原油煉制中減壓塔側線產品是精制潤滑油的原料，其粘度測量對生產有重要意義。通
過統計分析得出了影響側線產品粘度的主要因素，并采用支持向量機回歸方法建立粘度軟測量模型。針對支持向量機訓練參數確定問題，提出了采用差分進化算法的搜索策略，使模型訓練參數的調整過程按預定目標自動快速優化。所構建的粘度軟測量模型預報精度較高，趨勢跟蹤性能良好。
關鍵詞：軟測量；減壓塔；粘度；支持向量機；差分進化；

Abstract: The measurement of viscosity is important for the side-draw product quality of vacuum distillation column in oil refinery. The main factors, which influence the viscosity, were analyzed by statistic methods. Then soft-sensor models were developed based on support vector regression. A search method based on differential evolution was proposed to select proper training parameters. Using this strategy, the parameters can be optimized automatically and efficiently. The developed model got high accuracy and good trend tracking performance on the real industrial data.
Keywords: soft sensor; vacuum distillation column; viscosity; support vector machine;
differential evolution;