权益分类	功能权益	普通用户	{{item.name}}会员
{{category.name}}	{{benefitItem.name}}

Expert System of Reservoir Operation for Flood Control with Artificial Intelligent Techniques

人工智能技术水库防洪调度专家系统

基本信息

批准号：
03555116
负责人：
KOJIRI Toshiharu
金额：
$ 2.5万
依托单位：
Gifu University
依托单位国家：
日本
项目类别：
Grant-in-Aid for Developmental Scientific Research (B)
财政年份：
1991
资助国家：
日本
起止时间：
1991 至 1992
项目状态：
已结题

项目摘要

In this project, the expert system of real-time reservoir operation was proposed through the fuzzy inference theory and other artificial intelligent technologies. The release discharge at each control time stage in real-time operation was calculated with fuzzy inference or object oriented procedures considering predicted flood scale and monitored hydrological data. Concretely, Knowledge-based reservoir operation was proposed through fuzzy classification method and fuzzy inference engine. To sum up, the following results were listed :i) Historical typhoons were classified through soft classification method to get the typical typhoon courses. Typhoon course after current time stage as predicted through fuzzy inference.ii) Hyetographs in classified typhoon group were classified into a few patterns. Hydrograph was estimated with predicted hyetographs.iii) Knowledge-base of control rule were extracted applying dynamic programming to the classified hydrographs, the representative current discharge and representative storage volume.iv) Release discharge was calculated by using predicted hydrograph, observed discharge and observed storage volume for flood scale. Two levels inference engine were established for reservoir operation.v) Tuning reservoir operation was introduced to modify the hourly release volume into shorter control time step. Control rules were formulated by using the target storage level and the difference between current storage and target storage levels.vi) The integrated reservoir operation was simulated on big computer or work station for knowledge base and rule base, on personal computer for real time operation. These systems will be useful for dam operator in the isolated reservoir control office and the emergency flooding.

本课题通过模糊推理理论等人工智能技术，提出了水库实时调度专家系统。结合预测洪水规模和监测水文资料，采用模糊推理或面向对象方法计算实时运行中各控制时段的放水量。具体而言，通过模糊分类方法和模糊推理机，提出了基于知识的水库调度方法。综上所述，结果如下：1)采用软分类方法对历史台风进行分类，得到典型台风路径。通过模糊推理预测当前时间阶段后的台风走向。ii)分类台风群的静息图被划分为几种模式。用预测的水流图估计水流线。iii)将动态规划应用于分类水线、代表性电流流量和代表性蓄水量，提取控制规则知识库。iv)利用洪水尺度下的预测线、观测流量和观测蓄积量计算释放流量。建立了水库运行二级推理机。v)引入调校油藏操作，将每小时释放量修改为更短的控制时间步长。利用目标存储级别以及当前存储与目标存储级别的差值制定控制规则。(6)在大型计算机或工作站上模拟水库综合作业，建立知识库和规则库，在个人计算机上进行实时操作。这些系统将为大坝操作员在隔离水库控制室和紧急洪水中提供有用的信息。