Brought to you by EarthWeb
IT Library Logo

Click Here!
Click Here!

Search the site:
 
EXPERT SEARCH -----
Programming Languages
Databases
Security
Web Services
Network Services
Middleware
Components
Operating Systems
User Interfaces
Groupware & Collaboration
Content Management
Productivity Applications
Hardware
Fun & Games

EarthWeb Direct EarthWeb Direct Fatbrain Auctions Support Source Answers

EarthWeb sites
Crossnodes
Datamation
Developer.com
DICE
EarthWeb.com
EarthWeb Direct
ERP Hub
Gamelan
GoCertify.com
HTMLGoodies
Intranet Journal
IT Knowledge
IT Library
JavaGoodies
JARS
JavaScripts.com
open source IT
RoadCoders
Y2K Info

Previous Table of Contents Next


3.1. THE ENTIRE GENERATION PROCESS

3.1.1. KNOBOS

One real-time, predictive online expert system, developed in 1993 by Siemens of Germany to monitor and diagnose the overall power generation process was the Knowledge-Based Operator System (KNOBOS) (Lausterer et al., 1993). KNOBOS uses monitoring and diagnosis subsystems to examine instrumentation signals from machinery and uses these to determine if a deviation from a predetermined reference pattern is present. If a deviation does occur, the system will examine a set of predetermined variables and, if these exceed a certain threshold, it will then swap into a diagnostic state. If faults are identified, the system selectively informs the system operator of problems. This is done intelligently by the expert system so that false alarms are not triggered by small deviations from the norm.

When a problem is identified, KNOBOS informs the machine operator by displaying a message on the message board and highlighting the plant diagram, with the problem area shown in red. The system also provides suggestions about identified problems such as how to rectify it or prevent it from escalating into a critical problem. Along with the identified problems, and suggestions, KNOBOS also gives a detailed explanation of how it reached its conclusions.

KNOBOS was developed by knowledge engineers working with domain experts. After each knowledge acquisition session, the notes were transcribed and given to the domain expert for verification. This method proved to be time-consuming and error-prone, as errors and ambiguities were found within the knowledge. The KNOBOS system provides a built-in graphical knowledge acquisition tool that enables plant operators to update or change the system's knowledge base and plant structure models as time goes by.

KNOBOS represents knowledge by describing the entire plant process as a hierarchy with faults assigned to specific components. A structural model is used to model the plant, process components, and process media with a fault model representing the world of faults and symptoms.

3.1.2. VIAD

VIAD is a VIbration ADvisor expert system developed by Scheibel et al. in 1989. Initial versions of VIAD worked offline, with later versions being fully automatic. VIAD's knowledge base consists of over 300 rules, 50% of which are generic. The system can use over 200 symptoms to identify over 50 possible causes of vibration problems and give a conclusion along with a confidence factor. The VIAD system can be used with various plant equipment, including centrifugal pumps, fans, steam turbines, centrifugal compressors, electric motors, and gearboxes.

The initial industry response to VIAD was favorable, as the system knowledge was good for examining vibration data. However, some users felt that VIAD might exceed the data analysis limitations of the average user. The developers decided to re-implement VIAD as a multilevel expert system, where the first level would examine an initial set of data for possible faults and the next two levels (level two and level three) would be used to give a more detailed diagnosis. The first level, level 1, is completely automatic with level 2 and level 3 requiring user interaction (see Figure 1).


LEVEL FUNCTION
LEVEL 1 Requires information about the type of machine it is monitoring and the spectral data associated with that machine.
  From this information, it attempts to reach conclusions about possible causes of vibration problems.
LEVEL 2 Requires a dialog with the operator. Asks questions about recent maintenance work and current performance.
  New data are used to reinforce conclusions reached in level 1 or to redirect the user to other possible equipment faults.
LEVEL 3 Interactively recommends and considers additional information, such as types of measurements, that are not normally taken during the usual monitoring process.

FIGURE 1 The VIAD system.

A later version of VIAD was released by Entek Corporation as EXPLORE-EX and is still an important expert system application (Entek, 1992).

3.1.3. ADVISOR

The ADVISOR system (Bruel and Kjaer, 1996) also uses expert system technology to perform automatic machine diagnosis. ADVISOR is an add-on package for COMPASS (COMputerized Prediction Analysis and Safety System). COMPASS is designed to monitor any process-related parameters and display an alarm if the signal exceeds a given threshold. However, this is only part of the automation process, as COMPASS does not identify what the problem is. ADVISOR runs in the background and performs a diagnosis with the available data, with a list of possible faults displayed on-screen, ranked in order by a weighting or likelihood value. The system also supplies a list of explanations that go with the identified faults. This also helps novice engineers gain more experience, thus acting as an intelligent tutor.

ADVISOR's knowledge base started out as general rules for vibration analysis, supplied by Bruel and Kjaer's own team of expert machine analysts. The knowledge base, when implemented at a site, can be tailored for a specific machine by taking into account the operator's knowledge of the machine and his/her experiences with any special cases. The system's basic rules can also be modified by the machine operator if they are insufficient or over-reactive in certain situations.

The effectiveness of ADVISOR was shown in a case study (Bruel and Kjaer, 1996):

"After an overhaul of a single stage turbine with a centrifugal compressor and inward-flow turbine, previously absent vibration levels became apparent at 128Hz and 192Hz."

The ADVISOR system examined the vibration data, looking at the vibration component of the middle shaft output gearwheel. This component showed an increase relative to the reference, thus indicating a misalignment of the middle shaft of the gearbox. Trending of the vibration data before and after the overhaul of the turbine showed that the increase in vibration started immediately after the start-up of the turbine. The tooth-meshing vibration frequency on the output shaft gearwheel also showed a slight increase in amplitude. This aided in confirming the fault to be a misalignment between the output and middle shaft, which caused a sliding toothmesh instead of a rolling toothmesh (Bruel and Kjaer, 1996).


Previous Table of Contents Next

footer nav
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.