A simulated annealing algorithm for zoning in planning using parallel computing

Santé, Inés; Rivera, Francisco F.; Crecente, Rafael; Boullón, Marcos; Suárez, Marta; Porta, Juan; Parapar, Julio; Doallo, Ramón

doi:10.1016/j.compenvurbsys.2016.05.005

Cited by 16 publications

(14 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In previous studies that applied SA algorithms to solve MOLA problems, the target numbers of units allocated for each land use are usually exact values or percentages. The number of units allocated for each use were set in the initial solutions and kept unchanged during the solution update (Aerts and Heuvelink 2002;Brown 2005, 2007;Santé et al 2016). No penalty function is needed for this approach and the constraint violation problem can be avoided.…”

Section: Conclusion and Discussionmentioning

confidence: 99%

A comparison of three heuristic optimization algorithms for solving the multi-objective land allocation (MOLA) problem

Song

Chen

2018

Annals of GIS

View full text Add to dashboard Cite

Multi-objective land allocation (MOLA) can be regarded as a spatial optimization problem that allocates appropriate use to specific land units concerning some objectives and constraints. Simulating annealing (SA), genetic algorithm (GA), and particle swarm optimization (PSO) have been popularly applied to solve MOLA problems, but their performance has not been well evaluated. This paper applies the three algorithms to a common MOLA problem that aims to maximize land suitability and spatial compactness and minimize land conversion cost subject to the number of units allocated for each use. Their performance has been evaluated based on the solution quality and the computational cost. The results demonstrate that: (1) GA consistently achieves quality solutions that satisfy both the objectives and the constraints and the computational cost is lower. (2) The popular penalty function method does not work well for SA in handling the constraints. (3) The solution quality of PSO needs to be improved. Techniques that better adapt PSO for discrete variables in MOLA problems need to be developed. (4) All three algorithms take high computational costs to achieve quality solutions in handling the objective of maximizing spatial compactness. How to encourage compact allocation is a common problem for them.

show abstract

Section: Conclusion and Discussionmentioning

confidence: 99%

A comparison of three heuristic optimization algorithms for solving the multi-objective land allocation (MOLA) problem

Song

Chen

2018

Annals of GIS

View full text Add to dashboard Cite

show abstract

“…Given the inherent complexity of identifying sustainable land use patterns as discussed above, a number of heuristics have been developed to solve land use optimization problems, such as GA (Chandramouli et al, ; Cao et al, ; Cao, Huang, Wang, & Lin, ; Cao & Ye, ; Schwaab et al, ), simulated annealing (Caparros‐Midwood et al, ; Santé et al, ), particle swarm (Masoomi et al, ), and ant colony algorithms (Mi, Hou, Mi, & Song, ). For example, Aerts et al () applied both GA and simulated annealing to solve a goal programming model for land use allocation and found that the former had better performance in terms of both computational efficiency and quality of solutions.…”

Section: Related Researchmentioning

confidence: 99%

“…For example, Aerts et al () applied both GA and simulated annealing to solve a goal programming model for land use allocation and found that the former had better performance in terms of both computational efficiency and quality of solutions. Porta et al () and Santé et al () sought to improve the efficiency of GA and simulated annealing using parallel computing, respectively.…”

Section: Related Researchmentioning

confidence: 99%

“…One reason for this is that spatial optimization usually involves spatial query and evaluation of spatial relationships that adds to computational complexity (Murray, ). Exact solution approaches are often limited in practice (Porta et al, ; Santé et al, ). Further, land use planning generally involves multiple, often conflicting, objectives, so one best solution likely does not exist.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Evaluation and development of sustainable urban land use plans through spatial optimization

Yao

Murray

Wang

et al. 2019

Transactions in GIS

View full text Add to dashboard Cite

show abstract

“…Essentially, parallel computing is a form of HPC that utilizes multiple CPUs or GPUs to solve large and complex computational problems in a timely manner (Wilkinson and Allen 2005). Parallel computing techniques have been utilized to mitigate the computational burden of large spatial problems in landscape pattern analyses (Hazen and Berry 1997;Kalluri et al 2000;Tang, Bennett, and Wang 2011;Gong, Tang, and Thill 2012;Porta et al 2013;Hohl et al 2016;Santé et al 2016). For example, Tang, Bennett, and Wang (2011) developed a parallel agentbased model that simulates land use change; while Porta et al (2013) developed genetic algorithms to create land use plans in Guitiriz, Spain, while parallel computing techniques were applied to mitigate the computational burden due to a large number of land plots.…”

Section: Computational Challengesmentioning

confidence: 99%

An interactive platform for the analysis of landscape patterns: a cloud-based parallel approach

2019

View full text Add to dashboard Cite

Understanding spatial and temporal characteristics of landscape patterns is critical in ecology, since human interactions with their natural environment can significantly impact ecological processes. The common approach to detect changes in landscape patterns is to evaluate the spatial and temporal variation of well known, established metrics. Examples of such metrics include the composition of different land-use class types and the spatial heterogeneity of individual patches. However, computing such metrics over large geographic areas and at fine levels of granularity requires significant computing resources. In addition, conventional software often lack a visual component that is essential for the detection of changes in landscape patterns and knowledge discovery. In this paper, we propose a cloud-based framework to facilitate the estimation and visualization of landscape pattern analysis in both space and time, capitalizing on the cloud computing facilities provided by Amazon EC2. We illustrate the merit of our approach on landscape metrics across the USA for the years 1992, 2001, and 2011 at the county level. Leveraging cloud computing technology provides the flexibility, scalability and portability to different study regions and at variable scales.

show abstract

A simulated annealing algorithm for zoning in planning using parallel computing

Cited by 16 publications

References 35 publications

A comparison of three heuristic optimization algorithms for solving the multi-objective land allocation (MOLA) problem

A comparison of three heuristic optimization algorithms for solving the multi-objective land allocation (MOLA) problem

Evaluation and development of sustainable urban land use plans through spatial optimization

An interactive platform for the analysis of landscape patterns: a cloud-based parallel approach

Contact Info

Product

Resources

About