Abstract: Parts in compressors sometimes is deformed with time going.Therefore it is very important to monitor the deformation and predicate the trend to keep the performance of the compressors.In this paper，the base theories for time series predication model are introduced here.Parameter identification methods are also used.Based on the deformation data of muffle plate in compressor，a predication model for this part is established to monitor and forecast the deformation.Practical applications show that the model in this paper is reliable and useful.

摘要： 压缩机零件在使用时会随时间的变化发生变形，对于压缩机的性能会有很大的影响，因此在零件进行材料选择和几何尺寸设计时，对相关零件变形的检测和预报是至关重要的。介绍了时间序列分析的基本理论及其预测建模的方法；采用参数识别方法，结合压缩机消音板发生变形的动态数据，建立该压缩机零件变形的时序预测模型，对相关零件变形进行了监测和预报，具有很强的实际运用价值。