Recruit training is the basis of improving combat effectiveness. For scientific evaluation for recruit training scores, 19 indexes of recruit training are selected as evaluation factors, average training scores of 8 classes are selected for fuzzy clustering analysis, and 4-layers classification method is made according to scores. Put forward the optimized discrete Hopfield neural network evaluation model based on quantum genetic algorithm. Then, 4 soldiers’ training scores are used for Hopfield coding evaluation, and the model is verified by example. It provides theoretical basis for enforcing basic unit intellectualization training method.