In order to improve the searching performance(global optimal, computational cost, learning ability etc.) of wolf pack algorithm in solving complex functions, a novel algorithm---Smart wolf pack algorithm(SWPA) is proposed. First of all, intelligent hunting is proposed to improve the adaptive learning ability and reduce the computational cost. Then the bimodal Gaussian regeneration method is applied to enhance the global searching ability. The, Markov process is used to prove the convergence of SWPA. Finally, compared with some typical evolutionary algorithms, simulation on several Benchmark functions is analyzed. Results show that the SWPA has excellent searching performance on the global optimization ability, the convergence rate and the optimal precision.