【作者】 马丽丽；

【导师】 纪建伟；

【作者基本信息】 沈阳农业大学 ， 农业电气化与自动化， 2009， 博士

【摘要】 温室要在多变的自然气候条件下,为作物生长创造一个适宜的环境,以实现高产出、高品质、低成本、高效益、环境友好等可持续发展为目标,因而越来越引起人们重视。现代日光温室是目前我国设施农业生产的主力,在我国反季节蔬菜的生产中起着重要的作用,有着良好的发展前景。并成为一个具有重要意义的研究方向。其中温室小气候控制是现代温室中最重要的关键技术之一。在进行有效试验观测的基础上,论文研究的内容包括以下几个部分:（1）温度、湿度、光照和CO₂浓度是节能型日光温室内对作物生长有重要影响的主要环境因子。主要应用BP神经网络方法,基于环境因子的条件下,建立有效积温法的温室番茄叶面积指数模型,温室番茄果实横纵径的模型,温室番茄干物质积累和分配指数模型。在模型建成之后就可以通过番茄的生长发育天数来预测番茄的叶面积指数,果实横茎和果实纵茎。这种方法简单、快速、经济、不破坏番茄的叶片省去很多的麻烦。因此模型具有实际应用性。（2）农田蒸散量（Evapotranspiration,ET）是研究区域水分平衡和制定灌溉需水量计划的基本依据,其准确性直接影响着作物需水预报的精度,而且业已成为监测农业旱情、提高区域水资源利用效率的关键环节。本文建立了基于环境因子的水分蒸发模型。该模型揭示了环境因子与水分蒸发、土壤水分蒸发、作物蒸腾之间的数量关系,揭示了环境因子对水分蒸发,土壤水分蒸发变化的影响,和环境因子对植物水分生理变化的影响,可以利用气象参数预测日光温室内的作物蒸腾耗水量,然后依据预测值适时适量地供给植株水分,从而达到节水、高产、优质的目的。在水分蒸发和土壤水分蒸发的模型中由于温室内的环境因子时刻变化,水分和土壤蒸发也受其影响,因此如果可以预测模型中蒸发的最小值,这样不仅对植物的成长有利,而且也可以节约用水。遗传算法,是近些年来发展起来的一种崭新的自适应搜索的全局优化算法。它比传统的非线性规划方法收敛性好,适应性强,可以达到全局最优,故其应用前景十分广泛。本文应用遗传算法优化水分蒸发模型,可以预测模型中蒸发的最小值。体现了遗传算法广泛的应用性。该模型对于制定合理的温室湿度管理技术体系,减轻病害发生频率,以及制定合理的灌溉制度,在保证适时适量供给作物用水的基础上节约大量灌溉用水,都具有重要意义。（3）温室环境的建模是温室环境控制的关键,也是实现温室环境控制必须解决的一个难题。针对温室这样一个非线性、多变量藕合的复杂系统,无论是仅采用数据信息还是仅采用语言信息,都难以对它进行充分的描述。无论是机理建模还是实验建模都无法建立温室一个通用的模型,而且这些模型都对温室系统进行了简化和线性化处理。自适应模糊神经网络是一类具有学习功能的模糊逻辑系统,通过学习能自动修正其模糊规则,能够综合利用数据和语言两类信息,适宜于辨识时变非线性对象。本文建立了基于自适应模糊神经网络的温室环境模型,根据实验数据对模型中的参数进行调整,参数可最终收敛到某些确定的值,从而可得到温室环境自适应模糊神经网络模型。从建立的温室环境自适应模糊神经网络模型,对不同季节温室内部小气候的模拟结果来看,模型具有较高的精度,使用该模型对不同季节温室内部的温度、湿度、光照进行预测,预测值和实测值也具有较好的拟合关系。这说明可以使用自适应模糊神经网络模型对温室不同季节小气候进行模拟,并可以得到物理模型无法达到的精度,在温室内部的湿度模拟这一点上,表现尤为明显。如果对温室的定性描述较为全面,那么所建立的模型就会具有较好的适应性,克服了一般线性模型在这方面的局限。（4）根据建立的温室小气候模型,提出了运用计算机仿真的方法预测温室小气候的温度、湿度、光照,并结合温室实际运行情况,分析温室小气候特点以及各种调控设施的调控能力。针对春季、夏季、秋季、冬季,室内温度、室内湿度、室内光照、室外温度、室外湿度、室外光照不同特点采取不同的调控措施创造适宜作物生长的环境,为温室小气候的控制奠定前提条件和基础。（5）基于对番茄生长的日光温室多年环境数据的综合,建立了一套周年连续的温湿光数据库,称为通用数据库。应用在温室番茄专家系统病害预测模块中,建立的环境数据库主要根据温度湿度光照的记录结合温室常见病害发生条件,预测病害的发生,目的是从预防的角度来解决病害的危害。这样可以减少施药的次数和用量。目前这个系统只是从环境因子的角度进行预测,无法量化。随着我国温室的发展,温室环境数据库将在温室管理中发挥重要的作用,具有重要的意义。为生产上现代温室、特别是新建温室的环境管理提供重要的数据支持,实现现代温室生产的环境管理自动化、智能化。更多还原

【Abstract】 Greenhouse is to create the most suitable environment for crops cultivated inside such that the goal of high-yield,high-quantity,low-cost,best-benefit,environment friendly at harvest can be reached,so it has attracted attention.Modern greenhouse is the main of China’s facilities agricultural production,playing an important role in our off-season vegetable production and having good prospects for development.The control of the greenhouse microclimate is one of the key techniques in modern greenhouse and becoming increasingly important.On the basis of conducting effective test observation,the main contents of this paper include follows:(1)Temperature,humidity,light,carbon dioxide concentration are the major environmental factor having important impacts on crop growths in energy-saving solar greenhouse.In the use of the method that is applied of neural network and based on the environmental factors,the tomato leaf area index model of the effective temperature,the horizontal and longitudinal diameter model of tomato fruits,and dry matter accumulation and distribution exponential model of tomato are established.This method is simple,rapid,economic,and does not damage the leaves of tomato saving a lot of trouble.Thus the model has practical application.(2)Farmland evapotranspiration(Evapotranspiration,ET) is the fundamental basis to researchthe regional water balance and to plan to develop irrigation water requirement,and the accuracy of it have direct impact on crop water demand forecasting,so it has become the key link to monitor agricultural drought and to improve the regional water use efficiency.This paper established the water evaporation models based on environmental factors.The models show the relationship of environmental factors and water evaporation,soil evaporation,crop transpiration between the number,not only reveal the environmental factors of water evaporation,evaporation of soil moisture changes,and environmental factors on plant physiological changes of water,but also can be used to forecast crop transpiration water consumption with meteorological parameters,then based on predictive value to supply water plants timely and adequately,so as to achieve water-saving,high-yielding,high-quality purposes.In water evaporation and soil evaporation models,in the moment of change climatic factors,soil moisture and evaporation is also affected,if the model can predict the minimum evaporation,so that the growth of plants not only is beneficial,but also can save water.Genetic algorithm,referred to in recent years develop a new self-adaptive search algorithm for global optimization.It is more programming,and adaptable than the traditional method of convergence of nonlinear,you can reach the global optimum,so the applications prospects is a wide range.The application of genetic algorithm optimization model can predict the minimum evaporation model,this paper use genetic algorithm to optimize the water evaporation model.This models are of great significance for establishing a reasonable greenhouse moisture management technology system to reduce the frequency of disease,as well as to establish a reasonable irrigation system,in a timely manner to ensure adequate water supply of crops based on a large number of irrigation water conservation. (3)Modeling of greenhouse environment is not only the key to control the greenhouse environment,but also a problem of the achievement of greenhouse environmental controling. Greenhouse for such a non-linear,coupled multivariable complex system,whether it is used only data or only information on the use of language are difficult to adequately describe it.Modeling the mechanism of both the experimental model or are unable to establish a generic model of the greenhouse,and to greenhouse systems these models have simplifly and linearization ability.Whether it is the mechanism modeling or the experimental modeling we are unable to establish a common model of the greenhouse,and have ability to simplifly and linear models.Adaptive fuzzy neural network is a fuzzy logic system of learning function,by learning automatically correcting the fuzzy rules and with two types of utilization data and linguistic information,and it can identify nonlinear time-varying object suitablely.This paper established models of the greenhouse based on adaptive fuzzy neural network,according to the experimental datas adjusting the parameters of the models and the parameters converged to some final determined value in order to be adaptive fuzzy neural network models.From the simulated results that inside the greenhouse established microclimate models with adaptive fuzzy neural network,the models has higher precision,using the model in different seasons within the greenhouse temperature,humidity,light to carry out the forecast,then the forecast value and actual value also has a good fitting relationship.This shows that using adaptive fuzzy neural network model to simulate microclimate in the greenhouse and different seasons, it meet precision that the physical models may be unable to meet,and the performance of the internal simulation humidity is particularly evident.If the qualitative description of the greenhouse are more comprehensive,then the models will have better adaptability, overcoming these limitations that the general linear models have.(4)According to the establishment greenhouse microclimate models,the views are bring forward that the use of computer simulation to predict temperature,humidity,light and analyzed the characteristics of the greenhouse microclimate control facilities as well as ability to regulate and control in conjunction with the actual operation of the greenhouse.For spring, summer,fall,winter,indoor temperature,indoor humidity,indoor lighting,outdoor temperature, outdoor humidity,outdoor illumination of the different characteristics,different control measures are taken to create a suitable environment for crop growth,and characteristics and control facilities abilities are analyzed.It is fixed the premise conditions and the foundation for the control of greenhouse microclimate.(5)Based on environmental comprehensive datas for greenhouse tomato growth for many years,a set of successive anniversary temperature humidity solar radiation database is established,known as the universal database.Accoring to the establishment of the environmental databases light primarily on the basis of the record of temperature and humidity combined with the occurrence of common diseases of greenhouse conditions,used in greenhouse tomato disease forecasting expert system module,it predict the occurrence of diseases.The purpose is solving disease hazards in view of prevention.This will reduce the number of spraying and amount of application.At present the system only from the perspective of environmental factors predict,can not quantify.It proved an important data support for management of modern greenhouse,especially provide important data to support for the newly established greenhouse,and is helpful to automatization and intelligentized management for modern greenhouse environment.更多还原