An artificial intelligence approach to accelerator control systems
Abstract
An experiment was recently started at LAMPF to evaluate the power and limitations of using artificial intelligence techniques to solve problems in accelerator control and operation. A knowledge base was developed to describe the characteristics and the relationships of the first 30 devices in the LAMPF H+ beam line. Each device was categorized and pertinent attributes for each category defined. Specific values were assigned in the knowledge base to represent each actual device. Relationships between devices are modeled using the artificial intelligence techniques of rules, active values, and object-oriented methods. This symbolic model, built using the Knowledge Engineering Environment (KEE) system, provides a framework for analyzing faults, tutoring trainee operators, and offering suggestions to assist in beam tuning. Based on information provided by the domain expert responsible for tuning this portion of the beam line, additional rules were written to describe how he tunes, how he analyzes what is actually happening, and how he deals with failures. Initial results have shown that artificial intelligence techniques can be a useful adjunct to traditional methods of numerical simulation. Successful and efficient operation of future accelerators may depend on the proper merging of symbolic reasoning and conventional numerical control algorithms.
- Authors:
- Publication Date:
- Research Org.:
- Los Alamos National Laboratory (LANL), Los Alamos, NM (United States)
- OSTI Identifier:
- 6625116
- Report Number(s):
- LA-UR-87-739; CONF-870260-2
ON: DE87007472
- DOE Contract Number:
- W-7405-ENG-36
- Resource Type:
- Conference
- Resource Relation:
- Conference: International workshop on hadron facility technology, Santa Fe, NM, USA, 2 Feb 1987; Other Information: Portions of this document are illegible in microfiche products
- Country of Publication:
- United States
- Language:
- English
- Subject:
- 43 PARTICLE ACCELERATORS; 99 GENERAL AND MISCELLANEOUS//MATHEMATICS, COMPUTING, AND INFORMATION SCIENCE; ACCELERATORS; ARTIFICIAL INTELLIGENCE; USES; KNOWLEDGE BASE; PROTON BEAMS; TUNING; BEAMS; NUCLEON BEAMS; PARTICLE BEAMS; 430300* - Particle Accelerators- Auxiliaries & Components; 990210 - Supercomputers- (1987-1989)
Citation Formats
Schultz, D E, Hurd, J W, and Brown, S K. An artificial intelligence approach to accelerator control systems. United States: N. p., 1987.
Web.
Schultz, D E, Hurd, J W, & Brown, S K. An artificial intelligence approach to accelerator control systems. United States.
Schultz, D E, Hurd, J W, and Brown, S K. 1987.
"An artificial intelligence approach to accelerator control systems". United States. https://www.osti.gov/servlets/purl/6625116.
@article{osti_6625116,
title = {An artificial intelligence approach to accelerator control systems},
author = {Schultz, D E and Hurd, J W and Brown, S K},
abstractNote = {An experiment was recently started at LAMPF to evaluate the power and limitations of using artificial intelligence techniques to solve problems in accelerator control and operation. A knowledge base was developed to describe the characteristics and the relationships of the first 30 devices in the LAMPF H+ beam line. Each device was categorized and pertinent attributes for each category defined. Specific values were assigned in the knowledge base to represent each actual device. Relationships between devices are modeled using the artificial intelligence techniques of rules, active values, and object-oriented methods. This symbolic model, built using the Knowledge Engineering Environment (KEE) system, provides a framework for analyzing faults, tutoring trainee operators, and offering suggestions to assist in beam tuning. Based on information provided by the domain expert responsible for tuning this portion of the beam line, additional rules were written to describe how he tunes, how he analyzes what is actually happening, and how he deals with failures. Initial results have shown that artificial intelligence techniques can be a useful adjunct to traditional methods of numerical simulation. Successful and efficient operation of future accelerators may depend on the proper merging of symbolic reasoning and conventional numerical control algorithms.},
doi = {},
url = {https://www.osti.gov/biblio/6625116},
journal = {},
number = ,
volume = ,
place = {United States},
year = {Thu Jan 01 00:00:00 EST 1987},
month = {Thu Jan 01 00:00:00 EST 1987}
}