Undercarriage Wear Prediction Using Physics-Based Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies face challenges in effectively predicting wear of undercarriage components of machines, such as excavators, due to the complexity of wear patterns and the need for integrating diverse data sources for reliable wear prediction.
Innovation Solution
A system and method that collects high-frequency time series data, applies physics-based feature engineering, and establishes a statistical model using machine learning to predict wear conditions of undercarriage components by deriving features and determining coefficients based on historical data and inspection data from multiple machines.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional wear prediction methods are used, then the system complexity is low, but the prediction accuracy and reliability are insufficient
Solution Approach 1:
The patent segments the undercarriage into multiple components (track links, rollers, sprockets, shoes) and collects data separately for each component. The wear prediction model is also segmented to handle different component types with their specific wear patterns and failure modes, improving prediction accuracy while maintaining manageable system complexity through modular design
Solution Approach 2:
The patent introduces an intermediary data processing layer that collects, cleans, and transforms raw sensor data into meaningful features. This intermediary layer includes data normalization, feature extraction, and preprocessing functions that bridge the gap between raw sensor data and the prediction model, improving reliability without directly increasing model complexity
2Reliability
If multiple data sources are integrated for wear prediction, then the prediction reliability improves, but the data collection and processing complexity increases
Solution Approach 1:
The patent implements a universal data collection framework that can handle multiple data sources (sensors, historical records, operational parameters) through a single integrated system. The data processing pipeline is designed to accommodate various data types and sources uniformly, improving reliability through multi-source integration while avoiding the complexity of separate processing systems for each data source
Solution Approach 2:
The patent incorporates feedback mechanisms where wear predictions are compared with actual wear measurements from inspections. This feedback loop allows the system to continuously improve its accuracy by adjusting prediction models based on real-world data, enhancing reliability while managing data processing complexity through iterative refinement rather than requiring perfect initial data collection
3Measurement precision
If high-frequency time series data is collected, then the wear prediction precision improves, but the data processing requirements and system complexity increase
Solution Approach 1:
The patent extracts and isolates the most critical features from the high-frequency time series data that are directly related to wear mechanisms. Instead of processing all raw data, the system identifies and extracts key parameters (load, speed, operational mode) that have the most significant impact on wear, improving measurement precision while reducing data processing complexity by focusing only on essential features
Solution Approach 2:
The patent transforms high-frequency time series data into aggregated statistical parameters and features that capture wear trends over time. By changing the representation of data from raw high-frequency measurements to processed features (mean, variance, trend parameters), the system maintains measurement precision while significantly reducing data processing requirements and system complexity
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present disclosure is directed to systems and methods for wear prediction of an undercarriage of a target machine. The method includes (1) receiving wear measurements from a plurality of source machines, and the wear measurements are associated with a first set of components of undercarriages of the plurality of source machines; (2) establishing a statistical model based on the received wear measurements and physic-based features derived from the wear measurements; (3) determining coefficients for the statistical model at least partially based on inspection data of a second set of components of the undercarriage of the target machine; and (4) predicting a wear condition of the undercarriage of the target machine by the statistical model and the coefficients.