A Chaotic Artificial Bee Colony Algorithm Based On A Combination Of Multiple Probability Distributions To Optimize The Traversal Control Strategy

As an important branch of control engineering,optimization algorithms are the main methods for solving different optimization problems.At present,optimization algorithms have been widely used in control fields such as communication,electricity and transportation.With the development of complex optimization problems and artificial intelligence theories,swarm intelligence optimization algorithms came into being.Artificial bee colony(ABC)algorithm is a classic among them.Because of its outstanding optimization performance,it is highly sought after by many researchers.The ABC algorithm is essentially a random search algorithm,which has disadvantages such as low optimization accuracy and weak local refined search ability.Therefore,it is a research hotspot to further adjust the control strategies to enhance search capabilities in the current optimization problem domain.Chaos mapping has excellent pseudo-random characteristics,and is suitable for the optimization process of ABC algorithm.However,due to the uneven distribution of the time series generated by traditional chaotic mapping,the initial honey source generated by it is too concentrated in the local area selection value.This reduces the overall pioneering of the algorithm during the early traversal of the search space.Simultaneously,a single-distributed traversal control strategy is used by the chaotic ABC algorithm in the iterative search stage.It is not conducive to improving the search efficiency of the local fine optimization in the traversal algorithm.The ABC algorithm is based on traversal to optimize the search strategy.This paper proposes an improved idea of improving the efficiency of the traversal search by combining to optimize the initial honey source probability distribution characteristics.In the optimization process,when the feasible solution is in the global pioneering stage,the value is traversed according to the homogenizing distribution.When the feasible solution is in the local refinement search stage,the value is traversed according to the peak distribution.Furthermore,the paper proposes a chaotic ABC algorithm under the control of optimal traversal strategies based on a combination of different probability distributions.Firstly,use the basic principles of probability theory to derive the Logistic mapping shaping function that satisfies the principle of maximum information entropy.And build a complete dynamic system that outputs a homogenizing time series.The entropy spectrum analysis and NIST randomness test are used to verify the distribution and non-periodic characteristics of the generated time series.Secondly,the same initial value is used to generate a time series with homogenizing and peak characteristic distribution through the Logistic homogenization system and traditional mapping respectively.And the time series generates the honey source that participates in the traversal optimization.Throughout the optimization process,two honey sources with different distribution characteristics are combined to participate in iteration.Thirdly,The neighborhood space structure is changed by combining the bee activity law of searching from near to far,from face to point.The neighborhood boundary is no longer randomly selected,but is constructed around the better value selected in each iteration.And combined with the distribution characteristics of honey sources,on the one hand,the algorithm uses honey sources with homogenizing distribution characteristics to spread more evenly across the entire solution space.It can enhance the global pioneering ability and avoid entering the local precocious zone.On the other hand,the use of honey sources with peak distribution characteristics to strengthen the fine mining ability in the neighborhood,which increases the chance and intensity of searching near better values.Finally,through 12 standard test functions and solving the shortest path optimization problem,the comprehensive system performance evaluation is carried out.And comprehensive comparative analysis is carried out with other swarm intelligence optimization algorithms proposed recently.The results show the algorithm in this paper can well take into account the global development search ability and the local fine optimization ability.And it has high reliability and good optimization effect.