The invention discloses a hypersphere search algorithm based on differential evolution. An initial population number Npop, the number of hypersphere centers NSC; the upper and lower limits rmin and rmax of the hypersphere searching radius, a hypersphere motion angle change probability Prangle, a differential evolution crossover probability MR and a differential evolution mutation probability CR are defined, the algorithm comprise six steps of algorithm initialization, result optimal movement, differential evolution, virtual particle recovery, hypersphere center re-determination and judgment convergence, the six steps are closely related, the overall algorithm design is simplified, the process is clear, and the practicability is high. According to the algorithm, under the condition that theoptimal solution is not lost, the global search capability is improved, the accuracy of the obtained solution is enhanced, and a new optimization technology is provided for the field of rail transit.