Towards an all-terrain identification method for models of industrial processes

✍ Scribed by L Boillereaux; J.M Flaus

• Publisher: Elsevier Science
• Year: 1995
• Tongue: English
• Weight: 752 KB
• Volume: 3
• Category: Article
• ISSN: 0967-0661
• DOI: 10.1016/0967-0661(95)00078-9

✦ Synopsis

In this paper, a new strategy of identification, based on Predictive Control, is presented. This method has already been developed for two years (Shook et al., 1991(Shook et al., & 1992)). However, in this paper is shown how to identify a process when the output is disturbed by an offset plus a drift plus a white noise. This case is interesting to study because, most of the time, disturbances can be modelled both as deterministic and stochastic signals. The prediction calculation used for the minimisation criterion take drifts, offsets and white noise into account. Concerning the drift, the slope is estimated at the same time as the model parameters.