Data-driven Fault Detection for Networked Control System Based on Implicit Model Approach

Chen Zhaoxu,Fang Huajing
DOI: https://doi.org/10.1109/chicc.2016.7554806
2016-01-01
Abstract:In this paper, we propose a data-driven fault detection algorithm for the nonlinear system with network-induced time delay. Double-layer Quasi Takagi-Sugeno fuzzy modeling method is employed to represent the system. Quasi fuzzy model addresses the issue of time delay on the inner layer. Classical fuzzy model depicts the nonlinearity of the networked control system (NCS) on the outer layer. We apply implicit model approach to achieve the data-driven fault detection for the system. The implicit model approach provides an alternative perspective of the recently developed subspace-based data-driven approach. Inspired by the method of parallel distributed compensation, we construct the overall data-driven fault detection algorithm for the nonlinear NCS. The validation of the proposed algorithm is proved on the well-established Tennessee-Eastman process.
What problem does this paper attempt to address?