Research On Disruption Identification And Classification Methods For Dynamic Scheduling Of Steelmaking

With the development of the society, the special steel company is facing more and more competitive. Energy price, raw materials prices and fuel price are rising quickly and bringing cost great pressures to the special steel industry. At the same time, the society demands the products more and more personalized, small quantity and variability. While special steel production is a complex, long process route and process-cross-over with each other, it brings huge impact with the change of external environments. The production of special steel is full of uncertainty. Material processing time, material element, equipment capacity state and material production process and so on has great changes. Combined with cross-over process, it brings big difficulty to the production of the special steel. In order to respond to the pressure from outer and inner system, it needs to monitor the real time information, find out the abnormal information, and respond to the changes of the production. It can improve the production efficiency, reduce production costs and enhance product competitiveness.For the need of treatment to the disruption, it establishes a monitoring model of real time information. It can detect the disruption at any time and respond to the uncertainty to improve system sensitivity.It establishes a detecting model with Entropy. It can determine whether the uncertainty is in the control range. According to the decision threshold setting and the system dynamic disrupt deviation value, it can determine the state of the system and provide the basis to the further decisions.According to the characteristics and causes of the disruption, it establishes the fault tree model of production disruption. Through fault tree model of production disruption, analyze the causes of production disrupt, reveal the source of disruption, and make it more clearly through the way of graphical analysis. For orders and other dominant disturbance, it deals with them directly with the monitoring of xBOM. For the hidden and other difficult to identify disrupts, such as quality, equipment capacity, it establishes a Rule-Based Reasoning disrupt classification model. For the disruption difficult to determine by the type of rule-based reasoning, it uses Evidence Theory approach to determine the type of disruption to achieve rapid identification.Through the rapid identification of disruption, it supports the next re-scheduling and enhances the system's responsiveness, increases system stability.