QUALTES: A Domain Specific Tool for Electric Power Stations

QUALTES: A Domain Specific Tool for Electric Power Stations

Tsunehiko Suzuki(!), Ryuji Kudo< 2 ), Katsunori Ikami< 2 ), Ken lida< 2 ), Takao Terano< 3 )

(1) Computer Technology Integrator, Ltd., tss@cti.co.jp,

2-1-80, lnae, Minato-ku, Nagoya 455, Japan

(2) Nihon Unisys, Ltd., ryu@ksd.unisys.co.jp,

2-17-51 Akasaka, Minato-ku, Tokyo 107, Japan

1. Graduate School of Systems Management, The University of Tsukuba,

terano@gssm.otsuka.tsukuba.ac.jp, 3-29-1 Otsuka, Bunkyo-ku, Tokyo 112, Japan

Abstract

QUALTES is a domain specific tool for develop­ ing knowledge-based systems (KBSs) for elec­ tric power substations. The task of the KBSs is to support human operators to make and verify Command Table Sheets (CTSs). CTSs define qperation procedures for stoppage sequences at a specific substation. This paper describes the principles of QUALTES behind the functions shown in the videotape demonstration .

Background of QUALTES

While executing maintenance tasks at an electric power substation, complex operations on the facilities or equip­ menta are required. To prevent mistakes from occurring, prior to beginning the tasks, Command Table Sheets (CTSs) are reviewed. The CTSs have been first de­ scribed by human operators and then verified by human experts, so far. There are about 70 substations in Chubu Electric Power Company to maintain through the use of the CTS method. The objective of the QUALTES project is to provide the utility personnel with a cost­ effective way to develop specific knowledge-based sys­ tems (KBSs) to support the CTS method at each sub­ station with different facilities and operational heuristic rules[Suzuki et al., 1993].

Principle 1: Domain Specific Approach

There are two approaches to develop special purpose AI tools: the task specific one and the domain specific one [Terano, 1991] . The former is useful when the task level knowledge to solve a given problem requires complex in­ ference mechanisms [Chandrasekaran et al., 1992]. The latter is effective when the characteristics of target do­ main tasks are well specified and when there are needs to develop similar applications. The requirements on the QUALTES project meet the domain specific one.

Principle 2: Model-Based Approach

Similar to PLEXSYS [EPRI, 1989], QUALTES has plant component models, model building functions, model­ based reasoning capability, and simulation functions. Generic Knowledge Base (GKB) of QUALTES contains class libraries on general substation facilities, declarative descriptions on power systems, and generic simulators.

In this sense, we adopt the model-based approach to de­ velop QUALTES. However, unlike PLEXSYS, the task also requires empirical knowledge hardly represented only by the plant models. The simulator solves phys­ ical constraints problems with the empirical knowledge on operational regulations.

Principle 3: Graphic Oriented Interfaces

Sophisticated graphic interfaces are critical for KBS de­ velopment by end users. QUALTES has (1) an intel­ ligent graphical editor with functions for verifying the correctness of generated CTSs, (2) a mimic diagram board subsystem to represent the substation facilities on a display, and (3) integrated simulators and consultation functions to verify the CTSs.

Concluding Remarks

Using QUALTES, the complex task of implementing a specific knowledge-based system can become accom­ plished by the end users themselves at the substations. QUALTES has been highly accepted in field tests and in demonstration to human experts. Specific KBSs de­ veloped by QUALTES are now on field testing at two of the substations. After completing the evaluation phase, we have a plan to use QUALTES in company-wide scale.