Container Treatment Control Using Preform-Based Sub-Agents
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods require a separate statistical design of experiments for each combination of PET preform and PET container, making it inefficient to use an agent for parameter setting in container treatment systems.
Innovation Solution
A method that utilizes a main agent comprising multiple sub-agents, each assigned to a specific container type or preform, to automatically select the appropriate sub-agent based on preform characteristics for efficient control of the container treatment system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If a separate statistical design of experiments is carried out for each combination of PET preform and PET container, then the agent can be trained to learn parameter settings for that specific combination, but the complexity and time required for agent training increases significantly
Solution Approach 1:
The patent segments the agent training process by dividing preforms and containers into different groups based on their characteristics. Instead of training separate agents for each specific combination, multiple groups are formed where each group shares common characteristics, allowing one agent to learn parameter settings that can be applied across all combinations within that group.
Solution Approach 2:
The patent creates a universal agent that can handle multiple container types and preform combinations within a group. This multi-functional agent learns parameter settings that are applicable across various combinations sharing similar characteristics, eliminating the need for separate specialized agents for each specific pair.
2Manufacturing precision
If a separate statistical design of experiments is carried out for each combination of PET preform and PET container, then the agent can be trained to learn parameter settings for that specific combination, but the device complexity increases due to multiple agents required
Solution Approach 1:
The system segments container types and preform characteristics into distinct groups, where each group is handled by a single trained agent. This segmentation approach reduces the number of agents needed compared to having one agent per specific combination, while still maintaining appropriate specialization for each group.
Solution Approach 2:
Each agent is designed to be universal within its assigned group, capable of handling multiple container types and preform variations that share common characteristics. This multi-functionality reduces the total number of agents required in the system while preserving the ability to provide accurate parameter settings for diverse combinations.
Data Source
AI summary
The disclosure relates to a method for controlling a container treatment system, for example a blow molding machine, for producing containers from preforms, wherein the method comprises: providing a main agent comprising a plurality of sub-agents, each sub-agent being assigned to a container type or a preform, a target parameter set for controlling the container treatment system and a group of target values for a controlled variable being provided for each sub-agent; carrying out a determination of a characteristic of a preform; evaluating a result of the determination and automatically selecting a corresponding sub-agent from the plurality of sub-agents based on the evaluation; and using the corresponding sub-agent for controlling the container treatment system. The disclosure further relates to a device comprising a container treatment system and a control device for carrying out the method.