公开(公告)号：US20210165708A1

公开(公告)日：2021-06-03

申请号：US16700316

申请日：2019-12-02

Applicant: Accenture Inc.

Inventor： Badrinarayanan Damal Vijayaraghavan ,  Atul Goel ,  Harshith Vuppala ,  Rani Gopalakrishnan

IPC: G06F11/07 ,  G06N20/20 ,  G06N3/08 ,  G06N5/00 ,  G06N5/04 ,  G06N3/04

Abstract: Systems, methods, and computer-readable storage media configured to predict future system failures are disclosed. Performance metrics (e.g., key performance indicators (KPIs)) of a system may be monitored and machine learning techniques may utilize a trained model to evaluate the performance metrics and identify trends in the performance metrics indicative of future failures of the monitored system. The predicted future failures may be identified based on combinations of different performance metrics and the impact that the performance metric trends of the group of different performance metrics will have on the system in the future. Upon predicting that a system failure will occur, operations to mitigate the failure may be initiated. The disclosed embodiments may improve overall performance of monitored systems by: increasing system uptimes (i.e., availability); helping systems administrators maintain the monitored systems in a healthy state; and ensuring the functionality those systems provide is readily available to system users.

公开(公告)号：US11921573B2

公开(公告)日：2024-03-05

申请号：US16700316

申请日：2019-12-02

Applicant: Accenture Inc.

Inventor： Badrinarayanan Damal Vijayaraghavan ,  Atul Goel ,  Harshith Vuppala ,  Rani Gopalakrishnan

IPC: G06F11/00 ,  G06F11/07 ,  G06N3/04 ,  G06N3/08 ,  G06N5/01 ,  G06N5/04 ,  G06N20/20

CPC classification number: G06F11/0793 ,  G06F11/0703 ,  G06F11/0709 ,  G06F11/0751 ,  G06F11/0754 ,  G06N3/04 ,  G06N3/08 ,  G06N5/01 ,  G06N5/04 ,  G06N20/20 ,  G06F2201/81

Abstract: Systems, methods, and computer-readable storage media configured to predict future system failures are disclosed. Performance metrics (e.g., key performance indicators (KPIs)) of a system may be monitored and machine learning techniques may utilize a trained model to evaluate the performance metrics and identify trends in the performance metrics indicative of future failures of the monitored system. The predicted future failures may be identified based on combinations of different performance metrics and the impact that the performance metric trends of the group of different performance metrics will have on the system in the future. Upon predicting that a system failure will occur, operations to mitigate the failure may be initiated. The disclosed embodiments may improve overall performance of monitored systems by: increasing system uptimes (i.e., availability); helping systems administrators maintain the monitored systems in a healthy state; and ensuring the functionality those systems provide is readily available to system users.