Abnormality detection apparatus, computer-readable storage medium, and abnormality detection method

A non-transitory computer-readable storage medium having stored thereon a program that, when executed by an apparatus comprising one or more processors, causes the apparatus to function as an abnormality detection apparatus for detecting an abnormality in a movable apparatus including an operating part and capable of performing at least one task of machining or assembling performed on one or more works using the operating part, the abnormality detection apparatus comprising:

• a communication unit configured to at least receive operation-related data from the movable apparatus or from a sensor, the operation-related data resulting from an operation of the operating part of the movable apparatus;
• a target data generation unit configured to generate, based on the operation-related data, a plurality of target data, each of the plurality of target data being target data that includes a plurality of values, the operation-related data including data in a time series resulting from the operation of the operating part when the at least one task is performed on the one or more works, each of the plurality of target data corresponding to one of a plurality of portions of the operation-related data, each of the plurality of portions of the operation-related data containing data resulting from the same operation of the operating part performed at a different corresponding time, the plurality of portions of the operation-related data being temporally separated from each other such that each of the plurality of portions of the operation-related data is separated from a succeeding one of the plurality of portions of the operation-related data by a corresponding time interval, and the corresponding time interval being greater than a time interval between pairs of consecutive data in the time series of each of the plurality of portions of the operation-related data;
• a linked data generation unit configured to generate linked data that includes data obtained by linking the plurality of target data in a series, wherein the plurality of target data are linked in the series such that each of the plurality of target data in the series, except for a first one of the plurality of target data in the series, immediately follows another of the plurality of target data in the series in sequential order, to provide at least a portion of the linked data, wherein each of the plurality of target data has a wave form, a first end, and a second end, and wherein each of the plurality of target data in the series, except for the first one of the plurality of target data in the series, is appended to a preceding one of the plurality of target data in the series in end-to-end fashion such that the first end of each of the plurality of target data in the series, except for the first one of the plurality of target data in the series, is appended to the second end of the preceding one of the plurality of target data in the series;
• a detection processing execution unit configured to execute change detection processing on the linked data;
• an abnormality detection unit;
• a reference data storage unit configured to store in advance reference data, the linked data generation unit being configured to generate the linked data such that the linked data includes data obtained by linking the plurality of target data in the series and the reference data;
• and
• a reference data generation unit configured to generate the reference data based on a plurality of portions of the operation-related data corresponding to the reference data that are earlier in time compared to the plurality of portions of the operation-related data corresponding to the plurality of target data, and to cause the reference data to be stored in the reference data storage unit, wherein
• each of the plurality of portions of the operation-related data corresponding to the reference data and each of the plurality of portions of the operation-related data corresponding to the plurality of target data comprises a corresponding wave form having a plurality of values, the plurality of portions of the operation-related data corresponding to the reference data are temporally separated from each other such that each of the plurality of portions of the operation-related data corresponding to the reference data is separated from a succeeding one of the plurality of portions of the operation-related data corresponding to the reference data by a corresponding time interval,
• the linked data generation unit is configured to generate the linked data by including the reference data and linking the reference data in series with the plurality of target data that are linked in series in the linked data that is generated, the plurality of the portions of the operation-related data corresponding to the plurality of target data are later than the plurality of portions of the operation-related data corresponding to the reference data,
• the detection processing execution unit is configured to execute change detection processing on the linked data to generate a change detection result waveform, the change detection processing is performed by applying a moving window of change detection processing to a portion the linked data corresponding to the reference data and a portion of the linked data corresponding to at least recent data of the plurality of the target data, and by comparing their results,
• the abnormality detection unit is configured to determine a degree of abnormality by using a calculated numerical value as the degree of abnormality, the calculated numerical value is selected from the group consisting of a maximum value of the change detection result waveform, an average value of the change detection result waveform, and a standard deviation of the change detection result waveform;
• and
• the abnormality detection apparatus is configured to, based on the change detection result waveform, provide instructions to continue the performing of the at least one task of machining or assembling, or provide instructions not to continue the performing of the at least one task of machining or assembling.

An abnormality detection method that is executed by an apparatus comprising one or more processors for detecting an abnormality in a movable apparatus including an operating part and capable of performing at least one task of machining or assembling performed on one or more works using the operating part, the method comprising:

• receiving operation-related data from the movable apparatus or from a sensor, the operation-related data resulting from an operation of the operating part of the movable apparatus
• generating, based on the operation-related data, a plurality of target data, each of the plurality of target data being target data that includes a plurality of values, the operation-related data including data in a time series resulting from the operation of the operating part when the at least one task is performed on the one or more works, each of the plurality of target data corresponding to one of a plurality of portions of the operation-related data, each of the plurality of portions of the operation-related data containing data resulting from the same operation of the operating part performed at a different corresponding time, the plurality of portions of the operation-related data being temporally separated from each other such that each of the plurality of portions of the operation-related data is separated from a succeeding one of the plurality of portions of the operation-related data by a corresponding time interval, and the corresponding time interval being greater than a time interval between pairs of consecutive data in the time series of each of the plurality of portions of the operation-related data;
• generating linked data that includes data obtained by linking the plurality of target data in a series, wherein the plurality of target data are linked in the series such that each of the plurality of target data in the series, except for a first one of the plurality of target data in the series, immediately follows another of the plurality of target data in the series in sequential order, to provide at least a portion of the linked data,
• wherein the plurality of target data are linked in the series such that each of the plurality of target data in the series, except for a first one of the plurality of target data in the series, immediately follows another of the plurality of target data in the series in sequential order, to provide at least a portion of the linked data, wherein each of the plurality of target data has a wave form, a first end, and a second end, and wherein each of the plurality of target data in the series, except for the first one of the plurality of target data in the series, is appended to a preceding one of the plurality of target data in the series in end-to-end fashion such that the first end of each of the plurality of target data in the series, except for the first one of the plurality of target data in the series, is appended to the second end of the preceding one of the plurality of target data in the series;
• executing a change detection processing on the linked data;
• generating reference data based on a plurality of portions of the operation-related data corresponding to the reference data that are earlier in time compared to the plurality of portions of the operation-related data corresponding to the plurality of target data;
• storing the reference data,
• wherein each of the plurality of portions of the operation-related data corresponding to the reference data and each of the plurality of portions of the operation-related data corresponding to the plurality of target data comprises a corresponding wave form having a plurality of values, the plurality of portions of the operation-related data corresponding to the reference data are temporally separated from each other such that each of the plurality of portions of the operation-related data corresponding to the reference data is separated from a succeeding one of the plurality of portions of the operation-related data corresponding to the reference data by a corresponding time interval,
• generating linked data comprises generating the linked data such that the linked data includes data obtained by linking the plurality of target data in the series and the reference data,
• generating linked data further comprises generating the linked data by including the reference data and linking the reference data in series with the plurality of target data that are linked in series in the linked data that is generated, wherein the plurality of the portions of the operation-related data corresponding to the plurality of target data are later than the plurality of portions of the operation-related data corresponding to the reference data,
• wherein the change detection processing is performed by applying a moving window of change detection processing to a portion the linked data corresponding to the reference data and a portion of the linked data corresponding to at least recent data of the plurality of the target data, and by comparing their results;
• determining a degree of abnormality by using a calculated numerical value as the degree of abnormality, wherein the calculated numerical value is selected from the group consisting of a maximum value of the change detection result waveform, an average value of the change detection result waveform, and a standard deviation of the change detection result waveform;
• and
• based on the change detection result waveform, providing instructions to continue the performing of the at least one task of machining or assembling, or providing instructions not to continue the performing of the at least one task of machining or assembling.