摘要
针对砂土液化预测的非线性难题,在分析BP神经网络和混沌优化的各自优缺点的基础上,将混沌优化与梯度下降法相结合,构成了神经网络权值和阈值的一种新的组合优化算法(COBP),并将该组合优化算法用于砂土液化的预测建模。工程实例应用表明,该组合优化模型不仅搜索速度快,全局稳定性好,而且预测精度高,结果可靠,能达到工程应用的精度要求,为砂土液化的非线性预测提供了一种有效方法。
Based on the research of BP neural network and chaos optimization algorithm, a new hybrid optimization model is presented. This model integrates chaos optimization algorithm with BP algorithm, which not only has a BP algorithm＇s quick local search capability, but also can converge strongly to the global optimal result by using the chaos optimization＇ s global search character. The integrated optimization model is applied to predict sand liquefaction in practice. The results show that it is an effective and feasible method to predict sand liquefaction and can quickly converge to the global optimal result.
出处
《自然灾害学报》
CSCD
北大核心
2009年第3期111-116,共6页
Journal of Natural Disasters
基金
国家自然科学基金重点资助项目(50374084)
西南科技大学博士基金项目(08zx7119)
关键词
优化模型
砂土液化预测
混沌优化
BP神经网络
optimization model
sand liquefaction prediction
chaos optimization
BP neural networks
