Inundation depth prediction device, and inundation depth prediction method
An inundation depth prediction device includes: a flow speed value acquiring unit that acquires a flow speed value on the sea surface; and an inundation depth predicting unit that predicts an inundation depth on the ground by inputting the flow speed value acquired by the flow speed value acquiring unit to a learned inundation depth prediction model used for predicting the inundation depth on the ground from the flow speed value on the sea surface.
1 . An inundation depth prediction device comprising:
flow speed value acquiring circuitry to acquire a flow speed value on a sea surface; and
inundation depth predicting circuitry to predict an inundation depth on a ground by inputting the flow speed value acquired by the flow speed value acquiring circuitry directly to a learned inundation depth prediction model used for predicting the inundation depth on the ground from the flow speed value on the sea surface,
wherein the inundation depth predicting circuitry predicts a primary prediction value of the inundation depth by inputting the flow speed value acquired by the flow speed value acquiring circuitry directly to the learned inundation depth prediction model, and calculates a secondary prediction value of the inundation depth on a basis of the predicted primary prediction value and a past prediction value of the inundation depth predicted in a past.
2 . The inundation depth prediction device according to claim 1 , further comprising data preprocessing circuitry to perform preprocessing which is at least one of standardization and complementation of missing data on the flow speed value acquired by the flow speed value acquiring circuitry, wherein
the inundation depth predicting circuitry predicts the inundation depth by inputting the flow speed value preprocessed by the data preprocessing circuitry to the learned inundation depth prediction model.
3 . An inundation depth prediction device comprising:
flow speed value acquiring circuitry to acquire a flow speed value on a sea surface; and
inundation depth predicting circuitry to predict an inundation depth on a ground by inputting the flow speed value acquired by the flow speed value acquiring circuitry to a learned inundation depth prediction model used for predicting the inundation depth on the ground from the flow speed value on the sea surface,
wherein the flow speed value acquired by the flow speed value acquiring circuitry is time-series data indicating a flow speed value for each time,
the learned inundation depth prediction model used by the inundation depth predicting circuitry is a convolutional neural network model, and
the inundation depth predicting circuitry determines whether or not the time-series data acquired by the flow speed value acquiring circuitry includes a required amount of data for predicting the inundation depth using the learned inundation depth prediction model, and in a case where the inundation depth predicting circuitry determines that the time-series data does not include the required amount of data, the inundation depth predicting circuitry performs complementation of a shortage amount of data on the time-series data acquired by the flow speed value acquiring circuitry.
4 . An inundation depth prediction device comprising:
flow speed value acquiring circuitry to acquire a flow speed value on a sea surface; and
inundation depth predicting circuitry to predict an inundation depth on a ground by inputting the flow speed value acquired by the flow speed value acquiring circuitry directly to a learned inundation depth prediction model used for predicting the inundation depth on the ground from the flow speed value on the sea surface,
wherein the inundation depth predicting circuitry predicts a probability distribution indicating an occurrence probability for each inundation depth by inputting the flow speed value acquired by the flow speed value acquiring circuitry directly to the learned inundation depth prediction model.
5 . An inundation depth prediction method comprising:
acquiring a flow speed value on a sea surface; and
predicting an inundation depth on a ground by inputting the acquired flow speed value directly to a learned inundation depth prediction model used for predicting the inundation depth on the ground from the flow speed value on the sea surface,
wherein a primary prediction value of the inundation depth is predicted by inputting the flow speed value acquired directly to the learned inundation depth prediction model, and a secondary prediction value of the inundation depth is calculated on a basis of the predicted primary prediction value and a past prediction value of the inundation depth predicted in a past.
6 . An inundation depth prediction method comprising:
acquiring a flow speed value on a sea surface; and
predicting an inundation depth on a ground by inputting the acquired flow speed value to a learned inundation depth prediction model used for predicting the inundation depth on the ground from the flow speed value on the sea surface,
wherein the flow speed value acquired is time-series data indicating a flow speed value for each time,
the learned inundation depth prediction model used is a convolutional neural network model, and
it is determined whether or not the time-series data acquired includes a required amount of data for predicting the inundation depth using the learned inundation depth prediction model, and in a case where it is determined that the time-series data does not include the required amount of data, complementation of a shortage amount of data is performed on the time-series data acquired.
7 . An inundation depth prediction method comprising:
acquiring a flow speed value on a sea surface; and
predicting an inundation depth on a ground by inputting the acquired flow speed value directly to a learned inundation depth prediction model used for predicting the inundation depth on the ground from the flow speed value on the sea surface,
wherein a probability distribution indicating an occurrence probability for each inundation depth is predicted by inputting the flow speed value acquired directly to the learned inundation depth prediction model.