公开(公告)号：US10725463B1

公开(公告)日：2020-07-28

申请号：US16583678

申请日：2019-09-26

Applicant: Northrop Grumman Systems Corporation

Inventor： Sunil Dixit

IPC: G05B23/02 ,  B64D45/00 ,  G06F17/15 ,  G06N3/04

Abstract: A method implemented by a computing system identifies anomalies that represents potential off-normal behavior of components on an aircraft based on data from sensors during various modes of operations including in-flight operation of the aircraft. High-frequency sensor outputs monitor performance parameters of respective components. Anomaly criteria is dynamically selected and the high-frequency sensor outputs are compared with the anomaly criteria where the comparison results determine whether potentially off-normal behavior exists. A conditional anomaly tag is inserted in the digitized representations of first high-frequency sensor outputs where the corresponding comparison results indicate potentially off-normal behavior. At least the digitized representations of the first high-frequency sensor outputs containing the conditional anomaly tag are sent to a computer-based diagnostic system for a final determination of whether the conditional anomaly tag associated with the respective components represents off-normal behavior for the respective components.

公开(公告)号：US10089214B1

公开(公告)日：2018-10-02

申请号：US15268870

申请日：2016-09-19

Applicant: Northrop Grumman Systems Corporation

Inventor： Sunil Dixit

IPC: G06F11/00 ,  G06F11/36

Abstract: An exemplary method provides for automatically curing a detected behavior anomaly in executing target software during the continuing execution of the target software. Ranges of parameters of acceptable behaviors are stored. One behavior is detected that is outside the range of parameters for acceptable behavior for the corresponding behavior. A probability of success is predicted for restoring the corresponding behavior to acceptable behavior. First bytes associated with the behavior anomaly are replaced with the other bytes upon the predicted probability of success exceeding a predetermined success threshold, thereby automatically implementing a likely cure of a detected anomaly in the target software during the continuous execution of the target software.