Comparison of calculated and experimental power distributions in a reactor on the basis of mathematical statistcs methods

Optimum strategy of statistical estimation permitting to compare calculated and experimental distributions of energy release with a good degree of confidence is suggested. At initial stage of investigation the correlation-regression analysis is used. Application of orthogonal polynomials in regression equation permits to estimate separately the contribution of every term. Selection of polynomial regression order and estimation of accepted model parameters are carried out using optimum statistical criteria. A priory hypothesis on normal law of error distribution is verified on the basis of analysis of standardized residual errors for the adequate model. As an illustration the results of calculating energy release distribution in fuel elements of the model critical assembly with regular lattice are presented. 16 refs.; 2 figs.; 2 tabs