Research on a Service Load Prediction Method Based on VMD-GLRT

In this paper, a deep learning-based prediction model VMD-GLRT is proposed to address the accuracy problem of service load prediction. The combines Variational Mode Decomposition (VMD) and GRU-LSTM. At same time, incorporates residual networks self-attentive mechanisms improve model. VMD part decomposes original time series into several intrinsic mode functions (IMFs) part. other uses GRU-LSTM structure with ResNets Self-Attention learn features IMF model-building process focuses on three main aspects: Firstly, mathematical constructed based data characteristics workload. used decompose input multiple components efficiency in extracting from data. Secondly, long short-term memory (LSTM) network unit incorporated network, allowing correct predictions more accurately performance Finally, self-focus mechanism model, better capture over distances. This improves dependence output vector these features. To validate experiences were conducted using open-source datasets. experimental results compared learning statistical models, it was found that paper achieved improvements mean absolute percentage error (MAPE).