作　　者： ; ; ; ;

机构地区： 中南大学材料科学与工程学院粉末冶金国家重点实验室

出　　处： 《粉末冶金材料科学与工程》 2006年第5期272-276,共5页

摘　　要： 为了预测不同成分的树脂基摩擦材料的摩擦性能,建立了摩擦材料成分与摩擦因数之间的人工神经网络(ANN)预测模型。用收集到的30种不同组分的摩擦材料在100℃时的摩擦因数数据作为训练样本对网络进行训练,然后进行拟合,结果表明,网络拟合值与实验数据吻合很好。最后利用该模型对不同成分的4种摩擦材料进行摩擦因数的预测,并研究单一成分对树脂基摩擦材料因数的影响。结果表明,预测值与实测数据基本相符,100℃时预测的摩擦因数最小值μmin.p与实验中实测的摩擦因数最小值μmin.m的相对误差小于13%,相应的摩擦因数最大值μmax.p与μmax.m的相对误差小于9%。 A model for predicating friction coefficient（COF） of resin-based friction materials based on artificial neural network（ANN） was established to forecast its friction performance, and the ANN network model was trained by using 30 kinds of frication material＇s data of friction coefficient, and then fitted. The results show well anastomosis for experiment data and simulated results. Four styles of resin-based friction material were respectively discussed according to the trained ANN mode, at the same time, the influences of single element of friction properties was discussed according to the curves ploted by the ANN model. The simulated results show that the predicted and measured results are in good agreement; the relative errors are rather low, for example, lower than 13% for mini-COF （μmax）, and lower than 9% for max-COF（μmax）.

关 键 词： 树脂基摩擦材料 神经网络 摩擦因数 磨损率

领　　域： [化学工程] [冶金工程] [冶金工程]