Awareness and its use in Plug and Play Process Control

Assume a existing control system needs a new sensor or actuator due to deterioration or unsatisfactory performance for other reasons. If such a new device can be implemented in a completely automated fashion it will be a very powerful tool. This is exactly the vision for the research project “Plug and Play Process Control” P
³
C. The first step is to get the device on the network. The next is to decide on if it is useful and if yes for what. Methods to answer these questions are discussed and developed in this paper. The results are adaptive methods to estimate a correlation based “awareness”. These methods are successfully applied to a simulation experiment where control performance is improved. An estimate of the stochastic part of a system given the deterministic part is a necessary part of the above methods. In this paper, a prediction error method (PEM) with analytical gradient is developed for this problem.