Anti-false Alarm Method of Methanol Three-column Distillation Monitoring

The methanol three-column distillation process is often affected by noise and random interference, which causes the normal data to be misreported as fault data when monitoring the distillation process. In order to reduce the false positive rate and improve the process monitoring effect, this paper presents a novel method combining Butterworth Low-pass Filter (BLPF) and Principal component analysis (PCA). The method firstly uses BLPF to filter the original data, then establishes a PCA process monitoring model on the filtered data, and finally verifies the data of a methanol production plant in Zhejiang. The results show that this method has better monitoring effect than the traditional PCA method, and can effectively deal with the problem of high false positive rate caused by noise, and has practical application reference value.