Simulation of power plant transients with artificial neural networks: application to an existing combined cycle

To maintain the high performance of gas-turbine-based combined cycles, tran sients must be properly taken into account in the design phase and efficien tly monitored in the operational phase, because they are not negligible tim e intervals. The use of artificial intelligence techniques such as expert s ystems, fuzzy sets and neural networks (NNs), coupled with advanced measure ment and monitoring devices, can provide a reliable and efficient monitorin g system. An existing two-pressure-level combined cycle has been simulated by dividin g its simplified model into blocks representative of the main elements. An NN is associated with each of these blocks. Once the training and testing o f the NN are complete, using data from a simulator, the blocks are put eith er in a cascade arrangement or in a parallel arrangement, providing reliabl e systems that can predict the load-change transient behaviour of the entir e plant. The parallel approach was then tested on data from the real plant. The excessive simplification introduced with the simulator required the ad dition of selected real cases to the training set that are able to fit the NN response to reality. The results obtained are encouraging for use in an on-line monitoring system which evaluates the difference between the measur ed data and the predicted data.