Supply Chain Preprocessor for MILP Model Simplification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Supply chain planning problems, particularly mixed integer linear programming (MILP) issues, often require excessive computation time and resources due to their complexity, even with small numbers of variables, making it difficult to find optimal solutions within reasonable times.
Innovation Solution
A computer-implemented architecture that preprocesses supply chain planning problems using a set of rules to simplify and reduce the core model, transforming it into a more manageable mathematical model, which can then be solved efficiently, independent of the specific optimizer algorithm used. This includes redundancy elimination, remodelling, big number reduction, and bound reduction rules to improve numerical stability and model simplicity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If supply chain planning problems are solved using traditional optimization techniques, then optimal solutions can be found, but excessive computation time and resources are required
Solution Approach 1:
The patent applies preprocessing techniques before the main optimization process to simplify the problem structure. This includes identifying and eliminating redundant constraints, tightening bounds on variables, and reformulating the model to reduce computational complexity, thereby preparing the problem in advance for more efficient solving
Solution Approach 2:
The patent extracts and removes redundant elements from the supply chain planning model. Specifically, it identifies and eliminates constraints that do not affect the optimal solution, removes duplicate variables, and extracts unnecessary data elements, thereby reducing the overall problem size and computational burden
2Productivity
If the core model is simplified through preprocessing, then computation time is reduced, but the complexity of the preprocessing step increases
Solution Approach 1:
The patent divides the preprocessing task into distinct modular components: redundancy elimination module, bound reduction module, and remodelling module. Each module handles specific aspects of simplification independently, making the overall preprocessing process more manageable and maintainable despite its complexity
Solution Approach 2:
The patent introduces an intermediary preprocessing layer between the original core model and the optimization solver. This intermediary component transforms the complex original model into a simplified form that is easier to solve, acting as a mediator that bridges the gap between model complexity and solving efficiency
Data Source
AI summary
Methods, systems and computer-implemented architectures are provided for performing supply chain planning. In one implementation, a system is provided that comprises a database configured to store master data descriptive of supply chain items, a model generator coupled to the database and configured to derive a master data-based core model representative of a supply chain planning problem, and a solver configured to translate the core model into a mathematical model and determine a solution to the mathematical model. In addition, the system may comprise a preprocessor configured to preprocess the core model by subjecting the core model to a set of rules to derive a preprocessed core model, wherein the solver establishes the mathematical model from the preprocessed core model. By applying a set of preprocessing rules to the core model, where the business logic of the supply chain planning problem still exists, the solving process of the planning problem can be improved and the performance of the solver can be enhanced. Moreover, meaningful results obtained from the preprocessing can be reported to a customer.