Injection Molding State Determination Using Corrected ML Estimates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current state determination methods for injection molding machines require significant costs and time to collect learning data for various machine types and auxiliary equipment, leading to inefficient abnormality detection, especially when dealing with different specifications and materials.
Innovation Solution
A state determination device that uses machine learning to derive an abnormality degree correction value, allowing for numeric conversion of estimation results to account for machine and equipment differences, enabling a single learning model to be applied across various injection molding machines, with a correction function applied based on stored coefficients to determine abnormality states.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning is performed with various types of injection molding machines and auxiliary equipment to improve determination accuracy, then determination accuracy is improved, but cost and time for data collection increase significantly
Solution Approach 1:
The patent extracts only the essential machine specification parameters (motor rated torque, inertia, reduction gear reduction ratio) from the complex variety of injection molding machines. By isolating these key parameters that actually influence the check ring wear pattern, the system can perform numeric conversion on learning data to adapt it to different machine specifications without requiring complete re-collection of data for every machine type.
Solution Approach 2:
The patent applies parameter changes by performing numeric conversion on learning data based on machine specification parameters. Instead of collecting new data for each machine type, the system transforms existing learning data using the specific parameters of the target machine (motor torque, inertia, reduction ratio) to generate appropriate determination criteria, thereby avoiding time-consuming data re-collection.
2Loss of time
If a single learning model is used for various injection molding machines to reduce cost and time, then cost and time are reduced, but determination accuracy deteriorates due to divergence between measured values and learning data
Solution Approach 1:
The patent introduces machine specification parameters (motor rated torque, inertia, reduction gear reduction ratio) as intermediary elements between the learning data and the actual machine operation data. These parameters serve as mediators that enable numeric conversion of learning data to match different machine characteristics, allowing a single learning model to be adapted accurately to various injection molding machines without requiring separate models for each machine type.
3Measurement precision
If learning data is collected for each combination of machine components and auxiliary equipment to improve determination accuracy, then determination accuracy is improved, but cost and complexity increase significantly
Solution Approach 1:
The patent achieves universality by creating a single learning model that can be applied to various injection molding machines through numeric conversion. Instead of developing separate learning models for each machine type and configuration, the system uses one universal learning model that adapts to different machines by applying numeric conversion based on their specific specification parameters, thereby eliminating the need for multiple specialized models.
Data Source
AI summary
A state determination device acquires data related to an injection molding machine, stores a learning model obtained by learning an operation state of the injection molding machine with respect to the data, and performs estimation using the learning model based on the data. Further, the state determination device acquires a correction coefficient, which is associated with a type of the injection molding machine and equipment attached to the injection molding machine and numerically converts and corrects the estimation result with a predetermined correction function to which the acquired correction coefficient is applied.