Ice Breakup Date Forecast with Hybrid Artificial Neural Networks

A hybrid artificial neural network model combining particle swarm optimization (PSO) and back propagation (BP) was used for ice breakup date forecast in the top reach of the Yellow River, China. A comparison of PSO-BP model to other statistical models was also conducted to evaluate the performance of the PSO-BP model. The forecast results indicate a satisfactory performance in the ice breakup date forecast with the PSO-BP model. The study concludes that the hybrid artificial neural network model combining PSO and BP has the high practicability and good accuracy for describing complex nonlinear ice breakup processes.