Importance analysis deals with the influence of individual system component on system operation. Thus, a lot of failure data should be collected to make the analysis more accurate. This paper mainly focuses on the numerical estimation of component importance in complex mechanical system which is considered as a multi-state system with few failure data. In order to evaluate components’ failure probability distribution by small sample data, a time integral importance measure (TIIM) approach is proposed. In this measure, we aim to measure component importance using the change of system performance caused by wiping off component failure data. On this basis, the dynamic importance fluctuation of a component can be measured by calculating criticality of each state of the component. The approach has been verified by probability analysis of CNC machine tools. The main contribution of this work is the proposed dynamic importance measure which can be used to identify the key state of a component that influences system performance most by small-sample data.

This content is only available via PDF.
You do not currently have access to this content.