摘要
介绍了一种利用基因表达式编程算法进行多元非线性函数建模的方法,并应用于地震等级预测中,实验表明,GEP模型的预测精度远高于神经网络模型。最后,利用地震等级数据无后效性和平稳性的特征,通过马尔科夫链求得状态转移矩阵,并得到了GEP模型预测的状态区间以及相应概率,增加了预测结果的可信度。
The paper introduces a method of multivariate nonlinear function modeling using gene expression programming algorithm,and applies this method into predicting the rank of earthquake,the test results show that predictive precision of GEP model is much higher than neural network model.Finally,based on the non-aftereffect and stable property of the data of the earthquake rank,the transition matrix,the interval of GEP model and the corresponding probability can be obtained by Markov chain,and it increases the reliability of prediction result.
出处
《信息技术与信息化》
2017年第12期61-63,共3页
Information Technology and Informatization
关键词
基因表达式编程
函数挖掘
地震等级预测
Gene Expression Programming
Function mining
Markov chain
Earthquake Rank Forecasting
