Modeling And Prediction Of Gravity Tide Signal Based On RBF Neural Network

With the continuous development of the instrument industry technology,human understanding of the earth is more abundant.Gravity solid tide is an important research method for studying geophysics.More and more scholars are investing in the study of gravity solid tide.The gravity tide signal contains a large amount of geophysical information,which can help humans to further understand the earth we live on.Gravity tidal signals are affected by multiple celestial bodies,mainly due to changes in the relative positions of celestial orbits such as the sun and the moon.They are also affected by changes in geologic,hydrological,atmospheric and other geographical conditions,so they are both regular and periodic.The changing signals also contain abnormal information reflecting changes in geography,hydrology,and the atmosphere.Since the gravity solid tide signal is a mixed set of multiple signal components,in order to further verify the applicability of the improved radial basis neural network in the field of gravity solid tide,this paper uses independent component analysis according to the mechanism of gravity solid tide generation(Independent).The Component Analysis(ICA)algorithm solves the harmonic component of the gravity tidal signal,and decomposes the gravity tidal signal into a long-period wave parallel to the axis of the earth's rotation around the centroid and a daily and half-day parallel to the plane of the Earth's equator.Wave three harmonic components.Through the orthogonal decomposition model of the gravity tide,the information components of the gravity tide signal in three directions are clearly and clearly reflected.By modeling the gravity tide signal,the basic law of the periodic variation of the gravity tide signal can be reflected and predicted.By comparing the theoretical calculations,the abnormal change information in the gravity tide signal can be further extracted.In the experiment,an improved Radial Basis Function(RBF)neural network was trained by gravity solid tide signal to obtain an effective RBF neural network model of gravity solid tide signal.The model is used to predict the estimated value of gravity tidal signal,and compared with the traditional RBF neural networkmodel and Augmented Reality(AR)model prediction results,the results of the radial basis network model prediction of the improved training algorithm are more accurate.It is shown that the improved training algorithm of this paper is effective in the radial basis network modeling of gravity solid tide signals.In this paper,a simple evolutionary group intelligent optimization algorithm is used to optimize the weight of RBF neural network and optimize the hybrid matrix for independent component analysis.This group intelligent optimization algorithm is a pure random search algorithm,which defines multiple role states of particles to achieve search diversity.Sex,the structure is not complicated,iterating multiple central role positions,can maintain the performance of the algorithm stably,effectively achieve global optimization,and the final solution is more ideal.