188宝金博页面版

A Hyper-Heuristic Multi-Criteria Decision Support System for Eco-efficient Product life Cycle. John R. Woodward and Nabil Gindy The University of Nottingham Ningbo China (UNNC) ABSTRACT Decision support is required when complex situations arise during product development which takes into account the whole product life cycle. This is especially true when impacted by the ill-defined consequences on the environment in an ever increasingly eco-conscious world. Analytical Hierarchy process (AHP) is one method of providing decision support, and is an instance of a decision support heuristic. Machine learning methods have proved themselves on many well defined problems with clearly defined objectives. In particular, we focus on the recently emerging field of hyper-heuristics which is a blend of human designed heuristics, with the extension of machine designed heuristics. In essence humans can operate at the higher concept or abstract level, while machine heuristics can operate at a lower level. There are a number of issues within the proposed framework, including visualizing a multi-dimensional surface of designs along the Pareto front, as well as dealing with different types of data during the decision making process. It is proposed that Hyper-heuristics, supplemented with other methodologies to deal with vague or missing data, offer a framework in which to begin to address several of the complex compromises that arise during product development. Key words: Hyper-heuristics, genetic programming, analytical hierarchal process, decision support system. 1. THE ISSUE OF ECO-AWARE DESIGN AND MANUFACTURING. In recent years there has been a shift toward considering the environmental costs associated with manufacturing. While this trend continues, what is not agreed is that there no single universal unit of environmental damage, nor is there ever likely to be. The equivalent number of tons of carbon dioxide (carbon footprint), or the oil point method are candidates but fail to take fully into account other environmental issues such as resource depletion and energy usage. It remains difficult to choose between a product that requires a large quantity of energy during manufacture, a second design that is simpler to make but has a shorter life expectancy or consumes more energy in operation, and a third that contains certain quantities of mercury, cadmium or lead. Each poses different problems that can only be evaluated if their whole product lifecycle (see figure 1) is understood in-depth, thus there remains a need for a framework to formalize our understanding of the success factors, which is the aim of this paper One problems is environmental impact can be difficult to quantify, and to convert into a form that allows direct comparisons to be made. We do not have an analogous unit to the financial unit of the US dollar. Some environmental harm can be measured in terms of the anticipated ‘clean-up cost’ but it is hard to put a financial value on the harm that can occur. The cost to clean a polluted river might be quantifiable, but an