AbstractThe knowledge of the process state variables and coefficients is of substantial importance in many practical domains such as process control and fault detection and diagnosis. In practice, one encounters systems composed of both static and dynamic models owing to the natural difference in time response of various process variables. This paper presents an estimation algorithm for a system described by the coupled steady state and dynamic models via formulating the joint state and parameter estimation problem as a state estimation problem and using a filter for non‐linear systems. In the algorithm the error covariance matrix prior to the current measurements is updated using the filtered estimates rather than the predicted quantities, having a basal impact on the convergence behaviour. The other important aspect considered is the inclusion of the unmeasured disturbances in the parameter vector. This concept has an important implication for improving the on‐line detection and diagnosis of potential plant operational problems. A sodium‐cooled nuclear reactor is the application problem illustrating the effectiveness of the proposed s
展开▼
