Sensor fault detection and diagnosis in flight control systems

In the modern industry especially aircraft industry, due to the need for better performance, modern control systems are becoming increasingly complex. A conventional feedback control system due to error or fault of actuators, sensors or other components may not provide the targeted performance or stability. Faults or errors can changed dynamic of the system, reduce overall system performance or even lead to unstable work of the system. Therefore detection and diagnosis of faults in the design of a control system seems necessary. For managing unusual conditions of a process, it is necessary to determine whether the condition is unusual, and then identify the flaws causing the abnormal condition. Nowadays, process units are very complex and include numerous measurements of process variables that are used for controlling and monitoring the process. So, for monitoring a process, a user is often faced with high dimensional data that can cause confusion, and increase the volume of computations. Hence, designing an intelligent system for detection and diagnosis of faults that can act in a timely manner, and adequate immunity and economic, it has attracted the attention of many researchers. In this thesis, a longitudinal model of the F4 Phantom fighter jet was selected as an example of a flight control systems. In the model, the sensor faults were detected by using principal component analysis and independent component analysis methods based on multivariate statistical techniques. The faults was diagnosed by using support vector machines method.