Sugarcane harvest planning with yield and quality consideration

Abstract

In this paper, the sugarcane harvesting problem with sugarcane yield and sugar content (Commercial Cane Sugar: C.C.S.) consideration is addressed. The sugarcane yield and its quality are varied along its age and based on the growing environments. The farmer’s revenue is depended on the weight and C.C.S. of sugarcane they sold. Generally, cane weight and C.C.S. may not follow the same pattern. At a specific period, cane may reach high weight but low sugar content. Farmer normally pays more attention to yield than C.C.S even though mills are willing to pay an extra for sugarcane with high C.C.S. A framework which is an integration of artificial neural network (ANN) approach and mathematical model is proposed to solve the problem. Two sets of ANN models are developed to forecast sugarcane yield and C.C.S., separately. The mathematical model is developed to determine an optimal harvesting plan in order to maximize farmer revenues. The data used in this study are collected from one of the largest sugar mills in Thailand. The results showed that both ANN models are quite suitable in yield and C.C.S. forecasting with the average of mean absolute percent errors (MAPE) 11.30 % and 6.86 %, respectively. Farmer revenues could be raised up 11,847.86 Baht by this proposed plan. The optimal harvesting plan with maximizing farmer revenues can be obtained with reasonable computational time for the small size problems.