基于LS-SVM和Q(λ)学习的铁路绝缘子水冲洗定位研究

Research on Railway Insulator Flushing Location Based on Least Square Support Vector Machines and Q(λ) Learning

电气化铁路绝缘子智能冲洗的关键在于绝缘子跟踪、定位,通过对现有冲洗研究的分析,提出一种基于最小二乘支持向量机(LS-SVM)和Q(λ)学习的铁路绝缘子水冲洗定位方法。将水炮连续动作空间下的冲洗行为离散化,针对连续状态空间下Q(λ)学习可能面临维数灾难等问题,利用LS-SVM对Q值进行在线估计,为节省估计时间,训练样本采用滚动时间窗机制进行整体更新。Matlab仿真结果表明:该方法可将绝缘子定位时间缩短为0. 006 s,能够达到工程识别要求。

Insulator tracking and location is the key to realize the intelligent flushing on electrified railway,through the analysis of current flushing researches,a kind of location method is proposed based on least squares support vector machine(LS-SVM)and Q(λ)learning.Discretize flushing behavior of the water cannon under the continuous action space,then LS-SVM was proposed to estimate Q value in order to solve the curse of dimensionality problem caused by continuous state space,besides,the rolling time window mechanism was used to save the estimated time of training samples;The Matlab simulation result shows that this method can short the location time to 0.006 s,which can meet the engineering identification requirement.