Graph Network Substrate Scheduler for Throughput Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current substrate conveyance scheduling methods using simulation calculations are inefficient, requiring extensive calculation time and being unsuitable for real-time operations, and fail to achieve optimal throughput under varying conditions or in nonstationary states.
Innovation Solution
A scheduler that models processing conditions and constraints using graph network theory to calculate a substrate conveyance schedule based on the longest route length, allowing for reduced calculation time and flexibility in parameter settings, and adapts to changing conditions by preparing graph networks for each substrate and adjusting start times.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If simulation calculation is performed for combinations of many parameters to obtain excellent throughput, then throughput is improved, but calculation amount becomes huge and calculation time increases
Solution Approach 1:
The patent segments the substrate processing into multiple processing sections (e.g., pre-processing, main processing, post-processing sections) and applies graph network theory to model and optimize the conveyance schedule for each segment independently. This segmentation reduces the overall calculation complexity while maintaining throughput optimization.
Solution Approach 2:
The patent changes the approach from simulating many parameter combinations to using graph network theory with longest route length calculation. By transforming the scheduling problem into a graph theory problem, it achieves throughput optimization without requiring extensive parameter simulation, thereby reducing calculation time.
2Loss of time
If parameter range is narrowed down based on assumed process recipe condition, then calculation time is reduced, but excellent throughput cannot be achieved when conditions differ from assumed state
Solution Approach 1:
The patent employs dynamic graph network modeling where the processing conditions, processing times, and constraints are represented as nodes and edges that can be adjusted based on actual operating conditions. This dynamic model allows the system to adapt to varying process recipes and nonstationary states while maintaining efficient calculation through the graph theory framework.
Solution Approach 2:
The graph network theory approach serves as a universal scheduling method that can handle various process conditions, recipes, and apparatus states. Unlike simulation methods that require condition-specific parameter tuning, the graph network model provides a unified framework applicable to different scenarios, enhancing versatility without increasing calculation complexity.
3Productivity
If extensive pre-processing calculation is performed to prepare parameter range, then throughput optimization is achieved, but actual operation start is obstructed
Solution Approach 1:
The patent performs preliminary modeling of the substrate processing apparatus using graph network theory to establish the longest route length for each processing path. This preliminary action creates a ready-to-use scheduling framework that requires minimal computation during actual operation, allowing quick start of substrate processing while maintaining optimized throughput.
Data Source
AI summary
A calculation amount and calculation time for a substrate conveyance schedule are reduced. A scheduler is provided which is incorporated in a control section of a substrate processing apparatus including a plurality of substrate processing sections that process a substrate, a conveyance section that conveys the substrate, and the control section that controls the conveyance section and the substrate processing sections, and calculates a substrate conveyance schedule. The scheduler includes: a modeling section that models processing conditions, processing time and constraints of the substrate processing apparatus into nodes and edges using a graph network theory, prepares a graph network, and calculates a longest route length to each node; and a calculation section that calculates the substrate conveyance schedule based on the longest route length.