A Comparative Study of Unscented Kalman Filtering and Particle Filtering for State Predictions of Industrial Process

In the paper, we present a comparative study of Unscented Kalman Filtering (UKF) and Particle Filtering (PF) for state forecast of complex industrial processes. The performances of UKF and PF in data assimilation and state estimation are demonstrated with an example case study of simplified steam generator water level control system in nuclear power plants. The water level in the steam generator is predicted and compared for validity check. The comparison results shows both methods exhibit comparable prediction accuracy when handling the high nonlinear system process under Gaussian distribution. The increased time cost of particle filtering is primarily attributed to the larger number of particles sampled during the process.