A variational Bayesian robust linear dynamic system approach for dynamic process modelling and fault detection

In this work, a variational Bayesian robust linear dynamic system (VBRLDS) approach is proposed for dynamic process modeling and monitoring. Traditional linear dynamic system (LDS) constructed with Kalman filter is designed by Gaussian assumption which can be easily violated in outlier contaminated modeling situations. To deal with this issue, the conventional Gaussian-based Kalman filter is modified with heavy tailed Student's t-distribution so as to deal with modeling outliers. Then, a variational Bayesian expectation maximization (VBEM) algorithm is developed for learning parameters of the robust linear dynamic system. For process monitoring, traditional residual space modeling method of SPE statistics are modified. To explore the feasibility and effectiveness, our proposed method is applied into fault detection and is comparatively studied with several other methods on the Tennessee Eastman benchmark.