CAJ | 학술논문

立体栽培是将原地面的种植提升到温室内空间高度层面,这不仅使作物生长的感温层发生改变、接收到更多的日光照射,还解决了日光温室中的草莓生产管理、观光、摘菜的难题,是传统草莓种植的一次改革.但对于这样一种新的种植模式,目前还未有成熟的集高架栽培设施、滴灌、营养液配比、光照、通风、施药、温湿度等控制为一体的集成体系出现.该文介绍了一种电动式立体栽培机构,研究了一种日光温室草莓立体栽培智能控制系统.其核心内容论述了基于复杂农业大系统控制理念下,突破传统意义上的温室环境控制概念,考虑被控制量的解耦,将以往温室环境参量和作物生长参量分开,建立温室环境控制和作物生长控制2个并列的子控制系统,并在协调控制器作用下、以作物生长模型输出作为控制约束之一对各子系统进行调控.给出了基于Q学习的多Agent体协调算法,并以草莓采光随动控制和草莓根部加温控制进行了验证.该控制系统已在北京市第七届世界草莓大会日光温室草莓立体栽培中得到了成功应用.试验表明在控制草莓生长方面取得了满意的效果,草莓的挂果期比传统种植平均提前40d,草莓的果实体积平均增大22%,出产量平均提高33%.
입체재배시장원지면적충식제승도온실내공간고도층면,저불부사작물생장적감온층발생개변、접수도경다적일광조사,환해결료일광온실중적초매생산관리、관광、적채적난제,시전통초매충식적일차개혁.단대우저양일충신적충식모식,목전환미유성숙적집고가재배설시、적관、영양액배비、광조、통풍、시약、온습도등공제위일체적집성체계출현.해문개소료일충전동식입체재배궤구,연구료일충일광온실초매입체재배지능공제계통.기핵심내용논술료기우복잡농업대계통공제이념하,돌파전통의의상적온실배경공제개념,고필피공제량적해우,장이왕온실배경삼량화작물생장삼량분개,건립온실배경공제화작물생장공제2개병렬적자공제계통,병재협조공제기작용하、이작물생장모형수출작위공제약속지일대각자계통진행조공.급출료기우Q학습적다Agent체협조산법,병이초매채광수동공제화초매근부가온공제진행료험증.해공제계통이재북경시제칠계세계초매대회일광온실초매입체재배중득도료성공응용.시험표명재공제초매생장방면취득료만의적효과,초매적괘과기비전통충식평균제전40d,초매적과실체적평균증대22%,출산량평균제고33%.
Space planting for strawberry in solar greenhouse is one reform way with a few years development in Northern China. Comparing with the traditional planting on the ground, the space planting has distinct advantage, such as the temperature environment surrounding strawberry is changed and strawberry can get more sunlight based on planting field changing from ground to space in height of indoor solar greenhouse. On the other hand, with this new planting, strawberry production management including visiting and picking can be better resolved. According to control requirements of the new planting way, this paper presents the integration intelligent control system for strawberry space planting in solar greenhouse. The hardcore is that a new system structure of intelligent control is discussed special on harmonizing control model based on IA (Intelligent Agriculture) concept as well as complex large system intelligent control theory. The new control way, reforming traditional greenhouse environment control, is presented that the greenhouse control is divided two sub-systems corresponding to environment control and plant grow control, and with strawberry growing model output constrict condition as well as driving from harmonizing control unit, two sub-systems can parallel work. Q-learning algorithm basing on multi-Agent has been utilized to operate harmonizing control, and testing results have been presented by tracking daylighting and thermostatical control for strawberry rhizosphere. The presented control system had been applied in strawberry space planting in solar greenhouse during the 7th World Strawberry Beijing 2012. The result had been proved that control effect is satisfied, strawberry average bear fruit period was advanced 40 d, and strawberry body average size was expended 22%, average production was increased 20%compared to tradition planting and control.