Deep learning-based karst area underground water centralized drainage point pollution prediction method
The invention discloses a karst area underground water centralized drainage point pollution prediction method based on deep learning. The method comprises the following steps: S1, acquiring precipitation per hour, drainage point flow and pollutant concentration data of a target area within more than...
Gespeichert in:
Hauptverfasser: | , , , , , , |
---|---|
Format: | Patent |
Sprache: | chi ; eng |
Schlagworte: | |
Online-Zugang: | Volltext bestellen |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | The invention discloses a karst area underground water centralized drainage point pollution prediction method based on deep learning. The method comprises the following steps: S1, acquiring precipitation per hour, drainage point flow and pollutant concentration data of a target area within more than three months as sample data; s2, missing value checking and filling are carried out on the collected sample data, and then normalization processing is carried out on the data; s3, splitting the data into a training set and a test set; s4, establishing an LSTM neural network, and carrying out hyper-parameter adjustment by using Grid Search; s5, performing inverse normalization on the data, testing the data by using a test set, and evaluating the model by using RMSE and MAE functions; and S6, predicting the discharge point flow and pollutant concentration in a future time range based on the rainfall in the current time range by using the constructed LSTM model. The method can predict the pollution index of the under |
---|