Combined Forecasting Model of Precipitation Based on the CEEMD-ELM-FFOA Coupling Model

Precipitation prediction is an important technical mean for flood and drought disaster early warning, rational utilization, the development of water resources. Complementary ensemble empirical mode decomposition (CEEMD) can effectively reduce aliasing white noise interference; extreme learning machines (ELM) predict non-stationary data quickly easily; fruit fly optimization algorithm (FFOA) has better local ability. According to multi-scale characteristics precipitation time series, a new approach based on combination complementary (CEEMD), machine (ELM), proposed. The monthly measured in Zhengzhou City from 1951 2020 was taken as example conduct experiment compared with three models: ELM, EMD-HHT, CEEMD-ELM. research results show that sum annual predicted by CEEMD-ELM-FFOA model 577.33 mm, which higher than value 572.53 mm error 4.80 mm. average absolute 0.81 relative 1.39%. closely follow changing trend precipitation, shows effect other models be used regional prediction.