第五章 結論和未來研究方向
5.2 未來研究方向
經過了導引實驗和綜合實驗的分析後，在此整理分析結果對各種方法提出改 善方向如下：
(1) 初始解
原本的初始解是隨機產生RK(不合理解)後以逐步排程法求解得到 SRK(合理 解)，若能先行考慮專案問題的執行順序限制，只針對沒有執行順序限制的任務 來指定RK，就能一開始就產生出合理解，可能使初始解品質更好。
(3) 共識因子
依4.2.3 和 4.2.4 的分析結果來看，本研究在控制差異化的部分只使用 t1和 t2
做設定，還可以再做其他種方法的嘗試；抽樣數量的部分只使用母體大小或是抽 固定數量，然而抽樣數量對差異化和品質的影響這部分還未做詳細的實驗和分 析，是未來需加以研究的部分。
另外，由於共識矩陣是更新的關鍵，所以可以在連續幾代沒有更新最佳解 時，重新調整共識矩陣的樣本。產生新解時，也可以結合田口方法的概念，以田 口直交表結合共識矩陣中非0 或 1 的比率來更新，以提高求解績效。
(4) 更新解修正器
Alcaraz & Maroto (2001)在使用修正器時，在染色體中多加入一個基因來判 斷該SRK 是否進行修正，或是只修正一半，同樣是本研究可以考慮的改善方向。
參考文獻
吳政翰，“DHC 系統之任務指派與排程的新型基因算法＂，國立交通大學工業 工程與管理學系碩士論文，2006。
蘇朝墩，“品質工程＂，中華民國品質學會，2002。
黎正中，李家琪，“田口方法的應用與發展”，品質學報，第 7 卷第 2 期，頁 117-137，
2000。
Alcaraz J. and Maroto C.,“A robust genetic algorithm for resource allocation in project scheduling,＂Annals of Operations Research, 102, pp. 83-109, 2001.
Alcaraz J., Maroto C. and Ruiz R.,“Improving the performance of genetic algorithms for the RCPS problem,＂in Proceedings of the Ninth International Workshop on Project Management and Scheduling, Nancy, pp. 40-43, 2004.
Birbil S.I. and Fang S.C., “ An electromagnetism-like mechanism for global optimization,＂Journal of Global Optimization, Vol. 25, pp. 263-282, 2003.
Bouleimen K., Lecocq H.,“A new efficient simulated annealing algorithm for the resource-constrained project scheduling problem and its multiple modes version,＂
European Journal of the Operational Research, Vol. 149, pp. 268-281, 2006.
Brucker P., Drexl A., Möhring R., Neumann K. and Pesch E.,“Resource-constrained project scheduling: Notation, classification, models and methods,＂European Journal of the Operational Research, Vol. 112, pp. 3-41, 1999.
Brucker P., Knust S., Schoo A. and Thiele O.,“A branch & bound algorithm for the resource-constrained project scheduling problem, ＂ European Journal of the Operational Research, Vol. 107, pp. 272-288, 1998.
Cho J.H. and Kim Y.D.,“A simulated annealing algorithm for resource constrained project scheduling problems,＂Journal of the Operational Research Society, Vol. 48, pp. 736-744, 1997.
Debels D., Reych B.D., Leus R. and Vanhoucke M., “ A hybrid scatter search/electromagnetism meta-heuristic for project scheduling,＂European Journal of the Operational Research, Vol. 169, pp. 638-653, 2006.
Demeulemeester E. and Herroelen W.,“A branch-and-bound procedure for the
49
Demeulemeester E. and Herroelen W., “ New benchmark results for the resource-constrained project scheduling problem,＂Management Science, Vol. 43, pp.
1458-1492, 1997.
Glover F., “A template for scatter search and path relinking,＂Artificial Evolution Lecture Notes In Computer Science, Vol. 1363, pp. 3-51, 1998.
Glover F., Laguna M. and Martí R.,“Fundamental of scatter search and path relinking,＂Control and Cybernetics, Vol. 39, pp. 653-684, 2000.
Glover F., Laguna M. and Martí R., forthcoming, “Scatter search,＂in: Ghosh, A., Tsutsui S. (Eds.), Theory and Applications of Evolutionary Computation: Recent Trends, Springer-Verlag, New York.
Hartmann S.,“A competitive genetic algorithm for the resource-constrained project scheduling,＂Naval Research Logistics, Vol. 45, pp. 733-750, 1998.
Hartmann S., “ A self-adapting genetic algorithm for project scheduling under resource constraints,＂Naval Research Logistics, Vol. 49, pp. 433-448, 2002.
Hartmann S. and Kolisch R.,“Experimental evaluation of state-of-the-art heuristics for the resource-constrained project scheduling problem,＂European Journal of the Operational Research, Vol. 127, pp. 394-407, 2000.
Herroelen W., “ Project scheduling-Theory and Practice, ＂ Productions and Operations Management, Vol. 14, No. 4, pp. 413-432, 2005.
Herroelen W., Demeulemeester E. and De Reyck B.,“A classification scheme for project scheduling, ＂ in Weglarz J.(Ed.), Project Scheduling－Recent Model, Algorithms and Applications, International Series in Operations Research and Management Science, Vol 14, Kluwer Academic Publishers, Dordrecht, pp.
77-106(Chapter 1), 1998a.
Herroelen W., De Reyck B. and Demeulemeester E.,“Resource-constrained project scheduling: A survey of recent development,＂Computers and Operations Research, Vol. 25, No. 4, pp. 279-302, 1998b.
Kelley J.E. Jr.,“The critical-path method: Resources planning and scheduling,＂in J.F. Muth and G.L. Thompson (Eds.), Industrial Scheduling, Prentice-Hall, New Jersey, pp. 347-365, 1963.
Kochetov Y. and Stolyar A.,“Evolutionary local search with variable neighborhood for the resource constrained project scheduling problem,＂in Proceedings of the 3th International Workshop of Computer Science and Information Technologies, Russia, 2003.
Kolisch R..,“Serial and Parallel resource-constrained project scheduling methods revisited: Theory and computation,＂European Journal of the Operational Research, Vol. 90, pp. 320-333, 1996.
Kolisch R. and Hartmann S., “ Experimental investigation of heuristics for resource-constrained project scheduling: An update, ＂ European Journal of the Operational Research, Vol. 174, pp. 23-27, 2006.
Kolisch R. and Sprecher A.,“PSPLIB－A project scheduling library,＂European Journal of the Operational Research, Vol. 96, pp. 250-216, 1996.
Kolisch R., Sprecher A. and Drexl A.,“Characterization and Generation of a General Class of Resource-constrained Project Scheduling Problems,＂Management Science, Vol. 41, No. 10, pp. 1693-1703, 1995.
Li K.Y. and Willis R.J.,“An iterative scheduling technique for resource-constrained project scheduling,＂European Journal of the Operational Research, Vol. 56, pp.
370-379, 1992.
Martí R., Laguna M. and Glover F.,“Principles of scatter search,＂European Journal of the Operational Research, Vol. 169, pp. 359-372, 2006.
Mingozzi A., Maniezzo V., Ricciardelli S. and Bianco L.,“An exact algorithm for the resource-constrained project scheduling problem based on a new mathematical formulation,＂Management Science, Vol. 44, pp. 715-729, 1998.
Nonobe K. and Ibaraki T.,“Formulation and tabu search algorithm for the resource constrained project scheduling problem (RCPSP),＂in: Ribeiro C.C., Hansen P. (Eds.), Essays and Surveys in Metaheuristics. Kluwer Academic Publishers, Boston, pp.
557-588, 2002.
Őzdamar L. and Ulusoy G.,“A note on an iterative forward / backward scheduling technique with reference to a procedure by Li and Willis,＂European Journal of the Operational Research, Vol. 89, pp. 400-407, 1996.
Sprecher A.,“Scheduling resource-constrained projects competitively at modest resource requirements,＂Management Science, Vol. 46, pp. 710-723, 2000.
51
Taguchi G.,“Introduction to Quality Engineering: Designing Quality into products and Processes,＂Tokyo: Asian Productivity Organization, 1986.
Taguchi G.,“System of Experimental Design,＂White Plains, New York: UNIPUB / Krauss International, 1987.
Valls V., Ballestin F. and Quintanilla M.S.,“A hybrid genetic algorithm for RCPSP,＂
Technical report, Department of Statistics and Operations Research, University of Valencia, 2003a.
Valls V., Ballestin F. and Quintanilla M.S.,“A population-based approach to the resource-constrained project scheduling problem,＂Annals of Operational Research, Vol. 131, pp. 305-324, 2004.
Valls V., Ballestin F. and Quintanilla M.S.,“Justification and RCPSP: A technique that pays,＂European Journal of the Operational Research, Vol. 96, pp. 375-386, 2005.
Valls V., Quintanilla M.S. and Ballestin F.,“Resource-constrained project scheduling:
Acritical reordering heuristic,＂European Journal of the Operational Research, Vol.
149, pp. 282-301, 2003b.
Zhang H., Heng L. and Tam C.M., “ Particle swarm optimization for resource-constrained project scheduling, ＂ International Journal of Project Management, Vol. 24, pp. 83-92, 2006.
附錄
本附錄附上第四章所有實驗的詳細資料，包括總完工時間 ms的加總、Avg.
Dev. from best solutions(%)以及 480 或 600 個例子中，可以找到多少個最佳解。
每張表的左邊代表Debels et al. (2006)的實驗數據，右邊是本研究版本的數據，
因為Debels et al. (2006)在 J60 和 J120 比較的是 PSPLIB 在 2003 年提供的最佳解，
而本研究是使用2007 年的最佳解來比，為免混淆，所以 Debels et al. (2006)在 Avg.
Dev. Best 和共有多少問題找到最佳解的部分不放入，以”-”標示。
53
Debels et al., 2006(V0) v.s. V1(V39)
項目 停止條件 J30 (480 instances) J60 (480 instances) J120 (600 instances)
Debels et al., 2006(V0) v.s. V3
項目 停止條件 J30 (480 instances) J60 (480 instances) J120 (600 instances) 1000 28,410 28,407 38,765 38,788 76,736 -
5000 28,355 28,358 38,554 38,623 75,537 - 50,000 28,319 - 38,420 - 74,671 - Makespan 加總
PSPLIB Best Sum 28,316 38,316 73,766 1000 0.27% 0.25% - 0.97% - - 5000 0.11% 0.12% - 0.63% - - Avg. Dev. Best
50,000 0.01% - - -
1000 421 423 - 355 - -
5000 451 446 - 369 - -
共 有 幾 個 問 題 找 到最佳解
50,000 477 - - -
Debels et al., 2006(V0) v.s. V4(V40)
項目 停止條件 J30 (480 instances) J60 (480 instances) J120 (600 instances) 1000 28,410 28,396 38,765 38,810 76,736 76,946 5000 28,355 28,353 38,554 38,363 75,537 76,217 50,000 28,319 28,326 38,420 - 74,671 -
Makespan 加總
PSPLIB Best Sum 28,316 38,316 73,766
1000 0.27% 0.22% - 1.00% - 3.89%
5000 0.11% 0.10% - 0.65% - 3.06%
Avg. Dev. Best
50,000 0.01% 0.03% - - - -
1000 421 427 - 354 - 196
5000 451 444 - 358 - 212
共 有 幾 個 問 題 找 到最佳解
55
Debels et al., 2006(V0) v.s. V5(V41)
項目 停止條件 J30 (480 instances) J60 (480 instances) J120 (600 instances)
Debels et al., 2006(V0) v.s. V7(V42)
項目 停止條件 J30 (480 instances) J60 (480 instances) J120 (600 instances) 1000 28,410 28,410 38,765 38,772 76,736 76,766 5000 28,355 28,362 38,554 38,619 75,537 75,977 50,000 28,319 28,327 38,420 38,470 74,671 75,143 Makespan 加總
PSPLIB Best Sum 28,316 38,316 73,766
1000 0.27% 0.26% - 0.93% - 3.69%
5000 0.11% 0.13% - 0.62% - 2.80%
Avg. Dev. Best
50,000 0.01% 0.03% - 0.32% - 1.88%
1000 421 414 - 356 - 193
5000 451 433 - 358 - 211
共 有 幾 個 問 題 找 到最佳解
50,000 477 457 - 376 - 233
Debels et al., 2006(V0) v.s. V8
項目 停止條件 J30 (480 instances) J60 (480 instances) J120 (600 instances) 1000 28,410 28,407 38,765 38,792 76,736 -
5000 28,355 28,358 38,554 38,630 75,537 - 50,000 28,319 - 38,420 - 74,671 - Makespan 加總
PSPLIB Best Sum 28,316 38,316 73,766 1000 0.27% 0.25% - 0.97% - - 5000 0.11% 0.11% - 0.64% - - Avg. Dev. Best
50,000 0.01% - - -
1000 421 416 - 353 - -
5000 451 431 - 360 - -
共 有 幾 個 問 題 找 到最佳解
57
Debels et al., 2006(V0) v.s. V10
項目 停止條件 J30 (480 instances) J60 (480 instances) J120 (600 instances)
Debels et al., 2006(V0) v.s. V11
項目 停止條件 J30 (480 instances) J60 (480 instances) J120 (600 instances) 1000 28,410 28,406 38,765 38,789 76,736 -
5000 28,355 28,358 38,554 38,620 75,537 - 50,000 28,319 - 38,420 - 74,671 - Makespan 加總
PSPLIB Best Sum 28,316 38,316 73,766 1000 0.27% 0.25% - 0.97% - - 5000 0.11% 0.11% - 0.62% - - Avg. Dev. Best
50,000 0.01% - - -
1000 421 422 - 353 - -
5000 451 448 - 372 - -
共 有 幾 個 問 題 找 到最佳解
50,000 477 - - -
Debels et al., 2006(V0) v.s. V12
項目 停止條件 J30 (480 instances) J60 (480 instances) J120 (600 instances) 1000 28,410 28,413 38,765 38,792 76,736 -
5000 28,355 28,357 38,554 38,628 75,537 - 50,000 28,319 - 38,420 - 74,671 - Makespan 加總
PSPLIB Best Sum 28,316 38,316 73,766 1000 0.27% 0.27% - 0.98% - - 5000 0.11% 0.11% - 0.64% - - Avg. Dev. Best
50,000 0.01% - - -
1000 421 410 - 353 - -
5000 451 436 - 358 - -
共 有 幾 個 問 題 找 到最佳解
59
Debels et al., 2006(V0) v.s. V13
項目 停止條件 J30 (480 instances) J60 (480 instances) J120 (600 instances)
Debels et al., 2006(V0) v.s. V18
項目 停止條件 J30 (480 instances) J60 (480 instances) J120 (600 instances)
Debels et al., 2006(V0) v.s. V26
項目 停止條件 J30 (480 instances) J60 (480 instances) J120 (600 instances) 1000 28,410 28,479 38,765 39,130 76,736 -
5000 28,355 28,414 38,554 38,982 75,537 - 50,000 28,319 - 38,420 - 74,671 - Makespan 加總
PSPLIB Best Sum 28,316 38,316 73,766 1000 0.27% 0.46% - 1.67% - - 5000 0.11% 0.27% - 1.35% - - Avg. Dev. Best
50,000 0.01% - - -
1000 421 400 - 351 - -
5000 451 419 - 354 - -
共 有 幾 個 問 題 找 到最佳解
50,000 477 - - -
Debels et al., 2006(V0) v.s. V31
項目 停止條件 J30 (480 instances) J60 (480 instances) J120 (600 instances) 1000 28,410 28,493 38,765 39,163 76,736 -
5000 28,355 28,415 38,554 38,988 75,537 - 50,000 28,319 - 38,420 - 74,671 - Makespan 加總
PSPLIB Best Sum 28,316 38,316 73,766 1000 0.27% 0.50% - 1.73% - - 5000 0.11% 0.28% - 1.37% - - Avg. Dev. Best
50,000 0.01% - - -
1000 421 398 - 347 - -
5000 451 417 - 353 - -
共 有 幾 個 問 題 找 到最佳解
61
Debels et al., 2006(V0) v.s. V43
項目 停止條件 J30 (480 instances) J60 (480 instances) J120 (600 instances)
Debels et al., 2006(V0) v.s. V44
項目 停止條件 J30 (480 instances) J60 (480 instances) J120 (600 instances)
Debels et al., 2006(V0) v.s. V45
項目 停止條件 J30 (480 instances) J60 (480 instances) J120 (600 instances) 1000 28,410 28,407 38,765 38,776 76,736 76,725 5000 28,355 28,361 38,554 38,610 75,537 75,788 50,000 28,319 28,331 38,420 38,475 74,671 74,951 Makespan 加總
PSPLIB Best Sum 28,316 38,316 73,766
1000 0.27% 0.25% - 0.94% - 3.64%
5000 0.11% 0.13% - 0.61% - 2.60%
Avg. Dev. Best
50,000 0.01% 0.04% - 0.33% - 1.69%
1000 421 422 - 362 - 193
5000 451 448 - 368 - 209
共 有 幾 個 問 題 找 到最佳解
50,000 477 468 - 394 - 227
Debels et al., 2006(V0) v.s. V46
項目 停止條件 J30 (480 instances) J60 (480 instances) J120 (600 instances) 1000 28,410 28,408 38,765 38,793 76,736 76,824 5000 28,355 28,367 38,554 38,611 75,537 75,834 50,000 28,319 28,331 38,420 38,472 74,671 75,020 Makespan 加總
PSPLIB Best Sum 28,316 38,316 73,766
1000 0.27% 0.26% - 0.97% - 3.76%
5000 0.11% 0.14% - 0.61% - 2.71%
Avg. Dev. Best
50,000 0.01% 0.04% - 0.33% - 1.79%
1000 421 421 - 359 - 193
5000 451 440 - 368 - 208
共 有 幾 個 問 題 找 到最佳解
63
Debels et al., 2006(V0) v.s. V47
項目 停止條件 J30 (480 instances) J60 (480 instances) J120 (600 instances) 1000 28,410 28,417 38,765 38,773 76,736 76,605 5000 28,355 28,360 38,554 38,603 75,537 75,783 50,000 28,319 28,327 38,420 38,464 74,671 74,993 Makespan 加總
PSPLIB Best Sum 28,316 38,316 73,766
1000 0.27% 0.29% - 0.94% - 3.52%
5000 0.11% 0.12% - 0.59% - 2.60%
Avg. Dev. Best
50,000 0.01% 0.03% - 0.31% - 1.72%
1000 421 414 - 357 - 194
5000 451 448 - 367 - 207
共 有 幾 個 問 題 找 到最佳解
50,000 477 470 - 392 - 233