Process Control System Identification

The main purpose of this project is to develop black-box models. That can
introduce another modeling tool in process control system identification field. These
models need to address process that is difficult or complex to model using the
mathematical or empirical modeling approaches. The core activity for this project is
to develop a model for chemical process (single loop pressure control) using blackbox
identification techniques. This approach overcomes the difficulties that
encountered in modeling processes that characterized with low level of a priori
knowledge (operators have no priori knowledge about the system). Process control
system identification is implemented through conducting an experiment on a gas /air
pressure pilot plant where a set of input-output data are collected. Intelligent blackbox
modeling techniques are implemented for building process model that can be
utilized for further process control applications. Feedforward Neural network and
fuzzy clustering method are used to obtain a model from the plant data. The author
developed and compared the performance of black-box modeling techniques against
the performance of empirical model. According to the comparison achieved. It is
recommended to use black box modeling techniques for processes where process
control engineers mayface the lackof knowledge about the system or the plant.