Combining genetic algorithms and fault tree analysis in reliability/cost optimization for critical complex systems

The early evaluation of dependability requirements for complex critical systems is a hard task which is usually accomplished by means of well-established formal modeling techniques. Product cost optimization is also necessary due to well-known difficulty to perform modifications in a stable and assessed system architecture. In this paper we propose a new methodology aiming at the integration of optimization techniques with formal methods in order to provide an efficient search of optimal reliability/cost trade-offs for complex and critical systems. The proposed approach is based on the combined use of Multi-Objective Genetic Algorithms, for optimization of complex functions, and Fault Trees, used to model and evaluate reliability aspects of systems. The methodology has been applied on a generic Programmable Logic Controller reliability fault tree model.