【摘要】 在机床产品可靠性评估中,针对采集到的有效故障数据较少而导致失效分布无法唯一确定的问题,提出建立RBF神经网络故障数据扩充算法模型。该模型通过自组织聚类学习算法确定神经网络的径向基函数中心和扩展常数,然后利用学习算法的输出结果对神经网络进行训练,随后将随机生成的累积失效分布函数导入训练好的神经网络,得到与原数据组具有一致失效分布规律的仿真数据,从而确定机床故障分布类型。最后通过实例分析表明:此方法确定的机床故障分布类型为威布尔分布,对其进行的可靠性评估显示MTBF估计值为909.20h,可靠度估计值为0.4874,同时机床已经进入耗损失效期,在后续工作中应该制定相应预防性维修措施来延长机床使用寿命。更多还原

【Abstract】 In the reliability evaluation of machine tools,aiming at the problem that failure distribution cannot be uniquely determined due to the fewer effective data collected,the RBF neural network failure data extension algorithm is proposed.The radial basis function center and extension constant of the neural network are determined by self-organizing clustering learning algorithm.The neural network is trained with the output results of the learning algorithm,then,the randomly generated cumulative failure data are imported into the trained neural network to generate the expanded data with the same statistical law as the original data,so as to determine the failure distribution model.At last,through the example analysis shows that the failure distribution model conforms to Weibull distribution,the reliability evaluation shows that the MTBF estimated value of the machine tool is909.20 h and the reliability estimated value is 0.4874.At the same time,this machine has entered the wear-out failure period and should formulate corresponding preventive maintenance measures to prolong the service life.更多还原