Studies on fast neutron multiplicity measurement based on neural network

In the measurement of fast neutron multiplicity, multiplicity counting rates neutrons, including singles, doubles, and triplets, are often substituted into equation to solve quality problems. To simplify solution process directly obtain sample through S, D, T, a neural network multivariate nonlinear fitting used for analysis. First, multiple sets data measured detection system built with Geant4. After training back propagation network, corresponding relationship between m is established. It verified that there different degrees discrepancy predicted values simulated theoretical values. improve accuracy predictions, genetic algorithm optimization M coefficient correction introduced. analyze stability model, 10% error perturbation introduced T. The double rate has greatest influence on deviation value, indicating key parameter in analysis multiplicity. On this basis, functional obtained fitting, validation by simplifying technology equation.