Dynamic Rescheduling Expert System for Hybrid Flow Shop with Random Disturbance Procedia Engineering 15 ( 2011 ) 3921 – 3925 1877-7058 © 2011 Published by Elsevier Ltd. doi: 10.1016/j.proeng.2011.08.734 Available online at www.sciencedirect.com Dynamic Rescheduling Expert System for Hybrid Flow Shop with Random Disturbance Jing Yin a, Tieke Lib ,Baojiang Chen a , Bailin Wangb a* aDepartment of electromechanical and vehicle engineering Beijing University of Civil Engineering and Architecture, Beijing, 100044, China bSchool of Economics and Management University of Science and Technology Beijing, Beijing, 100083, China Abstract For the problem of hybrid flow shop with random disturbance, a dynamic rescheduling expert system is established in the paper. As the core of the system, the rescheduling solver is detailed described and the heuristic algorithm is proposed on the base of random disturbance modeling. The system is developed on the platform of G2. The simulation and computational results given at last verified that the system proposed in the paper can well meet the requirements of both function and performance. © 2011 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of [CEIS 2011] Keywords: Dynamic rescheduling; Hybrid Flow Shop ; expert system, time repair 1. Introduction The key of dynamic rescheduling is to obtain the scheduling solution quickly that can satisfy the production constraints and minimize the production cost when workshop condition change. The key of dynamic rescheduling is to obtain the scheduling solution quickly that can satisfy the production constraints and minimize the production cost when workshop condition change. Usually, for the problem of dynamic rescheduling, the way is respectively to generate a totally new solution and on-line local * Corresponding author. Tel.: +086-010-68325943; fax: +086-010-68322514 E-mail address: yinjing@bucea.edu.cn. Open access under CC BY-NC-ND license. Open access under CC BY-NC-ND license. CORE Metadata, citation and similar papers at core.ac.uk Provided by Elsevier - Publisher Connector https://core.ac.uk/display/81986590?utm_source=pdf&utm_medium=banner&utm_campaign=pdf-decoration-v1 http://creativecommons.org/licenses/by-nc-nd/3.0/ http://creativecommons.org/licenses/by-nc-nd/3.0/ 3922 Jing Yin et al. / Procedia Engineering 15 ( 2011 ) 3921 – 3925 modify the initial scheduling solution. Because of the practical use and its difficulty, the dynamic scheduling method has always been the focus of the academic research and enterprise application [1]. A great deal of efforts has been made to the rescheduling problems in recent years. Vieira et al. describe a framework of rescheduling system [2] and present an analytical model to predict the performance of one-machine systems under periodic and event-driven rescheduling strategies [3]. Guo et al. study the properties of a three-machine flow shop scheduling problem and present a trigger mechanism to instruct the rescheduling [4]. References [5]-[6] address rescheduling problem on single machine, parallel machines, flow shops and job shops, respectively. Though variations of existing methods improve the efficiency of dynamic rescheduling from different angles of view, the most difficulty point still lies in the contradiction between the real-time and stability of solution performance and the corresponding contradiction between the real-time ability and complexity of computation. Focusing on these difficulties, this paper studies the hybrid flow shop scheduling (HFS) with the random disturbance which typically exists in the manufacture practice, such as machine failures, arrival of rush orders, material shortages, tool breakages and so on. The remainder of this paper is organized as follows: In next section, the dynamic rescheduling solver embedded in the expert system is introduced. Then, two heuristic algorithms for disturbance of processing abnormal and machine breakdown are proposed. The realization of dynamic rescheduling expert based on G2 and computational results are illustrated next after. We conclude at last. Nomenclature l iO the No. i operation of job lJ , 1, 2,...,l n l ip the normal processing time of l iO 'lip the abnormal processing time of l iO l ist the start time of l iO l iet the completion time of l iO l iSlack the slackness of l iO breakt the time of machine breakdown covre ert the time of machine repaired mcap the remaining capacity of machine jM , j = 1,2,...,m kjb the No. k breakdown of jM hjro the processing time for emergency rush order h on the jM jx the processing time change of jM 2. Design of dynamic rescheduling solver 2.1. The Semantic Network As the core of the whole system, dynamic rescheduling solver consists of rules and processes and describes the problem of HFS with random disturbance. As shown in the Fig.1, there are six class definitions in the conceptual model, including schedule solution, job, operation, machine, constraint and 3923 Jing Yin et al. / Procedia Engineering 15 ( 2011 ) 3921 – 3925 random disturbance. On this basis, the knowledge network can be structured. The underline in the Fig.1 represents the key attributes of the class and the semantic network inference is realized by correlation, focus and multiple inheritances between different classes provided by G2. Fig.1 The semantic network of dynamic rescheduling solver 2.2. The Solving Procedure The solving procedure of dynamic rescheduling: 1. Initialization: Create entity { }1 2 nJ = J , J ,..., J ; input the initial scheduling solution 0S ; Constraint check; initialize attribute value. 2. Real-time monitoring: Executing scheduling; monitoring the shop state; identifying the disturbance. 3. Choose proper program: If the disturbance belongs to processing abnormal, then call the heuristic algorithm of time parameter repair (TPR); if the disturbance belongs to machine breakdown or emergency rush order, then call the heuristic algorithm based on the remaining capacity of machine (RCM); else transfer to human-computer interaction (dynamic editor). 4. Constraint Check: If rescheduling solution is infeasible, then call the conflicts resolution program 5. Output the rescheduling solution. Schedule solution job-num machine-1-num machine-1-start-time machine-1-process-time machine-2-num Reschedule Operation job-num operation-num machine-num machine-start-time machine-process-time machine-end-time opwork-lst-i opwork-lst-j opwork-ff Job job-num job-status job-start-time job-process-time Machine operation-num machine-num machine-status job-num start-time process-time job-sequence Random diturbance event-describe happen-time job-num operation-num machine-num anti-last-time min-gather doing-gather Time-Delay Machine-Break conflict-gather Constraint job-num time-constraint machine-constraint process-constraint 1 1 1 1, 2, 3 n 1 m n 1 1 1 1 Pre-Schedule Graphic illustration Focus Correlation inheritance 3924 Jing Yin et al. / Procedia Engineering 15 ( 2011 ) 3921 – 3925 Considering the production requirement for both the real-time ability of rescheduling efficiency and stability of rescheduling quality, the strategy of local repair is adopted. We take the advantage of the powerful real-time capability and concurrent multi-thread capacity of G2 which is developed by the Gensym Corporation, USA. The solving procedure of dynamic rescheduling is shown as follows: According to different kinds of disruption, different heuristic algorithm is dynamically exploited. The heuristic algorithm of time parameter repair (TPR): 1. Calculate the minimum operation set as follows: 1 ( || ) ( ), l l l k k l k l l i i i j j i j i iA O O O A O O O st > et O A (1) 2. For liO A ,Calculate the l iSlack as follows: 1min , l l l l k l l i i i i j i iSlack st p st st p st (2) 3. If l ' l li i iSlack p - p , then Calculate the start time as follows: For 1 l iO , 1 1 1 1' max , ' l l l l l i i i i ist st st et st (3) For kjO , 1 1' max , ' l k k l l i j j i ist st st et st (4) 4. Constrain propagation and check. { }liA = A - O , if A = , then stop; else, go to step 2. The heuristic algorithm based on the remaining capacity of machine (RCM): 1. Calculate the minimum operation set as follows: cov cov l l l l l i i break i re er i break i re erA O st t st t et t et t (5) 2. For liO A , calculate the mcap of the other normal machine available as follows: ( ) max min , max , , 0k j l l k l k l m i i j i j iM O m cap et st et et st st (6) 3. Assign the machine of max mcap to process l iO , { } l iA = A - O , if A = , then go to step 4 ; else go to step 2. 4. Constrain propagation and check. 5. If conflict exists, TPR adopted to modify the time parameters of related rest operation; else stop. 3. Experimental result Table 1 The computation result of RCM A Initialsolution Rescheduling solution Constrain propagation and check 9 1O CF-1 CF-1 Confict exists in 91O and 13 1O on CF-1, 35f 5 1O CF-3 CF-3 13 1O CF-2 CF-1 3925 Jing Yin et al. / Procedia Engineering 15 ( 2011 ) 3921 – 3925 Random Create an entity of machine breakdown. Assign the attribute value as follows: event- describe=breakdown, job-num=13, operation-num=1, machine-num=2, happen-time = 125, anti-last- time=35. The rescheduling result is shown in table 1 and in table 2. Table 2 The computation result of TPR A Operation Rescheduling solution Operation Rescheduling solution 13 1O 14 1O 13 2O 15 1O 14 2O 13 3O 14 3O 12 1O 15 2O 6 2O 15 3O 12 2O 16 2O 6 3O 16 3O 12 3O 18 2O 17 3O 18 3O 13 1O 13 1 160st 14 3O 14 3 265st 14 1O 14 1 195st 6 2O 6 2 255st 13 2O 13 2 195st 6 3O 6 3 300st 13 3O 13 3 225st 15 3O 15 3 305st 15 1O 15 1 230st 16 3O 16 3 345st 14 2O 14 2 230st 17 3O 17 3 375st 15 2O 15 2 265st 18 3O 18 3 405st 4. Summary When scheduling solution executed in the workshop, dynamic disturbance often unpredictable occurs, such as task change, processing abnormal, machine breakdown and etc. These make rescheduling attach comprehensive attention to the solving mechanism, considering both the real-time ability and solution quality. Focus on this problem, the dynamic rescheduling solver is designed and developed on the platform-G2 in the paper, including the semantic network, the solving procedure and the heuristic algorithm of TPR and RCM. The simulation and computational results show that the system is efficient and effective. References [1] Peter Cowling, Djamila Oueelhadj, Sanja Petrovic. A multi-agent architecture for dynamic scheduling of steel hot rolling. Intelligent Manufacturing, no. 14, p. 457- 470, 2003. [2] G.E. Vieira, J.W. Herrmann, and E. Lin, Rescheduling manufacturing systems: a framework of strategies, policies, and methods, Journal of Scheduling, vol. 6, no. 1, pp. 35-58, 2003. [3] G.E. Vieira, J.W. Herrmann, and E. Lin, Analytical Models to Predict the Performance of a Single- machine System under Periodic and Event-driven Rescheduling Strategies, International Journal of Production Research, vol. 38, no. 8, p. 1899-1915, 2000. [4] B. Guo, Y. Nonaka, Rescheduling and optimization of schedules considering machine failures, International Journal of Production Economics, p. 503–513, 1997. [5] M.S. Akturk, E. Gorgulu, Match-up scheduling under a machine breakdown, European Journal of Operational Research, vol. 112, p. 81-97, 1999. [6] Guo Dong-fen, Li Tie-ke, Constraint Satisfaction-based Method for Steelmaking-Continuous Casting Production Scheduling Problem, Information & control, vol. 34, no.6, p. 753-758, 2005.