基于Rabbit 2000温室智能采集控制器的研究开发

【作者】 李秀梅；

【导师】 刘华毅；

【作者基本信息】 大连理工大学 ， 控制理论与控制工程， 2004， 硕士

【摘要】 本课题研发的基于Rabbit2000温室智能采集控制器是将智能控制理论、电子技术和通讯技术集于一体，以实现温室智能控制系统为目的，对模糊神经网络控制器以及模型训练算法进行了探索和研究，并借此课题进一步掌握了智能控制系统在实践中的应用。 本文重点阐述了Rabbit2000温室采集控制器的硬件设计原理，主要包括以Rabbit2000处理器为核心的中央控制系统和通讯接口等。 最后提出采用模糊神经网络控制器算法实现温室智能控制的方案，并用改进的遗传算法训练模型结构。改进的遗传算法采用十进制编码，去掉了交叉算子，采用智能变异和随机变异，在搜索过程中不易陷入局部极值点，且不要求目标函数具有可微、非凹等特性，还具有编程简单的优点。 基于Rabbit2000温室智能采集控制器与被控对象构成的闭环控制系统鲁棒性强，在实际应用中控制性能好。更多还原

【Abstract】 This dissertation describes the Rabbit2000 intelligent collection controller of the greenhouse which involves intelligence control theory, electric and communication technique, it can help to study the fuzzy neural network controller model of the greenhouse during exploitation. The implement of the fuzzy neural network controller has been achieved in this dissertation.This dissertation presents in detail hardware design of control system for Rabbit2000 intelligent collection controller. It mainly includes control and collect system based on Rabbit2000 and communication interface and so on.Finally, the fuzzy neural network has been adopted in order to establish the controller model of greenhouse, and it is trained by improved genetic algorithm which has been brought forward in this dissertation. Calculating efficiency has been improved through the following means, the decimalization is adopted, the cross operator is got rid of, intelligent mutation and random mutation are used. It uneasily falls into the local minimum during the searching and don’t depend on the characteristic of the object function, furthermore it is programmed easily.The robust of the close loop feedback system which adopted Rabbit2000 collection controller is strong, and its control performance is fine.更多还原