Development of a Method for Quantifying the Reliability of Nuclear Safety-Related Software

[摘要] The work of our project is intended to help introducing digital technologies into nuclear power into nuclear power plant safety related software applications. In our project we utilize a combination of modern software engineering methods: design process discipline and feedback, formal methods, automated computer aided software engineering tools, automatic code generation, and extensive feasible structure flow path testing to improve software quality. The tactics include ensuring that the software structure is kept simple, permitting routine testing during design development, permitting extensive finished product testing in the input data space of most likely service and using test-based Bayesian updating to estimate the probability that a random software input will encounter an error upon execution. From the results obtained the software reliability can be both improved and its value estimated. Hopefully our success in the project's work can aid the transition of the nuclear enterprise into the modern information world. In our work, we have been using the proprietary sample software, the digital Signal Validation Algorithm (SVA), provided by Westinghouse. Also our work is being done with their collaboration. The SVA software is used for selecting the plant instrumentation signal set which is to be used as the input the digital Plant Protection System (PPS). This is the system that automatically decides whether to trip the reactor. In our work, we are using -001 computer assisted software engineering (CASE) tool of Hamilton Technologies Inc. This tool is capable of stating the syntactic structure of a program reflecting its state requirements, logical functions and data structure.

[发布日期] 2003-10-01 [发布机构] Massachusetts Institute of Technology (United States)

[关键词] Probability;Feedback;Computers;Safety;29 Energy Planning, Policy And Economy