Bottleneck Capacity Expansion On Network Under Uncertainty

The paper provide the capacity expansion model in network for uncertain conditions and study it after an analysis of capacity expansion in our transportation or conmunication network .This study is very helpful for the top management in transportation or conmunication network to use the optimal development strategy for the network capacity expansion with profit objective and minimum total expansion cost. we discuss the problem of net bottleneck capacity expansion in an uncertain circumstance, that is, unit expansion expanse is uncertain, and provide a uniform thinking to build models for those systems in this paper. Moreover, in order to figure out the results of those models, we design a hybrid intelligent algorithm, which is composed of network expansion algorithm, simulation, neural network and genetic algorithm. Finally, an experiment example is presented to show the great ability, fine stability and highly accuracy of this kind of hybrid intelligent algorithm how to solve the complexed problems. This paper has been divided into four parts: In the first part (i.e. Chapter Two), we introduce the problems of singlestage network bottleneck capacity expansion and multistage network bottleneck capacity expansion in a certain circumstance, and then put forward the relevant models and algorithms. Furthermore, we also introduce some algorithms, which are employed in the latter hybrid algorithm, such as genetic algorithm neural and neural network algorithm. In the second part (i.e. Chapter Three), we discuss the problem of singlestage network bottleneck capacity expansion in a stochastic circumstance, and assume that unit expansion expanse is a stochastic variable, which is accorded with probability distribution, in the network expansion process. Based on the uncertainty programming theory, we provide the expected value model, Chance-constrained programming model, and Dependent-chance programming model for this system, bring forth the hybrid intelligent algorithm for the three problems, validate the feasibility by examining the experiment example, and lastly achieve good results. Based on the problem of singlestage network bottleneck capacity expansion, according to different criteria, we bring forward three kinds of multistage network bottleneck capacity expansion programming, including expected value model, Chance-constrained programming model, and Dependent-chance programming mode, apply dynamic programming approach to deal with those multistage problems, and come out with the relevant hybrid intelligent algorithm in the third part (i.e. Chapter Four). In the forth part (i.e. Chapter Five), considering that some uncertain variables are not accord with probability distribution, we need to adopt fuzzy theory to denote them, so the problem of network bottleneck capacity expansion in a fuzzy circumstance has been discussed, the expected value model, Chance-constrained programming mode, and Dependent-chance programming mode for this system have been designed, in order to solve these models, data samples have been gained by employing fuzzy simulation approach to figure out the value of uncertain function, afterwards, a neural network has been trained and embedded in the genetic algorithm to get the hybrid intelligent algorithm. Finally, a great deal of experimental data approve that the algorithm is feasible and effective.