油液檢測手段主要有常規理化性能檢驗、油料光譜分析、鐵譜分析和顆粒計數四種，其目的在於控制油液的性能變化，控制設備狀態變化和診斷設備故障，分析零件失效。進行資訊提取主要目的是為了建立專家系統所必需的知識庫，所以要求資訊提取，真實準確，並能夠提供相應的數學模型，以便於資訊的提取和識別。本文對油液檢測手段中的油料光譜分析進行描述，以獲得專家系統所必需的知識資訊。

　　　　　There are four means in oil analysis including routine performance analysis, spectrometric analysis, ferrographic analysis and grain analysis in order to control and valuate the estate of used oil and equipments. We can analyze the reason of invalidation of equipments. First, we must pick-up and establish timely, accurately and largely character information from the oil used in these equipments. Second, we must put these character information into knowledge based and establish relevant mathematics model. This paper discusses and solves these problems in basic spectrometric analysis about how to put these character information into knowledge base and establish relevant mathematics model.