【作者】 刘立佳；

【导师】 满春涛；

【作者基本信息】 哈尔滨理工大学 ， 控制理论与控制工程， 2010， 硕士

【摘要】 多传感器信息融合技术是一项前沿技术,初期只是应用于军事领域的战场控制如C4I系统等,二战以后已逐渐被引入到非军事领域,例如：智能楼宇、遥感系统等等。其涉及的领域技术包括概率统计学、建模理论、模式识别、神经网络等。在温室内采用多个传感器对环境进行实时检测,有效地弥补了单一传感器失效无法对目标监测,进而无法进行数据传输的缺点。实现温室智能控制是根据作物自身的生理特点,部分或全部克服外界气候环境以及其它非自身因素的束缚,从而为作物生长创造最佳条件,达到增产和节能的目的。实现温室智能控制首先需要确定制约环境的各个因素,同时确定他们之间的关系,进而采取相应的管理、调控措施。系统根据农业领域专家的不断探索与实践最终确定了温度、光照度、湿度这三个量为农作物生长的三个主要影响因素。本温室环境控制系统以工控机为核心,采用现代信息处理技术,对以往更多依赖于农业经验知识技术判断各种作物生理状况的温室控制进行了改进。本文主要研究了多传感器信息融合的原理、结构和常用的算法,分析了多传感器系统的融合模型,在对比各个融合级别的信息融合算法的优缺点的基础上,重点介绍D-S证据理论相关知识和专家系统的内在含义。对于同质传感器输送的同源信息,采用分布图法去除干扰,达到了对信息进行初步优化处理的目的。对D-S组合算法进行了改进,同时对系统的知识构造识别框架进行了模糊化处理,在此基础上为D-S模型进行基本概率分配函数赋值,然后对经过分布图处理的温度、湿度和光照度信息进行分组融合,最终由专家系统做出决策。实验表明,这种方法提高了温室环境参数测控的决策准确性,可显著改善温室环境的控制效果。更多还原

【Abstract】 The multi-sensor information fusion technique is a cutting edge technology, initially only applied to the Battlefield control of the military, such as the C4I system, etc, this technology has gradually been introduced into the non-military areas after World WarⅡ, for example:intelligent buildings, remote sensing systems and so on. It relates to areas of technology which include probability statistics, modeling theory, pattern recognition and neural networks. Using multiple sensors to real-time detection on the environment in a greenhouse, it effectively compensates the shortcoming that the failure of a single sensor can not transmit data because a volume can not be monitored.According to the physiological characteristics of crops, wholly or partly to overcome the climatic and environmental factors outside world and other non-self-bondage, so as to create the best conditions for crop growth and achieved the purpose of increasing production and energy efficiency, firstly, to determine the various factors of the constrained environment and the relationship between them, then to take the appropriate management and control measures. Through the continuous exploration and practice of experts in the agricultural field, ultimately to determine the temperature, light intensity and humidity are greater impacts on the environment. Be seen as the main three factors to affect crop. Greenhouse environmental control systems often use IPC as the core. computer technology is more emphasized, for depending on the knowledge technology of agriculture and experience, greenhouse controls of judging the physiological state are clearly shortcomings.This paper mainly studies the principle of multi-sensor information fusion, structures and the commonly used algorithms, and analysis of multi-sensor system integration model, detailed analysis of the advantages and disadvantages of the different levels of information fusion algorithm, focuses on the D-S evidence theory knowledge and the intrinsic meaning of expert system. For the homology of the information transferred by homogeneous sensor, our purpose is to eliminate interference by compatibility matrix method and achieve the purpose of a preliminary optimization of information. Using of an improved D-S combination algorithm, the identification framework is constructed by knowledge of the expert system, and to be assignment for the basic probability assignment function in the D-S model, and then attempt to integrate for the processed information of temperature, humidity and illumination by compatibility matrix method, the final decision is given by the expert system. Experiments show that, this approach improves the accuracy of decision-making and control in parameters of greenhouse environmental, and to improve the effect of greenhouse environmental control.更多还原