Scheduling jobs on parallel machines applying neural network and heuristic rules

Dynamic adjustment of dispatching rule parameters in flow shops with sequence-dependent set-up times

“…The schedule resulting from applying the dispatching rule is post-processed by a simulated annealing algorithm for finetuning. Park, Kim, and Lee (2000) extended the approach from by adding an additional factor, the setup time range. To determine the scaling parameters k1 and k2 of the ATCS rule, they trained an artificial neural network with five input parameters and k1 and k2 as outputs with data from preliminary simulation runs of static scenarios.…”

“…Another extension has been presented by Chen, Pfund, and Fowler (2010) who consider the one-stage parallel machine problem with four characterizing factors as . However, Park, Kim, and Lee (2000) and Chen, Pfund, and Fowler (2010) only evaluated a one-stage production scenario.…”

“…They compared a neural network with an inductive decision tree to select the k -parameter of the BATC rule, which is an adaption of the ATC rule for batch machines (Balasubramanian, Mönch, Fowler and Pfund 2004). They considered dynamic job arrivals in their experiments and used system parameters similar to and Park, Kim, and Lee (2000). Their results showed that an improvement in total weighted tardiness over a fixed k setting could be gained.…”

Heuristics for Unrelated Parallel Machine Scheduling with Secondary Resource Constraints

“…Yalaoui & Chu (2002) derived some dominance properties and proposed a branch-and-bound procedure to minimize total tardiness. Park et al (2000) applied neural networks to obtain some look-ahead parameters which were used to calculate the priority index of each job to minimize total weighted tardiness. Lee & Pinedo (1997) presented a three-phase heuristic to minimize total weighted tardiness.…”

Identical Parallel Machine Scheduling with Dynamical Networks using Time-Varying Penalty Parameters

“…But a few papers are proposed for the solution of parallel machine scheduling problem using ANNs. Park et al (2000) presented a backpropagation network for solving identical parallel machine scheduling problems with sequence dependent set up times. They tried to find the sequence of jobs processed on each machine with the objective of minimizing weighted tardiness.…”