Towards a knowledge-based control of a complex industrial process

Difficulty to ensure optimal settings of all adjustable parameters is the dominant problem for complex and fast industrial processes such as cold rolling. The modern cold rolling mill is controlled by a distributed control system consisting of many nodes of several types. Proper tuning of all single controllers is pre-requisite. However, the quality of the product still depends on operator´s skills and experience due to large amount of many possible working modes and adjustments of the machine. The paper describes the project to develop a decision support tool that will provide advice for operators to help them in keeping adjustable mill parameters close to optimal settings. The main idea of the project is to extract valuable information from a huge amount of process data. The information obtained are then used for the decision-support tool to help operators to achieve the highest possible quality of the product.