Improved Ant Lion Optimizer and its application in modeling of Czochralski crystal growth

In order to overcome the shortcomings of low convergence speed and easily trapping in local optima in Ant Lion Optimizer (ALO), an improved ALO algorithm with Levy mutation and adaptive elite competition mechanism is proposed in this paper. Random number obeying Levy distribution is used to mutate poor individuals in the population, which can improve the diversity of population to increase global search ability of the algorithm. Moreover, adaptive elite competition mechanism that many elites lead the population to search simultaneously can expedite the convergence speed. To reduce the amount of calculation, the number of elites will decrease with the increase of iterations. Compared with the other two optimization algorithms (ALO, Particle Swarm Optimization (PSO)), test results show that convergence speed and optimization precision of improved ALO algorithm is better. Finally, the improved ALO algorithm is applied to the identification of thermal field temperature model in Czochralski silicon single crystal growth and experimental results demonstrate the efficiency of the proposed algorithm.