Fault Detection and Diagnosis Based on Neural Networks for Liquid Rocket propulsion System

摘要

Considering the health monitoring for liquid rocket engine with turbopump system, this paper tries to make use of state estimation techniques for nonlinear dynamic systems together with neural networks techniques to carry out fault detection and diagnosis. For fault detection of liquid rocket engine, in general, the method based on analytical models is employed. However, in order to improve the robustness and detection speed of conventional statistical testing methods on residuals or innovation sequences such as x~2-test, SPRT, etc., the paper proposes a new decision approach called intelligent decision algorithm, based on the characteristics of neural networks and SPRT criteria.