A critical look at deep neural network for dynamic system modeling

Neural network models become increasingly popular as dynamic modeling tools in the control community. They have many appealing features including nonlinear structures, being able to approximate any functions. While most researchers hold optimistic attitudes towards such models, this paper questions the capability of (deep) neural networks for the modeling of dynamic systems using input-output data. For the identification of linear time-invariant (LTI) dynamic systems, two representative neural network models, Long Short-Term Memory (LSTM) and Cascade Foward Neural Network (CFNN) are compared to the standard Prediction Error Method (PEM) of system identification. In the comparison, four essential aspects of system identification are considered, then several possible defects and neglected issues of neural network based modeling are pointed out. Detailed simulation studies are performed to verify these defects: for the LTI system, both LSTM and CFNN fail to deliver consistent models even in noise-free cases; and they give worse results than PEM in noisy cases.

Is Kalman Filter Optimal for Fault Detection?

Kalman filter is widely used for residual generation in fault detection. Kalman filter has several nice features: it leads to optimality in fault detection using some performance indices and it also leads to simple residual evaluation and threshold setting. In this work, using frequency-domain considerations, it will be shown that for common faults that are very low-pass, output error, or, zero-gain observer is better than Kalman filter; moreover, there exist optimal post-filters that can further improve detection performance; finally, it is shown that choosing KF-based residual then using the post-filters are also not optimal for common low-pass faults. Monte Carlo simulations are performed to validate the analysis.