Scheduling for Farm Work Planning based on Petri Net Model and Simulated Annealing

Abstract

A practical farm work planning is important for agricultural production corporations to manage the farming work systematically and efficiently. Since the farm work planning corresponds to a NP hard scheduling problem, a satisfactory and rational farm work planning is difficult to be generated by conventional approaches. This paper proposes a new concept of applying the Petri net and simulated annealing (SA) algorithm to develop a farm work planning for the agricultural production corporations. The necessary data for farm work planning including the data of farming process and changes of uncertainties is recorded by a cellular phone equipped with a GPS (Global Positioning System) function and an Internet connection. Petri net mathematically and graphically describes the farming process, and simulates the farming activation and resource allocation. The marking that is one of properties of Petri net facilitates mastering the farming progress and the online status of the farmland and resources. According to the formulation of farming process, we developed the SA algorithm to obtain high-quality solutions approximated to the optimum solution for the farm work planning. In the experimental evaluation on the simulation data between the SA algorithm and conventional local search algorithm, the scheduled length by the SA algorithm was much shorter than that by the local search algorithm. The result revealed that the SA algorithm had high superiority to generate high-quality solution for the farm work planning.