智能农业控制系统能根据实时的农田环境数据、农作物生长情况,借助于数据挖掘技术、图像处理技术,结合专家决策库给出决策规则,自动控制农田设备,调节农田环境参数,控制农作物生长情况。智能农业控制系统如图所示。
图X: 智能农业控制系统
1)基于图像处理技术的农作物生长阶段判断系统
该系统利用图像处理技术,对农作物生长不同阶段的不同图像进行特征提取,建立农作物生长阶段模型,对农作物的不同生长阶段进行识别。比如,基于植被系数建立水稻生长阶段模型,通过对水稻生长阶段的图像进行预处理和特征提取,实现对水稻不同生长阶段的自动判断。
2)基于图像处理技术的病虫害自动诊断和防治系统
农业专家可以依据农作物长势、叶片特征等对农作物病虫害进行诊断,根据这一原理,我们可以借助图像处理技术提取农作物生病或长虫时的长势情况、叶片颜色、叶片形状等特征,结合专家经验,建立农作物病虫害诊断规则库。通过对农作物图像进行特征提取,依据病虫害诊断规则库,实现病虫害的自动诊断,并针对农作物不同的病虫害给出针对性的防治措施。该系统不仅能实现病虫害的提前发现,尽早防治,而且不需要农民掌握病虫害防治知识,节约大量的人力成本。
3)基于数据和知识的专家决策系统
农业专家能根据农田环境参数和农作物长势情况提出病虫害防治、灌溉、施肥等指导性建议,但这依赖于专业领域知识,需要投入大量人力,对农民培训耗费大量的成本,农民不容易掌握,而且人工不能及早发现问题,具有滞后性。借助于计算机和数据处理技术,充分利用数据和专业领域知识构建专家决策系统,可以代替农业专家对农作生产进行指导。专家决策系统能充分利用数据和知识,能及早发现问题给出控制策略。
借助于传感器网络采集农田土壤温度、湿度、光照强度等环境参数数据,建立包含环境参数数据和农作物不同阶段的长势(生长情况)、产量、果实品质等数据的数据库,利用大数据处理技术和数据挖掘技术对这些长年累月收集的数据进行处理,结合专家经验知识,分析不同属性间的关联关系,利用数据融合技术,建立不同属性间的关联规则模型。
数据库中数据属性如下: [生长阶段 温湿度、光照强度等环境参数 土壤营养成分含量 生长情况 品质 产量]
生长情况包括:生长正常、营养元素缺乏(病害)、长虫(虫害)
品质是指作物生产的果实的质量;例如:小麦果实的品质是指麦粒营养成分、饱满度等;西瓜的大小、甜度等;
产量是指单位面积收获的果实的质量;例如:小麦的产量是指亩产小麦的质量;西瓜的产量是指亩产西瓜的质量。
以西瓜的种植为例,关联规则如下:
a、[结果期 土壤湿度降低 二氧化碳浓度增加 夜间温度降低 西瓜甜度增加]
b、[伸蔓期 虫害 喷药]
c、[伸蔓期 土壤氮磷钾含量偏低 施肥]
4)环境参数自动调节系统
在农作物生长的不同阶段甚至不同时刻,对环境参数要求有很大的差异,比如在西瓜的结果期,白天需要更强的光照强度和二氧化碳浓度以增强光合作用,而夜间则需要降低温度和增加二氧化碳浓度以降低呼吸作用。因此,为了使农田环境更好地适应农作物生长,需要实时调节农田环境参数,比如二氧化碳浓度、光照强度、温度、湿度等。借助于传感器网络对农作物生长环境数据的实时采集,通过对这些数据进行分析,由专家决策系统给出控制规则,并向底层控制系统发送控制命令,从而实现环境参数的自动调节。
5)自动施肥/灌溉系统
水和肥料在农作物的生长过程中起着至关重要的作用,在农作物生长的不同阶段,对水和肥料的需求有很大差异。通过对土壤水分和营养元素含量的实时监测,结合专家决策系统,实现农作物的自动灌溉和施肥以满足农作物不同时期对水和肥料的需求。
英文原文:
Intelligent Agricultural Control System
Based on real-time environmentdata of the farmland,and crop growth, with the data mining techniques and imageprocessing techniques, decision rules can be given combined with the expertdecision-making library.Intelligent Agricultural Control System can controlequipments automatically,adjust parameters of the farmland environment and helpcrops grow well.Intelligent agricultural control system is shown in figure X.
Figure X: Intelligent Agricultural Control System
1)The crops growing stage judgment system
Using image processing techniques can realize featureextraction of different images of different growing stages of crops, and cropsgrowing stage model can be given.The system can identify different growingstages of crops automatically. For example, the model of rice growing stage canbe established based on vegetation index, the system can automatically judgewhich growing stage rice is in with preprocessing and feature extraction ofdaily images of rice growing stage.
2)The automatic diagnosis and prevention system ofcrop diseases and pests
As is known to all,agricultural experts can realize the diagnosis of crops diseases and pestsbased on the growing condition and leaf characteristics of crops. According to thisprinciple, using image processing techniques,we can extract features of thecrops,such as the growing condition,the color and shape of leaves.And the libraryof the crops diseases and pests diagnosis rules can be established combinedwith the expert experience. By means of feature extraction of the crops images,thesystem can realize the automatic diagnosis of crop diseases and pests on thebasis of the library of diagnosis rules,and propose pertinent prevention andcontrol measures against different diseases and pests. The system can not only findcrop diseases and pests in advance,and then give a treatment as soon aspossible,but also it can reduce farmers’ workload and save a lot of labor asnot requiring farmers to master knowledge of disease and pests curing
3) The expert decision-making system based on data and knowledge
According to environmental parameters of the farmlandand crop growing condition,agricultural experts can put forward some guidance strategies,suchas the prevention and control of diseases and pests, irrigation, fertilization,butit depends on the professional knowledge,spends a lot of labor and cost ofthe farmers' training,and is verydifficult to master for farmers,and usually has a lag leading not to find the problem as early aspossible.With the aid of computer and data processing technology,the expert decision-making system can be established by making full use of data and professional knowledge,whichcan replace agricultural experts to guide agricultural production.The expertdecision-making system can make full use of the data and knowledge,findproblems as early as possible,and give some control strategies.
With the acquisition of farmland environmental data,suchas the soil temperature and humidity,light intensity,by using sensor networks,a database can be established,whichcontains a large amount of environment data, growth conditions of the crops,thecrop growing stage,yield,fruit quality,and so on. Using big data processing anddata mining techniques to deal with the data in database collected collectedover many years,combined with the expert knowledge,a model containing theassociation rules can be established with data fusion technology by analyzingthe correlation between different attributes.
The data attributes in the database are as follows:
[the growingstage;the environmental parameters;the content of soil nutrition constituent;thegrowth condition;fruit quality;yield]
Among them, the environmental parameters containtemperature, humidity,light intensity,concentrations of CO2,etc.The growthcondition consists of health, lack of nutrient elements,disease,pest. Qualityrefers to the quality of the fruit;for example, the quality of the wheat refersto the grain nutrients, plumpness,etc;the quality of the watermelon refers to thesize, sweetness, etc.
Association rule model is as follows:
Taking an example of watermelon, association rules areas follows:
A:[unripe fruit; soil moisture lower;concentrations of carbon dioxidehigher; night temperature lower;fruit sweetness increase]
B:[the vine growth stage;pests;spray];
C:[the vine growth stage; phosphorus contents are lower; phosphatefertilisers increases]
4)The automatic regulating system of environmental parameters
In the different crop growing stages,even at different times,the requirementsof crops for environmental parameters may be largely different,for example,inthe fruit period of watermelon,it needs stronger light intensity and CO2concentration in the daytime in order to enhance photosynthesis rate,then at night it needs to lowertemperature and increase concentration of the carbondioxide so as to reduce the respiration.Therefore,in orderto provide benign vegetal environment which is good for crop growth,some environmental parameters of thefarmland need to be adjusted and controlled in real time,such as theconcentration of CO2,light intensity,temperature,humidity,etc. Real-timeenvironmental data of the farmland can be acquired by sensor network,then throughthe analysis of these data,some decision rules can be given by the expert decision-making system,whichcan send control commands to the field control units,so as to realize automatic adjustment of theenvironmental parameters.
5)The automatic fertilization and irrigation system
Water and fertilizer plays a vital role in the growthof crops,and the demands of crops for water and fertilizer are very differentbetween the different crop growing stages.Based on the expert decision-makingsystem,it can realize the automatic irrigation and fertilization to meet demandsof crrops for water and fertilizer at different growth stages by taking real-time monitoring in the content of soil moisture and nutrition.
参考文献:references
[1]Satyanarayana G V, Mazaruddin S D.Wireless Sensor Based Remote Monitoring System for Agriculture Using ZigBee andGPS[C]//Conference on Advances in Communication and Control Systems. 2013:110-114.
[2]Chang K C, Liu P K, Kuo Z W, et al.Design of persimmon growing stage monitoring system using image recognitiontechnique[C]//2016 IEEE International Conference on Consumer Electronics-Taiwan(ICCE-TW). IEEE, 2016: 1-2.
[3]Soontranon N, Srestasathiern P, RakwatinP. Rice growing stage monitoring in small-scale region using ExG vegetationindex[C]//Electrical Engineering/Electronics, Computer, Telecommunications andInformation Technology (ECTI-CON), 2014 11th International Conference on. IEEE,2014: 1-5.