Numerical control machining method for polyvinylidene chloride soft sealing valve

Abstract

The invention belongs to the technical field of numerical control valves, and discloses a numerical control machining method for a polyvinylidene chloride soft sealing valve. The numerical control machining method comprises the steps that (1) three-dimensional scanning data of the surface of a rubber ring is obtained, a high-resolution point cloud model is generated, geometric parameters of local folds and a high-curvature area are extracted according to non-uniform curvature features, and initial surface feature distribution is obtained; (2) according to the initial surface feature distribution, generating a multi-scale grid structure by adopting a self-adaptive grid division algorithm, increasing the grid density in a complex area to cover wrinkles and high-curvature parts, reducing the density in a flat area to reduce the calculated amount, and determining an initial point location layout; and (3) spraying polyvinylidene chloride on the rubber ring according to the point position layout obtained in the step (2). The soft sealing valve has the advantage that the performance bottleneck problem of a traditional soft sealing valve in a high-temperature environment is solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of numerically controlled valves, and particularly relates to a method for numerically controlling the machining of a polyvinylidene chloride soft-sealed valve. Background Art

[0002] A soft-seal valve is a valve that uses an elastic or flexible material as a sealing element. Its core function is to achieve zero or extremely low leakage control of fluids through the deformation of soft materials. Most existing soft-seal valves use rubber gaskets for sealing, which have the following defects in application:

[0003] I. Insufficient thermal stability of materials:

[0004] The molecular structure of rubber is prone to degradation at high temperatures. Taking common fluororubber (VITON) as an example, although its short-term heat resistance can reach 250°C, when exposed to an environment above 230°C for a long time, the molecular chain will undergo cleavage and cross-linking reactions, resulting in a decrease in elastic modulus and an increase in the permanent deformation rate. Although silicone rubber (SIL) can withstand a wide temperature range of 55 - 250°C, its tensile strength is only 7 - 10 MPa, far lower than the mechanical properties of metal materials.

[0005] II. Deterioration of chemical compatibility:

[0006] High temperature exacerbates the chemical reaction between rubber and the medium. For example, when nitrile rubber (NBR) contacts aromatic media, the swelling rate reaches 20% at 120°C, and after exceeding 200°C, the swelling rate doubles, resulting in seal failure. Although hydrogenated nitrile rubber (HNBR) has excellent corrosion resistance, its tolerance to ester solvents at high temperatures is still limited, and it is prone to chemical decomposition especially in an environment above 230°C. This chemical erosion not only damages the seal structure but may also cause medium leakage and equipment corrosion.

[0007] III. Deterioration of mechanical properties:

[0008] The viscoelastic properties of rubber change significantly in a high-temperature environment. The compression set of rubber is about 20% at 150°C, and can reach more than 45% after rising to 230°C. This deformation causes a decrease in the seal contact stress and cannot maintain an effective seal. In addition, high temperature accelerates the oxidative aging process of rubber. Its annual attenuation rate of tensile strength is 15% at 200°C and can reach 30% at 250°C. This degradation of mechanical properties directly shortens the maintenance cycle of the valve.

[0009] IV. Limitations in process adaptability:

[0010] The manufacturing process of rubber seals is sensitive to high-temperature environments. During vulcanization, the low thermal conductivity of rubber (0.15 - 0.25 W / (m·K)) results in a temperature difference of more than 50 °C between the inside and outside of thick-walled products, easily causing the phenomenon of "outer cooked and inner raw", which affects the vulcanization uniformity. In addition, the bonding strength between rubber and metal-based materials decreases by 30% - 50% in environments above 200 °C, leading to interface peeling failure. This process defect further limits the reliability of rubber seals under high-temperature and high-pressure conditions.

[0011] Currently, to address the above defects, those skilled in the art have developed perfluoroether rubber (FFPM) as a material for soft-sealed valves. For example, a high-strength and high-modulus perfluoroether rubber material and its preparation method and application disclosed in Patent No. CN111925615B mainly use a material obtained by combining perfluoroether rubber with cross-linking agents, accelerators, etc. for valve applications. As is well known to those skilled in the art, although this material can withstand temperatures above 300 °C, its cost is 5 - 10 times higher than that of ordinary rubber, and the processing difficulty is large. Some also use metal-rubber composite materials as soft-sealed valves. For example, a rubber-sealed valve and its production process disclosed in Patent No. CN11315045613B disclose that by adhering silicone rubber to a metal valve, the high-temperature performance is improved. Although this soft-sealed valve can adapt to extreme environments of 800 °C, because it is only a simple composite of adhering silicone rubber to metal, the obtained valve has a high elastic modulus and poor sealing compensation ability, and is only suitable for static sealing occasions.

[0012] In summary, the soft-sealed valves of the prior art are difficult to work in environments above 230 °C, which has become a bottleneck in the technological development of this field. Therefore, breaking through the bottleneck of high-temperature application of soft-sealed valves has become a technical problem that those skilled in the art urgently need to solve. Summary of the Invention

[0013] (1) Technical Problems to be Solved

[0014] Aiming at the deficiencies of the prior art, the present invention provides a numerical control processing method for polyvinylidene chloride soft-sealed valves to solve the above technical problems.

[0015] (2) Technical Solutions

[0016] To achieve the above object, the present invention provides the following technical solutions:

[0017] A numerical control processing method for polyvinylidene chloride soft-sealed valves includes the following steps:

[0018] (1) Obtain three-dimensional scan data of the rubber seal surface, generate a high-resolution point cloud model, extract geometric parameters of local folds and high-curvature regions for non-uniform curvature features, and obtain the initial surface feature distribution;

[0019] (2) According to the initial surface feature distribution, an adaptive mesh generation algorithm is used to generate a multi-scale mesh structure. The mesh density is increased in complex regions to cover folds and high-curvature parts, and decreased in flat regions to reduce the computational amount, thereby determining a preliminary point layout.

[0020] (3) According to the point layout obtained in step (2), polyvinylidene chloride is sprayed on the rubber ring.

[0021] Polyvinylidene chloride is mostly used in the packaging of the food and pharmaceutical fields, such as food preservative films, etc., and also has industrial applications, such as coating aluminum molds used on chips for moisture protection. The present invention unexpectedly discovers that spraying it on the surfaces of existing commonly used soft-sealing valve materials such as NBR nitrile rubber, HNBR hydrogenated nitrile rubber, SIL silicone rubber, and VITON fluororubber can significantly improve the high-temperature resistance, which can be increased from the original 90 - 200 °C to 300 - 500 °C.

[0022] Preferably, in step (3), polyvinylidene chloride is dissolved in tetrachloroethane to obtain a spraying liquid, and then the spraying liquid is sprayed onto the rubber ring by an electrostatic spraying method. Polyvinylidene chloride is in a powder state under normal conditions. Tetrachloroethane is preferably selected to enable better dissolution. The electrostatic spraying method is preferably used for spraying. The electrostatic spraying technology realizes the directional adsorption of coating particles through a high-voltage electric field, which can significantly improve the utilization rate of the spraying liquid. For traditional spraying, the paint utilization rate is only 30% - 40% due to the scattering of paint mist, while for electrostatic spraying, through the charge adsorption principle, the coating particles are precisely attached to the surface of the workpiece along the electric power line, and the liquid paint utilization rate can reach 85% - 90%.

[0023] Preferably, 5 - 10 g of polyvinylidene chloride is dissolved in every 100 mL of tetrachloroethane, and the dissolution is carried out by a gradient temperature increase method. First, it is pre-dissolved at 40 °C for 0.3 - 0.5 h, the temperature is raised to 50 °C and stirred for dissolution for 0.3 - 0.5 h, and finally the temperature is raised to 60 °C and stirred for dissolution for 0.5 - 1 h. This mass ratio is preferably used, and combined with programmed temperature increase, the dissolution rate can be significantly improved, reaching a dissolution rate of more than 95%.

[0024] Preferably, the voltage of the electrostatic spraying is 50 - 60 kv, the atomizing air pressure is 0.2 - 0.6 Mpa, the distance from the spray gun to the rubber ring is 15 - 25 cm, and the liquid flow rate of the spraying liquid is 0.8 - 1.5 m / s. The preferably selected electrostatic spraying working condition parameters can cooperate well with the spraying liquid to evenly spray the spraying liquid on the surface of the rubber ring.

[0025] Preferably, the thickness of the polyvinylidene chloride sprayed on the rubber ring is 20-50 μm. If the selected thickness is too thin, it cannot achieve the function of improving the high temperature resistance of the rubber ring well. If the thickness is too thick, the cost will increase. Considering the cost, large-scale production, and the ability to achieve the high temperature resistance function well, the thickness of 20-50 μm is the best choice.

[0026] Further, the step (1) includes the following steps:

[0027] a. Collect three-dimensional scan data of the surface of the rubber ring through a laser scanner to generate a first point cloud model, and perform downsampling processing on the first point cloud model by using a voxel filtering method to obtain a second point cloud model;

[0028] b. Extract the non-uniformity features from the second point cloud model, calculate the normal vector and curvature value of each point for the curvature feature, and determine the first curvature distribution;

[0029] c. For the area where the curvature value in the first curvature distribution exceeds the preset threshold, extract the coordinate positions of the local folds and high-curvature areas to obtain a first area set;

[0030] d. Classify the points in the first area set through a random forest algorithm, judge the boundaries of the local folds and high-curvature areas, and generate a second area set;

[0031] e. Calculate the geometric parameters of each area from the second area set, including area, perimeter, and curvature mean, to obtain a first parameter set;

[0032] f. Perform dimensionality reduction processing on the first parameter set by using the principal component analysis method, extract the dominant components of the surface features, and generate a second parameter set;

[0033] g. Group the data points in the second parameter set through a clustering algorithm, determine the distribution pattern of the initial surface features, and obtain the final distribution result.

[0034] Further, the step (2) includes the following steps:

[0035] a. Generate a multi-scale grid structure through an adaptive grid division algorithm, adjust the grid density for the folded parts and curvature parts in the complex areas of the surface features, cover the coordinate positions of the local feature points, reduce the number of points in the flat area, and determine the preliminary grid point distribution;

[0036] b. Perform downsampling processing on the preliminary grid point distribution by using a voxel filtering method, retain the local feature details for the complex areas with a higher grid density, reduce the number of redundant points in the flat areas with a lower grid density, and obtain an optimized grid point set;

[0037] c. By calculating and optimizing the normal vectors and curvature values of each point in the set of grid points, local feature parameters are extracted for the wrinkled parts and the curvature parts, a smooth curvature distribution is obtained from the flat areas, and the curvature feature distribution pattern is determined;

[0038] d. The principal component analysis method is used to perform dimensionality reduction on the curvature feature distribution pattern, the dominant geometric components are extracted for local features in complex areas, and the secondary curvature variables are removed from the flat areas to obtain a dimensionality-reduced feature set;

[0039] e. The K-means clustering algorithm is used to group the data points in the dimensionality-reduced feature set, high-curvature clusters are divided for the wrinkled parts and the curvature parts, low-curvature clusters are separated from the flat areas, and the feature cluster distribution result is determined.

[0040] Furthermore, in step (3) during spraying, according to the preliminary point layout, the dynamic changes during the spraying process, and the spraying effect, dynamic deformation prediction factors are extracted. The stress and deformation trend of the rubber ring during dynamic spraying is simulated through finite element analysis. Combining with the transient spraying pressure data collected in real time, the influence of deformation on point offset is judged, and the spraying angle and spraying amount are dynamically adjusted.

[0041] Furthermore, step (3) also includes: when dynamically adjusting the spraying angle and spraying amount, according to the judged influence of deformation on point offset, the spraying points are optimized. The real-time calculation module is used to perform block processing on the area with a sharp increase in the calculation amount, obtain the spraying priority sequence for each block area, and perform spraying according to this priority sequence.

[0042] Furthermore, in step (3), the high-risk areas of stress concentration are extracted from the spraying priority sequence, and the positions of the points with relatively low adhesion uniformity are predicted through a support vector machine classifier to judge which points need to adjust the spraying parameters additionally to improve the consistency of the coating thickness.

[0043] Above, the present invention adopts this numerical control method for electrostatic spraying and has the following beneficial technical effects:

[0044] First, through the three-dimensional scanning and modeling technology, a high-resolution point cloud model of the rubber ring surface is obtained, and the geometric parameters of local wrinkles and high-curvature areas can be accurately extracted, providing a reliable basis for subsequent processing;

[0045] Second, the multi-scale grid structure generated by the adaptive grid division algorithm can increase the grid density in complex areas to cover local features, and at the same time reduce the density in flat areas to reduce the calculation amount, thereby optimizing the point layout;

[0046] Third, the electrostatic spraying method combined with the dynamic deformation prediction factor can adjust the spraying angle and spraying amount in real time to ensure that the coating thickness is consistent and uniform;

[0047] Fourth, by using parallel computing technology to allocate computing tasks to multi-core processors, optimize the spraying points, and perform spraying according to the priority sequence, the spraying efficiency is significantly improved;

[0048] Fifth, use a support vector machine classifier to predict the positions of high-risk areas of stress concentration and points with low adhesion uniformity, and adjust the spraying parameters additionally for these points, effectively improving the consistency of the coating thickness.

[0049] In summary, the present invention solves the performance bottleneck problem of traditional soft-sealed valves in high-temperature environments. At the same time, through three-dimensional scanning modeling, adaptive mesh generation, and dynamic spraying control technologies, the uniformity and adhesion performance of the polyvinylidene chloride coating are significantly improved, the processing time of soft-sealed valves is shortened, and the production efficiency is significantly increased, providing a reliable technical guarantee for the application of manufacturing soft-sealed valves that can still operate well in high-temperature environments. Brief Description of the Drawings

[0050] Figure 1 It is a flowchart of the processing method of Embodiment 1 of the present invention;

[0051] Figure 2 It is a specific flowchart of step (1) in Embodiment 2 of the present invention;

[0052] Figure 3 It is a specific flowchart of step (2) in Embodiment 2 of the present invention;

[0053] Figure 4 It is a flowchart of the processing method of Embodiment 3 of the present invention. Detailed Description of the Invention

[0054] Embodiment 1: The method for numerically controlling the processing of polyvinylidene chloride soft-sealed valves provided by the present invention, referring to Figure 1 as shown, the method includes the following steps:

[0055] (1) Obtain the three-dimensional scan data of the rubber ring surface, generate a high-resolution point cloud model, extract the geometric parameters of local wrinkles and high-curvature regions for non-uniform curvature features, and obtain the initial surface feature distribution;

[0056] (2) According to the initial surface feature distribution, use an adaptive mesh generation algorithm to generate a multi-scale mesh structure, increase the mesh density in complex regions to cover wrinkles and high-curvature parts, and reduce the density in flat regions to reduce the calculation amount, and determine the preliminary point layout;

[0057] (3) Spray polyvinylidene chloride on the rubber ring according to the point layout obtained in step (2).

[0058] The polyvinylidene chloride is dissolved in tetrachloroethane, with 5 g of polyvinylidene chloride dissolved in every 100 mL of tetrachloroethane. The gradient heating method is used for dissolution. First, it is pre-dissolved at 40°C for 0.3 h, then the temperature is raised to 50°C and stirred for dissolution for 0.3 h. Finally, the temperature is raised to 60°C and stirred for dissolution for 0.5 h to obtain the spraying liquid, and then the spraying liquid is sprayed onto the rubber ring by the electrostatic spraying method.

[0059] The working condition parameters of the electrostatic spraying are as follows: the voltage of the electrostatic spraying is 50 kv, the atomizing air pressure is 0.2 Mpa, the distance from the spray gun to the rubber ring is 15 cm, and the liquid flow rate of the spraying liquid is 0.8 m / s. The thickness of the polyvinylidene chloride sprayed on the rubber ring is 20 μm.

[0060] The applicant of the present invention unexpectedly found that for the soft-sealed valve made by coating polyvinylidene chloride on the rubber ring, the working temperature of the soft-sealed valves such as fluororubber and nitrile rubber commonly used in the prior art can be increased from 230°C to 300 - 500°C. At the same time, in order to ensure that the coating uniformly wraps the rubber ring, the present invention scans the three-dimensional surface of the rubber ring, and can clearly know the flat or concave-convex areas of the rubber ring. By increasing the grid density in the wrinkled areas and high-curvature parts, the spraying flow rate, spraying residence time, etc. can be more precisely controlled, so as to ensure that the coating thickness of the concave-convex parts and the flat parts is consistent, and improve the overall high-temperature resistance of the rubber ring.

[0061] Example 2: A method for numerically controlling the processing of a polyvinylidene chloride soft-sealed valve provided by the present invention includes the following steps: Dissolve polyvinylidene chloride in tetrachloroethane, with 10 g of polyvinylidene chloride dissolved in every 100 mL of tetrachloroethane. The gradient heating method is used for dissolution. First, it is pre-dissolved at 40°C for 0.5 h, then the temperature is raised to 50°C and stirred for dissolution for 0.5 h. Finally, the temperature is raised to 60°C and stirred for dissolution for 1 h to obtain the spraying liquid, and then the spraying liquid is sprayed onto the rubber ring by the electrostatic spraying method.

[0062] The voltage of the electrostatic spraying is 60 kv, the atomizing air pressure is 0.6 Mpa, the distance from the spray gun to the rubber ring is 25 cm, and the liquid flow rate of the spraying liquid is 1.5 m / s. Finally, the thickness of the polyvinylidene chloride sprayed on the rubber ring is 50 μm.

[0063] In order to ensure that the coating uniformly and consistently wraps the rubber ring on the surface of the rubber ring without dead corners in all directions, and at the same time can produce this soft-sealed valve with high yield, when spraying, the present invention further includes the following steps: (1), referring to Figure 2As shown, obtain the three-dimensional scan data of the rubber ring surface, generate a high-resolution point cloud model, extract the geometric parameters of local folds and high-curvature regions for non-uniform curvature features, and obtain the initial surface feature distribution: Collect the three-dimensional scan data of the rubber ring surface through a laser scanner to generate the first point cloud model, and use a voxel filtering method to downsample the first point cloud model to obtain the second point cloud model. Extract the non-uniformity features from the second point cloud model, calculate the normal vector and curvature value of each point for the curvature features, and determine the first curvature distribution. For the regions in the first curvature distribution where the curvature value exceeds the preset threshold, extract the coordinate positions of the local folds and high-curvature regions to obtain the first region set. Classify the points in the first region set through the random forest algorithm, judge the boundaries of the local folds and high-curvature regions, and generate the second region set. Calculate the geometric parameters of each region from the second region set, including area, perimeter, and mean curvature, to obtain the first parameter set. Use the principal component analysis method to perform dimensionality reduction processing on the first parameter set, extract the dominant components of the surface features, and generate the second parameter set. Group the data points in the second parameter set through the clustering algorithm, determine the distribution pattern of the initial surface features, and obtain the final distribution result;

[0064] Specific working principle: When obtaining the three-dimensional scan data of the rubber ring surface through a laser scanner, a laser beam with a wavelength of 905 nanometers is used to scan at a speed of 50,000 points per second to generate a high-resolution point cloud model. The point cloud model contains approximately 1 million data points with a resolution of 0.2 millimeters, ensuring that the minute details of the surface can be accurately captured. When processing the point cloud data, first use a point cloud filtering method based on the K-nearest neighbor algorithm, set the K value to 30, remove the noise points and smooth the data. Then, adopt a feature extraction algorithm based on curvature analysis to calculate the local curvature value of each point, and set the curvature threshold to 1 mm-1 to identify non-uniform curvature regions. For local folds and high-curvature regions, use a surface fitting method based on the least squares method, with the fitting radius range from 5 mm to 2 mm, and extract geometric parameters such as curvature radius and curvature direction. During the analysis process, calculate the main direction of the local surface through the principal component analysis (PCA) algorithm, and calculate the curvature eigenvalue using the curvature tensor to further determine the depth and width of the folds. Finally, compare the extracted geometric parameters with the initial surface feature distribution to generate a three-dimensional feature map of the rubber ring surface, providing data support for subsequent quality inspection and process optimization.

[0065] (2) According to the initial surface feature distribution, the specific operation process refers to Figure 3As shown, an adaptive mesh generation algorithm is used to generate a multi-scale mesh structure. The mesh density is increased in complex regions to cover the folds and high-curvature parts, and decreased in flat regions to reduce the computational amount, and the preliminary point layout is determined: A multi-scale mesh structure is generated by the adaptive mesh generation algorithm. The mesh density is adjusted for the fold parts and curvature parts in the complex regions of the surface features to cover the coordinate positions of local feature points. The number of points is reduced from the flat regions to determine the preliminary mesh point distribution. The voxel filtering method is used to downsample the preliminary mesh point distribution. The local feature details are retained for the complex regions with a high mesh density, and the number of redundant points is reduced from the flat regions with a low mesh density to obtain an optimized set of mesh points. By calculating the normal vector and curvature value of each point in the optimized set of mesh points, the local feature parameters are extracted for the fold parts and curvature parts, and the smooth curvature distribution is obtained from the flat regions to determine the curvature feature distribution pattern. The principal component analysis method is used to reduce the dimension of the curvature feature distribution pattern. The dominant geometric components are extracted for the local features in the complex regions, and the secondary curvature variables are removed from the flat regions to obtain a reduced-dimension feature set. The K-means clustering algorithm is used to group the data points in the reduced-dimension feature set. The high-curvature clusters are divided for the fold parts and curvature parts, and the low-curvature clusters are separated from the flat regions to determine the feature cluster distribution result. The coordinate positions of the boundary points are extracted for the high-curvature clusters in the feature cluster distribution result. The boundary points are classified by the random forest algorithm to judge the boundary range of the fold parts and curvature parts to obtain a set of boundary points.

[0066] Working principle: Based on the initial surface feature distribution, an adaptive mesh generation algorithm is used to generate a multi-scale mesh structure to optimize the computational efficiency and accuracy. First, the curvature value of each region is calculated by the curvature analysis algorithm, and the curvature threshold is set to 5 mm-1 to identify the high-curvature regions and fold parts. For the regions with a curvature greater than the threshold, a local refinement strategy is adopted to increase the mesh density to 10 nodes per square millimeter to ensure that the geometric features of the high-curvature parts are accurately captured. For the flat regions with a curvature less than the threshold, sparse mesh generation is used to reduce the mesh density to 2 nodes per square millimeter to reduce the consumption of computational resources. During the mesh generation process, an algorithm based on Delaunay triangulation is used to ensure the uniformity and continuity of the mesh. At the same time, the mesh node positions are adjusted by the iterative optimization algorithm to make the transition between the complex regions and non-complex regions of the mesh smoother. Finally, the generated multi-scale mesh structure contains approximately 500,000 nodes, and the mesh resolution is from 5 mm to 2 mm, covering all the feature regions on the surface of the rubber ring. This mesh structure provides an efficient and accurate computational basis for the subsequent finite element analysis and process simulation.

[0067] Example 3: A numerical control machining method for a polyvinylidene chloride soft-sealed valve provided by the present invention includes the following steps: The specific working process refers to Figure 4As shown, dissolve polyvinylidene chloride in tetrachloroethane. Dissolve 7 g of polyvinylidene chloride in every 100 mL of tetrachloroethane. Use the gradient heating method for dissolution. First, pre-dissolve at 40 °C for 0.4 h, then raise the temperature to 50 °C and stir for dissolution for 0.4 h. Finally, raise the temperature to 60 °C and stir for dissolution for 0.8 h to obtain the spraying liquid, and then use the electrostatic spraying method to spray the spraying liquid onto the rubber ring.

[0068] The voltage of electrostatic spraying is 55 kv, the atomizing air pressure is 0.4 Mpa, the distance from the spray gun to the rubber ring is 20 cm, and the liquid flow rate of the spraying liquid is 1.0 m / s. Finally, the thickness of the polyvinylidene chloride sprayed on the rubber ring is 35 μm.

[0069] To ensure that the coating is uniform and consistent on the surface of the rubber ring, and to completely wrap the rubber ring without dead angles in all directions, and at the same time to produce this soft-sealing valve with high yield, when spraying in the present invention, the following steps are also included: (1) Obtain the three-dimensional scan data of the rubber ring surface, generate a high-resolution point cloud model, extract the geometric parameters of local folds and high-curvature regions for non-uniform curvature features, and obtain the initial surface feature distribution: Collect the three-dimensional scan data of the rubber ring surface through a laser scanner to generate the first point cloud model, and use the voxel filtering method to downsample the first point cloud model to obtain the second point cloud model. Extract the non-uniformity features from the second point cloud model, calculate the normal vector and curvature value of each point for the curvature feature, and determine the first curvature distribution. For the regions in the first curvature distribution where the curvature value exceeds the preset threshold, extract the coordinate positions of local folds and high-curvature regions to obtain the first region set. Classify the points in the first region set through the random forest algorithm, judge the boundaries of local folds and high-curvature regions, and generate the second region set. Calculate the geometric parameters of each region from the second region set, including area, perimeter, and average curvature, to obtain the first parameter set. Use the principal component analysis method to perform dimensionality reduction processing on the first parameter set, extract the dominant components of the surface features, and generate the second parameter set. Group the data points in the second parameter set through the clustering algorithm, determine the distribution pattern of the initial surface features, and obtain the final distribution result;

[0070] Working principle: When obtaining the three-dimensional scan data of the rubber ring surface through a laser scanner, a laser beam with a wavelength of 905 nanometers is used to scan at a speed of 50,000 points per second, generating a high-resolution point cloud model. The point cloud model contains approximately 1 million data points with a resolution of 0.2 millimeters, ensuring that it can accurately capture the minute details on the surface. When processing the point cloud data, first, a point cloud filtering method based on the K-nearest neighbor algorithm is used, with the K value set to 30, to remove noise points and smooth the data. Then, a feature extraction algorithm based on curvature analysis is adopted to calculate the local curvature value of each point, and the curvature threshold is set to 1 millimeter^-1 to identify non-uniform curvature regions. For local wrinkles and high-curvature regions, a surface fitting method based on the least squares method is used, with the fitting radius range from 5 millimeters to 2 millimeters, to extract geometric parameters such as the curvature radius and curvature direction. During the analysis process, the principal direction of the local surface is calculated through the principal component analysis (PCA) algorithm, and the curvature eigenvalue is calculated using the curvature tensor to further determine the depth and width of the wrinkles. Finally, the extracted geometric parameters are compared with the initial surface feature distribution to generate a three-dimensional feature map of the rubber ring surface, providing data support for subsequent quality inspection and process optimization.

[0071] (2) According to the initial surface feature distribution, an adaptive mesh generation algorithm is used to generate a multi-scale mesh structure. The mesh density is increased in complex regions to cover wrinkles and high-curvature parts, and decreased in flat regions to reduce the computational amount, to determine the preliminary point layout: An adaptive mesh generation algorithm is used to generate a multi-scale mesh structure. The mesh density is adjusted for the wrinkle parts and curvature parts in the complex regions of the surface features to cover the local feature point coordinate positions, and the number of points is reduced in the flat regions to determine the preliminary mesh point distribution. A voxel filtering method is used to downsample the preliminary mesh point distribution. For the complex regions with a high mesh density, the local feature details are retained, and the redundant point numbers are reduced in the flat regions with a low mesh density to obtain an optimized mesh point set. By calculating the normal vector and curvature value of each point in the optimized mesh point set, local feature parameters are extracted for the wrinkle parts and curvature parts, and a smooth curvature distribution is obtained from the flat regions to determine the curvature feature distribution pattern. The principal component analysis method is used to perform dimensionality reduction on the curvature feature distribution pattern. The dominant geometric components are extracted for the local features in the complex regions, and the secondary curvature variables are removed from the flat regions to obtain a dimensionality-reduced feature set. The K-means clustering algorithm is used to group the data points in the dimensionality-reduced feature set. High-curvature clusters are divided for the wrinkle parts and curvature parts, and low-curvature clusters are separated from the flat regions to determine the feature cluster distribution result. The boundary point coordinate positions are extracted for the high-curvature clusters in the feature cluster distribution result, and the boundary points are classified through the random forest algorithm to determine the boundary range of the wrinkle parts and curvature parts, obtaining a boundary point set.

[0072] Working principle: Based on the initial surface feature distribution, an adaptive mesh generation algorithm is used to generate a multi-scale mesh structure to optimize the computational efficiency and accuracy. First, the curvature value of each region is calculated through a curvature analysis algorithm, and the curvature threshold is set to 5 mm-1 to identify high-curvature regions and wrinkled parts. For regions with a curvature greater than the threshold, a local refinement strategy is adopted to increase the mesh density to 10 nodes per square millimeter to ensure that the geometric features of the high-curvature parts are accurately captured. For flat regions with a curvature less than the threshold, sparse mesh generation is used to reduce the mesh density to 2 nodes per square millimeter to reduce computational resource consumption. During the mesh generation process, an algorithm based on Delaunay triangulation is used to ensure the uniformity and continuity of the mesh. At the same time, the mesh node positions are adjusted through an iterative optimization algorithm to make the transition between complex and non-complex regions of the mesh smoother. Finally, the generated multi-scale mesh structure contains approximately 500,000 nodes, with a mesh resolution ranging from 5 mm to 2 mm, covering all feature regions on the surface of the rubber ring. This mesh structure provides an efficient and accurate computational basis for subsequent finite element analysis and process simulation.

[0073] It also includes that during the spraying in step (3), according to the preliminary point layout, dynamic changes during the spraying process, and spraying effects, dynamic deformation prediction factors are extracted. The stress and deformation trend of the rubber ring during dynamic spraying are simulated through finite element analysis. Combining the transient spraying pressure data collected in real time, the influence of deformation on point offset is judged, and the spraying angle and spraying amount are dynamically adjusted. The spraying points are optimized according to the influence of deformation on point offset. A real-time calculation module is used to perform block processing on regions with a sharp increase in computational volume, obtain the spraying priority sequence for each region, and perform spraying according to this priority sequence. High-risk regions with stress concentration are extracted from the spraying priority sequence, and the positions of points with low adhesion uniformity are predicted through a support vector machine classifier to judge which points need to adjust the spraying parameters additionally to improve the coating thickness consistency and ensure the same spraying thickness and uniform spraying on the rubber ring.

[0074] The working principle is as follows: the initial geometric data of the rubber ring is collected through preliminary point layout, and the static force distribution before spraying is simulated by finite element analysis to obtain preliminary point layout data. According to the preliminary point layout data, combined with the real-time collected transient data of spraying pressure, the dynamic force change trend of the rubber ring during spraying is judged. For example, when the spraying pressure is 0.5MPa, the stress concentration coefficient of the edge area of the rubber ring reaches 1.8. Through the dynamic force change trend, the dynamic deformation prediction factor in the spraying process is extracted, the preliminary influence range of deformation on point offset is determined, and the area with deformation offset between 0.2mm and 0.5mm is calculated by finite element analysis. The dynamic deformation prediction factor is processed in blocks by the real-time calculation module to obtain the deformation distribution characteristics of each area. For example, the rubber ring is divided into 10 areas, and the deformation characteristics of each area are described by the Gaussian distribution model. The spraying priority sequence of each area is judged, for example, the calculation task is assigned to 8 processor cores using the MapReduce algorithm. According to the spraying priority sequence and combined with the dynamic deformation prediction factor, the distribution position of the high-risk area of stress concentration is obtained, for example, the stress concentration coefficient in area 3 and area 7 exceeds 2.0. The high-risk area of stress concentration is analyzed by the support vector machine classifier to determine the location of the point with low adhesion uniformity. For example, the radial basis function kernel classifier is used to identify 5 points with adhesion uniformity less than 90%. According to the location of the point with low adhesion uniformity and combined with the transient data of the spraying pressure, the point set that needs to adjust the spraying parameters is determined, for example, the spraying pressure at point A and point B needs to be adjusted from 0.5MPa to 0.6MPa. The point set that needs to adjust the spraying parameters is optimized by the real-time calculation module to obtain the dynamic adjustment scheme of the spraying angle and spraying amount, for example, the spraying angle is adjusted from 45 degrees to 50 degrees, and the spraying amount is increased from 0.2g to 0.25g.

[0075] It should also be noted that the electronic components used to complete the above work are all commercially available components. For example, the intelligent electrostatic spraying uses Hamanano PDS-PS series electrostatic spraying equipment. By setting pressure sensors, image acquisition equipment and other equipment at the output end of the electrostatic spraying, real-time collection of the pressure at the spraying output end and real-time feedback of the deformation data of the rubber ring caused by the spraying by the image acquisition equipment are completed. The main controller adopts the processing method described in this embodiment to ensure that the spray liquid is sprayed evenly and uniformly on the surface of the rubber ring without dead angles, thereby helping to improve the overall high temperature resistance of the rubber ring.

[0076] The following lists some experiments to illustrate the creative work done by the applicant of the present invention in obtaining the final technical solution:

[0077] 1. Set up a comparison example as follows:

[0078]

[0079]

[0080] II. For the spraying liquid and its spraying method according to Embodiments 1 to 3 of the present invention, polyvinylidene chloride is sprayed on the surface of an NBR nitrile rubber sealing ring to obtain a soft-sealing valve. Then, the spraying liquids of Comparative Examples 1-3 are used to spray polyvinylidene chloride on the surface of the NBR nitrile rubber sealing ring by the electrostatic spraying methods of Comparative Examples 1-3 to obtain soft-sealing valves for high-temperature resistance testing. The testing methods and results are as follows:

[0081]

[0082] When performing performance tests on the soft-sealing valves of the embodiments of the present invention and the soft-sealing valves of the comparative examples, it is found that when performing the fire resistance test on the valves of the embodiments of the present invention, the internal leakage of the valves of the embodiments of the present invention is required to be < 0.1×DN×P, while the valves of Comparative Examples 1-3 will have cracking or dissolution to varying degrees, resulting in unqualified internal leakage; when performing the thermal aging test, after the valves of this embodiment are tested, they can all exhibit the performance of tensile strength ≥ 80% and hardness change ≤ ±10 Shore A. However, the tensile strength of Comparative Examples 1-3 can no longer meet the requirement of ≥ 80%, and the hardness change is large; when performing the compression set test, the valves of this embodiment are compressed to 75% of the original thickness, placed at 200 °C for 22 hours, and the recovered thickness is measured after releasing the pressure. The calculated compression set rate ≤ 25%. However, the deformation of the valves of the comparative examples can be seen with the naked eye, and the calculated deformation rate is also much greater than 25%; when performing the high-temperature pressure cycle test, the leakage rate change of the valves of this embodiment is ≤ 0.1% / time, and the wear amount of the sealing surface ≤ 0.05 mm. However, the leakage amount of the valves of the comparative examples is large, and it can be seen with the naked eye that the wear is very serious, and the function of the valve can no longer be realized. At the same time, using the above experimental method for testing, this technology of the present invention can be used on NBR nitrile rubber, HNBR hydrogenated nitrile rubber, SIL silicone rubber, VITON fluororubber, FLS fluorosilicone rubber, EPDM ethylene-propylene-diene monomer rubber, CR chloroprene rubber, IIR butyl rubber, ACM acrylate rubber, NR natural rubber, and PU polyurethane rubber. Spraying polyvinylidene chloride on the surface of these rubbers can increase the working temperature from the original 90-250 °C to 300-500 °C.

[0083] It should be noted that in the above experimental verification of the present invention, at least 30 parallel repeated experiments were carried out on each sample, and the described effects are the means. For the specific testing methods and the judgment of compliance, the existing technologies are adopted, so they will not be repeated in detail here.

[0084] At the same time, in order to ensure that the valve product can indeed become a commercial product to promote the market, the applicant of the present invention also asked the partner to use the valve of the embodiment of the present invention in the petrochemical field for testing. Hot oil was passed into the pipeline at a temperature of 200-450°C and a pressure of 1.5 times the nominal pressure for 72 hours. After testing, it was found that the leakage rate and thermal aging of the valve of the embodiment of the present invention were up to standard after standard evaluation. The specific average leakage rate was approximately in the range of 0.06-0.08×DN×P, and the tensile strength retention rate after thermal aging was in the range of 82%-85%.

[0085] The applicant would also like to emphasize that, when obtaining the technical solution of the present invention, when looking for suitable spray liquid composition, electrostatic spraying and other working condition parameters, they were all obtained by chance after countless attempts, just like the above-mentioned test of valve performance. When developing the technical solution of the present invention, it was found that inappropriate spray liquid composition would corrode the rubber ring. When the spray condition parameters are not suitable, dangers such as overelectricity and sparks may occur. If the thickness of the spray coating is too thin, the effect is not good. If it is too thick, the cost will increase, but it will not help the high temperature resistance. These development processes are too redundant, so only some comparative examples are selected for illustration, and they will not be elaborated in detail here.

[0086] In addition, in order to produce the valve of the present invention with high output rate and low cost, the automated production method adopted by the present invention can ensure that the uniformity deviation of the coating thickness of the produced valve is about ±1%, and at the same time, it is highly efficient, and is very suitable for large-scale commercial promotion both in terms of production cycle and cost.

[0087] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A numerical control machining method for a polyvinylidene chloride soft-sealed valve, characterized in that, The method includes the following steps: (1) Obtain the three-dimensional scan data of the rubber ring surface, generate a high-resolution point cloud model, extract the geometric parameters of local folds and high-curvature regions for non-uniform curvature features, and obtain the initial surface feature distribution; (2) According to the initial surface feature distribution, use an adaptive mesh generation algorithm to generate a multi-scale mesh structure, increase the mesh density in complex regions to cover folds and high-curvature parts, reduce the density in flat regions to reduce the amount of calculation, and determine the preliminary point layout; (3) According to the point layout obtained in step (2), spray polyvinylidene chloride on the rubber ring.

2. The method according to claim 1, characterized in that, In step (3), polyvinylidene chloride is dissolved in tetrachloroethane to obtain a spraying solution, and then the spraying solution is sprayed onto the rubber ring by an electrostatic spraying method.

3. The method according to claim 2, wherein 5-10 g of polyvinylidene chloride is dissolved in every 100 mL of tetrachloroethane, and the dissolution is carried out by a gradient heating method. First, pre-dissolve at 40 °C for 0.3-0.5 h, then raise the temperature to 50 °C and stir to dissolve for 0.3-0.5 h, and finally raise the temperature to 60 °C and stir to dissolve for 0.5-1 h.

4. The method according to claim 2, wherein The voltage of electrostatic spraying is 50-60 kv, the atomizing air pressure is 0.2-0.6 Mpa, the distance from the spray gun to the rubber ring is 15-25 cm, and the liquid flow rate of the spraying solution is 0.8-1.5 m / s.

5. The method according to claim 1, characterized in that, The thickness of the polyvinylidene chloride sprayed on the rubber ring is 20-50 μm.

6. The method according to claim 1, wherein The said step (1) includes the following steps: a. Collect the three-dimensional scan data of the rubber ring surface through a laser scanner, generate a first point cloud model, and perform downsampling processing on the first point cloud model by a voxel filtering method to obtain a second point cloud model; b. Extract the non-uniformity features from the second point cloud model, calculate the normal vector and curvature value of each point for the curvature features, and determine the first curvature distribution; c. For the regions in the first curvature distribution where the curvature value exceeds the preset threshold, extract the coordinate positions of local folds and high-curvature regions to obtain a first region set; d. Classify the points in the first region set by a random forest algorithm, judge the boundaries of local folds and high-curvature regions, and generate a second region set; e. Calculate the geometric parameters of each region from the second region set, including area, perimeter, and mean curvature, to obtain a first parameter set; f. Perform dimensionality reduction processing on the first parameter set by a principal component analysis method, extract the dominant components of the surface features, and generate a second parameter set; g. Group the data points in the second parameter set by a clustering algorithm, determine the distribution pattern of the initial surface features, and obtain the final distribution result.

7. The method according to claim 1, characterized in that, The said step (2) includes the following steps: a. Generate a multi-scale mesh structure by an adaptive mesh generation algorithm, adjust the mesh density for the fold parts and curvature parts in the complex regions of the surface features, cover the coordinate positions of local feature points, reduce the number of points in the flat regions, and determine the preliminary mesh point distribution; b. Perform downsampling processing on the preliminary mesh point distribution by a voxel filtering method, retain the local feature details for the complex regions with a high mesh density, reduce the number of redundant points in the flat regions with a low mesh density, and obtain an optimized mesh point set; c. By calculating and optimizing the normal vectors and curvature values of each point in the grid point set, local feature parameters are extracted for the wrinkled parts and curvature parts, a smooth curvature distribution is obtained from the flat area, and the curvature feature distribution pattern is determined; d. The principal component analysis method is used to reduce the dimension of the curvature feature distribution pattern, the dominant geometric components are extracted for the local features in the complex area, and the secondary curvature variables are removed from the flat area to obtain a reduced-dimension feature set; e. The K-means clustering algorithm is used to group the data points in the reduced-dimension feature set, high-curvature clusters are divided for the wrinkled parts and curvature parts, low-curvature clusters are separated from the flat area, and the feature cluster distribution result is determined.

8. The method according to claim 1, wherein In step (3) during spraying, according to the preliminary point layout, dynamic changes during the spraying process, and spraying effects, dynamic deformation prediction factors are extracted. The stress and deformation trend of the rubber ring during dynamic spraying is simulated through finite element analysis. Combining with the transient spraying pressure data collected in real time, the influence of deformation on point offset is judged, and the spraying angle and spraying amount are dynamically adjusted.

9. The method according to claim 8, wherein Step (3) further includes: when dynamically adjusting the spraying angle and spraying amount, optimizing the spraying points according to the judged influence of deformation on point offset, using a real-time calculation module to perform block processing on the area with a sharp increase in the calculation amount, obtaining the spraying priority sequence for each area, and spraying according to this priority sequence.

10. The method according to claim 9, wherein In step (3), high-risk areas with stress concentration are extracted from the spraying priority sequence, the positions of points with relatively low adhesion uniformity are predicted through a support vector machine classifier, and it is judged which points need to additionally adjust the spraying parameters to improve the coating thickness consistency.

Images