%0 Journal Article
%T 采用神经网络的断路器传动机构磨损预测<br>Wear Prediction of Circuit Breaker Transmission Mechanism based on Neural Network
%A 刘创
%A 刘宏昭
%J 机械科学与技术
%D 2017
%X 针对某大型断路器机构系统磨损试验成本高的特点，通过建立磨损预测模型，对其传动机构危险关节的磨损量进行了预测分析。基于2种典型的预测模型建立方法，采用销盘磨损实验数据，分别建立磨损预测模型。对比分析表明Elman网络模型的预测精度较高，可准确的反映磨损率与接触压力、相对滑动速度和材料硬度之间的规律。考虑运动副间隙的存在，基于非线性弹簧阻尼模型，利用ADAMS软件仿真获得传动机构危险关节的动力学参数。基于Hertz接触理论对动力学参数进行变换，并将其作为预测模型的输入信息，对关节的磨损进行预测计算。通过迭代分析，发现随着断路器开断次数的增加，轴套表面一些特定位置的磨损越来越严重。对比采用固定系数下的Archard模型，表明预测模型计算的结果对磨损失效判定更具参考价值。<br>Considering the feature that the wear experiment cost for the circuit breaker mechanism system is very high, the wear prediction model is established for the dangerous joint. Based on the pin-on-disc experiments data, two kinds of typical models are established for the wear prediction. The comparative study shows that the model based on the Elman neural network can accurately reflect the inherent wear law between wear rate and contact pressure, sliding velocity and material hardness with a high prediction precision. Considering the clearance in revolute joint, the kinematic parameters are obtained by using ADAMS software based on the elastic-damping contact force model. Then, the trained Elman model is employed to predict the wear of dangerous joint, using the parameters transformed by Hertz contact model. It can be seen from iterative analysis that wear occurs seriously in some special areas on bushing surface. The comparison result indicates that the wear calculation from prediction model is more useful for wear failure criteria than the Archard model which uses constant coefficient
%K 断路器机构
%K 磨损预测
%K 神经网络
%K Archard模型&lt
%K br&gt
%K circuit breaker mechanism
%K wear prediction
%K neural network
%K archard model
%U http://journals.nwpu.edu.cn/jxkxyjs/CN/abstract/abstract6731.shtml