An artificial intelligence approach to accelerator control systems
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.
- Research Organization:
- Los Alamos National Laboratory (LANL), Los Alamos, NM (United States)
- DOE Contract Number:
- W-7405-ENG-36
- OSTI ID:
- 6625116
- Report Number(s):
- LA-UR-87-739; CONF-870260-2; ON: DE87007472
- 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
Similar Records
IEA/AIE-90; Proceedings of the 3rd International Conference on Industrial and Engineering Applications of Artificial Intelligence and Expert Systems, Charleston, SC, July 15-18, 1990. Vols. 1 and 2
A laboratory object-based expert system: The LOBES Project
Related Subjects
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)