The steel-making process requires a multi-stage production system that processes multiple items. Efficient production during the steel-making process is difficult to achieve because of the vast number of items, complex constraints and various changes in production requirements. This paper proposes a new method for maintaining high production efficiency by using prediction information to prevent large-scale production systems from experiencing process overload and full stockyards. First, the production system is modeled into a dynamic optimization problem capable of expressing complex steel-making process lot production constraints with some simplification. After that, the problem is solved using a new two-step algorithm. The first step decides appropriate lot size dynamically based on a Lagrangian decomposition coordination method, and the next step makes an optimal schedule that satisfies the constraints. We verify the performance of the proposed method by applying it to actual production data.
Railway traffic controllers are required to decide train operation restrictions. They instruct train drivers properly and promptly in order to ensure safety from natural hazards such as heavy rain, strong wind and big earthquakes. When an operation restriction is issued, controllers must make various decisions in a timely manner in view of the priority of each goal including minimizing the impact on train operations, in constantly changing situations. However, the competency to perform this task has not yet been realized.
The authors propose a method for analyzing controller competencies when train operation is restricted by extracting specific actions taken by skilled controllers to overcome difficult situations. Applying this method to controller instruction tasks, 209 competencies were obtained. One such competency is being able to give precise instructions to train drivers in scenes where situational awareness is difficult. For example, a situation where operation restrictions are issued from multiple locations, but then only one restriction is rescinded. Additionally, many competencies require the non-technical skills of situational awareness. In addition, many "competencies learned through personal experience", which had been regarded as the tacit knowledge of expert controllers, were revealed.