Attractor analysis of feature phase space

In this paper, the origin of feature phase space is presented, the construction method based on various features extracted from Lorenz model system is stated, and the validity of attractor analysis of feature phase space is tested by numerical experiments. Experimental results indicate that attractor analysis approach of feature phase space is an efficient way in pattern recognition, it determines the infimum of uncorrelated features in number, it has great immunity to noise, and it is very suitable to analyze the practical samples with noise and disturb. The practical application in the recognition of underground nuclear explosions from natural earthquakes has proved the validity of attractor analysis of feature phase space.