engleski

Predictive Model of the Heat Exchange Process

An approach to the control of a two-stage heat exchange process is presented. A warm liquid with low boling point is fed into destillation column. Strict maintenance of its value close to boling point is required. In the first stage, the liquid is heated nearly to desired temperature, and in the second stage the desired value is controlled within a permitted tolerance of +-1C. The two-stage heat exchange process is controlled by a PID controller. Input, output and control variables are registered. A trained artificial neural network is applied as integral process controller. Data collection and off-line processing have been performed with neural network and circular qualitative algebra as well. A more general process control model has been obtained with a predictive component. Results of the prediction model are presented.

heat exchange qualitative model; prediction; process control; neural network

nije evidentirano