Double-layer plastic blow molding detection method and system

Abstract

The invention discloses a double-layer plastic blow molding detection method and system, and relates to the field of plastic product production, and the method comprises the following steps: obtaining specification characteristics of a to-be-detected plastic plate, and screening out a difference grid region according to a pressure distribution thermodynamic diagram; obtaining a plurality of communication regions formed by the difference grid regions, screening out a target communication region from the communication regions, and extracting a pressure average gradient as a pressure feature of the to-be-detected plastic plate; collecting an echo signal set of the target connected region, obtaining the average echo density of the target connected region according to the echo signal set, and taking the average echo density as the ultrasonic characteristic of the to-be-detected plastic plate; performing feature-level fusion on the specification features, the pressure features and the ultrasonic features of the to-be-detected plastic plate to obtain multi-modal features; inputting the multi-modal features into a convolutional neural network to obtain a wall thickness detection result of the to-be-detected plastic plate; according to the invention, efficient and accurate wall thickness detection can be realized on plastic plates with different sizes and different materials.

Description

Technical Field

[0001] The invention relates to the field of plastic product production, and in particular to a double-layer plastic blow molding detection method and system. Background Art

[0002] Plastic sheets are widely used in packaging, automobile manufacturing, building materials and other industries. The uniformity of their wall thickness directly affects the mechanical properties, durability and service life of the products. However, in the production process of plastic sheets, due to the instability of process parameters such as extrusion, blow molding, and hot pressing, plastic sheets are prone to local uneven thickness, such as excessively thick areas caused by excessive local stacking, or too thin areas caused by uneven material stretching. This wall thickness deviation not only affects the quality stability of plastic sheets, but may also cause problems such as breakage and deformation of the product during actual use.

[0003] Existing methods for detecting the wall thickness of plastic sheets mainly include contact measurement, ultrasonic thickness measurement, optical measurement, etc. Among them, contact measurement methods (such as micrometers and thickness gauges) require manual operation, have a limited measurement range, and are difficult to cover the global thickness distribution of the entire sheet; although ultrasonic thickness measurement can achieve non-contact detection, it can only measure at a single point or a local area, and it is difficult to fully reflect the thickness changes of the entire plastic sheet; in addition, optical measurement methods work better on some transparent or translucent plastic sheets, but have poor adaptability to different materials and are easily affected by surface roughness and lighting conditions; therefore, existing detection technologies are difficult to balance comprehensiveness, accuracy and applicability, and are difficult to meet the needs of large-scale, automated, and high-precision quality inspection of plastic sheets.

[0004] In view of this, a new wall thickness detection method is urgently needed, which can improve the comprehensiveness and automation of detection while ensuring the detection accuracy. Summary of the invention

[0005] The object of the present invention is to provide a double-layer plastic blow molding detection method and system to solve the problems in the above-mentioned background technology.

[0006] In order to achieve the above object, the present invention provides the following technical solutions: In a first aspect, the present invention provides a double-layer plastic blow molding detection method, comprising: Obtaining the specification characteristics and pressure distribution thermodynamic map of the plastic sheet to be tested, and screening out the difference grid areas according to the pressure distribution thermodynamic map, wherein the pressure distribution thermodynamic map includes multiple grid areas and the measured pressure average value of each grid area; A plurality of connected regions composed of various difference grid regions are obtained, and a target connected region is screened out from the connected regions, and the average pressure gradient of the target connected region is extracted as the pressure feature of the plastic sheet to be tested; Collecting the echo signal set of the target connected area, obtaining the average echo density of the target connected area based on the echo signal set, and using the average echo density as the ultrasonic feature of the plastic sheet to be tested; The specification features, pressure features and ultrasonic features of the plastic sheet to be tested are fused at the feature level to obtain multi-modal features; The multimodal features are input into a convolutional neural network that has been pre-trained to perform wall thickness detection tasks to obtain the wall thickness detection results of the plastic sheet to be tested.

[0007] As a preferred technical solution of the first aspect of the present invention, the specification features include but are not limited to the size, type and designed thickness of the plastic sheet to be tested; The specific calculation formula for the measured mean pressure of each grid area is as follows: ; Where: is the measured pressure mean value of the grid area in the i-th row and j-th column; is the force collected by the rth pressure sensor corresponding to the grid area of ​​the i-th row and j-th column; is the grid area of ​​the grid area in the i-th row and j-th column, is the correction factor, and R is the total number of pressure sensors corresponding to each grid area.

[0008] As a preferred technical solution of the first aspect of the present invention, the correction factor is based on the distance between the pressure sensor and the center of the grid. The specific calculation formula is: , is a constant.

[0009] As a preferred technical solution of the first aspect of the present invention, the step of screening out the difference grid area includes: Calculate the global mean pressure of the plastic sheet to be tested and standard deviation ; Among them, the global pressure mean The expression is as follows: ; Among them, the standard deviation The expression is as follows: ; Generate pressure threshold intervals based on global pressure mean and standard deviation [ ]; The measured pressure mean of each grid area is compared with the pressure threshold interval; like or , then the corresponding grid area is determined to be a difference grid area; like and , then the corresponding grid area is determined to be a non-difference grid area; Repeat the above steps until each grid area is traversed and all difference grid areas are obtained.

[0010] As a preferred technical solution of the first aspect of the present invention, the step of selecting a target connected area from connected areas includes: Calculate the average pressure gradient of each connected area to obtain multiple average pressure gradients; The calculation formula of the average pressure gradient is as follows: ; Where: For connected areas The average pressure gradient, For Grid The gradient value of is the total number of difference grid regions in the connected region; Among them, the grid The calculation formula of the gradient value is as follows: ; Where: For Grid The measured mean pressure, For Grid The measured mean pressure, For Grid The measured mean pressure, For Grid The measured mean pressure, For Grid The width of For Grid Height; The multiple pressure average gradients are sorted from large to small in value, and the connected area corresponding to the first sorted pressure average gradient is marked as the target connected area.

[0011] As a preferred technical solution of the first aspect of the present invention, the echo signal set includes a plurality of ultrasonic echo signals; Wherein, obtaining the average echo density of the target connected area includes: Obtain the number of valid echo signals in the echo signal set and the area of ​​the target connected area; According to the number of effective echo signals and the area of ​​the target connected area, the average echo density of the target connected area is calculated; The specific calculation formula of the average echo density is as follows: ; Where: is the average echo density, is the number of valid echo signals in the target connected area, is the area of ​​the target connected region.

[0012] As a preferred technical solution of the first aspect of the present invention, the training method of the convolutional neural network for performing the wall thickness detection task is as follows: Acquire historical wall thickness detection training data, and divide the historical wall thickness detection training data into a wall thickness detection training set and a wall thickness detection test set; the historical wall thickness detection training data includes wall thickness detection features and their corresponding annotation labels; Wherein, the wall thickness detection features include specification features, pressure features and ultrasonic features; The marking labels include "0" and "1", "0" indicates that the wall thickness is unqualified, and "1" indicates that the wall thickness is qualified; A classifier for performing classification tasks is constructed based on a convolutional neural network as the basic architecture, the wall thickness detection features in the wall thickness detection training set are used as the input of the classifier, and the annotation labels are used as the output of the classifier. The classifier is trained to obtain an initial classification network; The initial classification network is model verified using the wall thickness detection test set, and the output of the initial classification network is greater than or equal to the preset test accuracy threshold as the convolutional neural network for performing wall thickness detection tasks.

[0013] In a second aspect, the present invention provides a double-layer plastic blow molding detection system, which is implemented based on the above-mentioned double-layer plastic blow molding detection method, and includes: A data acquisition module is used to obtain the specification characteristics and pressure distribution thermodynamic map of the plastic sheet to be tested, and screen out the difference grid area according to the pressure distribution thermodynamic map, wherein the pressure distribution thermodynamic map includes multiple grid areas and the measured pressure mean value of each grid area; The first feature extraction module is used to obtain multiple connected areas composed of various difference grid areas, screen out target connected areas from the connected areas, and extract the average pressure gradient of the target connected area as the pressure feature of the plastic sheet to be tested; The second feature extraction module is used to collect the echo signal set of the target connected area, obtain the average echo density of the target connected area based on the echo signal set, and use the average echo density as the ultrasonic feature of the plastic sheet to be tested; A feature fusion module is used to fuse the specification features, pressure features and ultrasonic features of the plastic sheet to be tested at the feature level to obtain multi-modal features; The blow molding detection module is used to input multimodal features into a convolutional neural network that has been pre-trained to perform wall thickness detection tasks to obtain the wall thickness detection results of the plastic sheet to be tested.

[0014] In a third aspect, the present invention provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements any of the above-mentioned double-layer plastic blow molding detection methods when executing the computer program.

[0015] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed, the double-layer plastic blow molding detection method described in any one of the above items is implemented.

[0016] In the above technical solution, the technical effects and advantages provided by the present invention are: The present invention uses a multi-point pressure sensor array to perform grid measurement on the pressure distribution of the plastic sheet, and combines the connected domain analysis to accurately screen out the abnormal wall thickness area, avoiding the problems of misjudgment and insufficient detection coverage caused by the single-point measurement method; compared with the existing method, the present application can identify local anomalies through the statistical characteristics of the global pressure distribution, and further combine ultrasonic detection to verify the thickness of the target connected area, making the wall thickness detection more accurate and stable; Moreover, by first screening out the target connected areas with drastic changes in wall thickness, the high-density measurement of the entire plastic sheet is reduced, the computational complexity is reduced, and the detection accuracy is not affected. In addition, the specification features, pressure features, and ultrasonic features of the plastic sheet are fused through the feature-level fusion method, and the wall thickness intelligent analysis is performed using a deep learning model, making the detection results more adaptable and generalizable, reducing human intervention and improving the degree of automation. Compared with the existing technology, the present invention can realize efficient and accurate wall thickness detection on plastic plates of different sizes and materials, ensure the comprehensiveness, stability and calculation efficiency of the detection, and provide more reliable technical support for the quality control of plastic product production. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0018] Figure 1 It is a flow chart of a double-layer plastic blow molding detection method of the present invention; Figure 2 It is a framework diagram of a double-layer plastic blow molding detection system of the present invention; Figure 3 The present invention provides a schematic structural diagram of an electronic device. DETAILED DESCRIPTION

[0019] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description disclosed herein will be more comprehensive and complete, and the concepts of the example embodiments will be fully conveyed to those skilled in the art. The accompanying drawings are only schematic illustrations of the disclosure of the present application and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and thus their repeated description will be omitted.

[0020] In addition, the described features, structures or characteristics may be combined in one or more example embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the example embodiments disclosed in the present application. However, those skilled in the art will appreciate that the technical solutions disclosed in the present application may be practiced while omitting one or more of the specific details, or other methods, components, steps, etc. may be adopted. In other cases, known structures, methods, implementations or operations are not shown or described in detail to avoid obscuring the various aspects disclosed in the present application.

[0021] Example 1 like Figure 1 As shown, this embodiment discloses a double-layer plastic blow molding detection method, comprising: Step 101: Obtaining the specification characteristics and pressure distribution thermodynamic map of the plastic sheet to be tested, and screening out the difference grid areas according to the pressure distribution thermodynamic map, wherein the pressure distribution thermodynamic map includes multiple grid areas and the measured pressure average value of each grid area; The specification features include but are not limited to the size (such as 30cm×30cm, 100cm×100cm, etc.), type (such as PET, PP, HDPE, PC, etc.) and design thickness (i.e., the standard thickness of the corresponding type of plastic sheet to be tested formulated before production) of the plastic sheet to be tested; Specifically, before obtaining the pressure distribution thermal map, it includes: Matching a corresponding division rule according to the size of the plastic sheet to be tested in the specification feature, and dividing the plastic sheet to be tested into m×n grid areas according to the corresponding division rule, where m and n are integers greater than zero; Each grid area is sequentially numbered to form a grid number (i.e., a position index of each grid area). For example, each grid area is numbered in order from left to right and from top to bottom to form a unique grid number. For example, in a 32×32 grid division, the grid number of the upper left corner is (1,1), and the grid number of the lower right corner is (32,32). It should be understood that: plastic sheets of different sizes to be tested are all matched with a division rule, which is specifically determined by the technicians' artificial association. For example, a size of 30cm×30cm is divided into 32×32 equal grids; It should be noted that: the measured pressure mean value is obtained based on the acquisition of the multi-point pressure sensor array; the pressure distribution heat map is generated based on the existing heat map tools, such as; Matplotlib (Python), OpenCV (Python), D3.js (JavaScript), etc.; wherein, the measured pressure mean value of each grid area is obtained by calculating the force acquired by each pressure sensor in the array after the plastic sheet to be tested is placed flat on the multi-point pressure sensor array; Specifically, the specific calculation formula for the measured mean pressure of each grid area is as follows: ; Where: is the measured pressure mean value of the grid area in the i-th row and j-th column; is the force collected by the rth pressure sensor corresponding to the grid area of ​​the i-th row and j-th column; is the grid area of ​​the grid area in the i-th row and j-th column, is the correction factor, R is the total number of pressure sensors corresponding to each grid area; The correction factor is based on the distance between the pressure sensor and the center of the grid. The specific calculation formula is: , is a constant to avoid division by zero errors; It should be noted that: the position of the pressure sensor can be measured by the built-in positioning unit, and the coordinates of the grid center are determined by the size of the plastic sheet to be measured and the grid number; For example, assuming that the size of the plastic sheet is W×H (width×height), and the number of grid divisions is m×n (number of rows×number of columns), the grid size is =W / m, h=H / n, where, For the grid area of ​​row i and column j, the center position of the grid is calculated as: , , where For the width of each grid, For the height of each grid, , is the grid center coordinate of the grid area in the i-th row and j-th column; In implementation, the step of screening out the difference grid area includes: Calculate the global mean pressure of the plastic sheet to be tested and standard deviation ; Among them, the global pressure mean The expression is as follows: ; Among them, the standard deviation The expression is as follows: ; Generate pressure threshold intervals based on global pressure mean and standard deviation [ ]; The measured pressure mean of each grid area is compared with the pressure threshold interval; like or , then the corresponding grid area is determined to be a difference grid area; like and , then the corresponding grid area is determined to be a non-difference grid area; Repeat the above steps until each grid area is traversed and all difference grid areas are obtained; It is understandable that: or When , it means that the corresponding grid area is too thick or too thin, that is, due to the blow molding pressure and other reasons, there is too much material piled up in some areas of the plastic sheet to be tested, or there are defects such as pores in some areas, thus resulting in uneven pressure distribution of the plastic sheet to be tested.

[0022] Step 102: obtaining a plurality of connected regions composed of the respective difference grid regions, screening out a target connected region from the connected regions, and extracting an average pressure gradient of the target connected region as a pressure feature of the plastic sheet to be tested; It should be noted that: there are multiple difference grid areas in each connected area; wherein, multiple connected areas are obtained by adopting the connected component analysis (CCA) method; the CCA method is to regard the difference grid area as the "foreground pixel" in the binary image, and to create a binary grid matrix by setting all non-difference grid areas to "0" and the difference grid area to "1", and then to use the depth first search (DFS) or breadth first search (BFS) to traverse the adjacent "1" in the connected area, so as to find multiple connected areas; In implementation, the step of selecting a target connected area from the connected areas includes: The average pressure gradient of each connected area is calculated to obtain multiple average pressure gradients; the average pressure gradient (MPG) is a value used to measure the average pressure gradient of all grids in a connected area; The calculation formula of the average pressure gradient is as follows: ; Where: For connected areas The average pressure gradient, For Grid The gradient value of is the total number of difference grid regions in the connected region; Among them, the grid The calculation formula of the gradient value is as follows: ; Where: For Grid The measured pressure average value, that is, relative to the grid The right grid pressure, For Grid The measured pressure average value, that is, relative to the grid The left grid pressure, For Grid The measured pressure average value, that is, relative to the grid The grid pressure below, For Grid The measured pressure average value, that is, relative to the grid The upper grid pressure, For Grid The width of is used to calculate the normalization factor of the X-direction gradient. For Grid The height of is used to calculate the normalization factor of the y-direction gradient; Sort the multiple pressure average gradients according to the values ​​from large to small, and mark the connected area corresponding to the first pressure average gradient in the sort as the target connected area; It should be understood that the larger the average pressure gradient, the more drastic the pressure change in the corresponding connected area, which usually corresponds to the part where the wall thickness changes most significantly. Therefore, by determining the target connected area through the average pressure gradient, and using the average pressure gradient of the target connected area as the pressure characteristic of the plastic sheet to be tested, the accuracy of the thickness detection of the plastic sheet to be tested can be significantly improved, while reducing the consumption of computing resources caused by calculating each local area.

[0023] Step 103: collecting an echo signal set of the target connected area, obtaining an average echo density of the target connected area based on the echo signal set, and using the average echo density as an ultrasonic feature of the plastic sheet to be tested; Wherein, the echo signal set includes a plurality of ultrasonic echo signals; It should be noted that: the echo signal set is obtained by scanning the target connected area using an ultrasonic detection device, and the ultrasonic detection device includes but is not limited to an ultrasonic detector, an ultrasonic array detector or an ultrasonic measuring instrument, etc.; In implementation, obtaining the average echo density of the target connected area includes: Obtain the number of valid echo signals in the echo signal set and the area of ​​the target connected area; It should be understood that the effective echo signal is obtained by screening according to the preset effective echo reflection time interval, that is, the echo reflection time of each ultrasonic echo signal in the echo signal set is obtained, and the echo reflection time is compared with the effective echo reflection time interval [ ] for comparison, if the echo reflection time is in the valid echo reflection time interval [ ], the corresponding ultrasonic echo signal is marked as a valid echo signal, otherwise, the corresponding ultrasonic echo signal is not marked as a valid echo signal; According to the number of effective echo signals and the area of ​​the target connected area, the average echo density of the target connected area is calculated; The specific calculation formula of the average echo density is as follows: ; Where: is the average echo density, is the number of valid echo signals in the target connected area, is the area of ​​the target connected region; It should be understood that by collecting the pressure characteristics and ultrasonic characteristics of the plastic sheet to be tested, the error existing in a single detection method and the resulting false detection phenomenon can be eliminated, and at the same time it is beneficial to improve the wall thickness detection accuracy of the plastic sheet.

[0024] Step 104: fusing the specification features, pressure features and ultrasonic features of the plastic sheet to be tested at the feature level to obtain multi-modal features; It should be noted that the feature-level fusion is achieved by direct feature concatenation or weighted feature fusion, and this embodiment does not impose too many limitations on this.

[0025] Step 105: input the multimodal features into a convolutional neural network that is pre-trained to perform a wall thickness detection task, so as to obtain a wall thickness detection result of the plastic sheet to be tested; Specifically, the training method of the convolutional neural network for performing the wall thickness detection task is as follows: Acquire historical wall thickness detection training data, and divide the historical wall thickness detection training data into a wall thickness detection training set and a wall thickness detection test set; the historical wall thickness detection training data includes wall thickness detection features and their corresponding annotation labels; Wherein, the wall thickness detection features include specification features, pressure features and ultrasonic features; The marking labels include "0" and "1", "0" indicates that the wall thickness is unqualified, and "1" indicates that the wall thickness is qualified; It should be noted that the wall thickness detection features in the historical wall thickness detection training data are obtained by technicians based on actual collection and recording of experimental data, and the annotation labels in the historical wall thickness detection training data are obtained by data annotators manually annotating. A classifier for performing classification tasks is constructed based on a convolutional neural network as the basic architecture, the wall thickness detection features in the wall thickness detection training set are used as the input of the classifier, and the annotation labels are used as the output of the classifier. The classifier is trained to obtain an initial classification network; The initial classification network is model verified using the wall thickness detection test set, and the output of the initial classification network is greater than or equal to the preset test accuracy threshold as the convolutional neural network for performing wall thickness detection tasks.

[0026] Example 2 like Figure 2 As shown, the part not described in detail in this embodiment is as shown in Example 1. This embodiment discloses a double-layer plastic blow molding detection system, including: The data acquisition module 201 is used to obtain the specification characteristics and pressure distribution thermodynamic map of the plastic sheet to be tested, and screen out the difference grid area according to the pressure distribution thermodynamic map, wherein the pressure distribution thermodynamic map includes multiple grid areas and the measured pressure average value of each grid area; The first feature extraction module 202 is used to obtain multiple connected areas composed of various difference grid areas, screen out a target connected area from the connected areas, and extract the average pressure gradient of the target connected area as the pressure feature of the plastic sheet to be tested; The second feature extraction module 203 is used to collect the echo signal set of the target connected area, obtain the average echo density of the target connected area based on the echo signal set, and use the average echo density as the ultrasonic feature of the plastic sheet to be tested; The feature fusion module 204 is used to fuse the specification features, pressure features and ultrasonic features of the plastic sheet to be tested at the feature level to obtain multi-modal features; The blow molding detection module 205 is used to input the multimodal features into a convolutional neural network that is pre-trained to perform the wall thickness detection task, so as to obtain the wall thickness detection result of the plastic sheet to be tested.

[0027] Example 3 See also Figure 3 As shown, this embodiment discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, any of the double-layer plastic blow molding detection methods provided by the above methods is implemented.

[0028] Since the electronic device introduced in this embodiment is an electronic device used to implement the double-layer plastic blow molding detection method in the embodiment of this application, based on the double-layer plastic blow molding detection method introduced in the embodiment of this application, the technical personnel in this field can understand the specific implementation of the electronic device in this embodiment and its various variations, so how the electronic device implements the method in the embodiment of this application is not described in detail here. As long as the technical personnel in this field implement the electronic device used in the double-layer plastic blow molding detection method in the embodiment of this application, it belongs to the scope of protection of this application.

[0029] Example 4 This embodiment discloses a computer-readable storage medium, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, any of the double-layer plastic blow molding detection methods provided by the above methods is implemented.

[0030] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters, weights and thresholds in the formula are set by technicians in this field according to actual conditions.

[0031] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website, computer, server or data center to another website, computer, server or data center via a wired network or a wireless network. The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media sets. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD) or a semiconductor medium. The semiconductor medium may be a solid-state hard disk.

[0032] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A double-layer plastic blow molding detection method, characterized in that: include: Obtaining the specification characteristics and pressure distribution thermodynamic map of the plastic sheet to be tested, and screening out the difference grid areas according to the pressure distribution thermodynamic map, wherein the pressure distribution thermodynamic map includes multiple grid areas and the measured pressure average value of each grid area; A plurality of connected regions composed of various difference grid regions are obtained, and a target connected region is screened out from the connected regions, and the average pressure gradient of the target connected region is extracted as the pressure feature of the plastic sheet to be tested; Collecting the echo signal set of the target connected area, obtaining the average echo density of the target connected area based on the echo signal set, and using the average echo density as the ultrasonic feature of the plastic sheet to be tested; The specification features, pressure features and ultrasonic features of the plastic sheet to be tested are fused at the feature level to obtain multi-modal features; The multimodal features are input into a convolutional neural network that has been pre-trained to perform wall thickness detection tasks to obtain the wall thickness detection results of the plastic sheet to be tested.

2. The double-layer plastic blow molding detection method according to claim 1 is characterized in that: The specification features include but are not limited to the size, type and design thickness of the plastic sheet to be tested; The specific calculation formula for the measured mean pressure of each grid area is as follows: ; Where: is the measured pressure mean value of the grid area in the i-th row and j-th column; is the force collected by the rth pressure sensor corresponding to the grid area of ​​the i-th row and j-th column; is the grid area of ​​the grid area in the i-th row and j-th column, is the correction factor, and R is the total number of pressure sensors corresponding to each grid area.

3. The double-layer plastic blow molding detection method according to claim 2 is characterized in that: The correction factor is based on the distance between the pressure sensor and the center of the grid The specific calculation formula is: , is a constant.

4. The double-layer plastic blow molding detection method according to claim 3 is characterized in that: The step of screening out the difference grid area includes: Calculate the global mean pressure of the plastic sheet to be tested and standard deviation ; Among them, the global pressure mean The expression is as follows: ; Among them, the standard deviation The expression is as follows: ; Where: m is the number of rows, n is the number of columns; Generate pressure threshold intervals based on global pressure mean and standard deviation [ ]; The measured pressure mean of each grid area is compared with the pressure threshold interval; like or , then the corresponding grid area is determined to be a difference grid area; like and , then the corresponding grid area is determined to be a non-difference grid area; Repeat the above steps until each grid area is traversed and all difference grid areas are obtained.

5. The double-layer plastic blow molding detection method according to claim 4, characterized in that: The step of selecting a target connected area from the connected areas includes: Calculate the average pressure gradient of each connected area to obtain multiple average pressure gradients; The calculation formula of the average pressure gradient is as follows: ; Where: For connected areas The average pressure gradient, For Grid The gradient value of is the total number of difference grid regions in the connected region; Among them, the grid The calculation formula of the gradient value is as follows: ; Where: For Grid The measured mean pressure, For Grid The measured mean pressure, For Grid The measured mean pressure, For Grid The measured mean pressure, For Grid The width of For Grid Height; The multiple pressure average gradients are sorted from large to small in value, and the connected area corresponding to the first sorted pressure average gradient is marked as the target connected area.

6. The double-layer plastic blow molding detection method according to claim 5, characterized in that: The echo signal set includes a plurality of ultrasonic echo signals; Wherein, obtaining the average echo density of the target connected area includes: Obtain the number of valid echo signals in the echo signal set and the area of ​​the target connected area; According to the number of effective echo signals and the area of ​​the target connected area, the average echo density of the target connected area is calculated; The specific calculation formula of the average echo density is as follows: ; Where: is the average echo density, is the number of valid echo signals in the target connected area, is the area of ​​the target connected region.

7. The double-layer plastic blow molding detection method according to claim 6, characterized in that: The training method of the convolutional neural network for performing the wall thickness detection task is as follows: Acquire historical wall thickness detection training data, and divide the historical wall thickness detection training data into a wall thickness detection training set and a wall thickness detection test set; the historical wall thickness detection training data includes wall thickness detection features and their corresponding annotation labels; Wherein, the wall thickness detection features include specification features, pressure features and ultrasonic features; The marking label includes "0" and "1", "0" indicates that the wall thickness is unqualified, and "1" indicates that the wall thickness is qualified; A classifier for performing classification tasks is constructed based on a convolutional neural network as the basic architecture, the wall thickness detection features in the wall thickness detection training set are used as the input of the classifier, and the annotation labels are used as the output of the classifier. The classifier is trained to obtain an initial classification network; The initial classification network is model verified using the wall thickness detection test set, and the output of the initial classification network is greater than or equal to the preset test accuracy threshold as the convolutional neural network for performing wall thickness detection tasks.

8. A double-layer plastic blow molding detection system, implemented based on the double-layer plastic blow molding detection method according to any one of claims 1 to 7, characterized in that: include: A data acquisition module is used to obtain the specification characteristics and pressure distribution thermodynamic map of the plastic sheet to be tested, and screen out the difference grid area according to the pressure distribution thermodynamic map, wherein the pressure distribution thermodynamic map includes multiple grid areas and the measured pressure mean value of each grid area; The first feature extraction module is used to obtain multiple connected areas composed of various difference grid areas, screen out target connected areas from the connected areas, and extract the average pressure gradient of the target connected area as the pressure feature of the plastic sheet to be tested; The second feature extraction module is used to collect the echo signal set of the target connected area, obtain the average echo density of the target connected area based on the echo signal set, and use the average echo density as the ultrasonic feature of the plastic sheet to be tested; A feature fusion module is used to fuse the specification features, pressure features and ultrasonic features of the plastic sheet to be tested at the feature level to obtain multi-modal features; The blow molding detection module is used to input multimodal features into a convolutional neural network that has been pre-trained to perform wall thickness detection tasks to obtain the wall thickness detection results of the plastic sheet to be tested.

9. An electronic device 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 double-layer plastic blow molding detection method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed, the double-layer plastic blow molding detection method according to any one of claims 1 to 7 is implemented.

Images