Font Size: a A A

Research On Intelligent Modeling Method Of Car Body Structure Oriented To Conceptual Design

Posted on:2022-01-22Degree:MasterType:Thesis
Country:ChinaCandidate:J F HanFull Text:PDF
GTID:2492306509494434Subject:Vehicle Engineering
Abstract/Summary:PDF Full Text Request
The conceptual design stage of the car body is an important stage to ensure the performance of the car body structure design.The predictive design and analysis of the car body structure in the early stage of design can reduce the risk in the subsequent development process and shorten the whole design cycle.In the conceptual design stage,in order to rapidly evaluate the design scheme of the body structure and shorten the design cycle,it is a key problem to quickly establish the parametric conceptual model of the body.Based on the deep learning method,this paper proposes a scheme to quickly establish a 3D conceptual model based on conceptual design stage sketch of the car body,and independently develops a parameterized intelligent modeling module of the car body conceptual structure.The main contents of this paper are as follows:Based on the deep learning-based face key point detection algorithm and human body key point detection algorithm,and a key point detection algorithm of car design sketch is proposed.By improving Hourglass Network,the 2D coordinate position of car design key points can be obtained directly from a design sketch.Data sets are built by collecting design sketches for different models,the 2D coordinates of the key points of the car body in the sketch are coded,two mainstream methods in the field of key point detection are compared and analyzed,they are coordinate regression method and heat map regression method respectively,finally choosing heat map methods to return the coordinates of point location,the heat map method can make the network easier to train and the model detection precision is higher.Based on an open-source geometry engine library called Open Cascade,a graphical user interface library called Py Qt and Excel spreadsheet,the intelligent modeling module of the body structure conceptual design system S-iVCD is developed and constructed.The module adopts the Wizard interface to realize the automatic detection of key points from importing the body design sketch.To establish the intelligent and integrated design process of the three-dimensional parametric conceptual model of the car body,The modeling time of the conceptual model of the body structure in the conceptual design stage is greatly shortened and the design efficiency of the model is improved.In order to realize the convenient human-computer interaction function,the intelligent modeling module developed four sub-modules of hard point,beam,chassis and plate,which guided the user to successfully complete the creation of the conceptual model of the body structure,and supported manual adjustment or modification of the model during the modeling process.At the same time,it also provides the data conversion function,which can output the STEP format concept model of the car body and realize the data interaction with other 3D CAD/CAE software.All the design parameters of the model are automatically stored in the external Excel spreadsheet to ensure the synchronization of the model and design parameters.Finally,an example is given to verify the results,input the design sketch of a certain vehicle into the system,and quickly establish the corresponding parametric geometric conceptual model of the vehicle by manipulating the interface,then show the opening effect of the model in UG software,which proves the feasibility of the proposed scheme and the effectiveness and practicability of the S-iVCD system intelligent modeling module.
Keywords/Search Tags:Concept model of body structure, Deep learning, Key point detection, Software engineering
PDF Full Text Request
Related items