State diagnosis apparatus and system
According to one embodiment, a state diagnosis apparatus includes a processing circuit. The processing circuit executes a model receiving, as an input, first data relating to a state of a device at each of a plurality of stages along a time series and outputting a first numerical value quantitatively indicating the first data for each of the stages. The processing circuit determines whether or not first numerical values output from the model monotonously change along the time series.
1. A state diagnosis apparatus comprising a processing circuit configured to:
execute a machine learning model receiving, as an input, first data relating to a state of a device at each of a plurality of successive points in time along a time series and outputting a first numerical value quantitatively indicating the first data for each of the plurality of successive points in time; and
determine whether or not first numerical values output from the machine learning model monotonously change along the time series.
2. The apparatus according to claim 1 , wherein
the processing circuit transmits at least one of the model and the first numerical values output from the model, when the first numerical values output from the model monotonously change along the time series.
3. The apparatus according to claim 1 , wherein
the processing circuit trains the model using second data relating to the state of the device as training data.
4. The apparatus according to claim 3 , wherein
the processing circuit changes at least one of the training data and a parameter of the model and trains the model, when the first numerical values output from the model do not monotonously change along the time series.
5. A state diagnosis apparatus comprising
a processing circuit configured to:
execute a machine learning model receiving, as an input, first data relating to a state of a device at each of a plurality of successive points in time along a time series and outputting a plurality of first numerical values quantitatively indicating the first data for each of the plurality of successive points in time;
calculate a variance and an average of the first numerical values output from the machine learning model for each of the plurality of successive points in time; and
determine whether or not calculated variances have a predetermined value or less along the time series and whether or not calculated averages monotonously change along the time series.
6. The apparatus according to claim 5 , wherein
the processing circuit transmits at least one of the model and the first numerical values output from the model, when the calculated variances have the predetermined value or less along the time series and that the calculated averages monotonously change along the time series.
7. The apparatus according to claim 6 , wherein
the processing circuit trains the model using second data relating to the state of the device as training data.
8. The apparatus according to claim 7 , wherein
the processing circuit changes at least one of the training data and a parameter of the model and trains the model, in at least one of the case where the calculated variances do not have the predetermined value or less along the time series and the case where the calculated averages do not monotonously change along the time series.
9. The apparatus according to claim 1 , wherein
the processing circuit generates the first data on the basis of third data relating to the state of the device at a predetermined point in time along the time series and fourth data indicating a difference from the third data.
10. The apparatus according to claim 1 , wherein
the first data is signal data relating to sound of the device at each of the plurality of successive points in time along the time series.
11. The apparatus according to claim 1 , wherein
the first data is real data or pseudo data.
12. A state diagnosis system comprising a state diagnosis apparatus, a first device, and a second device,
the state diagnosis apparatus including
a processing circuit configured to:
execute a machine learning model receiving, as an input, first data relating to a state of a device at each of a plurality of successive points in time along a time series and outputting a first numerical value quantitatively indicating the first data for each of the plurality of successive points in time;
determine whether or not first numerical values output from the machine learning model monotonously change along the time series; and
transmit the machine learning model to the second device when the first numerical values output from the machine learning model monotonously change along the time series,
the second device receiving, as an input, second data and executing the machine learning model transmitted from the state diagnosis apparatus, the second data acquired from the first device, different from the first data, and relating to the state of the first device at each of the plurality of successive points in time along the time series.
13. The system according to claim 12 , further comprising
a third device displaying a plurality of second numerical values output from the model in the second device and quantitatively indicating the second data.