(178c) Event-Based Optimisation for Harvesting Scheduling Under Precision Agriculture

Building on digital technologies (Internet of Things (IoT), big data), Agriculture 4.0 has the potential to enhance precision farming and improve farm system responsiveness and resilience. As part of IoT driven precision solutions, the increasing smart agricultural machinery is expected to catalyse the agriculture sector transformation by shifting the isolated farms towards performance-optimised and vehicle-interconnected agro-ecosystems. Despite the importance, development and application of computer-aided decision-making tool to optimise the precision operations by interconnected smart machinery remains largely unexplored. This study presents optimisation research under the Agriculture 4.0 vision with a particular focus of harvesting operation for a perennial crop, sugarcane.

Sugarcane is one of the most important commodities globally[1]; its production increased fourfold since 1965, achieving 1842 million metric tons in 2017[2]. Notably, BRICS countries representing the growing economy plays significant roles in global sugarcane market i.e. South Africa, Brazil, India and China. Sugarcane represents a highly perishable agro-product due to cane quality deterioration which can be affected by variety, harvesting, and pre- or post-harvest operations[3], thus increase the system complexity involving coordinated planning and scheduling[4]. Thereby, decision-making support to optimize operational planning and to estimate the harvest process quality will be of benefit to the agricultural sectors.

Almost all the mathematical models developed for the planning and scheduling of sugarcane harvesting are based on discretised time interval, which only provides a reference grid of time for all operations (e.g. harvesting, relocation) competing for shared resources (i.e. harvesters). For harvesting operations, the varying geometries of different fields, lead to variation in harvesting time; however, the discrete-time approaches can only provide approximate descriptions of the actual process, which can deviate substantially from the true solutions [5]. The real-time decision making and accurate optimisation solutions over continuous time horizon are of particular importance for coordinated operations and interconnected smart machineries with real-time communication (Agriculture 4.0 vision); this research challenge remains largely unexplored.

In this study we propose an event-based continuous-time formulation to optimise short-term sugarcane harvesting operations, accounting for the interconnected agro-system performances. To demonstrate the model functionality, we present a hypothetical case study based on sugarcane operation using smart harvester at Kwazulu Natal region in South Africa. KwaZulu Natal is a typical sugarcane farm region with an annual water precipitation of higher than 200 mm (historical statistics), which offers desirable climatic conditions for non-irrigated sugarcane growth. The model is formulated in Python 3.7 using Pyomo 5.6.2. In this study, we present a heuristics-based hybrid approach to solve the sugarcane harvesting scheduling problems, which significantly reduced computational time to enable real-time decision-making. Our research highlights the continuous-time algorithm development and applications in precision agriculture to provide accurate optimisation solutions for real-time scheduling of field operations, which are operated in a coordinated manner through smart machineries.

REFERENCES

1. Jena, S.D. and M. Poggi, Harvest planning in the Brazilian sugar cane industry via mixed integer programming. European Journal of Operational Research, 2013. 230(2): p. 374-384.
2. Statistica, World sugar cane production from 1965 to 2017 (in million metric tons). 2018.
3. Solomon, S., Post-harvest deterioration of sugarcane. Sugar Tech, 2009. 11(2): p. 109-123.
4. Kusumastuti, R.D., D.P.v. Donk, and R. Teunter, Crop-related harvesting and processing planning: a review. International Journal of Production Economics, 2016. 174: p. 76-92.
5. Floudas, C.A. and X. Lin, Continuous-time versus discrete-time approaches for scheduling of chemical processes: a review. Computers & Chemical Engineering, 2004. 28(11): p. 2109-2129.