Metaheuristic Algorithms for Spatial Multi-Objective Decision Making

Sammanfattning: Spatial decision making is an everyday activity, common to individuals and organizations. However, recently there is an increasing interest in the importance of spatial decision-making systems, as more decision-makers with concerns about sustainability, social, economic, environmental, land use planning, and transportation issues discover the benefits of geographical information. Many spatial decision problems are regarded as optimization problems, which involve a large set of feasible alternatives, multiple conflicting objectives that are difficult and complex to solve. Hence, Multi-Objective Optimization methods (MOO)—metaheuristic algorithms integrated with Geographical Information Systems (GIS) are appealing to be powerful tools in these regards, yet their implementation in spatial context is still challenging. In this thesis, various metaheuristic algorithms are adopted and improved to solve complex spatial problems. Disaster management and urban planning are used as case studies of this thesis.These case studies are explored in the four papers that are part of this thesis. In paper I, four metaheuristic algorithms have been implemented on the same spatial multi-objective problem—evacuation planning, to investigate their performance and potential. The findings show that all tested algorithms were effective in solving the problem, although in general, some had higher performance, while others showed the potential of being flexible to be modified to fit better to the problem. In the same context, paper II identified the effectiveness of the Multi-objective Artificial Bee Colony (MOABC) algorithm when improved to solve the evacuation problem. In paper III, we proposed a multi-objective optimization approach for urban evacuation planning that considered three spatial objectives which were optimized using an improved Multi-Objective Cuckoo Search algorithm (MOCS). Both improved algorithms (MOABC and MOCS) proved to be efficient in solving evacuation planning when compared to their standard version and other algorithms. Moreover, Paper IV proposed an urban land-use allocation model that involved three spatial objectives and proposed an improved Non-dominated Sorting Biogeography-based Optimization algorithm (NSBBO) to solve the problem efficiently and effectively.Overall, the work in this thesis demonstrates that different metaheuristic algorithms have the potential to change the way spatial decision problems are structured and can improve the transparency and facilitate decision-makers to map solutions and interactively modify decision preferences through trade-offs between multiple objectives. Moreover, the obtained results can be used in a systematic way to develop policy recommendations. From the perspective of GIS - Multi-Criteria Decision Making (MCDM) research, the thesis contributes to spatial optimization modelling and extended knowledge on the application of metaheuristic algorithms. The insights from this thesis could also benefit the development and practical implementation of other Artificial Intelligence (AI) techniques to enhance the capabilities of GIS for tackling complex spatial multi-objective decision problems in the future.