Aeroengine Condition Monitoring Based on Process Neural Networks and Aerothermo-parameters

Feedforward process neural networks are adopted to predict an aeroengine exhaust temperature, and the choice of input and output parameters of networks is discussed. Based on the orthogonal basis functions, the aggregation operation of the network is simplified. A transformation method from feedforward process neural networks to feedforward neural networks is proposed. Based on the prior knowledge of the feedforward neural networks, a learning algorithm is given. The simulation on the learning algorithm shows the results are satisfying.