Informed multi-objective decision-making in environmental management using Pareto optimality
Informed multi-objective decision-making in environmental management using Pareto optimality
复制标题
DOI：
10.1111/j.1365-2664.2007.01367.x
复制
发表时间：
2008-02-01
影响因子：
5.7
通讯作者：
Agee, James K.
中科院分区：
文献类型：
--
作者：
Kennedy, Maureen C.;Ford, E. David;Agee, James K.
1.环境管理中的有效决策需要考虑可能相互冲突的多个目标。常见的优化方法使用多个目标的权重来将它们聚合成单个值，忽略了对管理问题中目标之间关系的有价值的洞察力。我们提出了一个多目标优化过程，近似非支配帕累托边界，而不使用权重，允许可视化的权衡目标。非支配的帕累托前沿近似于一个向量目标函数的同时优化;如果一个目标的改进是在另一个目标的损害，则两个向量目标函数被定义为非支配的。我们演示的方法与案例研究的森林燃料处理，减少火灾对森林的影响的最佳分配。多重目标是保护濒危物种的栖息地，保护后期演替森林保护区，并尽量减少总面积处理。在三个优化搜索的比较中，非支配解的数量随着目标空间的维数而增加，但只有两个目标的搜索是无效的，在不同的景观类型中最小化火灾的影响。关键的挑战包括近似非支配集所需的大量计算时间，以及减少详细分析的解决方案的数量。合成与应用。所提出的多目标优化方案可以适用于其他环境管理问题，并很容易纳入广泛的量化目标。这一工具为决策者提供了一套备选办法，可对多种目标之间的各种权衡作出估计，并提供了一个共同基础，使对话能够在环境管理问题上达成知情的妥协和决定。
1. Effective decision-making in environmental management requires the consideration of multiple objectives that may conflict. Common optimization methods use weights on the multiple objectives to aggregate them into a single value, neglecting valuable insight into the relationships among the objectives in the management problem.2. We present a multi-objective optimization procedure that approximates the non-dominated Pareto frontier without the use of weightings, allowing for visualization of the trade-offs among objectives. The non-dominated Pareto frontier is approximated by the simultaneous optimization of a vector objective function; two vector objective functions are defined as non-dominated if improvement with respect to one objective is at the detriment of another objective.3. We demonstrate the method with a case study for the optimum distribution of forest fuels treatments that reduce the impact of fire on a forest. The multiple objectives are to protect habitat of an endangered species, protect late successional forest reserves and minimize the total area treated. In the comparison of three optimization searches, the number of non-dominated solutions increases with the dimensions of the objective space, but with only two objectives the search is ineffective in minimizing fire impact in the different landscape types. Key challenges include the extensive computation time required to approximate the non-dominated set, and reducing the number of solutions that are analysed in detail.4. Synthesis and applications. The multi-objective optimization program presented can be adapted to other environmental management problems, and easily incorporates a wide range of quantifiable objectives. This tool provides decision-makers with a set of alternatives that estimates the full range of trade-offs among multiple objectives and provides a common ground from which dialogue can come to an informed compromise and decision in environmental management problems.