Application Research Of Teaching And Learning Enhanced Whale Optimization Algorithm

Whale Optimization Algorithm(WOA)is a heuristic optimization algorithm to simulate the foraging behavior of humpback whales in the ocean.It has the characteristics of simple structure and powerful search ability.With the further study of the WOA algorithm,researchers found that the algorithm had problems such as low precision,slow convergence speed,and easy immature.In this paper,the shortcomings of WOA are analyzed and improved.At the same time,the improved algorithm is applied to practical optimization problems,aiming to further improve the WOA algorithm theory and expand the range of applications.The main work of this paper has the following three aspects:(1)To enhance the exploitation ability of the basic WOA and improve the convergence accuracy of the algorithm,inspired by the teaching-learning-based strategy,TWOA was formed by combining the teaching phase with WOA algorithm,which improved the quality of the algorithm candidate solutions.For the algorithm tends to fall into local optimization in the later stage,the simplex method is introduced into TWOA algorithm as a random variation strategy to form teaching and learning enhanced whale optimization algorithm(TSWOA)which enhance the diversity of the population,improves the exploration ability of the algorithm and effectively avoids the algorithm falling into local optimization too early.The experimental results show the TSWOA has an excellent performance of fast convergence speed and high precision.(2)The TSWOA algorithm was applied to optimize the structure of neural networks.Based on the standard UCI dataset,7 different datasets were selected for training.The experimental results show that compared with other meta-heuristic algorithms,the neural network optimized by the TSWOA algorithm has higher classification accuracy and convergence accuracy.(3)The traditional optimization algorithm depends on the initial value,easily falls into the local convergence,and the WOA algorithm has a poor calibration effect.Based on the principle of Levenberg-Marquardt(LM)algorithm,a hybrid WOA-LM algorithm is proposed,which makes full use of the global search ability of the WOA algorithm and the local optimization ability of LM algorithm to achieve high precision in star sensor parameter calibration process.