The BP Neural Network Based On MATLAB Application In Solar Flare

Solar flares are the most intense solar activity. The flare, a large number of high-energy particlesreach the earth, can affect the normal operation in the region of the earth’s orbit of the satellite, when theoperation had to temporarily escape, to avoid serious injury; at the same time astronauts aboard theinternational space station’s life is in danger. Flare X ray emitted and ultraviolet radiation reaching Earth’satmosphere, ionosphere, destruction of electromagnetic communication, so they can not be reflected radiowaves; Radio communication especially short wave communication, as well as the mobile phone, radio, etc.will be disturbed or even interrupted.Adopting efective method to study and predict the level of solar flares, to reduce and avoid tohave the very vital significance of solar flares to the people to bring the loss of property and life. As thesolar activity is complex and frequent outbreak, solar flares in is related to many factors, these factors andpredict solar flares level has a complex non-linear relationship. Artificial neural networks has strongnonlinear approximation ability to predict solar flares level, taking into account various factors can beintegrated in other areas, with with objectivity and effectiveness. Thus made use of BP neural networkalgorithm to predict solar flares level.Due to the need to achieve BP network model by means of computer programming language,more difficult to achieve, but powerful MATLAB software makes this problem is resolved. In this paper, bycalling MATLAB neural network toolbox to establish a BP neural network model to predict the level ofsolar flares, identified the link between factors and flare SPA VLF events occur in the ionosphere affected.And through to the Xinxiang city in1998January to June during theobservation records of Alpha very lowfrequency communication systemsignal navigation data anomalies occurred in the65groups of trainingsimulation, and carries on the forecast test.The experimental results show that the model is suitablefor flarelevel prediction, which has good application value.