摘要
为了提高组合预测精度,将最小二乘支持向量机(LS-SVM)用于确定组合预测的函数关系,提出了基于LS-SVM的非线性组合预测方法;为了提高LS-SVM的学习性能和泛化能力,提出了利用粒子群优化算法(PSO)和K-重交叉验证(CV)相结合的参数寻优方法;最后利用提出的方法对某导弹发射车液压系统的液压油污染度进行了预测,仿真结果表明了提出方法的优越性。
For the purpose of improving the predict accuracy of combination forecasting, firstly a so-called ' least squares support vector machine(LS-SVM) based nonlinear combining forecast' method was proposed,which use LSSVM to model the nonlinear relationship between the components and combine. Then, for the purpose of improving the learning performance and generalization ability of the LS-SVM, particle swarm optimization algorithm (PSO) combined with k-fold cross validation (CV) method was proposed to select the optimal parameters of LSSVM. Lastly, the proposed method was used to forecast the fluid contamination of a hydraulic system in missile launcher. Simulation results show its superiority.
出处
《传感技术学报》
CAS
CSCD
北大核心
2012年第5期712-717,共6页
Chinese Journal of Sensors and Actuators
关键词
油液污染度预测
组合预测
最小二乘支持向量机
粒子群优化
fluids contamination forecast
combination forecasting
least squares support vector machine ( LS-SVM )
particle swarm optimization (PSO)