Birefringence learning: A new global optimization technology model based on birefringence principle in application on artificial bee colony

Optimization technologies play a key role in addressing sophisticated optimization problems. However, with the increasing scale and dimension of problems, the performance of optimization technologies has been greatly challenged. To better address them, a new global optimization model is proposed inspired by the birefringence phenomenon, namely a birefringence learning (BRL) model which simulates the birefringence phenomenon of two refracted beams formed by the incident beam in the nature when the natural light enters some certain mediums. To substantiate the optimization performance of the model, it is applied to the artificial bee colony algorithm (ABC) to enhance its global optimization performance further and make its local optimization (exploitation) and global optimization (exploration) balance to some extent. In ABC, the BRL model as a mutation operator is employed in the phase of the scout bee to decrease the probability of ABC trapping into the local extreme region and then a novel ABC algorithm using the birefringence learning (NABC-BRL) is proposed. Via conducting abundant numerical experiments on some universally known benchmark functions, the experimental results and analysis indicate that it can obtain higher accuracy of solutions and faster convergence on majority of benchmark functions compared with some well-acknowledged algorithms, which also proves the effectiveness of global optimization of the BRL model.