Machine Diagnostics Using Context-Guided Signal Separation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In industrial settings, diagnostic apparatuses face challenges in accurately separating vibration and sound data from a target machine from those of other mechanical apparatuses, leading to difficulties in detecting anomalies with high precision.

Innovation Solution

A diagnostic apparatus and method that include a context information acquisition unit, a sensing information acquisition unit, a sensing information separation unit, a feature information extraction unit, and an anomaly determination unit, which acquire and separate physical quantity information from multiple mechanical apparatuses, extract feature information, and determine if the target machine's operation is normal based on context-specific models.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a sensing unit detects physical quantity in a factory building with multiple mechanical apparatuses, then the sensing unit can detect vibrations from all apparatuses, but the detection precision for the target machine deteriorates due to mixed signals from other apparatuses

Engineering Contradiction:
Improveanomaly detection precisionVSAvoidsignal contamination
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The patent segments the mixed vibration signal into multiple components corresponding to different mechanical apparatuses. The separation unit divides the composite signal detected by the sensing unit into individual apparatus signals based on characteristic frequencies and operational patterns, enabling precise anomaly detection for the target machine without interference from other apparatuses.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces context information as an intermediary element that mediates between the mixed signal and the target machine identification. By acquiring operational context (such as which apparatus is currently operating) and using it to guide the separation process, the system can accurately attribute vibration components to their respective sources even in complex multi-apparatus environments.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If context information is acquired from multiple mechanical apparatuses to separate physical quantity information, then the anomaly detection accuracy improves, but the system complexity increases

Engineering Contradiction:
Improvephysical quantity separation accuracyVSAvoidsystem structure complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent implements a universal diagnostic system that can handle multiple mechanical apparatuses with different types and operations. The context information acquisition unit and separation unit are designed to work with various apparatus configurations, making the system adaptable to different factory layouts and machine types without requiring separate dedicated systems for each apparatus.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

Each mechanical apparatus provides its own context information (operational status, type, current operation) to the diagnostic system. This self-service approach reduces the burden on the central diagnostic apparatus, as the distributed context information from each machine enables automatic signal separation without requiring complex centralized control or manual configuration.

Inventive Principle:
Principle #25Self-service

Data Source

PatentUS11385617B2Diagnostic apparatus and diagnostic method
Publication Date: 2022.07.12 FANUC LTD
  • US11385617B2 patent drawing
  • US11385617B2 patent drawing
  • US11385617B2 patent drawing

AI summary

A diagnostic apparatus includes a context information acquisition unit that acquires context information of operation being performed from a plurality of mechanical apparatuses, a sensing information acquisition unit that acquires physical quantity sensing information sensed by each of sensing units of a target machine and other mechanical apparatuses, a sensing information separation unit that separates the physical quantity information of the target machine and the physical quantity information of the other mechanical apparatuses from each other based on the context information and the sensing information of each of the plurality of mechanical apparatuses, a feature information extraction unit that extracts feature information of the physical quantity information, and an anomaly determination unit that determines, from a model corresponding to the acquired context information and the extracted feature information, whether or not operation of the target machine is normal.