A Novel Runoff Forecasting Model Based on the Decomposition-Integration-Prediction Framework

Abstract

Runoff forecasting is of great importance for flood mitigation and power generation plan preparation. To explore the better application of time-frequency decomposition technology in runoff forecasting and improve the prediction accuracy, this research has developed a framework of runoff forecasting named Decomposition-Integration-Prediction (DIP) using parallel-input neural network, and proposed a novel runoff forecasting model with Variational Mode Decomposition (VMD), Gated Recurrent Unit (GRU), and Stochastic Fractal Search (SFS) algorithm under this framework. In this model, the observed runoff series is first decomposed into several sub-series via the VMD method to extract different frequency information. Secondly, the parallel layers in the parallel-input neural network based on GRU are trained to receive the input samples of each subcomponent and integrate their output adaptively through the concatenation layers. Finally, the output of concatenation layers is treated as the final runoff forecasting result. In this process, the SFS algorithm was adopted to optimize the structure of the neural network. The prediction performance of the proposed model was evaluated using the historical monthly runoff data at Pingshan and Yichang hydrological stations in the Upper Yangtze River Basin of China, and seven various single and decomposition-based hybrid models were developed for comparison. The results show that the proposed model has obvious advantages in overall prediction performance, model training time, and multi-step-ahead prediction compared to several comparative methods, which is a reasonable and more efficient monthly runoff forecasting method based on time series decomposition and neural networks.

A novel ensemble deep reinforcement learning model for short‐term load forecasting based on Q‐learning dynamic model selection

“…The electricity load sequence data have non-stationary and non-linear characteristics, and the load data is preprocessed by VMD [29,30]. By iteratively searching for changing modes, the VMD decomposes the original load series into several mode components with finite bandwidth [31], as shown in Equation (1).…”

A Data-Driven Method and Hybrid Deep Learning Model for Flood Risk Prediction

“…Te results indicated that VMD decomposition could remove the interference noise in the data and improve the robustness of the model. In order to reduce noise, Xu and Yan et al [32,33] applied VMD to food data preprocessing. Experimental results showed that VMD decomposition could remove the interference noise in the water level data and improve the signal-to-noise ratio, further combining with the deep learning neural network algorithm to improve the food prediction accuracy [33].…”

Monthly Runoff Prediction for Xijiang River via Gated Recurrent Unit, Discrete Wavelet Transform, and Variational Modal Decomposition

“…In runoff prediction, the GRU model has shown superior performance compared to the SVM model, with further enhancement in prediction accuracy observed upon incorporating sequence processing technologies [15]. For instance, the RMSE of the DWT-GRU model decreased in comparison to the single GRU model [16]; the VMD-GRU model exhibited higher prediction accuracy than the standalone GRU model for monthly runoff prediction [17]; and the CEEMDAN-FE-VMD-SVM-GRU model outperformed eight other models for the daily runoff prediction [18]. Additionally, identifying the optimal combination of input variables, typically involving rainfall and runoff measurements [19], for a DWT-VMD-GRU runoff model is crucial.…”