Anas Ma’ruf’s research while affiliated with Bandung Institute of Technology and other places

Background: Accurate inventory management of intermittent spare parts requires precise demand forecasting. The sporadic and irregular nature of demand, characterized by long intervals between occurrences, results in a significant data imbalance, where demand events are vastly outnumbered by zero-demand periods. This challenge has been largely overlooked in forecasting research for intermittent spare parts. Methods: The proposed model incorporates the Synthetic Minority Oversampling Technique (SMOTE) to balance the dataset and uses focal loss to enhance the sensitivity of deep learning models to rare demand events. The approach was empirically validated by comparing the model’s Mean Squared Error (MSE) performance and Area Under the Curve (AUC). Results: The ensemble model achieved a 47% reduction in MSE and a 32% increase in AUC, demonstrating substantial improvements in forecasting accuracy. Conclusions: The findings highlight the effectiveness of the proposed method in addressing data imbalance and improving the prediction of intermittent spare part demand, providing a valuable tool for inventory management.

Purpose: This paper investigates a batch scheduling problem where pre-processing is required for parts before processing, considering time-changing effects from part deterioration and operator learning-forgetting.Design/methodology/approach: A mathematical model was developed with the decision variables of the number of batches, the number of pre-processings, batch sizes, and the schedule of processes and pre-processings to minimize total actual flow time. Different numbers of batches were gradually tried and increased until the objective function stopped improving. The minimum number of pre-processings that resulted in a feasible solution was examined at each number of batches.Findings: Our experiment showed that: (1) A faster operator learning led to a lower optimal number of batches and a lower total actual flow time, (2) A faster part deterioration brought a higher number of batches and a higher total actual flow time, (3) The model minimized the number of pre-processings by only scheduling pre-processings before the operations at machine 1, and (4) The model divided the parts into small batches to prevent increased processing time due to part deterioration.Research limitations: The research did not consider multi-due date and multi-item system which require pre-processings with different times and capacities.Practical implications: Production managers should assign fast learning operators to shorter batches and faster deteriorating parts.Originality/value: This research was the first to consider pre-processing in batch scheduling.

In recent years, the use of collaborative robots in assembly lines has promised productivity improvement. It provides more alternatives for the assembly line design, which are the alternative resources of the human, robot, or human-robot collaboration (HRC), and the alternative subsets of processes, termed alternative subgraphs, taking advantage of the variety of robotic tools or end-effectors. However, more alternatives make the assembly line balancing problem more complex. This situation is encountered frequently in modern electronics and automotive assembly lines. The contribution of this study is to provide a mathematical model and solution to the assembly line balancing problem that has both HRC and alternative subgraphs, which has not been discussed as an integrated problem in previous literature. To accomplish this optimization problem, a mixed-integer linear programming (MILP) model has been developed to assign tasks to stations and determine the type of resources required while minimizing the cycle time. Practical constraints such as the available number of robots and robotic end-effector types are also considered. Owing to the complexity of the problem, the exact method for MILP is extremely time-consuming for real-world applications. Therefore, a metaheuristic algorithm based on the ant colony optimization (ACO) approach has been developed to solve the problem more efficiently. The results show that the MILP model can obtain optimal solutions for small-sized problems, whereas the ACO algorithm has proven to be a practical solution for medium- to large-sized problems, providing good solutions within an acceptable computation time. The results also show that the presence of alternative subgraphs can give opportunities for better solutions.

Due to dynamism in job shop scheduling (JSS), the manufacturing environment is highly complex, dynamic, and unpredictable, where disturbing events such as new job arrival, machine breakdown, etc. are unavoidable and change the initial state of the manufacturing environment. To address these issues, this study employs the notion of a flexible shop floor that can meet specific product demands while also adapting to disturbing occurrences in real time. Under new job arrival, machine failure, order cancellation, and rush orders, flexible job shop scheduling (FJSS) is addressed for this purpose, and an efficient framework based on digital twin technology is provided, capable of recognizing and capturing the disturbing factors in real time and triggering timely re-scheduling to mitigate the negative impact in a cost-effective manner. This research could tackle disturbing factors effectively and efficiently in real time.

Predictive maintenance is one of the developments of maintenance activities to ensure the continuity of the production process, especially for companies with continuous production processes. However, predictive maintenance activities require support from other departments, one of which is the availability of spare parts. This study conducted an in-depth analysis of 11 previous studies regarding joint or integration in predictive maintenance. Based on the results of the research, it is known that only a few joint or integration models in predictive maintenance are included in the use of a data-based approach to utilize data collected in real-time. Furthermore, this study proposes a framework that can be used in developing a data-driven joint predictive maintenance and inventory model of spare parts in multi-components/ multi spare parts.

Human-robot collaboration (HRC) has arisen as a promising technology to improve the productivity of assembly processes. This paper discusses an assembly line balancing problem (ALBP) where manual, robotic, or HRC operations may be considered decision alternatives. Each assembly process task may be operated either by a human operator, a robot operator, or an HRC. This possibility of shared functions between humans and robots may result in a hybrid manual-robotic assembly line. This problem’s mathematical model is developed based on the simple ALBP and modifying the idea of two-sided ALBP, with additional aspects related to resource alternatives of human, robot, or HRC, and robot’s tool-type for the operations. The problem is formulated analytically in a mixed-integer linear programming model with a cost-oriented objective function. The exact method can be applied to obtain an optimal solution.

This paper presents a new mathematical model for mixed model assembly line balancing that includes human-robot collaboration. Assembly line with human-robot collaboration means that each workstation may consist of human only as an operator, a robot only, or a human and robot which work simultaneously. The proposed mathematical model will be solved with a mixed-integer linear programming model to minimizes the total relevant cost by assigning tasks to workstations and determine which kind of resources (human, robot, or human-robot collaboration) will be placed on workstations to produce various kinds of products. The type of assembly line is based on a simple assembly line: straight line, deterministic time, and mixed product variant model.

An assembly line model needs to be designed in order to be able to manage the assignment of operators and balance the workloads at once to meet the targeted cycle time. Each station has operations to do which need certain operators’ skills and classification. The proposed model is an analytical model of Mixed Integer Linear Programming (MILP) which uses the makespan as the objective. The inputs of this model are the historical demand, operation data, operator’s data and the corresponding precedence diagrams. The outputs are the operating placement on the workstation, operator assignment on exact operations, the start time of each operation, the last operation start time, and the last operation process time. The proposed model was able to reduce the makespan and reduce the number of operators. Furthermore, the computational results show that the proposed model was able to produce an assembly line which less sensitive by the given main parameter such as the demand (takt time) by comparing the line efficiency, smoothness index and its corresponding makespan in each parameter changes.