Anomaly detection in multidimensional sensor data
AssigneeFord Global Technologies, LLC
InventorsGaurav Pandey, Brian George Buss, Dimitar Petrov Filev
Life of this patent
Estimating the term…
Abstract
A computer includes a processor and a memory storing instructions executable by the processor to receive a time series of vectors from a sensor, determine a weighted moving mean of the vectors, determine an inverse covariance matrix of the vectors, receive a current vector from the sensor, determine a squared Mahalanobis distance between the current vector and the weighted moving mean, and output an indicator of an anomaly with the sensor in response to the squared Mahalanobis distance exceeding a threshold. The squared Mahalanobis distance is determined by using the inverse covariance matrix.
Claims20
A computer comprising a processor and a memory storing instructions executable by the processor to:
- receive a time series of vectors from a sensor;
- determine a weighted moving mean of the vectors;
- determine an inverse covariance matrix of the vectors;
- receive a current vector from the sensor;
- determine a squared Mahalanobis distance between the current vector and the weighted moving mean, wherein the squared Mahalanobis distance is determined by using the inverse covariance matrix;
- and
- output an indicator of an anomaly with the sensor in response to the squared Mahalanobis distance exceeding a threshold.
A method comprising:
- receiving a time series of vectors from a sensor;
- determining a weighted moving mean of the vectors;
- determining an inverse covariance matrix of the vectors;
- receiving a current vector from the sensor;
- determining a squared Mahalanobis distance between the current vector and the weighted moving mean, wherein the squared Mahalanobis distance is determined by using the inverse covariance matrix;
- and
- outputting an indicator of an anomaly with the sensor in response to the squared Mahalanobis distance exceeding a threshold.
Dependencies are inferred from the claim text.