Method, device and computer program product for sensor data analysis

Patent metadata

Field	Value
Publication number	US-11494282-B2
Application number	US-201916536599-A
Country	US
Kind code	B2
Filing date	Aug 9, 2019
Priority date	Apr 25, 2019
Publication date	Nov 8, 2022
Grant date	Nov 8, 2022

Abstract

Methods, devices and computer program products for data analysis are provided. For example, a method comprises: in response to receiving target data from a target sensor at a first time, determining one or more reference sensors based on location information of a neighbor sensor adjacent to the target sensor and a second time of receiving the latest data from the neighbor sensor; determining reference estimation data of the one or more reference sensors at the first time based on historical sensor data obtained from the one or more reference sensors; determining target estimation data of the target sensor at the first time based on the reference estimation data; and detecting abnormity of the target data based on the target data and the target estimation data. In this way, abnormity of the sensor data may be detected efficiently and accurately.

First claim

What is claimed is:

1. A method for sensor data analysis, comprising:
in response to receiving target data from a target sensor of a plurality of sensors at a first time, determining one or more reference sensors of the plurality of sensors based on location information of a neighbor sensor of the plurality of sensors adjacent to the target sensor of the plurality of sensors and receiving latest data from the neighbor sensor of the plurality of sensors at a second time;
determining reference estimation data of the one or more reference sensors of the plurality of sensors at the first time based on historical sensor data obtained from the one or more reference sensors of the plurality of sensors;
determining target estimation data of the target sensor of the plurality of sensors at the first time based on the reference estimation data of the one or more reference sensors of the plurality of sensors;
detecting an abnormality of the target data of the target sensor of the plurality of sensors based on the target data of the target sensor of the plurality of sensors and the target estimation data of the target sensor of the plurality of sensors; and
causing one or more actions to be initiated in one or more sensors of the plurality of sensors based on the detected abnormality of the target data of the target sensor of the plurality of sensors;
wherein the determining the one or more reference sensors of the plurality of sensors, the determining the reference estimation data of the one or more reference sensors of the plurality of sensors, the determining the target estimation data of the target sensor of the plurality of sensors, the detecting, and the causing steps are executed by at least one computing device coupled to the plurality of sensors.

2. The method according to claim 1 , wherein determining the one or more reference sensors of the plurality of sensors comprises:
determining a distance from the neighbor sensor of the plurality of sensors to the target sensor of the plurality of sensors;
determining a time difference between the second time and the first time; and
determining the one or more reference sensors of the plurality of sensors based on a weighted sum of the distance and the time difference.

3. The method according to claim 1 , wherein determining the one or more reference sensors of the plurality of sensors comprises:
determining a plurality of groups of neighbor sensors located in different directions of the target sensor of the plurality of sensors; and
selecting at most a predetermined number of reference sensors from each group of neighbor sensors of the plurality of groups of neighbor sensors.

4. The method according to claim 1 , wherein determining the reference estimation data of the one or more reference sensors of the plurality of sensors comprises:
using a Kalman filter to determine the reference estimation data of the one or more reference sensors of the plurality of sensors based on the historical sensor data.

5. The method according to claim 1 , wherein determining the reference estimation data of the one or more reference sensors of the plurality of sensors comprises:
inputting at least one of the historical sensor data and the second time into a prediction model to determine the reference estimation data of the one or more reference sensors of the plurality of sensors, wherein the prediction model is trained based on the historical sensor data and a reception time associated with the historical sensor data.

6. The method according to claim 1 , wherein determining the target estimation data of the target sensor of the plurality of sensors comprises:
determining a first weight associated with a first sensor based on a first distance from a first reference sensor of the one or more reference sensors of the plurality of sensors to the target sensor of the plurality of sensors;
determining a second weight associated with a second sensor based on a second distance from a second reference sensor of the one or more reference sensors of the plurality of sensors to the target sensor of the plurality of sensors; and
determining the target estimation data of the target sensor of the plurality of sensors based on a weighted sum of first reference estimation data of the first reference sensor of the one or more reference sensors of the plurality of sensors and second reference estimation data of the second reference sensor of the one or more reference sensors of the plurality of sensors.

7. The method according to claim 1 , wherein detecting the abnormality of the target data of the target sensor of the plurality of sensors comprises:
in response to a difference between the target data of the target sensor of the plurality of sensors and the target estimation data of the target sensor of the plurality of sensors exceeding a predetermined difference threshold, determining that the target data of the target sensor of the plurality of sensors is abnormal.

8. The method according to claim 1 , wherein detecting the abnormality of the target data of the target sensor of the plurality of sensors comprises:
processing the target data of the target sensor of the plurality of sensors and the target estimation data of the target sensor of the plurality of sensors by using a machine learning model to detect the abnormality of the target data of the target sensor of the plurality of sensors, wherein the machine learning model is trained based on the historical sensor data, sensor estimation data, and whether the historical sensor data is abnormal.

9. The method according to claim 1 , further comprising:
in response to detecting that the target data of the target sensor of the plurality of sensors is abnormal, providing an indication of the abnormality of the target data of the target sensor of the plurality of sensors.

10. A device for sensor data analysis, comprising:
at least one processing unit; and
at least one memory being coupled to the at least one processing unit and storing instructions for execution by the at least one processing unit, the instructions, when executed by the at least one processing unit, cause the device, to perform acts comprising:
in response to receiving target data from a target sensor of a plurality of sensors at a first time, determining one or more reference sensors of the plurality of sensors based on location information of a neighbor sensor of the plurality of sensors adjacent to the target sensor of the plurality of sensors and receiving latest data from the neighbor sensor of the plurality of sensors at a second time;
determining reference estimation data of the one or more reference sensors of the plurality of sensors at the first time based on historical sensor data obtained from the one or more reference sensors of the plurality of sensors;
determining target estimation data of the target sensor of the plurality of sensors at the first time based on the reference estimation data of the one or more reference sensors of the plurality of sensors;
detecting an abnormality of the target data of the target sensor of the plurality of sensors based on the target data of the target sensor of the plurality of sensors and the target estimation data of the target sensor of the plurality of sensors; and
causing one or more actions to be initiated in one or more sensors of the plurality of sensors based on the detected abnormality of the target data of the target sensor of the plurality of sensors;
wherein the determining the one or more reference sensors of the plurality of sensors, the determining the reference estimation data of the one or more reference sensors of the plurality of sensors, the determining the target estimation data of the target sensor of

Assignees

• Emc Ip Holding Co Llc

Inventors

• Lin Zhenzhen
• Wu Pengfei
• Chen Si

Classifications

• G06F11/3058 Primary

  Monitoring arrangements for monitoring environmental properties or parameters of the computing system or of the computing system component, e.g. monitoring of power, currents, temperature, humidity, position, vibrations (thermal management in cooling arrangements of a computing system G06F1/206) · CPC title

• G06N20/00 Primary

  Machine learning · CPC title

• G06F2218/00

  Aspects of pattern recognition specially adapted for signal processing · CPC title

• G06F18/2433

  Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection · CPC title

• G06F18/24147

  Distances to closest patterns, e.g. nearest neighbour classification · CPC title