Abstract: In order to enhance the performance of FA, this paper firstly analysed the two deceptive directions in objective function, then accordingly proposed an enhanced version of FA named Dynamic Fireworks Algorithm based on population evolution velocity(DFA).The parameter z in DFA changes dynamicly according to the population evolution velocity.DFA only changed parameter z's updating strategy without changing the structure of FA, thus the improved algorithm based on FA can be transplanted to DFA conveniently. Experimental evaluation on nine benchmark functions shows that DFA outperforms conventional FA in global solution accuracy capabilities.