样本数据的质量对BP神经网络模型学习能力和预测能力起直接决定作用。本文以DMC60H卧式铣削加工中心为平台,以铝合金壳体类零件为试验对象,采集数控加工铣削试验数据并进行分析研究,提出试验数据和样本数据处理的方法并分析出验证数据组成部分及各组成部分与样本数据的比例,完成了试验数据的优化选取同时删除了试验数据中有误的信息。 The quality of the sample data plays a direct role in the learning ability and predictive ability of the BP neural network model. In this paper, DMC60H horizontal milling machining center as a platform to aluminum alloy shell parts as the test object, collecting CNC machining milling test data and analysis, the test data and sample data processing methods and analysis of the verification data components and The ratio of each component to the sample data completes the optimization selection of the test data and deletes the incorrect information in the test data.