Research On Multi-objective Optimization Algorithm And Application Based On Membrane Particle Swarm

In recent years,some new evolutionary paradigms(such as swarm intelligence optimization,artificial immune system,distribution estimation algorithms,and co-evolutionary algorithms,etc.)have been introduced into the field of multi-objective optimization,becoming effective ways to solve multi-objective optimization problems.As a new branch of natural computing,membrane computing is a computational model inspired by the structure and function of living cells and organs or tissues composed of living cells.The membrane algorithm is an evolutionary computation method implemented using a membrane computing framework with distributed parallel features,and has achieved better performance in the application of optimization theory.As one of the typical optimization algorithms,particle swarm optimization algorithm updates the direction of particles through the speed and position formulas.Because of its advantages such as easy to understand and implement,it has received extensive attention in the fields of science and engineering,but its search strategy is relatively simple,making the algorithm It is difficult to obtain the Pareto front and easily fall into local optimum and infinite iteration.There are still some shortcomings in global search and convergence.Therefore,this paper combines the membrane algorithm with the particle swarm algorithm,and proposes a multi-object particle swarm optimization algorithm based on the framework of P system(PMOPSO).On the one hand,the particle swarm algorithm is used to implement a search strategy to update the local optimal solution in parallel in the basic film.The optimal solution in each basic film is transferred to the surface film through evolutionary rules such as the splitting,rewriting,communication,and dissolution of the P system.,Choose the global optimal solution set to improve the algorithm's convergence speed.On the other hand,in the surface layer,the non-dominated solution set and crowding distance mechanism in NSGA-II are used to sort the non-dominated solution set in descending order by using external files,which makes up for the shortcomings of the particle swarm algorithm's inefficient solution.While improving the fast convergence of the algorithm,it can also prevent the algorithm from falling into a local optimum,effectively balancing the global search and local optimization capabilities of the algorithm,and bringing the solution set closer to the true Pareto front.By selecting standard test functions of different dimensions for simulation experiments,the effectiveness of the proposed PMOPSO algorithm is verified,and compared with other multi-objective algorithms,it has better fast convergence and diversity of solution sets.Based on this,combined with the essence of radar source signal sorting,a multi-objective optimization algorithm of membrane particle swarm optimization is proposed to transform signal clustering and sorting into a multi-objective optimization problem to solve.The symbolized feature vector of the radar radiation source signal is used as the data set to be sorted.At the same time,the spectral time series of the radar radiation source signal is converted into a discrete symbol sequence,thereby capturing the signal characteristics in a wide range.The objective function is constructed by using the intra-class compactness and inter-class connectivity indicators in the MOCK cluster to ensure that the selected feature vector has good intra-class aggregation and inter-class separability,and automatically determines the number of classes at the same time.Finally,by computing the Pareto optimal solution set of the radar source signal symbol entropy feature data set,the multi-object clustering sorting of signals is achieved.Simulation results show that the algorithm achieves a higher accuracy rate of radar source signal sorting and recognition,verifying the effectiveness and feasibility of the algorithm,and its performance is superior to traditional clustering methods.