Research on the control method of turbine outlet temperature limitation protection based on nonlinear state estimation

Abstract Developing multivariable control schedules for turboshaft engines addresses critical safety demands in aero-engine performance. This study introduces a turbine outlet temperature (TOT) limitation protection method using nonlinear state estimation. A turboshaft engine multivariable controller is designed, complemented by a Bidirectional Long Short-Term Memory (BLSTM) network-based onboard model. Validation shows less than 2 % relative error between the BLSTM model and the component-level engine model. To enhance precision, an Unscented Kalman State Estimator (UKSE) leverages residuals from power turbine speed data to estimate TOT (parameter T45), achieving a 1.4 % estimation error. The UKSE framework enables proactive temperature limitation control by dynamically adjusting protection thresholds. Experimental tests under two flight conditions confirm the method’s efficacy: 1) TOT exceedance magnitude decreases by 75 %, 2) overtemperature duration shortens by over 50 %, and 3) temperature violations are fully eliminated in specific scenarios. These results highlight the integration of BLSTM modeling and UKSE-based estimation as a robust strategy for mitigating turbine thermal risks while maintaining operational stability. The approach bridges model-based control and real-time adaptive protection, offering a scalable solution for next-generation engine safety systems.