10th U6+ Consortium of African Universities International Conference

Pyrethrum (Tanacetum cinerariifolium) is a high-value industrial crop, primely cultivated for natural bioactive pyrethrin in the current global shift from synthetic pesticides. However, production remains constrained by manual, labour-intensive and selective harvesting of mature flower heads due to complex physico-mechanical characteristics, crop morphology and the necessity to preserve ratooning buds and perennial canopy. This work designed and simulated a small-scale mechanized pyrethrum harvester using crop physico-morpho-mechanical characteristics to optimize plucking efficiency, field capacity, and harvesting quality using artificial neural networks (ANN) and fuzzy logic optimizer. Results indicated that while a wide range of traits influence plant structure and post-harvest behaviour, only a few were critical for mechanized-oriented optimization. Mature stalk height (MSH) emerged as the dominant morphological trait influencing mechanised harvesting, while floral head diameter (FHD) was identified as the most important trait affecting plucking efficiency. However, plucking force (Pf) was variable across sites, while the cutting force (Cf) and coefficient of friction (μS) were established as design constraints rather than optimization variables. MSH and FHD predominantly influenced machine geometry and post-primary handling, while canopy width (FCW) and repose angle (ϴ) had post-harvest handling significance. The quantile-based analysis revealed the overall harvest efficiency, with the 70th percentile and a 2.5 mm tolerance emerging as the most favourable configuration, corresponding to a blade gap of approximately 10.33 mm and a harvest rate of 100%. Fuzzy logic optimization models delineated feasible cutting height ranging between 392.6 mm and 499.5 mm and ensured exclusion of non-target foliage at 70% harvester efficiency threshold. ANN simulation optimization revealed a blade gap of 10.3 mm and a cutting height of 392-500 mm for practical deployment of the harvesting header. Further, field simulation revealed 58.4 kghr⁻¹ throughput and 0.09 hahr⁻¹ effective field capacity. The harvester demonstrated preserved floral integrity at 76% harvesting efficiency and a damage rate of 5%. These findings provide optimized design and operational settings for mechanized pyrethrum harvesting across diverse field conditions.