Abstract: In order to meet the demand of design for testability (DFT) for aviation electromechanical systems, a novel approach for system testability modeling and analysis is proposed which is based on object oriented Bayesian networks（OOBN）and the state-test association sensitivity index. The modeling approach can describe clearly the inter-dependent relationship between system faults and test symptoms, and reflect the hierarchical structure of complex systems. Based on the cross entropy principle in information theory, a novel index named state-test association sensitivity index is proposed together with its calculating methods. The index can reflect the impact of uncertainties in complex electromechanical system tests, and overcome the shortcomings of the test point analysis approach based on Shannon entropy. Combined with the testability model information, it can be used in quantitative testability analysis. The approach is used on an aircraft fuel system testability mo-deling and analysis, and the result shows the practicability of the proposed approach and index in aviation electromechanical system DFT.

Key words: design for testability, modeling, object oriented Bayesian networks, cross entropy, uncertainty analysis