Improving the reliability of aero-engine control system via virtual sensor-assisted fault-tolerant control

Gas path sensors serve as critical information sources for the aero-engine control system, and the reliability of their measurements directly impacts engine safety. Constraints including installation space and maintenance cost limit further fault tolerance improvement through increasing hardware redundancy. To address this issue, this paper proposes a virtual sensor-assisted fault-tolerant control (VSFTC) method. Specifically, gas path virtual sensors are constructed using random vector functional link networks with correlation-based input selection and online sequential training, forming a triple-redundancy sensing architecture. A sensor fusion algorithm is developed to assess the confidence levels of physical sensors and generate high-confidence data for engine controller computations. Bias and drift faults are injected into the gas path sensors, and comparative simulations are performed to validate the proposed method. Verification results indicate that the VSFTC method exhibits robust fault tolerance against physical sensor failures, effectively mitigating high-pressure compressor surge risks and ensuring engine operational safety. Reliability analysis incorporating sensor diagnostic coverage demonstrates that the virtual sensor-assisted control system achieves a longer mean time between failures compared to the conventional dual hardware redundancy system.