Parameter estimation in degradation modelling: A case study using condition monitoring data from wood pole inspections

Estimating model parameters is a major challenge in degradation modelling. This paper presents an approach where sojourn time distribution parameters were estimated based on censored observations from condition monitoring. A maximum likelihood approach for parameter estimation is described, and the requirements related to type and amount of data are discussed. It is shown that feasible parameter estimates can be obtained when sufficient amount of data is available, even though the data is heavily censored. A case study is presented where data from wood pole inspections were applied for parameter estimation. The data were collected by a Norwegian electricity distribution company during condition monitoring of their power lines. The case study showed that parameter estimates can be established based on such type of data. However, the case study also revealed challenges and limitations because of missing or inconsistent information in the database. The paper discusses these aspects. Furthermore, suggestions on improvement of data collection and parameter estimation are given.