Method for realizing DET branch step length optimization based on Bayesian optimization search

Abstract

The invention discloses a method for realizing DET branch step length optimization based on Bayesian optimization search. The method comprises the following steps: S1, constructing a DET model; s2, determining a time parameter of an operation type branch event, performing DET model simulation, and calculating a reactor core failure probability; s3, performing dynamic time warping in combination with data stability classification to obtain a risk failure space range; s4, reconstructing an input and output mapping relation by using the Gaussian process model in the S2; s5, predicting a DET output result of the whole space by using a Gaussian process model; s6, gridding discretization and extreme value search are carried out, and the whole grid space is marked as a dangerous space and a safe space; s7, determining safe and dangerous space boundaries according to the search result, and obtaining a parameter range enveloping the risk boundary as a Bayesian optimization search result; and S8, integrating the risk failure space range in the step S3 and the Bayesian optimization search range in the step S7 to reduce the problem space range, and updating task time iterative calculation until the task time reaches the standard. According to the method disclosed by the invention, the key branch step length can be obtained by automatically optimizing and searching the discrete step length of the running type branch in the DET, repeated branches with similar accident simulation results under the equal-step-length branch rule are reduced, and the DET calculation efficiency is improved.

Images