摘要
根据机械部件磨损机理复杂、磨损量预测难精确的特点,提出基于免疫粒子群参数优化的最小二乘支持向量机方法预测磨损量.该算法采用免疫粒子群优化最小二乘支持向量机建模参数,避免了算法陷入局部最优解,实现了精确度高、泛化能力强的磨损量预测模型.对轴承钢试件磨损进行了试验研究,试验数据分析结果表明,基于免疫粒子群的最小二乘支持向量机预测方法优于前向反馈神经网络算法、遗传算法及蚁群算法,预测误差较小,具有很好的预测能力.
Because the wear loss is complicated and difficult to measure, a method of parameter optimized least square support vector machine is introduced to predict wear loss. The optimized algorithm is immunity-particle swarm which could avoide getting into local best place. The optimized model of wear loss was proposed. The training and measuring data set was obtained from an experiment about wear loss. The prediction results of the optimized LS-SVM and neural network, ant colony optimization, genetic algorithm were compared, it shows that the immuni- ty-particle swarm LS-SVM model is feasible and accurate for the prediction of wear loss.
出处
《摩擦学学报》
EI
CAS
CSCD
北大核心
2008年第6期562-566,共5页
Tribology
基金
国家高技术研究发展计划(863计划)资助项目(2006AA04Z427)
国家自然科学基金委员会与中国民用航空总局联合资助项目(60672164)
关键词
磨损量
最小二乘支持向量机
免疫粒子群
优化
wear loss, least square support vector machine, immunity-particle swarm, optimization
