Title :
Attractor analysis and its applications to seismic pattern recognition of nuclear explosion
Author :
Daizhi, Liu ; Star, Red ; Yinkang, Wei ; Ke, Zhao ; Weimin, Jia ; Juan, Su
Author_Institution :
Section 103, The Second Artillery Inst. of Eng., Xian, China
Abstract :
The phase space is first constructed by various features extracted from semi-signals in the time domain, then, the two-category source samples of underground nuclear explosions and natural earthquakes are investigated and analysed in the very phase space. The theoretical analysis and numerical calculation results indicate that there are two “attractors” in the phase space: one is the attractor for an underground nuclear explosion, another is for a natural earthquake. The dimensions of the attractors yielded in the light of the correlation dimension are of fractal dimensions, and less than that of the phase space. When the dimensions of phase space increased from low to high, the dimensions of the attractors become apparently saturated, and different attractors have different fractal dimensions. A new concept, namely, the “attraction center” is proposed. Moreover, the attractor, the attractor dimension, and the attraction center with their applications to seismic pattern recognition of an underground nuclear explosion are also explored
Keywords :
correlation methods; earthquakes; feature extraction; geophysical signal processing; nuclear explosions; pattern recognition; seismology; signal sampling; state-space methods; time-domain analysis; attraction center; attractor analysis; attractor dimension; correlation dimension; feature extraction; fractal dimensions; natural earthquakes; phase space; seismic pattern recognition; semisignals; source samples; time domain; underground nuclear explosion; underground nuclear explosions; Data preprocessing; Earthquakes; Explosions; Feature extraction; Fractals; Pattern analysis; Pattern recognition; Sampling methods; Signal processing; Time domain analysis;
Conference_Titel :
Signal Processing, 1996., 3rd International Conference on
Conference_Location :
Beijing
Print_ISBN :
0-7803-2912-0
DOI :
10.1109/ICSIGP.1996.566541