Harvest Particle Length Detection Using Real-Time Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining particle lengths in agricultural harvesters are inefficient, often requiring manual intervention or laboratory analysis, and struggle with chaotic material flows, leading to inaccuracies and long wait times.
Innovation Solution
A machine learning-based method using a neural network to identify particles with excessive lengths in real-time, reducing computational intensity by focusing on specific particle characteristics and optimizing image processing to enable real-time control of agricultural harvesters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional image processing methods are used to determine particle lengths, then the system can analyze particle sizes, but the computational intensity is too high for real-time processing and manual intervention is required
Solution Approach 1:
The patent extracts only the essential information needed for particle length determination from the complete image data. Instead of processing all pixel data and performing comprehensive image analysis, the system extracts minimal sufficient features (particle boundaries, length dimensions) using simplified algorithms that maintain measurement accuracy while dramatically reducing computational load for real-time processing
Solution Approach 2:
The patent segments the image processing task into discrete, manageable steps: identifying particle locations, extracting particle boundaries, measuring lengths along flow direction, and filtering results. This segmentation allows each step to be optimized independently and processed efficiently in real-time, avoiding the need for complex simultaneous analysis of all image data
2Quantity of substance
If comprehensive image analysis is performed to determine all particle sizes, then measurement completeness is improved, but computing requirements increase significantly
Solution Approach 1:
The patent applies different processing qualities to different parts of the analysis based on their importance. Instead of uniformly processing all particles with maximum detail, the system focuses computational resources on particles that meet specific criteria (e.g., particles exceeding threshold lengths) while using simpler processing for other particles, reducing overall energy consumption while maintaining quality where needed
Solution Approach 2:
The patent performs partial analysis sufficient for the specific application requirement rather than complete analysis of all particles. The system analyzes particles to determine if they exceed harmful length thresholds, which is the partial action needed for quality control, rather than measuring every particle dimension with full precision, thereby reducing computational energy requirements
3Measurement precision
If manual sampling and laboratory analysis are used to determine particle lengths, then measurement accuracy can be achieved, but time consumption increases significantly
Solution Approach 1:
The patent implements self-service analysis where the harvester system itself performs particle length measurement using onboard cameras and processing units during operation. Instead of requiring external laboratory analysis, the system autonomously captures images, processes them to determine particle lengths, and provides real-time feedback, eliminating time loss from sample collection, transport, and laboratory processing
Solution Approach 2:
The patent performs particle length measurement in advance during the harvesting operation itself rather than after the fact. By continuously monitoring particle lengths in real-time during harvest, the system can immediately identify and address issues with excessively long particles through on-the-spot adjustments to cutter head or cracker settings, eliminating the time delay inherent in post-harvest laboratory analysis
4Manufacturing precision
If the cutter head and cracker settings are adjusted to reduce excessively long particles, then particle quality improves, but the complexity of controlling the harvesting process increases
Solution Approach 1:
The patent implements feedback control where particle length measurements from image analysis are fed back to the control system, which automatically adjusts cutter head or cracker settings to reduce excessively long particles. This closed-loop feedback mechanism simplifies control complexity by using automated real-time adjustments based on measured particle characteristics rather than requiring complex manual control procedures
Solution Approach 2:
The patent replaces complex mechanical adjustment mechanisms with automated control systems that use image processing data to drive actuators for adjusting cutter head or cracker settings. This substitution of mechanical control with automated electro-mechanical control simplifies the overall system by using electronic sensing and actuation rather than complex mechanical linkages for adjustment
Data Source
AI summary
A method and system for identifying, using a computing unit, lengths of a particle in a material flow is disclosed. The computing unit is configured to analyze images of the material flow (13) in an analytical routine and to derive particle lengths of particles of the material flow contained in the images. The particle length derived in the analytical routine may comprise an excess length, with the analytical routine being based on a machine learning method trained to find or identify particles with the excess length.