【作者】 陈斯；

【导师】 濮文虹；

【作者基本信息】 华中科技大学 ， 环境工程， 2006， 硕士

【摘要】 由于水处理技术涉及知识面广,既有定性分析,还有定量计算,因此,为了满足临时多变的工程要求,开发一套具有水处理基本知识、通晓所有水处理技术又能单独深入处理专门问题的水处理专家系统是十分必要的。本文针对目前水处理专家系统的现状,系统地分析了专家系统的结构、组成及开发过程。在总结我国水处理技术的基础上,针对现有水处理专家系统的不足,重新规定了知识库所包括的水处理知识内容,提出了新的获取这些知识的途径和方法;并应用产生式规则建立了水处理知识库。本文详细介绍了构建知识库和实现推理机的主要数据结构和设计思想,同时,根据水处理要素不确定性的特点,将模糊数学引入水处理专家系统中,比较模糊规则和确定性规则两种形式,总结研究出一种适合水处理领域的模糊匹配和模糊推理方法。应用Vs.net和数据库设计语言开发了一套膜清洗诊断和水处理咨询专家系统。检测过程中,通过众多隶属函数值与阈值的比较可以得出与实际情况最相关的隶属值,具体实例对该系统的测试表明该系统运行结果准确。最后,在研究中发现知识录取的准确程度直接影响推理结果的精确性,并且考虑到专家系统的生存性,今后必须提高推理机制的算法和增加知识库的容量。更多还原

【Abstract】 The water treatment technology possesses a broad knowledge, including both qualitative analysis and quantitative calculation. To satisfy different requirements of project and dealing with a problem specially, developing an expert system with basic but skilled knowledge of water treatment technology is necessary.In this work, we analyzed the structure and composition of an expert system as well as the development process, aiming at the actuality of the water treatment expert system. Based on the summarization of water treatment technology and the shortage of the present water treatment expert system, new water treatment knowledge content was re-stipulated and a new method to acquire the knowledge was proposed in this thesis. Furthermore, the production rules were employed to develop the water treatment knowledge base. The main data structure and the design ideas used to build the knowledge base and perform the implement of the inference process were introduced in detail. According to the uncertain characteristic of the water treatment factors, fuzzy rules and definite rules were compared with the method of fuzzy mathematics in the water treatment expert system. A fuzzy matching strategy and fuzzy reasoning method that could be employed in the field of water treatment were summarized.Vs.net and database design language were employed to develop the water treatment expert system, including membrane fouling diagnosis and consultation. The most correlated membership value could be obtained by comparing the membership functions value and the threshold value in the monitoring process. The accuracy of the system operation was proved by the specific tests.At last, the knowledge enrollment was found to have a direct impacted on the accuracy of the conclusion results. Thus, taking into account the survival of the expert system, the algorithm of the reasoning mechanism must be improved and the capacity of the knowledge base should be increased in the future.更多还原