Because of case diversity and being difficult to get field knowledge in the case modification process, case modification becomes a key technique in case-based reasoning (CBR). According to the above-mentioned problems, taking advantage of the characteristics of genetic algorithm, genetic algorithm can be applied in the case modification. To find the optimal solution with the least delay possible, the similar cases in the case base are used as the initial population of genetic algorithm; The grey correlation calculation formula is the objective function, which can make the scheme that has the maximum similarity with the design problem continue to participate in the operation; Linear crossover operation and non-uniform mutation operation are chosen to avoid premature convergence of genetic algorithm; At last, take numerical control machine tool spindle as an example, MATLAB program is written to calculate the optimal design scheme according to the above ideas. Case study results show that this algorithm is a scientific and practical method.