Flight-control system architecture optimization for fly-by-wire airliners

The design problem of a flight-control system on a large fly-by-wire airliner is to find combinations of actuator(s), power circuit(s), computer(s) for each movable, so as to fulfill the constraints imposed by the safety regulations, while keeping the resulting system weight as low as possible. The trend towards more electrical aircraft makes it harder and harder to determine, in a reasonable computer time, optimal architectures by traditional trial-and-error methods. Actually new power sources and actuator technologies offer many more implementation choices, and lead to an extreme combinatorial complexity. Therefore, traditional design processes have to be assisted with automatic optimization. This paper introduces a flight-control architecture optimization process, intended as a decision aid for system engineers at early stages of the flight-control architecture definition. We present an optimization model for the design process, based on a safety constraint and a weight criterion. We start by reducing the initial search domain by introducing the notion of Surface possible architecture. Surface possible architectures are subsets of the initial search domain witch are constructed by taking into account technological constraints and empirical practices. These above subsets are generated outside the optimization loops. Once the search has been reduced, we use an adaptation of branch-and-bound methods to solve the remaining discrete optimization problem. Finally, an application to the Airbus A340 roll control system is adressed. An