Asset Anomaly Prediction Using Condition-Aware Sensor Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional asset monitoring systems fail to detect anomalies effectively due to the lack of consideration for varying operation conditions and interdependencies between sensor data from different components and devices.
Innovation Solution
A multi-tenant cloud solution that generates features accounting for different operation conditions and relationships between sensor data, using machine learning models to determine anomaly scores and predict faults, enabling intelligent detection and prediction of asset anomalies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional sensor alarm limits are set based on normal operating range values, then the system is simple to operate, but it fails to detect anomalies under varying operation conditions
Solution Approach 1:
The patent implements dynamic anomaly detection by training machine learning models on historical sensor data that captures various operating conditions. The models learn normal patterns across different operational states and dynamically adapt to detect anomalies, replacing static alarm limits with adaptive, condition-aware detection mechanisms.
Solution Approach 2:
The system transforms raw sensor data into meaningful features through data processing and feature engineering. These transformed parameters are then used to train machine learning models that can detect anomalies more accurately by understanding relationships between multiple sensor parameters rather than relying on single-parameter thresholds.
2Measurement precision
If sensor alarm limits are set without considering interrelationships between components, then the detection method is simple, but it fails to detect anomalies due to interdependencies
Solution Approach 1:
The patent combines data from multiple sensors and components into a unified machine learning model. By merging information across different sensor streams and considering their interrelationships, the system detects anomalies that result from complex interactions between components, rather than treating each sensor independently.
Solution Approach 2:
The machine learning model serves as an intermediary that processes and integrates data from multiple sensors. It learns the underlying relationships and dependencies between different sensor signals, acting as a mediator that understands how components interact and detects anomalies based on these learned relationships.
3Adaptability or versatility
If traditional anomaly detection methods are used, then the system is easy to implement, but it cannot detect anomalies across multiple assets with different specifications
Solution Approach 1:
The patent develops universal machine learning models that can be trained on historical data from multiple assets with different specifications. These models learn asset-type-specific patterns and can detect anomalies across diverse assets by understanding their unique operational characteristics, enabling a single system to handle multiple asset types effectively.
Solution Approach 2:
The system performs preliminary training phases using historical sensor data from multiple assets before deployment. During this preliminary action, the machine learning models learn from diverse operating conditions and asset specifications, preparing them to detect anomalies across different asset types without requiring complex asset-specific configuration during operation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Various embodiments described herein relate to intelligently detecting and predicting asset anomalies and faults. Such detection is enabled by generating feature that capture relationships between asset sensor values and different operation conditions or states of assets. In this regard, one or more features based at least on sensor data collected from a plurality of sensors associated with an asset are generated, and a data stream comprising data associated with the asset is received. An anomaly score for the data stream is then determined based at least on the one or more features. In accordance with determining whether the anomaly score is indicative of a potential fault of the asset, fault data indicative of the potential fault is generated, and presentation of the fault data and an indication of the one or more features considered in the determination of the anomaly score via the user interface is caused.