Context-Based Diagnostic Modeling for Abnormal Sound Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing diagnostic systems for image forming apparatuses and welding diagnostics face challenges in accurately identifying the source of abnormal sounds due to insufficient operation information, leading to incorrect diagnoses.
Innovation Solution
A diagnostic device and system that utilizes context information and sensing data to generate models for normal operation states, allowing for accurate determination of abnormal conditions by comparing real-time data against pre-learned models, and specifying the source of abnormalities with increased precision.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If operation timing chart is used to estimate abnormal sound generation source, then diagnostic function is provided, but diagnostic accuracy is insufficient due to inadequate operation information
Solution Approach 1:
The system performs preliminary learning to create operation models before actual diagnosis. Normal operation data is collected and stored as reference models, enabling accurate comparison during abnormality detection. This preliminary action ensures that sufficient operation information is available for accurate diagnosis.
Solution Approach 2:
The system creates copies of normal operation data in the form of operation models. These models replicate the characteristics of normal operations across different contexts, allowing the system to compare actual operations against these copied references to identify abnormalities accurately.
2Measurement precision
If simple operation timing chart is used, then device complexity is reduced, but measurement precision of abnormal sound source is insufficient
Solution Approach 1:
The operation information is segmented into multiple context types (operation mode, processing information, environment information). Each context is handled separately in the operation models, allowing precise measurement of abnormal sound sources while managing complexity through structured segmentation.
Solution Approach 2:
The system adds contextual dimensions to the operation information beyond simple timing. By incorporating operation mode, processing information, and environment information as additional dimensions, the system achieves higher measurement precision without excessive complexity increase.
3Reliability
If context information is incorporated into operation models, then diagnostic accuracy is improved, but device complexity increases
Solution Approach 1:
The operation models are designed to be universal across different contexts. A single model structure handles multiple context types (operation mode, processing information, environment information), reducing overall system complexity while maintaining high diagnostic accuracy through multi-functional modeling.
Solution Approach 2:
The system changes parameters within the operation models to reflect different contexts rather than creating entirely separate complex structures. By adjusting contextual parameters within a unified model framework, diagnostic accuracy is improved while device complexity is controlled.
Data Source
Figure 1~3
Figure 4~6
Figure 7
AI summary
A diagnostic device includes a reception unit and a determination unit. The reception unit is configured to receive context information and sensing information. The context information corresponds to a certain operation of a target item that constitutes a target device. The context information is a piece of a plurality of pieces of context information each describing an operation of the target item determined depending on a type of operation of the target device. The sensing information is on a physical quantity that varies in accordance with the operation of the target item. The determination unit is configured to determine a state of the target item based on the sensing information detected while the target item is performing the certain operation, and based on a model corresponding to the received context information. The model is a model of one or more models respectively defined for one or more pieces of the context information.