A risk-based active learning approach to inspection scheduling

Gaining the ability to make informed decisions regarding the operation and maintenance of structures and infrastructure provides motivation for the implementation of structural health monitoring (SHM) systems. However, descriptive labels for measured data corresponding to health-state information of the monitored system are often unavailable. This issue limits the applicability of fully-supervised machine learning paradigms for the development of statistical classifiers to be used in decision-supporting SHM systems. The current paper presents a risk-based active learning approach in which data-label querying is guided by the expected value of perfect information for incipient data points. In the context of SHM, the data-label querying process corresponds to the inspection of a structure to determine its health-state. The risk-based active learning process is demonstrated on a representative numerical case study. The results of the case study indicate that a decision-maker's performance can be improved via the risk-based active learning of a statistical classifier.