Early Warning of Intermittent Failure Based on Hybrid Algorithm

In this paper, the early warning of intermittent failure in complex industrial system is studied. Firstly, a target label generation strategy is proposed to preprocess the data. Secondly, combining RF with sliding window, a RF intermittent fault early warning model based on sliding window was constructed. Then, the ANT colony algorithm was used to fuse ARMA with RF based on sliding window, and a intermittent fault warning model based on hybrid algorithm was constructed. Finally, an argon air separation system is taken as an example to simulate these models, and the effectiveness and feasibility of the proposed method are verified.