摘要
运用ANN对机械密封进行寿命预测的方法建立了寿命预测系统的基本模型,提出了以增大磨损载荷(pcv)为手段的寿命加速试验方法,并给出了将加速试验的泄漏率q、端面温度Tf和剩余寿命t转化为工况条件的q’、Tf′和t′的理论方法。用若干组q′、Tf′和t′值训练具有2个输入单元、一个隐含层(3个单元)及1个输出单元的BP网络,再用其它几组值来检验网络可靠性。结果表明该网络具有较高精度,使用BP网络预测机械密封剩余寿命可行、高效。
The dements of BP Artificial Neural Networks (ANN) are clarified and the lifetime prediction method of mechanical seal based on ANN is come up. Firstly, lifetime prediction model is established, and then accelerated life test measure is used by augnenting the abrasion load (pcv) . The methods are also given to change the testing leakage rate q , face temperature Tf and residual life t to the operating mode leakage rate q , face temperature Tf and residual life t. Some q', T^1f and t' values are used to coach the BP ANN which comprises two input units, one hidden layer (three units) and one output trait. Some other q', T^1f and t' values should be used to detect the BP ANN, and the results proclaim that the networks is high definition. The networks are applied to prognosticate the residual life of inservice mechanical seal by inputting its leakage rate and face temperature into the networks.
出处
《流体机械》
CSCD
北大核心
2006年第3期19-23,共5页
Fluid Machinery
关键词
机械密封
BP人工神经网络
寿命预测
mechanical seal
BP artificial neural networks
lifetime prediction