Substrate Processing Schedule Creation Using Constraint-Guided ML
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Developers face a significant burden in creating schedules for substrate processing apparatuses due to the need to develop separate flow processes for each model, given their unique configurations, which hampers efficiency and increases development time.
Innovation Solution
A machine learning-based method for creating time schedules for substrate processing apparatuses, where patterns are sequentially placed in a timetable based on predefined constraints, with an evaluation system that selects patterns maximizing rewards and updating the schedule dynamically, allowing for efficient processing sequence optimization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If separate flow processes are developed for each substrate processing apparatus model, then the schedule creation can reflect the specific apparatus configuration, but the developer burden increases significantly
Solution Approach 1:
The schedule creation process is segmented into two independent components: a universal flow process that handles the common scheduling logic, and model-specific constraint conditions that define apparatus-specific rules. This segmentation allows the universal flow to remain unchanged across different models while adapting to specific configurations through constraints alone.
Solution Approach 2:
A universal flow process is created that can handle multiple substrate processing apparatus models simultaneously. This single flow process incorporates constraint conditions that can be configured for different models, making it multi-functional and eliminating the need to develop separate flows for each model.
2Manufacturing precision
If manual flow development is performed for each apparatus model, then the schedule can be optimized for that specific model, but the development time increases
Solution Approach 1:
Constraint conditions for different apparatus models are pre-configured and stored in the system. When creating a schedule, the appropriate pre-configured constraints are automatically selected and applied, eliminating the need for manual flow development for each model while maintaining optimized scheduling.
Solution Approach 2:
The system adapts to different apparatus models by changing constraint condition parameters rather than modifying the entire flow process. This allows the same universal flow to be optimized for different models through parameter adjustment alone, significantly reducing development time.
3Device complexity
If a universal flow process is used for all apparatus models, then the developer burden is reduced, but the schedule may not reflect specific apparatus configurations
Solution Approach 1:
The system uses constraint conditions as feedback mechanisms that inform the universal flow process about specific apparatus configuration requirements. The flow process receives constraint information and automatically adjusts its scheduling decisions to reflect the specific apparatus model being used.
Solution Approach 2:
Constraint conditions act as an intermediary layer between the universal flow process and the specific apparatus configurations. This intermediary translates apparatus-specific requirements into instructions that the universal flow can understand and execute, ensuring both universality and model-specific accuracy.
Data Source
AI summary
A schedule creation method is a method for creating a time schedule by executing a learning step multiple times. The learning step includes sequentially placing patterns each indicating a procedure in a processing sequence in a timetable for defining a time schedule for respective elements of a substrate processing apparatus. The sequentially placing patterns in a timetable includes: acquiring one or more placeable patterns that are allowed to be placed in the timetable from among the patterns based on a prescribed constraint condition; predicting and selecting through machine learning a pattern that makes an evaluation value maximum from among the one or more placeable patterns; and updating the timetable by placing the selected pattern in the timetable.