CAJ | 학술논문

为了提高基于最小二乘支持向量机（LSSVM）模型的时间序列预测方法的泛化能力及预测精度，研究一种基于自适应差分进化算法（ADE）的最小二乘支持向量机模型（ADE＿LSSVM）。首先利用相空间重构技术对样本数据进行相空间重构，再利用ADE对LSSVM的两个参数进行组合寻优，最后利用Lorenz系统对模型进行仿真试验并与未进行参数优化的LSSVM预测结果对比。结果表明，ADE＿LSSVM方法是一种可行的、有效的混沌时间序列预测方法，可为今后的科学研究提供新的理论思想。
위료제고기우최소이승지지향량궤（LSSVM）모형적시간서렬예측방법적범화능력급예측정도，연구일충기우자괄응차분진화산법（ADE）적최소이승지지향량궤모형（ADE＿LSSVM）。수선이용상공간중구기술대양본수거진행상공간중구，재이용ADE대LSSVM적량개삼수진행조합심우，최후이용Lorenz계통대모형진행방진시험병여미진행삼수우화적LSSVM예측결과대비。결과표명，ADE＿LSSVM방법시일충가행적、유효적혼돈시간서렬예측방법，가위금후적과학연구제공신적이론사상。
In order to improve the generalisation ability and prediction accuracy of time series prediction which is based on least squares support vector machine (LSSVM)model,we study an adaptive differential evolution algorithm (ADE)-based LSSVM (ADE_LSSVM) model.First,we use phase space reconstruction technology to reconstruct the phase space of sample data,and then use ADE to optimise two parameters of LSSVM in combination,finally we make use of Lorenz system to simulate the model in experiment,and compare it with the result of LSSVM model prediction without parameters optimisation.Results show that the ADE_LSSVM method is a feasible and effective prediction method for chaotic time series;it provides new theoretical idea for future scientific researches.