Leak detection method, device and system for pressure vessels

Abstract

The present application relates to the field of sealing detection technology, and specifically to a leak detection method, device and system for pressure vessels, the method comprising: respectively obtaining the acoustic emission signals of each detection sensor on the outer wall of the pressure vessel and the guard sensor directly above it, as well as the liquid level at each moment; judging the acoustic emission events that occur; obtaining each event set and its corresponding noise interference degree, and dividing the event set into a noise event set and a non-noise event set; determining the sound source estimation value and the droplet estimation value, obtaining the modal decomposition layer number, and locating the pressure vessel leakage position after performing signal reconstruction on all modal components corresponding to the non-noise through modal decomposition. The present application dynamically adjusts the decomposition layer number of the modal decomposition algorithm to avoid under-decomposition or over-decomposition of the acoustic emission signal mode, improves the filtering effect of the acoustic emission signal, and can more accurately locate the leakage point of the pressure vessel.

Description

Technical Field

[0001] The present application relates to the technical field of sealing detection, and in particular to a leak detection method, equipment and system for pressure vessels. Background Art

[0002] Pressure vessels are widely used in numerous industries, including chemical, energy, and food. Their safety is paramount. Leaks not only cause the loss of media and waste resources, but can also lead to serious accidents such as fires, explosions, and poisoning, threatening human life and damaging the ecological environment.

[0003] In the process of leak detection of pressure vessels using the traditional method of acoustic emission detection, the acoustic emission signal will be attenuated due to medium absorption, scattering and the presence of interference sources in the propagation path of large-volume pressure vessels, resulting in an increase in the influence of noise on the time delay estimation of the waveform of the acoustic emission signal. Secondly, for complex, variable, nonlinear and non-stationary acoustic emission signals, when filtering and denoising are performed using the modal decomposition algorithm, the number of decomposition layers of the modal decomposition algorithm needs to be set in advance, which is prone to modal aliasing or over-decomposition, affecting the accuracy of the final extraction of feature information in the acoustic emission signal, resulting in low accuracy in locating the pressure vessel leakage point. Summary of the Invention

[0004] In order to solve the above technical problems, a leak detection method, equipment and system for pressure vessels are provided to solve the existing problems.

[0005] The solution to the technical problem of this application is to provide a leak detection method, equipment and system for pressure vessels, including the following steps:

[0006] In a first aspect, an embodiment of the present application provides a leak detection method for a pressure vessel, the method comprising the following steps:

[0007] Acquire the acoustic emission signals of each detection sensor on the outer wall of the pressure vessel and the guard sensor directly above it, as well as the liquid level height at each moment;

[0008] The acoustic emission signal is divided into frames, and the upper envelope of each frame signal under each detection sensor is extracted. The occurrence of acoustic emission events is judged based on the deviation of the liquid level height at all times, the size of the signal amplitude in the upper envelope, and the distribution of extreme points.

[0009] Analyze the energy intensity of the signal corresponding to each acoustic emission event under each frame signal of each detection sensor and the duration of the event, combine the frequency distribution of the signal in the frequency domain, form a feature vector, classify the acoustic emission events of the same frame signal under each detection sensor and its adjacent detection sensors, and obtain each event set corresponding to each frame signal under each detection sensor; determine the noise interference degree of each event set based on the difference in the feature vectors between different acoustic emission events in each event set and the number of acoustic emission events, and divide the event set into a noise event set and a non-noise event set;

[0010] Based on the number of noise event sets and the number of non-noise event sets and the noise interference degree, the sound source estimation value of each frame signal under each detection sensor is determined; the energy intensity of the signal corresponding to each acoustic emission event in each frame signal under the guard sensor is analyzed, and the droplet estimation value is calculated. Combined with the sound source estimation value, the number of modal decomposition layers of each frame signal under each detection sensor is obtained, and each frame signal under each detection sensor is modally decomposed to obtain all modal components; the modal component corresponding to the non-noise is obtained through the correlation between each modal component and the remaining modal components, and the leakage position of the pressure vessel is located after signal reconstruction of all modal components corresponding to the non-noise.

[0011] Preferably, the judging of the acoustic emission event comprises:

[0012] The ratio of the average liquid level height at all times to the preset rated operating height is recorded as the liquid level ratio;

[0013] Effective threshold The calculation formula is: ,in, is the preset reference value, is the liquid level ratio, is the preset reference ratio, is an exponential function with a natural constant as its base;

[0014] Obtain the maximum and minimum points of the upper envelope; extend backward from any maximum point as the starting point, and record the moment when the first signal amplitude after the starting point in the upper envelope is less than the effective threshold value and the first minimum point after the starting point as critical points; and record the critical point closest to the starting point as the ending point;

[0015] For each frame signal of each detection sensor, the period between the starting point and the ending point corresponding to each maximum point is recorded as an acoustic emission event.

[0016] Preferably, the method for obtaining the feature vector is:

[0017] For each acoustic emission event corresponding signal in each frame signal of each detection sensor, calculate the duration of the signal corresponding to each acoustic emission event, which is recorded as the duration; calculate the product of the total energy of all signal amplitudes in the signal corresponding to each acoustic emission event and the duration as the evaluation value of each acoustic emission event;

[0018] Perform frequency domain analysis on the signal corresponding to each acoustic emission event to obtain a spectrum. The frequency corresponding to the maximum energy in the spectrum is recorded as the main frequency; the frequency corresponding to the central value after integrating all the energy in the spectrum is recorded as the center frequency.

[0019] Count the ratio of the number of signal amplitudes less than the effective threshold value in the signal corresponding to each acoustic emission event to the duration, and record it as the average frequency;

[0020] Normalization is performed on the main frequency, the center frequency, the average frequency, and the evaluation value respectively, and a feature vector is formed.

[0021] Preferably, the obtaining of each event set corresponding to each frame signal of each detection sensor includes: clustering the feature vectors corresponding to all acoustic emission events of the same frame signal of each detection sensor and all adjacent detection sensors to obtain multiple clusters, and recording all acoustic emission events belonging to the same cluster as each event set corresponding to each frame signal of each detection sensor.

[0022] Preferably, determining the noise interference degree of each event set and dividing the event set into a noise event set and a non-noise event set includes:

[0023] For each event set corresponding to each frame signal of any detection sensor, the average of the distances between the feature vectors of all two acoustic emission events belonging to any detection sensor in each event set is recorded as the dispersion of each event set;

[0024] Counting the number of all acoustic emission events in each event set, recording it as a total number; calculating the ratio of the number of all acoustic emission events belonging to any detection sensor in each event set to the total number, recording it as a relative ratio;

[0025] The noise interference degree is a normalized result of the product of the relative ratio and the dispersion;

[0026] Obtain a segmentation threshold of the noise interference degree of all event sets corresponding to each frame signal of any detection sensor, and record it as a first segmentation threshold; record the event set whose noise interference degree is greater than the first segmentation threshold as a noise event set, otherwise, record it as a non-noise event set.

[0027] Preferably, The next detection sensor is the next detection sensor Sound source estimation value of frame signal The calculation formula is: ,in, For the The next detection sensor The number of all non-noise event sets corresponding to the frame signal, For the The detection sensor is in the The number of all noise event sets corresponding to the frame signal, For the The next detection sensor The cumulative sum of the noise interference levels of all noise event sets corresponding to the frame signal, is the ceiling function.

[0028] Preferably, the calculation of the droplet estimation value, combined with the sound source estimation value, to obtain the modal decomposition layer number of each frame signal under each detection sensor includes:

[0029] Based on the occurrence period of each acoustic emission event in each frame signal of each detection sensor, the total energy of all signal amplitudes of the signal corresponding to the occurrence period in the same frame signal of the guard sensor directly above is recorded as the energy characteristic value;

[0030] Obtain the segmentation threshold of the energy characteristic values corresponding to all acoustic emission events in each frame of the guard sensor signal, recorded as the second segmentation threshold; count the number of acoustic emission events with energy characteristic values greater than the second segmentation threshold, as the droplet estimation value of each frame of the guard sensor signal;

[0031] The modal decomposition layer number is the sum of the sound source estimation value of each frame signal under each detection sensor and the droplet estimation value of the same frame signal of the guard sensor directly above.

[0032] Preferably, the process of acquiring the modal components corresponding to the non-noise is as follows: calculating the degree of correlation between each modal component and the remaining modal components; counting the number of modal components whose correlation degree is less than a preset threshold value between each modal component and all other modal components, and recording it as the judgment coefficient of each modal component; selecting the modal components whose judgment coefficient is less than the preset number threshold value, and recording them as the modal components corresponding to the non-noise.

[0033] In the second aspect, an embodiment of the present application also provides a leak detection device for pressure vessels, which includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any of the above-mentioned leak detection methods for pressure vessels.

[0034] In a third aspect, an embodiment of the present application further provides a leak detection system for a pressure vessel, wherein the system includes a signal analysis module, and the submodules of the module include:

[0035] The filtering processing submodule is used to obtain the reconstructed acoustic emission signal, including: framing the acoustic emission signal, extracting the upper envelope of each frame signal under each detection sensor; judging the acoustic emission events that meet the acoustic emission event judgment conditions according to the deviation of the liquid level height at all times, the size of the signal amplitude in the upper envelope and the distribution of extreme points; analyzing the energy intensity of the signal corresponding to each acoustic emission event under each frame signal of each detection sensor and the duration of the event, combining the frequency distribution of the signal in the frequency domain to form a feature vector, classifying the acoustic emission events of the same frame signal under each detection sensor and its adjacent detection sensors, and obtaining each event set corresponding to each frame signal under each detection sensor; and classifying the acoustic emission events of the same frame signal under each detection sensor by comparing the different acoustic emission events in each event set. The difference in the characteristic vectors between the two and the number of acoustic emission events is used to determine the noise interference degree of each event set, and the noise event set and the non-noise event set are obtained; based on the number of noise event sets and the number of non-noise event sets and the noise interference degree, the sound source estimation value of each frame signal under each detection sensor is determined; the energy intensity of the signal corresponding to each acoustic emission event in each frame signal under the guard sensor is analyzed, the droplet estimation value is calculated, and combined with the sound source estimation value, the modal decomposition layer number of each frame signal under each detection sensor is obtained, and all modal components are obtained by performing modal decomposition on each frame signal under each detection sensor; the modal component corresponding to the non-noise is obtained through the correlation between each modal component and the other modal components, and the signal of all modal components corresponding to the non-noise is reconstructed;

[0036] The time delay estimation submodule is used to estimate the time delay of the reconstructed acoustic emission signal;

[0037] The positioning algorithm submodule is used to locate the leakage position of the container bottom plate through the reconstructed acoustic emission signal and the result of time delay estimation.

[0038] This application has at least the following beneficial effects:

[0039] The present application extracts the upper envelope of each frame signal under each detection sensor, analyzes the change of signal amplitude after the maximum point in the upper envelope, and determines the acoustic emission events that meet the acoustic emission event judgment conditions. Its beneficial effect is that it takes into account the liquid level height and dynamically sets the effective threshold value of the acoustic emission signal. When the liquid level height is low, it improves the detection accuracy of the weak acoustic emission signal when the container bottom plate leaks. Then, it forms a feature vector, classifies the acoustic emission events of the same frame signal under each detection sensor and its adjacent detection sensors, and obtains each event set corresponding to each frame signal under each detection sensor. Its beneficial effect is that it takes into account the adjacent detection sensors detecting the same acoustic emission event. The possibility of the emission source signal is large. By classifying the acoustic emission events in the same frame signal under adjacent detection sensors, the events that may be of the same type are divided together, and then the noise interference degree of each event set is determined, and the noise event set and the non-noise event set are obtained. Its beneficial effect is that the number of acoustic emission events belonging to each detection sensor in the event set is used to illustrate the difference in signals between each detection sensor and adjacent sensors, so as to reflect the significant interference of the signals of each detection sensor by noise aliasing; determine the sound source estimation value of each frame signal under each detection sensor, calculate the droplet estimation value, and obtain the number of modal decomposition layers of each frame signal under each detection sensor, which has a beneficial effect. The result is that by analyzing the number of noise event sets and non-noise event sets, the acoustic emission sources appearing in each frame signal of each detection sensor are estimated. In addition, according to the significant situation of the energy of the signal corresponding to each acoustic emission event in each frame signal of the guard sensor, the acoustic emission events of the high-energy signals appearing in each frame signal of the guard sensor are estimated, so as to evaluate the interference effect of the high-energy signal in the guard sensor on the detection sensor below, so as to dynamically adjust the number of decomposition layers of the modal decomposition algorithm according to the signal mixing and superposition of the acoustic emission sources generated by the bottom plate and the top in the acoustic emission signal of the detection sensor, so as to avoid the mode of the acoustic emission signal being under-decomposed or over-decomposed. Decomposition; obtain the modal component corresponding to the non-noise through the correlation between each modal component and the remaining modal components, reconstruct the signal of all modal components corresponding to the non-noise, and locate the leakage position of the pressure vessel through the reconstructed acoustic emission signal. Its beneficial effect is that by analyzing the correlation between the modal components, the mode corresponding to the noise with weak correlation is eliminated, and the signal is reconstructed, thereby effectively eliminating the noise signal generated by the falling droplets, reducing the interference effect of the detection sensor on the bottom plate leakage detection, and can improve the filtering effect of the acoustic emission signal of the detection sensor, and improve the accuracy of locating the leakage point of the pressure vessel through the acoustic emission signal of the detection sensor. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The following is a detailed description of a leak detection method for pressure vessels of the present application in conjunction with the accompanying drawings.

[0041] Figure 1 A flowchart of a leak detection method for a pressure vessel provided in an embodiment of the present application;

[0042] Figure 2 A schematic diagram of the deployment of acoustic emission sensors on the outer wall of a pressure vessel provided in an embodiment of the present application;

[0043] Figure 3 A flowchart of the steps of the method for obtaining the number of modal decomposition levels provided in an embodiment of the present application;

[0044] Figure 4 A block diagram of a leak detection system for pressure vessels provided in an embodiment of the present application. DETAILED DESCRIPTION

[0045] To make the purpose, technical solutions, and advantages of this application more clearly understood, the following, in conjunction with the accompanying drawings and implementation examples, further describes in detail a leak detection method, device, and system for pressure vessels proposed in this application. It should be understood that the specific embodiments described herein are merely intended to explain this application and are not intended to limit this application.

[0046] Unless defined otherwise, 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 application belongs.

[0047] See also Figure 1 , which shows a flowchart of a leak detection method for a pressure vessel provided by an embodiment of the present application, the method comprising the following steps:

[0048] Step 1: Acquire the acoustic emission signals of each detection sensor on the outer wall of the pressure vessel and the guard sensor directly above it, as well as the liquid level height at each moment.

[0049] Pressure vessels generally refer to sealed devices that can withstand the pressure of fluid media, such as various pressure gas cylinders, glove boxes, and storage tanks. With the acceleration of industrialization, pressure vessels play an indispensable role in key industries such as petroleum, chemical industry, and electric power. Due to long-term use, the bottom plates of pressure vessels will corrode, rust, and impact, causing damage to the container and leading to dangers such as leakage.

[0050] Traditionally, pressure vessel leak testing involves opening the tank. However, this traditional method, which requires stopping production to open the tank, results in a significant waste of manpower, material, and financial resources. Online testing technology, however, can detect corrosion on the bottom of the container without opening it. Key online testing methods include acoustic emission (AE), ultrasonic guided waves, and robotics. AE, a non-destructive leak detection method based on weak acoustic signals generated by tiny internal displacements of the material, is widely used for leak detection in sealed pressure vessels.

[0051] When corrosion defects occur in the bottom plate of a pressure vessel, the material strength decreases, causing localized micro-deformation under the action of the liquid level, leading to the peeling and shedding of corrosion products, generating acoustic emission signals. Therefore, when a leak occurs, the flow of the medium will generate continuous acoustic emission signals.

[0052] Due to the temperature difference between the inside and outside of the pressure vessel, liquid droplets form on the top of the vessel. When these droplets fall to the liquid surface, they also generate acoustic emission signals. These signals are very similar to those generated by leaks on the container bottom. Conventional signal acquisition methods cannot distinguish the acoustic emission signals from the container bottom defect and the droplets on the top of the container. To solve this problem, guard sensors and detection sensors are deployed on the outer wall of the container. The guard sensors filter out the acoustic emission noise signals from the droplets on the top of the container.

[0053] Two groups of acoustic emission sensors are arranged on the outer wall of the container. One group receives the acoustic emission signal generated by the impact of liquid droplets on the liquid surface from the top of the container, which is the guard sensor, and the other group receives the acoustic emission signal from the bottom plate of the container, which is recorded as the detection sensor;

[0054] In this embodiment, a group of acoustic emission sensors are uniformly deployed along the outer wall surface at a height of 0.75m from the bottom plate, wherein the number of a group of acoustic emission sensors is 8, and this group of acoustic emission sensors is used as a detection sensor for receiving the acoustic emission signal of the bottom plate of the container; secondly, at a position where the height from the bottom plate of the container is 0.85 times the preset rated operating height of the container, another group of acoustic emission sensors are uniformly deployed along the outer wall surface at a height of 0.85 times the preset rated operating height of the container, and this group of acoustic emission sensors is used as a guard sensor, wherein the number of guard sensors is also 8. As other implementation methods, the implementer can deploy the acoustic emission sensors according to actual conditions. Therefore, by Converted into a digital signal to obtain the acoustic emission signal collected by each detection sensor and each guard sensor; in this embodiment, the guard sensor is directly above the detection sensor, the center frequency of the sensor is selected to be 140kHz, the frequency range is 50-200kHz, the sensitivity is selected to be >65, the acquisition frequency is 1MHz,, as other implementation methods, the implementer can set it according to the actual situation; secondly, the rated operating height refers to the maximum liquid level height allowed to be reached by the container under design and normal operating conditions, the maximum liquid level height is 0.8 or 0.9 times the container height, in this embodiment, the container height is 21.8, and 0.8 of the container height is used as the rated operating height.

[0055] In this embodiment, the schematic diagram of the deployment of the acoustic emission sensor on the outer wall of the pressure vessel provided in this embodiment is as follows: Figure 2 As shown, 201 is a detection sensor, 202 is a guard sensor, 203 is the liquid level of the container, and 204 is the outer wall of the container.

[0056] Among them, when a pressure vessel leaks, the liquid medium flows out from the leakage point and generates an acoustic emission signal. Due to the huge area of the container bottom plate, the acoustic emission signal at the leakage position is attenuated. The leakage position is recorded as the acoustic emission source. Then, when the acoustic emission sensor receives the signal at the acoustic emission source, interference signals will inevitably appear, causing the waveform of the acoustic emission sensor to be complex, resulting in the signal of the acoustic emission source being masked, seriously affecting the acoustic emission signal at the leakage point and the accuracy of the time delay estimation of the leak location. Therefore, it is necessary to reduce the noise of the acoustic emission signal and eliminate the interference signal.

[0057] Since acoustic emission detection of containers is dependent on the liquid level, a radar level meter is installed at the center of the top of the container to ensure that the radar wave can propagate vertically downward and reduce the interference of reflected waves. The radar level meter is used to monitor the liquid level of the container in real time and collect the liquid level at each moment. The sampling frequency of the radar level meter is 10 Hz. As other implementation methods, the implementer can set it according to the actual situation.

[0058] Thus, according to the signal acquisition module, the acoustic emission signals of each detection sensor and the guard sensor directly above it, as well as the liquid level height at each moment are obtained.

[0059] Step 2: Frame the acoustic emission signal and extract the upper envelope of each frame signal under each detection sensor; judge the acoustic emission event based on the deviation of the liquid level height at all times, the size of the signal amplitude in the upper envelope, and the distribution of extreme points.

[0060] Secondly, to facilitate the subsequent feature analysis of the signal at the acoustic emission source on the bottom plate of the container, the signals collected by the detection sensor and the guard sensor are framed and processed separately. The acoustic emission signal is divided into multiple frame segments. Each frame signal of each detection sensor and each frame signal of the guard sensor directly above each detection sensor are obtained.

[0061] In this embodiment, the frame length of each frame during the frame processing is 100 ms, that is, the time length of each frame signal is 100 ms. As other implementation methods, the implementer can set it according to actual conditions.

[0062] During the acoustic emission detection process, in addition to the signals generated by phenomena such as leakage from the container bottom, cracking of the substrate, and stress corrosion cracking, there will also be various background noises and other interference signals, such as interference caused by liquid droplets dripping from the top of the container. Among them, changes in the liquid level will affect the propagation and amplitude of the acoustic emission signal collected by the detection sensor. The pressure of the liquid on the bottom and side walls of the container increases with the increase of the liquid level. When the liquid level is low, the pressure intensity of the liquid on the bottom of the container is low; when the liquid level is high, the pressure intensity of the liquid on the bottom of the container is high. During the acoustic emission detection process, the valve of the pressure vessel will be closed, and the degree of change in the liquid level will be extremely small. The effective threshold value is set according to the liquid level to preliminarily filter the interference signal and identify the acoustic emission event. Specifically:

[0063] The ratio of the average liquid level height at all times to the preset rated operating height is recorded as the liquid level ratio;

[0064] In this embodiment, the preset rated operating height refers to the maximum liquid level height of the pressure vessel, and the preset rated operating height is 0.8 times the container height, where the container height is 21.8 meters. The container height is obtained through the container production data. As other implementation methods, the implementer can determine it according to actual conditions.

[0065] The effective threshold value is determined as follows:

[0066]

[0067] in, is the effective threshold value, is the preset reference value, is the liquid level ratio, is the preset reference ratio, is an exponential function with a natural constant as its base.

[0068] In this embodiment, the preset reference value Set to 15dB, preset reference ratio The preset reference ratio is set to 0.85, wherein the selection of the preset reference ratio is based on the industry standard "Atmospheric Pressure Storage Tank Corrosion Monitoring Method and Implementation Guide", which stipulates that the online detection liquid level of the tank bottom plate acoustic emission should be located at 85%-105% of the highest operating liquid level. The pressure vessel of this embodiment takes the storage tank as an example, and 0.85 is selected as the preset reference ratio. As for other implementation methods, the implementer can set it according to actual conditions.

[0069] It should be noted that if , indicating that the pressure intensity of the liquid in the pressure vessel is low and the acoustic emission signal is weakened. In order to detect these weaker acoustic emission signals more sensitively, the effective threshold value needs to be lowered.

[0070] Furthermore, when leak detection is performed on the bottom plate of a pressure vessel, leakage is a continuous process that generates continuous acoustic emission signals. Acoustic emission signals such as cracking of the bottom plate substrate and stress corrosion cracking are mutation signals that appear as obvious spikes or pulses. By analyzing the changing trend of each frame of the signal and combining it with the effective threshold value, the acoustic emission event is determined, specifically:

[0071] Adopting envelope extraction algorithm, extract the upper envelope of each frame signal of each detection sensor;

[0072] In this embodiment, the Hilbert transform algorithm is used to extract the upper envelope, wherein the Hilbert transform algorithm is a well-known technology and will not be described in detail here. As other implementation methods, the implementer can adopt other methods of the existing technology, such as the cepstrum method, etc. This embodiment does not impose any special restrictions on this.

[0073] Obtaining the maximum and minimum points of the upper envelope;

[0074] In this embodiment, the AMPD (Automatic multiscale-based peak detection) algorithm is used to obtain extreme points. The AMPD algorithm is a well-known technology and will not be described in detail here.

[0075] Extending backward from any maximum point as the starting point, the moment when the first signal amplitude after the starting point in the upper envelope is less than the effective threshold value and the first minimum point after the starting point are both recorded as critical points;

[0076] The critical point closest to the starting point is recorded as the ending point;

[0077] For each frame signal of each detection sensor, the time period between the starting point and the ending point corresponding to each maximum point is recorded as an acoustic emission event;

[0078] It should be noted that the maximum point is taken as the starting point to reflect the beginning of a potential acoustic emission event. An acoustic emission event may include acoustic emission signals of container bottom plate leakage, or may include sudden acoustic emission signals such as substrate cracking and stress corrosion cracking.

[0079] At this point, the acoustic emission event is obtained.

[0080] Step 3: Analyze the energy intensity of the signal corresponding to each acoustic emission event under each frame signal of each detection sensor and the duration of the event, combine the frequency distribution of the signal in the frequency domain, form a feature vector, classify the acoustic emission events of the same frame signal under each detection sensor and its adjacent detection sensors, and obtain each event set corresponding to each frame signal under each detection sensor; determine the noise interference degree of each event set through the difference in the feature vectors between different acoustic emission events in each event set and the number of acoustic emission events, and obtain the noise event set and non-noise event set.

[0081] Secondly, the continuous acoustic emission signal of the container bottom plate leakage lasts longer than the sudden acoustic emission signal such as substrate cracking and stress corrosion cracking, and the overall intensity of the continuous signal is higher during the duration. Therefore, the duration of the signal under each acoustic emission event and its signal energy are analyzed to calculate the evaluation value, which is specifically:

[0082] Calculate the duration of the signal corresponding to each acoustic emission event in each frame signal of each detection sensor, and record it as duration;

[0083] It should be noted that, for the sake of convenience, for the rth frame signal under the mth detection sensor, if During this period, an acoustic emission event occurs, and the r-th frame signal of the m-th detection sensor is The signal of this period is recorded as the corresponding signal of the acoustic emission event in the r-th frame signal under the m-th detection sensor; the signal in the r-th frame signal under the guard sensor directly above the m-th detection sensor is recorded as the corresponding signal of the acoustic emission event in the r-th frame signal under the m-th detection sensor. The signal of this period is recorded as the signal corresponding to the acoustic emission event in the rth frame signal under the guard sensor directly above.

[0084] It should be noted that the duration is the time interval between the starting point and the ending point of each acoustic emission event.

[0085] Calculate the product of the energy of all signal amplitudes in the signal corresponding to each acoustic emission event in each frame signal of each detection sensor and the duration as the evaluation value of each acoustic emission event;

[0086] In this embodiment, the calculation of the energy of all signal amplitudes in the signal corresponding to each acoustic emission event is a well-known technique, and the energy is calculated by calculating the sum of the squares of all signal amplitudes in the signal corresponding to each acoustic emission event.

[0087] It should be noted that the longer the duration is, the longer the event lasts, and the stronger the continuity of the signal is. Secondly, the greater the total energy is, the higher the overall intensity of the signal is during the duration, and the higher the possibility of it being a continuous acoustic emission signal. The larger the obtained evaluation value is, the more likely it is that the acoustic emission event is a continuous signal caused by a bottom plate leakage.

[0088] Furthermore, by analyzing the characteristic differences of the acoustic emission event corresponding signals between adjacent detection sensors in the same frame signal, the acoustic emission events are classified as follows:

[0089] Perform frequency domain analysis on the signal corresponding to each acoustic emission event in each frame of the signal from each detection sensor to obtain a spectrum diagram. The frequency corresponding to the maximum energy in the spectrum diagram is recorded as the main frequency.

[0090] In this embodiment, a Fourier transform algorithm is used to obtain a frequency spectrum diagram, wherein the Fourier transform algorithm is a well-known technology and will not be described in detail here.

[0091] The frequency corresponding to the center value after integrating all the energy in the spectrum is recorded as the center frequency;

[0092] Count the number of times the signal amplitude is less than the effective threshold value in each acoustic emission event in each frame signal of each detection sensor, record it as the number of ringings, and record the ratio of the number of ringings to the duration as the average frequency;

[0093] It should be noted that the number of ringings reflects the activity of the signal in this event, which may correspond to more intense acoustic emission activity. Among them, continuous signals such as leakage events may have more ringing times because their signal duration is long, the energy is relatively stable, and it fluctuates frequently around the effective threshold value; while sudden signal events such as substrate cracking events may have relatively few ringing times, but the signal intensity may be high and the amplitude is large.

[0094] Normalizing the main frequency, the center frequency, the average frequency, and the evaluation value respectively, and forming a feature vector;

[0095] In this embodiment, the maximum and minimum normalization method is used for normalization processing, wherein the maximum and minimum normalization method is a well-known technology and will not be described here in detail. As other implementation methods, the implementer can adopt other methods of the existing technology, such as the Z-score normalization method, and this embodiment does not impose any special restrictions on this.

[0096] The remaining detection sensors adjacent to any detection sensor are recorded as adjacent sensors;

[0097] Clustering the feature vectors corresponding to all acoustic emission events of the same frame signal under any detection sensor and all adjacent sensors to obtain multiple clusters;

[0098] All acoustic emission events belonging to the same cluster are recorded as event sets corresponding to each frame signal of any detection sensor;

[0099] In this embodiment, the density peak clustering algorithm (DPC) is used for clustering. The DPC algorithm is a well-known technology and will not be described in detail here. As other implementation methods, the implementer can adopt other methods of the existing technology, such as hierarchical clustering algorithms, etc. This embodiment does not impose any special restrictions on this.

[0100] It should be noted that, if the acoustic emission events in the cluster have similar feature vectors, they are more likely to correspond to acoustic emission sources of the same position and type of container bottom plate. Therefore, each event set corresponds to the same type of acoustic emission events.

[0101] Furthermore, as a dynamic non-destructive testing method, acoustic emission detection technology can be masked or distorted by background noise and environmental interference during the propagation process, affecting the accuracy of event detection and, in turn, the positioning results. However, since the distance between adjacent detection sensors is relatively close, the acoustic emission attenuation is low, and the correlation between acoustic emission events is high. If the noise interference level of acoustic emission events is low, they are easily classified into the same category. By analyzing the number of sensors to which the acoustic emission events belong in each event set corresponding to any detection sensor, the noise interference level is calculated, specifically:

[0102] The average of the distances between the feature vectors of all two acoustic emission events belonging to any detection sensor in each event set is used as the dispersion of each event set corresponding to each frame signal of any detection sensor;

[0103] In this embodiment, the distance is measured by measuring the DTW distance of the feature vectors of any two acoustic emission events belonging to any detection sensor in each event set. As other implementation methods, the implementer can adopt other methods of the prior art, such as Euclidean distance, etc., and this embodiment does not impose any special restrictions on this.

[0104] Counting the number of all acoustic emission events in each event set, recording it as a total number; calculating the ratio of the number of all acoustic emission events belonging to any detection sensor in each event set to the total number, recording it as a relative ratio;

[0105] Normalizing the product of the relative ratio and the dispersion as the noise interference degree of each event set corresponding to each frame signal of any detection sensor;

[0106] In this embodiment, the The next detection sensor The frame signal corresponds to Taking an event set as an example, the calculation formula for its noise interference degree is:

[0107]

[0108] in, For the The next detection sensor The frame signal corresponds to The noise interference degree of the event set, For the The frame signal corresponds to The event set belongs to The number of all acoustic emission events of a detection sensor, For the The frame signal corresponds to The number of all acoustic emission events in an event set, For the The next detection sensor The frame signal corresponds to The dispersion of the event set, is a normalization function. In this embodiment, the softmax function is used for normalization processing. The softmax function is a well-known technology and will not be described in detail here. As other implementation methods, the implementer can adopt other methods of the prior art, such as the sigmoid function, etc. This embodiment does not impose any special restrictions on this. For relative comparison.

[0109] It should be noted that the larger the relative ratio is, the The event set belongs to If the number of acoustic emission events of a detection sensor is large, the adjacent sensor and the The greater the difference in the signal characteristics of the same acoustic emission source for each detection sensor, the greater the difference in the signal characteristics of the same acoustic emission source. Ideally, for the same type of acoustic emission event, the first The number of acoustic emission events of the first detection sensor and the adjacent sensor should be close. If the number of acoustic emission events of a detection sensor is large, it means that The detection sensor is in the The more likely there is a larger noise aliasing situation under the event corresponding to the event set; the larger the dispersion, the more likely there is a larger noise aliasing situation under the event corresponding to the event set; The lower the correlation between similar acoustic emission events of different detection sensors, the more likely there is to be greater noise aliasing, and the greater the resulting noise interference.

[0110] Secondly, based on the noise interference level, the event set is screened, specifically:

[0111] Obtaining a segmentation threshold of the noise interference degree of all event sets corresponding to each frame signal of any detection sensor, recorded as a first segmentation threshold;

[0112] In this embodiment, the maximum inter-class variance method is used to obtain the segmentation threshold, wherein the maximum inter-class variance method is a well-known technology and will not be described here. As other implementation methods, the implementer can use the cross-validation method, etc., and this embodiment does not impose any special restrictions on this.

[0113] The event set in which the noise interference degree of each frame signal under any detection sensor is greater than the first segmentation threshold is recorded as a noise event set, otherwise it is recorded as a non-noise event set;

[0114] It should be noted that the more likely an acoustic emission event in the noise event set is to have greater noise aliasing, the greater the degree to which its signal is interfered with by noise.

[0115] At this point, all noise event sets and non-noise event sets corresponding to each frame signal of each detection sensor are obtained.

[0116] Step 4: Based on the number of noise event sets and the number of non-noise event sets and the noise interference degree, determine the sound source estimation value of each frame signal under each detection sensor; analyze the energy intensity of the signal corresponding to each acoustic emission event in each frame signal under the guard sensor, calculate the droplet estimation value, and combine it with the sound source estimation value to obtain the number of modal decomposition layers of each frame signal under each detection sensor.

[0117] Furthermore, based on the number of events in the noise event set and the non-noise event set, the sound source estimation value is determined, specifically:

[0118]

[0119] in, For the The next detection sensor The sound source estimation value of the frame signal, For the The next detection sensor The number of all non-noise event sets corresponding to the frame signal, For the The next detection sensor The number of all noise event sets corresponding to the frame signal, For the The next detection sensor The cumulative sum of the noise interference levels of all noise event sets corresponding to the frame signal, is the ceiling function.

[0120] It should be noted that for the non-noise event set, the acoustic emission event noise of the acoustic emission source is less mixed, corresponding to one bottom plate acoustic emission source; for the noise event set, the acoustic emission event noise of the acoustic emission source is more mixed, and there may be more noise interference sources. In order to avoid insufficient variational mode decomposition, the estimated value of the sound source at the bottom plate of the container should be larger, so that the number of decomposed modes in the subsequent modal decomposition is more reasonable.

[0121] Secondly, when testing for leaks in containers, when the liquid phase medium is highly volatile, a gas phase region forms at the top of the container, which condenses on the inner wall of the container top. As the condensation point continues to increase, a droplet effect forms, which drips back onto the liquid phase medium. The signal energy generated by the falling droplet is high, which has a significant impact on acoustic emission detection. Once these noise signals are received by the detection sensor, they can easily lead to misidentification of the container bottom plate leak, seriously affecting the accuracy of the bottom plate inspection results. These noise signals unrelated to the container bottom plate need to be eliminated.

[0122] Therefore, the energy significance of the signal corresponding to each acoustic emission event in each frame signal under the guard sensor is analyzed, and the droplet estimation value is calculated, specifically:

[0123] According to the occurrence period of each acoustic emission event in each frame signal of each detection sensor, the signal corresponding to the same frame signal of the guard sensor directly above in the same occurrence period is recorded as the signal corresponding to each acoustic emission event in each frame signal of the guard sensor;

[0124] Calculate the energy of all signal amplitudes corresponding to each acoustic emission event in each frame of signal under the guard sensor directly above each detection sensor, and record it as the energy eigenvalue;

[0125] Obtain the segmentation threshold of the energy characteristic values corresponding to all acoustic emission events in each frame signal of the guard sensor, recorded as the second segmentation threshold;

[0126] In this embodiment, the maximum inter-class variance method is used to obtain the segmentation threshold, wherein the maximum inter-class variance method is a well-known technology and will not be described here. As other implementation methods, the implementer can use the cross-validation method, etc., and this embodiment does not impose any special restrictions on this.

[0127] Count the number of acoustic emission events in each frame of the signal under the guard sensor directly above each detection sensor, whose energy characteristic value is greater than the second segmentation threshold, as the droplet estimation value of each frame of the signal under the guard sensor;

[0128] It should be noted that the larger the energy characteristic value, the more likely the acoustic emission event is a high-energy signal generated by a falling droplet. The droplet estimation value reflects the number of high-energy significant signals that appear in each frame of signal when a droplet falls from the top of the container. These high-energy signals are likely to be received by the detection sensor below and cause interference to the detection sensor below.

[0129] Furthermore, based on the sound source estimation value and the droplet estimation value, the modal decomposition layer number is obtained, specifically:

[0130] The sum of the sound source estimation value of each frame signal under each detection sensor and the droplet estimation value of the same frame signal under the guard sensor directly above is used as the modal decomposition layer number of each frame signal under each detection sensor;

[0131] It should be noted that the energy of the signal corresponding to the acoustic emission event whose energy eigenvalue is greater than the second segmentation threshold is higher, and these high-energy signals are likely to be received by the detection sensor below. The larger the number of modal decomposition layers, the more acoustic emission sources there are on the bottom plate and top of the container. In order to avoid mixing the acoustic emission signal modes between the acoustic emission sources or omitting effective modes to cause under-decomposition, when decomposing the acoustic emission signal of the detection sensor by the variational modal decomposition algorithm, the larger the number of decomposition layers of the variational modal decomposition algorithm should be. The step flow chart of the method for obtaining the number of modal decomposition layers provided in the embodiment of the present application is as follows. Figure 3 shown.

[0132] At this point, the modal decomposition layer number of each frame signal under each detection sensor is obtained.

[0133] In step 5, based on the modal decomposition layer number, modal decomposition is performed on each frame signal of each detection sensor to obtain all modal components; the modal components corresponding to non-noise are obtained through the correlation between each modal component and the remaining modal components, and the signals of all modal components corresponding to non-noise are reconstructed. The leakage position of the pressure vessel is located through the reconstructed acoustic emission signal.

[0134] The modal decomposition number is used as the decomposition number of the modal decomposition algorithm, and a plurality of modal components are obtained by performing modal decomposition on each frame signal of each detection sensor; wherein the number of modal components is the same as the number of the modal decomposition layers;

[0135] In this embodiment, a variational modal decomposition algorithm is used to perform modal decomposition, wherein the variational modal decomposition algorithm is a well-known technology and will not be described in detail here.

[0136] Secondly, based on the characteristics that the correlation between the acoustic emission signals received by the detection sensor of the same acoustic emission source is strong, and the correlation between the noise is weak, the correlation between each modal component and all other modal components is analyzed, the modal components are screened, and the modes corresponding to the noise are eliminated. Specifically:

[0137] Calculate the correlation between each modal component and the other modal components;

[0138] In this embodiment, the degree of correlation is determined by calculating the inverse of the DTW distance between each modal component and the other modal components. As other implementation methods, the implementer may adopt other methods of the prior art, such as cosine similarity, etc. This embodiment does not impose any special restrictions on this.

[0139] Counting the number of correlations between each modal component and all other modal components that are less than a preset threshold, and recording this as a judgment coefficient;

[0140] Selecting the modal components whose judgment coefficients are less than a preset number threshold and recording them as the modal components corresponding to non-noise;

[0141] In this embodiment, the preset threshold value is 1, and the preset number threshold value is 5. As other implementation methods, the implementer can set it according to actual conditions.

[0142] Reconstruct all modal components corresponding to the non-noise of each frame signal under each detection sensor to obtain the reconstructed frame signal, and splice all the reconstructed frame signals under each detection sensor to obtain the reconstructed acoustic emission signal of each detection sensor, wherein the reconstructed acoustic emission signal eliminates the interference signal of the noise. Therefore, the reconstructed acoustic emission signal is the acoustic emission signal after filtering;

[0143] It should be noted that the signal reconstruction process is a well-known technology and will not be described in detail here.

[0144] According to the reconstructed acoustic emission signal, the time delay is estimated by the improved quadratic correlation algorithm, and the leakage position of the container bottom plate is located by the overdetermined positioning algorithm based on the sequential quadratic programming algorithm to complete the leak detection of the container.

[0145] It should be noted that the improved quadratic correlation algorithm and the overdetermined positioning algorithm based on the sequential quadratic programming algorithm are well-known technologies and will not be described in detail here.

[0146] An embodiment of the present application also provides a leak detection device for pressure vessels, comprising a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the steps of any one of the above-mentioned leak detection methods for pressure vessels are implemented.

[0147] Based on the same inventive concept as the above method, an embodiment of the present application further provides a leak detection system for a pressure vessel, the system comprising:

[0148] An acoustic emission sensor array is used to deploy acoustic emission sensors on the outer wall of the pressure vessel;

[0149] The signal conditioning module is used to pre-process the collected signals, specifically:

[0150] Acoustic emission sensors lack built-in signal conditioning circuits and output weak, high-impedance charge signals, making them difficult to collect. By connecting a signal conditioning module to the acoustic emission sensor array, it is used to amplify and process the weak electrical signals collected by the acoustic emission sensors, improving the signal-to-noise ratio for subsequent analysis.

[0151] In this embodiment, AD8606 is selected as the charge amplifier chip, and a compensation circuit composed of capacitor C1 is connected to the operational amplifier to eliminate self-oscillation. A second-order bandpass filter circuit is used for filtering. The bandwidth B of the bandpass filter is set to 10 and the center frequency is 140 kHz. As other implementation methods, the implementer can set them according to actual conditions.

[0152] A signal acquisition module is used to obtain the acoustic emission signals of each detection sensor on the outer wall of the pressure vessel and the guard sensor directly above it, as well as the liquid level height at each moment;

[0153] The signal transmission module is connected to the signal acquisition module and supports uploading real-time signals to the host computer through the USB2.0 high-speed transmission protocol;

[0154] The signal analysis module is used to process the received acoustic emission signals to detect the location of the container leakage. It specifically includes the following submodules:

[0155] The filtering processing submodule is used to obtain the reconstructed acoustic emission signal, including: framing the acoustic emission signal, extracting the upper envelope of each frame signal under each detection sensor; judging the acoustic emission events that meet the acoustic emission event judgment conditions according to the deviation of the liquid level height at all times, the size of the signal amplitude in the upper envelope and the distribution of extreme points; analyzing the energy intensity of the signal corresponding to each acoustic emission event under each frame signal of each detection sensor and the duration of the event, combining the frequency distribution of the signal in the frequency domain to form a feature vector, classifying the acoustic emission events of the same frame signal under each detection sensor and its adjacent detection sensors, and obtaining each event set corresponding to each frame signal under each detection sensor; and classifying the acoustic emission events of the same frame signal under each detection sensor by comparing the different acoustic emission events in each event set. The difference in the characteristic vectors between the two and the number of acoustic emission events is used to determine the noise interference degree of each event set, and the noise event set and the non-noise event set are obtained; based on the number of noise event sets and the number of non-noise event sets and the noise interference degree, the sound source estimation value of each frame signal under each detection sensor is determined; the energy intensity of the signal corresponding to each acoustic emission event in each frame signal under the guard sensor is analyzed, the droplet estimation value is calculated, and combined with the sound source estimation value, the modal decomposition layer number of each frame signal under each detection sensor is obtained, and all modal components are obtained by performing modal decomposition on each frame signal under each detection sensor; the modal component corresponding to the non-noise is obtained through the correlation between each modal component and the other modal components, and the signal of all modal components corresponding to the non-noise is reconstructed;

[0156] The time delay estimation submodule is used to estimate the time delay of the reconstructed acoustic emission signal;

[0157] In this embodiment, an improved quadratic correlation algorithm is used to perform time delay estimation on the acoustic emission signal processed by the filtering processing submodule, thereby solving the problems of poor noise immunity, large computational complexity, and low estimation accuracy for nonlinear and non-periodic signals in acoustic emission source positioning.

[0158] The positioning algorithm submodule is used to locate the leakage position of the container bottom plate through the reconstructed acoustic emission signal and the result of time delay estimation.

[0159] In this embodiment, the leakage position of the container bottom plate is located by using the acoustic emission signal processed by the filtering processing submodule and the result of time delay estimation and an overdetermined positioning algorithm based on a sequential quadratic programming algorithm.

[0160] It should be noted that the improved quadratic correlation algorithm and the overdetermined positioning algorithm based on the sequential quadratic programming algorithm are well-known technologies and will not be described in detail here.

[0161] Among them, the block diagram of a leak detection system for pressure vessels provided in the embodiment of the present application is as follows: Figure 4 shown.

[0162] It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. In addition, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0163] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification. The above embodiments only express several implementation methods of the present application. The description is relatively specific and detailed, but it cannot be understood as a limitation of the present application. It should be pointed out that for ordinary technicians in this field, without departing from the concept of the present application, several variations and improvements can be made. Therefore, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present application without departing from the content of the technical solution of the present application shall fall within the scope of protection of the technical solution of the present application.

Claims

1. A leak detection method for a pressure vessel, characterized in that: The method comprises the following steps: Acquire the acoustic emission signals of each detection sensor on the outer wall of the pressure vessel and the guard sensor directly above it, as well as the liquid level height at each moment; The acoustic emission signal is divided into frames, and the upper envelope of each frame signal under each detection sensor is extracted. The occurrence of acoustic emission events is judged based on the deviation of the liquid level height at all times, the size of the signal amplitude in the upper envelope, and the distribution of extreme points. Analyze the energy intensity of the signal corresponding to each acoustic emission event under each frame signal of each detection sensor and the duration of the event, combine the frequency distribution of the signal in the frequency domain, form a feature vector, classify the acoustic emission events of the same frame signal under each detection sensor and its adjacent detection sensors, and obtain each event set corresponding to each frame signal under each detection sensor; determine the noise interference degree of each event set based on the difference in the feature vectors between different acoustic emission events in each event set and the number of acoustic emission events, and divide the event set into a noise event set and a non-noise event set; Based on the number of noise event sets and the number of non-noise event sets and the noise interference degree, the sound source estimation value of each frame signal under each detection sensor is determined; the energy intensity of the signal corresponding to each acoustic emission event in each frame signal under the guard sensor is analyzed, and the droplet estimation value is calculated. Combined with the sound source estimation value, the number of modal decomposition layers of each frame signal under each detection sensor is obtained, and each frame signal under each detection sensor is modally decomposed to obtain all modal components; the modal component corresponding to the non-noise is obtained through the correlation between each modal component and the remaining modal components, and the leakage position of the pressure vessel is located after signal reconstruction of all modal components corresponding to the non-noise.

2. A leak detection method for a pressure vessel according to claim 1, characterized in that: The judging of the acoustic emission event includes: The ratio of the average liquid level height at all times to the preset rated operating height is recorded as the liquid level ratio; Effective threshold The calculation formula is: ,in, is the preset reference value, is the liquid level ratio, is the preset reference ratio, is an exponential function with a natural constant as its base; Obtain the maximum and minimum points of the upper envelope; extend backward from any maximum point as the starting point, and record the moment when the first signal amplitude after the starting point in the upper envelope is less than the effective threshold value and the first minimum point after the starting point as critical points; and record the critical point closest to the starting point as the ending point; For each frame of signal from each detection sensor, the period between the starting point and the ending point corresponding to each maximum point is recorded as an acoustic emission event.

3. A leak detection method for a pressure vessel according to claim 2, characterized in that: The method for obtaining the feature vector is: For each acoustic emission event corresponding signal in each frame signal of each detection sensor, calculate the duration of each acoustic emission event corresponding signal and record it as duration; Calculating the product of the total energy of all signal amplitudes in the signal corresponding to each acoustic emission event and the duration as the evaluation value of each acoustic emission event; Perform frequency domain analysis on the signal corresponding to each acoustic emission event to obtain a spectrum. The frequency corresponding to the maximum energy in the spectrum is recorded as the main frequency; the frequency corresponding to the central value after integrating all the energy in the spectrum is recorded as the center frequency. Count the ratio of the number of signal amplitudes less than the effective threshold value in the signal corresponding to each acoustic emission event to the duration, and record it as the average frequency; Normalization is performed on the main frequency, the center frequency, the average frequency, and the evaluation value respectively, and a feature vector is formed.

4. A leak detection method for a pressure vessel according to claim 1, characterized in that: The obtaining of each event set corresponding to each frame signal of each detection sensor includes: clustering the feature vectors corresponding to all acoustic emission events of the same frame signal of each detection sensor and all adjacent detection sensors to obtain multiple clusters, and recording all acoustic emission events belonging to the same cluster as each event set corresponding to each frame signal of each detection sensor.

5. A leak detection method for a pressure vessel according to claim 1, characterized in that: The step of determining the noise interference degree of each event set and dividing the event set into a noise event set and a non-noise event set includes: For each event set corresponding to each frame signal of any detection sensor, the average of the distances between the feature vectors of all two acoustic emission events belonging to any detection sensor in each event set is recorded as the dispersion of each event set; Counting the number of all acoustic emission events in each event set, recording it as a total number; calculating the ratio of the number of all acoustic emission events belonging to any detection sensor in each event set to the total number, recording it as a relative ratio; The noise interference degree is a normalized result of the product of the relative ratio and the dispersion; Obtain a segmentation threshold of the noise interference degree of all event sets corresponding to each frame signal of any detection sensor, and record it as a first segmentation threshold; record the event set whose noise interference degree is greater than the first segmentation threshold as a noise event set, otherwise, record it as a non-noise event set.

6. A leak detection method for a pressure vessel according to claim 1, characterized in that: No. The next detection sensor Sound source estimation value of frame signal The calculation formula is: ,in, For the The next detection sensor The number of all non-noise event sets corresponding to the frame signal, For the The detection sensor is in the The number of all noise event sets corresponding to the frame signal, For the The next detection sensor The cumulative sum of the noise interference levels of all noise event sets corresponding to the frame signal, is the ceiling function.

7. A leak detection method for a pressure vessel according to claim 1, characterized in that: The droplet estimation value is calculated and combined with the sound source estimation value to obtain the modal decomposition layer number of each frame signal under each detection sensor, including: Based on the occurrence period of each acoustic emission event in each frame signal of each detection sensor, the total energy of all signal amplitudes of the signal corresponding to the occurrence period in the same frame signal of the guard sensor directly above is recorded as the energy characteristic value; Obtain the segmentation threshold of the energy characteristic values corresponding to all acoustic emission events in each frame of the guard sensor signal, recorded as the second segmentation threshold; count the number of acoustic emission events with energy characteristic values greater than the second segmentation threshold, as the droplet estimation value of each frame of the guard sensor signal; The modal decomposition layer number is the sum of the sound source estimation value of each frame signal under each detection sensor and the droplet estimation value of the same frame signal of the guard sensor directly above.

8. The leak detection method for a pressure vessel according to claim 1, characterized in that: The process of obtaining the modal components corresponding to the non-noise is as follows: calculating the correlation between each modal component and the remaining modal components; Counting the number of times the correlation between each modal component and all other modal components is less than a preset threshold, and recording this as a judgment coefficient of each modal component; The modal components whose judgment coefficients are less than a preset number threshold are selected and recorded as the modal components corresponding to non-noise.

9. A leak detection device for a pressure vessel, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the leak detection method for a pressure vessel as described in any one of claims 1 to 8 are implemented.

10. A leak detection system for a pressure vessel, applied to the leak detection method for a pressure vessel according to claim 1, wherein the system comprises a signal analysis module, wherein the submodules of the module include: The filtering processing submodule is used to obtain the reconstructed acoustic emission signal, including: framing the acoustic emission signal and extracting the upper envelope of each frame signal under each detection sensor; judging the acoustic emission events that meet the acoustic emission event judgment conditions based on the deviation of the liquid level height at all times, the size of the signal amplitude in the upper envelope and the distribution of extreme points; analyzing the energy intensity of the signal corresponding to each acoustic emission event under each frame signal of each detection sensor and the duration of the event, combined with the frequency distribution of the signal in the frequency domain, Composing feature vectors, classifying the acoustic emission events of the same frame signal under each detection sensor and its adjacent detection sensors, and obtaining each event set corresponding to each frame signal under each detection sensor; determining the noise interference degree of each event set based on the difference in feature vectors between different acoustic emission events in each event set and the number of acoustic emission events, and obtaining a noise event set and a non-noise event set; Based on the number of noise event sets and the number of non-noise event sets and the noise interference degree, the sound source estimation value of each frame signal under each detection sensor is determined; the energy intensity of the signal corresponding to each acoustic emission event in each frame signal under the guard sensor is analyzed, and the droplet estimation value is calculated. Combined with the sound source estimation value, the modal decomposition layer number of each frame signal under each detection sensor is obtained. By performing modal decomposition on each frame signal under each detection sensor, all modal components are obtained; The modal component corresponding to the non-noise is obtained through the correlation between each modal component and the other modal components, and the signal of all modal components corresponding to the non-noise is reconstructed; The time delay estimation submodule is used to estimate the time delay of the reconstructed acoustic emission signal; The positioning algorithm submodule is used to locate the leakage position of the container bottom plate through the reconstructed acoustic emission signal and the result of time delay estimation.

Images