A Stacked Auto-Encoder Based Fault Diagnosis Model for Chemical Process

摘要

Fault detection and diagnosis (FDD) is one of the key technologies to ensure the safe operation of chemical processes.With the widespread application of automation technology in chemical plants and the era of big data, data-based methods have become a hot research topic in the field of fault diagnosis.How to effectively extract the fault characteristics from the data and determine the cause of the fault is the key to help the operator deal with the abnormal conditions.Stack auto-encoder is a deep learning model with strong feature extraction and generalization capabilities.This paper proposes a SAE-based chemical process fault diagnosis model and applies it to Tennessee Eastman process.The performance of the SAE-based model is illustrated by comparison with the results of other methods.