The Forecast Of The Coronal Mass Ejection's Velocity At Interplanetary L1 Point

Coronal mass ejection(CME)is one of the main causes of disturbance in near-earth space.After ejecting from the solar and transmits into the interplanetary space,a CME is called an interplanetary coronal mass ejection(ICME).ICME is transmitting to the earth through interplanetary space,which makes the near-earth space suffer from severe disturbance in a short time,and has a huge impact on human activities such as aerospace,satellite navigation,long-distance power transmission and oil transmission network,and geological exploration.CMEs-ICMEs event has a significant impact on human production activities and is related to national security and development of economic and social.Therefore,it is urgent to predict the occurrence of space weather in advance so as to avoid the disastrous weather caused by space weather.In this paper,the ICME disturbed velocity at interplanetary L1 point is researched.Firstly,we use the multicollinearity(MC)technique to estimate the collinearity between coronal parameters.As the result,the coronal parameters with collinearity problem are judged.Then,the generalized additive model(GAM)is used to fit the coronal parameters and ICME velocities,and the correlation between the coronal parameters and ICME velocities is obtained.Through the analysis of MC technology and GAM nonlinear fitting,the input parameters of the model can be determined.Thus,we establishe the BPNN,GAM,MC+BPNN and MC+GAM models.In order to compare the prediction effects of different models,the Correlation Coefficient(CC)and Root-Mean-Square Error(RMSE)are selected as the comparison parameters.The results show that BPNN,MC+BPNN and GAM models don't achieve great forecasting results of ICME velocity,so MC technology never significantly improve the result of BPNN model.However,the CC of predicted value and observed value of MC+GAM is up to 0.71,far exceeding the empirical relationship proposed by Manohara et al.[2006].Meanwhile,its RMSE value of 83 km/s is also the smallest in these comparison models.The results show that MC technology has successfullyimproved the prediction effect of GAM model.In summary,the MC + GAM model is the best one among the four models in predicting the 25 CMEs-ICMEs events.