Identifying bias modules from reference populations for machine diagnostics

• Main Authors: YPMA ALEXANDER, GKOROU DIMITRI, KAREVEN ZAKARIY, DUNAEV DMITRY, VAN HEUDEN PIETER, RIJFJATER, JEROEN, VAN SCHUYBROECK JAN KLAAS
• Format: Patent
• Language: Chinese; English
• Published: 23.07.2024
• Subjects: APPARATUS SPECIALLY ADAPTED THEREFOR; CALCULATING; CINEMATOGRAPHY; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTROGRAPHY; HOLOGRAPHY; MATERIALS THEREFOR; ORIGINALS THEREFOR; PHOTOGRAPHY; PHOTOMECHANICAL PRODUCTION OF TEXTURED OR PATTERNED SURFACES,e.g. FOR PRINTING, FOR PROCESSING OF SEMICONDUCTORDEVICES; PHYSICS

Faults in a production plant of interest suspected of deviating machines are identified based on whether it is possible to train a machine learning model to distinguish first sensor data derived from the production plant of interest and second sensor data derived from one or more other production plants assuming normal performance. Thus, the discernible ability of the machine learning model serves as an indicator to discern a population of faulty machines from healthy machines. 基于是否可能训练机器学习模型以区分来源于所关注的生产设备的第一传感器数据和来源于假定为正常表现的一个或更多个其它生产设备的第二传感器数据而识别被怀疑为偏差机器的所关注的生产设备中的故障。因此，所述机器学习模型的可辨别能力用作辨别有故障的机器与健康的机器的群体的指示器。