The stretch forming process is used to make structural sheet metal parts in the aerospace industry. The development of stretch forming tools has long been plagued by significant challenges. First, the low production volumes within the aerospace industry and the large numbers of stretch formed parts make the process capital intensive. Second, the development of stretch forming tooling has long been more of an art than a science. This results in poorly designed tools, poor quality parts, and lengthy tooling development cycles. A stretch forming tool capable of rapid reconfiguration was previously designed to address these issues. This tool is used in conjunction with a self-tuning shape control algorithm, which guides the die to the correct shape. There have been many simulations, and lab scale successes with these algorithms, but production scale implementations have experienced difficulties. These problems are related to the method of system identification and process variation. To better understand these issues, analysis and simulation are performed on the various forms of the algorithm. These investigations led to a greater understanding of the algorithms and the synthesis of an improved algorithm. In conclusion, a greater understanding of previously developed algorithms is presented. The system identification is mapped as a Point Spread Function applied through a cyclic convolution. This view provides insight into how the system identification is applied and allows system coupling to be quantified. Furthermore, through improved understanding a new algorithm is synthesized. This new algorithm offers an implementable solution that is optimized for performance, robustness to variation, and ease of use.
展开▼
