Abnormality detection system, abnormality detection apparatus, and abnormality detection method

Patent metadata

Field	Value
Publication number	US-12013679-B2
Application number	US-202016850806-A
Country	US
Kind code	B2
Filing date	Apr 16, 2020
Priority date	Jun 14, 2019
Publication date	Jun 18, 2024
Grant date	Jun 18, 2024

Abstract

It is an object of the present invention to provide a technique capable of easily extracting a section signal of a specific sub-process. The anomaly detection system includes an extraction unit for extracting a specific subsequence to be an object of anomaly detection from among a plurality of subsequences from a composite sequence included in a monitor signal. The extraction unit determines an optimal warping path from the composite sequence and a reference sequence, which is an example of the composite sequence acquired in advance, by a dynamic time warping method. The extraction unit identifies a start point and an end point of a specific subsequence based on the optimal warping path and the start point and end point of the subsequence of the reference sequence. The extraction unit extracts a specific subsequence based on a start point and an end point of the specific subsequence.

First claim

What is claimed is:

1. An anomaly detection system for detecting an anomaly of a manufacturing device, the anomaly detection system comprising:
a central processing unit (CPU) configured to:
receive, from a sensor for measuring an operational parameter of the manufacturing device, the operational parameter as an analog monitor signal;
convert the analog monitor signal into a digital monitor signal comprising a composite sequence that is a series of sequences and includes a plurality of subsequences and indicating a status of the manufacturing device;
extract a specific subsequence of the manufacturing device from the plurality of subsequences of the composite sequence included in the digital monitor signal,
determine a first optimal warping path by a dynamic time warping method from the composite sequence and a reference sequence that is an example of a composite sequence acquired in advance;
identify start and end points of the specific subsequence based on the first optimal warping path, and start and end points of a subsequence of the reference sequence acquired in advance;
extract the specific subsequence based on the start and end points of the specific subsequence;
adjust a sampling interval of the composite sequence by a process of extracting or complementing an average value, a maximum value, a minimum value or a median value of adjacent data points, a number of which data points is specified by a down-sampling factor;
determine a second optimal warping path by a dynamic time warping method from the composite sequence of the adjusted sampling interval, and a reference sequence that is acquired in advance and is adjusted a sampling interval by a process of extracting or complementing an average value, a maximum value, a minimum value or a median value of adjacent data points, a number of which data points is specified by a down-sampling factor;
identify start and end areas of the specific subsequence based on the second optimal warping path, and start and end points of a subsequence of the reference sequence acquired in advance;
identify start and end points of the specific subsequence based on the second optimal warping path, and the start and end areas of the subsequence of the composite sequence;
identify start and end areas of the subsequence of the reference sequence based on the second optimal warping path, and the start and end areas of the subsequence of the composite sequence;
identify the start point of the specific subsequence based on a third optimal warping path for two sequences of the start area of the subsequence of the reference sequence and the composite sequence, and the start point of the subsequence of the reference sequence; and
identify the end point of the specific subsequence based on the third optimal warping path for two sequences of the end area of the subsequence of the reference sequence and the composite sequence, and the end point of the subsequence of the reference sequence; and
exclude down-sampling for a number of data points starting from a start point and an end point of the composite sequence, the number of data points being specified by a down-sampling exclusion length,
wherein the CPU, based on the extracted specific subsequence, outputs an output signal or indicates the status of the manufacturing device.

2. The anomaly detection system according to claim 1 , wherein a dynamic time warping method that determines any or a combination of the first to third optimal warping path modifies a step size constraint.

3. The anomaly detection system according to claim 1 , wherein the CPU is configured to extract a subsequence different from the specific subsequence among the plurality of sub-sequences based on the first optimal warping path, and start and end points of a subsequence different from the subsequence of the reference sequence acquired in advance.

4. The anomaly detection system according to claim 1 , wherein the CPU is further configured to extract the digital monitor signal comprising a time interval indicated by a trigger signal as the composite sequence having a finite width.

5. The anomaly detection system according to claim 4 , wherein the CPU is further configured output the digital monitor signal and the trigger signal as a time-series digital signal.

6. The anomaly detection system according to claim 1 , wherein the CPU is further configured to perform anomaly detection processing on the extracted specific subsequence based on an anomaly detection condition.

7. The anomaly detection system according to claim 4 ,
wherein the anomaly detection system further comprises a first memory configured to store a plurality of composite sequences extracted by the CPU,
wherein the CPU is further configured to control a display configured to display the plurality of composite sequences stored in the first memory,
wherein the CPU is further configured to set a subsequence extraction condition,
wherein the anomaly detection system further comprises a second memory configured to store a plurality of subsequences extracted by the set subsequence extraction condition, and
wherein the CPU is further configured to generate an anomaly detection condition based on the plurality of sub-sequences extracted.

8. An anomaly detection apparatus for detecting an anomaly of a manufacturing device, the anomaly detection apparatus comprising:
a central processing unit; and
a memory,
wherein the central processing unit is configured to:
receive, from a sensor for measuring an operational parameter of the manufacturing device, the operational parameter as an analog monitor signal;
convert the analog monitor signal into a digital monitor signal comprising a composite sequence that is a series of sequences and includes a plurality of subsequences and indicating a status of the manufacturing device;
determine a first optimal warping path by a dynamic time warping method from the composite sequence and a reference sequence that is an example of a composite sequence acquired in advance;
identify start and end points of a specific subsequence subject to anomaly detection among the plurality of subsequences based on the first optimal warping path, and start and end points of a subsequence of the reference sequence stored in the memory;
extract the specific subsequence based on the start and end points of the specific subsequence; and
perform anomaly detection processing on the extracted specific subsequence based on an anomaly detection condition,

wherein the central processing unit is configured to:
adjust a sampling interval of the composite sequence by a process of extracting or complementing an average value, a maximum value, a minimum value or a median value of adjacent data points, a number of which data points is specified by a down-sampling factor;
determine a second optimal warping path by a dynamic time warping method from the composite sequence of the adjusted sampling interval, and a reference sequence that is stored in the memory in advance and is adjusted a sampling interval by a process of extracting or complementing an average value, a maximum value, a minimum value or a median value of adjacent data points, a number of which data points is specified by a down-sampling factor;
identify start and end areas of the specific subsequence based on the second optimal warping path, and start and end points of a subsequence of the reference sequence stored the memory; and
identify start and end points of the specific subsequence based on the second optimal warping path, and the start and end areas of the subsequence of the composite sequence,

wherein the central processing unit is configured to:
identify start and end areas of the subsequence of the reference sequence based on the second optima

Assignees

• Renesas Electronics Corp

Inventors

• Ishikawa Shinji

Classifications

• G06F17/18

  for evaluating statistical data {, e.g. average values, frequency distributions, probability functions, regression analysis (forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04)} · CPC title

• G05B2219/34465

  Safety, control of correct operation, abnormal states · CPC title

• G05B2219/24065

  Real time diagnostics · CPC title

• G05B19/406 Primary

  characterised by monitoring or safety (G05B19/19 takes precedence) · CPC title

• G05B23/0243 Primary

  model based detection method, e.g. first-principles knowledge model · CPC title