A Gaussian weighted sum clustering method of on-board adaptive hybrid model was proposed for health parameter estimation of turbo-shaft engine in the full flight envelope.The on-board adaptive hybrid model was composed of Kalman filter and neural network.Gaussian weighted sum model was used to cluster a huge amount of flight data in real-time and update the neural network weighting.The working range of adaptive hybrid model was increased automatically with this method.The simulation results show the validity of the proposed method.Adaptive hybrid model based on Gaussian weighted sum clustering increases the accuracy for turbo-shaft engine performance tracking in the full flight envelope.