Wrinkle-predicting process for hydroforming
AssigneeTextron Innovations Inc.
InventorsPravin N. Kulkarni, Ashwini Shashikiran, Prabhakar Shashikiran, Subhrangshu Datta, Mark Virginia
Life of this patent
Estimating the term…
Abstract
Disclosed is a system and process which is a computer aided application using artificial intelligence modeling. This artificial intelligence model is able to predict the propensity for sheet metal parts to wrinkle when manufactured using a hydro forming process. The model uses geometric and material parameters to predict defects and facilitates defect predictions certainty into the ordinarily arduous redesign process directed at eliminating defects.
Claims9
A system for predicting wrinkling in a part to be manufactured according to a new metal part design, said system comprising:
- a computing device;
- a computer process running on said computing device, said process adapted to receive:
- (i) a plurality of part design parameters for said metal part design, and (ii) and a part design material for said metal part design into said computer process;
- wherein said process accesses a wrinkle-prediction model developed using an artificial neural network and then mathematically manipulates said plurality of part design parameters using said wrinkle-prediction model to predict the existence of said wrinkling in an actually manufactured part;
- and
- wherein said process provides a prediction of the possibility of said wrinkling of the part based on geometric and material characteristics of the part design.
A method for predicting wrinkling in a part design for a part to be constructed of a sheet metal material, said design including a plurality of selected design parameters, said method comprising:
- developing a wrinkle-prediction model, said model established by (i) conducting physical experiments using test parts comprised of said sheet metal material according to a plurality of experimental parameter sets;
- (ii) recording wrinkle information derived from results obtained regarding said test parts manufactured according to said experimental parameter sets, and then (iii) mathematically manipulating said wrinkle information to develop an artificial neural network for predicting said wrinkling;
- receiving said selected design parameters including geometric and material characteristics of a part to be designed;
- and
- using said wrinkle-prediction model to provide a prediction of the possibility of said wrinkling of the metal part based on geometric and material characteristics of the part design.
A method of predicting wrinkling in a part design, said part design including a plurality of selected part design parameters and a metal part design material, said method comprising:
- developing a wrinkle-prediction model on a database using an artificial neural network;
- providing a computer process;
- receiving said part design parameters and said metal part design material into said computer process, wherein the design parameters include geometric and material characteristics of a part to be manufactured in accordance with said part design, said geometric characteristics including at least one of part radius, sector angle, sheet thickness, fillet radius, and flange length;
- accessing said wrinkle-prediction model on said database and using said computer process;
- predicting said wrinkling in said part design using said wrinkle-prediction model;
- and
- displaying a result to a user;
- said result including predicting the possibility of said wrinkling of the part if the part were to be manufactured in accordance with said part design parameters.
Dependencies are inferred from the claim text.