在对实际检测数据进行深入分析的基础上，提出了对机械加工尺寸数据序列进行在线建模和预测的时间序列分析原理和方法，通过加工实验对所提出的原理和方法进行了验证，表明所建的自回归滑动平均（ＡＲＭＡ）模型及基于此模型的预测结果是相当令人满意的 On the basis of deep analysis of the actual test data, the principle and method of time series analysis for on-line modeling and forecasting of machined size data series are proposed. The proposed principles and methods are verified through machining experiments, The built autoregressive moving average (ARMA) model and the predictions based on this model are quite satisfactory