Development of an Expert System for flood Control Supporting based on Co-operating Knowledge-based Systems

设计了一个基于协同知识系统的防洪保障环境。本文提出的系统在子系统之间有两个层次的协作。在合作的上层，用数学算法处理定量数据的程序性知识系统与用语言表达的定性信息和知识的推理系统相联系。它们通过各系统的信息处理和系统间的信息交换，综合支持流量控制工作。较低层次的合作发生在推理系统中基于知识的系统之间。本文提出的推理系统由多个独立的基于知识的系统组成，每个系统都是一个专家系统，具有推理机、知识库和数据库。每个基于知识的系统处理洪水控制问题的部分问题。通过各个基于知识的系统中的推理过程和系统间的通信，解决水库应该放水多少等防洪问题。在本研究中，根据每个系统要处理的信息类型，我们采用两种类型的基于知识的系统：一种是基于生产系统的推理过程，另一种是基于模糊推理过程。研究了以原型形式设计的专家系统的改进方法，提出了基于知识的系统的层次结构。在该框架中，根据决策依据的数据类型，将上述设计的基于知识的系统划分为多个基于知识的系统，并引入元知识系统对这些系统进行管理。

A flood control support environment, which is based on cooperatin knowledge-based systems, is designed. The system proposed herein has two levels of cooperation among its subsystems.On the upper level of cooperation, a procedural knowledge system, which processes quantitative data with mathematical algorithm, is linked with an inference system, which deals with qualitative information and knowledge expressed in the form of language. They synthetically support floow control work through information processing in each system and information interchange between them.The lower level of cooperation takes place among knowledge-based systems in the inference system. The inference system proposed here is composed of several self-contained knowledge-based systems, each of which is an expert system with an inference engine, a knowledge base and a data base. Each knowledge-based system deals with a partial issue of a flood control problem. Flood control problems, such as how much water should be released from a storage reservoir, are solved through the inference process in each knowledge-based system and the communication among them. In this study, we adopt two types of knowledge-based systems according to the information types to be processed in each system : one with an inference process based on a production system and the other with a fuzzy inference process.The way how to imprive an expert system designed as a prototype is also investigated and the hierachical structure of knowledge-based systems is proposed. In this framework, the knowledge-based system designed above is divided into several knowledge-based systems according to the type of data used as grounds of decision and the meta knowledge-based system is inroduced to manage these systems.