|
|
||
|
|
|
|
This is Fig. 3 from the 2007 axialcomp paper. This slide shows the PANDA2 results for Case 1 in Table 4 (listed two slides ago). This slide shows design iterations during an execution of SUPEROPT, a PANDA2 processor the purpose of which is to seek a “global” optimum design.
The word, "global", is in quotes because PANDA2 cannot rigorously find the global optimum, but finds an optimum design the weight of which might be close to that of the global optimum because optimum designs are found starting from many different points in design space during a single execution of the PANDA2 processor, SUPEROPT.
Each “spike” in the plot shown here corresponds to a new starting design, which (as explained in [1D, 1K]) is generated randomly in a manner consistent with all linking and inequality constraints.
Five PANDAOPTs per AUTOCHANGE were specified for the execution of SUPEROPT. See Table 3 of the paper for a typical PANDA2 runstream that includes several executions of SUPEROPT.
(From: AIAA 48th Structures, Structural Dynamics, and Materials Conference, Paper no. AIAA-2007-2216, 2007)
Page 40 / 49