Optimum calibration of a corrosion rate instrument using information gain criterion within a Bayesian framework

Structural health monitoring (SHM) and non-destructive testing (NDT) provide necessary information in assessing the current condition of an ageing structure. However, measurements obtained from SHM and NDT can be erroneous and it is important that the uncertainties associated with these measurement errors are quantified through instrument calibration. Considering that the process of calibration is expensive and time consuming, a novel optimum instrument calibration methodology is proposed in this work, which uses an information theoretic criterion to decide on the optimum number of calibration data points required. The proposed methodology is demonstrated through the calibration of a corrosion rate instrument with measurement uncertainty. The calibrated model is validated using Bayes' factor as the validation metric. Case studies demonstrate 42.5% and more reduction in the required number of calibration data points. The proposed method also suggests an optimum sequence of data collection for calibration.