Low earth satellite(LEO) with near-circular orbit has a short-time cycle, strongly affected by perturbative forces , so it is extraordinary complex . However, the orbit elements in short time varies much simpler, as the obvious characteristics include two main parts, polynomial and periodic term. So we can establish mathematic models to fit and predict satellite orbit, avoiding from computing complicated perturbative models. The study is engaged in the real orbit data analyzing. Spectrum of satellite first no-singularity orbit elements is analyzed, and then mathematic models are made in which the parameters are calculated with the least squares. The results show that the analyzing of LEO orbit is right with the simple mathematic models, the right fitting results and the function of predicting in short time. The method can be used as a meaning of smoothing and forecasting in positioning results of GPS onboard.