通过三层BP人工神经网络 ,对溶剂系统进行模式识别。以经过规格化处理的Xt、Xe、Xm、Xn、Xd这 5个溶剂性质的选择性参数作为判别指标 ,5类共 19种不同的溶剂作为训练样本 ,得到最佳网络参数。然后以十多种有代表性的溶剂作为待测样本进行了模式识别。研究结果表明 ,人工神经网络用于溶剂系统的模式识别 ,识别结果与实际一致。本文结果有助于色谱分析及分离技术中溶剂的选择。 Through three-layer BP artificial neural network, the solvent system pattern recognition. The normalized parameters of the five solvent properties of Xt, Xe, Xm, Xn, Xd were used as discriminant indexes, and 19 kinds of different solvents in 5 categories were selected as training samples to get the best network parameters. Then, a dozen kinds of representative solvents were used as the sample for pattern recognition. The results show that the artificial neural network is used for the pattern recognition of the solvent system, and the recognition results are in accordance with the actual ones. The results of this paper contribute to the choice of solvent in chromatographic and separation techniques.