RL Return Decision Environment for Supply Chain Cost Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing supply chain systems face challenges in managing large volumes of returns due to complex infrastructure, logistical inefficiencies, and high operational costs, with traditional methods relying on legacy infrastructure and explicit rules, leading to suboptimal decision-making and increased wastage.
Innovation Solution
A processor-implemented method and system utilizing Reinforcement Learning (RL) agents trained with OpenAI gym toolkits to optimize return decisions by determining whether to re-stock, transfer, or return items to regional distributor centers, based on pre-processed input data including SKU, store numbers, and historical sales and returns data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If traditional legacy infrastructure and explicit rules are used for return management, then system simplicity is maintained, but decision-making quality deteriorates and operational costs increase
Solution Approach 1:
The patent replaces traditional mechanical rule-based systems with AI/ML-based intelligent systems. Machine learning models analyze historical return data, product attributes, and customer behavior to automatically generate return decisions, substituting explicit programming rules with learned patterns that adapt to changing conditions and improve decision quality without proportionally increasing system complexity
Solution Approach 2:
The patent introduces an AI/ML intermediary layer between data input and return decisions. This intermediary processes and interprets complex data patterns, transforming raw data into actionable insights that improve decision-making quality while keeping the overall system architecture manageable through modular design
2Adaptability or versatility
If complex monolithic infrastructure is used for supply chain management, then system functionality is comprehensive, but logistical efficiency deteriorates and operational costs increase
Solution Approach 1:
The patent segments the monolithic supply chain system into modular components: data collection modules, AI/ML processing modules, decision-making modules, and execution modules. Each module performs a specific function independently, allowing parallel processing and reducing bottlenecks, thereby improving logistical efficiency while maintaining comprehensive functionality through modular integration
Solution Approach 2:
The patent introduces dynamic adaptability through machine learning models that continuously learn from new data and adjust return decisions in real-time. The system transitions from static rule-based logic to dynamic adaptive decision-making, improving responsiveness and efficiency while handling diverse return scenarios through learned patterns rather than rigid predefined rules
3Ease of operation
If traditional return processing methods are used, then operational simplicity is maintained, but recovery value deteriorates due to damage and obsolescence
Solution Approach 1:
The patent applies preliminary action by using AI/ML models to predict the optimal disposition of returned items before they are processed. The system forecasts which items can be recovered, refurbished, or resold based on historical data and current conditions, enabling proactive decision-making that maximizes recovery value before damage or obsolescence occurs, while maintaining operational simplicity through automated predictions
4Reliability
If explicit rules and SQL databases are used for return optimization, then system reliability is maintained, but cost optimization deteriorates
Solution Approach 1:
The patent changes the fundamental parameters of the decision-making system from fixed explicit rules to flexible learned parameters from AI/ML models. These learned parameters adapt to changing business conditions, product types, and market dynamics, enabling continuous cost optimization without sacrificing reliability, as the models are trained on historical data that captures successful decision patterns while maintaining consistency through probabilistic predictions
Data Source
AI summary
The embodiments of present disclosure herein address unresolved problems in existing initiatives to optimize costs and streamline the return decisions which are based on legacy infrastructure and explicit rules such as a SQL database. Embodiments herein provide a method and system for streamlining return decision in a supply chain network and optimizing costs. The system is configured to create a returns decision environment using an OpenAI gym base class. Created classes lend extensibility for Reinforcement Learning (RL) applications through a supply chain management environment base class and more specific returns decision environment class. These encapsulate all of the environment functions including exploration of contextual information in the dataset.