Research And Application Of Bald Eagle Search Optimization Algorithm

Optimization aims at the decision space to search the value of variables according to a certain method,so that the objective function can obtain the optimal value.It is always the goal of optimization researchers to design a reasonable,fast and easy to execute search strategy.The bald eagle search optimization(BES)algorithm,which simulates the survival hunting behaviour of bald eagles,is a stochastic optimization algorithm with fast convergence and high optimization capability,but suffers from a lack of population diversity,a tendency to collapse into local optimums,and difficulty in balancing exploration and exploitation performance.To this end this paper proposes three improved BES algorithms and uses them for feature selection and engineering optimization,PV model parameter estimation,instance constrained optimization problems and path planning for mobile robots respectively,to improve the performance of the BES algorithm while promoting research into the application of swarm intelligence algorithms.The details of the research are as follows:(1)A bald eagle search optimization algorithm incorporating a spherical random shrinkage mechanism is proposed(INMBES).Taking advantage of the fact that for a certain surface area the range contained by the ball is the largest and the principle that spherical coordinates have more accurate ranging,the BES parameters are updated using a spherical coordinate transformation to expand the range of the bald eagle population searching for prey while improving the accuracy of the bald eagle targeting prey,improving the global exploration capability of the algorithm and the convergence speed of the algorithm.In the prey capture phase of the bald eagle,the populations are sorted and grouped,and the group of elite individuals is updated using the elite chaos strategy to enhance the exploitation performance of the algorithm;the group of non-elite individuals is grouped again,using the simplex strategy,which makes the vultures learn from the best individuals,thus correcting the worst individuals in each group,balancing the exploration and exploitation performance of the algorithm.Experimental results from 30 test functions in the CEC2017 test set show that the INMBES algorithm outperforms nine competitive comparison algorithms;the effectiveness of the INMBES algorithm is verified by using it on solving three engineering constrained optimization problems for vehicle side impact design,I-shaped beam design and photovoltaic design,as well as the feature selection problem.(2)An adaptive bald eagle search optimization algorithm with global information enhancement is proposed(VABES).The VABES algorithm uses the energy of the whole population information to construct virtual particles,and increases the strategies of learning from virtual particles and the flight pattern of the annular neighbor,so as to improve the diversity of the population.New adaptive probabilistic selection schemes are incorporated to induce appropriate transitions between aggregation and dispersion behaviour of the populations,new variable-step variation and crossover strategies are designed to enhance the exploitation performance of the algorithm,improving the ability of the algorithm to find optimal solutions.Experimental results from a single-strategy comparison at CEC2020 and a high-dimensional comparison at CEC2017 show that the VABES algorithm has good outcome-seeking performance.VABES is compared with several other representative algorithms on three PV model parameter optimization problems while it is used to solve 57 real-world problems.The results show that VABES has better real-world optimization-seeking capability compared to other algorithms.(3)A bald eagle search optimization algorithm combining heterogeneous comprehensive learning and complete information search mechanism is proposed(HFSBES).In order to enhance the diversity of the population,the λ-heterogeneous integrated learning strategy was used to select the sample group and sample points of each individual,and the better quality particles were selected in preference to enhance the global exploration performance of the population.The time-varying parameters were designed to expand the exploration range of the population in the early stage of iteration,and to enhance the ability of local development of the population in the late stage of iteration,which balanced the performance of global exploration and local development of the algorithm.The perfect information search mechanism for the optimal individual increases the probability of higher quality solutions.The experimental results show that HFSBES has a strong ability to solve optimization problems compared with other comparison algorithms.The results show that HFSBES has a strong ability to solve optimization problems in the test sets CEC2019,CEC2020 and CEC2021.Applying HFSBES algorithm to CEC2020 real constraint optimization problem and robot path planning problem in different environments further verifies its superior actual optimization performance.