基于PSO优化LS-SVM的小样本非线性协整检验与建模研究

针对小样本非线性时间序列，根据非线性协整的定义，利用基于粒子群优化最小二乘支持向量机的方法，对小样本非线性协整关系检验与非线性误差修正模型建模进行研究，设计了方法的 逻辑流程. 对舰船维修费指数与物价指数进行实证研究，在协整关系类型判断的基础上，实现了小样本非线性协整关系的检验，建立了预测舰船维修费指数的非线性误差修正模型，并与线 性向量自回归模型进行分析比较. 研究表明：基于粒子群优化最小二乘支持向量机的小样本非线性协整检验与建模方法，刻画了小样本系统的非线性协整关系，所建立的非线性误差修正模 型具有较好的预测效果，能够有效地预测小样本非线性系统.

Nonlinear cointegration test and error correction modeling based on LS-SVM optimized by PSO in small sample

Aiming at the small sample time series, a nonlinear cointegration test and error correction modeling method based on the least squares support vector machine (LS-SVM) which is optimized by particle swarm optimization (PSO) is put forward according to the definition of nonlinear cointegration, and the logic process is designed. Then the empirical study on the ship maintenance price index (SMPI) and several price indexes is analyzed through the method introduced in this paper. The nonlinear cointegration test among the empirical data is realized after the judgement of cointegration types, and the nonlinear error correction model (NECM) of SMPI is established. The results indicate that the method of nonlinear cointegration test and error correction modeling based on LS-SVM optimized by PSO in small sample describes the nonlinear cointegration of small sample system well, and the NECM established by this method can availably forecast the small sample nonlinear system comparing to the linear vector auto regressive model.