Modelling and control of generating units using neural network techniques

The increasing complexity of modern power systems highlights the need for advanced modelling and control techniques. Linear methods may not offer the best solution to these problems, since most industrial systems are inherently nonlinear. This paper presents an initial investigation into neural network modelling of power generating plant as an alternative to established linear methods. Comparisons are made between identified linear and neural models of a nonlinear boiler-generator simulation, and the potential of using these techniques on a real plant is discussed