基于神经网络的专家系统在温室控制中的应用

为了实现温室控制，针对温室环境的多输入、多输出、非线性和难以建立数学模型等特点，提出一种基于BP神经网络的专家系统并用于温室控制。该方法将传感器采集的温度、湿度等信息输入到神经网络专家系统，在获得决策结果的同时通知控制部分执行相应的决策。这种方法不仅解决了传统专家系统知识获取的瓶颈问题、推理能力差、智能化低的缺点，而且克服了神经网络不具有解释功能的问题。总之，成功实现了神经网络和专家系统功能上的互补，较好的用于温室控制。

The Application of Greenhouse Control Based on Neural-network Expert System

In order to implement greenhouse control, in accordance with greenhouse environment, which has multipleinput, multiple-output, non-linear, and difficult to establish mathematical models, an expert system based on BackPropagation neural network has been proposed and used in greenhouse control. Data as temperature or humidity which achieve from sensors axe input in the expert system. When we receive the decision, the system also informs the control block to react. The method not only solves the drawbacks of conventional expert system such as knowledge acquiring bottleneck, poor capability in reasoning and lower intelligent level, but also overcomes shortcomings that the neural network doesn＇t have interpreting function. In conclusion, the method which succeeds integrating merits of neural network and expert system can be used in greenhouse control.