Naphtha Cracking Scheduling with Multi-Agent Reinforcement Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for manufacturing system scheduling, such as mixed-integer linear programming, are not effectively applied in real-world factory settings due to complex and mathematically challenging production constraints, limiting the optimization of processes like naphtha cracking centers.
Innovation Solution
A method using reinforcement learning with multi-agents to determine optimal scheduling for naphtha cracking centers, where each agent manages a specific process, including incoming tank, mixing tank, and cracking furnace operations, while adhering to constraints, using a reward system based on profit and cost considerations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If mixed-integer linear programming is used for scheduling optimization, then mathematical optimization can be achieved, but it cannot be effectively applied due to complex production constraints
Solution Approach 1:
The patent replaces traditional mathematical optimization methods (mixed-integer linear programming) with reinforcement learning-based AI agents. This substitution allows the system to handle complex production constraints that are difficult to model mathematically, while still achieving scheduling optimization. The AI agents learn optimal scheduling policies through interaction with the production environment, bypassing the need for explicit mathematical formulation of complex constraints.
2Productivity
If traditional scheduling methods are used, then implementation is straightforward, but profitability and efficiency are limited
Solution Approach 1:
The patent divides the scheduling problem into multiple independent AI agents, each responsible for specific scheduling decisions (e.g., incoming tank scheduling, mixing tank scheduling, cracking furnace scheduling). This segmentation allows each agent to specialize in particular aspects of the production process, improving overall profitability while managing system complexity through modular architecture. Each agent can be trained and optimized independently.
Solution Approach 2:
The patent changes the fundamental parameter of the scheduling system from static rule-based decision-making to dynamic reinforcement learning-based decision-making. By training AI agents to learn optimal policies through trial and error in the production environment, the system adapts to changing conditions and optimizes profitability dynamically, rather than relying on fixed scheduling rules.
3Productivity
If expert decision-making is used for scheduling, then human expertise is utilized, but profitability is limited compared to AI optimization
Solution Approach 1:
The patent implements self-service through AI agents that autonomously make scheduling decisions without requiring continuous human intervention. The agents learn from historical data and production environment interactions to independently optimize scheduling decisions, achieving higher profitability than expert decision-making while reducing the need for human operational complexity.
Data Source
AI summary
A method for scheduling a naphtha cracking center by at least one processor, includes the steps of: obtaining input information; determining, by the at least one processor, incoming tank information using a first agent based on the input information, wherein the first agent is a first artificial intelligence device configured to be learned by reinforcement learning; determining, by the at least one processor, mixing tank combination information using a second agent, wherein the second agent is a second artificial intelligence device configured to be learned by reinforcement learning; and determining, by the at least one processor, cracking furnace operation information using a third agent, wherein the third agent is a third artificial intelligence device configured to be learned by reinforcement learning.