SYSTEM, AND DETERMINATION METHOD OF PROCESSING CONDITION
To embody reinforcement learning for optimizing processing flows of a semiconductor manufacturing process in a situation when the status is unobtainable in a direct manner.SOLUTION: A system for determining processing flows including a plurality of processings for controlling a subject, includes a l...
Saved in:
Main Authors | , , , |
---|---|
Format | Patent |
Language | English Japanese |
Published |
12.12.2019
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Summary: | To embody reinforcement learning for optimizing processing flows of a semiconductor manufacturing process in a situation when the status is unobtainable in a direct manner.SOLUTION: A system for determining processing flows including a plurality of processings for controlling a subject, includes a learning section which executes learning processing for determining a processing condition for each of the plurality of processings. The learning section obtains, from a device controlling the subject based on the processing flows, under a prescribed processing condition, a physical quantity correlating with a status of the subject to which the processing is executed, calculates a pseudo status corresponding to the status of the subject based on the physical quantity, and determines the processing condition for each of the plurality of processings for embodying a desired status of the subject by executing the learning processing using a value function.SELECTED DRAWING: Figure 1
【課題】状態を直接取得できない状況において、半導体製造プロセスの処理フローを最適化するための強化学習を実現する。【解決手段】対象を制御するための複数の処理を含む処理フローを決定するシステムであって、複数の処理の各々の処理条件を決定するための学習処理を実行する学習部を有し、学習部は、処理フローに基づいて対象を制御する装置から、所定の処理条件の下で、処理が実行された対象の状態と相関がある物理量を取得し、物理量に基づいて、対象の状態に対応する擬状態を算出し、価値関数を用いた学習処理を実行することによって、目的とする対象の状態を実現するための複数の処理の各々の処理条件を決定する。【選択図】図1 |
---|---|
Bibliography: | Application Number: JP20180107419 |