基于IPSO-BP算法的城轨列车轮对故障率预测模型研究
|
摘要
为了提高城轨列车轮对故障率的预测精度,文章采用人工神经网络方法代替传统维修策略模型中基于经验的故障率分布显示表达式,以避开故障分布模型的选择;建立了IPSO-BP(improved particle swarm optimization-backpropagation)预测模型,并通过与常规的BP(backpropagation)及PSO-BP(particleswarmoptimization-backpropagation)预测模型进行对比来验证其高效性。仿真结果显示,IPSO-BP神经网络模型的预测误差范围为0～5.5%,输出值的相对误差百分比为0～10%,预测精度均优于常规方法,可为预防性维修决策提供理论参考和方法支撑。
In order to improve the prediction accuracy of wheel set failure rate of urban rail trains, the artificial neural network method was used to replace the empirical expression of failure rate distribution in the traditional maintenance strategy model to avoid the manaul selection of fault distribution models. The IPSO-BP(improved particle swarm optimization-back propagation) prediction model was established and compared with the conventional BP(back propagation) and PSO-BP(particle swarm optimization-back propagation)prediction model. Comparisons are made to verify its efficiency. The simulation results show that the prediction error range of IPSOBP neural network model is 0～5.5%, and the relative error percentage of output value is 0～10%. The prediction accuracy of IPSO-BP neural network model is better than that of conventional methods, which can provide theoretical reference and methodological support for preventive maintenance decision-making.
In order to improve the prediction accuracy of wheel set failure rate of urban rail trains, the artificial neural network method was used to replace the empirical expression of failure rate distribution in the traditional maintenance strategy model to avoid the manaul selection of fault distribution models. The IPSO-BP(improved particle swarm optimization-back propagation) prediction model was established and compared with the conventional BP(back propagation) and PSO-BP(particle swarm optimization-back propagation)prediction model. Comparisons are made to verify its efficiency. The simulation results show that the prediction error range of IPSOBP neural network model is 0～5.5%, and the relative error percentage of output value is 0～10%. The prediction accuracy of IPSO-BP neural network model is better than that of conventional methods, which can provide theoretical reference and methodological support for preventive maintenance decision-making.
引文
[1]李建伟,程晓卿,秦勇,等.基于BP神经网络的城市轨道交通车辆可靠性预测[J].中南大学学报(自然科学版),2013(增刊1):42-46.
[2]董锡明.机车车辆故障与可靠性分析[J].铁道机车车辆,2004,26(6):42-47.
[3]尹怀仙,王凯,张铁柱,等.基于PSO-BP神经网络的城轨列车转向架轮对轴箱故障预测[J].复杂系统与复杂性科学,2015,12(4):97-103.
[4]李笑梅,贺德强,谭文举,等.基于EEMD和RBFNN的列车滚动轴承故障诊断研究[J].铁道科学与工程学报,2017,14(5):1056-1062.
[5]马锐.人工神经网络原理[M].北京:机械工业出版社,2010.
[6]GAO H B,ZHOU C,GAO L.General particle swarm optimization model[J].China Journal of Computers,2005,28(12):1980-1987.
[7]KENNEDY J,EBERHART R.Particle swarm optimization[C]//Proceeding of IEEE International Conference on Neutral Networks.Piscataway,NJ:IEEE,1995,4:1942-1948.
[8]SHI Y,EBERHART R.Modified particle swarm optimizer[C]//Proc.of IEEE ICEC Conference.Anchorage:IEEE,1999:69-73.
[2]董锡明.机车车辆故障与可靠性分析[J].铁道机车车辆,2004,26(6):42-47.
[3]尹怀仙,王凯,张铁柱,等.基于PSO-BP神经网络的城轨列车转向架轮对轴箱故障预测[J].复杂系统与复杂性科学,2015,12(4):97-103.
[4]李笑梅,贺德强,谭文举,等.基于EEMD和RBFNN的列车滚动轴承故障诊断研究[J].铁道科学与工程学报,2017,14(5):1056-1062.
[5]马锐.人工神经网络原理[M].北京:机械工业出版社,2010.
[6]GAO H B,ZHOU C,GAO L.General particle swarm optimization model[J].China Journal of Computers,2005,28(12):1980-1987.
[7]KENNEDY J,EBERHART R.Particle swarm optimization[C]//Proceeding of IEEE International Conference on Neutral Networks.Piscataway,NJ:IEEE,1995,4:1942-1948.
[8]SHI Y,EBERHART R.Modified particle swarm optimizer[C]//Proc.of IEEE ICEC Conference.Anchorage:IEEE,1999:69-73.