Robot Condition Monitoring via Manifold Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current mechanical condition monitoring algorithms are operation-dependent, requiring similar robotic operations for comparison, which impedes efficient operation and can lead to robot breakdowns if maintenance monitoring is forgone to increase uptime.
Innovation Solution
A condition monitoring device that uses manifold alignment to project operational data onto a common subspace with baseline data, allowing for independent operation monitoring by distinguishing between changes in health and operation, employing techniques like principal component analysis and short-time Fourier transform for feature extraction, and unsupervised domain adaptation methods like low-rank alignment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If operation-dependent monitoring algorithms are used to ensure accurate health assessment, then measurement precision is improved, but productivity deteriorates due to required operational cessations for comparison
Solution Approach 1:
The patent segments the monitoring process into offline baseline establishment and online health assessment phases. The complex operation-dependent comparisons are performed offline during baseline creation, while online monitoring uses simplified metrics that don't require operational cessations. This segmentation allows accurate health assessment without sacrificing productivity during actual robot operation.
Solution Approach 2:
The patent performs preliminary action by establishing comprehensive baseline models offline before actual production operations begin. All the complex computations, model training, and reference data creation are completed in advance, so that during production the system can quickly compare against pre-established baselines without requiring operational stoppages, thus maintaining both accuracy and productivity.
2Productivity
If maintenance monitoring is forgone to increase uptime, then productivity is improved, but reliability deteriorates due to potential robot breakdowns
Solution Approach 1:
The patent enables the robot to essentially self-monitor its own health condition through automated sensing and comparison against baseline models. The system continuously assesses health metrics without requiring external intervention or operational cessations, allowing the robot to maintain both high uptime and reliability through autonomous health monitoring during production operations.
3Productivity
If operation-independent monitoring is implemented to maintain continuous operation, then productivity is improved, but measurement precision deteriorates due to inability to distinguish health changes from operation changes
Solution Approach 1:
The patent extracts and isolates health-related metrics from operation-dependent variables by identifying and removing operation-specific features from the baseline model. This allows the system to focus on health-relevant signals while filtering out operation-related variations, enabling continuous monitoring without sacrificing the ability to accurately detect actual health changes.
Solution Approach 2:
The patent transforms the monitoring parameters by changing from operation-specific metrics to operation-independent health indicators. By selecting and weighting parameters that are insensitive to operational variations but sensitive to health degradations, the system achieves both continuous operation capability and accurate health change detection through parameter transformation and selection.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Mechanical condition monitoring of robots can be used to detect unexpected failure of robots. Data taken from a robot operation is processed and compared against a health baseline. Features extracted during the monitoring stage of robot operation are aligned with features extracted during the training stage in which the health baseline is established by projecting both onto a common subspace. A classifier which can include a distance assessment such as an L2-norm is used within the common subspace to assess the condition of the robot. Excursions of the distance assessment from a criteria indicate a failure or potential failure.