Knowledge Based Prediction through Artificial Neural Networks and Evolutionary Strategy for Power Plant Applications

Artificial Intelligence (AI) as a field has grown and adapted to the changing times. Strong AI or the very fact that machines can exhibit human like intelligence still remains a controversial & distant dream. But the gap between strong AI and weak AI has been converging. A mathematical proof for this has been rendered by Schapire [2] who concludes that the notions of strong and weak learning are equivalent by training them through boosting algorithms. Noted experts in the field Kurt Gödel, Noam Chomsky and Roger Penrose [3] have said that strong AI is impossible. Well this is a matter of continuous debate with Ray Kurzweil, the futurist, pointing out that the Singularity is near. In fact present day machine learning algorithms can do a whole lot of things that were not possible in the past including occupations like nursing, teaching manifested through robotics. Genetics, Nanotechnology and Robotics are the mainstay for knowledge boosting algorithms in future. There is a large class of algorithms in AI based on computational intelligence or soft computing. Some are connectionist, simulating the human cognitive processes which can be largely grouped under the term Artificial Neural Networks (ANN), yet others are evolutionary like Evolutionary Strategy based Genetic Algorithms. This paper is focused on prediction of power of induced draft fan in a thermal power plant based on a variety of independent features amounting to 33 which was reduced to 10 by feature selection using Evolutionary Strategy.