Abstract: The standard particle swarm optimization algorithm（PSO）is widely used in bird foraging research. It is a group intelligence
algorithm, but it has the disadvantages of premature convergence and low convergence precision. To further improve the performance
and extension practicability of the standard particle swarm optimization algorithm is an urgent problem in need to be solved. In
response to the above shortcomings, by simulating the three types of circles that human access information, i. e. "neighborhood circle",
"friend circle" and "media", the present research has improved the standard particle swarm algorithm in two ways. On the one hand,
the information exchange between particles is enhanced through weighted averaging of the particle position of the "neighborhood
circle". On the other hand, by grouping the particles according to the "friend circle", and making full use of the search ability of
"experienced" particles, the algorithm is better improved. In addition, the two-dimensional Girewank function, Rosenbrock function
and Schwefel function are used as test functions to compare the improved algorithm with the standard PSO algorithm, which proves the
superiority of the improved algorithm.

Key words: particle swarm optimization, swarm intelligence algorithm, improvement, information exchange, experience