基于微分进化算法的逻辑斯谛沉降预测模型

Logistic Settlement Prediction Model Based on Differential Evolution

预测沉降的逻辑斯谛曲线模型,其函数表达式是一非线性方程,采用传统方法对模型参数进行回归处理往往因为计算复杂和人为因素的影响,而使预测结果带有较大的误差。为此,用微分进化算法来直接估计逻辑斯谛曲线模型的待定参数,从而建立了基于微分进化算法的逻辑斯谛沉降预测模型。基于MATLAB环境编写了算法的实现程序,结合实际工程的沉降观测数据进行了模型的应用研究。实例计算结果表明,基于微分进化算法的逻辑斯谛曲线模型具有良好的适应性,能很好地反映地基沉降变形的规律,拟合实测沉降数据精度较高,该模型具有一定的参考和使用价值;作为一种简单、鲁棒的算法,微分进化算法在求解高维复杂非线性问题中表现出求解速度快、自动化程度高、程序通用性强、初值区间选择无关性等优点。

The formula expression of logistic settlement prediction model is a super-set and nonlinear equation. In order to avoid the high margin of error which resulted form traditional method because of its complicated calculation process or artificial factors, the logistic settlement prediction model based on differential evolutionary （DE） has been established. Based on the software MATLAB environment, the DE solution program is compiled to estimate the parameters of logistic settlement prediction model directly. Then, the logistic settlement prediction model based on DE was applied to fitting the settlement data observed from practical projects. The results make clearly that the logistic settlement prediction model based on DE has a good adaptability that can fit the observed data well and reflect the settlement-time relationship effectively. Hence the logistic settlement prediction model based on DE has some reference value in practice. Besides, as a new, simple and robust method, DE is such a high effective algorithm in solving high-dimensional and complex nonlinear problems with many good advantages such as high efficiency, high precision, fast computing speed, small initial value affection, etc.