Research On Mechanical Performance Design System Based On 3D Printing And Deep Learning

This paper proposes a new idea of mechanical property design based on deep learning and 3D printing technology.With the development of the economy,more and more complex super-engineerings have been built,and the construction has placed higher demands on the performance and construction of the structure.How to deal with more complex stress situations,more complex structural systems,and more stringent construction conditions is one of the major challenges in the future development of civil engineering.However,the existing structural design process is difficult to adapt to the new structural design.In this context,,this paper proposes a new performance-based direct design-manufacturing system.The main contents are as follows:Firstly,this paper describes in detail a performance-based direct designmanufacturing system,including its background,content and significance.The advantages of using this method are demonstrated by comparison with existing methods.Secondly,this paper studies the mechanical performance design method based on deep learning technology,and introduces the logical framework and main content of the design method in detail.The main content is divided into three stages,namely preparation based on traditional methods,modeling based on deep learning and prediction based on model.For different stages,this paper introduces the parameterization method,data set acquisition method,deep learning method,model training and prediction method in detail.The feasibility of the design method was verified by a simple supported I-beam case with an elliptical hole in the web,and a detailed template using the design method was provided.The forms of construction designed using deep learning technology are mostly unconventional and complex,which can just take advantage of 3D printing technology.Thirdly,this paper studies the performance of typical 3D printed materials and the typical 3D printing component form.In this paper,the research is carried out from the beam,column and node level,and the resin material is processed by fused deposition modeling(FDM)technology.16 porous cylindrical uniaxial compression tests,11 sandwich beam three-point bending tests and two assembled node frame loading test were carried out.Through the above experiments,this paper verifies that the complex structural form will bring about the diversity of mechanical behavior,not only can optimize the structural form,but also can obtain new mechanical behavior by changing the structural form.In this paper,the 3D Concrete Printing process is used to process concrete,which gives the unique properties of 3D printed concrete materials.The compressive strength test,splitting tensile test and elastic modulus test of a total of 20 test blocks were carried out.Finally,combined with deep learning and 3D printing technology,to build a lunar habitat as a case background,the application process of the direct design-manufacturing system is introduced through application cases.This paper introduces the technical path of proposing the expected mechanical behavior requirements,applying deep learning to mechanical design,solving the corresponding structural structure,and directly using 3D printing technology to produce and construct.The feasibility and accuracy of the proposed method are verified,which provides a reference for engineering design.