Complexity Analysis for Model-Based Fault Diagnosis Systems

Three methodologies were examined to determine complexity of the Model-based Off-Nominal State Identification and Detection (MONSID) software including straightforward loop counting, comparisons to graph theory, and empirical validation. As predicted by straightforward counting, MONSID fits within O(n
⁴
), for nominal data, where n is a representation of some combination of Components, Nodes, and Connections. To support the analysis, different model topologies were defined to reduce coupling among the objects such that their influence could be parameterized. Five different model families were utilized to validate the analysis. Empirical testing has shown that the largest influencers to processing time are Components and the number of Connections. This corresponds with the analysis that compared MONSID Models with traditional graphs where the order is O(Components + Connections).