Fault detection and isolation of aeroengine combustion chamber based on unscented Kalman filter method fusing artificial neural network

Continuously improving the operational efficiency of aeroengine is an important part to reduce the use of fossil energy and environmental pollution. This makes the combustion chamber hotter and more prone to fault, so fault detection and isolation is even more important. Therefore, a fault detection and isolation method for engine combustion chamber based on the unscented Kalman filter and artificial neural network is proposed. A twin-spool turbofan engine model for combustion chamber fault detection and isolation study is established for the first time. The concept of distributed flame tube health state parameters is proposed, and Kalman filter technique is used to achieve parameter estimation and monitor the health state. Further, in order to address the issue that Kalman filter requires the explicit state and measurement functions to be specified a priori, which are usually rough approximations of reality, this paper proposes to learn rich and dynamic representations of state and measurement models from data by using multilayer perceptron. Besides, it is compared with the method based on exhaust gas temperature dispersion. The experimental results show that the proposed method exhibits excellent performance for both abrupt and ramp faults of the flame tube, and is robust to interference from changes in working conditions.