go-v.orgdoi

Enhancing Combine Harvester Performance: A Data-Driven Recommendation System for Improving Initial Machine Settings Based on Operational Context

Altaleb, Mohamed · osnaDocs (Osnabrück University) · 2024

  • DOI 10.48693/574
  • Publisher Osnabrück University
  • Type dissertation
  • Issued 2024

Not cached yet. This record was fetched live for this request and is stored after the lookup. Reopening the page should serve the cached copy.

osnaDocs (Osnabrück University) · Osnabrück University · dissertation · 2024

Abstract

In agriculture, optimizing machinery is crucial; therefore, modern combine harvesters provide an onboard system with a closed-loop to optimize performance, increasing efficiency and profitability. The harvesters begin with initial settings recommended by the manufacturers as factory defaults. These settings are refined and optimized as the harvesting progresses according to the harvesting conditions, which constitute the operational context. Diverse sectors, including e-commerce, utilize big data to develop recommendation systems from historical data. Modern combines also gather big data, allowing advancements through utilizing the optimized settings that likely exist within this data.

This work aims to develop a recommendation system to propose suitable initial settings for the combine harvester using data mining techniques. There are several generic process models to guide through the standard activities of performing data mining; however, each industrial domain has its own characteristics and traits. This research develops supplementary materials to extend a standard data mining process model to cover the essential characteristics of the agricultural machinery domain. These materials are utilized to develop several recommendation models using machine learning and analytics approaches.

An influencing factors schema has been defined based on the combine operational context. The research has analyzed and identified correlations between combine settings and various factors such as harvest strategy, weather, climate regions, and more. Various recommendation models are developed, utilizing data from thousands of machines. The models are evaluated using three approaches; all have favored these models over the traditional system. The models have enhanced combine performance by reducing losses and, broken grain and increasing cleanliness. This is achieved by adjusting the settings to align with harvest conditions rather than using generic conditions.

Open the resolved locator · 10.48693/574

Metadata licence: OpenAlex metadata (CC0)