Robot Joint Fault Detection via Torque Distribution Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current condition monitoring and fault detection methods for industrial robots are inadequate due to the difficulty in constructing accurate friction models that account for speed, load, temperature, and wear, leading to high maintenance costs and inefficiencies.
Innovation Solution
A method that utilizes torque measurements and kernel density estimators to detect faults by comparing distribution characteristics of torque values obtained under recurring conditions, with filtering and weighting techniques to enhance fault indication, allowing for more robust condition monitoring.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If friction monitoring is used to detect wear in robot joints, then wear detection capability is improved, but measurement precision deteriorates because accurate friction models accounting for speed, load, temperature and wear are difficult to construct
Solution Approach 1:
The patent replaces the mechanical friction model-based approach with a statistical distribution-based approach. Instead of using a physics-based friction model that requires accurate parameters for speed, load, temperature and wear, the invention uses kernel density estimators to create probability distributions from actual torque measurement data. This substitution eliminates the need for complex friction modeling while maintaining wear detection capability.
Solution Approach 2:
The patent introduces distribution characteristics (kernel density estimators) as an intermediary between raw torque measurements and wear detection. Rather than directly comparing torque values or using friction models, the invention compares distribution characteristics of torque measurements taken under recurring conditions. This intermediary layer filters out variations due to speed, load and temperature while preserving wear-related changes.
2Reliability
If preventive scheduled maintenance is implemented to improve equipment safety and reliability, then availability is improved, but maintenance costs increase due to unnecessary maintenance actions
Solution Approach 1:
The patent implements a feedback-based condition monitoring system that continuously compares distribution characteristics of torque measurements against a reference distribution. When the comparison exceeds a threshold, it indicates actual wear and triggers maintenance. This feedback mechanism replaces scheduled maintenance with condition-based maintenance, ensuring maintenance is performed only when actually needed based on real wear indicators.
Solution Approach 2:
The system enables the robot joint to essentially monitor its own condition through torque measurements and distribution comparisons. The automatic detection of wear through statistical analysis of operational data allows the system to self-diagnose its condition, eliminating the need for external scheduled maintenance planning and reducing unnecessary maintenance interventions.
3Loss of energy
If condition based maintenance is implemented to reduce overall costs, then maintenance costs are reduced, but the challenge of defining methods to determine equipment condition automatically increases
Solution Approach 1:
The patent changes the parameter used for condition monitoring from physical friction model parameters (speed, load, temperature coefficients) to statistical distribution parameters (kernel density estimator characteristics). This parameter transformation simplifies the condition determination method by replacing complex physics-based parameters with data-driven statistical parameters that are automatically computed from torque measurements.
Solution Approach 2:
The invention substitutes complex friction modeling with statistical distribution analysis. Instead of implementing complex mechanical models that require multiple parameters and assumptions, the system uses kernel density estimation to directly analyze torque measurement distributions. This substitution dramatically reduces the complexity of defining condition determination methods while maintaining effectiveness.
Data Source
AI summary
A method for detecting a fault in a robot joint includes the steps of: performing a first torque measurement at the robot joint to thereby obtain a first set of torque values; calculating a first distribution characteristic reflecting a distribution of the first set of torque values; performing a second torque measurement at the robot joint to thereby obtain a second set of torque values; calculating a second distribution characteristic reflecting a distribution of the second set of torque values; and comparing the first and the second distribution characteristics to determine whether a fault is present or not. A difference in the distributions of torque measurements is a robust fault indicator that makes use of the repetitive behavior of the system.