Characteristics of Data Processing for Industrial Process Modeling

2 Characteristics of Data Processing for Industrial Process Modeling 

Industrial production of chemical and metallurgical processes accumulates a large amount of data every day. Useful information in these data sets can be extracted by data processing for twofold purposes (1) to provide the mathematical models for the optimal control of the industrial processes, in order to realize the energy saving, yield increasing, pollution control, and production cost reduction (2) to find the mathematical model of the fault in production for the fault diagnosis, in order to find the cause of the fault, or to find the operation condition to avoid the fault. 

Compared with the more accurate laboratory data, the data in industrial records have their special characteristics [22] (1) The data in industrial records usually have a higher noise/signal ratio. Even in a modem factory, it is still inevitable to meet many uncontrollable impacts affecting the production processes, such as the fluctuation of the composition of raw materials, the change of requirements of products, the fluctuation of environments of production processes, the impact of 

Since it is very dangerous to put a wrong mathematical model in practice for chemical industry, we have to be particularly cautious in 

This is an interesting example of the accordance between the empirical data processing and theoretical research of physical chemistry [30]. 

---