药物气相色谱分析专家系统的柱系统推荐

Column System Recommendation of Expert System for Gas Chromatography Used in Pharmaceutical Analysis

报道了药物气相色谱分析柱系统推荐专家系统 .该系统可提供文献报道的 GC分析方法及文献出处 ;判断样品能否直接采用 GC分析 ,推荐固定相、柱温、固定液用量范围、检测器、载气等 ,并能对所推荐的色谱初始条件进行优化 .对固定相 ,系统可分别从 3个通道给予推荐 ,且选择通道的优先顺序依次为特殊组分 (同系物 ,异构体 )、组分的 Kováts指数以及组分的结构和极性 .本文以 Turbo Prolog为程序设计语言 ,采用产生式规则和谓词逻辑的知识表达方式以及深度优先搜索的正向链推理控制策略 ,在 IBM PC机上编程实现了这一系统 .

An expert system for column system recommendation of gas chromatography used in pharmaceutical analysis is reported. The expert system is able to give GC analytical method(s) reported by literature(s), judge if a sample can be directly analyzed by GC, recommend stationary phase, column temperature, stationary phase content, detector, carrier gas etc. and optimize the chromatographic operation conditions recommended by itself. Stationary phase recommendation can be done from three different passages which are selected by the following order: firstly special components (homologues and isomers), then the Kováts index of components and thirdly the structure and polarity of components. The knowledge base contains about 2 000 Kováts indices got on eight stationary phases and more than 1 000 GC analytical methods of different drugs and related compounds. Production rule and predicate logic are used to represent knowledge and forward chaining method is applied to give inference. The expert system is implemented by using AI language Turbo Prolog and operated on IBM PC.