Development of an Advisory System Based on a Neural Network for the Operation of a Coal Fired Power Plant

This paper describes the application of neural networks to the operaton of a coal fired power plant. The performance of a pulverized coal fired power plant depends on the design and operation of the plant and the quality of the coals. Incorrect operation of a power plant may lead to lower efficiencies and higher emissions of toxic gaseous elements. During the operation of the boiler the performance has to be controlled and optimized by adjusting the various boiler parameters. This is a difficult task for the operators because of the large number of parameters and the mutual dependence of the performance parameters. The complexity is even more enhanced ub case the quality of the coals changes frequently. In order to support the operator in performing the above task, KEMA, in cooperation with EZH and Schelde is developing an advisory system. Core of the advisory system will be a neural network model of the boiler. The model predicts accurately the various performance variables for a wide range of operating conditions. The applicability of neural networks to power plants is herewith demonstrated.