CAJ | 학술논문

以物联网为代表的现代信息技术在农机作业管理领域的发展应用，实现了农机作业过程的定位监控，但现有农机远程监管系统对海量农机空间位置数据仅实现了远程存储、显示和简单分析，难以满足农机精准管理和数据智能处理的要求。该文采用数据挖掘中的聚类和空间数据分析方法，结合农机空间运行轨迹的特点，研究了基于空间运行轨迹点的农机作业状态自动识别算法；设计实现了典型农机运行状态自动识别方法，定量分析了农机作业班次内田间作业时间、空行转移时间、停歇时间的量化构成。农机试验表明：发展的基于空间索引和网格密度的聚类算法精度达89%以上。农机作业状态自动识别为农机作业生产率、农机利用率和作业成本核算提供了定量依据。
이물련망위대표적현대신식기술재농궤작업관리영역적발전응용，실현료농궤작업과정적정위감공，단현유농궤원정감관계통대해량농궤공간위치수거부실현료원정존저、현시화간단분석，난이만족농궤정준관리화수거지능처리적요구。해문채용수거알굴중적취류화공간수거분석방법，결합농궤공간운행궤적적특점，연구료기우공간운행궤적점적농궤작업상태자동식별산법；설계실현료전형농궤운행상태자동식별방법，정량분석료농궤작업반차내전간작업시간、공행전이시간、정헐시간적양화구성。농궤시험표명：발전적기우공간색인화망격밀도적취류산법정도체89%이상。농궤작업상태자동식별위농궤작업생산솔、농궤이용솔화작업성본핵산제공료정량의거。
With development and wide application of modern information technology as presented by the Internet of things, the position monitoring of operation process of agricultural machinery has been realized easily. But the existing remote monitoring system for agricultural machinery has only realized the remote storage, display and simple analysis. It is difficult to meet the requirements of fine management and intelligent data processing of agricultural machinery. In this paper, combining with the characteristics of the space track of agricultural machinery, the methods of clustering in data mining and spatial data analysis method were used to study automatic segmentation algorithm of field operation area based on spatial track of agricultural machinery. The procedure is as follows: firstly, data preprocessing, including velocity threshold and projection transformation methods, is preparing for further gridding and density slicing for the next step; secondly, spatial track of agricultural machinery was gridded; thirdly, density slicing removed low density cell-grids and preserved high density cell-grids; fourthly, spatial partition tree and spatial index are constructed, which is able to accelerating query speed of spatial nearest neighbors; finally, the cluster analysis is executed to connect high density cell-grids adjacently. The automatic identification of typical agricultural machinery operation method was designed and achieved. The quantitative analysis of the agricultural machinery operation divisions within the field operation time, transfer time and idle time of the agricultural machinery operation were divided and analyzed quantitatively. From May 27, 2012 to June 20, 2012, which is the season of wheat harvest, in order to verify the state of agricultural machinery operation cut automatic identification algorithm, the agricultural machinery tests were carried out in the Shijia Agricultural Machinery Cooperation in Xuchang City, Henan Province. Nine wheat harvesters, installing the embedded vehicle terminal based on GPS (positioning accuracy: 10 m, velocity measurement precision: 0.1 m/s) and GPRS (general packet radio service), were selected to do real harvesting task to produce spatial track data of agricultural machinery. The data collected by embedded vehicle terminal was used to verify the state of agricultural machinery operation cut automatic identification algorithm above. It showed that the accuracy of clustering algorithm based on spatial index and density slicing was above 89% in agricultural machinery test. After comparing clustering algorithm used in this paper with K-Means clustering algorithm and DBSCAN clustering algorithm, it is found that the clustering algorithm used in this paper has the best time efficiency, and the K-Means algorithm is better, and the DBSCAN algorithm has the worst run-time efficiency. It is shown that the construction of spatial partition tree and spatial index is very effective to accelerating query speed of spatial nearest neighbors, so it improves the clustering algorithm efficiency. But the current algorithm model is sensitive to the parameters of grid size and density threshold. The parameter setting depends on statistical analysis of spatial track of agricultural machinery. The grid size and density threshold are closely related to equipment width, working speed of agricultural machinery, and the frequency of GPS data uploading. The study direction in future is how to establish the quantitative relationship among the above parameters.