Injection Molding Condition Detection with Selectable ML Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing condition judgment devices for injection molding machines lack the ability to prepare and select a wide variety of learning models for accurate condition detection and do not effectively collect sufficient sensor value data for generating or updating these models.
Innovation Solution
An information processing method and apparatus that acquire sensor value data from manufacturing devices, store it in databases, generate or update learning models using machine learning, and allow entities to select and use the appropriate models for condition detection, enabling the collection and utilization of sensor data from multiple entities to generate a wide variety of learning models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple learning models are generated for different manufacturing steps, then condition detection accuracy is improved, but device complexity increases
Solution Approach 1:
The system segments the manufacturing process into multiple steps (injection, holding, cooling, etc.) and generates dedicated learning models for each step. Each learning model focuses on detecting abnormalities specific to its corresponding manufacturing step, thereby improving detection accuracy without requiring a single complex model to handle all scenarios
Solution Approach 2:
The system creates a universal framework that can generate multiple specialized learning models within a single platform. The condition judgment device serves multiple functions: data collection, model generation for different steps, model storage, and selective model execution, thereby managing complexity through modular multi-functionality
2Reliability
If sensor value data is collected from multiple entities, then learning model quality is improved, but data management complexity increases
Solution Approach 1:
The system merges data from multiple entities (different manufacturing lines, plants, or companies) into a unified data collection. By combining sensor value data from multiple sources, the system generates more robust and reliable learning models that are trained on diverse operational scenarios, thereby improving model quality while centralizing data management
3Measurement precision
If learning models are generated for each manufacturing step, then detection precision is improved, but processing time increases
Solution Approach 1:
The system dynamically selects which learning model to execute based on the current manufacturing step. Rather than running all models simultaneously or using a single static model, the system adapts its processing by activating only the relevant model for the current step, thereby maintaining high detection precision while minimizing processing time
Data Source
AI summary
An information processing method for detecting conditions of a plurality of manufacturing devices respectively used by a plurality of entities includes acquiring sensor value data obtained by detecting physical quantities related to the manufacturing devices; individually storing collected sensor value data in a plurality of databases; generating by machine learning a plurality of learning models based on the stored sensor value data; and calculating a condition of the manufacturing device of one of the entities by inputting sensor value data acquired from the manufacturing device of the one of the entities to the one or more of the learning models selected by the one of the entities.