FAILURE PREDICTION SYSTEM, FAILURE PREDICTION DEVICE, LEARNING DEVICE, FAILURE PREDICTION METHOD, AND PROGRAM
To predict failure of a device in a plant with high accuracy.SOLUTION: A training information acquisition unit 340 acquires training information based on sensor detection information being information obtained by detecting a state of a device by a sensor 100 and worker explanation information being...
Gespeichert in:
Hauptverfasser: | , , , , , |
---|---|
Format: | Patent |
Sprache: | eng ; jpn |
Schlagworte: | |
Online-Zugang: | Volltext bestellen |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | To predict failure of a device in a plant with high accuracy.SOLUTION: A training information acquisition unit 340 acquires training information based on sensor detection information being information obtained by detecting a state of a device by a sensor 100 and worker explanation information being information generated by a portable terminal 200 of a worker using the device and obtained by explaining the state of the device by the worker. A trained model generation unit 350 generates a trained model showing a relation between states of the device and failures by machine learning using preliminarily acquired training information. A trained model output unit 370 outputs the trained model. A trained model acquisition unit 510 acquires the trained model. An inference result information generation unit 560 accepts input of information for inference based on newly acquired sensor detection information and worker explanation information to the trained model to generate inference result information being information indicative of a result obtained by inferring a failure of the device. An inference result information output unit 570 outputs the inference result information.SELECTED DRAWING: Figure 2
【課題】工場の機器の故障を精度良く予測する。【解決手段】学習用情報取得部340は、センサ100が機器の状態を検知した情報であるセンサ検知情報と、機器を用いる作業者の携帯端末200が生成した機器の状態について作業者が説明した情報である作業者説明情報とに基づく学習用情報を取得し、学習済モデル生成部350は、予め取得した学習用情報を用いた機械学習によって機器の状態と故障との関係性を示す学習済モデルを生成し、学習済モデル出力部370は、学習済モデルを出力する。学習済モデル取得部510は、学習済モデルを取得し、推論結果情報生成部560は、新たに取得したセンサ検知情報と作業者説明情報とに基づく推論用情報を学習済モデルに入力して機器の故障を推論した結果を示す情報である推論結果情報を生成し、推論結果情報出力部570は、推論結果情報を出力する。【選択図】図2 |
---|