Implicit Kalman filter algorithm for nuclear reactor analysis

Conference · Wed Jan 01 00:00:00 EST 1986 · Trans. Am. Nucl. Soc.; (United States)

OSTI ID:6839449

Hassberger, J A; Lee, J C

Artificial intelligence (AI) is currently the hot topic in nuclear power plant diagnostics and control. Recently, researchers have considered the use of simulation as knowledge in which faster than real-time best-estimate simulations based on first principles are tightly coupled with AI systems for analyzing power plant transients on-line. On-line simulations can be improved through a Kalman filter, a mathematical technique for obtaining the optimal estimate of a system state given the information contained in the equations of system dynamics and measurements made on the system. Filtering can be used to systemically adjust parameters of a low-order simulation model to obtain reasonable agreement between the model and actual plant dynamics. The authors present here a general Kalman filtering algorithm that derives its information of system dynamics implicitly and naturally from the discrete time step-series of state estimates available from a simulation program. Previous research has demonstrated that models adjusted on past data can be coupled with an intelligent controller to predict the future time-course of plant transients.

• Research Organization: Univ. of Michigan, Ann Arbor
• OSTI ID: 6839449
• Report Number(s): CONF-861102-
• Journal Information: Trans. Am. Nucl. Soc.; (United States), Vol. 53; Conference: American Nuclear Society and Atomic Industrial Forum joint meeting, Washington, DC, USA, 16 Nov 1986
• Country of Publication: United States
• Language: English

Related Subjects

22 GENERAL STUDIES OF NUCLEAR REACTORS
NUCLEAR POWER PLANTS
REACTOR CONTROL SYSTEMS
REACTOR KINETICS
ARTIFICIAL INTELLIGENCE
COMPUTERIZED SIMULATION
ALGORITHMS
DIFFERENTIAL EQUATIONS
ON-LINE CONTROL SYSTEMS
TRANSIENTS
CONTROL SYSTEMS
EQUATIONS
KINETICS
MATHEMATICAL LOGIC
NUCLEAR FACILITIES
ON-LINE SYSTEMS
POWER PLANTS
SIMULATION
THERMAL POWER PLANTS
220400* - Nuclear Reactor Technology- Control Systems
220100 - Nuclear Reactor Technology- Theory & Calculation