Pseudo-Inverse Determination of Design Parameter Boundaries for Systems Subject to Stochastic Disturbances and Uncertainties

In this paper a method for efficient determination of acceptable design parameter ranges during early development phases is introduced, which takes un-certainties into account. Development of safety-critical systems is related to require-ments for critical events that must not be violated with a probability higher than a usually very low threshold. During requirements derivation and specification, ad-missible design parameter combinations must be determined that fulfill the proba-bilistic top level requirements and hence can be used for subsequent system design. This paper describes a gradient-based method for determination of admissible de-sign parameter ranges for dynamic systems subject to plant uncertainties and sto-chastic disturbances. The presented algorithm uses knowledge of a single point on the design parameter boundary to evaluate the whole parameter boundary using un-certain gradients. Considerable performance enhancements are achieved, especially if there is only a minor correlation between the effects of individual design param-eters. Functionality and performance of the algorithm are demonstrated exemplarily by deriving requirements for an automatic flight control system for close formation flight of two fixed-wing aircraft.     «

In this paper a method for efficient determination of acceptable design parameter ranges during early development phases is introduced, which takes un-certainties into account. Development of safety-critical systems is related to require-ments for critical events that must not be violated with a probability higher than a usually very low threshold. During requirements derivation and specification, ad-missible design parameter combinations must be determined that fulfill the proba-bilistic top le...     »