Data-driven fault detection for large-scale network systems: A mixed optimization approach

Highlights：

• The fault detection problem for large-scale network systems with unknown system dynamic matrices is investigated within the data-driven framework.

• A data-driven subspace identification approach is developed to estimate the unmeasurable interconnection signals composed by neighboring subsystems states.

• An H−=H∞ mixed optimization method is proposed to improve the fault detection performance by enhancing the sensitivity of residual to faults and the robustness against measurement noises. By means of this method, the actuator faults with smaller magnitude can be detected compared with the existing methods ignoring fault sensitivity.

• Moreover, the proposed H−=H∞ mixed optimization approach is also applicable to other types of faults, such as sensor faults.