TABLE OF CONTENTS The use of inferential sensors is a common task for online fault detection in various control applications. A problem arises when sensors fail when the system is designed to make a decision based on the data from those sensors. Various techniques to handle missing data are discussed in this chapter. First, a novel algorithm that classifies and regresses in the presence of missing data online is presented. The algorithm was tested for using both classification and regression problems and was compared to an off-line trained method that used auto-associative networks as well as a Hybrid Genetic Algorithm (HGA) method and a Fast Simulated Annealing (FSA) technique. The results showed that the presented methods performed well for online missing data estimation. Second, an online estimation algorithm that uses an ensemble of multi-layer perceptron regressors, HGA and FSA and genetic programming is presented for missing data estimation and compared with a similar procedure that was trained off-line.
Fault detection is one of the most active research areas with several applications. There are a number of challenges faced by online detection and identification systems (Marwala, 2001; Benitez-Perez, Garcia-Nocetti, & Thompson, 2007; Vilakazi & Marwala, 2006; Vilakazi & Marwala, 2007a&b; Marwala, 2007; Hulley & Marwala, 2007). Marwala and Hunt (2000) proposed pseudo modal energies for fault detection in structures. The problem with the method they proposed is that for its successful implementation it had to be continuously retrained and this process required human intervention (Marwala, 2001). One of the biggest problems hindering...
