Detecting apparatus, detection method, and program

• US 10,379,001 B2
• Filed: 09/17/2015
• Issued: 08/13/2019
• Est. Priority Date: 09/17/2014
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Takahashi, Toshihiro
• Primary Examiner(s): Breene, John E
• Assistant Examiner(s): Liao, Christine Y
• Application Number:
  US14/857,053
• Publication Number:
  US 20160076970A1
• Time in Patent Office: 1,426 Days
• Field of Search: 702 33
• CPC Class Codes: G01H 1/003 of rotating machines G01H1/10 takes precedence
• CPC Class Codes: G01M 13/02 Gearings; Transmission mechanisms
• CPC Class Codes: G01M 13/045 Acoustic or vibration analysis
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/00 Computing arrangements using knowledge-based models
• CPC Class Codes: G06N 5/04 Inference or reasoning models
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks

Abstract

Provided is a detecting apparatus for detecting one or more anomalies of an operating body. The detecting apparatus includes a processor communicatively coupled to a dividing unit, a model learning unit, a difference calculating unit, and an anomaly detecting unit. The dividing unit divides a data sequence corresponding to an operation of the operating body into a plurality of divided data sequences in a time direction. The model learning unit learns each of the plurality of divided data sequences according to a regression model in the time direction, and calculates a model component modeling each of the divided data sequences. The difference calculating unit calculates a difference component indicating a difference between each of the plurality of divided data sequences and their corresponding model component. The anomaly detecting unit detects one or more anomalies of the operating body based on the calculated difference components.

Claims

1. A system for detecting defects in a physical operating body, the system comprising:

a set of sensors configured to measure physical motion of the operating body to generate one or more data sequences that correspond to the physical motion of the operating body;

a memory; and

a processor communicatively coupled to the memory and the set of sensors, wherein the processor is configured to perform a method comprising;

receiving the one or more data sequences from the set of sensors;

temporally dividing the one or more data sequences into a plurality of divided data sequences, the plurality of divided data sequences being arranged according to an operating period of the operating body;

obtaining a plurality of model components that correspond to a normal operation state of the operating body based on the operating period, each model component corresponding to a divided data sequence in the plurality of divided data sequences;

calculating a plurality of non-periodic components using the plurality of divided data sequences and the plurality of model components;

determining an operation state for the operating body based on at least the plurality of non-periodic components; and

transmitting a notification that the operating body has an anomaly to another system in response to determining that the operation state for the operating body is not the normal operation state.

2. The system of claim 1, wherein the operation state for the operating body is selected from a group consisting of the normal operation state, an initial damage state, and a damaged state.

3. The system of claim 1, wherein determining the operation state for the operating body includes detecting one or more anomalies of the operating body.

4. The system of claim 1, wherein the set of sensors includes at least one three-dimensional acceleration sensor and a rotational sensor, wherein the method further comprises determining the operating period for the operating body, and wherein the receiving the one or more data sequences includes:

receiving a first data sequence from each of the at least one three-dimensional acceleration sensor;

receiving a second data sequence from the rotational sensor.

5. The system of claim 1, wherein obtaining the plurality of model components that correspond to the normal operation state of the operating body comprises:

receiving a first set of training data, the first set of training data corresponding to operation of the operating body when the operating body is in the normal operation state;

receiving a second set of training data, the second set of training data corresponding to operation of the operating body when the operating body is in a damaged operation state; and

processing the first and second sets of training data to learn a regression model for the operating body.

6. The system of claim 1, wherein calculating the plurality of non-period components includes subtracting each of the plurality of model components from a corresponding divided data sequence, wherein the method performed by the processor further comprises:

calculate a feature quantity for each of the divided data sequences by performing statistical analysis on the plurality of non-periodic components, wherein the feature quantity is selected from a group consisting of a dispersion of the non-periodic components, a deviation of the non-periodic components, and an average of the non-periodic components,

wherein the determining the operation state for the operating body is based on the calculated feature quantities.

7. The system of claim 6, wherein determining the operation state for the operating body comprises:

ranking each of the feature quantities according to a magnitude of values of the feature quantities;

calculating a probability of anomaly for the operating body by performing logistic regression on the ranked feature quantities; and

comparing the probability of anomaly to a predetermined threshold.

8. The system of claim 1, wherein determining the operation state for the operating body comprises:

calculating a degree of anomaly for each divided data sequence according to the operation period of the operating body;

determining an average value of the plurality of degrees of anomaly; and

comparing the average value of the plurality of degrees of anomaly to a threshold.