Intelligent metal gasket automatic detection and grading system and method

Abstract

The present invention discloses an intelligent metal gasket automatic detection and grading system and method, which specifically relates to the field of metal gasket automatic detection, and is used to solve the problem of balancing complex defect detection and assembly line efficiency. The defect complexity of the metal gasket is accurately quantified by comprehensively considering the geometric complexity of the defect edge and the uneven energy distribution, and is graded accordingly. The gasket of the fuzzy signal triggers the accommodation space adjustment mechanism, dynamically adjusts the size of the temporary accommodation space before the assembly line, increases the buffer area, and adjusts the gasket interval to ensure that the gasket with complex defects obtains sufficient detection time and resources. After the detection is completed, the temporary accommodation space is automatically compressed according to the signal type of the subsequent gasket, and the transmission of the clear signal gasket is accelerated to maintain the continuity of the assembly line. In this way, the accurate detection and resource allocation of complex defects are realized, the detection accuracy is improved, the efficient operation of the assembly line is guaranteed, and the quality risk is reduced.

Description

Technical Field

[0001] The present invention relates to the field of automated detection of metal gaskets, and more specifically, to an intelligent automated detection and grading system and method for metal gaskets. Background Art

[0002] Metal gaskets are widely used in industrial production, especially in high-pressure sealing, mechanical equipment and other fields. The quality of gaskets directly affects the safety and stability of the equipment. With the popularization of automated assembly lines, the demand for gasket inspection is increasing, and accurate and efficient quality inspection has become the key. Traditional gasket inspection methods mostly rely on manual or simple automated equipment, which makes it difficult to effectively deal with the accurate identification of complex defects on the gasket surface, especially in mass production. Detailed analysis of complex defects often affects the production rhythm. Therefore, there is an urgent need for an intelligent solution that can detect gasket surface defects in real time and efficiently on the assembly line.

[0003] The existing metal gasket assembly line inspection technology faces obvious bottlenecks in practical applications. Since metal gaskets may have a variety of defects during the production process, such as surface scratches, cracks, corrosion and other complex problems, the characteristics of these defects are often difficult to be fully identified and analyzed within the scheduled inspection time. If the inspection is not comprehensive enough, the gasket will flow directly into the next process, which may cause quality problems and affect the final product quality of the entire production line. However, if the inspection time of each gasket is extended to ensure the accuracy and coverage of the inspection, the overall inspection efficiency of the production line will inevitably be significantly reduced, resulting in a decrease in the operating speed of the assembly line and affecting the progress of large-scale production. It is difficult for existing technologies to achieve a balance between inspection accuracy and efficiency, and it is impossible to flexibly adjust the inspection time and resource allocation according to the complexity of the actual defects of the gasket. This not only creates high-risk quality risks, but also makes it difficult to maintain the continuity and efficiency of the inspection process in a mass production environment.

[0004] In order to solve the above problems, a technical solution is now provided. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides an intelligent metal gasket automatic detection and grading system and method, which accurately quantifies the defect complexity of the metal gasket by comprehensively considering the geometric complexity of the defect edge and the uneven energy distribution, and grades them according to the evaluation results. For gaskets judged as fuzzy signals, the accommodation space adjustment mechanism is triggered, the size of the temporary accommodation space before the assembly line is dynamically adjusted, the buffer area is increased, and the gasket interval is adjusted to ensure that gaskets with complex defects obtain sufficient detection time and resource investment. After the detection is completed, the temporary accommodation space is automatically compressed according to the signal type of the subsequent gasket, and the transmission of the clear signal gasket is accelerated to ensure the continuity of the assembly line. The entire process realizes the accurate detection of complex defects and the reasonable allocation of resources through the dynamic adjustment of the structure-specific redundancy index, which not only improves the detection accuracy, but also ensures the efficient operation of the assembly line and reduces quality risks, so as to solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] S1, collects image data of the metal gasket surface through image sensors, applies deep learning algorithms based on convolutional neural networks, extracts defect features, analyzes the defect complexity of each gasket, and generates a structure-specific redundancy index;

[0008] S2, comparing the structure-specific redundancy index with the set threshold, those exceeding the set threshold are judged as fuzzy signals, and those not exceeding the threshold are judged as clear signals;

[0009] S3, for fuzzy signal gaskets, trigger the accommodation space adjustment mechanism, dynamically adjust the temporary accommodation space before the pipeline according to the structure-specific redundancy index, increase the buffer space and adjust the gasket interval to ensure that the fuzzy signal gasket has sufficient detection time.

[0010] In a preferred embodiment, step S1 includes the following contents:

[0011] S1.1, use relevant equipment to scan the surface of each metal gasket at multiple angles to obtain holographic image data of the gasket surface;

[0012] S1.2, after image data acquisition, the image is first subjected to noise filtering and contrast enhancement;

[0013] S1.3, after the image preprocessing is completed, the defect features of the image are automatically extracted through the convolutional neural network model;

[0014] S1.4, after the defect feature extraction is completed, the geometric morphology feature and the energy distribution feature are calculated and obtained, wherein the geometric morphology feature includes the fractal dimension of the defect edge, and the energy distribution feature includes the surface energy dispersion coefficient.

[0015] In a preferred embodiment, the process of obtaining the fractal dimension of the defect edge is as follows:

[0016] Use an image sensor to obtain the original grayscale image of the metal gasket surface ;

[0017] Apply anisotropic diffusion filter Smooth the image to preserve edge details while reducing noise; ; Represents the image processed by an anisotropic diffusion filter, reducing noise while preserving edges; represents the preprocessed image;

[0018] The Marr-Hildreth edge detection algorithm is used, and the Laplace-Gaussian operator is used to Convolve the preprocessed image to extract edge information and obtain an edge response image; ;in, is a Gaussian function; is the scale parameter; represents the edge response image;

[0019] Use adaptive threshold Convert edge response image to binary covered edge image : ;

[0020] Place the covered edge image at a scale of On a square grid, the initial box size is ;

[0021] Define a set of box dimensions ,in , until the minimum box size is reached ; Indicates the number of box dimensions; It represents the index of the scale, i.e. The size of the box;

[0022] For each scale , calculate the number of non-empty boxes needed to cover the edge image , the following conditions are met: ;in, , Indicates There are edge points in the box. Refers to the side length of the smallest circumscribed square area of ​​the edge covering the image; Indicated in scale The number of non-empty boxes required to cover the edge image; Indicates The size of the box, ; Indicated in scale The total number of boxes required to cover the entire edge image area;

[0023] by is the horizontal axis, As the vertical axis, draw a relationship diagram in the double logarithmic coordinate system;

[0024] The least squares method is used to perform linear fitting on the data points. The slope of the fitted line is the fractal dimension of the defect edge. : ;in Indicates the number of different box scales used for the computation.

[0025] In a preferred embodiment, the process of obtaining the surface energy spread coefficient is as follows:

[0026] Using the gradient energy method, the local energy of each pixel on the gasket surface is calculated by Sobolev gradient; the specific formula is: ;in, Indicates that the image is The local energy of the position is based on the square sum of the transverse gradient and the longitudinal gradient of the image, reflecting the intensity of the change of the local energy on the gasket surface; represents partial derivative;

[0027] The surface of the metal gasket is divided into several regions of equal area, denoted as , each region contains a set of pixels ; For each region , calculate its total energy , the formula is as follows: ; Indicates sub-regions; Indicates area The set of pixels in ; where Indicates area Total energy;

[0028] Based on the local energy distribution, analyze the energy changes between adjacent areas; define adjacent areas and The energy divergence between , the formula is as follows: ;in, and represent the total energy of adjacent regions, For Region and The length of the boundary between the two regions measures the impact of energy changes between the two regions on their boundary;

[0029] The energy dispersion of all adjacent areas is combined to calculate the surface energy dispersion coefficient of the entire gasket surface , to measure the degree of change and unevenness of surface energy distribution; the formula is as follows: ;in, is the energy divergence of adjacent regions, is the total number of regions;

[0030] The fractal dimension of the defect edge and the energy dispersion coefficient are calculated comprehensively to obtain the structure-specific redundancy index.

[0031] In a preferred embodiment, step S2 includes the following contents:

[0032] The structure-specific redundancy index of each gasket was compared with the set threshold Perform real-time comparison; the comparison formula is as follows: ;

[0033] Ambiguous signal: If , indicating that the defect complexity of the corresponding metal gasket is high, and the defect characteristics are difficult to accurately detect by conventional means. At this time, a more detailed analysis is required, so it is determined to be a fuzzy signal;

[0034] Clear signal: If , it indicates that the defect of the corresponding metal gasket is relatively simple or not obvious, and can be detected through conventional detection procedures and determined as a clear signal.

[0035] In a preferred embodiment, step S3 includes the following contents:

[0036] S3.1, when the metal gasket is determined to be a fuzzy signal, the adjustment mechanism of the temporary accommodation space is first triggered; the temporary accommodation space is located before the pipeline and serves as a buffer for the redundant gaskets to be detected; based on the value of the current structure-specific redundancy index, the additional accommodation space that needs to be reserved for subsequent metal gaskets is calculated ; The adjustment formula is as follows: ;in: Indicates the new temporary storage space size; Indicates the base holding space, which indicates the default buffer space size.

[0037] In a preferred embodiment, step S3 further includes the following contents:

[0038] S3.2, while adjusting the temporary accommodation space, adjust the interval between the fuzzy signal gasket that has entered the detection area and the subsequent gaskets on the assembly line through the structure-specific redundancy index; the interval adjustment is dynamically matched according to the load and complexity of the current detection area to ensure that the gaskets with higher complexity have sufficient detection time; the interval adjustment formula is as follows: ;in: Indicates the adjusted shim spacing; Indicates the base interval, which indicates the system default spacer interval; Represents the adjustment coefficient, which is used to dynamically adjust the interval according to the load of the detection system.

[0039] In a preferred embodiment, step S3 further includes the following contents:

[0040] S3.3, in order to ensure that the adjustment of the accommodation space matches the interval adjustment, further adjustment is made according to the signal type after the fuzzy signal gasket detection is completed; if the subsequent gasket is detected as a clear signal, the temporary accommodation space is dynamically compressed according to the need to shorten the interval; the compression formula of the temporary accommodation space is as follows: ;in: Indicates the adjusted size of the temporary storage space.

[0041] Intelligent metal gasket automatic detection and grading system, including: defect extraction module, grading determination module and detection adjustment module;

[0042] Defect extraction module: collects image data of the metal gasket surface through image sensors, applies deep learning algorithms based on convolutional neural networks, extracts defect features, analyzes the defect complexity of each gasket, generates a structure-specific redundancy index, and sends the structure-specific redundancy index to the grading judgment module;

[0043] Grading judgment module: compares the structure-specific redundancy index with the set threshold, and judges the signal exceeding the set threshold as a fuzzy signal, and the signal not exceeding the threshold as a clear signal, and sends the obtained fuzzy signal and clear signal to the detection adjustment module;

[0044] Detection adjustment module: For fuzzy signal gaskets, the accommodation space adjustment mechanism is triggered, and the temporary accommodation space before the pipeline is dynamically adjusted according to the structure-specific redundancy index, the buffer space is increased and the gasket interval is adjusted to ensure that the fuzzy signal gasket has sufficient detection time.

[0045] The technical effects and advantages of the intelligent metal gasket automatic detection and grading system and method of the present invention are as follows:

[0046] The present invention comprehensively considers the geometric complexity of the defect edge and the uneven energy distribution, accurately quantifies the defect complexity of the metal gasket, and grades them according to the evaluation results. For gaskets judged as fuzzy signals, the accommodation space adjustment mechanism is triggered, the size of the temporary accommodation space in front of the assembly line is dynamically adjusted, the buffer area is increased, and the gasket interval is adjusted to ensure that gaskets with complex defects obtain sufficient detection time and resource investment. After the detection is completed, the temporary accommodation space is automatically compressed according to the signal type of the subsequent gasket, and the transmission of clear signal gaskets is accelerated to ensure the continuity of the assembly line. The entire process realizes the accurate detection of complex defects and the rational allocation of resources through the dynamic adjustment of the structure-specific redundancy index, which not only improves the detection accuracy, but also ensures the efficient operation of the assembly line and reduces quality risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 It is a schematic diagram of the process of the intelligent metal gasket automatic detection and grading method of the present invention.

[0048] Figure 2 It is a structural schematic diagram of the intelligent metal gasket automatic detection and grading system of the present invention. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] Embodiment 1: Figure 1 The present invention provides an intelligent metal gasket automatic detection and grading method, comprising:

[0051] S1, collects image data of the metal gasket surface through image sensors, applies a deep learning algorithm based on convolutional neural networks, extracts defect features, analyzes the defect complexity of each gasket, and generates a structure-specific redundancy index.

[0052] S2, the structure-specific redundancy index is compared with the set threshold. The signal exceeding the set threshold is judged as a fuzzy signal, and the signal below the threshold is judged as a clear signal.

[0053] S3, for fuzzy signal gaskets, trigger the accommodation space adjustment mechanism, dynamically adjust the temporary accommodation space before the pipeline according to the structure-specific redundancy index, increase the buffer space and adjust the gasket interval to ensure that the fuzzy signal gasket has sufficient detection time.

[0054] Step S1 includes the following contents:

[0055] S1.1, use relevant equipment (industrial camera or optical sensor) to scan the surface of each metal gasket at multiple angles to obtain holographic image data of the gasket surface. The sensor has a high dynamic range (HDR) function to ensure that clear images can be obtained under different lighting conditions, especially to accurately capture tiny defect features.

[0056] S1.2, after the image data is collected, the image is first subjected to noise filtering and contrast enhancement. The bilateral filtering technique is used to remove background noise, maintain the sharpness of edge features, and ensure that defect details in the image (such as scratches, cracks, etc.) are not masked by noise. Subsequently, the adaptive histogram equalization algorithm is used to optimize the brightness of the image to enhance the visibility of scratches and corrosion areas.

[0057] S1.3, after image preprocessing, the defect features of the image are automatically extracted through the Convolutional Neural Network (CNN) model. The convolutional neural network model is trained in advance with a large number of defect sample data of metal gaskets, and can automatically identify and classify different types of defects, including scratches, cracks, corrosion, etc.

[0058] Feature extraction layer: The first few layers of the convolutional neural network are responsible for extracting low-level features in the image (such as edges, lines, textures, etc.). Through the superposition of convolutional layers and pooling layers, higher-level features are gradually obtained.

[0059] Specific defect identification: In the middle layer of the network, the residual network (ResNet) structure is used to identify subtle features such as scratches, cracks, corrosion, etc. Through multi-layer feature fusion technology, even small defects can be identified at multiple scales.

[0060] S1.4, after the defect feature extraction is completed, the complexity parameters of multiple dimensions are constructed according to the extracted defect type and area, depth, distribution location and other information. Based on these parameters, the geometric morphology features and energy distribution features are calculated and obtained, where the geometric morphology features include the fractal dimension of the defect edge, and the energy distribution features include the surface energy dispersion coefficient.

[0061] Among them, the process of obtaining the fractal dimension of the defect edge is as follows:

[0062] Use an image sensor to obtain the original grayscale image of the metal gasket surface , ensuring clear capture of subtle defect features.

[0063] Apply anisotropic diffusion filter Smooths the image, preserving edge details while reducing noise. ; Represents the image processed by an anisotropic diffusion filter, reducing noise while preserving edges; represents the preprocessed image;

[0064] The Marr-Hildreth edge detection algorithm is used, and the Laplace-Gaussian operator is used to Convolution is performed on the preprocessed image to extract edge information and obtain an edge response image. ;in, is a Gaussian function; is the scale parameter; Represents an edge response image.

[0065] Use adaptive threshold Convert edge response image to binary covered edge image : ;

[0066] Place the covered edge image at a scale of On a square grid, the initial box size is .

[0067] Define a set of box dimensions ,in , until the minimum box size is reached . Indicates the number of box dimensions; It represents the index of the scale, i.e. The size of a box.

[0068] For each scale , calculate the number of non-empty boxes needed to cover the edge image , the following conditions are met: ;in, , Indicates There are edge points in the box. Refers to the side length of the smallest circumscribed square area of ​​the edge covering the image; Indicated in scale The number of non-empty boxes required to cover the edge image; Indicates The size of the box, ; Indicated in scale The total number of boxes required to cover the entire edge image area.

[0069] by is the horizontal axis, As the vertical axis, draw a relationship diagram in the double logarithmic coordinate system.

[0070] The least squares method is used to perform linear fitting on the data points. The slope of the fitted line is the fractal dimension of the defect edge. : ;in Indicates the number of different box scales used for the computation.

[0071] The fractal dimension of the defect edge is used to represent and reflect the complexity and irregularity of the defect edge on the surface of the metal gasket, mainly reflecting the self-similarity and detail richness of the defect geometry at different scales. In the process of automated inspection of metal gaskets, defects often have complex geometric features. Especially when defects such as scratches, cracks and corrosion show nonlinear and subtle changes, simple geometric size descriptions are not sufficient to accurately reflect the nature of the defects. By calculating the fractal dimension of the defect edge, the complexity of the edge can be quantified, thereby revealing the changes and anomalies of the defect in the microstructure. The higher the fractal dimension, the more complex and irregular the defect edge is, which usually means that the defect has a greater impact on the structural integrity of the gasket. In this scenario, the introduction of fractal dimension not only improves the accuracy of detection, but also provides an important basis for subsequent defect classification and resource allocation, helping the system to reasonably allocate detection resources and adjust detection strategies according to the complexity of the defect.

[0072] The process of obtaining the surface energy dispersion coefficient is as follows:

[0073] Using the gradient energy method, the local energy of each pixel on the gasket surface is calculated by Sobolev gradient. The specific formula is: ;in, Indicates that the image is The local energy of the position is based on the square sum of the transverse gradient and the longitudinal gradient of the image, reflecting the intensity of the change of the local energy on the gasket surface; represents partial derivative;

[0074] The surface of the metal gasket is divided into several regions of equal area, denoted as , each region contains a set of pixels For each region , calculate its total energy , the formula is as follows: ; Indicates sub-regions; Indicates area The set of pixels in ; where Indicates area The total energy of the defect reflects the energy distribution caused by defects in the area.

[0075] Based on the local energy distribution, the energy changes between adjacent regions are analyzed. Define adjacent regions and The energy divergence between , the formula is as follows: ;in, and represent the total energy of adjacent regions, For Region and The length of the boundary between the two regions measures the impact of energy changes between the two regions on their boundary.

[0076] The energy dispersion of all adjacent areas is combined to calculate the surface energy dispersion coefficient of the entire gasket surface , to measure the degree of change and unevenness of surface energy distribution. The formula is as follows: ;in, is the energy divergence of adjacent regions, is the total number of regions. By performing the square sum operation on the energy divergence of each region, the influence of the region with large energy variation on the overall energy divergence can be strengthened, reflecting the difference in energy distribution of defective regions on the surface of the metal gasket.

[0077] The energy dispersion coefficient is used to represent and reflect the uneven energy distribution between different areas on the surface of the metal gasket, especially the energy difference between adjacent areas and the degree of change on the boundary. In the automated inspection scenario of metal gaskets, surface defects such as cracks, scratches and corrosion can cause energy anomalies in local areas, and these energy fluctuations are reflected in the image energy distribution through gradient changes. The energy dispersion coefficient measures the amplitude of energy changes caused by these defects and their propagation effects between local areas by calculating the energy difference and boundary length of adjacent areas. The higher the coefficient, the more uneven the energy distribution on the gasket surface, which means that the defects have a greater impact on the gasket structure, especially the degree of energy anomaly diffusion caused by local defects is more significant. Therefore, the energy dispersion coefficient can not only quantify the energy effect of local defects, but also provide an important basis for subsequent defect classification and detection strategies, effectively improving the accuracy and efficiency of complex defect detection.

[0078] The structural specific redundancy index is obtained by comprehensively calculating the defect edge fractal dimension and energy dispersion coefficient. , for example, can be calculated by the following formula: ;in, and are weight parameters respectively.

[0079] : Used to adjust the contribution of the product term of the defect edge fractal dimension and the surface energy divergence coefficient. This parameter controls the impact of the direct interaction between the fractal dimension and the energy divergence on the final calculation results. The larger the weight, the more significant the role of the interaction term between the two in the overall structure-specific redundancy index.

[0080] : Used to adjust the contribution of the sum of squares of fractal dimension and energy divergence. The weight controls the weight of the squared values ​​of fractal dimension and energy divergence in the overall calculation. The larger the weight, the more weight the independent characteristics of fractal dimension and energy divergence (rather than their interaction) will account for in the structure-specific redundancy index.

[0081] The structure-specific redundancy index is mainly used to analyze the complexity of surface defects of metal gaskets and help the detection system dynamically adjust the detection strategy and resource allocation. The index quantifies the impact of defects on the overall detection complexity of gaskets by integrating the geometric complexity of the defect edge and the unevenness of energy distribution. The larger the value, the more complex the defect morphology and the more significant the energy anomaly, requiring more time and resources for detailed detection to ensure high-precision identification and processing; the smaller the value, the relatively simple defect, the lower the detection complexity, and can be processed in a conventional way. The index plays a role in priority sorting and resource optimization allocation during the detection process, ensuring that complex defects receive sufficient detection resources while improving detection efficiency.

[0082] Step S2 includes the following contents:

[0083] In order to distinguish metal gaskets of different complexity, it is first necessary to set a reasonable threshold based on historical test data and experimental results. . This threshold is determined based on the detection requirements of the metal gasket and the processing capacity of the equipment. Usually, the threshold can be defined by statistically analyzing the distribution of the structure-specific redundancy index values ​​in historical data. The specific method is to use quantile analysis or to determine the appropriate value based on the upper limit of detection resource consumption in actual detection, so that it can effectively distinguish between high-complexity fuzzy signals and low-complexity clear signals.

[0084] For each metal gasket to be inspected, its structure-specific redundancy index is calculated through the S1 process. This index quantifies the complexity of gasket surface defects, combines the edge fractal dimension and surface energy dispersion coefficient, and reflects its complexity and resource requirements.

[0085] The structure-specific redundancy index of each gasket was compared with the set threshold Real-time comparison. In specific operations, the index value can be dynamically read and automatically compared through a pre-programmed control system or algorithm. The comparison formula is as follows: ;

[0086] Ambiguous signal: If , indicating that the defect complexity of the corresponding metal gasket is high, and the defect characteristics are difficult to accurately detect by conventional means. At this time, a more detailed analysis is required, so it is judged as a fuzzy signal.

[0087] Clear signal: If , it indicates that the defect of the corresponding metal gasket is relatively simple or not obvious, and can be detected through conventional detection procedures and determined as a clear signal.

[0088] Step S3 includes the following contents:

[0089] S3.1, when the detection system determines that the metal gasket is a fuzzy signal, the adjustment mechanism of the temporary accommodation space is triggered first. The temporary accommodation space is located before the assembly line and serves as a buffer for the redundant gaskets to be detected. Based on the value of the current structure-specific redundancy index, the additional accommodation space that needs to be reserved for subsequent metal gaskets is calculated. , to avoid buildup during detection of obscure signal pads.

[0090] The adjustment formula is as follows: ;in: Indicates the new temporary storage space size; Indicates the base holding space, which indicates the default buffer space size.

[0091] Through the nonlinear adjustment of the square of the structure-specific redundancy index value, the temporary accommodation space is dynamically increased to ensure that when the fuzzy signal gasket enters the pipeline, the system can reserve enough buffer space for the subsequent gaskets to be detected.

[0092] S3.2, while adjusting the temporary accommodation space, adjust the interval between the fuzzy signal gasket that has entered the detection area and the subsequent gaskets on the assembly line through the structure-specific redundancy index. The interval adjustment is dynamically matched according to the load and complexity of the current detection area to ensure that the gasket with higher complexity has sufficient detection time and the subsequent gasket will not interfere with its detection process.

[0093] The interval adjustment formula is as follows: ;in: Indicates the adjusted shim spacing; Indicates the base interval, which indicates the system default spacer interval; Represents the adjustment coefficient, which is used to dynamically adjust the interval according to the load of the detection system.

[0094] The interval adjustment is performed according to the cubic relationship of the structure-specific redundancy index to ensure that the fuzzy signal shims with higher complexity can obtain sufficient analysis time in the pipeline detection area without interfering with the subsequent shims.

[0095] S3.3, in order to ensure that the adjustment of the accommodation space matches the interval adjustment, further adjustments are made according to the signal type (fuzzy signal or clear signal) after the fuzzy signal gasket detection is completed. If the subsequent gasket is detected as a clear signal, the temporary accommodation space is dynamically compressed according to the need to shorten the interval. At this time, the size of the temporary accommodation space Reduced according to the aforementioned reduction in the spacing between the gaskets.

[0096] The compression formula for temporary accommodation space is as follows: ;in: Indicates the adjusted size of the temporary storage space.

[0097] The size of the compressed accommodation space is calculated by the ratio of the interval reduction to ensure that the subsequent clear signal gaskets can enter the assembly line smoothly and will not accumulate due to insufficient space, while ensuring the continuity and efficiency of the entire assembly line.

[0098] The present invention comprehensively considers the geometric complexity of the defect edge and the uneven energy distribution, accurately quantifies the defect complexity of the metal gasket, and grades them according to the evaluation results. For gaskets judged as fuzzy signals, the accommodation space adjustment mechanism is triggered, the size of the temporary accommodation space in front of the assembly line is dynamically adjusted, the buffer area is increased, and the gasket interval is adjusted to ensure that gaskets with complex defects obtain sufficient detection time and resource investment. After the detection is completed, the temporary accommodation space is automatically compressed according to the signal type of the subsequent gasket, and the transmission of clear signal gaskets is accelerated to ensure the continuity of the assembly line. The entire process realizes the accurate detection of complex defects and the rational allocation of resources through the dynamic adjustment of the structure-specific redundancy index, which not only improves the detection accuracy, but also ensures the efficient operation of the assembly line and reduces quality risks.

[0099] Embodiment 2: Figure 2 The invention provides an intelligent metal gasket automatic detection and grading system, including: a defect extraction module, a grading determination module and a detection adjustment module;

[0100] Defect extraction module: collects image data of the metal gasket surface through image sensors, applies deep learning algorithms based on convolutional neural networks, extracts defect features, analyzes the defect complexity of each gasket, generates a structure-specific redundancy index, and sends the structure-specific redundancy index to the grading judgment module.

[0101] Grading judgment module: compares the structure-specific redundancy index with the set threshold, and judges the signal exceeding the set threshold as a fuzzy signal, and the signal below the threshold as a clear signal. The obtained fuzzy signal and clear signal are sent to the detection adjustment module.

[0102] Detection adjustment module: For fuzzy signal gaskets, the accommodation space adjustment mechanism is triggered, and the temporary accommodation space before the pipeline is dynamically adjusted according to the structure-specific redundancy index, the buffer space is increased and the gasket interval is adjusted to ensure that the fuzzy signal gasket has sufficient detection time.

[0103] 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 in the formula are set by technicians in this field according to actual conditions.

[0104] 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.

[0105] It should be noted that, in this article, if there are relational terms such as first and second, etc., they are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "including a..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0106] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. Intelligent metal gasket automatic detection and grading method, characterized in that: Includes steps: S1, collects image data of the metal gasket surface through image sensors, applies deep learning algorithms based on convolutional neural networks, extracts defect features, analyzes the defect complexity of each gasket, and generates a structure-specific redundancy index; S2, comparing the structure-specific redundancy index with the set threshold, those exceeding the set threshold are judged as fuzzy signals, and those not exceeding the threshold are judged as clear signals; S3, for the fuzzy signal gasket, trigger the accommodation space adjustment mechanism, dynamically adjust the temporary accommodation space before the pipeline according to the structure-specific redundancy index, increase the buffer space and adjust the gasket interval to ensure that the fuzzy signal gasket has enough detection time; Step S1 includes the following contents: S1.1, use relevant equipment to scan the surface of each metal gasket at multiple angles to obtain holographic image data of the gasket surface; S1.2, after image data acquisition, the image is first subjected to noise filtering and contrast enhancement; S1.3, after the image preprocessing is completed, the defect features of the image are automatically extracted through the convolutional neural network model; S1.4, after the defect feature extraction is completed, the geometric morphology feature and the energy distribution feature are calculated and obtained, wherein the geometric morphology feature includes the fractal dimension of the defect edge, and the energy distribution feature includes the surface energy dispersion coefficient; Among them, the process of obtaining the fractal dimension of the defect edge is as follows: Use an image sensor to obtain the original grayscale image of the metal gasket surface ; Apply anisotropic diffusion filter Smooth the image to preserve edge details while reducing noise; ; Represents the image processed by an anisotropic diffusion filter, reducing noise while preserving edges; represents the preprocessed image; The Marr-Hildreth edge detection algorithm is used, and the Laplace-Gaussian operator is used to Convolve the preprocessed image to extract edge information and obtain an edge response image; ;in, is a Gaussian function; is the scale parameter; represents the edge response image; Use adaptive threshold Convert edge response image to binary covered edge image : ; Place the covered edge image at a scale of On a square grid, the initial box size is ; Define a set of box dimensions ,in , until the minimum box size is reached ; Indicates the number of box dimensions; It represents the index of the scale, i.e. The size of the box; For each scale , calculate the number of non-empty boxes needed to cover the edge image , the following conditions are met: ;in, , Indicates There are edge points in the box. Refers to the side length of the smallest circumscribed square area of ​​the edge covering the image; Indicated in scale The number of non-empty boxes required to cover the edge image; Indicates The size of the box, ; Indicated in scale The total number of boxes required to cover the entire edge image area; by is the horizontal axis, As the vertical axis, draw a relationship diagram in the double logarithmic coordinate system; The least squares method is used to perform linear fitting on the data points. The slope of the fitted line is the fractal dimension of the defect edge. : ;in represents the number of different box scales used for the calculation; The process of obtaining the surface energy dispersion coefficient is as follows: Using the gradient energy method, the local energy of each pixel on the gasket surface is calculated by Sobolev gradient; the specific formula is: ;in, Indicates that the image is The local energy of the position is based on the square sum of the transverse gradient and the longitudinal gradient of the image, reflecting the intensity of the change of the local energy on the gasket surface; represents partial derivative; The surface of the metal gasket is divided into several regions of equal area, denoted as , each region contains a set of pixels ; For each region , calculate its total energy , the formula is as follows: ; Indicates sub-regions; Indicates area The set of pixels in ; where Indicates area Total energy; Based on the local energy distribution, analyze the energy changes between adjacent areas; define adjacent areas and The energy divergence between , the formula is as follows: ;in, and represent the total energy of adjacent regions, For Region and The length of the boundary between the two regions measures the impact of energy changes between the two regions on their boundary; The energy dispersion of all adjacent areas is combined to calculate the surface energy dispersion coefficient of the entire gasket surface. , to measure the degree of change and unevenness of surface energy distribution; the formula is as follows: ;in, is the energy divergence of adjacent regions, is the total number of regions; The structural specific redundancy index is obtained by comprehensively calculating the defect edge fractal dimension and energy dispersion coefficient. Step S2 includes the following contents: The structure-specific redundancy index of each gasket was compared with the set threshold Perform real-time comparison; the comparison formula is as follows: ; Ambiguous signal: If , indicating that the defect complexity of the corresponding metal gasket is high, and the defect characteristics are difficult to accurately detect by conventional means. At this time, a more detailed analysis is required, so it is determined to be a fuzzy signal; Clear signal: If , it indicates that the defect of the corresponding metal gasket is relatively simple or not obvious, and the detection is completed through the conventional detection process and determined as a clear signal; Step S3 includes the following contents: S3.1, when the metal gasket is determined to be a fuzzy signal, the adjustment mechanism of the temporary accommodation space is first triggered; the temporary accommodation space is located before the pipeline and serves as a buffer for the redundant gaskets to be detected; based on the value of the current structure-specific redundancy index, the additional accommodation space that needs to be reserved for subsequent metal gaskets is calculated ; The adjustment formula is as follows: ;in: Indicates the new temporary storage space size; Indicates the base accommodation space, indicating the default buffer space size; S3.2, while adjusting the temporary accommodation space, adjust the interval between the fuzzy signal gasket that has entered the detection area and the subsequent gaskets on the assembly line through the structure-specific redundancy index; the interval adjustment is dynamically matched according to the load and complexity of the current detection area to ensure that the gaskets with higher complexity have sufficient detection time; the interval adjustment formula is as follows: ;in: Indicates the adjusted shim spacing; Indicates the base interval, which indicates the system default spacer interval; Represents the adjustment coefficient, which is used to dynamically adjust the interval according to the load of the detection system; S3.3, in order to ensure that the adjustment of the accommodation space matches the interval adjustment, the adjustment is made according to the signal type after the fuzzy signal gasket detection is completed; if the subsequent gasket is detected as a clear signal, the temporary accommodation space is dynamically compressed according to the need to shorten the interval; the compression formula of the temporary accommodation space is as follows: ;in: Indicates the adjusted size of the temporary storage space.

2. An intelligent metal gasket automatic detection and grading system, used to implement the intelligent metal gasket automatic detection and grading method according to claim 1, characterized in that: include: Defect extraction module, classification determination module and detection adjustment module; Defect extraction module: collects image data of the metal gasket surface through image sensors, applies deep learning algorithms based on convolutional neural networks, extracts defect features, analyzes the defect complexity of each gasket, generates a structure-specific redundancy index, and sends the structure-specific redundancy index to the grading judgment module; Grading judgment module: compares the structure-specific redundancy index with the set threshold, and judges the signal exceeding the set threshold as a fuzzy signal, and the signal not exceeding the threshold as a clear signal, and sends the obtained fuzzy signal and clear signal to the detection adjustment module; Detection adjustment module: For fuzzy signal gaskets, the accommodation space adjustment mechanism is triggered, and the temporary accommodation space before the pipeline is dynamically adjusted according to the structure-specific redundancy index, the buffer space is increased and the gasket interval is adjusted to ensure that the fuzzy signal gasket has sufficient detection time.

Images