Web-based decision system for effective process planning in network manufacturing environment
Summary (2 min read)
Introduction
- Keywords - Meta-heuristics, Simulated Annealing, Discrete Artificial Bee Colony, Neighborhood Structures, TSP.
- I. INTRODUCTION Complex Combinatorial Problems (COPs) such as the Travelling Salesman Problem (TSP) is a classic combinatorial optimization problem, which has been applied in, logistics, transportation, networking and commercial domains [1-7].
- A Simulated Annealing (SA) and Artificial Bee Colony algorithm (DABC) are used to approach TSP and to determine how much the chosen NS impacts the performance of either meta-heuristics.
- Furthermore, a statistical analysis of the performance of the NS is presented.
- There are two distinct classes of meta-heuristics: the single solution based meta-heuristics and population based metaheuristics.
B. Discrete Artificial Bee Colony
- At the phase of working bees, each one will explore a solution in the neighbourhood (vi) of its food supply, and if it yields a superior performance than the current food source, the new one should replace that food source.
- A food source is abandoned once l iterations did occur without improving the performance.
- NS2, or 2-opt, is a more tradional choice for TSP problems, it selects two paths at random, disconnects and reconnects them in another manner [6].
- Since meta-heuristics are bestowed with mechanisms to move from the local-optimums, the NS should be presented as another parameter, which interacts with the other parameters to determine the overall balance between the intensity/diversity of the meta-heuristic search procedure.
B. Computational Results
- Both meta-heuristics were run five times for each problem and achieved computational results close to the optimum.
- In figure 4 the top row represents the computational result of SA, and the bottom row represents the computational results of DABC, for 1 and 10 seconds, with each neighborhood structure results represented in a different color.
- Once more, all the NS improved their performance in the 10 seconds limit, nevertheless, in opposition to the SA, there appears to be a difference between the improvement of the NS1 and NS2 more intense searches and the NS3 or NS4 more diverse searches.
- In the 10 seconds experiment the performance of both meta-heuristics improved, but there appears to be a difference in improvement of NS1 and NS2, and NS3 and NS4 in DABC, where NS1 and NS2 did not improve as the NS with more diverse searches.
C. Statistical Results
- In the computational trials, there appears to be a difference in the evolution in the NS performance, however to examine this difference in more detail the computational results need to be normalized to compare the results across all instances.
- In SA the difference in the improvement of NS1 and NS2, and NS3 and NS4 is not as prominent as in DABC.
- D. Karaboga and B. Gorkemli, “A Combinatorial Artificial Bee Colony Algorithm for Traveling Salesman Problem,” Proc. of the International Symposium on Innovations in Intelligent Systems and Applications , 2011, 50-53. [16].
Did you find this useful? Give us your feedback
Citations
36 citations
1 citations
Additional excerpts
...Networked manufacturing system architecture [41]....
[...]
References
3,568 citations
"Web-based decision system for effec..." refers background in this paper
...These exchange messages need engineering ontologies (Daconta et al. 2003; Lin and Harding, 2007; Uschold and Grueninger, 1996) to develop....
[...]
505 citations
233 citations
"Web-based decision system for effec..." refers background in this paper
...…manufacturing is the MADE (Manufacturing Automation and Design Engineering) an American research project, Several researches (Cutskosky et al. 1996; Petrie, 1996; Whitney et al. 1995; Bryant et al. 1996; Will, 1996) involved in handling the above mentioned MADE program and their contributions are…...
[...]
206 citations
184 citations
Related Papers (5)
Frequently Asked Questions (7)
Q2. How many seconds of deviations for a second?
In KroE, the deviations for 1 second are 64.047%, 4.953%, 64.279% and 44.408%, for 10 seconds the deviations improve to 54.341%, 2.655%, 12.534% and 7.722%.
Q3. What is the metric to compare different meta-heuristics?
Since the optimal solutions KroA, KroB, KroC, KroD and KroE, are available in TSPLIB, this document will use mean percent devotion from the optimal solution, which is the best know metric to compare different meta-heuristics [23].
Q4. What is the stoppage condition in the DABC?
Since this paper will study the performance of each NS bellow 1 and 10 seconds of computational time, the Stoppage Condition will be the number of iterations, while the NS will be the one presented in the previous point.
Q5. How many times will the results of the meta-heuristics be tested?
Moreover each meta-heuristics will be tested with parameters that find solutions in less that 1 and 10 seconds, to examine how the performance of each NS varies with the increase of the computational time, to examine the evolution of the solutions.
Q6. What is the meta-heuristic for the scout bees?
Through this meta-heuristic the search within the solution space is performed by three types of bees: worker bees, opportunistic bees, and scout bees.
Q7. How many iterations of the NS will be used?
Like in SA, the stop criterion will be the number of iterations, since the purpose is to examine the performance of each NS bellow 1 and 10 seconds of computational time.