WSEAS Transactions on Systems
Print ISSN: 1109-2777, E-ISSN: 2224-2678
Volume 15, 2016
A Hybrid Solution for an Abstract-Level Daily-Volume Problem and Hourly-Level Multilayer Network Problem
Author:
Abstract: In this paper, we propose a method for operational planning of water transportation systems, by mutually and independently solving problems at an abstract level, for daily volume planning, and at a detailed level, for planning hourly-based flow and storage volumes, by utilizing reservoir water level restoration as a constraint. At the abstract level, a daily-volume plan is formulated, in the form of an interactive, multi-objective planning problem, for flexible application to human solutions. In the detailed level plan, we propose a method for automatically generating the costs of a multilayer network model constrained by reservoir water level restoration, based on the results at the abstract level, and obtain a detailed plan through minimum cost flow calculation. A network transportation problem having holders can be solved by reverting to a multilayer extended network model. In this case it is necessary to define the expense (cost) for arcs, taking into account economics and safety. However, as the scale of the network increases, the work of defining all the costs for arcs becomes difficult. In addition, costs are tuned by a trial-and-error while planning results are compared with an ideal design. Another problem is that it is difficult to flexibly correct or modify a solution to match the desires of designers. As a way of satisfying the multiple aspirations of designers, one approach is to formulate them as a multi-objective planning problem having multiple objective functions. Even in this case, if we apply a scalarizing technique that suitably weights and combines vector-valued objective functions, it becomes essentially impossible to solve the problem as a single objective function problem. One proposed planning method to address this issue involves interactively extracting the preference information from decision makers utilizing the concept of “aspiration level”. However, if a water supply system is formulated as a multi-objective planning problem in a multilayer network model as it is, the scale of the problem becomes extremely large, thereby necessitating enormous computing resources. In order to resolve the issues with these traditional methods, this paper discusses the validity of separating problems into an abstract level and a detailed level and proposes a method for automatically generating the costs in a multilayer network model based on the results of establishing a plan at the abstract level, where daily volumes are set. Furthermore, by separating the problem into two layers—an abstract level for handling daily volumes (abstract planning level) and a detailed level for handling hourly-based flow rates (time-series flow planning level)—the method enables the application of solutions that take advantage of the features of each level. The method utilizes an interactive multi-objective planning method as a user interface and produces a Pareto solution as an abstract plan, through a process of interactively obtaining aspiration levels. If an abstract plan that satisfies the designer can be created, a “cost generator” can produce the costs for the arc variables of a multilayer network model, as described above, based on the abstract plan. Then by determining the minimum cost flow corresponding to these dynamically generated costs, it is possible to arrive at a detailed plan that reflects the various requirements, without having to tune costs in advance by trial and error. In order to validate this proposed method, we formulated a plan using data from an existing water supply network. The results of this show that through variation in the water level, any water reservoir can absorb demand fluctuations and that after 24 hours, the water level restores to its original value, while even for the smoothing ratio, the above objective is satisfied. Furthermore, our results showed that computational effort is greatly reduced and that computing time is shortened.
Search Articles
Keywords: electric vehicle, energy resurrection, limitation of battery capacity, maximization of traveling point, round cost, monotonic increase
Pages: 262-274
WSEAS Transactions on Systems, ISSN / E-ISSN: 1109-2777 / 2224-2678, Volume 15, 2016, Art. #28