Fault Prediction for Valve Hall Fitting Based on Multivariate State Estimation
摘要
Valve hall fittings are important for a system’s security, operation, and maintenance. The fault estimation of valve hall fittings has attracted extensive research interest. Currently, most fault estimation schemes typically employ theoretical and numerical modeling methods. However, these methods often necessitate simplification and assumptions about the analysis object, making it challenging meeting engineering requirements. This paper introduces a novel Multivariate State Estimation Technique (MSET) method for valve hall fittings’ fault estimation. Specifically, based on CTGAN and a clustering-based memory matrix generation strategy, the paper proposes a data augmentation method in order to solve the shortcomings of the traditional MSET method. The experiment results in real conditions verify the effectiveness of the proposed method.