Statistical Model for Power Plant Performance Monitoring and Analysis

A novel approach for monitoring and analysing power plant operation and performance is presented utilizing statistical modelling technology, specifically linear partial least squares (PLS) and non-linear radial basis function (RBF-PLS) models. For the RBF neural network, a genetic algorithm (GA) is employed to optimise the model parameters. The potential of these models for signal and error prediction, and performance analysis is demonstrated utilizing data from a combined cycle gas turbine (CCGT).