Intelligent control method for heat insulation sheet grinding machine

Abstract

The present invention discloses an intelligent control method for a heat insulation plate grinder, which relates to the field of intelligent control technology, including: obtaining K grinding process nodes and K grinding process node parameters; performing grinding defect frequent item mining to obtain K grinding defect data sets; performing defect degree analysis to obtain K process node abnormalities; retrieving the K grinding defect data sets, and respectively screening the search results for control points to obtain K feedback control point sets; performing process node grinding quality identification on the target heat insulation plate that is polished according to the K grinding process node parameters to obtain a grinding quality identification result; judging whether the grinding quality identification result meets the grinding process node quality requirement, and if not, performing feedback adjustment on the target grinder. The present invention solves the technical problems of insufficient grinding accuracy and poor quality consistency in the prior art, and achieves the technical effect of improving the grinding quality of the heat insulation plate.

Description

Technical Field

[0001] The invention relates to the technical field of intelligent control, and in particular to an intelligent control method for a heat insulation sheet grinder. Background Art

[0002] With the rapid development of industrial manufacturing technology, thermal insulation sheets, as a key material, have been widely used in aerospace, automobile manufacturing, electronic equipment and other fields. However, with the growing demand for thermal insulation sheets and the increasing requirements for product performance, its manufacturing process is also facing higher technical challenges. The traditional thermal insulation sheet grinding method is difficult to adapt to the complex and changing production environment due to its reliance on fixed parameters, especially in precision manufacturing and large-scale production scenarios, exposing many problems.

[0003] The existing thermal insulation sheet grinding process usually has technical bottlenecks such as insufficient grinding accuracy and poor quality consistency. For example, since it is impossible to dynamically adjust the grinding parameters according to the material properties and shape differences of the thermal insulation sheet, it is easy to cause problems such as excessive surface roughness, obvious edge burrs, and uneven thickness distribution. In addition, once quality problems are found during the production process, manual parameter adjustments are often required, which is time-consuming and labor-intensive, and the adjustment effect is unstable. These problems seriously affect the manufacturing quality and production efficiency of the thermal insulation sheet. Summary of the invention

[0004] The present application provides an intelligent control method for a heat insulation sheet grinder, which is used to solve the technical problems of insufficient grinding accuracy and poor quality consistency in the prior art.

[0005] In view of the above problems, the present application provides an intelligent control method for a thermal insulation sheet grinder.

[0006] The present application provides an intelligent control method for a heat insulation sheet grinding machine, the method comprising:

[0007] Based on the basic information of the target thermal insulation sheet, the grinding process node configuration is performed to obtain K grinding process nodes and K grinding process node parameters, wherein K is an integer greater than or equal to 1; based on the basic information and the K grinding process nodes, the grinding defect frequent item mining is performed to obtain K grinding defect data sets; the K grinding defect data sets are traversed to perform defect degree analysis to obtain K process node abnormalities; the K grinding defect data sets are retrieved with the defect position and defect degree as indexes, and the search results are respectively screened for control points to obtain K feedback control point sets; based on the K feedback control point sets and the preset grinding index set, the process node grinding quality identification is performed on the target thermal insulation sheet that is polished by the target grinder at the K grinding process nodes according to the K grinding process node parameters to obtain K process node grinding quality identification results; it is judged whether the K process node grinding quality identification results meet the K grinding process node quality requirements, and if not, feedback adjustment is performed on the target grinder based on the K process node grinding quality identification results.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] The present application configures the grinding process nodes based on the basic information of the target thermal insulation sheet, obtains K grinding process nodes and K grinding process node parameters, wherein K is an integer greater than or equal to 1; performs grinding defect frequent item mining based on the basic information and the K grinding process nodes, and obtains K grinding defect data sets; traverses the K grinding defect data sets to perform defect degree analysis, and obtains K process node abnormalities; retrieves the K grinding defect data sets with defect location and defect degree as indexes, and performs control point screening on the retrieval results respectively, and obtains K feedback control point sets; based on the K feedback control point sets and the preset grinding index set, performs process node grinding quality identification on the target thermal insulation sheet that is polished by the target grinder at the K grinding process nodes according to the K grinding process node parameters, and obtains K process node grinding quality identification results; determines whether the K process node grinding quality identification results meet the K grinding process node quality requirements, and if not, performs feedback adjustment on the target grinder based on the K process node grinding quality identification results. The present invention solves the technical problems of insufficient grinding accuracy and poor quality consistency in the prior art. By constructing grinding process nodes based on the basic information of the target thermal insulation sheet, combined with frequent item mining of grinding defects, defect degree analysis, control point screening and quality identification, accurate feedback control and dynamic adjustment of the thermal insulation sheet grinding process are achieved, thereby achieving the technical effect of improving the grinding quality of the thermal insulation sheet. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0011] Figure 1 A schematic flow chart of an intelligent control method for a heat insulation sheet grinding machine provided in an embodiment of the present application;

[0012] Figure 2 A schematic diagram of a flow chart for obtaining abnormality degrees of K process nodes in an intelligent control method for a heat insulation plate grinder provided in an embodiment of the present application. DETAILED DESCRIPTION

[0013] The present application provides an intelligent control method for a thermal insulation sheet grinder, which is used to solve the technical problems of insufficient grinding accuracy and poor quality consistency in the prior art. By constructing a grinding process node based on the basic information of the target thermal insulation sheet, combined with frequent item mining of grinding defects, defect degree analysis, control point screening and quality identification, accurate feedback control and dynamic adjustment of the thermal insulation sheet grinding process are achieved, thereby achieving the technical effect of improving the grinding quality of the thermal insulation sheet.

[0014] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0015] It should be noted that any variations of the terms "include" and "have" are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules that are not explicitly listed or inherent to these processes, methods, products or devices.

[0016] Examples, such as Figure 1 As shown, the present application provides an intelligent control method for a heat insulation sheet grinder, the method comprising:

[0017] Step S100: Perform polishing process node configuration based on basic information of the target thermal insulation sheet to obtain K polishing process nodes and K polishing process node parameters, where K is an integer greater than or equal to 1.

[0018] In the embodiment of the present application, the basic information of the thermal insulation sheet, including the size specifications and edge profile, is first collected by a 3D scanner to ensure a comprehensive understanding of the physical properties of the thermal insulation sheet. The surface roughness is then measured using a surface profiler to assess the initial quality of the thermal insulation sheet surface. Next, the thickness uniformity is tested using an ultrasonic thickness gauge to determine the precise data of the thermal insulation sheet thickness distribution.

[0019] Then, based on the basic information collected, the grinding process is configured with reference to the preset rule library. The rule library provides process node recommendations for different materials, sizes, and roughness. For example, for an aluminum thermal insulation sheet, if the initial roughness is Ra6.5µm, the target roughness is Ra0.8µm, and there are burrs on the edge profile, the rule library recommends dividing the process into three nodes (K=3): rough grinding, semi-finishing grinding, and finishing grinding.

[0020] Finally, set specific grinding parameters for each process node. For example, in the rough grinding stage, select the abrasive size of P80, set the grinding pressure to 8N, and the grinding speed to 3000rpm; in the semi-finishing stage, select the size of P320, set the pressure to 5N, and the speed to 2000rpm; in the fine grinding stage, select the size of P1000, reduce the pressure to 2N, and the speed to 1000rpm. The selection of each parameter is based on the corresponding best practice case in the rule base.

[0021] Through the above steps, K polishing process nodes and K polishing process node parameters are obtained, where K is an integer greater than or equal to 1.

[0022] Step S200: performing grinding defect frequent item mining based on the basic information and the K grinding process nodes to obtain K grinding defect data sets.

[0023] In the embodiment of the present application, historical data consistent with the current basic information and K grinding process nodes are first extracted from the historical database. These data include historical grinding records that match the material characteristics, size specifications, surface roughness of the target thermal insulation sheet, and the grinding parameters (such as pressure, speed, and abrasive particle size) of each process node, as well as the defect types (such as scratches, pits, burrs) corresponding to these conditions and their locations. These historical data are automatically collected and stored by the equipment recording system in the production process to ensure high relevance to the current grinding task.

[0024] The extracted historical data is then preprocessed. First, missing value filling techniques are used to handle missing items in the data, such as filling missing grinding parameters based on the historical mean of similar conditions. Subsequently, the parameter data is standardized using the normalization method to convert grinding parameters of different dimensions (such as pressure and speed) into a unified range to ensure the accuracy and reliability of subsequent mining and analysis.

[0025] After preprocessing, the historical data is analyzed using association rule mining algorithms (such as the Apriori algorithm). Through this algorithm, the data of each process node is mined to extract the most common defect patterns in each node. For example, in the historical data of the rough grinding stage, it may be found that edge burrs are easily generated under conditions of larger abrasive grains and higher pressure; in the historical data of the fine grinding stage, it may be found that higher grinding speeds are prone to cause fine scratches. The mined results reveal the high-frequency occurrence conditions and characteristics of defects in each process node.

[0026] The mining results are then organized into K grinding defect data sets. Each set corresponds to a grinding process node, including the defect type, the location of the defect, and the grinding parameters related to the defect. For example, for the rough grinding node, the defect data set may include edge burrs (the associated parameters are particle size P80 and pressure 8N) and excessive surface roughness (the associated parameters are particle size P120 and pressure 6N). For the fine grinding node, the defect data set may include fine scratches (the associated parameters are speed 3000rpm and pressure 2N). These data sets provide the distribution of defects and their cause analysis for each process node.

[0027] Through the above steps, K polishing defect data sets are mined based on historical data that is consistent with the current basic information and process nodes.

[0028] Step S300: traverse the K polishing defect data sets to perform defect degree analysis and obtain K process node abnormality degrees.

[0029] Further, such as Figure 2 As shown, in the method provided in the embodiment of the application, the K polishing defect data sets are traversed to perform defect degree analysis to obtain K process node abnormalities, and the method also includes:

[0030] Based on a preset set of standard values ​​of polishing indicators, the K polishing defect data sets are traversed to perform defect degree identification to obtain K polishing defect degree sets; the means of the K polishing defect degree sets are respectively calculated to obtain K polishing defect degree means; the K polishing defect degree means are taken as K starting center anomalies, and the K polishing defect degree sets are centrally searched to obtain the K process node anomalies.

[0031] In the embodiment of the present application, when traversing K grinding defect data sets for defect degree analysis, firstly, based on the preset grinding index standard value set, each grinding defect data set is traversed one by one, and the difference between the defect data and the corresponding standard value is calculated, and the difference is divided by the standard value to obtain the defect degree of each defect. For example, if the standard value of surface roughness is Ra0.8µm, and an actual detection value is Ra1.2µm, the defect degree is calculated as (1.2-0.8) / 0.8=0.5. This process quantifies each defect data and generates K grinding defect degree sets.

[0032] Then, the mean of each grinding defect set is calculated, and the defect mean of each process node is calculated using a simple arithmetic mean method. The mean reflects the average severity of all defects in the process node. For example, for the defect set {0.2, 0.5, 0.4} of a process node, its mean is calculated as (0.2+0.5+0.4) / 3=0.367. Through this step, K grinding defect means are obtained.

[0033] Then, the average of K polishing defects is used as the abnormality of the starting center, the positive direction axis of the defect is constructed, the defect value is mapped to a real point on the axis, and the initial center neighborhood is constructed around the starting center according to the preset leap step. The diffusion center neighborhood is generated by diffusing the neighborhood edges to the left and right, and the diffusion center neighborhood is combined with the initial center neighborhood for iterative analysis to finally determine the target center neighborhood.

[0034] Finally, the set of real number points of defect degree contained in the neighborhood of the target center is counted, and the values ​​of these points are averaged to obtain the abnormality degrees of K process nodes.

[0035] Furthermore, the method provided in the application embodiment also includes:

[0036] Obtain an initialized positive direction number axis, take the K starting center abnormalities as K number axis center points respectively, and construct K defect positive direction number axes in combination with the K polishing defect sets, wherein each polishing defect is a real number point on the corresponding defect positive direction number axis; construct K initial center neighborhoods of the K number axis center points on the K defect positive direction number axes according to a preset leap step; perform neighborhood edge diffusion in the left and right directions for the K initial center neighborhoods according to the preset leap step respectively to obtain K diffusion center neighborhoods; perform iterative analysis on the K diffusion center neighborhoods and the K initial center neighborhoods to obtain K target center neighborhoods; traverse and count the K real number point sets contained in the K target center neighborhoods, and perform mean calculation on the K polishing defect degree sets corresponding to the K real number point sets respectively to obtain the K process node abnormalities.

[0037] In an embodiment of the present application, the positive direction axis is first initialized, with the mean of K polishing defects as the starting center, which is used as the center point of the positive direction axis. The positive direction axis is established by a mathematical modeling method to intuitively represent the defect distribution, and each polishing defect value is mapped to a real point on the axis. These points reflect the defect degree distribution of each process node, forming K positive direction axes. Then, according to the preset spanning step size (for example, 0.1 or 0.2), an initial neighborhood of a fixed range is defined around the center point of the axis. For example, when the center point is 0.5 and the step size is 0.1, the initial neighborhood range is [0.4, 0.6]. All real points within the neighborhood are counted to form K initial center neighborhoods.

[0038] Then, the neighborhood edges of the K initial center neighborhoods are diffused to the left and right directions according to the preset step length. The diffusion process uses the step length iteration method to expand the neighborhood range to the left and right sides with the set step length, for example, expanding the neighborhood from [0.4, 0.6] to [0.3, 0.7] to form a diffusion center neighborhood. After each expansion, the defect data points within the expansion range are re-counted to capture a wider distribution area. After the diffusion is completed, K diffusion center neighborhoods are obtained.

[0039] Then, the neighborhood density calculation and iterative analysis are performed. The neighborhood density (i.e., the ratio of the number of defect data points to the width of the neighborhood range) of each initial center neighborhood and diffusion center neighborhood is calculated. If the diffusion neighborhood density is greater than or equal to the initial neighborhood density, the diffusion center neighborhood is used as the iteration center neighborhood and continues to expand; otherwise, the expansion is stopped and the current initial center neighborhood is used as the target center neighborhood. After multiple iterative analyses, K target center neighborhoods are finally obtained.

[0040] Next, we traverse and count the defect data point sets within each target center neighborhood. These sets correspond to all defect distribution points within the target center neighborhood. The K defect real point sets obtained by statistics are matched with the original K polishing defect sets respectively, and the relevant data points are extracted to ensure the accuracy of the correlation. The mean of the defect data points in each set is calculated to generate the abnormality of each process node. For example, if the defect data points contained in the target center neighborhood are {0.4, 0.5, 0.6}, the corresponding abnormality is calculated as (0.4+0.5+0.6) / 3=0.5.

[0041] Finally, the traversal statistics and calculation of the neighborhoods of the K target centers are completed, and the abnormality of the K process nodes is obtained.

[0042] Furthermore, in the method provided in the embodiment of the application, the K diffusion center neighborhoods and the K initial center neighborhoods are iteratively analyzed to obtain K target center neighborhoods, and further includes:

[0043] The neighborhood densities of the K diffuse center neighborhoods and the K initial center neighborhoods are calculated respectively to obtain K diffuse neighborhood densities and K initial neighborhood densities; when the K diffuse neighborhood densities are greater than or equal to the K initial neighborhood densities, the K diffuse center neighborhoods are used as K iteration center neighborhoods, and the K iteration center neighborhoods are continuously iterated according to the preset leap step until a preset number of iterations is met, and the K iteration center neighborhoods obtained in the last iteration are used as the K target center neighborhoods; when the K diffuse neighborhood densities are greater than or equal to the K initial neighborhood densities, the K initial center neighborhoods are used as the K target center neighborhoods.

[0044] In an embodiment of the present application, the neighborhood densities of K diffuse center neighborhoods and K initial center neighborhoods are first calculated respectively. Specifically, the neighborhood density is calculated by counting the number of defect data points contained in the neighborhood and dividing it by the width of the neighborhood range. For example, for the initial center neighborhood [0.4, 0.6], if it contains 5 defect data points and the neighborhood width is 0.2, its neighborhood density is calculated as 5 / 0.2=25. Through this method, the initial neighborhood density of each initial center neighborhood and the diffuse neighborhood density of the diffuse center neighborhood are obtained respectively, that is, K diffuse neighborhood densities and K initial neighborhood densities are obtained.

[0045] Then compare the diffusion neighborhood density of each node with the initial neighborhood density. If the diffusion neighborhood density is greater than or equal to the initial neighborhood density, the diffusion center neighborhood is used as the new iterative center neighborhood and continues to expand. For example, for the first process node, after the diffusion neighborhood is expanded from [0.4, 0.6] to [0.3, 0.7], the number of data points is recalculated and the new density is calculated. If the diffusion neighborhood density is less than the initial neighborhood density, the expansion is stopped and the current initial center neighborhood is used as the first target center neighborhood. Repeat this process for all K process nodes, and determine whether the expansion needs to continue one by one. For nodes that need to continue to expand, use the step-size iterative expansion method to expand the neighborhood range to the left and right sides with a preset leap step size (such as 0.1 or 0.2). For example, for the second process node, after its initial neighborhood is expanded from [0.5, 0.7] to [0.4, 0.8], the number of data points in the expanded range is recalculated and the neighborhood density is updated. This process is iterated for each node until the termination condition is met, including reaching the maximum number of iterations (such as 5 times) or the neighborhood density change is less than the set threshold (such as the change is less than 0.01). After all iterations are completed, the K iterative center neighborhoods obtained from the last expansion are used as the K target center neighborhoods.

[0046] When the diffusion neighborhood density is less than the initial neighborhood density, the expansion is stopped and the current K initial center neighborhoods are directly used as the K target center neighborhoods without further iterative expansion.

[0047] Through the above process, K target center neighborhoods are finally determined.

[0048] Step S400: using the defect position and defect degree as indexes, searching the K grinding defect data sets, and filtering the search results by control points to obtain K feedback control point sets.

[0049] Furthermore, in the method provided in the embodiment of the application, the K polishing defect data sets are retrieved with the defect location and defect degree as indexes, and the control points of the retrieval results are respectively screened to obtain K feedback control point sets, and the method also includes:

[0050] Obtain K grinding defect position sets corresponding to the K grinding defect degree sets; perform control point screening based on the two dimensions of defect degree size and defect position, combining the K grinding defect degree sets and the K grinding defect position sets, to obtain K feedback control point sets.

[0051] In the embodiment of the present application, firstly, the polishing defect data set of each process node is retrieved one by one through two types of index conditions, namely, defect location and defect degree, to extract relevant defect data points. The defect location index indicates the specific location of the defect on the thermal insulation sheet, such as the coordinate information of the edge area or the center area; the defect degree index quantifies the severity of the defect, such as the deviation value of the surface roughness or the depth value of the burr. By matching these two types of indexes, the defect data points that meet the conditions in each node are screened out to form K polishing defect degree sets and corresponding K polishing defect location sets.

[0052] Next, control points are screened from the two dimensions of defect size and defect location, combined with K polishing defect sets and K polishing defect location sets. Specifically, the first polishing defect set and the corresponding first polishing defect location set are extracted from the K polishing defect sets and K polishing defect location sets. M positions are randomly extracted from the first polishing defect location set as the initial control points, and a control loss function is introduced. This function combines the influence of defect size and defect location, and selects the control point with the smallest loss value as the first feedback control point set by randomly selecting the optimized control point combination multiple times. Finally, based on the control loss function and the data set of all nodes, screening is performed to obtain K feedback control point sets.

[0053] Furthermore, the method provided in the application embodiment also includes:

[0054] A first grinding defect degree set and a corresponding first grinding defect position set are extracted from the K grinding defect degree sets and the K grinding defect position sets; M first grinding defect positions are randomly extracted from the first grinding defect position set as M first control points, where M is an integer greater than or equal to 3; a control loss function is constructed from two dimensions of defect degree size and defect position, where the control loss function is:

[0055] ;

[0056] in, To control the output value of the loss function, is the first grinding defect degree corresponding to the jth first grinding defect position in the first grinding defect position set whose distance to the i-th control point is within the preset threshold, is the jth first grinding defect position in the first grinding defect position set whose distance to the i-th control point is within the preset threshold, is the first grinding defect degree of the i-th first control point among the M first control points, is the first grinding defect position of the i-th control point among the M first control points, To balance the weights of defect size and defect location, is the total number of first grinding defect positions in the first grinding defect position set whose distance to the ith control point is within a preset threshold, and M is the total number of first control points; feedback control loss analysis is performed based on the M first control points and the control loss function, and the M first control points corresponding to the minimum value in the output value set of the corresponding control loss function are selected as the first feedback control point set by randomly selecting the M first control points for a preset number of times; based on the control loss function, control points are screened in combination with the K grinding defect degree sets and the K grinding defect position sets to obtain K feedback control point sets.

[0057] In the embodiment of the present application, firstly, the first polishing defect degree set and the first polishing defect position set corresponding to each process node are randomly extracted from K polishing defect degree sets and K polishing defect position sets.

[0058] Then, from the extracted first polishing defect position set, M first polishing defect positions are randomly selected as the initial M first control points, where M is an integer greater than or equal to 3. This step is performed by a random sampling method to ensure that the control points cover multiple key positions and provide diverse initial conditions for subsequent optimization. For example, a node may randomly select two points on the edge and one point in the center as the initial control points.

[0059] Then, from the two dimensions of defect size and defect location, a control loss function is constructed to optimize the control point selection. The control loss function is:

[0060] ;

[0061] in, To control the output value of the loss function, is the first grinding defect degree corresponding to the jth first grinding defect position in the first grinding defect position set whose distance to the i-th control point is within the preset threshold, is the jth first grinding defect position in the first grinding defect position set whose distance to the i-th control point is within the preset threshold, is the first grinding defect degree of the i-th first control point among the M first control points, is the first grinding defect position of the i-th control point among the M first control points, To balance the weights of defect size and defect location, the default value is set. is the total number of first polishing defect positions in the first polishing defect position set whose distance to the i-th control point is within a preset threshold, and M is the total number of first control points.

[0062] Through the above formula, the control loss function comprehensively evaluates the severity and distribution uniformity of the defect points and balances the contribution of each control point.

[0063] In the optimization stage, feedback control loss analysis is performed on the initial M control points based on the control loss function. Through the preset number of random searches, multiple control point combinations are randomly generated, and the control loss function value of each combination is calculated. For example, for each combination, 3 control points are randomly selected, their loss values ​​are calculated, and the results are recorded. Finally, a set of control points with the smallest loss value is selected as the first feedback control point set. This process ensures that the control points can cover the severe areas of defects and be reasonably distributed, thereby improving the accuracy and efficiency of feedback control.

[0064] Finally, the grinding defect degree set and grinding defect location set of all nodes are combined to comprehensively screen each process node and finally obtain a set of K feedback control points.

[0065] Step S500: Based on the K feedback control point sets and the preset polishing index set, the target thermal insulation sheet polished by the target grinder at the K polishing process nodes according to the K polishing process node parameters is subjected to process node polishing quality identification to obtain K process node polishing quality identification results.

[0066] Furthermore, the method provided in the application embodiment also includes:

[0067] The preset polishing index set includes surface roughness, flatness, edge quality and thickness uniformity.

[0068] In the embodiment of the present application, first, the target grinder performs actual grinding operations on the target thermal insulation sheet according to the grinding parameters of each process node (such as abrasive particle size, grinding pressure, grinding speed, etc.). After the grinding is completed, the key detection area on the thermal insulation sheet is focused on based on the feedback control point set. These feedback control points represent locations where defects may exist or need to be focused, such as burr areas on the edge or rough areas in the center, to ensure the accuracy and pertinence of the detection.

[0069] After grinding is completed, quality inspection is performed on each feedback control point one by one based on the preset grinding index set. The preset grinding index set includes surface roughness, flatness, edge quality and thickness uniformity. In surface roughness detection, the actual roughness value (such as Ra value) of each feedback control point is obtained by a roughness measuring instrument; in flatness detection, a three-dimensional profilometer or an optical plane measuring instrument is used to evaluate the flatness deviation of the area around the feedback control point; the edge quality detection uses a high-resolution image recognition system to identify the burr depth or smoothness of the feedback control point; for thickness uniformity, an ultrasonic thickness gauge is used to detect the thickness distribution deviation value of the feedback control point and its surroundings. The above detection methods cover all key indicators of the feedback control point in grinding quality, and generate specific detection data for each feedback control point.

[0070] After quality inspection, K clusters of grinding index eigenvalues ​​of feedback control points are formed, and each cluster contains the inspection results of all feedback control points. These eigenvalues ​​fully reflect the actual quality performance of the feedback control points in terms of surface roughness, flatness, edge quality and thickness uniformity. For example, a control point may have a surface roughness Ra = 0.7µm, flatness deviation = 0.03mm, edge burr depth = 0.02mm, and thickness uniformity deviation = ±0.01mm. Subsequently, these eigenvalue clusters are input into the pre-built quality identifier for comprehensive evaluation. The quality identifier judges the quality status of each feedback control point and identifies its quality by comparing it with the preset grinding index standard value.

[0071] Finally, the recognition results of all feedback control points in each process node are summarized to generate K process node polishing quality recognition results.

[0072] Furthermore, in the method provided in the embodiment of the application, based on the K feedback control point sets and the preset grinding index set, the target thermal insulation sheet that is polished by the target grinder at the K grinding process nodes according to the K grinding process node parameters is subjected to process node grinding quality identification to obtain K process node grinding quality identification results, and further includes:

[0073] Based on the preset polishing index set and the K feedback control point sets, the target thermal insulation sheet polished by the target polishing machine at the K polishing process nodes according to the K polishing process node parameters is subjected to feedback control point quality identification to obtain K feedback control point polishing index characteristic value clusters; the K feedback control point polishing index characteristic value clusters are subjected to quality identification using a pre-built quality identifier to obtain K process node polishing quality identification results.

[0074] In the embodiment of the present application, after the polishing of each process node is completed, the polishing quality of the target thermal insulation sheet is tested by focusing on the key detection area based on the feedback control point set. The preset polishing index set includes four specific numerical standards, namely surface roughness, flatness, edge quality and thickness uniformity.

[0075] During the inspection process, the quality of each control point is evaluated based on the position of the feedback control point. The surface roughness value Ra is measured by a roughness meter, the flatness deviation is evaluated by a three-dimensional profilometer, the edge burr depth is detected by an image recognition system, and the thickness distribution uniformity is detected by an ultrasonic thickness gauge. All the inspection results of each feedback control point form a complete indicator feature set. For example, the surface roughness Ra of a certain feedback control point is 0.7µm, the flatness deviation is 0.03mm, the edge burr depth is 0.02mm, and the thickness deviation is ±0.01mm. Through this process, K feedback control point polishing indicator feature value clusters are generated for all feedback control points.

[0076] These feedback control point indicator feature value clusters are then input into the pre-built quality identifier for comprehensive evaluation. The quality identifier classifies the actual test results of each feedback control point based on the comparison with the preset rules. Specifically, the quality identifier classifies the test results into four levels: "excellent", "good", "medium" or "poor" according to the preset rules. For example, when all indicators of a feedback control point are significantly better than the preset standards, such as Ra of 0.6µm (better than the required 0.8µm) and burr depth of 0.02mm (better than the required 0.03mm), the feedback control point is rated as "excellent"; if some indicators are slightly lower than the standard but still within the allowable range, such as Ra of 0.75µm (slightly close to the required 0.8µm), the point is rated as "good"; if some indicators do not meet the standard but are close to the requirement, such as Ra of 0.85µm (exceeding 0.8µm but closer), the point is rated as "medium"; if multiple indicators are obviously not up to standard, such as Ra of 1.0µm and flatness deviation of 0.06mm (both exceeding the standard), it is rated as "poor".

[0077] Finally, the quality identification results of all feedback control points in each process node are summarized to form the corresponding K process node polishing quality identification results.

[0078] Step S600: Determine whether the polishing quality identification results of the K process nodes meet the quality requirements of the K polishing process nodes. If not, feedback adjustment is performed on the target grinder based on the polishing quality identification results of the K process nodes.

[0079] In the embodiment of the present application, when judging whether the polishing quality identification results of K process nodes meet the quality requirements of the respective K polishing process nodes, the unique quality identification result of each process node is compared with the corresponding preset quality requirements one by one. If the quality identification result of a node is "excellent" or "good", it is determined that the node meets the quality requirements, no adjustment is required, and the subsequent process operations are directly executed; if the identification result is "medium" or "poor", it is determined that the node does not meet the quality requirements, and the feedback adjustment process is started.

[0080] For nodes that do not meet the requirements, first analyze the possible causes of the problem based on the specific quality indicators of the failure (such as high surface roughness or excessive edge burrs) and the corresponding feedback control point area. For example, if the surface roughness does not meet the standard, it may be caused by coarse abrasive particles or excessive grinding speed; if the edge quality is poor, it may be caused by insufficient grinding pressure or inaccurate path control. Combined with these analyses, generate targeted adjustment plans. For example, to improve surface roughness, use finer abrasives or reduce the grinding speed; to improve edge quality, increase the grinding pressure or optimize the grinding accuracy of the edge path.

[0081] The generated adjustment plan is transmitted and implemented in real time through the target grinder's automated control system. After receiving the adjustment plan, the grinder dynamically modifies the parameters of the corresponding process node, such as increasing the grinding pressure, changing the abrasive particle size, or adjusting the grinding speed. These adjustments are directly applied to subsequent grinding operations to ensure that the grinding quality of the problem area is optimized in real time.

[0082] After the feedback adjustment is completed, the corrected process node is re-polished and quality identification is performed again to verify the adjustment effect. If the adjusted identification result still does not reach "excellent" or "good", the parameters are further optimized according to the new identification result, and the feedback adjustment process is repeated until the quality identification result of the node meets the preset requirements. This closed-loop feedback mechanism gradually improves the overall polishing quality of the target thermal insulation sheet through multiple iterations.

[0083] In the embodiments of the present application, in summary, the embodiments of the present application have at least the following technical effects:

[0084] The present application configures the grinding process nodes based on the basic information of the target thermal insulation sheet, obtains K grinding process nodes and K grinding process node parameters, wherein K is an integer greater than or equal to 1; performs grinding defect frequent item mining based on the basic information and the K grinding process nodes, and obtains K grinding defect data sets; traverses the K grinding defect data sets to perform defect degree analysis, and obtains K process node abnormalities; retrieves the K grinding defect data sets with defect location and defect degree as indexes, and performs control point screening on the retrieval results respectively, and obtains K feedback control point sets; based on the K feedback control point sets and the preset grinding index set, performs process node grinding quality identification on the target thermal insulation sheet that is polished by the target grinder at the K grinding process nodes according to the K grinding process node parameters, and obtains K process node grinding quality identification results; determines whether the K process node grinding quality identification results meet the K grinding process node quality requirements, and if not, performs feedback adjustment on the target grinder based on the K process node grinding quality identification results. The present invention solves the technical problems of insufficient grinding accuracy and poor quality consistency in the prior art. By constructing grinding process nodes based on the basic information of the target thermal insulation sheet, combined with frequent item mining of grinding defects, defect degree analysis, control point screening and quality identification, accurate feedback control and dynamic adjustment of the thermal insulation sheet grinding process are achieved, thereby achieving the technical effect of improving the grinding quality of the thermal insulation sheet.

[0085] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. The processes depicted in the accompanying drawings do not necessarily require the specific order and continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0086] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

[0087] This specification and drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. An intelligent control method for a heat insulation sheet grinding machine, characterized in that: The method comprises: performing polishing process node configuration based on basic information of a target thermal insulation sheet, obtaining K polishing process nodes and K polishing process node parameters, wherein K is an integer greater than or equal to 1; Perform grinding defect frequent item mining based on the basic information and the K grinding process nodes to obtain K grinding defect data sets; Traversing the K polishing defect data sets to perform defect degree analysis, and obtaining K process node abnormality degrees, including: Based on a preset grinding index standard value set, traverse the K grinding defect data sets to perform defect degree identification, and obtain K grinding defect degree sets; Calculating the mean of the K grinding defect sets respectively to obtain K grinding defect mean values; Taking the mean values ​​of the K polishing defects as K starting center abnormalities, performing centralized search on the K polishing defect sets to obtain the K process node abnormalities; The K grinding defect data sets are retrieved with the defect location and defect degree as indexes, and the control points of the retrieval results are screened to obtain K feedback control point sets, including: Obtaining K grinding defect position sets corresponding to the K grinding defect degree sets; From the two dimensions of defect size and defect location, the control points are screened by combining the K polishing defect sets and the K polishing defect location sets to obtain K feedback control point sets; Extracting a first grinding defect degree set and a corresponding first grinding defect position set from the K grinding defect degree sets and the K grinding defect position sets; Randomly extracting M first polishing defect positions from the first polishing defect position set as M first control points, where M is an integer greater than or equal to 3; Considering the two dimensions of defect size and defect location, a control loss function is constructed, where the control loss function is: ； in, To control the output value of the loss function, is the first grinding defect degree corresponding to the jth first grinding defect position in the first grinding defect position set whose distance to the i-th control point is within the preset threshold, is the jth first grinding defect position in the first grinding defect position set whose distance to the i-th control point is within the preset threshold, is the first grinding defect degree of the i-th first control point among the M first control points, is the first grinding defect position of the i-th control point among the M first control points, To balance the weights of defect size and defect location, is the total number of first polishing defect positions in the first polishing defect position set whose distance to the i-th control point is within a preset threshold, and M is the total number of first control points; Performing feedback control loss analysis based on the M first control points and the control loss function, by randomly selecting the M first control points for a preset number of times, taking the M first control points corresponding to the minimum value in the output value set of the corresponding control loss function as the first feedback control point set; Based on the control loss function, control point screening is performed in combination with the K grinding defect degree sets and the K grinding defect position sets to obtain K feedback control point sets; Based on the K feedback control point sets and the preset grinding index set, the target thermal insulation sheet that is polished by the target grinder at the K grinding process nodes according to the K grinding process node parameters is subjected to process node grinding quality identification to obtain K process node grinding quality identification results; It is determined whether the polishing quality identification results of the K process nodes meet the quality requirements of the K polishing process nodes. If not, feedback adjustment is performed on the target grinder based on the polishing quality identification results of the K process nodes.

2. The intelligent control method of a heat insulation sheet grinding machine according to claim 1, characterized in that: Taking the K grinding defect means as K starting center abnormalities, performing centralized search on the K grinding defect sets to obtain the K process node abnormalities, including: Obtain an initialized positive direction axis, use the K starting center abnormalities as K axis center points respectively, and construct K defect positive direction axes in combination with the K polishing defect sets, wherein each polishing defect is a real number point on the corresponding defect positive direction axis; According to a preset leap step, construct K initial center neighborhoods of the center points of the K number axes on the K defect degree positive direction number axes; Diffusion of the neighborhood edges of the K initial center neighborhoods in the left and right directions according to the preset leap step length is performed to obtain K diffuse center neighborhoods; Iteratively analyzing the K diffusion center neighborhoods and the K initial center neighborhoods to obtain K target center neighborhoods; The K real number point sets contained in the neighborhoods of the K target centers are traversed and counted, and the K polishing defect degree sets corresponding to the K real number point sets are respectively averaged to obtain the abnormality degrees of the K process nodes.

3. The intelligent control method of a heat insulation sheet grinding machine according to claim 2, characterized in that: Iteratively analyzing the K diffusion center neighborhoods and the K initial center neighborhoods to obtain K target center neighborhoods includes: Calculating the neighborhood densities of the K diffuse center neighborhoods and the K initial center neighborhoods respectively, to obtain K diffuse neighborhood densities and K initial neighborhood densities; When the K diffusion neighborhood densities are greater than or equal to the K initial neighborhood densities, the K diffusion center neighborhoods are used as K iteration center neighborhoods, and the K iteration center neighborhoods are continuously iterated according to the preset leap step until the preset number of iterations is met, and the K iteration center neighborhoods obtained in the last iteration are used as the K target center neighborhoods; When the densities of the K diffuse neighborhoods are less than the densities of the K initial neighborhoods, the K initial center neighborhoods are used as K target center neighborhoods.

4. The intelligent control method of a heat insulation sheet grinding machine according to claim 1, characterized in that: The preset polishing index set includes surface roughness, flatness, edge quality and thickness uniformity.

5. The intelligent control method of a heat insulation sheet grinding machine according to claim 1, characterized in that: Based on the K feedback control point sets and the preset grinding index set, the target thermal insulation sheet that is polished by the target grinder at the K grinding process nodes according to the K grinding process node parameters is subjected to process node grinding quality identification, and the K process node grinding quality identification results are obtained, including: Based on the preset grinding index set and the K feedback control point sets, the target thermal insulation sheet that is grinded by the target grinder at the K grinding process nodes according to the K grinding process node parameters is subjected to feedback control point quality identification to obtain K feedback control point grinding index characteristic value clusters; The pre-built quality identifier is used to perform quality identification on the K feedback control point polishing index characteristic value clusters to obtain K process node polishing quality identification results.

Images