Research Of Swarm Intelligence Algorithm Based Particle Swarm Optimization And Chicken Swarm Optimization

In a large number of swarm intelligence optimization algorithms,particle swarm optimization and chicken swarm optimization algorithms are widely used to solve these problems because of their simple concept,easy realization,fast convergence and so on.Particle swarm optimization and chicken swarm optimization algorithms have their advantages,but they still have shortcomings.In order to effectively improve the performance of particle swarm optimization and chicken swarm optimization algorithm,the algorithm is studied from the following angles.1.The basic particle swarm optimization algorithm and the chicken swarm optimization algorithm are described and their characteristics are analyzed.The particle swarm optimization algorithm,the chicken swarm optimization algorithm and the krill swarm optimization algorithm are compared.and the advan-tages and disadvantages of the particle swarm optimization algorithm and the chicken swarm optimization algorithm are analyzed.Then,we propose several particle swarm optimization algorithms and an improved strategy of chicken swarm optimization algorithm.2.Because the convergence speed of particle swarm optimization algorithm is too fast,the diversity of the algorithm is insufficient in the later period,and it often falls into the local optimal solution,which makes the search of the algorithm stagnant.So three strategies are introduced to improve the particle swarm optimization algorithm.(1)The stochastic perturbation term is introduced to improve the speed updating formula of particle swarm optimization,and the best particle fast optimization is guided by the Newton descent direction.A modified particle swarm optimization algorithm with an adaptive disturbance with Newton direction is proposed.This algorithm can make the trapped particles jump out of the local trap and thus realize the whole problem.The global optimization is realized.The convergence characteristics of the algorithm are proved theoretically.The numerical experiment shows that the proposed algorithm has better global optimization ability and higher computing precision.(2)In this paper,the clone strategy and local optimization method are introduced,and a particle swarm optimization algorithm with cloned acceleration strategy is proposed.The algorithm uses particle swarm optimization algorithm for global optimization,cloning strategy expands global optimization range and local optimization algorithm accel-erates convergence.Numerical experiments show that the algorithm has fast convergence speed and high accuracy.(3)The speed and position updating formula is improved by fuzzy reasoning to enhance the global searching ability of particles,and then the central particle and crossover operator are introduced to enhance the information sharing among particles,thus the local search capability is increased.Finally,the particle is disturbed by Logistic mapping.A particle swarm optimization algorithm based on fuzzy reasoning is proposed.The algorithm can quickly converge to the global optimal solution,and has high global optimization ability and computational accuracy.(4)Based on the analysis of the model of particle swarm optimization(PSO),the Markov chain model is established,and some properties of the Markov chain are analyzed.It is proved that the particle swarm state sequence is a finite homogeneous Markov chain.Combining the convergence criteria of the random algorithm,it is proved that the particle swar-m optimization algorithm can satisfy the 2 criteria of global convergence of the random algorithm.The global convergence of the particle swarm optimization algorithm is guaranteed.3.In the process of convergence,chicken swarm optimization algorithm often falls into the local op-timal solution when the chicken follows the hen,and for the high dimensional complex problems,the convergence accuracy is not enough.In this paper,several improvement strategies are proposed in this paper.(1)The updating formula of hen and chicken is improved by fuzzy reasoning,and the parameters are adjusted adaptively.The diversity of the later period is enhanced with Tent disturbance.The numerical experiment shows that the convergence accuracy and convergence speed of the algorithm are obviously improved.(2)Two kinds of differential mutation operators are introduced,and an improved chicken swar-m optimization algorithm is proposed.There are two populations and two levels in the algorithm,which are the bottom and top layers respectively.The top layer is made up of the individual best points,and the bottom layer is made up of all the particles,and the top particles are distributed to each sub group.The particle has two different control parameters and two different control differential mutation algorithms.Therefore,a sub group has good search ability,and the other sub group has good exploration ability.Obviously,when the top particles guide the underlying particles,the underlying particles have good ex-ploration and search capability at the same time.Under this structure,the double layer chicken swarm optimization algorithm has global search ability and search efficiency.Numerical experiments show that it has good convergence speed and convergence accuracy.(3)The chicken swarm optimization algorithm and artificial swarm algorithm are all new optimization methods.When solving high dimensional com-plex optimization problems,the CSO algorithm may fall into the local solution because of its low global search efficiency.The ABC algorithm lacks strong local search ability and makes the convergence rate slow.Therefore,the local search phase of CSO and the global ABC are global.In the search phase,the CS-ABC algorithm is proposed.In the iteration,the algorithm adjusts the particle according to the pbest of each particle.The numerical experiment proves that CS-ABC is an effective and fast conver-gent method.(4)In the chicken group optimization algorithm,the iterative formula of the cock,hen and chicken is simplified,the characteristics of the iterative formula are analyzed.The convergence analysis of the chicken group optimization algorithm is carried out,and the range of parameters in the algorithm is corrected.Finally,the inertia weight is introduced to the iterative formula in the chicken swarm optimiza-tion algorithm,and the convergence analysis is used.The value of inertia weight is obtained.Numerical experiments also verify the global convergence of the algorithm.(5)Because the chicken is easy to get into local solution in the chicken swarm optimization algorithm,the optimization effect of the algorithm is not ideal,so the chicken is analyzed and the finite information is introduced into the iterative formula of the chicken,so that the chicken can learn from the chickens around the chicken,so that more useful information is passed to the chicken,so that the chicken can better find the food.The chaos mapping is used to disturb the chicken group optimization algorithm,and the diversity of the algorithm is guaranteed.The numerical experiment shows that the algorithm has better convergence precision and convergence speed.(6)The chicken group optimization algorithm is inspired by the chicken group searching for food and grade system.It is often used to solve the nonlinear global optimization problem.Since proposed,various improved algorithms have come out in endlessly.However,the second term of the hen's updating formula in the chicken group is the difference between the rooster and the hen's individual in the hen group,third For the difference between the random individual and the hen individual in the rooster group,this subtraction mechanism makes the algorithm fall into premature or stagnant in the iterative process.This paper proposes a normal update mechanism of chicken group optimization algorithm to overcome this problem,and then uses 13 test functions to verify the feasibility of the algorithm.4.In the multi-objective particle swarm optimization algorithm,in order to guide the particles to converge to the Pareto front end and maintain diversity,first,the points of the non inferior concentration are mapped to the parallel grid coordinate system,and the Pareto entropy and the differential entropy are introduced to evaluate the status of the particles,and the parameters of the algorithm are improved by the evaluation results,and then the clonal immune operation is used.The external file is maintained and chaotic map is used to perturb the algorithm.Numerical experiments show that the algorithm has good convergence accuracy and diversity.5.The rooster updating formula of the chicken swarm optimization algorithm is improved.The chick-en group optimization algorithm is extended from the single target algorithm to the multi target algorithm.The algorithm is improved with the non dominant domain.The numerical experiment shows that the multi target chicken group algorithm has better convergence precision and diversity than the multi-target particle swarm optimization algorithm.6.The power system is a complex large system.In the background of increasing demand for many quality indicators such as security,reliability,economy,quality and low carbon,the multi-objective par-ticle swarm optimization algorithm is applied to the distribution network.A model is proposed,which is both cheap and environmentally friendly.A non dominated neighborhood particle swarm optimization al-gorithm is used to solve the problem.The improved multi-objective chicken group optimization algorithmis applied to the charging and discharging application of electric vehicle,and the mathematical model is set up.The penalty function is used to transform the constrained optimization model into an unconstrained optimization model,and then the improved algorithm is used to solve the problem.