Solar flares are well-known events on the solar disk while most of them being non-geoeffective. The key factors that makes them geo-effective are weather they have CME accompanied and the features of CME as well. But among the hundreds of CME, only few of them could cause significant geomagnetic disturbances, which mainly depends on whether they are towards to the earth. In this work, the relationship between the geomagnetic disturbances and the energetic proton flux （ACE-EPAM） data, together with the parameter of the solar flares that related to the CME was carefully investigated. The preliminary result is that, more than 90% of the enhancement of the particle flux followed by shock could be measured by ACE. But the correlation between flux of the particles and magnitudes of the geomagnetic disturbances was not much clear. Other factors that related to the characteristics of the CME have also to be taken into consideration. The position of the flare, which may affect the direction of the CME, the flare scale, which may decide the velocity, and the duration, which could relate to the magnetic field strength. But through statistical work, it was found that the relationship between the magnitudes of the geomagnetic disturbance Ap index and all those parameters mentioned above was non-liner, so neural network method was introduced to simulate the relation automatically. After the neural work being trained with the historical data range from 1986 to 2002, a model to predict the geomagnetic storms after the solar eruptive events was constructed. Test result showed that the error of the model comparing with the measurement was less than 20%.