摘要
为准确预测数控车削时机床的振动情况,搭建了一个能用于测量机床振动信号的实验平台,并运用Levenberg-Marquardt算法改进BP神经元网络,建立了改进BP神经网络的非线性系统预测模型,实现了对机床刀架的振动趋势进行多歩预测,为提高工件加工质量、加工精度和进行故障诊断提供了依据。结果表明L-M算法的收敛速度和预测精度均明显优于标准的BP算法。
In order to predict the vibration of machine tool during turning, an experiment platform is set up for testing the machine tool vibration signal. On this basis, the predictive model of nonlinear system is set up based on improved BP neural network （IBNN） to exactly predict the vibration of the cutting tool. Then the multi-step prediction for vibration trend of the tool carriage is realized. The prediction data provides evidence for improving the processing quality and machining accuracy. The results show that the convergence and forecasting accuracy of Levenberg-Marquardt algorithm is largely superior to that of the standard BP algorithm.
出处
《制造技术与机床》
CSCD
北大核心
2009年第4期63-65,共3页
Manufacturing Technology & Machine Tool
