基于遗传算法的重力固体潮信号独立元分析与地震前兆信息提取

Independent component analysis of gravity earth tide and extraction of seismic precursory signals based on genetic algorithm

在对重力固体潮信号的分析中,应用了一种将固体潮信号所含有的各谐波信息分解到3个正交分量上的三维正交分解模型;为了更好地提取各个分量谐波信号,克服传统独立分量分析(ICA)收敛速度慢、局部搜索等缺陷,使用了传统ICA与实数遗传算法(GA)相结合的方法,经验证此方法可行有效;实验中对实测信号与理论信号相结合的信号进行了谐波提取,并对提取出的谐波信号进行分析,捕捉地震前兆信息;分析发现地震前某不定时刻点,在长周期波中波形出现一定程度的奇异,充分反映了震前能量积蓄的变化.因其与地震发生时间之间的不确定性,所以还不能将其作为地震的预测方法.

A three—dimensional orthogonal decomposition model was used to decompose the harmonic information of the tidal signal into 3 orthogonal components. In order to extract the components of harmonic signaland overcome the defects of the traditional independent component analysis（ ICA）,such as slow convergence speed, local search and others defects, this paper uses the traditional ICA and real coded genetic algorithm（ GA） combined method,which has been proved is feasible and effective by the experiments. By the way this paper uses the combination of the measured signal and the theoretical signal to extract the harmonic information of the tidal signal, which is used to analyze the harmonic signal and capture the earthquake precursor information;There are some singular waveform in the long period waves before the earthquake pointindefinitely,which reflects the change of the earthquake energy before the point. Because of the uncertainty between the singular waveform point and when the earthquake would happen, so it can ’t be used as the prediction method of earthquake.