Abstract: An improved discrete differential evolution algorithm (PA-DDE) with mechanism of parameter adaptive was proposed,based on the research and analysis of discrete differential evolution algorithm．Firstly,the parameters of continuous domains are adaptively adjusted in the process of evolution to balance the global search and local search,and also to coordinate the contradiction of population diversity and convergence speed．Secondly,the co-evolution processing is guided by the feedback information of discrete encoding of the successful evolutionary individuals on the corresponding discrete domains．Simulation results on the knapsack problem show that the proposed algorithm has good convergence efficiency and stability.

Key words: Discrete differential evolution,Parameter control,Discrete encoding,Co-evolution,Multidimensional knapsack problem