Software Maintenance Policy Optimization Model Design And Analysis Based On ACO

With widespread use of software in almost every walk of life, the softwaremaintenance cost is rising annually, and its significant influence has received attentionfrom increasing number of software practitioners. Numerous researchers also try toconstruct some models and employ several methods to maintain the software and save thesoftware maintenance cost better. This thesis, taking Ertan information platform asbackground, conducts some research about planning optimal software maintenance policyon information platform.In this thesis, an ACO based software maintenance policy model is built. The modelis aim to optimize both the software maintenance cost and software reliability of the entiresystem. It divides the entire system into subsystem, takes the sub-model of the subsystemas the minimum research object, and converts the model into travel salesman problemwhich obtain ideal results by using ACO. The concept of penalty factor was introducedinto the object function and problem was transformed into a single-objective one. Toverify the validity and the feasibility of the model, an ACO based simulation algorithm isdesigned. In addition an example was mentioned and simulated by Matlab. Simulationresults show that the model and the algorithm are effective.Based on the constructed ACO based software maintenance policy, this thesis alsotakes the real operation of the Ertan information platform into consideration and makesanalysis on the selected Material dispatching management system in details. A complicatenetwork based on this system is built, after analyzed the operation flow of each subsystemand influence relationship among them. Moreover, the system’s software maintenancepriorities are received by setting weight on each operation flow. The simulated results alsoindicate good performance of this model and algorithm.The ACO based software maintenance policy model established in this thesis providesa reference to study problems of the same kind, the simulation optimization algorithmoffers a reference to solve similar problems for future.