【摘要】：針對最小二乘支持向量機(jī)模型傳統(tǒng)參數(shù)選擇方法費(fèi)時且效果差的問題,利用蝙蝠算法的模型簡單、快速收斂和全局搜索能力強(qiáng)的特點(diǎn),優(yōu)化模型的正則化參數(shù)和核函數(shù)參數(shù),對水文時間序列建立最小二乘支持向量機(jī)預(yù)測模型�；谖鹘饔騼�(nèi)的柳州水文站2000-2014年月徑流資料對模型進(jìn)行訓(xùn)練和預(yù)測,并與使用粒子群算法優(yōu)化參數(shù)確定的最小二乘支持向量機(jī)模型,網(wǎng)格搜索及交叉驗(yàn)證優(yōu)選參數(shù)確定的最小二乘支持向量機(jī)模型及BP神經(jīng)網(wǎng)絡(luò)模型進(jìn)行比較。計算結(jié)果表明,基于蝙蝠算法優(yōu)化最小二乘支持向量機(jī)模型具有很好的適用性和較高的預(yù)測精度,為利用最小二乘支持向量機(jī)模型解決非線性的水文時間序列問題提供了新的方向。
[Abstract]:Aiming at the problem that the traditional parameter selection method of least squares support vector machine (LS-SVM) is time-consuming and ineffective, the regularization parameters and kernel function parameters of the model are optimized by using the characteristics of the bat algorithm, such as simple model, fast convergence and strong global search ability. The prediction model of least squares support vector machine is established for hydrological time series. Based on the monthly runoff data of Liuzhou hydrologic station in Xijiang River Basin from 2000 to 2014, the model is trained and predicted, and the least squares support vector machine (LS-SVM) model is used to optimize the parameters by using particle swarm optimization (PSO). The least square support vector machine (LS-SVM) model and the BP neural network model are compared with each other. The results show that the optimization of least-squares support vector machine model based on bat algorithm has good applicability and high prediction accuracy. It provides a new direction for solving nonlinear hydrological time series problems using least square support vector machine (LS-SVM) model.
【作者單位】： 華北電力大學(xué)經(jīng)濟(jì)與管理學(xué)院;
【基金】：河北省社會科學(xué)基金項(xiàng)目(HB16YJ075)
【分類號】：P338;TP18
，

本文編號：2378831