(Georgia Institute of Technology, 1999-01)
Budianto, Irene Arianti; Olds, John R.; Baker, Nelson C.
Most of the analysis codes used in the design of aerospace systems are complex, requiring some expertise to set up and execute. Usually the Program to Optimize Simulated Trajectories (POST) fails to converge when its control variables are given a bad set of initial guesses, causing the trajectory to remain in the infeasible design region throughout the computations. The user then analyzes the output produced and relies on a set of heuristics, typically gained from experience with the program, to determine the appropriate modification to the problem setup that will guide POST in finding a feasible region and eventually converge to a solution. The potential benefits of employing knowledge-based system within a design environment have long been well known. Various methods of utilization have been identified. As a postprocessing guide, an expert system can distill information obtained from an analysis code, such as POST, into knowledge. The system then can emulate the human analyst's decision-making capability based on this collected knowledge. This paper describes the implementation of POST expertise in a knowledge-based system called CLIPS and demonstrates the feasibility of utilizing this integrated system as a design tool.
(Georgia Institute of Technology, 1998-01)
Olds, John R.; Budianto, Irene Arianti
Future space transportation vehicles may well rely on high speed airbreathing propulsion (ramjets and scramjets) to supply much of their motive power. Because of the tradeoff relationship between engine thrust and vehicle airframe weight, ascent trajectories are typically simulated using a constant dynamic pressure phase during airbreathing acceleration; dynamic pressure is increased to benefit vehicle thrust up to some fixed limit imposed by the vehicle structure. The constant dynamic pressure portion of the trajectory typically begins around Mach 2-3 and continues to the maximum airbreathing Mach number or until some convective aeroheating limit is reached. We summarize comparative research on three candidate guidance methods suitable for simulating constant dynamic pressure trajectories. These are generalized acceleration steering, linear feeedback control, and cubic polynomial control. All methods were implemented in POST (Program to Optimize Simulated Trajectories), an industry standard trajectory simulation code. Both quantitative and qualitative comparisons of these methods (i.e. in terms of computer processing time, number of required iterations for convergence, sensitivity to quality of initial values, accuracy and program robustness) are presented. Of the three methods, the linear feedback control approach is found to be the most efficient and robust, with good accuracy.