deviation of the ratio corresponding to the average pressure drop rate refers to the degree of fluctuation in the pressure drop rate of the same specification oil tank under the same test conditions in historical test data. Specifically, it can be achieved by statistically analyzing the historical dataset using variance calculation methods. This standard deviation reflects the inherent differences in pressure decay rate under different test environments and the uncertainty of the measurement system. The pressure decay deviation is a quantitative indicator of the degree of deviation between the measured net pressure decay rate and the ideal net pressure decay rate relative to the historical fluctuation range. Specifically, it can be calculated by dividing the absolute difference between the measured value and the ideal value by the historical standard deviation. This indicator can eliminate the influence of differences in different test batches or environments on the pressure decay assessment.

[0085] Specifically, after the pressure holding phase, the average pressure drop rate data of all valid test records of the same model of fuel tank in historical testing is first retrieved from the database. The standard deviation of these data ratios is used to obtain a parameter characterizing the system uncertainty. Then, the difference between the measured net pressure decay rate and the ideal net pressure decay rate in the current test area is calculated, and this difference is divided by the historical standard deviation to obtain the dimensionless pressure decay deviation. By introducing the historical standard deviation as a normalization factor, measurement errors caused by individual differences in fuel tanks or fluctuations in test conditions can be effectively eliminated, making the deviation calculation results comparable across test batches.

[0086] Calculate the pressure attenuation deviation using the formula. :

[0087] ,

[0088] in, This is a dimensionless parameter; the larger its value, the more significant the deviation between the measured pressure decay behavior and the ideal sealing state. If the pressure decay deviation is within the allowable range of measurement uncertainty, it can be considered that the pressure behavior conforms to the ideal state.

[0089] Through the above technical solution, this application can effectively distinguish between pressure decay caused by actual leakage and pressure changes caused by inherent fluctuations in the test environment, solve the problem of misjudgment caused by ignoring the volatility of historical data in traditional methods, and improve the consistency and reliability of leakage detection results.

[0090] In some embodiments, acoustic activity weights are determined using the acoustic activity of acoustic signals in an acoustic event and the average energy of background noise in multiple test areas of the acoustic event, including the following operations:

[0091] The energy of the acoustic signal collected by the background noise sampling window of the target size is obtained, and the average energy of the background noise is determined by the sum of squares of the signals corresponding to each frame in the background noise sampling window and the mean of the sum of squares of all frames in the background noise sampling window.

[0092] Obtain acoustic activity.

[0093] The acoustic activity weights are determined using acoustic activity degree and average background noise energy.

[0094] The average background noise energy refers to the energy statistics of acoustic signals obtained through a sampling window of fixed duration. Specifically, a sliding window mechanism can be used to acquire acoustic signals in real time, calculate the sum of squares of each frame within the window, and then take the mean. This parameter quantifies the baseline level of environmental noise, providing a dynamic reference threshold for the effective identification of subsequent acoustic events. Acoustic activity reflects the degree of fluctuation in the pressure curve. It is obtained by calculating and normalizing the smoothness deviation between the ideal and measured pressure curves. This parameter characterizes the degree of abnormality in pressure changes. When irregular fluctuations occur in the pressure curve, the acoustic activity increases, indicating a possible leak-related acoustic event.

[0095] Specifically, within the background noise sampling window, the acoustic signal is divided into multiple frames for energy calculation; for example, the duration of each frame can be set to 10 milliseconds. The energy of each frame is calculated by the sum of the squares of the signal amplitudes of all sampling points within that frame, and the average energy of all frames within the window is the average background noise energy. Acoustic activity is obtained by comparing the smoothness of the pressure curve under ideal sealing conditions with the smoothness of the measured pressure curve, for example, by normalizing the difference in their standard deviations. Finally, the acoustic activity weight is dynamically adjusted by the ratio of acoustic activity to the average background noise energy. When the background noise is high, the acoustic activity weight is automatically reduced to decrease noise interference; when the acoustic activity significantly increases, the weight is increased accordingly to enhance the detection sensitivity of abnormal signals.

[0096] Through the above technical solution, this application can optimize the acoustic signal processing strategy according to the real-time environmental noise level, reduce the false alarm rate in noisy working conditions, and improve the ability to capture small leakage signals in stable environments, thus solving the technical defect of poor adaptability of traditional fixed threshold methods in complex industrial scenarios.

[0097] In some embodiments, obtaining acoustic activity includes the following operations:

[0098] Obtain the smoothness of the ideal pressure curve and the smoothness of the measured pressure curve.

[0099] The smoothness deviation of the curve is determined by using the smoothness of the ideal pressure curve and the smoothness of the measured pressure curve.

[0100] The acoustic activity is obtained by normalizing the curve smoothness deviation.

[0101] Among them, the smoothness of the ideal pressure curve refers to the stability index of the pressure sequence under leak-free conditions, which can be calculated using the reciprocal of the standard deviation of the ideal pressure sequence. A larger value indicates more stable pressure changes. The smoothness of the measured pressure curve refers to the stability index of the pressure sequence during actual testing, which can also be calculated using the reciprocal of the standard deviation of the measured pressure sequence. This value reflects the severity of pressure fluctuations. The curve smoothness deviation refers to the degree of difference between the stability of the ideal and measured pressure curves, which can be calculated using the ratio of the absolute difference between the two to the ideal value. This parameter quantifies the degree of pressure fluctuation anomalies. Normalization involves mapping the curve smoothness deviation to the 0-1 range, specifically using the maximum historical deviation value as the normalization benchmark. This operation makes the acoustic activity comparable.

[0102] Specifically, during the pressure holding phase, ideal pressure sequences and measured pressure sequences are simultaneously acquired, and the reciprocal of their standard deviations is calculated as a smoothness index. By calculating the absolute difference between the two and dividing it by the ideal smoothness value, a deviation parameter characterizing the degree of pressure fluctuation anomaly is obtained. The current deviation is further normalized using the historical maximum deviation value, mapping the result to the 0-1 interval to form an acoustic activity parameter. This parameter effectively reflects the degree of pressure fluctuation anomaly; when the measured pressure curve shows intermittent fluctuations or slight oscillations, the acoustic activity will increase significantly, indicating a possible leak.

[0103] Through the above technical solution, this application effectively solves the problem of insufficient accuracy in identifying dynamic features of pressure curves. By quantifying the stability differences of pressure sequences, it significantly improves the accuracy of acoustic activity parameters. This technical solution can accurately distinguish between normal temperature fluctuations and pressure oscillations caused by leakage. Even under conditions with intermittent minor leaks, it can maintain high detection sensitivity and reduce the risk of misjudgment caused by changes in the smoothness of the pressure curve.

[0104] In some embodiments, the following operations are also included:

[0105] Multiple signal segments identified as acoustic events are acquired, and the energy integral of the acoustic events during the holding period is obtained to determine the average energy of the measured acoustic events corresponding to each signal segment.

[0106] The acoustic activity weights are determined by using the measured average energy of acoustic events, acoustic activity, and average energy of background noise.

[0107] Acoustic events are defined as signal segments exceeding a background noise threshold. They can be identified using a dynamic thresholding method, which filters out signal segments with significant energy characteristics by setting a threshold related to the background noise energy. Energy integral refers to the cumulative energy of an acoustic event over time. This can be calculated using a numerical integration algorithm to calculate the sum of squares at each sampling point within the signal segment, quantifying the total energy of the acoustic event. The measured average energy of acoustic events is the average of the energy integrals of all identified acoustic events, achieved by summing and dividing by the total number of events, characterizing the overall intensity of the acoustic events. Acoustic activity refers to the degree of deviation of the frequency of acoustic events from the ideal state. This can be achieved by normalizing the measured event frequency to a range of 0 to 1, quantifying the abnormal level of acoustic activity. The average background noise energy is the energy benchmark of environmental noise under leak-free conditions, obtained by calculating the average signal energy within a preset time window, used to establish a noise energy reference value. Acoustic activity weight refers to the contribution of acoustic features to leakage determination. Specifically, it can be dynamically adjusted by combining the ratio of acoustic event energy to background noise energy and acoustic activity, and is used to adaptively assess feature reliability under different noise environments.

[0108] Specifically, during the pressure holding period, acoustic signals are monitored in real time, and a sliding window method is used to detect event segments exceeding the background noise threshold. Each event segment can last for 3 seconds. For each identified acoustic event, its energy integral is calculated using a numerical integration method, such as summing the squares of the signal amplitude over the time domain to obtain the total energy of the event. The sum of the total energy of all events is divided by the number of events to obtain the measured average energy of the acoustic events. Simultaneously, the frequency of the currently detected events is normalized by analyzing the frequency of the maximum acoustic events under known leakage conditions in historical data to obtain the acoustic activity level. The average energy of the background noise is determined by collecting environmental noise signals during the silent phase before detection begins and calculating their average energy. Finally, the ratio of the measured average energy of acoustic events to the average energy of the background noise is multiplied by the acoustic activity level to obtain the acoustic activity weight. When a sudden increase in background noise causes a decrease in the signal-to-noise ratio, this weight is automatically reduced to decrease the influence of acoustic characteristics on leakage determination.

[0109] For example, the negative pressure method measures the total loss of gas medium over a period of time, but its location capability is poor and it cannot distinguish the type of leak. Analysis shows that high-speed airflow generates strong turbulence and vortices at the leak point. This turbulence causes rapid oscillations in local pressure, which propagate outwards in the form of ultrasonic waves.

[0110] Therefore, a background noise sampling window of size 3 seconds was set; the acquired acoustic signals were... The energy is calculated in short frames (e.g., every 10ms). Specifically, the sum of squares of the signals at the corresponding sampling points in each frame within the calculation window is calculated. Then, the mean of the sum of squares of all frames within the window is obtained, thus yielding the mean background noise energy. .

[0111] Simultaneously, the standard deviation of the background noise energy within the window is obtained; based on In principle, a threshold is set as follows: ;in, The standard deviation of the background noise energy is denoted as .

[0112] Within the detection time Δt, count the number of signal pulses that exceed this threshold, and record it as . The larger the value, the more likely it is that there are continuous ultrasonic pulses exceeding the background noise in the region, which usually corresponds to a continuous, stable leak; the smaller the value, the more likely it is that the leak is intermittent or very weak, with only occasional signals that can exceed the threshold, which may correspond to a loose flange occasionally venting under pressure, or a very fine crack; this characteristic is the third characteristic.

[0113] Finally, for each signal segment identified as an acoustic event, its energy integral over the entire segment's time is calculated, i.e. The average energy is obtained by summing the energy values ​​of each counted event and then dividing by the total number of events. A higher value indicates that each leak event releases a strong ultrasonic energy. This usually means a high leakage velocity, a large leak orifice diameter, or a large pressure difference between the inside and outside of the leak point; this characteristic serves as the fourth characteristic.

[0114] Ideal pressure curve smoothness The calculations are also based on an ideal pressure sequence. The core monitoring section is the target, and the standard deviation is calculated according to the method in step two. It should be noted that, under ideal conditions, there are no pressure fluctuations due to leakage, and the temperature change is a continuous and stable process. The values ​​will be significantly higher than the actual measurements. .

[0115] Through the above technical solution, this application solves the problem of misjudgment of acoustic features caused by background noise fluctuations. By dynamically calculating the acoustic activity weight, it can accurately distinguish between real leakage signals and random noise in noisy industrial environments, thereby improving the reliability and environmental adaptability of leakage detection.

[0116] In some embodiments, an area leakage comprehensive index is determined using acoustic activity weighting and pressure attenuation deviation, and then the leakage detection status of the fuel tank is determined using the area leakage comprehensive index, including the following operations:

[0117] Get the preset minimum value.

[0118] The pressure deviation weight is determined by using the pressure attenuation deviation.

[0119] The comprehensive leakage index of the region is determined by using pressure attenuation deviation, pressure deviation weight, acoustic activity weight, and a preset minimum value.

[0120] The leakage detection status of the fuel tank is determined by using the regional leakage comprehensive index. If the regional leakage comprehensive index of the target test area is greater than a first threshold, it is determined that there is a leakage in the target test area where the regional leakage comprehensive index is greater than the first threshold; or, if at least one of the multiple regional leakage comprehensive indices of the target test area is less than the first threshold, it is determined that there is interference in the target test area.

[0121] The preset minimum value refers to a positive number approaching zero, specifically a value on the order of 1×10^-6, used to prevent mathematical errors caused by a zero denominator during weighted calculations. The pressure deviation weight is a coefficient reflecting the importance of pressure attenuation deviation in the comprehensive judgment, specifically achieved by directly assigning a value to the pressure attenuation deviation; a larger weight indicates a higher contribution of pressure characteristics to leakage judgment. The regional leakage comprehensive index is a quantitative indicator integrating pressure and acoustic characteristics, specifically achieved by dividing the weighted sum of the pressure deviation weight and acoustic activity weight by the sum of the total weight and the minimum value. This index comprehensively reflects the synergistic effect of pressure changes and acoustic signals. The first threshold is a pre-set critical value for judging leakage, specifically determined through historical data statistics or experimental calibration, used to distinguish between normal fluctuations and actual leaks. Interference refers to signal anomalies caused by non-leakage factors, specifically identified through repeated testing combined with background noise analysis to eliminate misjudgments caused by accidental factors.

[0122] Specifically, when determining the comprehensive leakage index for a region, the pressure characteristic contribution value is first obtained by multiplying the pressure attenuation deviation by the pressure deviation weight, and the acoustic characteristic contribution value is obtained by multiplying the acoustic activity weight by the acoustic activity level. Then, the pressure deviation weight and the acoustic activity weight are added together and a preset minimum value is added as the denominator, and the sum of the pressure characteristic contribution value and the acoustic characteristic contribution value is used as the numerator. The comprehensive index is obtained through ratio calculation. When this index exceeds a preset threshold, a leak is determined to exist in the corresponding region. For example, when the pressure attenuation deviation significantly deviates from the ideal value and the acoustic event energy is high, the comprehensive index will rapidly rise above the determination threshold, triggering a leak alarm.

[0123] After the pressure holding phase, the system calculates a comprehensive leakage index for each test area. When the index for a certain area exceeds a preset first threshold, a leakage alarm is triggered, and a repeat verification process is initiated. If the index exceeds the threshold in two consecutive tests, a leak is confirmed. If some indices for the same area fall below the threshold in multiple tests, the system automatically identifies this as environmental interference or random noise, terminates the leakage determination, and prompts a check of the test conditions. This determination logic, combining a dynamic threshold and a repeat verification mechanism, effectively balances detection sensitivity and anti-interference capability.

[0124] Through the above technical solution, this application effectively solves the problem of misjudgment caused by unreasonable weight allocation during multimodal feature fusion, thereby improving the accuracy of leak detection in complex industrial environments. Especially under conditions of drastic temperature fluctuations or sudden changes in background noise, dynamically adjusting feature weights avoids the impact of single feature anomalies on the overall judgment result, significantly improving the environmental adaptability and reliability of the detection system. Furthermore, this application can accurately distinguish between real leaks and transient interference in complex industrial environments, reducing the risk of misjudgment caused by environmental noise or equipment fluctuations, and improving the reliability and repeatability of detection results. Simultaneously, by dynamically adjusting the judgment logic, it avoids incorrectly marking areas with intermittent interference, ensuring the objectivity and accuracy of the detection conclusions.

[0125] In some embodiments, the comprehensive index of regional leakage is determined using pressure attenuation deviation, pressure deviation weight, acoustic activity weight, and a preset minimum value, including the following operations:

[0126] The first value is determined by using the pressure attenuation deviation and the pressure deviation weight.

[0127] The second value is determined using acoustic activity weights and acoustic activity degrees.

[0128] The third value is determined by using pressure deviation weight, acoustic activity weight, and preset minimum value.

[0129] The comprehensive regional leakage index is determined using the first, second, and third values.

[0130] The pressure attenuation deviation refers to the difference between the measured net pressure attenuation rate and the ideal net pressure attenuation rate. Specifically, it can be calculated by dividing the absolute difference between the two by the historical standard deviation. This parameter quantifies the deviation of pressure changes from the ideal state. The pressure deviation weight is the dynamic weighting coefficient of the pressure attenuation deviation in the comprehensive index, directly using the value of the pressure attenuation deviation. This parameter reflects the contribution of pressure characteristics to leak detection. The acoustic activity weight is the dynamic weighting coefficient of acoustic events in the comprehensive index, specifically achieved by multiplying acoustic activity by the acoustic event's energy signal-to-noise ratio. This parameter characterizes the reliability of acoustic characteristics. The preset minimum value is a minimal constant used to prevent the denominator from being zero; specifically, it can be a value on the order of 1×10^-6. This parameter ensures the numerical stability of the calculation process.

[0131] Specifically, after the pressure holding phase, the pressure attenuation deviation is first calculated using data collected from pressure and temperature sensors. This parameter reflects the degree of deviation between the measured pressure change and the ideal model. Simultaneously, acoustic activity weights are calculated based on acoustic activity level and background noise energy obtained from acoustic signal analysis. This parameter reflects the validity of the ultrasonic event. The first value, representing the decision contribution of pressure characteristics, is obtained by multiplying the pressure attenuation deviation by its own weight. The second value, representing the decision contribution of acoustic characteristics, is obtained by multiplying the acoustic activity weight by the acoustic activity level. The third value, representing the minimum sum of the pressure deviation weight and the acoustic activity weight, forms a dynamic normalized benchmark. Finally, the sum of the first and second values ​​is divided by the third value to obtain the regional leakage comprehensive index. This index integrates multimodal features through a dynamic weighting mechanism to avoid misjudgment based on a single feature.

[0132] In some specific implementations, the standard deviation of historical test data can be used as an uncertainty factor when calculating the pressure attenuation deviation. For example, the standard deviation of the pressure attenuation rate in 100 historical tests of the same type of fuel tank can be selected. Energy signal-to-noise ratio correction can be introduced when calculating the acoustic activity weights. When the average energy of an acoustic event is lower than the background noise, the signal-to-noise ratio is forcibly set to 1 to avoid weight distortion. The setting of the preset minimum value must meet numerical stability requirements; for example, it should be on the order of 1e-6 in floating-point operations.

[0133] For example, in existing transformer tank leakage detection technology, when auxiliary features such as acoustics are introduced, it is impossible to adaptively adjust according to the real-time changes in the quality of different feature signals such as pressure and acoustics in each detection. For example, under the condition of sudden increase in background noise, the reliability of acoustic features will be significantly reduced, but the system still assigns them a fixed weight, which will introduce interference and reduce the accuracy and reliability of decision-making.

[0134] First, based on historical testing records, the average rate of pressure drop in the tested cavity is obtained for transformer tanks of the same specifications during the testing process. The standard deviation of the ratio of the average rate R to the total number of historical detections across all historical detection processes is denoted as . The larger this value, the greater the uncertainty of the overall pressure decay caused by various subjective and objective factors during the leakage detection process of the transformer tank of the current specification.

[0135] Calculate the pressure attenuation deviation using the formula. :

[0136] ,

[0137] in, This is a dimensionless parameter; the larger its value, the more significant the deviation between the measured pressure decay behavior and the ideal sealing state. If the pressure decay deviation is within the allowable range of measurement uncertainty, it can be considered that the pressure behavior conforms to the ideal state.

[0138] Next, curve smoothness deviation. Calculate using the formula:

[0139] ,

[0140] in, Also a dimensionless parameter, its value ranges from 0 to 1; the smaller the value, the smaller the deviation between the measured pressure curve and the ideal state, and the no obvious abnormality; the larger the value, the more significant the fluctuation of the pressure curve, and the possible leakage or unstable sealing problem.

[0141] Analysis shows that, under ideal sealing conditions, there are no leakage acoustic signals. The theoretical value is 0, therefore the acoustic activity degree can directly characterize the deviation between the measured acoustic characteristics and the ideal characteristics; therefore, based on the maximum frequency of acoustic events for this type of oil tank under the known maximum leakage in historical data, the above abnormal frequencies are normalized; thus, the acoustic activity degree is obtained. ;like This indicates that there are no valid acoustic events, which meets the ideal condition; if the acoustic activity level The larger the value, the more frequent the acoustic activity, and the higher the possibility of leakage.

[0142] This allows us to obtain the corresponding weights for pressure and acoustics. Among these, the pressure deviation weight... Directly using pressure attenuation deviation ,Right now The larger this value, the greater the contribution of pressure deviation to leakage determination, and the higher its weight should be assigned; acoustic activity weight. The frequency of acoustic events and the energy signal-to-noise ratio must be considered simultaneously, therefore:

[0143] ;

[0144] in, To measure the average energy of acoustic events, The average energy of the background noise is obtained using the aforementioned calculation method. If this parameter is less than 1, it indicates that the energy of the acoustic event is lower than the background noise, which may be a misjudgment. Therefore, 1 is taken as a correction value to avoid abnormal weighting. The larger the value, the higher the frequency of the acoustic event, the higher the signal-to-noise ratio, the higher the contribution of acoustic features to leakage determination, and the greater the weight.

[0145] Ultimately, this yields the comprehensive regional leakage index. The formula is:

[0146] ;

[0147] in, The first value, The second value, The third value, To preset a minimum value (taken as 1 × 10^-6), avoid a denominator of 0; by pressing... and The final result is obtained by weighted summation of proportions. It directly characterizes the degree of deviation between the current regional state and the ideal sealing state. The value ranges from 0 to 1; the larger the value, the higher the possibility of leakage.

[0148] For example, if the first threshold is set to 0.7, the system will test all test areas. Perform individual judgments; if a certain area If so, it is determined that there is a leak in the area.

[0149] Meanwhile, to avoid random errors from a single test, the system is equipped with a re-verification mechanism: if a leak is initially detected in a certain area, that area must be tested again; if the recalculation is successful... If so, then a leak in that area is confirmed; if a recalculation is performed... If the result is not found, it is determined to be accidental interference. A third test is required after checking the test environment, and the result of the third test will be used as the final determination.

[0150] In addition, the system needs to record key data during the decision-making process, including data from various regions. The number of repeated tests and results provide data support for subsequent result visualization and traceability report generation, ensuring that decision-making results are traceable and verifiable. At the same time, on the system interface, the test curves for each region are displayed as hyperbolas: one is the measured pressure curve, and the other is the ideal pressure curve predicted by the virtual twin. Users can intuitively see the separation between the two curves. The actual deviation of each feature is marked, clearly indicating which region and feature (pressure / acoustics) is significantly deviated from the ideal state based on the leakage judgment.

[0151] Through the above technical solution, this application effectively solves the problem of rigid weight allocation when fusion of multi-sensor data. When the pressure sensor is affected by environmental interference or the acoustic sensor is affected by noise pollution, the feature weight ratio is dynamically adjusted to avoid misjudgment caused by the abnormality of a single data source, and significantly improves the robustness and accuracy of the leak detection system in complex industrial environments.

[0152] This application also provides a leakage detection system for transformer tank production.

[0153] Please see Figure 2 , Figure 2 This is a schematic diagram of the structure of one embodiment of this application.

[0154] like Figure 2 As shown, a leakage detection system 200 for transformer tank production includes: a division module 210 for dividing the tank under test into multiple test areas; a first determination module 220 for obtaining an ideal net pressure decay rate by using the average rate of pressure drop in the tank cavity, the pressure change rate corresponding to temperature change, and the volume change in the target test area within a target time period; a second determination module 230 for determining the pressure decay deviation by using the ideal net pressure decay rate and the measured net pressure decay rate; a third determination module 240 for determining the acoustic activity weight by using the acoustic activity of the acoustic signal in the acoustic event and the average background noise energy of the multiple test areas in the acoustic event; and a fourth determination module 250 for determining the regional leakage comprehensive index by using the acoustic activity weight and the pressure decay deviation, and then using the regional leakage comprehensive index to determine the leakage detection status of the tank.

[0155] This application provides a leakage detection method and system for transformer tank production. By combining pressure attenuation deviation and dynamic acoustic activity weighting to calculate the regional leakage comprehensive index, it effectively eliminates the interference of temperature changes on pressure attenuation rate and adaptively adjusts acoustic signal weights. It has the advantages of improving leakage detection accuracy, effectively eliminating environmental temperature interference, and dynamically adapting to changes in background noise.

[0156] In other words, this application employs a zoned collaborative and multimodal data synchronous acquisition approach. Based on a high-precision CAD model, the test area is automatically divided, and pressure, acoustic, and temperature data are acquired synchronously. Unified timestamps and filtering preprocessing ensure the spatiotemporal correlation of the data. Secondly, in the feature extraction stage, the concept of net pressure decay rate is introduced. The ideal gas law and temperature compensation are used to remove thermodynamic interference, accurately characterizing the pressure changes caused by actual leakage. Simultaneously, by combining the static stability characteristics of the pressure sequence and the frequency and energy analysis of acoustic events, continuous and intermittent leaks are effectively distinguished, and the leakage intensity is identified. A physical model is constructed based on a digital twin model to simulate pressure behavior under ideal sealing conditions. By calculating the pressure decay deviation and curve smoothness deviation, the deviation between the measured and ideal states is quantified, and combined with acoustic activity, multi-feature fusion is achieved. Finally, a dynamic weighted fusion decision is adopted, adaptively adjusting the weights based on the real-time signal-to-noise ratio of pressure and acoustic features to improve sensitivity to micro-leakage and anti-interference capabilities, significantly reducing the false alarm rate.

[0157] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0158] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A leakage detection method for transformer tank production, characterized in that, include: The fuel tank under test is divided into multiple test areas; The ideal net pressure decay rate is determined by using the average rate of pressure drop, the rate of pressure change corresponding to temperature change, and the volume change of the fuel tank cavity in the target test area within the target time period. The pressure attenuation deviation is determined using the ideal net pressure attenuation rate and the measured net pressure attenuation rate. The acoustic activity weight is determined by using the acoustic activity of the acoustic signal in an acoustic event and the average energy of background noise in multiple test areas of the acoustic event. This includes: acquiring the energy of the acoustic signal collected by a background noise sampling window of a target size, and determining the average energy of the background noise using the sum of squares of the signals corresponding to each frame within the background noise sampling window and the mean of the sum of squares of all frames within the background noise sampling window; acquiring the smoothness of the ideal pressure curve and the smoothness of the measured pressure curve; determining the curve smoothness deviation using the ideal pressure curve smoothness and the smoothness of the measured pressure curve; normalizing the curve smoothness deviation to obtain the acoustic activity; acquiring multiple signal segments identified as acoustic events, and acquiring the energy integral of the acoustic event during the pressure holding period to determine the average energy of the measured acoustic event corresponding to each signal segment; and determining the acoustic activity weight using the average energy of the measured acoustic event, the acoustic activity, and the average energy of the background noise. Using the acoustic activity weights and the pressure attenuation deviation, a comprehensive regional leakage index is determined, and then the leakage detection status of the fuel tank is determined using the comprehensive regional leakage index.

2. The leakage detection method for transformer tank production according to claim 1, characterized in that, The determination of the ideal net pressure decay rate by utilizing the average rate of pressure drop, the rate of pressure change corresponding to temperature change, and the volume change in the tank cavity within the target test area over a target time period includes: Obtain the initial pressure and initial ambient temperature of each test area at the start of the pressure holding period, and obtain the end pressure and end ambient temperature at the end of the pressure holding period; Using the initial pressure, the final pressure, the start time of the pressure holding period, and the end time of the pressure holding period, the average rate of pressure drop in the tank cavity of each test area within the target time period is determined; The ideal pressure value is determined using the initial pressure, the initial ambient temperature, and the change in ambient temperature. The ideal net pressure decay rate is determined using the ideal pressure value, the start time of pressure holding, the end time of pressure holding, and the volume change of the tank cavity during the pressure holding period.

3. The leakage detection method for transformer tank production according to claim 2, characterized in that, The determination of the ideal net pressure decay rate using the ideal pressure value, the start time of pressure holding, the end time of pressure holding, and the volume change of the tank cavity during the pressure holding period includes: Obtain the time difference between the start and end times of the pressure holding, and use the ratio between the time difference and the ideal pressure value to determine the pressure change rate corresponding to the temperature change; Obtain the initial volume of the fuel tank cavity, use the initial volume to determine the volume change, and use the volume change to determine the corrected ideal pressure value; The ideal net pressure decay rate is determined using the corrected ideal pressure value, the average rate of pressure drop, and the rate of pressure change.

4. The leakage detection method for transformer tank production according to claim 1, characterized in that, The determination of pressure attenuation deviation using the ideal net pressure attenuation rate and the measured net pressure attenuation rate includes: Obtain the average rate of pressure drop in the tank cavity for each test area within the target time period, and obtain the standard deviation of the ratio corresponding to the average rate of pressure drop in all historical test processes; The pressure attenuation deviation is determined using the standard deviation, the ideal net pressure attenuation rate, and the measured net pressure attenuation rate.

5. The leakage detection method for transformer tank production according to claim 1, characterized in that, The process of determining a comprehensive regional leakage index using the acoustic activity weights and the pressure attenuation deviation, and then using the comprehensive regional leakage index to determine the leakage detection status of the fuel tank, includes: Get the preset minimum value; The pressure attenuation deviation is used to determine the pressure deviation weight; The regional leakage comprehensive index is determined using the pressure attenuation deviation, the pressure deviation weight, the acoustic activity weight, and the preset minimum value. The leakage detection status of the fuel tank is determined using the regional leakage comprehensive index. If the regional leakage comprehensive index of the target test area is greater than a first threshold, it is determined that there is a leakage in the target test area where the regional leakage comprehensive index is greater than the first threshold. Alternatively, if at least one of the regional leakage comprehensive indices of the target test area is less than the first threshold, it is determined that there is interference in the target test area.

6. The leakage detection method for transformer tank production according to claim 5, characterized in that, The determination of the regional leakage comprehensive index using the pressure attenuation deviation, the pressure deviation weight, the acoustic activity weight, and the preset minimum value includes: The first value is determined using the pressure attenuation deviation and the pressure deviation weight; The second value is determined using the acoustic activity weights and the acoustic activity degree; The third value is determined using the pressure deviation weight, the acoustic activity weight, and the preset minimum value; The comprehensive leakage index of the area is determined using the first value, the second value, and the third value.

7. A leakage detection system for transformer tank production, characterized in that, The leakage detection system for transformer tank production includes: The partitioning module is used to divide the fuel tank under test into multiple test areas; The first determination module uses the average rate of pressure drop in the tank cavity, the rate of pressure change corresponding to temperature change, and the volume change in the target test area within the target time period to obtain the ideal net pressure decay rate. The second determining module uses the ideal net pressure decay rate and the measured net pressure decay rate to determine the pressure decay deviation. The third determining module uses the acoustic activity degree of the acoustic signal in the acoustic event and the average energy of background noise in multiple test areas of the acoustic event to determine the acoustic activity weight. This includes: acquiring the energy of the acoustic signal collected by a background noise sampling window of a target size, and using the sum of squares of the signals corresponding to each frame within the background noise sampling window and the mean of the sum of squares of all frames within the background noise sampling window to determine the average energy of the background noise; acquiring the smoothness of the ideal pressure curve and the smoothness of the measured pressure curve; using the smoothness of the ideal pressure curve and the smoothness of the measured pressure curve to determine the curve smoothness deviation; normalizing the curve smoothness deviation to obtain the acoustic activity degree; acquiring multiple signal segments identified as acoustic events, and acquiring the energy integral of the acoustic event during the pressure holding period to determine the average energy of the measured acoustic event corresponding to each signal segment; and using the average energy of the measured acoustic event, the acoustic activity degree, and the average energy of the background noise to determine the acoustic activity weight. The fourth determination module uses the acoustic activity weight and the pressure attenuation deviation to determine the regional leakage comprehensive index, and then uses the regional leakage comprehensive index to determine the leakage detection status of the oil tank.

Images