Any complex system must be controlled by some basic laws, including macroscopic level submicroscopic level and microscopic level laws. How to discover its necessity-laws from these contingency phenomena (observed data) is the most important task of data mining (DM) and KDD, and it is the goal of this paper too. Based on the evolutionary computation and natural fractal, a multi-scale dynamic prediction system is proposed, which models the macro-behavior of the system by ordinary differential equations while models the micro-behavior of the system by natural fractals. The financial data such as the stock market data of Jun抋n stock price and the scientific observed data such as rainfall data of Wuhan in flood season are used as the test data for simulated test of analysis and prediction. The experimental results show that this system fits the data very well, and the simulated prediction is good too, even for modeling the time series with large undulating.