Unit health state prediction based on VMD-TCN

Abstract To address the problem that it is difficult to accurately predict health status of hydropower units using real-time monitoring data, this paper proposes a method for predicting trend based on variational modal decomposition (VMD) and time convolution network (TCN); considering nonlinear factors series unit vibration index, prediction model VMD-TCN established achieve accurate status. And multiple sets comparison experiments are designed verify higher accuracy faster proposed model.