Browse Theses

In reservoir operation it is known that the river reservoir system regulation is highly dependent upon the experience of operators and processing of real-time data. Lately it has been observed that this experience level is likely to decline basically because the staff is showing a high turnover rate (retirement, alternate jobs) and the new replacement staff member create uncertainty and risk because the inexperience of operations.

An efficient way to avoid the loss of valuable knowledge is to develop an expert system which incorporates hydrologic, hydraulic, system analysis, environmental, economical knowledge based on years of practical experience gained by those operating the system.

This dissertation develops a method for the operation of real-time multipurpose, multireservoir system via knowledge-based expert systems. The method explains, in detail, the process of knowledge transfer from experienced operators to knowledge bases where the expertise is stored. The expert system designed also shows the interaction of knowledge bases with weather databases, real-time hydrologic field data, call upon and utilize its own or others mathematical models. The resulting expert system may be not only an advanced decision support system in the management of the operations of multipurpose reservoir systems, but also can provide the knowledge and the assistance resulting from both theory and experience in a logical and organized way for training of new staff.

The benefits obtained on the development of a knowledge-based expert system are multiple. In an expert system very valuable knowledge from highly experienced water managers is stored. Novice operators can quickly gain this information by accessing the expert system. The knowledge gain through expert system is consistent, non-perishable, easy to access, transfer and document and affordable. The data (heuristic and algorithmic) is processed quickly using mostly inferential processes with effective manipulation of large knowledge bases and data structures.

A case study for the Mark Twain River Reservoir System (St. Louis, Missouri) has been developed, tested and it is operational on a real-time basis. The system demonstrates a multireservoir multipurpose operation (hydropower, water supply, flood protection, conservation, water quality, recreation and navigation) for which a knowledge-based real-time operational system has been developed.