城镇绿地智能化灌溉决策系统研究

发布时间：2018-05-24 19:48

本文选题：智能化灌溉 + PSO-BP神经网络　；参考：《西安建筑科技大学》2014年硕士论文

【摘要】：水是维持人类生存和社会发展的基本要素。我国水资源短缺问题日益严重。随着我国城镇化进程的加快，城市绿地面积迅速增加，绿地灌溉需水量也大幅度增加，加之水污染等环境问题的威胁，致使城市供水问题形势更加严峻。提高水资源利用率，走节水、省水路线，是解决我国城镇供水紧张问题的根本途径。城镇绿地灌溉势必以精量化、智能化节水灌溉模式取代当前的粗放式灌溉模式。本文首先建立PSO-BP神经网络预测参考作物蒸腾量ET0，在此基础上建立灌水系数模糊推理系统以及灌溉模糊决策系统，最终实现对绿地适时、适量地灌溉。本文选用9组不同气象因子组合作为输入量，建立BP神经网络，通过理论分析以及大量重复实验，确定网络结构。然后采用PSO算法优化BP网络的权值和阈值，建立PSO-BP神经网络。利用训练好的PSO-BP模型预测参考作物蒸腾量。将不同输入得到的预测结果与PM ET0对比分析，结果表明，只需选用平均温度、平均相对湿度、风速和日照时数等四项影响因子作为输入，即可通过PSO-BP模型得到较为精确的参考作物蒸腾量预测值，为气象资料缺失情况下准确预测参考作物蒸腾量提供了一个可行的方法。本文利用经模糊处理的绿地作物系数估算绿地作物实际蒸腾量。以实际蒸腾量和土壤有效含水量为输入，，灌水系数为输出，建立灌水系数模糊推理系统。在此基础上结合灌水定额公式及水量平衡方程等，建立灌溉模糊决策系统，由此系统决策灌水量和灌水时间。用2005年的相关数据对灌溉模糊决策系统进行测试，结果表明，利用本系统进行灌溉决策，可使土壤有效含水量较为稳定的保持在适宜作物生长的范围内。在保证作物健康生长的前提下最大限度的节约了水资源，对节水灌溉的广泛实施具有一定的指导意义。
[Abstract]:Water is the basic element to maintain human survival and social development. The shortage of water resources in China is becoming more and more serious. With the acceleration of urbanization in our country, the area of urban green space is increasing rapidly, the irrigation water demand of green land is also increasing by a large margin, and the threat of environmental problems, such as water pollution, makes the situation of urban water supply more serious. To improve the utilization rate of water resources, to save water and to save water is the fundamental way to solve the shortage of water supply in cities and towns in China. Urban greenbelt irrigation is bound to be quantified, intelligent water-saving irrigation mode will replace the current extensive irrigation mode. In this paper, the PSO-BP neural network is established to predict the reference crop transpiration et _ 0. Based on this, a fuzzy inference system for irrigation coefficient and a fuzzy decision-making system for irrigation are established. In this paper, 9 groups of different combinations of meteorological factors are selected as inputs, and BP neural network is established. The network structure is determined by theoretical analysis and a large number of repeated experiments. Then the weight and threshold of BP neural network are optimized by PSO algorithm, and the PSO-BP neural network is established. The reference crop transpiration was predicted by the trained PSO-BP model. The predicted results obtained from different inputs are compared with PM ET0. The results show that only four factors, such as mean temperature, mean relative humidity, wind speed and sunshine duration, are selected as input. A more accurate prediction value of reference crop transpiration can be obtained by PSO-BP model, which provides a feasible method for predicting reference crop transpiration in the absence of meteorological data. In this paper, the actual transpiration of green land crop is estimated by fuzzy processing. The fuzzy inference system of irrigation coefficient was established by taking actual transpiration and soil effective water content as input and irrigation coefficient as output. On the basis of this, the irrigation fuzzy decision system is established by combining the irrigation quota formula and the water balance equation, so that the irrigation quantity and irrigation time can be determined by the system. The fuzzy decision system of irrigation was tested with the relevant data in 2005. The results showed that the effective water content of soil could be kept within the suitable range of crop growth by using the system to make irrigation decision. On the premise of ensuring the healthy growth of crops, the water resources are saved to a great extent, which has certain guiding significance for the wide implementation of water-saving irrigation.
【学位授予单位】：西安建筑科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TU985.14

【相似文献】