A data processing method for chemical industry based on sensor data and process monitoring

In order to recognizing the feature of abnormality in advance and fix the possible dirty data simultaneously, this article proposed a method to construct a proper mode for a production process producing specific product, a method to estimate the stability of a linear process and a group of data cleaning rule. Both the data processing procedure and algorithm are given detailedly with some samples and mathematical derivation in this article.

I Introduction

In chemical industry, production safety is an important problem. Chemical production process may use many kinds of hazardous article, and the reaction conditions are hardly ambient temperature and pressure. If an accident happens, it will break out suddenly, spread rapidly and last for a long time. Therefore, we must try to avoid accident happens, otherwise, it will lead to serious consequences. For this purpose, we should find a way to recognize the feature of accident, and try to remove the fault before accident happens.