A treatment system for optimizing the wear resistance of hose outer membranes
By performing multi-angle image acquisition and laser scanning on the hose surface, a three-dimensional model is constructed and the spray quality evaluation function is used to optimize the spray parameters, the problem of insufficient wear resistance of the hose surface in the prior art is solved, and more efficient spraying is achieved Quality evaluation and process optimization.
Patent Information
- Application Number
- CN202411134980.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-19
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-08-19
AI Technical Summary
The existing hose wear resistance process optimization method cannot accurately evaluate the spray quality of the wear-resistant materials on the hose surface, resulting in unclear objectives of the spray process optimization, increasing the difficulty of process optimization, and resulting in poor wear resistance on the hose surface.
Provide a processing system for optimizing the wear resistance of the outer membrane of the hose, including a target coating data acquisition module, a contour extraction module, a point cloud fitting module, a spray evaluation coefficient generation module and a spray parameter optimization module, through multiple angles Image acquisition, laser scanning, three-dimensional model construction and spray quality evaluation functions, optimize spray parameters to improve spray quality.
The precision and accuracy of the spray quality evaluation of the wear-resistant material on the hose surface is improved, the spray parameter settings are optimized, and the wear resistance of the outer membrane of the hose is improved.
Smart Images

Figure CN118657957B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of data processing, and in particular to a processing system for optimizing the wear resistance of an outer film of a hose. Background Art
[0002] Improving the wear resistance of the hose surface by spraying is an effective method. The spraying process involves applying a special wear-resistant coating on the hose surface, which can not only protect the hose from damage from the external environment, but also significantly improve its wear resistance.
[0003] At present, when evaluating the spraying quality of wear-resistant coatings on hose surfaces, conventional detection methods have certain limitations. Due to the thin thickness of the hose surface coating, there will be large measurement errors through traditional visual inspection or tool measurement methods. Especially when the hose surface shape is more complex, the measurement error will be more obvious, resulting in insufficient precision and accuracy in coating quality assessment.
[0004] In summary, the existing hose wear-resistant process optimization method cannot accurately evaluate the spraying quality of the wear-resistant material on the hose surface, resulting in unclear goals for spraying process optimization, making it difficult to optimize the spraying process, and causing technical problems such as poor wear resistance on the hose surface. Summary of the invention
[0005] The purpose of this application is to provide a processing system for optimizing the wear resistance of the outer film of the hose, so as to solve the technical problem that the existing hose wear resistance process optimization method cannot accurately evaluate the spraying quality of the wear-resistant material on the hose surface, resulting in unclear goals for spraying process optimization, making the spraying process optimization more difficult, resulting in poor wear resistance on the hose surface.
[0006] In view of the above problems, the present application provides a processing system for optimizing the wear resistance of the outer film of the hose. The system includes: a target coating data acquisition module, which is used to obtain the attribute information and scene information of the hose to be sprayed, and retrieve the target coating data online with the attribute information and scene information as constraints and the expected usage time as a limit, wherein the target coating data includes the target coating type and the target coating thickness; a contour extraction module, which is used to perform multi-angle image acquisition of the hose to be sprayed, perform contour extraction on the multi-angle hose image based on a predetermined contour analysis channel, and output the initial hose contour; a point cloud fitting module, which is used to perform laser scanning on the hose to be sprayed, obtain a hose point cloud data set, and fit the hose point cloud data set according to the initial hose contour to obtain a point cloud fitting result and generate an initial hose three-dimensional model; a hose A three-dimensional model acquisition module is used to set predetermined spraying parameters based on the target coating type and target coating thickness, use a spraying device to spray the hose to be sprayed, and perform multi-angle image acquisition and laser scanning of the sprayed hose to fit the hose three-dimensional model; a spraying evaluation coefficient generation module is used to perform grid traversal comparison on the initial hose three-dimensional model and the hose three-dimensional model in a three-dimensional visualization platform, use a spraying quality evaluation function to evaluate the traversal comparison results, and generate a spraying evaluation coefficient; a spraying parameter optimization module is used to perform simulation optimization of the spraying parameters based on the spraying quality evaluation function if the spraying evaluation coefficient does not meet the expected quality coefficient, and obtain optimized spraying parameters to perform the spraying operation of the hose to be sprayed.
[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:
[0008] The target coating data acquisition module is used to obtain the attribute information and scene information of the hose to be sprayed, and the target coating data is retrieved online with the attribute information and scene information as constraints and the expected use time as a limit, wherein the target coating data includes the target coating type and the target coating thickness; the contour extraction module is used to perform multi-angle image acquisition of the hose to be sprayed, perform contour extraction on the multi-angle hose image based on a predetermined contour analysis channel, and output the initial hose contour; the point cloud fitting module is used to perform laser scanning on the hose to be sprayed, obtain a hose point cloud data set, and fit the hose point cloud data set according to the initial hose contour to obtain a point cloud fitting result, and generate an initial hose three-dimensional model; the hose three-dimensional model A module is provided for setting predetermined spraying parameters based on the target coating type and the target coating thickness, spraying the hose to be sprayed using a spraying device, and performing multi-angle image acquisition and laser scanning of the sprayed hose to obtain a three-dimensional model of the hose by fitting; a spraying evaluation coefficient generation module is provided for performing a grid traversal comparison between the initial hose three-dimensional model and the hose three-dimensional model in a three-dimensional visualization platform, evaluating the traversal comparison result using a spraying quality evaluation function, and generating a spraying evaluation coefficient; a spraying parameter optimization module is provided for performing a simulation optimization of the spraying parameters based on the spraying quality evaluation function if the spraying evaluation coefficient does not meet the expected quality coefficient, and obtaining the optimized spraying parameters to perform the spraying operation of the hose to be sprayed. That is to say, by taking the attribute information and scene information of the hose to be sprayed as constraints, the target coating type and target coating thickness are retrieved online, which can improve the rationality and accuracy of the target coating type and target coating thickness; then, the image and point cloud of the hose to be sprayed are respectively collected by the image sensor and the laser point cloud scanning device, and the initial hose contour is generated based on the image acquisition results, and the hose point cloud data is fitted based on the initial hose contour to construct an initial hose three-dimensional model; based on the target coating type and target coating thickness, the hose to be sprayed is sprayed and a hose three-dimensional model is constructed; then, in the three-dimensional visualization platform, the initial hose three-dimensional model and the hose three-dimensional model are meshed and compared, and the traversal comparison results are evaluated using the spray quality evaluation function to generate a spray evaluation coefficient, which can improve the precision and accuracy of the hose spray quality evaluation; finally, when the spray evaluation coefficient does not meet the expected quality coefficient, a spray evaluation model is constructed based on the spray quality evaluation function, and a simulation optimization of the spray parameters is performed to obtain the optimized spray parameters to execute the spray operation of the hose to be sprayed. The precision and accuracy of the spray quality assessment of wear-resistant materials on the hose surface can be improved, thereby improving the accuracy and rationality of the optimized spray parameter settings, achieving the technical effect of optimizing the hose spraying process and improving the wear resistance of the hose outer film.
[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented according to the contents of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically cited below. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS
[0010] In order to more clearly illustrate the technical solutions in the present application or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.
[0011] Figure 1 A schematic diagram of the structure of a processing system for optimizing the wear resistance of the outer film of a hose in this application;
[0012] Figure 2 A schematic diagram of a process for obtaining target coating data in a processing system for optimizing the wear resistance of a hose outer film in the present application.
[0013] Description of reference numerals:
[0014] Target coating data acquisition module 11, contour extraction module 12, point cloud fitting module 13, hose three-dimensional model acquisition module 14, spraying evaluation coefficient generation module 15, spraying parameter optimization module 16. DETAILED DESCRIPTION
[0015] This application provides a processing system for optimizing the wear resistance of the hose outer film, which solves the existing hose wear resistance process optimization method. Since the spraying quality of the wear-resistant material on the hose surface cannot be accurately evaluated, the target of the spraying process optimization is unclear, which makes the optimization of the spraying process more difficult and causes the technical problem of poor wear resistance on the hose surface. The precision and accuracy of the spraying quality evaluation of the wear-resistant material on the hose surface can be improved, thereby improving the accuracy and rationality of the optimized spraying parameter setting, achieving the technical effect of optimizing the hose spraying process and improving the wear resistance of the hose outer film.
[0016] Below, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited to the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application. It should also be noted that, for the convenience of description, only the parts related to the present application are shown in the accompanying drawings, rather than all of them.
[0017] Please see attached Figure 1 The present application provides a processing system for optimizing the wear resistance of the outer film of a hose, comprising:
[0018] The target coating data acquisition module 11 is used to acquire the attribute information and scene information of the hose to be sprayed, and retrieve the target coating data online with the attribute information and scene information as constraints and the expected use time as a limit, wherein the target coating data includes the target coating type and the target coating thickness;
[0019] Specifically, first, the attribute information and scene information of the hose to be sprayed are obtained, wherein the attribute information includes data such as the material, surface shape, and size of the hose to be sprayed; the scene information refers to the environmental information in the target application scene of the hose to be sprayed, including information such as the environmental medium, environmental temperature, environmental humidity, and friction state. The environmental medium includes acidic medium, corrosive medium, etc. The above information can be set according to actual conditions.
[0020] Big data technology is a series of technologies and methods that enable the value contained in massive data to be mined and displayed, including data planning, data collection, analysis and mining and other technical means. Then, based on big data technology, with the attribute information and scene information as retrieval constraints, information retrieval is performed online to obtain a sample hose coating data set, and the sample hose coating data set is screened according to the expected usage time to obtain a hose coating data set that is greater than the expected usage time; the hose coating data set is further analyzed to determine the target coating type and target coating thickness, wherein the target coating type refers to the wear-resistant coating material adapted to the hose to be sprayed, and the target coating thickness refers to the minimum coating thickness that meets the expected usage time. By using the attribute information and scene information as constraints to obtain the target coating type and target coating thickness based on big data retrieval, the rationality and accuracy of the target coating type and target coating thickness settings can be improved, providing a basis for subsequent spraying operations of the hose to be sprayed.
[0021] A contour extraction module 12 is used to perform multi-angle image acquisition of the hose to be sprayed, perform contour extraction on the multi-angle hose image based on a predetermined contour analysis channel, and output an initial hose contour;
[0022] Specifically, the image sensor is used to collect multi-angle images of the hose to be sprayed, wherein the image sensor is a high-precision image sensor, such as a CCD image sensor, an industrial camera, etc., to obtain a multi-angle hose image. A predetermined contour analysis channel is constructed based on a convolutional neural network, and the contour of the multi-angle hose image is extracted through the predetermined contour analysis channel to obtain multiple hose contour extraction results; then the Bezier curve fitting algorithm is used to fit the contour of the multiple hose contour extraction results to obtain an initial hose contour. By constructing a predetermined contour analysis channel based on a convolutional neural network to perform multi-angle contour extraction, and fitting the multi-angle hose contour using a Bezier curve fitting algorithm to obtain an initial hose contour, the precision of the initial hose contour analysis can be improved, thereby improving the accuracy and efficiency of obtaining the initial hose contour.
[0023] A point cloud fitting module 13 is used to perform laser scanning on the hose to be sprayed, obtain a hose point cloud data set, and fit the hose point cloud data set according to the initial hose contour to obtain a point cloud fitting result, and generate an initial hose three-dimensional model;
[0024] Specifically, the hose to be sprayed is laser scanned by a laser scanning device to obtain a hose point cloud data set; then the hose point cloud data set is fitted based on the initial hose contour to obtain a point cloud fitting result. Then, in a three-dimensional visualization platform, the hose to be sprayed is three-dimensionally modeled based on the point cloud fitting result to generate an initial hose three-dimensional model. By fitting the point cloud data using the initial hose contour, the interference of point cloud noise data can be reduced, and the accuracy of the hose point cloud fitting result can be improved, thereby improving the precision and accuracy of the initial hose three-dimensional model construction.
[0025] A hose three-dimensional model obtaining module 14 is used to set predetermined spraying parameters based on the target coating type and target coating thickness, use a spraying device to spray the hose to be sprayed, and perform multi-angle image acquisition and laser scanning of the sprayed hose to obtain a hose three-dimensional model by fitting;
[0026] Specifically, a target spraying device is determined, and the target spraying device can be set according to the attribute information of the hose to be sprayed, such as a spray gun device. Taking the target coating thickness as the spraying purpose, the adaptive predetermined spraying parameters are set based on the target coating type and the operating parameters of the target spraying device, wherein the predetermined spraying parameters include a predetermined spraying speed, a predetermined spraying pressure, and a predetermined spraying distance. Then, based on the predetermined spraying parameters, the target spraying device is used to spray the hose to be sprayed to obtain a sprayed hose. The sprayed hose is further subjected to multi-angle image acquisition and laser scanning, and three-dimensional modeling is performed based on the point cloud fitting result of the sprayed hose using the same method as that of constructing the initial hose three-dimensional model to obtain a hose three-dimensional model.
[0027] A spraying evaluation coefficient generating module 15 is used to perform mesh traversal comparison on the initial hose 3D model and the hose 3D model in a 3D visualization platform, evaluate the traversal comparison result using a spraying quality evaluation function, and generate a spraying evaluation coefficient;
[0028] Specifically, the three-dimensional visualization platform is a data display platform based on computer technology and visualization technology. It uses 3D graphics and visualization technology to present data to users in a 3D form, so that users can understand and grasp the data more intuitively. In the three-dimensional visualization platform, the initial hose three-dimensional model and the hose three-dimensional model are first meshed, and then the coating deviation in the mesh is traversed and compared based on the meshing result to obtain a mesh coating deviation data set. Then, the mesh coating deviation data set is calculated according to the spraying quality evaluation function to obtain a spraying evaluation coefficient.
[0029] By meshing the initial hose 3D model and the hose 3D model in the 3D visualization platform and performing deviation comparison based on the meshing results, the precision of the coating deviation comparison can be improved. At the same time, by constructing a spraying quality evaluation function to evaluate the comparison results, the coating spraying quality of the hose can be obtained more intuitively, thereby providing support for the subsequent simulation optimization of the spraying parameters.
[0030] The spraying parameter optimization module 16 is used to perform simulation optimization of the spraying parameters based on the spraying quality evaluation function if the spraying evaluation coefficient does not meet the expected quality coefficient, and obtain the optimized spraying parameters to perform the spraying operation of the hose to be sprayed.
[0031] Specifically, an expected quality coefficient is obtained, and the expected quality coefficient can be set by a person skilled in the art according to actual quality requirements, wherein the higher the quality requirements, the larger the expected quality coefficient. Then, the spraying evaluation coefficient is judged based on the expected quality coefficient. If the spraying evaluation coefficient is less than the expected quality coefficient, the spraying parameters are simulated and optimized using an optimization algorithm based on the spraying quality evaluation function to obtain optimized spraying parameters, and finally the spraying operation of the hose to be sprayed is performed according to the optimized spraying parameters.
[0032] Further, as attached Figure 2 As shown, the target coating data acquisition module 11 includes:
[0033] Based on big data, with the attribute information and scene information as constraints, information retrieval is performed online to obtain multiple sample hose coating data, wherein the sample hose coating data includes sample coating type, sample coating thickness and sample scrapping time;
[0034] Select sample hose coating data whose sample scrapping time is greater than the expected service time to construct a qualified hose coating data set;
[0035] Clustering the qualified hose coating data set based on the sample coating type to obtain a plurality of qualified hose coating data sets of a plurality of similar sample coating materials;
[0036] The coating thickness scalar is calculated sequentially based on the plurality of qualified hose coating data sets, the coating resource consumption is evaluated according to the coating thickness scalar and the unit thickness resource, and the coating type and coating thickness scalar with the minimum coating resource consumption are set as the target coating data.
[0037] Specifically, big data technology is a series of technologies and methods that enable the value contained in massive data to be mined and displayed, including data planning, data collection, analysis and mining and other technical means. Based on big data technology, with the attribute information and scene information as constraints, information retrieval is performed online to obtain multiple sample hose coating data, wherein the sample hose coating data includes sample coating type, sample coating thickness and sample scrapping time, and the sample coating type, sample coating thickness and sample scrapping time correspond one to one. Then select the sample hose coating data whose sample scrapping time is greater than the expected use time to construct a qualified hose coating data set, wherein the qualified hose coating data represents the hose coating data that meets the expected wear resistance requirements.
[0038] Then, the qualified hose coating data set is clustered based on the sample coating type, that is, the qualified hose coating data corresponding to the same sample coating type are clustered into one category, and multiple qualified hose coating data sets of the same sample coating material are obtained. Coating thickness scalar calculations are performed based on the multiple qualified hose coating data sets in sequence, wherein the coating thickness scalar refers to the coating thickness mean of the qualified hose coating data set, and multiple coating thickness scalars corresponding to the multiple sample coating types are obtained. The unit thickness resource of each sample coating type is obtained, wherein the unit thickness resource refers to the resource consumption per unit thickness, such as the cost required for the unit thickness coating, wherein different coating types have different corresponding unit costs. Then, the coating resource consumption is calculated based on the coating thickness scalar and the unit thickness resource, and the coating type and coating thickness scalar corresponding to the minimum coating resource consumption are set as the target coating data.
[0039] Further, the contour extraction module 12 includes:
[0040] Constructing a profile analysis sub-channel based on a convolutional neural network, and using a twin network to perform twin replication on the profile analysis sub-channel to generate N profile analysis sub-channels, where N is an integer greater than 5;
[0041] Taking the attribute information as a constraint, retrieve and obtain a sample data set, and divide the sample data set into N equal parts to obtain N sample training sets, wherein the sample data includes a sample hose image and a sample hose contour;
[0042] Using the N sample training sets, iteratively train and cross-validate the N profile analysis sub-channels to obtain N converged profile analysis sub-channels, and build a profile analysis layer;
[0043] The predetermined contour analysis channel is generated based on the contour analysis layer and the contour fitting layer, wherein the contour fitting layer is embedded with a Bezier curve fitting algorithm.
[0044] Specifically, a convolutional neural network is a deep learning model that is usually used for classification and recognition tasks of image data. It extracts features through operations such as convolution and pooling, maps the input data into a high-dimensional feature space, and then classifies or regresses the features through a fully connected layer. A contour analysis subchannel is constructed based on a convolutional neural network. The contour analysis subchannel includes an input layer, multiple hidden layers, and an output layer, wherein the input data of the input layer is a hose image, and the output data of the output layer is a hose contour. The contour analysis subchannel is then twinned using a twin network to generate N contour analysis subchannels, where N is an integer greater than 5, and the specific value of N can be set according to actual needs.
[0045] Taking the attribute information as a constraint, a sample data set is retrieved online, and the sample data set is further divided into N equal parts to obtain N sample training sets, wherein the sample data includes a sample hose image and a sample hose profile. The N profile analysis sub-channels are iteratively trained using the N sample training sets. When the profile analysis sub-channel is in a converged state, the interactive sample training set is used as a sample verification set to cross-validate the N profile analysis sub-channels until N converged profile analysis sub-channels that meet the expected convergence constraints are obtained. The expected convergence constraints are expected output accuracy indicators that can be set according to actual needs. Then, based on the N converged profile analysis sub-channels, the profile analysis layer is built in parallel.
[0046] A contour fitting layer is constructed based on the Bezier curve fitting algorithm, wherein the Bezier curve fitting algorithm is a mathematical curve that describes a smooth curve through control points, which is composed of a starting point, an end point and a number of control points. By adjusting the position and weight of the control points, the Bezier curve is made as close as possible to the given data point, thereby achieving the purpose of optimizing the fitting effect. The contour fitting layer is connected to the output layer of the contour analysis layer, and is used to fit multiple hose contours output by the contour analysis layer into one hose contour. Finally, the predetermined contour analysis channel is constructed based on the contour analysis layer and the contour fitting layer.
[0047] By constructing a predetermined contour analysis channel based on a convolutional neural network to perform multi-angle contour extraction, and using a Bezier curve fitting algorithm to fit the multi-angle hose contour to obtain the initial hose contour, the fineness of the initial hose contour analysis can be improved, thereby improving the accuracy and efficiency of obtaining the initial hose contour.
[0048] Furthermore, the point cloud fitting module 13 includes:
[0049] Based on the initial hose contour, performing random fitting in the hose point cloud data set to obtain a first point cloud fitting result and a first fitting accuracy rate, wherein the first fitting accuracy rate is a ratio of the number of point clouds in the initial hose contour to the number of point clouds in the hose point cloud data set;
[0050] Setting a fitting accuracy threshold based on a laser point cloud device error, and determining whether the first fitting accuracy satisfies the fitting accuracy threshold;
[0051] If not, iterative random fitting is performed in the hose point cloud data set based on the initial hose contour until the fitting accuracy threshold is met or a preset number of iterations is reached, and a point cloud fitting result with the maximum fitting accuracy is output and set as the point cloud fitting result.
[0052] Specifically, in the initial hose contour, random fitting is performed in the hose point cloud dataset, that is, the point cloud data in the hose point cloud dataset is randomly placed in the initial hose contour to obtain a first point cloud fitting result, and a first fitting accuracy is calculated based on the first point cloud fitting result. The first fitting accuracy is the ratio of the number of point clouds in the initial hose contour to the number of point clouds in the hose point cloud dataset, wherein the more the number of point clouds falling into the initial hose contour, the higher the fitting accuracy.
[0053] The acquisition accuracy error of the laser point cloud device is obtained, and a fitting accuracy threshold is set based on the acquisition accuracy error, wherein the acquisition accuracy error and the fitting accuracy threshold are negatively correlated, that is, the smaller the acquisition accuracy error, the larger the fitting accuracy threshold. It is further determined whether the first fitting accuracy is greater than or equal to the fitting accuracy threshold. If the first fitting accuracy is less than the fitting accuracy threshold, iterative random fitting is performed in the hose point cloud data set based on the initial hose contour until a point cloud fitting result greater than or equal to the fitting accuracy threshold is obtained; or if the point cloud fitting result does not meet the fitting accuracy threshold when the number of iterative random fittings reaches a preset number of iterations, the point cloud fitting result with the highest current fitting accuracy is output and set as the point cloud fitting result.
[0054] By fitting the point cloud data based on the initial hose contour, the interference of point cloud noise data can be reduced, and the accuracy of the hose point cloud fitting results can be improved, thereby improving the precision and accuracy of the initial hose three-dimensional model construction.
[0055] Further, the spraying evaluation coefficient generation module 15 includes:
[0056] In the three-dimensional visualization platform, the initial hose three-dimensional model and the hose three-dimensional model are fused, and the deviation area in the fusion result is color-coded to obtain a marked deviation three-dimensional model;
[0057] Dividing the identification deviation three-dimensional model according to a predetermined grid size, and performing deviation traversal comparison based on the grid division result and a preset deviation index to obtain a grid deviation data set;
[0058] The grid deviation data set is evaluated using a spraying quality evaluation function to obtain the spraying evaluation coefficient.
[0059] Specifically, the three-dimensional visualization platform is a data display platform based on computer technology and visualization technology. It uses 3D graphics and visualization technology to present data to users in a 3D form, so that users can understand and master the data more intuitively. In the three-dimensional visualization platform, the initial hose three-dimensional model and the hose three-dimensional model are first fused, and then the deviation area in the fusion result is color-marked, where the deviation area is the coating area, and the marked deviation three-dimensional model is obtained. The marked deviation three-dimensional model is further divided based on a predetermined grid size, and the predetermined grid size can be set based on the deviation comparison accuracy requirement, wherein the higher the deviation comparison accuracy requirement, the smaller the predetermined grid size, for example: the predetermined grid size can be set to a square with a side length of 1 mm; then based on the grid division result, the grid division result is traversed and compared according to the preset deviation index to obtain a grid deviation data set. Finally, the grid deviation data set is sprayed according to the spray quality evaluation function to evaluate the spray quality, and a spray evaluation coefficient is obtained, wherein the higher the spray evaluation coefficient, the better the spray quality of the hose surface.
[0060] Furthermore, the spraying evaluation coefficient generation module 15 also includes:
[0061] Obtaining an expected thickness tolerance threshold, taking an area where the coating thickness is greater than an upper limit of the expected thickness tolerance threshold as a coating overlap area, and taking an area where the coating thickness is less than a lower limit of the expected thickness tolerance threshold as a coating missing area;
[0062] The preset deviation indicators include a coating thickness deviation mean, a coating missing ratio, and a coating overlap ratio, wherein the coating thickness deviation mean, the coating missing ratio, and the coating overlap ratio have a first weight, a second weight, and a third weight identifier, and the second weight is greater than the first weight and greater than the third weight.
[0063] Specifically, first, obtain the expected thickness tolerance threshold, which is the allowable range of the thickness deviation of the coating, which can be set according to the actual situation. The expected thickness tolerance threshold includes the expected thickness lower limit and the expected thickness upper limit, and the area where the coating thickness is greater than the upper limit of the expected thickness tolerance threshold is used as the coating overlap area, and the area where the coating thickness is less than the lower limit of the expected thickness tolerance threshold is used as the coating missing area. The preset deviation index includes the coating thickness deviation mean, the coating missing ratio and the coating overlap ratio. The coating thickness deviation mean is the average value of the thickness deviation of the coating area in the grid, the coating missing ratio refers to the area ratio of the coating missing area in the grid, and the coating overlap ratio refers to the area ratio of the coating overlapping area in the grid. The coating thickness deviation mean, the coating missing ratio and the coating overlap ratio have a first weight, a second weight and a third weight identifier, the sum of the first weight, the second weight and the third weight is 1, and the second weight is greater than the first weight and greater than the third weight, wherein the first weight, the second weight and the third weight can be set according to the degree of influence of the indicator on the overall coating wear resistance, wherein the greater the degree of influence, the greater the corresponding weight. By setting multiple deviation indicators and setting adaptation weights for the deviation indicators based on the degree of influence, the accuracy of hose surface spray quality evaluation can be improved.
[0064] Furthermore, the spraying evaluation coefficient generation module 15 also includes:
[0065] The expression of the spraying quality evaluation function is:
[0066] ;
[0067] in, Characterizes the overall spraying evaluation coefficient, N is the number of comparison grids, is the importance of the i-th grid, is the first weight, is the second weight, is the third weight, is the mean coating thickness deviation of the i-th grid, is the coating missing ratio of the i-th grid, is the coating overlap ratio of the i-th grid.
[0068] Specifically, the expression of the spraying quality evaluation function is:
[0069] ; In the spraying quality evaluation function, the overall spraying evaluation coefficient is characterized, wherein the larger the spraying evaluation coefficient is, the better the overall spraying quality of the hose surface is characterized; N is the number of comparison grids; is the importance of the i-th grid, where the i-th grid is any one of the N grids, and the grid importance is set based on the position of the grid in the hose, where the more critical the position, the greater the importance; is the first weight, is the second weight, is the third weight; is the mean coating thickness deviation of the i-th grid; is the coating missing ratio of the i-th grid; is the coating overlap ratio of the i-th grid.
[0070] Further, the spraying parameter optimization module 16 includes:
[0071] Taking the attribute information and scene information as constraints, a model training data set is retrieved online, wherein the model training data includes sample spraying parameters and sample spraying evaluation coefficients, and the sample spraying evaluation coefficients can be obtained by evaluating historical sample data through the spraying quality evaluation function;
[0072] The spraying evaluation model built based on the feedforward neural network is supervised and trained by using the model training data set to obtain a spraying evaluation model that is in convergence;
[0073] Based on the spraying evaluation model, the spraying parameters are simulated and optimized with the adjustment range of the spraying parameters as the optimization space, and the optimized spraying parameters are output.
[0074] Specifically, firstly, with the attribute information and scene information as constraints, the model training data set is retrieved online, wherein the model training data includes sample spraying parameters and sample spraying evaluation coefficients. The sample spraying evaluation coefficients can be obtained by evaluating historical sample data through the spraying quality evaluation function, that is, the sample hose spraying parameters, sample hose images and sample hose point cloud data are retrieved, and then the same method as above is used to perform grid deviation comparison, and the comparison results are evaluated using the spraying quality evaluation function to obtain the sample spraying evaluation coefficient.
[0075] Then, a spraying evaluation model is constructed based on a feedforward neural network. The spraying evaluation model is a BP neural network model that can be iteratively optimized in machine learning. It is obtained by supervised training through a training data set. The input data of the spraying evaluation model is the spraying parameters, and the output data is the spraying evaluation coefficient. Then, the spraying evaluation model is supervised and trained using the model training data set as sample training data to obtain a spraying evaluation model that is in convergence. By constructing a spraying evaluation model based on a feedforward neural network, spraying quality evaluation can be performed based on spraying parameters, while improving the efficiency and accuracy of spraying quality evaluation, providing support for subsequent simulation optimization of spraying parameters. Finally, based on the spraying evaluation model, the spraying parameters are simulated and optimized using an optimization algorithm with the adjustment range of the spraying parameters as the optimization space to obtain the optimized spraying parameters.
[0076] Furthermore, the spraying parameter optimization module 16 also includes:
[0077] The adjustment range of the spraying parameters includes a spraying speed threshold, a spraying pressure threshold and a spraying distance threshold;
[0078] Based on the spraying speed threshold, the spraying pressure threshold, and the spraying distance threshold, a plurality of spraying parameter combinations are randomly generated and set as a plurality of preliminary solutions;
[0079] The plurality of preliminary solutions are evaluated by the spraying evaluation model to obtain a plurality of spraying evaluation coefficients, which are set as a plurality of fitnesses;
[0080] Arrange the multiple preliminary solutions in descending order of fitness to generate a preliminary solution sequence, and select the first K preliminary solutions of the preliminary solution sequence as the head solution and the last J preliminary solutions as the tail solution;
[0081] Based on the K head solutions, the J tail solutions are randomly clustered to generate K domains, and within the K domains, the head solution is used as the optimization direction, and the tail solutions within the domain are updated once according to the preset three-dimensional step size to obtain K updated domains, wherein within the K updated domains, if the updated tail solution does not satisfy the adjustment range of the spraying parameters, the tail solution is not updated, and if there is a situation where the fitness of the tail solution is greater than the fitness of the head solution, the tail solution is used to replace the head solution within the domain;
[0082] Continue to iterate and optimize until the preset number of iterations is met, output the current K fields, and select the head solution in the optimal field to set as the optimized spraying parameter. The optimal field is the field with the largest fitness mean among the current K fields.
[0083] Specifically, the adjustment range of the spraying parameters includes a spraying speed threshold, a spraying pressure threshold, and a spraying distance threshold. Then, any parameter is randomly extracted within the spraying speed threshold, the spraying pressure threshold, and the spraying distance threshold to form a spraying parameter combination, wherein the spraying parameter combination includes any spraying speed, any spraying pressure, and any spraying distance, and a plurality of spraying parameter combinations are obtained, and the spraying parameter combinations are set as preliminary solutions to obtain a plurality of preliminary solutions. The plurality of preliminary solutions are further evaluated by the spraying evaluation model to obtain a plurality of spraying evaluation coefficients, and the plurality of spraying evaluation coefficients are set as a plurality of fitness levels.
[0084] Then, the multiple preliminary solutions are arranged from large to small according to the fitness, and a preliminary solution sequence is generated. The first K preliminary solutions of the preliminary solution sequence are selected as the head solutions, and the last J preliminary solutions are selected as the tail solutions, where the sum of K and J is the number of preliminary solutions in the preliminary solution sequence, and K is much smaller than J. The specific values of K and J can be set according to the number of preliminary solutions. Then, the J tail solutions are randomly clustered based on the K head solutions to generate K fields, where the number of tail solutions in each field is the same. Then, within the K fields, the head solution is used as the optimization direction, and the tail solutions within the field are adjusted once according to the preset three-dimensional step size, where the preset three-dimensional step size includes the spraying speed adjustment step size, the spraying pressure adjustment step size, and the spraying distance adjustment step size, which can be set according to the actual situation to obtain K updated fields, where in the tail solution adjustment process, if the updated tail solution does not meet the adjustment range of the spraying parameters, the tail solution will not be updated. Then, within the K update domains, if the fitness of the tail solution is greater than the fitness of the head solution, the head solution within the domain is replaced with the tail solution to obtain K update domains after replacement.
[0085] Continue to iterate and optimize using the above method until the preset number of iterations is met. The preset number of iterations can be set according to the optimization requirements, and the current K fields are output; then the spray quality evaluation function is used to calculate the fitness of the head solution and the tail solution in the current K fields, and the field with the largest fitness mean in the current K fields is set as the optimal field, and finally the head solution in the optimal field is selected and set as the optimized spray parameter. By using the optimization algorithm to simulate the optimization of the spray parameters, the optimization efficiency can be improved, while saving production resources. In addition, since the algorithm has a strong global search capability, the accuracy and comprehensiveness of the optimized spray parameters can be improved.
[0086] Furthermore, the spraying parameter optimization module 16 also includes:
[0087] Expanding the attribute information and the scene information based on a predetermined tolerance strategy to generate an attribute expansion threshold and a scene expansion threshold;
[0088] Based on the classification decision tree, the attribute expansion threshold is used as the first child node, the scenario expansion threshold is used as the second child node, and the target coating type, target coating thickness and optimized spraying parameters are used as parallel leaf nodes of the first child node and the second child node to construct a first decision tree branch and add it to the hose spraying resource library;
[0089] The subsequent hose spraying processing is performed through the hose spraying resource library, and the real-time update of the hose spraying resource library is completed at the same time.
[0090] Specifically, the attribute information and scene information are expanded based on a predetermined tolerance strategy, and the predetermined tolerance strategy is set based on the specific indicator type of the attribute information and scene information. For example, when the ambient temperature in the scene information is 30 degrees Celsius, but the error temperature is 1 degree Celsius, the deviation of the wear resistance is small, and the tolerance threshold of the ambient temperature can be set to 29 to 31 degrees Celsius to obtain the attribute expansion threshold and the scene expansion threshold.
[0091] Classification decision tree refers to dividing data into different categories by segmenting the data set. This segmentation process is based on a series of features and thresholds, and finally forms a tree structure. Each node represents a feature and its threshold, and each leaf node represents a classification result. Based on the construction principle of the classification decision tree, the attribute extension threshold is the first child node, the scene extension threshold is the second child node, and the target coating type, target coating thickness and optimized spraying parameters are the parallel leaf nodes of the first child node and the second child node. That is, the leaf node can only be obtained when the first child node and the second child node are satisfied at the same time; the first decision tree branch is constructed, and then the first decision tree branch is added to the hose spray resource library to complete the update of the hose spray resource library. Finally, the subsequent hose spraying processing is performed through the hose spray resource library, and the real-time update of the hose spray resource library is completed at the same time.
[0092] By constructing a hose spraying resource library based on the principle of classification decision tree, the analysis time of spraying data can be reduced and the efficiency and quality of subsequent hose spraying processing can be improved.
[0093] In summary, the processing system for optimizing the wear resistance of the hose outer film provided by the present application has the following technical effects:
[0094] 1. By taking the attribute information and scene information of the hose to be sprayed as constraints, the target coating type and target coating thickness are retrieved online to improve the rationality and accuracy of the target coating type and target coating thickness; then, the image and point cloud of the hose to be sprayed are collected respectively by the image sensor and the laser point cloud scanning device, and the initial hose contour is generated based on the image acquisition results, and the hose point cloud data is fitted based on the initial hose contour to construct an initial hose three-dimensional model; based on the target coating type and target coating thickness, the hose to be sprayed is sprayed to construct a hose three-dimensional model; then, in the three-dimensional visualization platform, the initial hose three-dimensional model and the hose three-dimensional model are meshed and compared, and the traversal comparison results are evaluated using the spray quality evaluation function to generate a spray evaluation coefficient, which can improve the precision and accuracy of the hose spray quality evaluation; finally, when the spray evaluation coefficient does not meet the expected quality coefficient, a spray evaluation model is constructed based on the spray quality evaluation function, and the simulation optimization of the spray parameters is performed to obtain the optimized spray parameters to execute the spray operation of the hose to be sprayed. The precision and accuracy of the spray quality assessment of wear-resistant materials on the hose surface can be improved, thereby improving the accuracy and rationality of the optimized spray parameter settings, achieving the technical effect of optimizing the hose spraying process and improving the wear resistance of the hose outer film.
[0095] 2. By constructing a predetermined contour analysis channel based on a convolutional neural network to perform multi-angle contour extraction, and using the Bezier curve fitting algorithm to fit the multi-angle hose contour to obtain the initial hose contour, the fineness of the initial hose contour analysis can be improved, thereby improving the accuracy and efficiency of obtaining the initial hose contour.
[0096] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.
[0097] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and its equivalent technology, the present application is also intended to include these modifications and variations.
Claims
1. A treatment system for optimizing the wear resistance of the outer membrane of a hose, characterized in that: The system comprises: A target coating data acquisition module is used to acquire the attribute information and scene information of the hose to be sprayed, and retrieve the target coating data online with the attribute information and scene information as constraints and the expected use time as a limit, wherein the target coating data includes the target coating type and the target coating thickness; A contour extraction module, used to perform multi-angle image acquisition of the hose to be sprayed, perform contour extraction on the multi-angle hose image based on a predetermined contour analysis channel, and output an initial hose contour; A point cloud fitting module is used to perform laser scanning on the hose to be sprayed, obtain a hose point cloud data set, and fit the hose point cloud data set according to the initial hose contour to obtain a point cloud fitting result, and generate an initial hose three-dimensional model; A hose three-dimensional model obtaining module is used to set predetermined spraying parameters based on the target coating type and target coating thickness, use a spraying device to spray the hose to be sprayed, and perform multi-angle image acquisition and laser scanning of the sprayed hose to obtain a hose three-dimensional model by fitting; A spraying evaluation coefficient generation module is used to perform mesh traversal comparison on the initial hose 3D model and the hose 3D model in a 3D visualization platform, evaluate the traversal comparison result using a spraying quality evaluation function, and generate a spraying evaluation coefficient; A spraying parameter optimization module, for performing simulation optimization of spraying parameters based on the spraying quality evaluation function if the spraying evaluation coefficient does not meet the expected quality coefficient, and obtaining optimized spraying parameters to perform the spraying operation of the hose to be sprayed; Wherein, the spraying evaluation coefficient generation module is used for: In the three-dimensional visualization platform, the initial hose three-dimensional model and the hose three-dimensional model are fused, and the deviation area in the fusion result is color-coded to obtain a marked deviation three-dimensional model; Dividing the identification deviation three-dimensional model according to a predetermined grid size, and performing deviation traversal comparison based on the grid division result and a preset deviation index to obtain a grid deviation data set; Using a spraying quality evaluation function to evaluate the grid deviation data set, to obtain the spraying evaluation coefficient; Wherein, the spraying evaluation coefficient generation module is also used for: Obtaining an expected thickness tolerance threshold, taking an area where the coating thickness is greater than an upper limit of the expected thickness tolerance threshold as a coating overlap area, and taking an area where the coating thickness is less than a lower limit of the expected thickness tolerance threshold as a coating missing area; The preset deviation index includes a coating thickness deviation mean, a coating missing ratio, and a coating overlap ratio, wherein the coating thickness deviation mean, the coating missing ratio, and the coating overlap ratio have a first weight, a second weight, and a third weight identifier, and the second weight is greater than the first weight and greater than the third weight; Wherein, the expression of the spraying quality evaluation function is: ; in, Characterizes the overall spraying evaluation coefficient, N is the number of comparison grids, is the importance of the i-th grid, is the first weight, is the second weight, is the third weight, is the mean coating thickness deviation of the i-th grid, is the coating missing ratio of the i-th grid, is the coating overlap ratio of the i-th grid, wherein the expected thickness tolerance threshold includes an expected thickness lower limit and an expected thickness upper limit, and the area where the coating thickness is greater than the upper limit of the expected thickness tolerance threshold is taken as the coating overlap area, and the coating missing ratio refers to the area ratio of the coating missing area in the grid; Wherein, the spraying parameter optimization module is used for: Taking the attribute information and scene information as constraints, a model training data set is retrieved online, wherein the model training data includes sample spraying parameters and sample spraying evaluation coefficients, and the sample spraying evaluation coefficients are obtained by evaluating historical sample data through the spraying quality evaluation function; The spray evaluation model built based on the feedforward neural network is supervised and trained by the model training data set to obtain a spray evaluation model in convergence, wherein the spray evaluation model is a BP neural network model that is iteratively optimized in machine learning, which is obtained by supervised training with the training data set, and the input data of the spray evaluation model is the spray parameter, and the output data is the spray evaluation coefficient; Based on the converged spray evaluation model, taking the adjustment range of the spray parameters as the optimization space, the spray parameters are simulated and optimized, and the optimized spray parameters are output; Wherein, the spraying parameter optimization module is also used for: The adjustment range of the spraying parameters includes a spraying speed threshold, a spraying pressure threshold and a spraying distance threshold; Based on the spraying speed threshold, the spraying pressure threshold, and the spraying distance threshold, a plurality of spraying parameter combinations are randomly generated and set as a plurality of preliminary solutions; The plurality of preliminary solutions are evaluated by the spraying evaluation model to obtain a plurality of spraying evaluation coefficients, which are set as a plurality of fitnesses; Arrange the multiple preliminary solutions in descending order of fitness to generate a preliminary solution sequence, and select the first K preliminary solutions of the preliminary solution sequence as the head solution and the last J preliminary solutions as the tail solution; Based on the K head solutions, the J tail solutions are randomly clustered to generate K domains, and within the K domains, the head solution is used as the optimization direction, and the tail solutions within the domain are updated once according to the preset three-dimensional step size to obtain K updated domains, wherein within the K updated domains, if the updated tail solution does not satisfy the adjustment range of the spraying parameters, the tail solution is not updated, and if there is a situation where the fitness of the tail solution is greater than the fitness of the head solution, the tail solution is used to replace the head solution within the domain; Continue to iterate and optimize until the preset number of iterations is met, output the current K fields, and select the head solution in the optimal field to set as the optimized spraying parameter. The optimal field is the field with the largest fitness mean among the current K fields.
2. The system according to claim 1, characterized in that The target coating data acquisition module is used for: Based on big data, with the attribute information and scene information as constraints, information retrieval is performed online to obtain multiple sample hose coating data, wherein the sample hose coating data includes sample coating type, sample coating thickness and sample scrapping time; Select sample hose coating data whose sample scrapping time is greater than the expected service time to construct a qualified hose coating data set; Clustering the qualified hose coating data set based on the sample coating type to obtain a plurality of qualified hose coating data sets of a plurality of similar sample coating materials; The coating thickness scalar is calculated sequentially based on the plurality of qualified hose coating data sets, the coating resource consumption is evaluated according to the coating thickness scalar and the unit thickness resource, and the coating type and coating thickness scalar with the minimum coating resource consumption are set as the target coating data.
3. The system according to claim 1, characterized in that The contour extraction module is used for: Constructing a profile analysis sub-channel based on a convolutional neural network, and using a twin network to perform twin replication on the profile analysis sub-channel to generate N profile analysis sub-channels, where N is an integer greater than 5; Taking the attribute information as a constraint, retrieve and obtain a sample data set, and divide the sample data set into N equal parts to obtain N sample training sets, wherein the sample data includes a sample hose image and a sample hose contour; Using the N sample training sets, iteratively train and cross-validate the N profile analysis sub-channels to obtain N converged profile analysis sub-channels, and build a profile analysis layer; The predetermined contour analysis channel is generated based on the contour analysis layer and the contour fitting layer, wherein the contour fitting layer is embedded with a Bezier curve fitting algorithm.
4. The system according to claim 1, characterized in that The point cloud fitting module is used for: Based on the initial hose contour, performing random fitting in the hose point cloud data set to obtain a first point cloud fitting result and a first fitting accuracy rate, wherein the first fitting accuracy rate is a ratio of the number of point clouds in the initial hose contour to the number of point clouds in the hose point cloud data set; Setting a fitting accuracy threshold based on a laser point cloud device error, and determining whether the first fitting accuracy satisfies the fitting accuracy threshold; If not, iterative random fitting is performed in the hose point cloud data set based on the initial hose contour until the fitting accuracy threshold is met or a preset number of iterations is reached, and a point cloud fitting result with the maximum fitting accuracy is output and set as the point cloud fitting result.
5. The system according to claim 1, characterized in that The spraying parameter optimization module is also used for: Expanding the attribute information and the scene information based on a predetermined tolerance strategy to generate an attribute expansion threshold and a scene expansion threshold; Based on the classification decision tree, the attribute expansion threshold is used as the first child node, the scenario expansion threshold is used as the second child node, and the target coating type, target coating thickness and optimized spraying parameters are used as parallel leaf nodes of the first child node and the second child node to construct a first decision tree branch and add it to the hose spraying resource library; The subsequent hose spraying processing is performed through the hose spraying resource library, and the real-time update of the hose spraying resource library is completed at the same time.
