Journal of Integrative Agriculture ›› 2017, Vol. 16 ›› Issue (02): 252-265.DOI: 10.1016/S2095-3119(16)61430-2

• 论文 • 上一篇    下一篇

  

  • 收稿日期:2016-04-05 出版日期:2017-02-20 发布日期:2017-01-31

An adaptive Mealy machine model for monitoring crop status

Berk Üstünda?    

  1. Agricultural and Environmental Informatics Application & Research Center/Faculty of Informatics and Computer Engineering, Istanbul Technical University, Istanbul 34469, Turkey

  • Received:2016-04-05 Online:2017-02-20 Published:2017-01-31
  • Contact: Berk üstünda?, Tel: +90-212-2856702, E-mail: bustundag@itu.edu.tr
  • Supported by:

    This study was funded by Turkish Ministry of Development as a part of Agricultural Monitoring and Information Systems Project (2011A020100) and the relevant joint project funded by Ministry of Food, Agriculture and Livestock, Turkey.

Abstract: Variation in phenological stage is the major nonlinearity in monitoring, modeling and various estimations of agricultural systems.  Indices are used as a common means of evaluating agricultural monitoring data from remote sensing and terrestrial observation systems, and many of these indices have linear characteristics.  The analysis of and relationships between indices are dependent on the type of plant, but they are also highly variable with respect to its phenological stage.  For this reason, variations in the phenological stage affect the performance of spatiotemporal crop status monitoring.  We hereby propose an adaptive event-triggered model for monitoring crop status based on remote sensing data and terrestrial observations.  In the proposed model, the estimation of phenological stage is a part of predicting crop status, and spatially distributed remote sensing parameters and temporal terrestrial monitoring data are used together as inputs in a state space system model.  The temporal data are segmented with respect to the phenological stage-oriented timing of the spatial data, so instead of a generalized discrete state space model, we used logical states combined with analog inputs and adaptive trigger functions, as in the case of a Mealy machine model.  This provides the necessary nonlinearity for the state transitions.  The results showed that observation parameters have considerably greater significance in crop status monitoring with respect to conventional agricultural data fusion techniques.

Key words:  plant phenology, monitoring, yield prediction, finite automata, Mealy machine, remote sensing