Font Size: a A A

Research On System Reliability Optimization Problems Based On Intelligent Optimization Algorithm

Posted on:2018-07-29Degree:DoctorType:Dissertation
Country:ChinaCandidate:Y B LiuFull Text:PDF
GTID:1318330515976120Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
Reliability optimization problems are the complex nonlinear mathematical programming problems in the engineering field,and the traditional optimization techniques cannot obtain the optimal solutions for a limited time.The emergence of intelligent optimization algorithm provides a new effective way to solve this kind of problem.It can get optimal solution in the limited time for the complex optimization problem,and it solved the problem that the traditional optimization technique is difficult to solve.It started a new phase in the field of optimization techniques.At present the research on intelligent optimization algorithm has made many results,but the algorithm itself still has some shortcomings,such as premature convergence,difficult control parameter setting and adjustment,low accuracy of the optimal solutions.The research work of this dissertation focuses on the shortcomings and insufficiency of intelligent optimization algorithm,takes solving the reliability optimization problems as the main target.The improvement research on intelligent optimization algorithm is done to resolve reliability optimization.The main works are:(1)An improved differential evolution algorithm is proposed for the problem of premature convergence in the basic differential evolution algorithm.In the proposed algorithm,the adjustment factor F and the crossover probability CR are modified with the adaptive adjustment parameters.For adjustment factor F,it will be taken a new value Fi on each generation evolution.With the increase of the number of iterations of the algorithm,the value of Fi is more than the average value Fmean of the previous n-1 generation.For the crossover probability CR also uses a similar strategy,the difference is that the CRi value is less than the average value CRmean of the previous n-1 generation with the increase of the number of iterations.At the same time,the selection strategy of the algorithm is also improved.The improved algorithm not only enhances the diversity of the population in the process of iteration to avoid the premature convergence,but also the algorithm has a fast convergence speed in the later stage.The simulation results show that the method gets the better results in solving the problems of series system.Aiming at the balance of global search and local search in the working process of the intelligent optimization algorithm,the differential evolution algorithm combined with Lévy flight is proposed.For searching the solution space,the ideal effect is the global search and local search to achieve a good balance,that is,a large area of global search is done to enhance the diversity of the solution at starting,the latter phase of search should be efficient local search in order to increase the convergence speed of the algorithm.The proposed method is well balanced by the global search and local search.Through the simulation experiments on the series-parallel system reliability optimization problem,the results show that the proposed method has better balance the global search and local search and have achieved good results.(2)An improved bat algorithm is proposed to overcome the shortcomings of slow convergence speed and low accuracy.The global search strategy and local search strategy are improved in the proposed method.Through the simulation experiments on the the complex system reliability optimization problem,the results show that the proposed method has improved convergence speed and solution accuracy significantly.It also has better stability.(3)Internal search algorithm(ISA)is studied to resolve the large-scale system reliability optimization problem.The parameter ? of the ISA algorithm is improved,and the adaptive parameter ISA algorithm is proposed.In the ISA algorithm,the bigger the parameter ? value is,the better the diversity of the solution is.But the smaller the parameter ?,the stronger the local search ability of the algorithm is.The initial value of the parameter ? of ISA algorithm is set to the maximum value in a range,then the value of the parameter ? is gradually smaller and smaller with the increase of the number of iterations,and the parameter ? is changed to the minimum value finally.This can enhance the diversity of solutions and avoid falling into local optimum,as well as improve the convergence speed in the later.The simulation results show that the algorithm gets the better results,and the algorithm on all aspects of performance index is improved.(4)For the problem of low accuracy of the ISA,two stage ISA is proposed.In this algorithm,the working process of ISA is divided into two stages: the first stage the initial optimal solution is obtained by using the basic ISA;the second stage a further concentrated search is done to get improved optimal solution based on the preliminary optimal solution.Experimental results show that the two stage algorithm can greatly improve the accuracy of solution and convergence speed for the large-scale system.And the better results are obtained.(5)In the process of iteration,the quality of the new generated solution has a great influence on precision and convergence speed of the algorithm.The differential strategy has a better ability to produce high quality new solution.With this advantage,the differential strategy is introduced into the ISA to generate new candidate solutions.In the process of the iteration,the mirror search strategy and the differential strategy are used to generate new candidate solutions at the same time,and use the competitive strategy to select the better one as a new candidate solution.Experimental results show that the algorithm has achieved good results in solving the large-scale system reliability optimization problem,and the better optimal solution is acquired.
Keywords/Search Tags:differential evolution algorithm, bat algorithm, interior search algorithm, reliability optimization problem, combination optimization, nonlinear programming
PDF Full Text Request
Related items