Harvester Fill-Level Prediction Using Yield-Corrected Path Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current combine harvesters face inefficiencies due to difficulty in predicting when the grain tank will be full, leading to idle time and suboptimal deployment of harvesting machines and haulage units, as existing yield prediction methods are inaccurate and fail to account for real-time data effectively.
Innovation Solution
An agricultural harvesting machine equipped with a path processing system that uses a priori geo-referenced vegetative index data and historical data to predict yield, combined with real-time sensor data to generate a georeferenced probability distribution of fill capacity, allowing for optimized routing and operation to minimize idle time and enhance logistical efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If yield prediction is based only on a priori geo-referenced vegetative index data or historical data, then the prediction can be made in advance for planning purposes, but the accuracy is insufficient due to large error ranges and failure to account for real-time conditions
Solution Approach 1:
The system continuously monitors actual yield data from yield sensors during harvesting operations and feeds this information back to update the yield prediction model in real-time. This feedback mechanism allows the system to correct prediction errors and adapt to actual field conditions, significantly improving prediction accuracy while maintaining the ability to provide timely unloading predictions to reduce idle time.
Solution Approach 2:
The system performs preliminary yield predictions using a priori geo-referenced vegetative index data and historical data before harvesting begins. These preliminary predictions provide advance planning information for deploying haulage units, while the system is prepared to update predictions in real-time as harvesting progresses and actual yield data becomes available.
2Productivity
If multiple haulage units are deployed without accurate real-time yield prediction, then coverage of the field can be maintained, but inefficiency increases due to wrong haulage units going to wrong harvesters and operators not knowing when capacity is reached
Solution Approach 1:
The system acts as an intermediary information hub that collects real-time yield data from multiple harvesters, processes this information through the yield prediction model, and distributes accurate predictions to the appropriate haulage unit operators. This intermediary function ensures that each haulage unit receives specific information about which harvester will be full first and when, enabling optimized routing and coordination without requiring operators to manually monitor multiple harvesters.
3Loss of time
If the harvester operates without real-time yield correction, then the system complexity is reduced, but the ability to accurately predict fill capacity and optimize unloading timing is compromised
Solution Approach 1:
The system replaces complex mechanical monitoring and manual calculation methods with an electronic information processing system. Yield sensors automatically collect data, processors continuously update predictions using algorithms that integrate a priori data, historical data, and real-time measurements, and communication systems automatically transmit predictions to haulage units. This substitution of mechanical/manual processes with electronic automation manages system complexity while enabling real-time yield correction and accurate fill capacity prediction.
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
An agricultural harvesting machine (100) comprises a path processing system (174) that obtains a predicted crop yield at a plurality of different field segments along a harvester path on a field, and obtains field data corresponding to one or more of the field segments generated based on sensor data as the agricultural harvesting machine is performing a crop processing operation. A yield correction factor is generated based on the received field data and the predicted crop yield at the one or more field segments. Based on applying the yield correction factor to the predicted crop yield, a georeferenced probability metric is generated indicative of a probability that the harvested crop repository will reach the fill capacity at a particular geographic location along the field. A control signal generator (176) generates a control signal to control the agricultural harvesting machine based on the georeferenced probability metric.