Abnormality determining apparatus, abnormality determining method, and abnormality determining system

An abnormality determining method for a motor driven mechanism, comprising:

• acquiring time-series observing data with respect to an input to and an output from a motor which drives the motor driven mechanism in an observation state, the time-series observing data comprising a plurality of observing data points acquired in a time series manner;
• determining whether each of the observing data points of the time-series observing data acquired have data abnormality;
• and
• determining, based on at least one of a frequency of acquisition of the observing data points that are determined to have the data abnormality, an acquisition time of the observing data points that are determined to have the data abnormality, an acquisition frequency of the observing data points that are determined to have the data abnormality, and continuously acquiring period of the observing data points that are determined to have the data abnormality, whether mechanical abnormality in the motor driven mechanism occurs,
• wherein determining whether each of the observing data points of the time-series observing data acquired have data abnormality comprises:
• calculating a data abnormality determining threshold;
• calculating a sample average based on time-series sample data acquired by driving the motor driven mechanism in a normal state of the motor driven mechanism;
• calculating a sample covariance matrix based on the time-series sample data;
• calculating a Mahalanobis distance based on the sample average, the sample covariance matrix, and the time-series observing data acquired by driving and observing the motor driven mechanism;
• and
• determining the data abnormality of the observing data point based on comparison between the data abnormality determining threshold and the Mahalanobis distance.

An abnormality determining system for a motor driven mechanism, comprising:

• a first calculator to
• determine a false information rate,
• calculate a chi-square distribution,
• calculate a data abnormality determining threshold based on the false information rate and the chi-square distribution,
• acquire time-series sample data with respect to an input to and an output from a motor which drives the motor driven mechanism while the motor driven mechanism is in a normal state, and
• calculate a sample average and a sample covariance matrix based on the time-series sample data;
• and
• a second calculator to
• acquire time-series observing data with respect to an input to and an output from the motor by driving and observing the motor driven mechanism,
• calculate a Mahalanobis distance based on the sample average, the sample covariance matrix, and the time-series observing data,
• detect data abnormality in the time-series observing data by comparing the data abnormality determining threshold with the Mahalanobis distance, and
• determine, based on the data abnormality, whether mechanical abnormality in the motor driven mechanism occurs.

A motor control apparatus for a motor which drives a motor driven mechanism, comprising:

• circuitry configured to:
• acquire time-series observing data with respect to an input to and an output from the motor which drives the motor driven mechanism in an observation state, the time-series observing data comprising a plurality of observing data points acquired in a time series manner;
• determine whether each of the observing data points of the time-series observing data acquired have data abnormality;
• and
• determine, based on at least one of a frequency of acquisition of the observing data points that are determined to have the data abnormality, an acquisition time of the observing data points that are determined to have the data abnormality, an acquisition frequency of the observing data points that are determined to have the data abnormality, and continuously acquiring period of the observing data points that are determined to have the data abnormality, whether mechanical abnormality in the motor driven mechanism occurs,
• wherein the circuitry is configured to:
• calculate a data abnormality determining threshold;
• calculate a sample average based on time-series sample data acquired by driving the motor driven mechanism in a normal state of the motor driven mechanism;
• calculate a sample covariance matrix based on the time-series sample data;
• calculate a Mahalanobis distance based on the sample average, the sample covariance matrix, and the time-series observing data acquired by driving and observing the motor driven mechanism;
• and
• determine the data abnormality of the observing data point based on comparison between the data abnormality determining threshold and the Mahalanobis distance.