摘要
针对可靠性预测精度低的问题提出了一种支持向量机回归预测模型。在可靠性预测过程中,开发了一种结合正弦映射和自适应策略来更新惯性权重的自适应粒子群优化算法,通过增强算法的局部挖掘能力与全局搜索能力,在一定程度上提高了粒子群算法的精度和收敛效率。基于8种测试基准函数将提出的算法与其他粒子群算法进行比较验证,结果表明,提出的自适应粒子群算法相比于其他算法具有更好的搜索能力。在此基础上,提出了一种新的自适应粒子群优化-支持向量机回归混合可靠性预测模型,对支持向量机回归的参数进行调整并预测涡轮增压器和工业机器人系统的可靠性,结果表明该混合模型在可靠性预测方面可达到实际工程精度要求。
Aiming at the problem of low reliability prediction accuracy,a support vector regression prediction model is proposed.In the process of reliability prediction,an adaptive particle swarm optimization algorithm that combines sine mapping and adaptive strategies to update inertia weights is developed.By enhancing the local mining capabilities and global search capabilities of the algorithm,it is improved to a certain extent.The accuracy and convergence efficiency of the particle swarm algorithm are verified.Based on 8 benchmark functions,the proposed algorithm is compared and verified with other particle swarm algorithms.The results show that the adaptive particle swarm optimization algorithm has better search capabilities than other algorithms.On this basis,a new adaptive particle swarm optimization-support vector machine regression hybrid reliability prediction model is proposed to adjust the parameters of support vector regression and predict the reliability of turbochargers and industrial robot systems.The results show that the hybrid model can meet the actual engineering accuracy requirements in terms of reliability prediction.
作者
周策
白斌
叶楠
ZHOU Ce;BAI Bin;YE Nan(School of Mechanical Engineering,Hebei University of Technology,Tianjin 300401;SANY Heavy Industry Co.,Ltd.,Changsha 400100)
出处
《机械工程学报》
EI
CAS
CSCD
北大核心
2023年第14期328-338,共11页
Journal of Mechanical Engineering
基金
国家重点研发计划(2017YFB1301300)
国家自然科学基金(11772011)
河北省自然科学基金(E2020202217)
河北省教育厅基金重点(ZD2020180)
国防科学与技术重点实验室粒子输运与富集技术开放基金(STPTS202111)
湖南省教育厅基金(21C1235)资助项目。
关键词
可靠性预测
支持向量机
粒子群算法
涡轮增压器
工业机器人
reliability prediction
support vector machine
particle swarm algorithm
turbocharger
industrial robot
