To this end, this work introduces a novel learning-based RHO framework, termed L-RHO, designed to accelerate RHO for long-horizon COPs by identifying overlapping decision variables that do not need to be re-optimized between consecutive iterations. Long-horizon FJSPs represent a complex class of COPs involving interdependent assignment and scheduling decisions over extended time horizons, significantly increasing the difficulty compared to the basic JSP variant. The core of L-RHO is to intelligently fix operations assigned to the same machine across successive RHO iterations. By employing a customized network with tailored contextual embeddings, L-RHO learns to imitate a look-ahead oracle, effectively reducing the size of subproblems, significantly improving solve times, and delivering notable improvements in solution quality. Mathematical optimization problems including a time dimension abound.

In the present paper, we will use a multi-facility procedure similar to the allocation approach by Meyr (2009) and Azevedo et al. (2016). The simulation infrastructure proposed by Herding et al. (2017) is extended for the rolling horizon experiments carried out in the course of the research for this paper. rolling horizon approach This means that production control is related to lots that are already released on the shop floor. The order penetration point is at the interface between planning and control.

Products and services

All already promised but unfinished orders are considered within a STDSM approach (Fleischmann and Meyr 2004) taking into account the available supply and capacity. STDSM approaches are desirable in semiconductor supply chains due to the long cycle times and the process and demand uncertainty (Mönch et al. 2018b). Note that the proposed planning approach is somehow similar to the FE- and BE-based production planning decomposition procedure used in the decision support system IMPReSS (Leachman et al. 1996). However, orders are not considered in IMPReSS in contrast to the present paper. Next, we will describe the different ingredients of the proposed planning approach.

• RHO breaks the problem into overlapping subproblems with shorter planning horizons rolling forward over time, allowing for much better scalability.
• The STDSM function is similar to batch promising, however, all already promised orders compete for the supply and the capacity, while only the orders arriving within the batch interval are considered in batch order promising.
• Such unique challenges, coupled with the inherent NP-hardness and large-scale nature of the problems, call for advanced temporal decomposition strategies Du and Pardalos, 1998, Hentenryck and Bent, 2006, Yang et al., 2013.

Using data from a large semiconductor manufacturer, it is shown by designed experiments that average stock levels are reduced and the overall service level is increased. This is especially true for customers that provide truthful forecasts. An allocation planning model similar to the model of Meyr (2009) is proposed by Babarogić et al. (2012). Customers are assigned to priority groups based on the size of their orders.

2 Detailed Comparison with RHO Baselines Under Different FJSP Variants

In this paper, a STDSM approach for semiconductor supply chains was proposed. The approach is based on a decomposition that takes into account the structure of the semiconductor supply chain. The integration of the STDSM approach into a hierarchical planning approach that includes master planning, allocation planning, and production planning was discussed.

Furthermore, when evaluating the performance on 600 operations FJSP (10, 20, 30) in Table 1, we see that option (1) and (2) , results in a longer solve time but an improved makespan from the architecture without attention. We also note that option (3) is strictly dominated by the performance of the architecture without attention. The results of the rolling horizon experiments are shown in Table 5. 95% confidence intervals are presented instead of the values of point estimates to obtain statistically reasonable results. Two values are provided for each factor level and performance measure.

A.5.6 Performance of RHO methods under the same RHO parameter.

The term rolling horizon is used to indicate that a time-dependentmodel is solved repeatedly, and in which the planning interval is movedforward in time during each solution step. With the facilitiesintroduced in the previous sections setting up such a model isrelatively easy. This section outlines the steps that are required toimplement a model with a rolling horizon, without going into detailregarding the contents of the underlying model. RHO breaks the problem into overlapping subproblems with shorter planning horizons rolling forward over time, allowing for much better scalability. However, such overlaps often lead to redundant computations that reduce the efficiency especially when only a small subset of variables needs re-optimization. This presents an opportunity to accelerate RHO by identifying such redundancies, an approach that, to our knowledge, has not yet been explored in the combinatorial optimization context.

Both the STDSM and the RBR approach require allocation planning, i.e. solving instances of the model (A1)–(A6). Note that the average computing time for a single STDSM decision in the case of the SSC-S supply chain is less than 5 min. Discrete-event simulation is crucial for implementing rolling horizon schemes in a risk-free environment due to the fact that the dynamics and the uncertainty of the supply chain can be covered. Several early papers mention demand fulfillment-related subsystems of semiconductor supply chain planning systems. For instance, a module of the IMPReSS production planning system at Harris Corporation calculates product availability for the quotation and order entry system (Leachman et al. 1996). Requirement and system specification efforts are described by Soares et al. (2000) for an order promising module of a decision support system for semiconductor supply chains, but computational results are not reported.

• L-RHO outperforms both traditional solvers (CP-SAT, GA) and the learning-based DRL solver in the standard (offline) FJSP setting.
• Long-horizon FJSPs represent a complex class of COPs involving interdependent assignment and scheduling decisions over extended time horizons, significantly increasing the difficulty compared to the basic JSP variant.
• For instance, a module of the IMPReSS production planning system at Harris Corporation calculates product availability for the quotation and order entry system (Leachman et al. 1996).
• This is indicated by the surrounding frame for the BE facilities (Step 3) in Fig.
• Moreover, we use an iterative approach that extends the delivery time windows of the orders.
• The algorithm to implement the rolling horizon can be outlined asfollows.

Assessing this interaction in a rolling horizon setting under process and demand uncertainty is highly desirable. AB – We address the dynamic design of supply chain networks in which the moments of demand distribution function are uncertain and facilities’ availability is stochastic because of possible disruptions. To incorporate the existing stochasticity in our dynamic problem, we develop a multi-stage stochastic program to specify the optimal location, capacity, inventory, and allocation decisions. Further, a data-driven rolling horizon approach is developed to use observations of the random parameters in the stochastic optimization problem. In contrast to traditional stochastic programming approaches that are valid only for a limited number of scenarios, the rolling horizon approach makes the determined decisions by the stochastic program implementable in practice and evaluates them. The stochastic program is presented as a quadratic conic optimization, and to generate an efficient scenario tree, a forward scenario tree construction technique is employed.

Various generalizations are shown to be captured by straightforward modifications of our model. I am trying solve a scheduling problem adopting a rolling horizon approach. I have developed an Integer programming model and seek to speed up execution.

However, only some preliminary computational results for the interaction of master planning and rule-based online order promising and repromising are presented in this paper. Demand fulfillment and order management are important functions in semiconductor supply chains to interact with customers. In this paper, an iterative short-term demand supply matching (STDSM) algorithm based on mixed-integer linear programming (MILP) is proposed.