| Author | Warisa Wisittipanich |
| Call Number | AIT Diss. no.ISE-12-09 |
| Subject(s) | Production scheduling
|
| Note | A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Engineering in Industrial and Manufacturing Engineering |
| Publisher | Asian Institute of Technology |
| Series Statement | Dissertation ; no. ISE-12-09 |
| Abstract | The purposes of this dissertation are to develop efficient Differential Evolution (DE) algorithms and broaden their applications for solving single -objective and multi- objective scheduling problems. Particularly, the research focuses on the applications of the proposed algorithms on the two types of NP -hard scheduling models; Job shop scheduling problems (JSP) and Flexible job shop scheduling problems (FJSP). In the first part, the dissertation focuses on the development of efficient DE algorithms for sing le-objective scheduling problems. The preliminary development of a one -stage DE (1ST -DE) is carried out to demonstrate its potential for solving JSP. The preliminary results show that the mechanisms to search for better solutions in JSP of DE are more effe ctive than that of Particle Swarm Optimization (PSO). The 1ST -DE is then extended to solve FJSP with a new proposed regulated decoding rule with the preference of assigned machines. The experimental results also show that DE provides very competitive resul ts compared with those from other existing solution methods. To further enhance the efficiency of the search and remedy the common problem of premature convergence in the 1ST -DE, this dissertation proposes two new DE algorithms with an aim to dynamically b alance exploration and exploitation ability. The performances of the two enhanced D E algorithms are evaluated on set s of benchmark JSP and compared with the state- of-the -art 2S -PSO algorithm, using two separated single objective functions; the minimizatio n of makespan and the minimization of total weighted tardiness. The numerical results demonstrate that the proposed algorithms are highly competitive in terms of solution quality and computational cost. The DE algorithms demonstrate faster convergence beha vior by providing better solutions while using fewer number of function evaluations. In the second part, the dissertation concentrates on the development of a novel multi - objective DE (MODE) algorithm to search for non- dominated or Pareto solutions in mul ti- objective (MO) problems. The emphasis of MODE is on the development of mutation strategies dictating the movement direction of DE population in order to obtain the high quality Pareto front. Five potential mutation strategies which possess distinct sear ch behaviors are proposed in the MODE framework. The performance of MODE is evaluated on the standard test problems, and the results show that MODE algorithm is very competitive when compared to other existing approaches with respect to convergence, distri bution and robustness. The MODE is then extended to solve multi- objective JSP and FJSP with the objective to simultaneously minimize makespan and total tardiness of jobs . The performances of MODE algorithm with different mutation strategies are evaluated on set s of benchmark scheduling problems and compared with results obtained from an existing algorithm. The experimental results demonstrate that the MODE algorithm is a highly competitive approach and is capable of finding diverse and high quality non- dom inated solutions. |
| Year | 2011 |
| Corresponding Series Added Entry | Asian Institute of Technology. Dissertation ; no. ISE-12-09 |
| Type | Dissertation |
| School | School of Engineering and Technology (SET) |
| Department | Department of Industrial Systems Engineering (DISE) |
| Academic Program/FoS | Industrial Systems Engineering (ISE) |
| Chairperson(s) | Voratas Kachitvichyanukul |
| Examination Committee(s) | Huynh Trung Luong ;Do Ba Khang |
| Scholarship Donor(s) | Royal Thai Government Fellowship |
| Degree | Thesis (Ph. D.) - Asian Institute of Technology, 2012 |