An artificial intelligence-based hermeticity monitoring system for a seal tester

Abstract

The application relates to the technical field of air tightness testing, and particularly discloses a sealing test machine air tightness monitoring system based on artificial intelligence, which comprises the following modules: a sensor module that collects air pressure change parameters in real time through a pressure sensor; an image processing module that collects bubble dynamic image information by using a camera and obtains the inner surface contour image of a sealing cavity; a model construction module that inputs historical air pressure change parameters and historical bubble dynamic image information into a preset convolutional neural network model for training, constructs a prediction model, and obtains bubble motion parameters; a data analysis module that inputs the real-time obtained air pressure change parameters and bubble motion parameters into the prediction model for tracking detection, obtains air tightness prediction parameters, executes an evaluation strategy based on the air tightness prediction parameters, judges whether the sealing test machine air tightness has problems according to an evaluation result, and analyzes surface defects of the inner surface contour image of the sealing cavity of the sealing space; and a display module that displays an alarm result.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of air tightness testing, in particular to an air tightness monitoring system of a sealing test machine based on artificial intelligence. BACKGROUND

[0002] In industrial production, sealing test is an important link to ensure product quality, especially for products that require high air tightness, such as electronic products, automobile parts, etc. Traditional air tightness monitoring methods usually rely on manual observation or simple pressure sensors, which are not only inefficient, but also difficult to ensure accuracy.

[0003] With the development of artificial intelligence technology, an air tightness monitoring system of a sealing test machine based on artificial intelligence has emerged; this system usually combines computer vision, machine learning and deep learning technologies, and can automatically identify and analyze air tightness problems in the sealing test process. For example, through image analysis of the air bubbles generated during the sealing test process, the location and degree of air leakage can be accurately detected. In addition, based on artificial intelligence, the system can continuously optimize the detection algorithm through learning from a large amount of test data, improving the accuracy and efficiency of detection.

[0004] Common air tightness monitoring systems of sealing test machines usually test the cavity tightness of the test object, and determine whether the equipment sealing of the test machine will leak by directly observing the state of the connected air bubbles, and then determining the location and size of the leak; however, only the analysis of the state and number of air bubbles during the air tightness product detection process cannot accurately test the specific situation of the test object leakage, and the existing dry air tightness detection system combined with machine vision tracking algorithm and combined with the geometric model and dynamic model of the air bubble to obtain the leakage amount of the workpiece, but cannot combine the pressure state and air bubble state of the sealing test machine to determine the sealing performance of the test machine and the specific leakage degree. SUMMARY

[0005] The purpose of the present application is to provide an air tightness monitoring system of a sealing test machine based on artificial intelligence, which solves the following technical problems:

[0006] How to combine the analysis of pressure state and tracking of air bubble state based on artificial intelligence technology to accurately determine the sealing performance of the sealing test machine.

[0007] The purpose of the present application can be achieved by the following technical solutions:

[0008] An air tightness monitoring system of a sealing test machine based on artificial intelligence, the sealing test machine comprising a test sealing space and a water box; the test sealing space comprising a sealing cavity and a device under test; the system comprising:

[0009] A sensor module is configured to collect, in real time, a pressure change parameter of the inflatable device in the test sealed space through a pressure sensor;

[0010] An image processing module is configured to collect, by using a camera, bubble dynamic image information in a water box connected to the test sealed space through an air pipe; and obtain a sealed cavity inner surface profile image through image processing technology;

[0011] A model construction module is configured to input historical pressure change parameters and historical bubble dynamic image information into a preset convolutional neural network model for training, construct a prediction model, and obtain a bubble motion parameter;

[0012] A data analysis module is configured to input the real-time obtained pressure change parameter and bubble motion parameter into the prediction model for tracking detection, and obtain a gas tightness prediction parameter;

[0013] and execute an evaluation strategy based on the gas tightness prediction parameter, and determine whether there is a problem with the gas tightness of the sealing test machine according to the evaluation result:

[0014] If yes, it is determined that the gas tightness of the current sealing test machine is unqualified, an alarm signal is generated, and surface defect analysis is performed on the sealed cavity inner surface profile image of the sealed space;

[0015] If no, it is determined that the gas tightness of the current sealing test machine is qualified, and the gas tightness monitoring is continued;

[0016] A display module is configured to display the alarm result.

[0017] Preferably, the gas pressure change parameter is obtained in the following manner:

[0018] S1, obtaining a first air pressure value at the air inlet of the sealed space;

[0019] S2, obtaining a second air pressure value in the sealed space after continuous inflation for a set time period;

[0020] S3, calculating the difference between the first air pressure value and the second air pressure value as the air pressure difference value in the set time period;

[0021] S4, obtaining a curve of continuous time change of the air pressure difference value in a preset time period, and calculating the slope of the curve to determine the air pressure change parameter.

[0022] Preferably, the bubble motion parameter is obtained in the following manner:

[0023] SS1, determining a time interval between adjacent frames of bubble dynamic images in advance;

[0024] SS2, obtaining a bubble number change in the continuous frames of bubble dynamic images in a preset time period and a bubble displacement distance change corresponding to all time intervals.

[0025] SS3, obtaining the bubble motion speed according to the bubble number change and the bubble displacement distance change;

[0026] SS4, dividing the motion rhythm in each time interval as the bubble motion parameter according to the bubble motion speed.

[0027] Preferably, the data analysis module inputs the real-time obtained air pressure change parameter and the bubble motion parameter into the prediction model, and the specific process of the tracking detection includes:

[0028] normalizing the coding of the air tightness related features;

[0029] the prediction model calculates the loss function between the prediction value of the air tightness related features and the target value thereof;

[0030] using a stochastic gradient descent algorithm to automatically adjust the model parameters of the prediction model to minimize the loss function;

[0031] after multiple iterations until the prediction model converges, outputting the air tightness prediction parameter.

[0032] Preferably, the calculation method of the air tightness prediction parameter is:

[0033]

[0034] wherein, is the air tightness prediction parameter of the i-th test sample; is a preset function; is a preset first conversion function, is a second preset conversion function; is the air pressure change parameter of the i-th test sample; is a standard air pressure change parameter; is a preset deviation value of the air pressure change parameter of the i-th test sample; is the bubble motion parameter of the i-th test sample; is a standard bubble motion parameter; is a preset deviation value of the bubble motion parameter of the i-th test sample. Preferably, the air tightness prediction parameter is compared with a preset standard air tightness prediction parameter allowed threshold

[0035] Preferably, the air tightness prediction parameter is compared with a preset standard air tightness prediction parameter allowed threshold

[0036] If > ​​​​, it is determined that the current tested equipment has a leakage risk and generates an early warning message;

[0037] like ≤ , it is determined that the current device under test is normal.

[0038] Preferably, the process of the data analysis module executing the evaluation strategy according to the airtightness prediction parameters includes:

[0039] By formula Calculate the evaluation coefficient ;

[0040] in, is the total number of test samples, and ∈ ; For the The preset weight coefficient of each test sample; is the standard airtightness prediction parameter, and ≤ .

[0041] Preferably, judging whether there is an airtightness problem of the sealing test machine based on the evaluation results specifically includes:

[0042] The evaluation coefficient and preset evaluation coefficient threshold To compare:

[0043] like < , it is determined that there is an airtightness problem with the sealing test machine and an alarm signal is generated;

[0044] like ≥ , it is judged that there is no airtightness problem in the sealing test machine.

[0045] Preferably, the process of performing surface defect analysis on the inner surface contour image of the sealed cavity of the sealed space is:

[0046] Identify the inner surface contour image of the sealed cavity and perform grayscale processing;

[0047] The Canny algorithm is used to extract the contour of abnormal areas in the inner surface contour image;

[0048] Calculate the area S and perimeter C of the abnormal area contour by the formula Calculate the abnormal area structure value .

[0049] Preferably, it also includes:

[0050] The abnormal area structure value With preset threshold and compared, wherein, < :

[0051] If ≤ , it is judged that the cavity surface of the measured device in the sealed space has a slight leakage;

[0052] If < < , it is judged that the cavity surface of the measured device in the sealed space has a moderate leakage;

[0053] If ≥ , it is judged that the cavity surface of the measured device in the sealed space has a serious leakage.

[0054] Advantages of the present application:

[0055] (1) The present application realizes the acquisition of real-time air pressure change parameter data through the pressure sensor by setting the sensor module, realizes the acquisition of real-time bubble movement image in the water box connected through the air pipe on one side of the sealed space through the image processing module, acquires bubble dynamic image information by using the camera, ensures the acquisition of continuous bubble movement state, and also acquires the inner surface contour image of the sealed cavity through image processing technology, realizes the sealing degree of the sealed cavity of the sealed measured device through the inner surface contour image of the sealed cavity, realizes the construction model of the operation and processing process of machine learning and computer vision in artificial intelligence through the setting of the model construction module, inputs the historical air pressure change parameter and historical bubble dynamic information image into the preset convolutional neural network model for training to construct a prediction model, acquires the bubble movement parameter in the sealing process through machine training, and helps to judge and early warning process the air tightness of the sealing test machine through the bubble movement parameter.

[0056] (2) The present application realizes the tracking detection of air pressure change parameter and bubble movement parameter through the setting of the data analysis module, and further acquires the air tightness prediction parameter to realize the air tightness prediction of the sealing test machine, and outputs the air tightness prediction parameter according to the tracking detection; and the data analysis module also designs an evaluation strategy based on the air tightness prediction parameter, acquires an evaluation result through the execution of the evaluation strategy to judge whether there is a problem with the air tightness of the sealing test machine, generates a warning for the case that there is an air tightness problem and analyzes the surface defects of the current inner surface contour image of the sealed cavity, realizes the combination of the pressure state and the bubble state of the sealing test machine through the data analysis module, and judges the specific leakage degree of the sealing performance of the test machine according to the evaluation result.

[0057] Of course, implementing any product of the application does not necessarily require achieving all the advantages described above at the same time. BRIEF DESCRIPTION OF DRAWINGS

[0058] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for the description of the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0059] Figure 1 A module diagram of the air tightness monitoring system of the sealing test machine based on artificial intelligence of the present application is shown in the figure.

[0060] Figure 2 A step diagram of the air pressure change parameter acquisition method of the present application is shown in the figure.

[0061] Figure 3 A step diagram of the bubble movement parameter acquisition method of the present application is shown in the figure. DETAILED DESCRIPTION

[0062] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0063] The common air tightness monitoring system of the sealing test machine usually detects the cavity tightness of the test object, judges whether the equipment sealing of the test machine will appear leakage by directly observing the state of the connected bubble, and further determines the position and size of the leakage. However, only the analysis of the bubble state and quantity is carried out in the product detection process of air tightness, which cannot accurately test the specific situation of the leakage of the test object. The existing leakage amount of the workpiece is obtained by combining the tracking algorithm of the dry air tightness detection system with machine vision, and combining the geometric model and dynamic model of the bubble, but the specific leakage degree of the sealing performance of the test machine cannot be judged by combining the pressure state and bubble state of the sealing test machine.

[0064] Referring to Figure 1 To solve the above technical problems, the present application designs an air tightness monitoring system of a sealing test machine based on artificial intelligence. The sealing test machine includes a test sealing space and a water box. The test sealing space includes a sealing cavity and a device under test. The sealing test machine comprises:

[0065] The sensor module is used to collect the air pressure change parameter of the device under test after inflation in the test sealing space in real time through the pressure sensor.

[0066] An image processing module is configured to collect bubble dynamic image information in the water box connected to the test sealed space through the air pipe by using the camera; and to obtain a sealed cavity inner surface profile image by using image processing technology;

[0067] A model construction module is configured to input historical air pressure change parameters and historical bubble dynamic information images into a preset convolutional neural network model for training, to construct a prediction model, and to obtain bubble motion parameters;

[0068] A data analysis module is configured to input real-time obtained air pressure change parameters and bubble motion parameters into the prediction model for tracking detection, to obtain air tightness prediction parameters;

[0069] and to perform an evaluation strategy based on the air tightness prediction parameters, to determine whether there is a problem with the air tightness of the sealed tester according to the evaluation result:

[0070] If yes, it is determined that the air tightness in the current sealed tester is unqualified, an alarm signal is generated, and surface defect analysis is performed on the sealed cavity inner surface profile image of the sealed space;

[0071] If no, it is determined that the air tightness in the current sealed tester is qualified, and air tightness monitoring is continued;

[0072] A display module is configured to display the alarm result.

[0073] In the above technical solution, the air tightness monitoring process is realized based on the cavity sealing detection technology, the test sealed space, the water box, and the sealing monitoring mode in which the test sealed space and the water box are connected are set; wherein the test sealed space includes a sealed cavity and a device under test, and the device under test is packaged by a sealing wall; the sealing tester air tightness monitoring system based on artificial intelligence specifically includes a sensor module, an image processing module, a model construction module, a data analysis module, and a display module to realize accurate judgment of the sealing performance and leakage degree of the sealed tester by combining artificial intelligence technology with analysis of pressure state and tracking of bubble state.

[0074] The sensor module is arranged to obtain the data of the real-time air pressure change parameter through the pressure sensor, the real-time air pressure value is obtained through the pressure sensor arranged in the sealed space, the real-time pressure value in the continuous inflation process is analyzed through the data processing technology arranged in the pressure sensor, and then the air pressure change parameter in the continuous inflation process is obtained, the image processing module is arranged to obtain the real-time image of the real-time bubble movement in the water box connected with the air pipe on one side of the sealed space, the camera is used to collect the bubble dynamic image information, so that the continuous movement state of the bubble is obtained, and the image processing technology is used to obtain the inner surface contour image of the sealed cavity, and the sealing degree of the sealed cavity for the measured equipment is realized through the inner surface contour image of the sealed cavity; the model construction module is arranged to construct a model based on the operation processing process of machine learning and computer vision in artificial intelligence, and the construction of the model is based on the image data processing process in the convolutional neural network model, the historical air pressure change parameter and the historical bubble dynamic information image are input into the preset convolutional neural network model for training, a prediction model is constructed, the bubble movement parameter in the sealing process is obtained through machine training, and the air tightness of the sealing test machine is judged and early warning processing is performed through the bubble movement parameter.

[0075] The data analysis module is arranged to track and detect the air pressure change parameter and the bubble movement parameter, and then obtain the air tightness prediction parameter to realize the air tightness prediction of the sealing test machine. Specifically, the real-time obtained air pressure change parameter and bubble movement parameter are input into the prediction model for tracking and detection, and the air tightness prediction parameter is output according to the tracking and detection; and the data analysis module also designs an evaluation strategy based on the air tightness prediction parameter, obtains an evaluation result by executing the evaluation strategy to judge whether there is a problem with the air tightness of the sealing test machine, generates an early warning for the case that there is an air tightness problem and performs surface defect analysis on the inner surface contour image of the current sealed cavity, and continues to monitor the case that there is no air tightness problem. The data analysis module is arranged to realize the combination of the pressure state and the bubble state of the sealing test machine, and the evaluation result is used to judge the specific leakage degree of the sealing performance of the test machine; the display module is arranged to display the early warning result in time, so that the leakage information is processed through corresponding measures.

[0076] As an embodiment of the present application, please refer to Figure 2 The air pressure change parameter is obtained as follows:

[0077] S1, obtaining a first air pressure value at the inflation pipe opening in the sealed space;

[0078] S2, obtaining a second air pressure value in the sealed space after continuous inflation reaches a set time length;

[0079] S3, calculate the difference between the first air pressure value and the second air pressure value as the air pressure difference value in the set time length;

[0080] S4, obtain a curve of continuous time change of the air pressure difference value in the preset time period, and determine the air pressure change parameter by slope calculation on the curve.

[0081] In the above technical solution, the air pressure change in the inflation state is monitored through the air pressure change parameter. Specifically, the initial state of air pressure filling is first judged by the first air pressure value, and the air pressure of the initial state is set as the first air pressure value. Then, the air pressure value in the sealed space after the set inflation time is reached is taken as the second air pressure value. Next, the difference between the first air pressure value and the second air pressure value is obtained as the air pressure difference value in the set time length. Finally, a curve of continuous time change of the air pressure difference value is generated according to the real-time change of the air pressure difference value in the preset time period of the closed detection, and the air pressure change parameter is determined by slope calculation on the curve.

[0082]

[0083] is the slope value of the i-th time point.

[0084] As an embodiment of the present application, please refer to FIG. 1, and the obtaining step of the bubble movement parameter is as follows: Figure 3

[0085] SS1, determine the time interval between adjacent frames of bubble dynamic images in advance;

[0086] SS2, obtain the change of the number of bubbles and the change of the bubble displacement distance corresponding to all time intervals in the continuous frames of bubble dynamic images in the preset time period;

[0087] SS3, obtain the bubble movement speed according to the change of the number of bubbles and the change of the bubble displacement distance;

[0088] SS4, divide the movement rhythm in each time interval as the bubble movement parameter according to the bubble movement speed.

[0089] ​​​​​​​​​In the technical solution, the bubble movement parameter is acquired to ensure that the state of the bubbles overflowing from the sealed cavity space during the sealing test of the device under test is analyzed in data, and the next tracking detection is judged according to the size of the bubble movement parameter. The specific process is as follows: first, the dynamic bubble image acquired by the camera is divided into multiple images, and the time interval between adjacent frames of dynamic bubble images is determined in advance; then, the bubble quantity change in the dynamic bubble image of the continuous frame in the preset time period and the bubble displacement distance change obtained according to the time interval are acquired; then, the bubble movement speed is calculated according to the bubble quantity change and the bubble displacement distance change. Specifically, the method of image tracking is used to record the displacement y of the bubble along the axis in each frame from the generation to the approach to the liquid surface. Meanwhile, the frame rate of the camera, i.e., the time interval of two frames of images, is determined in advance. The curve of the displacement y of the bubble along the Y-axis direction with respect to time t can obviously be represented by a quadratic polynomial.

[0090] In the design, the least square method is used to fit the movement trajectory of the bubble along the Y-axis direction, for example, polynomial fitting or spline fitting is used, and then the derivative of the fitting function is obtained to obtain the instantaneous speed. Specifically, according to the continuous data of the bubble movement trajectory, the instantaneous speed of a point on the trajectory can be obtained by calculating the tangent slope of the point. The bubble movement speed is acquired. Finally, the average speed of the bubble in each time interval is calculated. These average speeds can be used as the movement rhythm of the bubble, and the proportion value of each movement rhythm is obtained by numerical simulation according to the movement rhythm. The bubble movement parameter is obtained by fitting.

[0091] As an embodiment of the present application, the data analysis module inputs the real-time acquired air pressure change parameter and bubble movement parameter into the prediction model to perform the specific process of tracking detection, which includes:

[0092] The air tightness related features are encoded and normalized after being encoded;

[0093] The prediction model calculates the loss function between the prediction value of the air tightness related features and the target value thereof;

[0094] The random gradient descent algorithm is used to automatically adjust the model parameters of the prediction model to minimize the loss function;

[0095] After multiple iterations, the prediction model converges, and the air tightness prediction parameter is output.

[0096] In the above technical solution, the process of machine simulation of the obtained air pressure change parameters and bubble movement parameters by the data analysis module is mainly to input them into the trained model, that is, the prediction model, for further tracking and detection to obtain air tightness prediction parameters. The air tightness prediction parameters reflect the predicted state of the air tightness state in the sealing cavity during the monitoring process of the current sealing test. The specific air tightness prediction parameters are obtained through machine learning: first, the characteristics related to air tightness are determined, including the material properties of the container (such as permeability, thickness, etc.), the geometric parameters of the sealing structure (such as sealing width, gap size, etc.), and environmental factors (such as temperature, pressure, etc.). This design is optimized by obtaining the air pressure change parameters and bubble movement parameters as reference values ​​for air tightness characteristics; and the reference values ​​of the air tightness characteristics are encoded as features and then normalized to ensure the unification of subsequent data.

[0097] Then, the normalized airtightness-related feature data is input into the prediction model; the prediction model will output the predicted value of the airtightness-related feature; the difference between the predicted value and the target value is measured by the loss function, and the common loss function is the mean square error (MSE); the loss function is then minimized, and in each iteration, the stochastic gradient descent algorithm (SGD) updates the model parameters according to the gradient of the loss function to the model parameters; finally, the SGD algorithm is used multiple times to update the model parameters, and the loss function will gradually decrease as the number of iterations increases; when the value of the loss function no longer decreases significantly or reaches a pre-set convergence condition (such as the value of the loss function is less than a certain threshold, or the change of the loss function in several consecutive iterations is less than a certain value), it proves that the prediction model has converged and outputs the airtightness prediction parameters.

[0098] As an embodiment of the present invention, the calculation method of the airtightness prediction parameter is:

[0099]

[0100] in, For the Airtightness prediction parameters of test samples; is the preset function; is the preset first conversion function, is the second preset conversion function; For the Air pressure variation parameters of the test samples; is the standard air pressure variation parameter; For the The preset deviation value of the air pressure change parameter of each test sample; For the Bubble motion parameters of the test samples; is the standard bubble motion parameter; For the The bubble motion parameter preset deviation value of each test sample.

[0101] In the above technical solution, the calculation process of the airtightness parameter is through the formula Get the first Airtightness prediction parameters of test samples , and according to the airtightness prediction parameters The size of the equipment can be used to estimate the risk of leakage in advance. It is a prediction function set according to historical data to ensure that the airtightness prediction parameters are within a specific reasonable range; the first conversion function is preset , the second preset conversion function All of them are obtained by fitting the test data of the sealing test machine under normal operation. Therefore, by substituting the pressure change parameter range of different test samples and the range of variation of bubble motion parameters Then, the airtightness prediction parameters are obtained To determine whether there is a leak on the device under test; 、 These are standard data values ​​obtained based on historical experience and will not be detailed here.

[0102] As an embodiment of the present invention, the airtightness prediction parameter The threshold value of the airtightness prediction parameter is consistent with the preset standard To compare:

[0103] like > , it is determined that the current tested equipment has a leakage risk and generates an early warning message;

[0104] like ≤ , it is determined that the current device under test is normal.

[0105] In the above technical solution, the standard air tightness prediction parameters are obtained through the machine learning process, and the prediction of whether the equipment of the sealing tester has the risk of leakage is determined through comparative analysis, so as to carry out early warning processing.

[0106] As an embodiment of the present invention, the process of the data analysis module executing the evaluation strategy according to the airtightness prediction parameters includes:

[0107] By formula Calculate the evaluation coefficient ;

[0108] in, is the total number of test samples, and ∈ ; For the The preset weight coefficient of each test sample; is the standard airtightness prediction parameter, and ≤ .

[0109] In the above technical solution, by setting the evaluation strategy and executing the evaluation analysis process, the impact of the airtightness prediction parameter size on the airtightness problem is evaluated, and then the airtightness problem is determined according to the evaluation coefficient; specifically, the formula Calculate the evaluation coefficient ; Analyze the changes in the airtightness prediction parameters of the cumulative test samples, and determine the airtightness problem of the sealing test machine based on the evaluation coefficient obtained by calculation. Among them, the standard airtightness prediction parameter It is the standard data value obtained based on historical experience; the preset weight coefficient Those selected based on historical experience will not be described in detail here.

[0110] As an embodiment of the present invention, judging whether there is an airtightness problem of the sealing tester based on the evaluation result specifically includes:

[0111] The evaluation coefficient and preset evaluation coefficient threshold To compare:

[0112] like < , it is determined that there is an airtightness problem with the sealing test machine and an alarm signal is generated;

[0113] like ≥ , it is judged that there is no airtightness problem in the sealing test machine.

[0114] In the above technical solution, the evaluation results are judged by comparative analysis to determine if the evaluation coefficient Less than the evaluation coefficient threshold , it is considered that the sealing test machine has an airtightness problem and needs to be repaired. The inner surface contour image of the sealing cavity of the sealed space is identified to determine the structural defects. Otherwise, it is judged to be in normal state. Therefore, the sealing test machine does not have an airtightness problem.

[0115] As an embodiment of the present invention, the process of performing surface defect analysis on the inner surface contour image of the sealed cavity of the sealed space is as follows:

[0116] Identify the inner surface contour image of the sealed cavity and perform grayscale processing;

[0117] The Canny algorithm is used to extract the abnormal area contour in the inner surface contour image.

[0118] The area S and the perimeter C of the abnormal area contour are calculated, and the abnormal area structure value is calculated by the formula The abnormal area structure value is calculated .

[0119] In the technical solution, the gray image of the inner surface contour image of the sealed cavity is recognized, the test area contour in the image is screened, the abnormal area contour is obtained through screening, the shape feature of the abnormal area contour is calculated, the abnormal area structure value is obtained through calculation, the defect of the inner surface contour of the sealed cavity is determined, and then the leakage state of the cavity surface of the measured device is determined according to the defect feature analysis.

[0120] As an embodiment of the application, it further includes:

[0121] The abnormal area structure value is compared with a preset threshold value , and , wherein < :

[0122] If ≤ , it is judged that the cavity surface of the measured device in the sealed space has slight leakage.

[0123] If < < , it is judged that the cavity surface of the measured device in the sealed space has moderate leakage.

[0124] If ≥ , it is judged that the cavity surface of the measured device in the sealed space has serious leakage.

[0125] Each embodiment in the specification is described in a progressive manner, and the same and similar parts between each embodiment can be referred to each other. Each embodiment mainly describes the difference from other embodiments. Especially, the device, equipment and non-volatile computer storage medium embodiments are basically similar to the method embodiments, so the description is relatively simple, and the related parts can be referred to the part of the method embodiment.

[0126] The above described embodiments of the present description have been described. Other embodiments are within the scope of the following documents. In some cases, the actions or steps described in this application can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multi-task processing and parallel processing are possible or can be advantageous.

[0127] The above is only an example and illustration of the concept of the present application, and those skilled in the art can make various modifications or supplements to the described specific embodiments or replace them with similar ways, as long as they do not deviate from the concept of the invention or exceed the scope defined in this application.

Claims

1. An artificial intelligence-based airtightness monitoring system for a sealing tester, the sealing tester comprising a test sealing space and a water box; the test sealing space comprising a sealed cavity and a device under test; characterized in that: The system comprises: The sensor module is used to collect the air pressure change parameters of the device under test in the test sealed space in real time through the pressure sensor after the device under test is inflated; An image processing module is used to use a camera to collect dynamic image information of bubbles in the water box connected to one side of the test sealed space through the air pipe; and to obtain an image of the inner surface contour of the sealed cavity through image processing technology; A model building module is used to input historical air pressure change parameters and historical bubble dynamic information images into a preset convolutional neural network model for training, build a prediction model, and obtain bubble motion parameters; A data analysis module is used to input the real-time acquired air pressure change parameters and bubble movement parameters into the prediction model for tracking detection to obtain air tightness prediction parameters; And it is used to execute the evaluation strategy based on the airtightness prediction parameters and judge whether there is an airtightness problem of the sealing test machine according to the evaluation results: If so, the airtightness in the current sealing test machine is judged to be unqualified, an alarm signal is generated, and a surface defect analysis is performed on the inner surface contour image of the sealing cavity of the sealed space; If not, the airtightness in the current sealing test machine is judged to be qualified and the airtightness monitoring is continued; Display module, used to display alarm results; The air pressure change parameter is obtained as follows: S1. Obtaining a first air pressure value at an air filling pipe opening in a sealed space; S2. Obtaining a second air pressure value in the sealed space after the continuous inflation reaches a set time; S3, calculating the difference between the first air pressure value and the second air pressure value as the air pressure difference within the set time period; S4. Obtain a curve showing the continuous change of the air pressure difference within a preset time period, and calculate the slope of the curve to determine the air pressure change parameter; The steps for obtaining the bubble motion parameters are as follows: SS1, predetermine the time interval between adjacent frames of bubble dynamic images; SS2. Obtain the change in the number of bubbles in the bubble dynamic image of consecutive frames within a preset time period and the change in the bubble displacement distance corresponding to all time intervals; SS3. Obtain the bubble movement speed according to the change of bubble quantity and bubble displacement distance; SS4, dividing the movement rhythm in each time interval according to the bubble movement speed as the bubble movement parameter; The calculation method of the airtightness prediction parameter is: in, For the Airtightness prediction parameters of test samples; is the preset function; is the preset first conversion function, is the second preset conversion function; For the Air pressure variation parameters of the test samples; is the standard air pressure variation parameter; For the The preset deviation value of the air pressure change parameter of each test sample; For the Bubble motion parameters of the test samples; is the standard bubble motion parameter; For the The bubble motion parameter preset deviation value of each test sample.

2. The airtightness monitoring system for a sealing tester based on artificial intelligence according to claim 1, characterized in that: The data analysis module inputs the real-time acquired air pressure change parameters and bubble movement parameters into the prediction model. The specific process of tracking detection includes: Statistical airtightness-related features are coded and then normalized; The prediction model performs tracking detection to calculate the loss function between the predicted value of the airtightness-related feature and its target value; Automatically adjust the model parameters of the prediction model using the stochastic gradient descent algorithm to minimize the loss function; After multiple iterations, the prediction model converges and the airtightness prediction parameters are output.

3. The airtightness monitoring system for a sealing tester based on artificial intelligence according to claim 1, characterized in that: The airtightness prediction parameters The threshold value of the airtightness prediction parameter is consistent with the preset standard To compare: like > , it is determined that the current tested equipment has a leakage risk and generates an early warning message; like ≤ , it is determined that the current device under test is normal.

4. The airtightness monitoring system for a sealing tester based on artificial intelligence according to claim 1, characterized in that: The process of the data analysis module executing the evaluation strategy according to the airtightness prediction parameters includes: By formula Calculate the evaluation coefficient ; in, is the total number of test samples, and ∈ ; For the The preset weight coefficient of each test sample; is the standard airtightness prediction parameter, and ≤ .

5. The airtightness monitoring system for a sealing tester based on artificial intelligence according to claim 4, characterized in that: The step of judging whether there is an airtightness problem of the sealing test machine based on the evaluation results specifically includes: The evaluation coefficient and preset evaluation coefficient threshold To compare: like < , it is determined that there is an airtightness problem with the sealing test machine and an alarm signal is generated; like ≥ , it is judged that there is no airtightness problem in the sealing test machine.

6. The airtightness monitoring system for a sealing tester based on artificial intelligence according to claim 1, characterized in that: The process of performing surface defect analysis on the inner surface contour image of the sealed cavity of the sealed space is as follows: Identify the inner surface contour image of the sealed cavity and perform grayscale processing; The Canny algorithm is used to extract the contour of abnormal areas in the inner surface contour image; Calculate the area S and perimeter C of the abnormal area contour by the formula Calculate the abnormal area structure value .

7. The airtightness monitoring system for a sealing tester based on artificial intelligence according to claim 6, characterized in that: Also includes: The abnormal area structure value With preset threshold and For comparison, < : like ≤ , it is judged that there is a slight leakage on the cavity surface of the device under test in the sealed space; like < < , it is judged that there is moderate leakage on the cavity surface of the device under test in the sealed space; like ≥ , it is judged that there is a serious leakage on the cavity surface of the device under test in the sealed space.

Images