Dynamic Adjustment Strategies of Inertia Weight in Particle Swarm Optimization Algorithm

The high search speed and efficiency, and simple algorithm of particle swarm optimization algorithm make it suitable for actual-value processing. Starting from the angle of weight, this paper studies several improved particle swarm optimization algorithms and divides the improvement into three types as linear decreasing weight strategy, self-adaptive weight strategy and random weight strategy. Furthermore, this paper also demonstrates the principles of these three improved algorithms and tests and analyzes the three algorithms with test function. It is suggested by the result of tests that random weight strategy can make the algorithm more stable, linear decreasing weight strategy can improve the effect of optimization, while self-adaptive weight strategy can accelerate the convergence. However, the operation of self-adaptive weight strategy takes obviously more time than that of the other two strategies.