Degradation level prediction method and degradation level prediction system
A degradation level of a noise source is predicted by performing deep learning using frequency data indicating a vibration of the noise source during a time period and degradation data indicating degradation of the noise source during the predetermined period to determine a state estimation model. First vibration signals are measured at the noise point during a monitoring period. The first vibration signals are converted into a frequency domain, and the frequency domain converted first vibration signals are input into the state estimation model to estimate time series degradation data of degradation of the noise source during the monitoring period. A degradation prediction model is determined by deep learning using the time series degradation data. Degradation data indicating predicted degradation of the noise source during a prediction period are predicted via the degradation prediction model. A remaining useful life of the noise source may be calculated.
1 . A method comprising:
determining a state estimation model by performing deep learning using:
a plurality of frequency data items indicating a plurality of vibrations of a noise source during a predetermined period, and
a plurality of degradation data items indicating corresponding degrees of degradation of the noise source during the predetermined period;
measuring a plurality of first vibration signals, from the noise source, at a noise point during a monitoring period;
estimating, by the state estimation model, time series degradation data indicating degrees of degradation of the noise source during the monitoring period by inputting a plurality of first frequency data items, obtained by converting the plurality of first vibration signals into a frequency domain, into the state estimation model;
determining a degradation prediction model by performing second deep learning using the time series degradation data estimated by the state estimation model as a training input for the second deep learning;
predicting, based on the degradation prediction model, degradation data indicating degrees of degradation of the noise source at at least one target point in time within a prediction period later than the monitoring period; and
calculating, based on the time series degradation data and the degradation data at the target point in time, a remaining useful life of the noise source.
2 . The method of claim 1 , wherein
the determining the state estimation model comprises:
providing a frequency data item, indicating a vibration measured at the noise point at a first point in time, as an input to the state estimation model;
predicting a first degradation data item indicating a first degree of degradation of the noise source according to the frequency data item at the first point in time; and
modifying the state estimation model based on an error between a data item indicating the degree of degradation of the noise source at the first point in time and the first degradation data item.
3 . The method of claim 1 , wherein
the determining the degradation prediction model comprises:
providing the time series degradation data, as an input, to the degradation prediction model;
predicting, via the degradation prediction model, a first degradation data item indicating a second degree of degradation of the noise source at a second point in time later than a first period corresponding to a plurality of consecutive first points in time, according to data corresponding to the first period among the time series degradation data; and
modifying the degradation prediction model according to an error between data corresponding to the second point in time among the time series degradation data and the first degradation data item.
4 . The method of claim 3 , wherein the degradation prediction model comprises a long short term memory (LSTM) model.
5 . The method of claim 1 , wherein
the predicting the degradation data indicating the degrees of degradation of the noise source comprises generating time series prediction data by flattening the time series degradation data and data indicating the degrees of degradation of the noise source corresponding to the prediction period.
6 . The method of claim 5 , wherein
the calculating the remaining useful life of the noise source comprises:
generating a trend graph corresponding to the time series prediction data by performing curve fitting on the time series prediction data;
deriving a first degradation level of the noise source corresponding to one reference point in time, among the at least one target point in time, from the trend graph; and
determining, based on degradation level of the noise source at a current point in time and the first degradation level of the noise source, the remaining useful life of the noise source at the current point in time.
7 . The method of claim 1 , wherein the noise source comprises one or more of a motor reducer, a brake, or a tire.
8 . A degradation level prediction system comprising:
an input unit configured to acquire vibration signals from a noise point configured to measure noise from a noise source;
an estimation model learning unit configured to determine a state estimation model by performing deep learning using:
a plurality of frequency data items indicating a plurality of vibrations of the noise point during a predetermined period, and
a plurality of degradation data items indicating degrees of degradation of the noise source during the predetermined period; and
a controller configured to:
measure a plurality of first vibration signals at the noise point during a monitoring period,
estimate, by the state estimation model based on a plurality of first frequency data items input into the state estimation model, time series degradation data indicating degrees of degradation of the noise source during the monitoring period, wherein the plurality of first frequency data items are obtained by converting the plurality of first vibration signals into a frequency domain,
predict, via a degradation prediction model performing second deep learning using the time series degradation data estimated by the state estimation model as a training input for the second deep learning, degradation data indicating a degree of degradation of the noise source at at least one target point in time within a prediction period later than the monitoring period, and
calculate, based on the time series degradation data and the predicted degradation data at the at least one target point in time, a remaining useful life of the noise source.
9 . The degradation level prediction system of claim 8 , wherein
the estimation model learning unit is configured to:
provide a frequency data item, indicating a vibration measured at the noise point at a first point in time, as an input to the state estimation model,
predict a first degradation data item indicating the degree of degradation of the noise source according to the frequency data item at the first point in time, and
modify the state estimation model according to an error between a data item indicating a degree of degradation of the noise source at the first point in time and the first degradation data item.
10 . The degradation level prediction system of claim 8 , further comprising:
a prediction model learning unit configured to:
provide the time series degradation data as an input to the degradation prediction model,
predict, based on data corresponding to a first period among the time series degradation data, a second degradation data item indicating a degree of degradation of the noise source at a second point in time later than the first period, wherein the first period corresponds to a plurality of consecutive first points in time, and
modify the degradation prediction model according to an error between a portion of the time series degradation data corresponding to the second point in time and the second degradation data item.
11 . The degradation level prediction system of claim 10 , wherein the degradation prediction model comprises a long short term memory (LSTM) model.
12 . The degradation level prediction system of claim 8 , wherein
the degradation prediction model is configured to generate time series prediction data by flattening the time series degradation data and data indicating the degrees of degradation of the noise source corresponding to the prediction period.
13 . The degradation level prediction system of claim 12 , wherein
the controller is configured to:
generates a trend graph corresponding to the time series prediction data by performing curve fitting on the time series prediction data,
derive, based on the trend graph, a first degradation level of the noise source corresponding to one reference point in time among the at least one target point in time, and
calculate, based on degradation level of the noise source at a current point in time and the first degradation level of the noise source, the remaining useful life of the noise source at the current point in time.
14 . The degradation level prediction system of claim 8 , wherein the noise source comprises one or more of a motor reducer, a brake, or a tire.
15 . The degradation level prediction system of claim 8 , wherein the state estimation model is configured to estimate, based on vibration data associated with a vehicle component generating noise and a location of the vehicle component, a state of the vehicle component.
16 . The degradation level prediction system of claim 11 , wherein the LSTM model is configured to learn a temporal pattern of the time series degradation data.
17 . The method of claim 1 , wherein the state estimation model is configured to estimate, based on vibration data associated with a vehicle component generating noise and a location of the vehicle component, a state of the vehicle component.
18 . The method of claim 4 , wherein the LSTM model is configured to learn a temporal pattern of the time series degradation data.
19 . The method of claim 1 , wherein state estimation model is configured to perform a learning process to learn a correlation between vibration frequency data and degradation states, and
wherein the degradation prediction model is configured to perform a learning process to learn temporal patterns of the time-series degradation data estimated by the state estimation model.