Device Failure Monitoring Using Fleet and Individual Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques fail to accurately model the probability of failure for new or relatively new devices due to limited data, leading to uncertain predictions of their lifetime and potential failures.
Innovation Solution
A method that combines fleet measurement data with device-specific data to determine a probability of failure by calculating effective data measurements, utilizing both distributions to simulate future device behavior through statistical techniques like Monte Carlo simulations, and adjusting the weight of fleet versus device data as more device data becomes available.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If fleet measurement data is used to model new devices, then more data is available for analysis, but the accuracy of failure probability prediction deteriorates due to device individuality differences
Solution Approach 1:
The patent combines fleet measurement data from multiple devices with device-specific measurement data from the monitored device. The system merges these two data sources to create a hybrid dataset that leverages the quantity of fleet data while incorporating the individuality of the specific device, thereby improving prediction accuracy without sacrificing data availability.
Solution Approach 2:
The patent dynamically adjusts the weighting between fleet data and device-specific data based on the amount of available device data. As more device-specific measurements become available, the system transitions from relying primarily on fleet data to relying more on device-specific data, optimizing the balance between data quantity and prediction accuracy over time.
2Measurement precision
If device-specific measurement data is used for prediction, then individual device behavior is captured accurately, but the quantity of available data is insufficient for reliable statistical analysis
Solution Approach 1:
The patent merges device-specific measurement data with fleet measurement data from multiple similar devices. This combination allows the system to maintain the high precision of individual device modeling while supplementing it with the larger quantity of fleet data, enabling reliable statistical analysis even when device-specific data is limited.
Solution Approach 2:
The patent uses fleet data as an intermediary to bridge the gap when device-specific data is insufficient. The fleet data serves as a statistical proxy that provides additional information about device behavior patterns, which is then integrated with the available device-specific data to improve prediction reliability.
3Quantity of substance
If only fleet data is used for new devices, then data availability is high, but the reliability of prediction deteriorates due to lack of individual device characteristics
Solution Approach 1:
The patent combines fleet data with device-specific measurement data to create a hybrid approach. This merging ensures that the benefits of large data quantities from the fleet are maintained while incorporating individual device characteristics, thereby improving the reliability of failure predictions for new devices.
Solution Approach 2:
The patent applies preliminary action by using fleet data to establish initial prediction models for new devices before sufficient device-specific data is available. As device-specific data accumulates, the system updates and refines the predictions, transitioning from fleet-based preliminary models to device-specific accurate models.
4Quantity of substance
If fleet data weighting is maintained for new devices, then data utilization is maximized, but the adaptability to individual device behavior deteriorates
Solution Approach 1:
The patent implements a dynamic weighting mechanism that adjusts the relative importance of fleet data versus device-specific data over time. For new devices, fleet data is weighted higher to maximize data utilization. As device-specific data accumulates, the weighting dynamically shifts to increase the influence of device-specific data, thereby improving adaptability to individual device behavior while maintaining efficient data utilization throughout the device lifecycle.
Data Source
AI summary
A method for device monitoring includes: providing a plurality of fleet measurement data for at least one device, where the plurality of fleet measurement data includes a number of fleet device data measurements “Nfleet;” providing a plurality of device measurement data for a monitored device, where the plurality of device measurement data includes a number of monitor device data measurements “Ndata;” determining a fleet distribution “Fdist” determined as a distribution of differences between consecutive measurements of the plurality of fleet measurement data; determining a device distribution “Ddist” determined as a distribution of differences between consecutive measurements of the device measurement data; determining an effective number of data measurements “Neff” and determining a probability of failure of the monitored device, the determining including utilising the fleet distribution “Fdist,” the device distribution “Ddist,” the number of monitor device data measurements “Ndata,” and the effective number of data measurements “Neff.”