Abnormality detection apparatus, abnormality detection system, and learning apparatus, and methods for the same and non-temporary computer-readable medium storing the same

Patent metadata

Field	Value
Publication number	US-11989013-B2
Application number	US-201917421521-A
Country	US
Kind code	B2
Filing date	Jan 18, 2019
Priority date	Jan 18, 2019
Publication date	May 21, 2024
Grant date	May 21, 2024

Abstract

An abnormality detection apparatus ( 200 ) includes storage means ( 210 ) for storing a learned self-encoder ( 211 ) including predetermined number of two or more of elements as input layers, extraction means ( 220 ) for extracting a target data group of a predetermined period including data pieces from time series data measured by one or more sensors, the number of the data pieces being the predetermined number, conversion means ( 230 ) for converting the target data group into multi-dimensional vector data including the predetermined number of elements, identifying means ( 240 ) for identifying a time period in which there may be a cause of an abnormality from the predetermined period based on a difference between output vector data having the predetermined number of elements obtained by inputting the multi-dimensional vector data to the self-encoder ( 211 ) and the multi-dimensional vector data, and output means ( 250 ) for outputting abnormality detection information including the identified time period.

First claim

What is claimed is:

1. An abnormality detection apparatus comprising:
at least one memory configured to store instructions and a learned self-encoder, the learned self-encoder including a predetermined number of two or more elements as input layers, and
at least one processor configured to execute the instructions to:
extract, from time series data measured by one or more sensors during a first period of time, a target data group including a number of data pieces equal to the predetermined number of the elements of the learned self-encoder, the target data group being for a predetermined period of time which is at least part of the first period;
convert the target data group into multi-dimensional vector data having a number of elements equal to the predetermined number of elements of the learned self-encoder;
input the multi-dimensional vector data to the self-encoder to obtain output vector data having a number of elements equal to the predetermined number of elements of the learned self-encoder;
identify a time period in which there may be an abnormality within the predetermined period based on a difference between the output vector data and the multi-dimensional vector data; and
output abnormality detection information including the identified time period.

2. The abnormality detection apparatus according to claim 1 , wherein
the predetermined period is a time period in which a number of combinations of the one or more sensors and a measured time measured by the one or more sensors becomes the predetermined number.

3. The abnormality detection apparatus according to claim 1 , wherein the at least one processor is further configured to execute the instructions to
convert the target data group into the multi-dimensional vector data including each data of a combination of the one or more sensors and a measured time measured by the sensor as a corresponding element.

4. The abnormality detection apparatus according to claim 1 , wherein the at least one processor is further configured to execute the instructions to
compare the output vector data with the multi-dimensional vector data for each corresponding element and calculate the difference, and
identify a time period including a measured time when the difference exceeds a predetermined threshold.

5. The abnormality detection apparatus according to claim 1 , wherein
the time series data includes data measured at a plurality of measured times measured by each of two or more sensors, and
wherein the at least one processor is further configured to execute the instructions to
further identify the sensor that may be a cause of the abnormality from among the two or more sensors based on the difference, and
include the identified time period with the identified sensor in association with each other in the abnormality detection information and output the abnormality detection information.

6. The abnormality detection apparatus according to claim 5 , wherein
the self-encoder comprises:
an encoder including a plurality of sub-input layers corresponding to the two or more sensors, respectively; and
a decoder coupled to the encoder and includes a plurality of sub-output layers corresponding to the plurality of sub-input layers, respectively,

each of the plurality of sub-input layers includes each data corresponding to the plurality of measured times measured by the corresponding sensor as a corresponding element, and
each of the plurality of sub-output layers includes a same number of the elements as that of the elements of the sub-input layer.

7. The abnormality detection apparatus according to claim 1 , wherein
the self-encoder includes an encoder which is a fully-connected neural network in which elements of the input layer are connected in a round robin fashion.

8. The abnormality detection apparatus according to claim 1 , wherein the at least one processor is further configured to execute the instructions to
extract a second target data group of the predetermined period from second time series data measured by the one or more sensors when a measuring target is normal,
convert the second target data group into second multi-dimensional vector data, and
input the second multi-dimensional vector data to the input layers, learn a parameter of the self-encoder, and store the self-encoder as the learned self-encoder in the memory.

9. An abnormality detection method performed by a computer and comprising:
extracting, from time series data measured by one or more sensors during a first period of time, a target data group including a number of data pieces equal to a predetermined number of elements of a learned self-encoder, the target data group being for a predetermined period of time which is at least part of the first period, the elements being input layers of the learned self-encoder;
converting the target data group into multi-dimensional vector data having a number of elements equal to the predetermined number of elements of the learned self-encoder;
inputting the multi-dimensional vector data to the self-encoder to obtain output vector data having a number of elements equal to the predetermined number of elements of the learned self-encoder;
identifying a time period in which there may be an abnormality within the predetermined period based on a difference between the output vector data and the multi-dimensional vector data; and
outputting abnormality detection information including the identified time period.

10. A non-transitory computer readable medium storing an abnormality detection program which causes a computer to execute:
extracting, from time series data measured by one or more sensors during a first period of time, a target data group including a number of data pieces equal to a predetermined number of elements of a learned self-encoder, the target data group being for a predetermined period of time which is at least part of the first period, the elements being input layers of the learned self-encoder;
converting the target data group into multi-dimensional vector data having a number of elements equal to the predetermined number of elements of the learned self-encoder;
inputting the multi-dimensional vector data to the self-encoder to obtain output vector data having a number of elements equal to the predetermined number of elements of the learned self-encoder;
identifying a time period in which there may be an abnormality within the predetermined period based on a difference between the output vector data and the multi-dimensional vector data; and
outputting abnormality detection information including the identified time period.

Assignees

• Nec Corp

Inventors

• Yoshida Kosuke

Classifications

• G06N3/0499

  Feedforward networks · CPC title

• G06N3/0455

  Auto-encoder networks; Encoder-decoder networks · CPC title

• G05B23/0221 Primary

  Preprocessing measurements, e.g. data collection rate adjustment; Standardization of measurements; Time series or signal analysis, e.g. frequency analysis or wavelets; Trustworthiness of measurements; Indexes therefor; Measurements using easily measured parameters to estimate parameters difficult to measure; Virtual sensor creation; De-noising; Sensor fusion; Unconventional preprocessing inherently present in specific fault detection methods like PCA-based methods · CPC title

• G06F18/2132

  based on discrimination criteria, e.g. discriminant analysis · CPC title

• G06N3/04

  Architecture, e.g. interconnection topology · CPC title