Abstract：In order to address the intrinsic challenge of attractor reconstruction from the time series of a chaotic system, such as strong nonlinearity and random like properties, an approximation scheme was propsed based on cloud transformation, which is the natural result of cloud theory. The method is considered nonlinear time series as the implementation of a number of alternative sequences in the sense of a certain probability. Furthermore, the feasibility of the method was verified via the chaotic series reconstruction of Lorenz chaotic system and Newton-Leipnik chaotic system with and without externally added noises.

Keyword：cloud model;cloud transform;model approximation;chaotic system;