Intelligent production method and equipment of composite insulation board and medium

Abstract

The embodiment of the invention discloses an intelligent production method and equipment of a composite insulation board, a medium, application and an intelligent production technology, which are used for solving the problem of unstable quality of an existing production mode. The method comprises the following steps: determining target product data of a to-be-detected composite insulation board according to workshop order information and historical detection data; detecting the defect connected graph and the boundary contour according to a preset camera and ultrasonic equipment, performing regional grouping on the defect graph in combination with the layer structure data, inputting a classification model to obtain a first detection result, and comparing the boundary contour to determine a second detection result. And positioning an optimizable production link based on the first detection result and the second detection result, performing simulation operation on an optimizable production environment by using a digital twin model, and identifying variable bottleneck parameters. And inputting the variable bottleneck parameters, the production link parameter range and historical production data into a least square model, calculating tuning data, and adjusting the composite insulation board according to the tuning data to realize closed-loop production.

Description

Technical Field

[0001] The present invention relates to the field of intelligent production technology, and in particular to an intelligent production method, equipment and medium for a composite insulation board. Background Art

[0002] Composite insulation board is a new type of building material that integrates insulation, fireproofing and waterproofing. It is widely used in building exterior walls, roofs and other parts. With the increasing global awareness of energy conservation and environmental protection, the construction industry, as one of the main areas of energy consumption and carbon emissions, is facing unprecedented transformation pressure. As an efficient, environmentally friendly and energy-saving building material, composite insulation board has been widely used in modern buildings due to its excellent insulation, sound insulation and fireproofing properties. In the production process of composite insulation board, quality control is a crucial link, and the most fundamental guarantee for stable quality is the control of production process. Being able to timely detect the quality of composite insulation board and adjust the production process is an important means to ensure the stable quality of composite insulation board.

[0003] However, in the current production process of composite insulation boards, the production parameters of composite insulation boards are adjusted mainly based on manual experience to ensure the quality of composite insulation boards. However, due to the influence of manual experience, the quality of different batches of composite insulation boards may vary greatly, resulting in unstable quality of composite insulation boards. Moreover, the method that relies on manual experience lacks the support of real-time production information, making it difficult to make timely adjustments to production. In addition, the method based on sensor detection lacks a real-time feedback mechanism, making it difficult to flexibly adjust the production of composite insulation boards based on actual production quality issues. Summary of the invention

[0004] In order to solve the above technical problems, one or more embodiments of the present specification provide an intelligent production method, equipment and medium for a composite insulation board.

[0005] One or more embodiments of this specification adopt the following technical solutions:

[0006] One or more embodiments of the present specification provide an intelligent production method for a composite insulation board, the method comprising:

[0007] According to the order information and historical inspection data corresponding to the current workshop, the target product data of the composite insulation board to be inspected is determined; wherein the target product data includes: layer structure data and target boundary data;

[0008] According to the preset camera and the preset ultrasonic equipment of the current workshop, the defect connectivity map and the boundary contour of the composite insulation board to be inspected are detected and obtained;

[0009] The defect connectivity graph is grouped by regions according to the layer structure data to obtain a region set corresponding to each of the defect connectivity graphs, so as to input the defect connectivity graph into a classification model corresponding to each of the region sets, obtain a first detection result of the composite thermal insulation board to be detected, and determine a second detection result of the composite thermal insulation board to be detected according to the boundary contour and the target boundary contour of the composite thermal insulation board to be detected;

[0010] According to the first test result and the second test result, locate the optimizable production link of the composite thermal insulation board to be tested, simulate the operation of the optimizable production link based on the preset digital twin model, and identify the variable bottleneck parameters of the optimizable production link;

[0011] The variable bottleneck parameters, the range of each parameter in the optimizable production link and the historical production data are input into a preset least squares model to obtain the tuning data of the optimizable production link, and the production of the composite insulation board is adjusted according to the tuning data.

[0012] Optionally, in one or more embodiments of the present specification, according to the preset camera and the preset ultrasonic equipment of the current workshop, detecting and obtaining the defect connectivity map and the boundary contour of the composite insulation board to be detected specifically includes:

[0013] Based on the preset camera of the product inspection workshop, a product video of the current composite insulation board corresponding to the composite insulation board to be inspected is obtained, and a detection target image of the composite insulation board to be inspected is determined according to the product video;

[0014] Acquire a grayscale image of the detection target image, determine a surface defect connectivity graph of the current composite thermal insulation board according to the grayscale image, and perform convolution processing on the grayscale image by using a plurality of convolution templates corresponding to a preset multi-directional edge detection operator to obtain a boundary contour of the composite thermal insulation board to be detected;

[0015] The ultrasonic sequence images collected by the preset ultrasonic equipment are obtained to determine the defect connectivity map of the composite thermal insulation board to be inspected based on the ultrasonic sequence images and the surface defect connectivity map.

[0016] Optionally, in one or more embodiments of the present specification, determining the detection target image of the composite thermal insulation board to be detected according to the product video specifically includes:

[0017] Acquire adjacent image sequences in the product video, and determine the acquisition interval of each adjacent image sequence according to the moving speed of the conveyor belt in the current workshop and the acquisition frequency of the adjacent image sequences; wherein the acquisition interval is in a linear proportional relationship;

[0018] Based on the acquisition interval and the camera imaging principle of the product video, scaling each of the adjacent image sequences to obtain a scaled adjacent image sequence;

[0019] Performing differential processing on each of the scaled adjacent image sequences to obtain a differential image, and determining the target image to be detected in each of the differential images according to the intersection-and-union ratio of each of the differential images;

[0020] Comparing the specifications of the composite thermal insulation board to be detected with the target images to be detected in each of the differential images, classifying each of the target images to be detected, and obtaining a first target image to be detected with occlusion and a second target image to be detected without occlusion;

[0021] If the composite thermal insulation board to be detected corresponds to a first target image to be detected, a plurality of first target images to be detected in continuous frame images are acquired based on the video image, and the first target images to be detected are superimposed to obtain a superimposed image;

[0022] According to the number of superpositions and the pixel value of each pixel point in the superposition image, the superposition image is averaged to obtain a complete detection target image of the first target image to be detected;

[0023] The complete detection target image and the second target image to be detected are summarized to determine the detection target image of the composite thermal insulation board to be detected.

[0024] Optionally, in one or more embodiments of the present specification, a plurality of convolution templates corresponding to preset multi-directional edge detection operators are used to perform convolution processing on the grayscale image to obtain the boundary contour of the composite insulation board to be detected, specifically including:

[0025] Acquire a standard image of the composite thermal insulation board to be inspected, and perform block processing on the standard image and the grayscale image to obtain a standard sub-image and a grayscale sub-image;

[0026] Determining a regional structural similarity value between the standard sub-image and the grayscale sub-image, and determining an overall structural similarity value between the standard sub-image and the grayscale sub-image based on a mean value of the regional structural similarity value;

[0027] Selecting a preset multi-directional edge detection operator in the multi-directional edge detection operator template according to the overall structural similarity value;

[0028] Performing convolution processing on the grayscale image based on multiple convolution templates corresponding to the preset multi-directional edge detection operator to obtain gradient data of each pixel point in the grayscale image; wherein the gradient data includes: gradient direction and gradient value;

[0029] The edge pixel points of the complete detection target image are determined based on the gradient value and the preset gradient threshold, the contour edge data of the complete detection target image is determined based on the position and gradient direction of the edge pixel points, and the contour edge data are sequentially connected to determine the boundary contour of the composite insulation board to be detected.

[0030] Optionally, in one or more embodiments of the present specification, obtaining the ultrasonic sequence images collected by the preset ultrasonic device to determine the defect connectivity map of the composite thermal insulation board to be inspected based on the ultrasonic sequence images and the surface defect connectivity map specifically includes:

[0031] Collecting each of the composite thermal insulation boards to be tested according to the ultrasonic equipment preset in the current workshop to obtain an ultrasonic sequence image; wherein the ultrasonic sequence image includes a plurality of ultrasonic images;

[0032] Acquire the grayscale value of each ultrasonic image pixel in the ultrasonic sequence images to determine the sudden change pixel of the ultrasonic sequence images based on the difference between each pixel and adjacent pixel points;

[0033] According to the positions of the mutation pixel points and the adjacent mutation pixel points of the mutation pixel points, the bending direction and the curvature of each of the mutation pixel points are determined, and the pixel points to be fitted corresponding to each of the mutation pixel points are determined according to the bending direction and the curvature, so as to obtain an ultrasonic defect connectivity map by fitting the pixel points to be fitted based on a preset fitting algorithm;

[0034] Based on the union of the ultrasonic defect connectivity map and the surface defect connectivity map, the defect connectivity map of the composite thermal insulation board to be inspected is determined.

[0035] Optionally, in one or more embodiments of the present specification, according to the first test result and the second test result, the optimizable production link of the composite insulation board to be tested is located, and the operation of the optimizable production link is simulated based on a preset digital twin model to identify the variable bottleneck parameters of the optimizable production link, specifically including:

[0036] Based on the first detection result and the second detection result, determine the current defect type, defect position and defect parameter value of the composite insulation board to be detected; wherein the current defect type includes: pore type, crack type, inclusion type, loose type, interface crack type;

[0037] Determine the allowable range of defect parameters of each current defect type based on the current defect type and the historical factory data of the composite thermal insulation board to be tested;

[0038] Determine the first defect level of the composite thermal insulation board to be detected according to the order information of the composite thermal insulation board to be detected and the defect position, and determine the second defect level of the composite thermal insulation board to be detected according to the defect parameter value and the allowable range of the defect parameter;

[0039] If it is determined that the quality of the composite thermal insulation board to be inspected is unqualified based on the first defect level and the second defect level, locating the optimizable production link of the composite thermal insulation board to be inspected based on the current defect type;

[0040] The operation of the optimizable production link is simulated by using the preset digital twin model corresponding to the production workshop to identify the variable bottleneck parameters of the optimizable production link.

[0041] Optionally, in one or more embodiments of the present specification, before simulating the operation of the optimizable production link based on the preset digital twin model and identifying the variable bottleneck parameters of the optimizable production link, the method further includes:

[0042] Collect the workshop data and building data of the production workshop corresponding to the composite thermal insulation board to be tested; wherein the workshop data includes: equipment data, material data, and environmental data;

[0043] Determine the physical entity data in the workshop data, so as to input the physical data and the building data into a preset modeling tool to obtain an initial digital twin model corresponding to the production workshop;

[0044] Determine a mapping relationship between the twin data in the workshop data and the initial digital twin model, so as to add the twin data to the initial digital twin model based on the mapping relationship to obtain a current digital twin model;

[0045] The workshop operation data of the current digital twin model is compared with the actual workshop operation data of the production workshop, so as to adjust the model parameters of the current digital twin model based on the comparison result to obtain the preset digital twin model.

[0046] Optionally, in one or more embodiments of the present specification, before inputting the variable bottleneck parameter, the range of each parameter in the optimizable production link, and the historical production data into a preset least squares model to obtain the tuning data of the optimizable production link, and adjusting the production of the composite insulation board according to the tuning data, the method further includes:

[0047] Acquire historical tuning data of the production workshop, and sort the historical tuning data based on time sequence to obtain the historical tuning data sequence; wherein the historical tuning data includes variable bottleneck parameters corresponding to the tuning process, the range of each parameter in the optimizable production link, historical production data and tuning data;

[0048] Normalizing the historical tuning data sequence, and using the normalized historical tuning data sequence as training set data;

[0049] Determining an input term expression and an output term expression of a least squares model according to a preset delay time and a preset embedding dimension of a production data time series, so as to construct an initial least squares model according to the input term expression and the output term expression;

[0050] The initial least squares model is trained based on the training set data to obtain the preset least squares model.

[0051] One or more embodiments of this specification provide an intelligent production device for a composite insulation board, the device comprising:

[0052] at least one processor; and,

[0053] a memory communicatively connected to the at least one processor; wherein,

[0054] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to: execute any of the above-mentioned methods.

[0055] One or more embodiments of the present specification provide a non-volatile computer storage medium storing computer executable instructions, wherein the computer executable instructions are configured to execute any of the above-described methods.

[0056] At least one of the above technical solutions adopted in the embodiments of this specification can achieve the following beneficial effects:

[0057] Combining the first test result and the second test result, the optimizable production links of the composite insulation board to be tested can be accurately located, providing a clear direction for subsequent tuning work, and then simulating and optimizing the optimizable production links through the digital twin model, identifying the variable bottleneck parameters, so as to quickly respond to and improve the production process. Through the simulation and optimization of the digital twin model, the production links can be adjusted and improved without affecting the actual production, avoiding rework and waste caused by quality problems, improving production efficiency and reducing production costs. This process forms a closed loop of continuous improvement, and by continuously simulating, optimizing and verifying the production links, it can promote the optimization and upgrading of product production. In addition, using the preset camera and ultrasonic equipment, the defect connectivity map and boundary contour of the composite insulation board to be tested can be obtained non-contact and efficiently, avoiding the damage and errors that may be caused by traditional detection methods. According to the layer structure data, the defect connectivity map is grouped into regions, and the classification model is used for defect identification, which can achieve accurate classification and positioning of different types of defects, providing strong support for production tuning. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions in the embodiments of this specification or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art description. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. In the drawings:

[0059] Figure 1 A schematic diagram of a process flow of an intelligent production method of a composite thermal insulation board provided in an embodiment of this specification;

[0060] Figure 2 A schematic diagram of the structure of an intelligent production device for a composite thermal insulation board provided in an embodiment of this specification;

[0061] Figure 3 A schematic diagram of the structure of a non-volatile storage medium provided in an embodiment of this specification. DETAILED DESCRIPTION

[0062] The embodiments of this specification provide an intelligent production method, equipment and medium for a composite insulation board.

[0063] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this specification.

[0064] like Figure 1 As shown, the embodiment of this specification provides a schematic diagram of the intelligent production method of a composite insulation board. Figure 1 It can be seen that in one or more embodiments of this specification, an intelligent production method of a composite insulation board specifically includes:

[0065] S101: Determine target product data of the composite insulation board to be tested according to the order information and historical test data corresponding to the current workshop; wherein the target product data includes: layer structure data and target boundary data.

[0066] Composite insulation boards are usually composed of multi-layer structures, such as insulation layers, waterproof layers, and anti-cracking layers. The layer structure data describes in detail the key information of these layers, such as the material, thickness, and arrangement order. Based on these layer structure data, it is helpful to judge whether the composite insulation board has quality problems such as delamination or cracks when conducting quality inspection of the composite insulation board. The boundary data usually includes data such as the size and shape of the composite insulation board, which can ensure the precise match between the produced composite insulation board and the building wall. Therefore, in the intelligent production process corresponding to, for example, the composite micro-nano flame retardant board (GPDF) exterior wall fire insulation system and the composite micro-nano flame retardant board (GPDF) external formwork cast-in-place concrete insulation system, in order to determine that the produced composite insulation board can meet the quality requirements, it is necessary to determine the target product data of the composite insulation board to be inspected based on the order information and historical inspection data corresponding to the current workshop.

[0067] S102: According to the preset camera and preset ultrasonic equipment of the current workshop, a defect connectivity map and a boundary contour of the composite thermal insulation board to be inspected are detected and obtained.

[0068] Specifically, in one or more embodiments of the present specification, according to the preset camera and preset ultrasonic equipment in the current workshop, the defect connectivity map and boundary contour of the composite insulation board to be detected are detected and obtained, which specifically includes the following process:

[0069] First, based on the preset camera in the product inspection workshop, the product video of the current composite insulation board corresponding to the composite insulation board to be inspected is obtained, and the inspection target image of the composite insulation board to be inspected is determined based on the product video. By using the camera and ultrasonic equipment for inspection, no physical damage will be caused to the composite insulation board, avoiding the problem of finished product waste caused by destructive inspection. Then, the grayscale image of the inspection target image is obtained, and then the surface defect connectivity map of the current composite insulation board is determined according to the grayscale image, and the grayscale image is convoluted through multiple convolution templates corresponding to the preset multi-directional edge detection operator to obtain the boundary contour of the composite insulation board to be inspected. The ultrasonic sequence image collected by the preset ultrasonic equipment is obtained to determine the defect connectivity map of the composite insulation board to be inspected based on the ultrasonic sequence image and the surface defect connectivity map. In this process, the camera can capture the detailed information of the composite insulation board, and the defect connectivity map and boundary contour can be accurately extracted through image processing technology. The ultrasonic equipment can penetrate the composite insulation board, detect internal defects, and improve the accuracy and comprehensiveness of the inspection. By detecting the defect connectivity map and boundary contour of the composite insulation board, quality problems existing in the production process can be discovered in time, such as cracks, bubbles, delamination and other defects, which helps to timely discover potential safety hazards such as crack expansion, delamination and detachment, and ensure the safety and reliability of building use.

[0070] Further, in one or more embodiments of the present specification, determining the detection target image of the composite thermal insulation board to be detected according to the product video specifically includes the following process:

[0071] Acquire the adjacent image sequence in the product video, and then determine the acquisition spacing of each adjacent image sequence according to the moving speed of the conveyor belt in the current workshop and the acquisition frequency of the adjacent image sequence; wherein, it should be noted that the acquisition spacing is in a linear proportional relationship. Then, based on the acquisition spacing and the camera imaging principle of the product video, scale each adjacent image sequence to obtain the scaled adjacent image sequence. According to the camera imaging principle, the adjacent image sequence is scaled to ensure that the size and proportion of the image are consistent with the actual bottle body, thereby reducing the redundant error caused by the differential calculation due to different sizes, and improving the accuracy of subsequent classification. Then, the scaled adjacent image sequence is differentially processed to obtain a differential image, and the target image to be detected in each differential image is determined according to the intersection and union ratio of each differential image. Compare the specifications of the composite insulation board to be detected with the target image to be detected in each differential image. If the specifications of the target image to be detected and the composite insulation board to be detected are inconsistent, it can be determined that the target to be detected is blocked, that is, each target image to be detected can be classified to obtain a first target image to be detected with occlusion and a second target image to be detected without occlusion. If the composite insulation board to be detected corresponds to the first target image to be detected, in order to fill the obstructed part of the composite insulation board body bottle to be detected, multiple first target images to be detected in the continuous frame image will be obtained based on the video image, that is, with the help of the movement of the production line, the composite insulation board to be detected that is obstructed changes in position, and its display position in the camera field of view also changes. Therefore, the first target image to be detected is superimposed in the product video to obtain a superimposed image. Then, according to the number of superpositions and pixel values ​​of each pixel point in the superimposed image, the superimposed image is averaged to obtain a complete detection target image of the first target image to be detected. By summarizing the complete detection target image and the second target image to be detected, the detection target image of each composite insulation board to be detected can be determined. The above process not only improves the accuracy of detection, but also enhances the ability to handle occlusion situations, providing strong support for the quality control of composite insulation boards.

[0072] Further, in one or more embodiments of the present specification, a plurality of convolution templates corresponding to the preset multi-directional edge detection operator are used to perform convolution processing on the grayscale image to obtain the boundary contour of the composite insulation board to be detected, which specifically includes the following process:

[0073] First, obtain a standard image of the composite insulation board to be detected, and perform block processing on the standard image and the grayscale image to obtain a standard sub-image and a grayscale sub-image. Then, determine the regional structural similarity value between the standard sub-image and the grayscale sub-image, and determine the overall structural similarity value between the standard sub-image and the grayscale sub-image based on the mean of the regional structural similarity value. The defects of the composite insulation board material itself and the problem of unclear edges are fully considered by obtaining the overall structural similarity. Then, select the preset multi-directional edge detection operator in the multi-directional edge detection operator template according to the overall structural similarity value. Among them, it should be noted that the preset multi-directional edge detection operator can be a sobel operator or other operators such as canny. After obtaining the preset multi-directional edge detection operator, the grayscale image is convolved according to the multiple convolution templates corresponding to the preset multi-directional edge detection operator, so as to obtain the gradient data of each pixel point in the grayscale image; among them, it should be noted that the gradient data includes: gradient direction and gradient value. Then, the edge pixel points of the complete detection target image are determined according to the gradient value and the preset gradient threshold, so as to determine the contour edge data of the complete detection target image according to the position of the edge pixel points and the gradient direction, and the contour edge data are sequentially connected to determine the boundary contour of the composite insulation board to be detected. By adaptively obtaining a preset multi-directional edge detection operator based on the overall structural similarity value, the extraction of edge pixel points is more adapted to the actual situation of the composite insulation board to be detected, thereby improving the accuracy of detection.

[0074] Furthermore, in order to avoid the problem that the existing defects identified by machine vision are only surface defects and it is difficult to identify internal defects, resulting in the problem of missed identification of product defects, in one or more embodiments of this specification, an ultrasonic sequence image collected by a preset ultrasonic device is obtained to determine the defect connectivity map of the composite insulation board to be inspected based on the ultrasonic sequence image and the surface defect connectivity map, specifically including the following process:

[0075] First, the ultrasonic sequence images are obtained by collecting data on each composite insulation board to be tested according to the ultrasonic equipment preset in the current workshop. It should be noted that the ultrasonic sequence images contain multiple ultrasonic images. In a certain application scenario, the ultrasonic sequence images are obtained by collecting data on each composite insulation board to be tested according to the ultrasonic equipment preset in the current workshop. The process is as follows: the initial ultrasonic sequence images are obtained by collecting data on each composite insulation board to be tested according to the preset ultrasonic equipment; it should be noted that the initial ultrasonic sequence images contain multiple initial ultrasonic images. Then, the initial ultrasonic sequence images are normalized and standardized to obtain the standard ultrasonic sequence images corresponding to the initial ultrasonic sequence images. The standard ultrasonic sequence images are then filtered according to the pre-set bilateral filter to obtain the denoised ultrasonic sequence images of the standard ultrasonic sequence images. Since the detail information and structural information in the image are distributed in different frequency ranges, in order to more accurately obtain the detail information and improve the accuracy of defect recognition, and then effectively control the factory instructions of the product, the denoised ultrasonic sequence image is subjected to discrete wavelet transform in the embodiment of this specification, so as to obtain multiple detail image information and multiple structural image information of the denoised image, and the detail image information and the structural image are secondary decomposed according to the preset low-pass and high-pass filters, so as to obtain detail image information components and structural image information components distributed at different levels. In order to perform enhancement processing on different information components, the image contrast value corresponding to each frequency range is determined in the embodiment of this specification, so as to determine the adaptive enhancement function of each denoised image in each frequency range according to the image contrast value and the preset contrast threshold, and then the structural image information component and the detail image information component are processed according to the adaptive enhancement function to obtain the first enhancement component and the second enhancement component. The information components are stratified through multi-level cascade decomposition, so as to realize adaptive enhancement processing of the information components, which is helpful for the enhancement processing of denoised ultrasonic sequence images, that is, after performing discrete wavelet inverse transform on the first enhancement component and the second enhancement component, the enhanced sequence image of the denoised ultrasonic sequence image is obtained, and the enhanced sequence image is used as the ultrasonic sequence image.

[0076] Then, in order to determine the defects such as scratches, bubbles, uneven thickness, etc. on the composite insulation board based on the ultrasonic sequence image, the grayscale value of each ultrasonic image pixel in the ultrasonic sequence image is obtained, and then according to the difference between each pixel and the adjacent pixel, if there is a pixel and the adjacent pixel The difference is greater than the preset difference threshold, then the pixel is determined to be a mutation pixel in the ultrasonic sequence image. Then, the mutation pixel points determined in the ultrasonic sequence image are extracted and the adjacent mutation pixel points of each mutation pixel point are determined. According to the position of the mutation pixel point and the adjacent mutation pixel point, the bending direction and curvature of each mutation pixel point are determined. Therefore, according to the bending direction and curvature of each mutation pixel point, the pixel point to be fitted corresponding to each mutation pixel point is determined, so as to fit the pixel point to be fitted based on the preset fitting algorithm to obtain an ultrasonic defect connectivity map. Then, according to the union of the ultrasonic defect connectivity map and the surface defect connectivity map, the defect connectivity map of the composite insulation board to be detected is determined. By taking the union of the ultrasonic defect connectivity map and the initial defect connectivity map, two different defect detection results can be comprehensively considered, and the defect connectivity map of the composite insulation board to be inspected can be more comprehensively identified, thereby improving the comprehensiveness and accuracy of defect detection, and avoiding the problem of unreliable quality inspection caused by missed judgments when quality inspection is based on machine vision and only based on product image recognition.

[0077] S103: performing region grouping on the defect connectivity map according to the layer structure data, obtaining a region set corresponding to each of the defect connectivity maps, inputting the defect connectivity map into a classification model corresponding to each of the region sets, obtaining a first detection result of the composite insulation board to be detected, and determining a second detection result of the composite insulation board to be detected according to the boundary contour and the target boundary contour of the composite insulation board to be detected.

[0078] In order to make each classification model focus on processing a specific type of defect or a specific area of ​​defects, so as to improve the pertinence and accuracy of the model. In the embodiment of this specification, the defect connectivity map will be regionally grouped according to the layer structure data, that is, the boundary position of each layer of the insulation board to be detected is determined based on the layer structure data, and the defect connectivity map is regionally grouped based on the boundary position and the position corresponding to the defect connectivity map, so as to obtain the region set corresponding to each defect connectivity map. Then the defect connectivity map is input into the classification model corresponding to each region set to obtain the first detection result of the composite insulation board to be detected. That is, by analyzing each region through a special classification model, false positives and false negatives can be reduced, and the overall detection accuracy can be improved. Among them, it should be noted that the classification model is obtained after training the existing neural network model based on the defect connectivity map sample, and its training process will not be specifically described here. In addition, by comparing the boundary contour with the target boundary contour of the composite insulation board to be detected, the detection result can be further refined to identify more specific defect positions and shapes. Based on the boundary contour and the target boundary contour of the composite insulation board to be detected, the second detection result of the composite insulation board to be detected is determined. The boundary contour is extracted through the previous steps, and the target boundary contour can be a predefined standard boundary contour or a target boundary determined according to the requirements. By comparing the boundary contour and the target boundary contour, it can be determined whether the composite insulation board to be inspected meets the requirements. For example, if the boundary contour completely matches the target boundary contour, then the second detection result can be "the boundary contour is consistent with the target boundary contour and meets the requirements." If there is an inconsistency between the boundary contour and the target boundary contour, then the second detection result is a result containing the defect type, defect location and defect parameter value. Through the above process, the first detection result and the second detection result of the composite insulation board to be inspected can be obtained. Through the joint judgment of the boundary contour and the defect connectivity graph, the effective detection of multi-dimensional quality defects of the composite insulation board to be inspected is realized, and the coverage and reliability of quality inspection are improved by comprehensively evaluating the quality of the composite insulation board.

[0079] S104: Based on the first test result and the second test result, locate the optimizable production link of the composite insulation board to be tested, simulate the operation of the optimizable production link based on the preset digital twin model, and identify the variable bottleneck parameters of the optimizable production link.

[0080] In order to accurately locate the problematic links in the production process of composite insulation boards, feedback adjustment can be implemented to achieve the purpose of intelligent production of composite insulation boards. In the embodiment of this specification, by combining the first test result and the second test result, the problematic links in the production process of composite insulation boards can be accurately located to point to specific production links. Then, based on the simulation operation of the digital twin model, the variable bottleneck parameters can be identified, that is, those parameters that have a key impact on production efficiency and product quality. By optimizing and adjusting these parameters, production efficiency can be significantly improved while ensuring the stability and consistency of product quality. In addition, the optimization of production links can also reduce unnecessary resource consumption, such as raw materials, energy and manpower, and further reduce production costs.

[0081] Specifically, in one or more embodiments of the present specification, according to the first detection result and the second detection result, the optimizable production link of the composite thermal insulation board to be detected is located, and the operation of the optimizable production link is simulated based on the preset digital twin model, and the variable bottleneck parameters of the optimizable production link are identified, which specifically includes:

[0082] Based on the first test result and the second test result obtained in the above steps, the current defect type, defect location and defect parameter value of the composite insulation board to be tested are determined; wherein the current defect type includes: pore type, crack type, inclusion type, loose type and interface crack type, etc. Then, according to the current defect type and the historical factory data of the composite insulation board to be tested, the allowable range of defect parameters of each current defect type is determined. Then, according to the order information and defect location of the composite insulation board to be tested, the first defect level of the composite insulation board to be tested is determined, and according to the defect parameter value and the allowable range of the defect parameter, the second defect level of the composite insulation board to be tested is determined. For example: assuming that in a certain embodiment, the first test result: defect type such as delamination, bubble, crack, etc.; defect location such as board surface, edge, inside, etc.; defect parameter value such as length, width, depth, area, etc.; second test result: may involve the matching degree of boundary contour, such as edge unevenness, dimensional deviation, etc. Then, according to the current defect type and the historical factory data of the composite insulation board, the allowable range of defect parameters of each current defect type is determined. This is usually done by analyzing a large amount of historical data to derive the qualified standards for each defect type under different conditions. Next, the first defect level of the composite insulation board to be tested is determined based on the order information and defect location of the composite insulation board to be tested. For example, if the order requires that the board surface must be flat and defect-free, and the actual test results show that the board surface has delamination, then the board surface is judged to be the first defect level. At the same time, the second defect level of the composite insulation board to be tested is determined based on the defect parameter value and the allowable range of the defect parameter. For example, if the allowable range of the delamination length is between 0mm and 5mm and is considered qualified, and the actual test results show that the delamination length is 6mm, then the delamination defect level is the second defect level. Finally, the overall quality level of the composite insulation board to be tested is determined by superimposing the first defect level and the second defect level. For example, if the first defect level is specifically that the A-level board surface has serious defects, and the second defect level is specifically determined to be B-level delamination length of 6mm, then the overall quality level at this time can be determined as C-level unqualified. If the quality of the composite insulation board to be tested is determined to be unqualified according to the first defect level and the second defect level, the optimizable production link of the composite insulation board to be tested is located according to the current defect type. Then, the operation of the optimizable production link is simulated through the preset digital twin model corresponding to the production workshop, so as to identify the variable bottleneck parameters of the optimizable production link.

[0083] Once it is determined that the quality of the composite insulation board is unqualified in this process, the optimizable production links of the composite insulation board to be tested can be quickly located according to the current defect type, and the variable bottleneck parameters can be identified through simulation and optimization through the digital twin model, so as to quickly respond to and improve the production process. Through the simulation and optimization of the digital twin model, adjustments and improvements can be made to the production links without affecting the actual production, avoiding rework and waste caused by quality problems, improving production efficiency and reducing production costs. This process forms a closed loop of continuous improvement, which can promote the optimization and upgrading of product production by continuously simulating, optimizing and verifying the production links.

[0084] Furthermore, in one or more embodiments of the present specification, before simulating the operation of the production link that can be optimized based on the preset digital twin model and identifying the variable bottleneck parameters of the production link that can be optimized, the method also includes the following process:

[0085] First, collect the workshop data and building data of the production workshop corresponding to the composite insulation board to be tested; it should be noted that the workshop data includes: equipment data, material data and environmental data, etc. Then, determine the physical entity data in the workshop data, and input the physical data and building data into the preset modeling tool to obtain the initial digital twin model corresponding to the production workshop. After determining the mapping relationship between the twin data in the workshop data and the initial digital twin model, the twin data can be added to the initial digital twin model according to the mapping relationship to obtain the current digital twin model. Then, compare the workshop operation data of the current digital twin model with the actual workshop operation data of the production workshop, so as to adjust the model parameters of the current digital twin model based on the comparison results and obtain the preset digital twin model.

[0086] The physical entity data in the workshop data is determined, and it is input into the preset modeling tool together with the building data to obtain the initial digital twin model corresponding to the production workshop. This fusion method enables the model to not only include the operating status of the production equipment, but also consider the impact of building structure and environmental factors on production, thereby improving the accuracy and practicality of the model. The mapping relationship between the twin data in the workshop data and the initial digital twin model is determined, and the twin data is added to the initial digital twin model according to the mapping relationship to obtain the current digital twin model. This dynamic mapping mechanism ensures that the model can reflect the changes in the production workshop in real time, providing a reliable basis for subsequent simulation and optimization. After obtaining the preset digital twin model, the operation of the production link that can be optimized can be simulated based on the model, so as to quickly locate the variable bottleneck parameters. This fast positioning method makes the production optimization process more efficient and accurate, and reduces unnecessary trial and error costs. Through the simulation and optimization of the digital twin model, problems in the production process can be discovered and solved in a timely manner, avoiding rework and waste caused by quality problems. At the same time, the optimized production links can operate more efficiently, improving production efficiency and reducing production costs.

[0087] S105: Input the variable bottleneck parameter, the range of each parameter in the optimizable production link and the historical production data into a preset least squares model to obtain the tuning data of the optimizable production link, and adjust the production of the composite insulation board according to the tuning data.

[0088] After obtaining the variable bottleneck parameters based on the above steps, in order to achieve adjustments to production, the variable bottleneck parameters, the range of each parameter in the optimizable production link, and the historical production data are input into the preset least squares model in the embodiment of this specification, thereby obtaining the tuning data for the optimizable production link, and adjusting the production of the composite insulation board based on the tuning data.

[0089] Further, in one or more embodiments of the present specification, before inputting the variable bottleneck parameter, the range of each parameter in the optimizable production link, and the historical production data into the preset least squares model to obtain the tuning data for optimizing the production link, and adjusting the production of the composite insulation board according to the tuning data, the method further includes the following process:

[0090] First, the historical tuning data of the production workshop is obtained, and the historical tuning data is sorted according to the time sequence to obtain the historical tuning data sequence. It should be noted that the historical tuning data includes the variable bottleneck parameters corresponding to the tuning process, the range of each parameter in the production link that can be optimized, the historical production data and the tuning data. Then, the historical tuning data sequence is normalized, and the normalized historical tuning data sequence is used as the training set data. According to the preset delay time and the preset embedding dimension of the production data time series, the input term expression and the output term expression of the least squares model are determined to construct the initial least squares model according to the input term expression and the output term expression. Then, the initial least squares model is trained according to the training set data to obtain the preset least squares model. In this process, by obtaining the historical tuning data of the production workshop and sorting it according to the time sequence, the temporal continuity and integrity of the data can be ensured, providing a reliable basis for subsequent model training. Normalizing the historical tuning data sequence can eliminate the dimensional differences between different parameters, so that the model can more fairly consider the influence of each parameter during the training process, thereby improving the accuracy and generalization ability of the model. By selecting the appropriate delay time and embedding dimension, the dynamic characteristics and nonlinear relationships in the production process can be fully captured, thereby improving the prediction accuracy and tuning effect of the model. The trained preset least squares model can quickly and accurately provide tuning data based on the current variable bottleneck parameters, the range of parameters in the production link that can be optimized, and historical production data. This allows production managers to control each link in the production process more accurately, thereby improving product quality and production efficiency.

[0091] like Figure 2 As shown, the present specification provides a schematic diagram of the structure of an intelligent production device for a composite thermal insulation board. Figure 2 It can be seen that in one or more embodiments of this specification, an intelligent production device for a composite insulation board includes:

[0092] at least one processor; and,

[0093] a memory communicatively connected to the at least one processor; wherein,

[0094] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to: execute any of the above-mentioned methods.

[0095] like Figure 3 As shown, the present specification provides a schematic diagram of the structure of a non-volatile storage medium. Figure 3It can be seen that in one or more embodiments of the present specification, a non-volatile storage medium stores computer executable instructions 301, and the computer executable instructions 301 can execute any of the above-mentioned methods.

[0096] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0097] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0098] The above description is only one or more embodiments of this specification and is not intended to limit this specification. For those skilled in the art, one or more embodiments of this specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of this specification shall be included in the scope of the claims of this specification.

Claims

1. An intelligent production method for a composite insulation board, characterized in that: The method comprises: According to the order information and historical inspection data corresponding to the current workshop, the target product data of the composite insulation board to be inspected is determined; wherein the target product data includes: layer structure data and target boundary data; According to the preset camera and the preset ultrasonic equipment of the current workshop, the defect connectivity map and the boundary contour of the composite insulation board to be inspected are detected and obtained; The defect connectivity graph is grouped by regions according to the layer structure data to obtain a region set corresponding to each of the defect connectivity graphs, so as to input the defect connectivity graph into a classification model corresponding to each of the region sets, obtain a first detection result of the composite thermal insulation board to be detected, and determine a second detection result of the composite thermal insulation board to be detected according to the boundary contour and the target boundary contour of the composite thermal insulation board to be detected; According to the first test result and the second test result, locate the optimizable production link of the composite thermal insulation board to be tested, simulate the operation of the optimizable production link based on the preset digital twin model, and identify the variable bottleneck parameters of the optimizable production link; The variable bottleneck parameters, the range of each parameter in the optimizable production link and the historical production data are input into a preset least squares model to obtain the tuning data of the optimizable production link, and the production of the composite insulation board is adjusted according to the tuning data.

2. The intelligent production method of a composite thermal insulation board according to claim 1, characterized in that: According to the preset camera and the preset ultrasonic equipment of the current workshop, the defect connectivity map and the boundary contour of the composite insulation board to be detected are detected and obtained, specifically including: Based on the preset camera of the product inspection workshop, a product video of the current composite insulation board corresponding to the composite insulation board to be inspected is obtained, and a detection target image of the composite insulation board to be inspected is determined according to the product video; Acquire a grayscale image of the detection target image, determine a surface defect connectivity graph of the current composite thermal insulation board according to the grayscale image, and perform convolution processing on the grayscale image by using a plurality of convolution templates corresponding to a preset multi-directional edge detection operator to obtain a boundary contour of the composite thermal insulation board to be detected; The ultrasonic sequence images collected by the preset ultrasonic equipment are obtained to determine the defect connectivity map of the composite thermal insulation board to be inspected based on the ultrasonic sequence images and the surface defect connectivity map.

3. The intelligent production method of a composite thermal insulation board according to claim 2, characterized in that: Determining the detection target image of the composite thermal insulation board to be detected according to the product video specifically includes: Acquire adjacent image sequences in the product video, and determine the acquisition interval of each adjacent image sequence according to the moving speed of the conveyor belt in the current workshop and the acquisition frequency of the adjacent image sequences; wherein the acquisition interval is in a linear proportional relationship; Based on the acquisition interval and the camera imaging principle of the product video, scaling each of the adjacent image sequences to obtain a scaled adjacent image sequence; Performing differential processing on each of the scaled adjacent image sequences to obtain a differential image, and determining the target image to be detected in each of the differential images according to the intersection-and-union ratio of each of the differential images; Comparing the specifications of the composite thermal insulation board to be detected with the target images to be detected in each of the differential images, classifying each of the target images to be detected, and obtaining a first target image to be detected with occlusion and a second target image to be detected without occlusion; If the composite thermal insulation board to be detected corresponds to a first target image to be detected, a plurality of first target images to be detected in continuous frame images are acquired based on the video image, and the first target images to be detected are superimposed to obtain a superimposed image; According to the number of superpositions and the pixel value of each pixel point in the superposition image, the superposition image is averaged to obtain a complete detection target image of the first target image to be detected; The complete detection target image and the second target image to be detected are summarized to determine the detection target image of the composite thermal insulation board to be detected.

4. The intelligent production method of a composite thermal insulation board according to claim 3, characterized in that: By presetting a plurality of convolution templates corresponding to a multi-directional edge detection operator, the grayscale image is convoluted to obtain the boundary contour of the composite thermal insulation board to be detected, specifically including: Acquire a standard image of the composite thermal insulation board to be inspected, and perform block processing on the standard image and the grayscale image to obtain a standard sub-image and a grayscale sub-image; Determining a regional structural similarity value between the standard sub-image and the grayscale sub-image, and determining an overall structural similarity value between the standard sub-image and the grayscale sub-image based on a mean value of the regional structural similarity value; Selecting a preset multi-directional edge detection operator in the multi-directional edge detection operator template according to the overall structural similarity value; Performing convolution processing on the grayscale image based on multiple convolution templates corresponding to the preset multi-directional edge detection operator to obtain gradient data of each pixel point in the grayscale image; wherein the gradient data includes: gradient direction and gradient value; The edge pixel points of the complete detection target image are determined based on the gradient value and the preset gradient threshold, the contour edge data of the complete detection target image is determined based on the position and gradient direction of the edge pixel points, and the contour edge data are sequentially connected to determine the boundary contour of the composite insulation board to be detected.

5. The intelligent production method of a composite thermal insulation board according to claim 2, characterized in that: Acquiring the ultrasonic sequence images collected by the preset ultrasonic equipment to determine the defect connectivity map of the composite thermal insulation board to be inspected based on the ultrasonic sequence images and the surface defect connectivity map, specifically includes: Collecting each of the composite thermal insulation boards to be tested according to the ultrasonic equipment preset in the current workshop to obtain an ultrasonic sequence image; wherein the ultrasonic sequence image includes a plurality of ultrasonic images; Acquire the grayscale value of each ultrasonic image pixel in the ultrasonic sequence images to determine the sudden change pixel of the ultrasonic sequence images based on the difference between each pixel and adjacent pixel points; According to the positions of the mutation pixel points and the adjacent mutation pixel points of the mutation pixel points, the bending direction and the curvature of each of the mutation pixel points are determined, and the pixel points to be fitted corresponding to each of the mutation pixel points are determined according to the bending direction and the curvature, so as to obtain an ultrasonic defect connectivity map by fitting the pixel points to be fitted based on a preset fitting algorithm; Based on the union of the ultrasonic defect connectivity map and the surface defect connectivity map, the defect connectivity map of the composite thermal insulation board to be inspected is determined.

6. The intelligent production method of a composite thermal insulation board according to claim 1, characterized in that: According to the first test result and the second test result, the optimizable production link of the composite thermal insulation board to be tested is located, and the operation of the optimizable production link is simulated based on the preset digital twin model to identify the variable bottleneck parameters of the optimizable production link, specifically including: Based on the first detection result and the second detection result, determine the current defect type, defect position and defect parameter value of the composite insulation board to be detected; wherein the current defect type includes: pore type, crack type, inclusion type, loose type, interface crack type; Determine the allowable range of defect parameters of each current defect type based on the current defect type and the historical factory data of the composite thermal insulation board to be tested; Determine the first defect level of the composite thermal insulation board to be detected according to the order information of the composite thermal insulation board to be detected and the defect position, and determine the second defect level of the composite thermal insulation board to be detected according to the defect parameter value and the allowable range of the defect parameter; If it is determined that the quality of the composite thermal insulation board to be inspected is unqualified based on the first defect level and the second defect level, locating the optimizable production link of the composite thermal insulation board to be inspected based on the current defect type; The operation of the optimizable production link is simulated by using the preset digital twin model corresponding to the production workshop to identify the variable bottleneck parameters of the optimizable production link.

7. The intelligent production method of a composite thermal insulation board according to claim 1, characterized in that: Before simulating the operation of the optimizable production link based on the preset digital twin model and identifying the variable bottleneck parameters of the optimizable production link, the method further includes: Collect the workshop data and building data of the production workshop corresponding to the composite thermal insulation board to be tested; wherein the workshop data includes: equipment data, material data, and environmental data; Determine the physical entity data in the workshop data, so as to input the physical data and the building data into a preset modeling tool to obtain an initial digital twin model corresponding to the production workshop; Determine a mapping relationship between the twin data in the workshop data and the initial digital twin model, so as to add the twin data to the initial digital twin model based on the mapping relationship to obtain a current digital twin model; The workshop operation data of the current digital twin model is compared with the actual workshop operation data of the production workshop, so as to adjust the model parameters of the current digital twin model based on the comparison result to obtain the preset digital twin model.

8. The intelligent production method of composite thermal insulation board according to claim 1, characterized in that , inputting the variable bottleneck parameter, the range of each parameter in the optimizable production link and the historical production data into the preset least squares model to obtain the tuning data of the optimizable production link, and before adjusting the production of the composite insulation board according to the tuning data, the method also includes: Acquire historical tuning data of the production workshop, and sort the historical tuning data based on time sequence to obtain the historical tuning data sequence; wherein the historical tuning data includes variable bottleneck parameters corresponding to the tuning process, the range of each parameter in the optimizable production link, historical production data and tuning data; Normalizing the historical tuning data sequence, and using the normalized historical tuning data sequence as training set data; Determining an input term expression and an output term expression of a least squares model according to a preset delay time and a preset embedding dimension of a production data time series, so as to construct an initial least squares model according to the input term expression and the output term expression; The initial least squares model is trained based on the training set data to obtain the preset least squares model.

9. An intelligent production equipment for composite insulation board, comprising: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can: execute any of the methods described in claims 1-8.

10. A non-volatile storage medium storing computer executable instructions, characterized in that: The computer executable instructions can: execute the method described in any one of claims 1 to 8.

Images