COLLABORATIVE LEARNING SYSTEM AND MONITORING SYSTEM

To overcome a bottleneck of a handling process, and upgrade the efficiency of the entire process.SOLUTION: A collaborative learning system of an embodiment can be adapted to process monitoring to be performed by defining a monitoring model for plural processes that are time-sequentially continuous with a predetermined transition time interval between adjoining processes. The system time-sequentially stores first monitoring data of a first process, second monitoring data of a second process upstream or downstream of the first process, and a result of monitoring of the first process which a first monitoring model outputs using the first monitoring data as an input parameter. The system performs master model learning processing for the first monitoring model using the first monitoring data and the result of monitoring outputted by the first monitoring model, and slave model learning processing for a second monitoring model using a result of monitoring, which is outputted by the first monitoring model at a first time, as teacher data, and using second monitoring data, which is obtained at a second time different from the first time by the transition time, as an input parameter.SELECTED DRAWING: Figure 2 【課題】処理プロセスのボトルネックを改善し、プロセス全体の効率アップを図る。【解決手段】実施形態の協調型学習システムは、所定の遷移時間間隔で時系列に連続する複数の工程別に監視モデルが設けられたプロセス監視に利用することができる。本システムは、第１工程の第１監視データと、第１工程に対する上流又は下流の第２工程の第２監視データと、第１監視データを入力パラメータとして第１監視モデルが出力する第１工程の監視結果と、を時系列に記憶する。そして、第１監視データと第１監視モデルの監視結果とを用いて、第１監視モデルに対する親モデル学習処理と、第１時刻の第１監視モデルの監視結果を教師データとして用い、第１時刻に対して遷移時間分異なる第２時刻における第２監視データを入力パラメータとした第２監視モデルに対する子モデル学習処理と、を行う。【選択図】図２