A METHOD FOR PREDICTING DIAMETRAL CREEP OF CANDU REACTOR PRESSURE TUBES

摘要

PURPOSE: A method for estimating the diametric creep of a heavy-water reactor pressure tube is provided to reduce a diametric creep estimation error by applying FNN(Fuzzy Neural Network) based on creep data. CONSTITUTION: Creep data is classified into training data and verification data(S1). An initial gene including a membership function variable is arbitrarily generated(S2). A fitness function of gene algorithm is calculated(S3). It is determined whether the fitness function is larger than a preset value(S4). Variables are changed by gene manipulation like selection, crossing, and deformation(S5). A membership function is generated based on a subtractive clustering method(S6). A consequent parameter of a fuzzy regulation function is obtained by using the least square method(S7). An FNN creep estimation model is corrected by using the membership function and the consequent parameter(S8). The uncertainty of the FNN creep estimation model is evaluated(S9). The final FNN creep estimation model is calculated(S10).