Measuring Dependencies in Cyber-Physical Systems: Overhead Cranes Case Study

Abstract

Complex cyber-physical systems with multifunctional duties generally degrade simultaneously due to common factors such as the designed structure, the operating environment, and the exploitation history (we refer to exploitation when we include operation and maintenance processes together). Risk-oriented modeling is a well-established methodology for degradation analysis and prediction, and one of the major challenges is the modelling features of the marginals that describe, and store information related to the degradation process. Usually, the modelling process for each component of the system is analyzed independently, but in recent years the multivariate copulas developed and implemented in formal programming languages allow us to consider both sides of the modelling process in one (independent and dependent components). In this paper, we discuss the impact through sensitivity analysis of the paradigm shift between the assumption of independence or dependencies between components. The paper measures the impact of the decision in a parameterized scenario showing significant changes.