Design an algorithm for damage detection of a liquid propellant engine based on neural network in order to classification and identification of quantity and place of damage

The aim of this paper is to design an Algorithm for damage detection of the open cycle liquid propellant engine which is based on artificial neural networks in combination with stochastic analysis. Damage is simulated as cavitation in pumps (oxidizer or fuel pump) and fouling in some path of engine. The key stone of the method is feed-forward multi-layer neural network with back propagation algorithm. This network uses output signals of unhealthy system to detect place and quantity of damage. It is impossible to obtain appropriate training set for real engine, so stochastic analysis using mathematical model is carried out and dynamic simulation is made to get training set virtually. Result of dynamic simulation of engine is validated with experimental result. In this plan, percentage of variation of output signals of engine such as output pressure of subsystem and revolution of turbine, considered as best input data for neural network. This data is obtained from output parameters of simulated unhealthy engine. Finally, this damage detection approach was carried out using laboratory hot test.