To cope with fluctuating production demands in the volatile markets, a new-type seru production system is adopted due to its efficiency, flexibility, and responsiveness advantages. Seru loading problems are receiving tremendous attention, however, full downward substitution and uncertainties in product demand and yield are seldom considered. Accordingly, a combinatorial optimization seru loading model is constructed to address these concerns so as to maximize system profits, which, however, is notoriously challenging to solve with exact algorithms. Therefore, an improved genetic-simulated annealing algorithm (IGSA) is designed to obtain optimal loading results. To validate the effectiveness and efficacy of the proposed IGSA, algorithm comparisons with adaptive genetic algorithm (A-GA) and simulated annealing (SA) algorithm are conducted. Results show that the proposed model is effective for addressing the seru loading problem and IGSA is robust in solving the seru loading model.

Seru production is one of the latest manufacturing modes arising from Japanese production practice. Seru can achieve efficiency, flexibility, and responsiveness simultaneously. To accommodate the current business environment with volatile demands and fierce competitions, seru has attracted more and more attention both from researchers and practitioners. A new planning management system, just-in-time organization system (JIT-OS), is used to manage and control a seru production system. The JIT-OS contains two decisions: seru formation and seru loading. By seru formation, a seru system with one or multiple appropriate serus is configured; by seru loading, customer ordered products are allocated to serus to implement production plans. In the process of seru formation, workers have to be assigned to serus. In this paper, a seru loading problem with worker assignment is constructed as a bi-level programming model, and the worker assignment on the upper level is to minimize total idle time while the lower level is to minimize the makespan by finding out optimal product allocation. A product lot can be splitted and allocated to different serus. The problem of this paper is shown to be NP-hard. Therefore, a simulated annealing and genetic algorithm (SA-GA) is developed. The SA is for the upper level programming and the GA is for the lower level programming. The practicality and effectiveness of the model and algorithm are verified by two numerical examples, and the results show that the SA-GA algorithm has good scalability.

This paper is concerned with the scheduling problem in a new-type seru production system by consideration of DeJong's learning effect to minimise the total weighted completion time, so as to achieve efficiency, flexibility, and fast responsiveness to cope with the current volatile market. A combinatorial optimisation model is constructed and then reformulated to a binary quadratic assignment program. Accordingly, after presenting the necessary and sufficient condition for the locally optimal solution, a tabu search with strategic oscillation based on 3-opt as a diversification strategy is designed as the solution approach. A set of test problems are generated, and computational experiments with large-scale cases are made finally. The results indicate that the proposed heuristic algorithm is promising in solving seru scheduling problems and has a good performance in term of solution quality, efficiency, and scalability.