Kouichi Matsuda

Raw material ship operation scheduling system in steelworks

We adopt the simulated annealing method for dealing with raw material transportation ship operation scheduling. The scheduling algorithm should satisfy the proper stock for each raw material level by satisfying the restriction conditions of the transportation ship and the harbour equipment. The developed scheduling algorithm consists mainly of two parts: 1) the operation of the ship is simulated by decision variables, such as the order of the operation of ship and the type of raw material on-board, for calculating the amount of each raw material unloaded and the unloading time; and 2) the simulation calculates the change in the stock level of each raw material type based on the simulation result by which the first simulation is operated. The evaluation function, comprising the operating cost of the ship and the propriety of the stock level, can be calculated by using these two simulation results.

Scheduling System for Raw Material Ship Operation

Abstract The raw material ship operation scheduling should satisfy the proper stock level for each raw material by satisfying the restriction. The developed scheduling algorithm consists of two simulators. First of all, the operation of the ship is simulated by decision variables like the order of the operation of the ship, etc. The second simulation calculates the change in the stock level of each raw material. The evaluation function, comprising the operating cost of the ship and the propriety of the stock level, can be calculated and to minimize the evaluation function, the simulated annealing is introduced.

Application of Artificial Intelligence to Operation Control of Blast Furnace

Abstract This paper describes about the application of artificial intelligence to the operation control of blast furnace. This operation control system consists of three parts, that is, the forecasting system for the blast furnace heat level, the action guidance system for controlling the blast furnace heat level, and the learning system for forecasting. The forecasting system has two prediction models. One is the rule - based model useful for the short term heat level variation and the other is the statistical quantitative model for the long one. Applying the fuzzy inference theory to these models, a new forecasting system has been developed available for both the short and long term hea t level variation. The action guidance system is an expert system which calculates the operations conditions quantitatively according to the result of the forecasting system. The learning system estimates the actual heat level quantitatively using fuzzy inference . After comparing this result with the forecasted one, this system can calculate the accuracy of the forecasting system, and update the parameters of the system. Applying this system to the actual operation, the failure ratio of the forecasting system proved to be less than 5%, and the heat level control operation could be taken more accurately than before . As a result, the furnace heat level has been stabilized and furnace fuel ratio has been improved to a higher level. Furthermore, the changes in fuel ratio have resulted in decrease hot metal Si content.