厚板製造工期の確率モデルと製造標準工期算出技術の開発

• Published in: 鉄と鋼 Vol. 101; no. 11; pp. 574 - 583
• Main Authors: 塩谷, 政典, 森, 純一, 伊藤, 邦春, 水谷, 泰, 鳥飼, 健司
• Format: Journal Article
• Language: Japanese
• Published: 一般社団法人 日本鉄鋼協会 2015
• Subjects: decision tree; maximum likelihood estimation; plate mill; probability distribution; production time
• ISSN: 0021-1575; 1883-2954
• DOI: 10.2355/tetsutohagane.TETSU-2015-024

Accurate estimation of the standard production time in steel plate mills is crucial for making successful production plans. Due to the complicated and stochastic processes of the steel plate mills, it is challenging and time-consuming to build precise mechanistic models which enable the estimation of the precise production time. To overcome this limitation, we propose a method to estimate the accurate standard production time from the historical process data instead of mechanistic models. In our method, decision trees are employed to identify the process flow for each order. Then the probability distribution of the process time for each process is computed by means of the maximum likelihood estimation. These probability distributions are combined into one probability distribution of the total production time in accordance with the predicted process flow. Finally, the standard production time is defined as the corresponding time with cumulative density function of the probability distribution at the specified confidence level. Real world steel production process data have been used to examine the effectiveness of the proposed approach. The results demonstrate that the new standard production time can increase the rate of production completion no later than the deadlines as well as shorten the average production time one to three days.