基于 CEEMDAN-LSTM 的陶岔渠首水深预测.

The water depth was predicted using a fully empirical mode decomposition-long short-term memory neural network model (CEEMDAN-LSTM) based on adaptive noise. First, the data is preprocessed by the median average filtering method, and then the historical water depth sequence is decomposed by the CEEMDAN method to obtain the high, low frequency and residual sequence of the historical water depth, and then the obtained components are predicted by LSTM neural network, and finally stacked The prediction value of each component reconstructs the water depth prediction result. Taking Taocha Canal Head as the research object, the test results of CEEMDAN-LSTM model show that the model has stronger performance than support vector machine regression, BP neural network, long short-term memory neural network, empirical mode decomposition-long short-term memory neural network model. prediction performance. [ABSTRACT FROM AUTHOR]