Sensitivity-Analysis Method for Inverse Simulation Application

Abstract

An important criticism of traditional methods of inverse simulation that are based on the Newton–Raphson
algorithm is that they suffer from numerical problems. In this paper these problems are discussed and a new method
based on sensitivity-analysis theory is developed and evaluated. The Jacobian matrix may be calculated by solving a
sensitivity equation and this has advantages over the approximation methods that are usually applied when the
derivatives of output variables with respect to inputs cannot be found analytically. The methodology also overcomes
problems of input-output redundancy that arise in the traditional approaches to inverse simulation. The sensitivityanalysis
approach makes full use of information within the time interval over which key quantities are compared,
such as the difference between calculated values and the given ideal maneuver after each integration step.
Applications to nonlinear HS125 aircraft and Lynx helicopter models show that, for this sensitivity-analysis method,
more stable and accurate results are obtained than from use of the traditional Newton–Raphson approach.