This paper proposes an algorithm for the Unrelated Parallel Machine Scheduling Problem(UPMSP) without setup times, aiming to minimize total tardiness. As an NP-hard problem, the UPMSP is hard to get an optimal solution. Consequently, practical scenarios are solved by relying on operator's experiences or simple heuristic approaches. The proposed algorithm has adapted two methods: a policy network method, based on Transformer to compute the correlation between individual jobs and machines, and another method to train the network with a reinforcement learning algorithm based on the REINFORCE with Baseline algorithm. The proposed algorithm was evaluated on randomly generated problems and the results were compared with those obtained using CPLEX, as well as three scheduling algorithms. This paper confirms that the proposed algorithm outperforms the comparison algorithms, as evidenced by the test results.

This study focuses on a job-shop scheduling problem with the objective of minimizing total tardiness for the job orders that have different due dates and different process flows. We suggest the dispatching rule based scheduling algorithm to generate fast and efficient schedule. First, we show the delay schedule can be optimal for total tardiness measure in some cases. Based on this observation, we expand search space for selecting the job operation to explore the delay schedules. That means, not only all job operations waiting for process but also job operations not arrived at the machine yet are considered to be scheduled when a machine is available and it is need decision for the next operation to be processed. Assuming each job operation is assigned to the available machine, the expected total tardiness is estimated, and the job operation with the minimum expected total tardiness is selected to be processed in the machine. If this job is being processed in the other machine, then machine should wait until the job arrives at the machine. Simulation experiments are carried out to test the suggested algorithm and compare with the results of other well-known dispatching rules such as EDD, ATC and COVERT, etc. Results show that the proposed algorithm, MET, works better in terms of total tardiness of orders than existing rules without increasing the number of tardy jobs.

This paper considers a parallel-machine scheduling problem with dedicated and common processing machines using GA (Genetic Algorithm). Non-identical setup times, processing times and order lot size are assumed for each machine. The GA is proposed to minimize the total-tardiness objective measure. In this paper, heuristic algorithms including EDD (Earliest Due-Date), SPT (Shortest Processing Time) and LPT (Longest Processing Time) are compared with GA. The effectiveness and suitability of the GA are derived and tested through computational experiments.

Lot-order assignment is the process of assigning items in lots being processed in a production facility to orders to meet due-dates of the orders. In this study, we consider the lot-order assignment problem (LOAP) with the objective of minimizing total tardiness of the orders with distinct due dates. We address similarity relationships between the LOAP and the single machine total tardiness scheduling problem (SMTTSP) and suggest priority rules for the LOAP based on those for the SMTTSP. Performances of the priority rules are compared with each other and with that of the commercial optimization software package in computational experiments.

본 연구는 반도체 제조공정에서 사용되는 단일 Burn-In oven에서의 Total weighted earliness와 Tardiness를 최소화하기 위한 생산 스케줄링을 결정하는 문제를 다룬다. 본 연구에서는 모든 작업은 상시에 시작가능하고 각각은 서로 다른 가중치를 가지고 있다고 가정하였다. 일반적으로 단일 Burn-In oven은 다양한 작업들이 동시에 가능한 Batch processing 기계이다. 따라서 다양한 작업들로 구성된 하나의 Batch

This paper deals with a batch processor model in which the batch processing times depend on the jobs assigned to the batch. Each job has a distinct processing time which is determined as not the exact value but the range from the lower limit to the upper,