![]() |
|||
![]()
|
![]() |
![]() |
![]() |
3.3. STEAM TURBINE DIAGNOSTICS3.3.1. STES The Steam Turbine Expert System (STES) (Armor, 1993) uses all available turbine monitoring information such as phase angle, temperature, shaft position, and vibration readings from various machine running states. Data samples are captured at a specific time interval and stored in a database that is then examined automatically by the expert system. The system extracts features from the data that might indicate a major flaw in the condition of a turbine and then examines these features in more detail for specific flaws as well as the associated cause. The system provides an in-depth diagnosis of any faults that it finds. Faults that can be examined include:
The system also has the facility to automatically schedule for repair after a diagnosis has been made. The system was built with the GEN-X software package, with the knowledge base consisting of a collection of production rules and the inference mechanism using both forward and backward chaining. Probability values were assigned to a conclusion when a diagnosis was made. A separate subsystem was developed for the conversion of numerical readings into a symbolic form with which the expert system could manipulate and reason. The system also has a database that enables the operator to plot historical data, and an easy-to-use interface facilitates interaction between the expert system and the operator. A large version of STES (called LSTES, Large Steam Turbine Expert System) was released; it had the capability to interface to a turbine's instrumentation and take readings online. The LSTES system also facilitates the use of a modem for remote access purposes, so that the system could be used off-site. 3.4. GENERATORS3.4.1. GEMS A system developed by EPRI for use with generators was GEMS (Generator Expert Monitoring System). Failures on generators are rare, but a severe failure of a major insulation or core can force an extended plant outage. An effective monitoring system would provide the operator with early warning of developing generator faults, which would reduce the duration and frequency of generator down-time. The GEMS system interfaces with existing generator sensor equipment and provides operators with warnings of developing problems and recommends corrective action (Armor et al., 1993). The knowledge base for GEMS consists of an extensive set of files put together by experts aimed at covering every possible occurrence of trouble within a generator. GEMS was originally developed as a rule-based system; however, subsequent application of causal networks revealed a number of advantages over the original confidence calculations. The causal networks also made the knowledge acquisition from domain experts easier as (Armor et al., 1993):
GEMS works in four operation modes, with different data interpretation for each mode. The four operating modes are for different operational states of the turbines:
The system can diagnose problems with stator winding, stator core and frame, rotor winding and body, bearings and lubrication oil, excitation, hydrogen cooling, stator cooling, and oil seals. The system also has an installation advisor that allows the machine operator to configure the expert system to a specific generator by including details of the generator, sensors, and the particular operating policies of the power generation company. For field testing, prototype versions of GEMS were installed on a 500-MW Parsons generator at Ontario Hydro's Nanticoke station and on a 700-MW Westinghouse generator at Nigra Mohawk's Oswego station. During the testing phase, no major problems occurred on any of the machines; however, GEMS did manage to correctly diagnose minor problems at both locations. 3.4.2. MICCA Another EPRI developed expert system is MICCA (Machine Insulation Condition Assessment Advisor) which was designed for use with motor and generator insulation systems. The cost of critical failure of such a system is estimated to be in the region of $250,000. A survey of approximately 7,500 large motors in power generation facilities revealed that 37% of failures were due to stator winding problems and that an additional 10% of failures were due to rotor winding problems. The MICCA system relies heavily on user interaction, with a menu-based interface and an extensive context sensitive help that describes technical terms and explains about 50 tests and inspections. The knowledge base contains over 50 separate databases, including information on each major motor and generator insulation system, the most probable aging mechanism of these systems, and diagnostic and inspection tests for confirming or rejecting the operation of a particular aging mechanism. MICCA uses information from these databases to deduce the most likely aging process occurring within the stator and rotor windings and recommends diagnostic tests to indicate what the operator or engineer should look for in a visual inspection and how to interpret the results of these inspections. The results from the visual inspections are inputted into the MICCA system by the use of multiple choice questions. Results from the visual inspection are used to reinforce conclusions or re-direct diagnostics to another area.
|
![]() |
|
Use of this site is subject certain Terms & Conditions. Copyright (c) 1996-1999 EarthWeb, Inc.. All rights reserved. Reproduction in whole or in part in any form or medium without express written permission of EarthWeb is prohibited. Please read our privacy policy for details. |