Neural Network WIP Control for Semiconductor Cycle Time Bottlenecks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional methodologies are inadequate for identifying key factors to reduce cycle time in complex manufacturing processes like semiconductor manufacturing, where dozens of stages with multiple steps are involved, leading to unreliable results.
Innovation Solution
A system and method that groups tools by similar process steps, uses managing computers to collect and analyze process profile data, calculates key performance indicators (KPIs) such as standard deviation of output from bottleneck tool groups, and employs a neural network model to analyze the impact of KPIs on work-in-progress (WIP), allowing for optimized tool allocation and management to reduce total WIP.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional methodologies (linear programming, computer simulation, queue model) are used to reduce cycle time, then the approach is simple and traditional, but the reliability is insufficient for complex manufacturing processes
Solution Approach 1:
The patent segments the complex manufacturing process into multiple stages and identifies bottleneck tool groups within each stage. By dividing the overall process into manageable segments and analyzing them individually, the system can reliably identify key factors affecting cycle time even in complex semiconductor manufacturing with dozens of stages and hundreds of steps.
Solution Approach 2:
The patent introduces a neural network model as an intermediary between the complex manufacturing process data and the cycle time optimization goal. This intermediary processes the complex data from multiple tool groups and stages, extracting key patterns and relationships that traditional methodologies cannot reliably identify, thereby resolving the contradiction between handling complexity and achieving reliable results.
2Productivity
If more tool groups and stages are added to handle complex manufacturing, then the manufacturing capability increases, but the cycle time becomes harder to control
Solution Approach 1:
The patent implements a feedback mechanism where the neural network model continuously analyzes data from multiple tool groups and stages, identifies bottleneck areas, and provides guidance for optimization. This feedback loop enables effective cycle time control by focusing improvements on the most critical bottleneck stages rather than attempting to control the entire complex process uniformly.
Solution Approach 2:
The patent changes the analytical parameters from traditional aggregate metrics to stage-specific and tool group-specific metrics including standard deviation of output. By focusing on specific parameters at bottleneck stages rather than overall process parameters, the system can effectively control cycle time in complex manufacturing environments with multiple tool groups and stages.
3Loss of time
If bottleneck tool groups are identified and controlled, then the cycle time can be reduced, but the measurement and detection of bottlenecks becomes more difficult
Solution Approach 1:
The patent replaces traditional mechanical and manual bottleneck detection methods with a neural network-based intelligent system. The neural network automatically analyzes data from multiple tool groups and stages, identifying bottleneck areas without manual intervention. This substitution makes bottleneck detection feasible even in complex environments with dozens of stages and hundreds of steps, thereby enabling cycle time reduction.
Solution Approach 2:
The neural network model serves as an intermediary that simplifies the difficult task of bottleneck detection. It processes complex data from multiple sources and translates it into clear identification of bottleneck tool groups and stages, making the detection process manageable despite the complexity of the manufacturing system.
Data Source
AI summary
A method for improving a cycle time of a process of a product is provided. The method includes: collecting process profile data from a plurality of tool groups running the process, and calculating values of a plurality of key-performance-indicators (KPIs) of each tool group including calculating a standard deviation of an output of a stage of a bottleneck tool group of the tool groups; feeding the values of the KPIs and a work-in-progress (WIP) of each tool group into a neural network model in order to output an impact on the WIP for each KPI of each tool group by the neural network model; selecting a set of major KPIs of each tool group from the KPIs according to the impact of each tool group; and controlling the tool groups according to the impact of the set of major KPIs of each tool group in order to reduce a total WIP.