Techniques for handling missing data: Applications to online condition monitoring

The use of inferential sensors is a common task in online fault detection in various control applications. A problem arises when sensors fail while the control system is designed to make a decision based on the data from those sensors. Various techniques to handle missing data are discussed in this paper. Firstly, a novel algorithm that classifies and regresses in the presence of missing data is proposed. The algorithm is tested for both classification and regression problems. Secondly, an estimation algorithm that uses an ensemble of regressors is proposed. Hybrid genetic algorithms and fast simulated annealing are used to predict the missing values and their results are compared. Results show that fast simulated annealing is slightly faster than the hybrid GA for the problem investigated. Results provide a valuable insight into dealing precisely with missing data. ICIC International