A design method, device and storage medium for a shield waterproof gasket

Abstract

The application discloses a shield waterproof sealing gasket design method and device and a storage medium. The method comprises the following steps: obtaining existing shield waterproof sealing gasket parameters to form a data set; constructing a shield waterproof sealing gasket identification network; training the shield waterproof sealing gasket identification network by using the data set to obtain a trained network; inputting a to-be-detected sealing gasket cross-section picture into the trained network to obtain completed sealing gasket cross-section geometric information; and obtaining a mapping relationship between waterproof capacity and assembly pressure and the geometric information of the completed sealing gasket cross-section and materials by using a machine learning method. The sealing gasket is designed under the constraint condition by using the constraint condition and the machine learning method. The application uses the big data of the elastic sealing gaskets of existing engineering examples to intelligently design the sealing gasket by using the machine learning, and the optimal design scheme under the constraint condition can be obtained, so that the application has great popularization and application value.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of shield waterproof gasket design, and particularly relates to a shield waterproof gasket design method, device and storage medium. BACKGROUND

[0002] The shield tunnel is a fabricated structure, which is formed by pipe segments connected by bolts, and there are a large number of joints. These joints are the main channels for water leakage of the tunnel, and joint waterproofing is an important guarantee for normal use of the shield tunnel. In China, the joints of the shield tunnel mainly use extrusion of the EPDM sealing gasket to achieve the purpose of waterproofing. By changing the cross-sectional shape and the number, shape and size of the holes of the sealing gasket, elastic sealing gaskets with different waterproofing capacities can be obtained.

[0003] At present, the cross-sectional design method of the elastic sealing gasket of the shield tunnel joint mainly adopts a method of numerical simulation combined with waterproofing test, which needs a large amount of nonlinear numerical simulation calculation and waterproofing test, and at least has the following problems:

[0004] 1. The cross section of the elastic sealing gasket used in the numerical calculation has a certain experience, and a relatively optimal cross section can be obtained only by repeated trial calculation, which leads to low design efficiency and increases the design workload;

[0005] 2. According to the numerical simulation, an optimal cross section needs to be selected by combining with the waterproofing test, and a plurality of repeated tests need to be performed, which is not economical;

[0006] 3. In the waterproofing test, it is difficult to determine the waterproofing capacity of the elastic sealing gasket. SUMMARY

[0007] In order to solve the technical problems of low efficiency, poor economy and difficulty in determining the waterproofing capacity in the existing shield waterproof gasket design, the present application provides a shield waterproof gasket design method, device and storage medium.

[0008] Specifically, the method comprises the following steps:

[0009] S1, acquiring existing collectable shield waterproof sealing gasket parameters to form a data set;

[0010] S2, constructing a shield waterproof sealing gasket recognition network;

[0011] S3, training the shield waterproof sealing gasket using the data set to obtain a trained network;

[0012] S4, inputting a to-be-detected sealing gasket cross-sectional picture into the trained network to obtain completed sealing gasket cross-sectional geometric information;

[0013] S5, according to the geometric information of the completed gasket section and other gasket parameters, the gasket design under the constraint condition is completed by using the constraint condition and the machine learning method.

[0014] A storage medium stores instructions and data for implementing a shield waterproof gasket design method.

[0015] A shield waterproof gasket design device comprises a processor and the storage medium; the processor loads and executes the instructions and data in the storage medium to implement a shield waterproof gasket design method.

[0016] The present application provides beneficial effects: using the existing engineering example of the elastic gasket big data, the intelligent design of the gasket is carried out through the machine learning, the optimal design scheme under the constraint condition can be obtained, and the method has great popularization and application value. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 is a schematic diagram of the method flow of the present application;

[0018] Figure 2 is the sectional form of several typical different types of gaskets;

[0019] Figure 3 is the input picture and output picture effect schematic diagram of the network;

[0020] Figure 4 is a schematic diagram of the gasket section size;

[0021] Figure 5 is a working schematic diagram of the hardware device of the present application. DETAILED DESCRIPTION

[0022] In order to make the purpose, technical scheme and advantages of the present application clearer, the embodiments of the present application will be further described below with reference to the drawings.

[0023] Please refer to Figure 1 , Figure 1 is a schematic diagram of the method flow of the present application;

[0024] The present application provides a shield waterproof gasket design method, comprising the following steps:

[0025] S1, obtain the existing shield waterproof gasket parameters that can be obtained, form a data set;

[0026] The shield waterproof gasket parameters include: gasket geometric information, material, waterproof capacity and assembly pressure.

[0027] Please refer to Figure 2 , Figure 2 is the sectional form of several typical different types of gaskets.

[0028] In the present application, the cross section, material parameters, waterproof ability and assembly pressure data of the shield waterproof sealing pad in the public literature and engineering cases are collected,

[0029] form a data set;

[0030] S2, build a shield waterproof sealing pad identification network;

[0031] It should be noted that the shield waterproof sealing pad identification network adopts MaskR-CNN network. Mask R-CNN is a method for end-to-end training of target detection and image segmentation tasks using deep neural networks, and its main features include:

[0032] 1. Use FPN (Feature Pyramid Network) for feature extraction. This network can extract and fuse features of different sizes into the final feature map.

[0033] 2. Use a standard convolutional neural network as the backbone network for target detection. This network can extract high-level features and directly regress the bounding box coordinates when detecting targets.

[0034] 3. Provide a differentiable Rol pooling module that can extract features of different size ROIs.

[0035] 4. Introduce a bounding box regression loss function and a mask classification loss function in the prediction stage to optimize the target detection results by updating the network parameters.

[0036] S3, train the shield waterproof sealing pad using the data set to obtain a trained network;

[0037] S4, input the sealing pad cross section picture to be detected into the trained network to obtain the completed sealing pad cross section geometric information;

[0038] As an embodiment, please refer to Figure 3 , Figure 3 is a schematic diagram of the input picture and output picture of the network;

[0039] In the case of incomplete input picture geometric size annotation, the size is completed after network processing.

[0040] As another embodiment, Opencv can also be used to identify the image based on the existing geometric size of the sealing pad image, obtain the geometric information of the sealing pad, and perform digital processing to obtain the complete geometric information of the elastic sealing pad.

[0041] When processing by Opencv, the outer contour shape, outer contour area, closed hole shape and area, open hole shape and area can be calculated. At the same time, the boundary of the hole can be parameterized to complete the completion of the geometric information.

[0042] The completed picture and the corresponding shield waterproof sealing gasket parameters can be supplemented to the data set, and the data set is updated.

[0043] S5, according to the geometric information of the completed sealing gasket surface and other sealing gasket parameters, using constraint conditions and machine learning method, complete the sealing gasket design under the constraint condition.

[0044] Please refer to Figure 4 , Figure 4 is a schematic diagram of the sealing gasket surface size;

[0045] Step S5 is specifically:

[0046] By machine learning method, the mapping relationship P between waterproof capacity and completed sealing gasket surface geometric information, material is obtained c =f(b i ,h j ,x k ,y k ,H a ) and the relationship P between assembly pressure and sealing gasket surface geometric information, material I =f(b i ,h j ,x k ,y k ,H a ); wherein P c represents waterproof capacity; b i represents the width of sealing gasket at different heights (mm), i=1, 2, 3, 4; h i represents the height of sealing gasket width change position (mm), j=1, 2, 3, as shown in Figure 4 x k represents the horizontal coordinate of sealing gasket hole boundary; y i represents the vertical coordinate of sealing gasket hole boundary, wherein k represents the hole boundary control point number; H a represents the Shore hardness of sealing gasket material (degree); P I represents the sealing gasket assembly pressure (N).

[0047] When the relationship P c =f(b i ,h j ,x k ,y k ,H a ) and P I =f(bi ,h j ,x k ,y k H a After that, the geometric information of the sealing gasket is determined based on the segment trench information. i and h j And the Shore hardness H of the sealing gasket material determined based on engineering experience. a The range, combined with the constraint condition, assembly pressure upper limit P I By using machine learning methods, the x-coordinate of the sealing gasket pore boundary that meets the design waterproofing requirements is obtained. k and the vertical coordinate y k This also yielded information including the outer contour shape and size b. i h j The geometric information of the sealing gasket, including the location and shape of the hole and material information.

[0048] Optionally, when relation P is obtained c =f(b i ,h j ,x k ,y k H a ) and P I =f(b i ,h j ,x k ,y k H a After that, the Shore hardness H of the sealing gasket material can be determined based on engineering experience. a The selectable range, combined with the upper limit of assembly pressure P under the constraint conditions. I Using machine learning methods, the geometric information of the outer contour of the sealing gasket that meets the design requirements for waterproofing is obtained. i and h j x-coordinate of the boundary of the sealing gasket hole k and the vertical coordinate y k This also yielded information including the outer contour shape and size b. i h j The geometric information of the sealing gasket, including the location and shape of the hole and material information.

[0049] As one example, the machine learning method used in this application can be linear regression algorithm, support vector machine algorithm, decision tree algorithm, random forest algorithm, neural network algorithm, etc.

[0050] As one embodiment, the location and shape of the holes are described using B-spline curves.

[0051] Specifically, the x-coordinate of the sealing gasket hole boundary is obtained. k and the vertical coordinate yk Then, the hole position and shape of the sealing gasket are described by a B-spline curve as follows:

[0052]

[0053] where P k =(x k ,y k ) T (k=0,…,n) are the control point coordinates of the closed B-spline curve, and n is the number of control points.

[0054] N k,l(u) is the l-th B-spline basis function of the corresponding k-th control point, which is generated by a monotone non-decreasing real number sequence U={u0,u1,…,u m}(m=n+l+1). The shape curve of the hole is determined by the control point coordinates on the hole.

[0055] According to the geometric parameters b i ,h j of the sealing gasket and the horizontal coordinates x k and the vertical coordinates y k of the hole boundary, the corresponding sealing gasket design diagram is matched from the updated data set.

[0056] Please refer to Figure 5 , Figure 5 is a hardware device working schematic diagram of an embodiment of the present application, and the hardware device specifically comprises: a shield waterproof sealing gasket design device 401, a processor 402, and a storage medium 403.

[0057] The shield waterproof sealing gasket design device 401: the shield waterproof sealing gasket design device 401 realizes the shield waterproof sealing gasket design method.

[0058] The processor 402: the processor 402 loads and executes the instructions and data in the storage medium 403 to realize the shield waterproof sealing gasket design method.

[0059] The storage medium 403: the storage medium 403 stores instructions and data; and the storage medium 403 is used to realize the shield waterproof sealing gasket design method.

[0060] The present application has the beneficial effect that: by using the existing engineering example elastic sealing gasket big data, the intelligent design of the sealing gasket is carried out through machine learning, the optimal design scheme under the constraint condition can be obtained, and the present application has great popularization and application value.

[0061] The above merely describes preferred embodiments of the present application and is not used to limit the present application, and any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method of designing a shield waterproofing sealant gasket, characterized by: The method comprises the following steps: S1, obtaining existing collectable shield waterproof sealing pad parameters to form a data set; S2, constructing a shield waterproof sealing pad recognition network; S3, training the shield waterproof sealing pad recognition network using the data set to obtain a trained network; S4, inputting a to-be-detected sealing pad cross-section picture to the trained network to obtain completed sealing pad cross-section geometric information; The completed picture and the corresponding shield waterproof sealing pad parameters can be supplemented to the data set to update the data set; S5, according to the geometric information of the completed sealing pad cross-section and other sealing pad parameters, using constraint conditions and machine learning methods, completing the sealing pad design under the constraint conditions; The shield waterproof sealing pad parameters include sealing pad geometric information, material, waterproof capacity and assembly pressure; The shield waterproof sealing pad recognition network adopts a MaskR-CNN network; Step S5 is specifically: The relationship between waterproof ability and geometric information of the gasket cross section after completion, material parameters is obtained by machine learning method P c = f ( b i ,h j ,x k ,y k ,H a ) and the relationship between assembly pressure and geometric information of the gasket cross section, material parameters P I = f ( b i ,h j ,x k ,y k ,H a ); wherein P c represents waterproof ability; b i represents the width of the gasket at different heights, i =1, 2, 3, 4; h j represents the height of the gasket width change position, j =1, 2, 3; x k represents the horizontal coordinate of the gasket hole boundary; y k represents the vertical coordinate of the gasket hole boundary; H a represents the Shore hardness of the gasket material; P I represents the gasket assembly pressure; When the relationship P c = f b i ,h j ,x k ,y k ,H a is obtained, the sealing gasket part geometric information b i and h j , and the sealing gasket material Shore hardness H a range determined according to engineering experience, and the constraint condition assembly pressure P I is not greater than the assembly force that the shield machine can provide, while P I is not greater than the assembly force that causes the shield segment groove to be damaged, through the constraint condition and the machine learning method, the sealing gasket design under the constraint condition is completed, and the sealing gasket geometric parameters satisfying the design requirements are obtained b i ,h j and the hole boundary abscissa x k and the ordinate y k ;​ After obtaining the transverse coordinates of the gasket hole boundary x k and the longitudinal coordinates y k Then, the B-spline curve is used to describe the position and shape of the gasket hole, as follows: In the formula P k = (x x k , y k ) T , k = 0, …, n is the coordinate of the control point of the closed B-spline curve, n is the number of control points; N k,l(u) for the corresponding first k control point of the l order B-spline curve basis function generated by a monotonically non-decreasing real number sequence U = { u 0, u 1, …, u m}, m = n + l +1; the shape curve of the hole is determined by the control point coordinates on the hole; According to the gasket geometry b i ,h j and the x-coordinate of the hole boundary x k and the y-coordinate y k Automatically match from the updated dataset to get the corresponding gasket design drawing.

2. A storage medium, characterized by: The storage medium stores instructions and data for realizing the shield waterproof sealing pad design method of claim 1.

3. A shield waterproof gasket design apparatus, characterized by: Comprise: A processor and a storage medium; the processor loads and executes the instructions and data in the storage medium to realize the shield waterproof sealing pad design method of claim 1.

Images