Water Level Data Preprocessing Method Based on Savitzky-Golay Filter

摘要

Using historical water level data to establish a model to estimate the future water level is a common means of water level prediction. In this type of method, the accuracy of water level prediction is closely related to the quality of historical water level data. However, due to measurement accuracy and system deviations, water level data collected by real-time water level observatories often have abnormal or erroneous data. Aiming at this situation, this paper studies the water level data preprocessing method based on Savitzky-Golay filter. This method removed the noise of historical water level data and smoothed the data, which could better establish the water level prediction model. This paper used the water level prediction model established by LSTM to verify the water level data preprocessing method. The results showed that this method can improve the accuracy of the prediction model.