Dynamic Topology Phased Evolutionary Particle Swarm Optimization Algorithmand Multi-state System Reliability Redundant Allocation

During the two sessions this year,Premier Li Keqiang stressed the need to further implement the "made in China 2025",to achieve the strategic goal of manufacturing power,and quality is the lifeline of building a powerful country,therefore,the quality of technology has become the most important task of the enterprise.The reliability optimization technology is the key technology for enterprises to improve the performance of products,therefore,more and more people attach importance to it.With the development of science and technology,the product structure is more and more complex,The combination of different states leads to the polymorphism of the system,resulting in a lot of problems in the reliability optimization of multi-state systems to be solved.At the same time,the parameters of the components become more and more diversified.The combination of different parameters leads to the exponential growth of the solution space,therefore,the performance requirements of the optimization algorithm are also increasingly stringent.PSO algorithm is a typical swarm intelligence optimization algorithm which has been successfully used in the complex reliability optimization problems,but its own shortcoming of premature convergence reduces the accuracy of the optimization results.Therefore,the paper carries out research on improving particle swarm optimization algorithm and its application in reliability optimization.First,for single force rule can not meet the performance requirements of the population diversity and convergence speed in different periods,phased evolutionary particle swarm optimization algorithm is proposed.Based on phase search strategy the search process is divided into two phases,the rules of the global optimization attractive repulsive force and the adaptive attractive repulsive force are constructed.The performance of the proposed algorithm is tested by optimizing standard test functions,the test results are compared with other improved particle swarm algorithm,the effectiveness of the proposed algorithms are further verified.Secondly,based on the research of static topology,a dynamic topology based on the improved fitness model is designed,which is combined with the evolutionary particleswarm optimization algorithm,a dynamic topological phased evolutionary PSO algorithm is proposed.Three typical static topologies are selected and combined with the phased evolution PSO algorithm,study the changes of population diversity and search performance and analysis the influence of topological structure parameters on the performance of the algorithmand.Combined with the study of static topology,in order to simulate the growth characteristics of real networks,take the evolutionary theory of the fittest in natural selection and groups of organisms which exhibite self-organization,a dynamic population topology structure is studied.Adding new particles to improve particle swarm activity and particle scale increase in the early stage of the structural evolution,the poor particles are removed and the new particles are derived in the later stage of the structural evolution.When algorithm evolution stop,combine the population structure evolution and the evolutionary of algorithm,and compared to the test of the proposed algorithm,verify the effectiveness of the proposed algorithmFinally,for the multi-state system reliability and the interval reliability redundancy allocation of series parallel system,based on the theory of interval analysis and the theory of universal generating function,use the optimization algorithm to design the redundancy allocation scheme.Finally,the system structure with higher reliability and lower cost is obtained.Meanwhile,the ability of solving the practical reliability optimization problems are verified.