An Improved Wolf Pack Algorithm

This paper aims to overcome the shortcomings of original wolf pack algorithm (WPA). The shortcomings of original WPA include fixed scouting directions, scouting step length that cannot be adjusted dynamically, lack of information exchange between each other in artificial wolves and lead wolf does not participate in intelligent behavior. An improved wolf pack algorithm is proposed in this paper. The improved algorithm provides scout wolves variable scouting directions and scouting step length could adjust dynamically based on the Fibonacci Sequence by the improved algorithm. Furthermore, ferocious wolves could exchange information with scout wolves during raiding. Finally, the hunting principle of lead wolf is provided in the improved algorithm. Thus, WPA is improved from four aspects: global optimization, convergence speed, convergence accuracy and robustness. The simulation results show that compared with original WPA, the improved wolf pack algorithm not only increase the global search capability, but also avoids early-maturing of the algorithm. Furthermore, it has strong robustness and significant improvements in the search accuracy and rate of convergence.