Research On The Dynamic Reliability Allocation Of Complex Software Systems Based On Differential Evolution Algorithm

In this rapidly developed informationlized society,we can say software everywhere,it affects all aspects of society in every regard,they bring a lot of convenience while leaving many hidden dangers at the same time.Unreliable software is the root of most hidden trouble.Hence the research of software reliability optimization becomes a focus.Considering the existing study of software reliability allocation problem,the discussion of the software system whose structure was fixed,ignoring the actual situation that the structure of the software system are often not fixed,therefore in this paper,we proposed a dynamic reliability allocation optimization model of complex software systems,in this model software system structure is variational,and then research is carried out around this model.The main work of the thesis is as follows:1.In view of the existing complex software system reliability allocation model cannot simulate the fact that most software in the real world is constantly updated,a dynamic reliability allocation model of complex software system is proposed,and points out that the key of dynamic model lies in the history of reservations.The structure of the complex software system in the dynamic reliability allocation model changes over time,which simulates the process of software update and shortens the distance between theory and practice.2.In view of the illegal code may appear,when using differential evolution algorithm(DE)for resolving the model,a coding correction strategy based on weight sorting is proposed.It can effectively improve the efficiency and effectiveness of the differential evolution algorithm.3.Discussesing the strategy of reserve historical solution of dynamic reliability allocation model of complex software system,in order to shorten the distance between initial population and the target population of differential evolution algorithm with Euclidean distance as the goal,three strategies for preserving historical solutions to the solution of the model is proposed:promote the evolution of the DE history population level;the reasonable choice of historical preservation solution;sorting the DE history population according to the objective function.Finally,the simulation results show that these three strategies can effectively improve the results of the model.