Abstract: Predicting the remaining useful life (RUL) holds great significance in ensuring the operational safety of complex structures. To enhance the accuracy of RUL prediction for hydraulic gates, we propose a multi-feature information fusion-based approach. Initially, we employ the gamma process to simulate the corrosion evolution of gates and analyze the corrosion-caused degradation of characteristic parameters, including stress, natural vibration frequency, and dry/wet modal shapes through numerical simulations. Subsequently, we screen the feature parameters considering monotonicity and discreteness. We construct a health index by fusing these features based on principal component analysis. To model the gate degradation process, we employ a non-linear Wiener process and utilize the particle filtering method to obtain RUL prediction results for the gate at different operating times. Finally, we validate the reliability and effectiveness of our proposed method through engineering examples and finite element simulations. Our results demonstrate that the fusion of multiple information sources enables a more comprehensive reflection of the gate’s degradation state. The root mean square error (RMSE) of the prediction accuracy evaluation index is 1.395 5, the mean absolute error (MAE) is 1.262 8, and the absolute error of variance (VAE) is 0.352 8, showcasing high accuracy. This method can serve as a basis for gate health management and safety assessment.

Key words: hydraulic steel gate, information fusion, remaining useful life prediction, particle filter