Engine Performance Prediction via Particle Image Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current engine performance prediction methods, such as optical atomic spectroscopy, lack repeatability and cannot analyze particles greater than 5 µm in diameter, failing to characterize individual particles effectively, and there is a need for improved methods to predict engine performance based on similarity with reference engines.
Innovation Solution
A zoning and profiling approach is used to extract features from fluid samples, including chemical composition and size/aspect ratio, to determine correlation indices between the target engine and reference engines, allowing for the prediction of performance similarity and likelihood of issues based on historical data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If optical atomic spectroscopy is used for fluid analysis, then elemental analysis of total oil sample is achieved, but repeatability among different equipment is poor and particles greater than 5 µm cannot be analyzed
Solution Approach 1:
The patent replaces optical atomic spectroscopy with image analysis technology that uses digital imaging to capture and analyze particle characteristics. This substitution enables direct visualization and measurement of individual particles, achieving both elemental analysis capability and improved repeatability across different equipment through standardized digital image processing
Solution Approach 2:
The patent creates digital copies (images) of particles for analysis instead of relying on spectroscopic measurements. By capturing digital images of particles and analyzing their characteristics through image processing, the system achieves consistent, repeatable measurements across different equipment while maintaining the ability to analyze particles of various sizes including those greater than 5 µm
2Quantity of substance
If optical atomic spectroscopy is used for fluid analysis, then total oil sample analysis is performed, but individual particles cannot be characterized
Solution Approach 1:
The patent segments the analysis process into two complementary approaches: analyzing the total oil sample to determine overall particle concentration and composition, and simultaneously analyzing individual particles through image capture to characterize their specific properties such as size, shape, and morphology. This segmentation preserves all information from both aggregate and individual particle perspectives
Solution Approach 2:
The patent adds a spatial dimension to the analysis by capturing images of individual particles, which provides dimensional information (size, shape, aspect ratio) that is lost in traditional spectroscopic analysis. This dimensional enhancement allows simultaneous characterization of both the total sample composition and individual particle properties
3Measurement precision
If particle analysis methods are developed to characterize individual particles, then particle size and shape information is obtained, but analysis of particles greater than 5 µm is still limited
Solution Approach 1:
The patent changes the measurement parameters by using image analysis technology that can detect and measure particles across a wide size range, specifically removing the 5 µm upper size limitation. The system captures digital images that allow measurement of particles from sub-micron to millimeter scales, and adjusts analysis parameters such as aspect ratio and size thresholds to accommodate different particle types and sizes
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
Systems and methods for predicting engine performance are provided. A fluid sample having particles suspended therein is received from a first engine (10). A plurality of particles are extracted from the fluid sample. Features of the plurality of particles extracted from the fluid sample (201) and features of particles of reference fluid samples from a plurality of reference engines are obtained (203). A plurality of correlation indices indicative of a level of correlation between the first engine and each one of the plurality of reference engines is determined (205). The correlation indices are compared to a threshold to determine a subset of the plurality of reference engines (207). Performance history for the engines in the subset is obtained (209). From the performance history, the first engine is determined as having a similarity in performance with the engines in the subset (210). An output is generated (211) indicating a predicted performance for the first engine (10).