Adaptive control technologies that incorporate learning algorithms have been proposed to enable automatic flight control and vehicle recovery, autonomous flight, and to maintain vehicle performance in the face of unknown, changing, or poorly defined operating environments. In order for adaptive control systems to be used in safety-critical aerospace applications, they must be proven to be highly safe and reliable. Rigorous methods for adaptive software verification and validation must be developed to ensure that control system software failures will not occur. Of central importance in this regard is the need to establish reliable methods that guarantee convergent learning, rapid convergence (learning) rate, and algorithm stability. This paper presents the major problems of adaptive control systems that use learning to improve performance. The paper then presents the major procedures and tools presently developed or currently being developed to enable the verification, validation, and ultimate certification of these adaptive control systems. These technologies include the application of automated program analysis methods, techniques to improve the learning process, analytical methods to verify stability, methods to automatically synthesize code, simulation and test methods, and tools to provide on-line software assurance.

No abstract

[1] Measurements of NO Y condensation on cirrus particles during the SOLVE-I field campaign are analyzed and segregated based on altitude. Significant amounts of NO Y were found on the upper tropospheric ice particles; therefore condensation on ice appears to be an important method of NO Y removal from the gas phase at the low temperatures of the Scandinavian upper troposphere. For the data set collected on 23 January 2000, NO Y condensation on cirrus particles has different properties depending on whether the ice particles are sampled in the upper troposphere, where HNO 3 does not dominate NO Y , or in the lower stratosphere, where HNO 3 does dominate NO Y . Nitric acid becomes enriched in the gas phase as NO Y condenses on upper tropospheric ice crystals, indicating that a non-HNO 3 component of NO Y is condensing on upper tropospheric ice particles much faster and at higher concentrations than HNO 3 alone on this day. It is unclear which non-HNO 3 constituent of NO Y is condensing on upper tropospheric ice particles, although N 2 O 5 is the most likely species. This condensation of a non-HNO 3 component of NO Y is not universal in the upper troposphere but depends on the conditions of the air parcel in which sampling occurred, notably exposure to sunlight.

No abstract

[1] Measurements of NO Y condensation on cirrus particles found in stratospherically influenced air sampled during the SOLVE-I mission are analyzed and compared with data from other field studies of HNO 3 or NO Y condensation on ice. Each field study exhibits an order of magnitude data spread for constant HNO 3 pressures and temperatures. While others assumed this distribution is due to random error, the data spread exceeds instrument precision errors and instead suggests HNO 3 removal had not attained equilibrium at the time of sampling. During the SOLVE-I mission, condensation on ice was a significant sink for HNO 3 despite submonolayer surface coverages; we therefore propose condensation of HNO 3 on lower-stratospheric cirrus particles is controlled by kinetics and will occur at a kinetically limited rate. Furthermore, we suggest the low accommodation coefficient for HNO 3 on ice combined with relatively short-lived clouds causes highly scattered, limited HNO 3 uptake on cirrus particles. We couple laboratory data on the accommodation coefficient of HNO 3 on ice with field surface coverage data in order to generate a ''cloud clock'': a calculation to determine the age of a cloud parcel. Data from the aforementioned field studies are compared to theoretical models for equilibrium surface coverage on the basis of laboratory data extrapolated to atmospheric temperatures and HNO 3 pressures. This comparison is difficult because most of the atmospheric data are probably not at equilibrium and follow a condensation time curve rather than an equilibrium surface coverage curve. Finally, we develop a simple mathematical solution for the time required for HNO 3 condensation on ice.