A Behavioral Approach to Direct Data-Driven Fault Detection
Please login to view abstract download link
Efficient and reliable fault detection methods are needed for monitoring and evaluation of processes, e.g., in structural health assessment. Most existing methods, however, rely on a priori given model. Thus, these methods require a model suitable for fault detection. Obtaining such a model is nontrivial and is often the bottleneck in applications. Direct data-driven methods were recently developed in signal processing and control. These methods avoid the model identification step and were shown to outperform state-of-the-art model-based methods in practical applications. In this paper, we propose a direct data-driven method for fault detection. The monitored process is modeled as a linear time-invariant system with unobserved deterministic disturbance. We use the behavioral approach to systems theory in order to define a representation invariant measure for the distance between data and model. The main contribution of the paper is computing the distance directly from offline and online data without parametric model identification. A second contribution of the paper is a direct data-driven method for input estimation.