A Fuzzy Logic Methodology for Fault-Tree Analysis in Critical Safety Systems

Conference · Sun Jun 20 04:00:00 EDT 1993 · Transactions of the American Nuclear Society

OSTI ID:5892679

Erbay, A. ^[1]; Ikonomopoulos, A. ^[1]

---

1. University of Tennessee, Knoxville, TN (United States)

A new approach for fault-tree analysis in critical safety systems employing fuzzy sets for information representation is presented in this paper. The methodology is based on the utilization of the extension principle for mapping crisp measurements to various degrees of membership in the fuzzy set of linguistic Truth. Criticality alarm systems are used in miscellaneous nuclear fuel processing, handling, and storage facilities to reduce the risk associated with fissile material operations. Fault-tree methodologies are graphic illustrations of tile failure logic associated with the development of a particular system failure (top event) from basic subcomponent failures (primary events). The term event denotes a dynamic change of state that occurs to system elements, which may include hardware, software, human, or environmental factors. A fault-tree represents a detailed, deductive, analysis that requires extensive system information. The knowledge incorporated in a fault tree can be articulated in logical rules of the form "IF A is true THEN B is true." However, it is well known that this type of syllogism fails to give an answer when the satisfaction of the antecedent clause is only partial. Zadeh suggested a new type of fuzzy conditional inference. This type of syllogism (generalized modus ponens) reads as follows: Premise: A is partially true Implication: IF A is true THEN B is true Conclusion: B is partially-true. In generalized modus ponens, the antecedent is true only to some degree; hence, it is desired to compute the grade to which the consequent is satisfied. Fuzzy sets provide a natural environment for this type of computation because fuzzy variables (e.g., B) can take fuzzy values (e.g., partially-true).

• Research Organization: University of Tennessee, Knoxville, TN (United States)
• Sponsoring Organization: USDOE National Nuclear Security Administration (NNSA), Nuclear Criticality Safety Program (NCSP)
• OSTI ID: 5892679
• Report Number(s): CONF-930601--
• Conference Information: Journal Name: Transactions of the American Nuclear Society Journal Volume: 68
• Country of Publication: United States
• Language: English

Related Subjects

054000* -- Nuclear Fuels-- Health & Safety
11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS
42 ENGINEERING
420203 -- Engineering-- Handling Equipment & Procedures
98 NUCLEAR DISARMAMENT, SAFEGUARDS, AND PHYSICAL PROTECTION
ALARM SYSTEMS
CRITICALITY
ENGINEERING
FAULT TREE ANALYSIS
FUZZY LOGIC
Fault-Tree Analysis
Fuzzy Sets
Information Representation
MATHEMATICAL LOGIC
NUCLEAR FACILITIES
New Approach
Nuclear Criticality Safety Program (NCSP)
SAFETY ANALYSIS
SAFETY ENGINEERING
SYSTEM FAILURE ANALYSIS
SYSTEMS ANALYSIS