Cost Constrained Integrated Management for Multiple Farm Fields in Agriculture for Areas without Historical Data

There is a lot of previous research on profit maximization in agriculture. In a previous research, crop rotation problem and cultivation management problem are solved as one integrated management problem for multiple farm fields in order to maximize the profit in agriculture under cost constraints. In the previous research, the profit is maximized under the condition that probabilities are known. Sufficient historical data is required to estimate probabilities. The previous method cannot be applied to areas where there is no historical data. In this research, a new integrated management method for areas where there is no historical data is proposed. Integrated management is modeled by Markov decision processes with unknown probabilities. The proposed method maximizes profit with reference to a Bayes criterion by dynamic programming. In the numerical calculation results, adaptive crop selection according to posterior probabilities and adaptive cultivation action selection according to cost constraints are confirmed. This research is fundamental research, and future extended research is required.