基于神经网络的温、湿度控制系统辨识

Identification of the Temperature and Humidity Control System Based on Neural Network

针对温、湿度需要同时进行控制的这种复杂的非线性系统,采用神经网络辨识的方法,从工程实用角度出发,考虑辨识精度和辨识学习时间,确定出合理简单的网络结构及网络训练参数。考虑影响温湿度控制的各种因素,用不同测试条件下得到的检测样本对训练后的网络进行检测,均能满足一定的精度要求。在网络训练中实行训练和测试间隔进行,最终确定出网络参数,这样训练出来的神经网络有较好的泛化能力,能在较宽的范围内正确映射控制系统的输入输出,为对温、湿度控制系统进行进一步的研究,进而为神经网络控制器的设计打下可靠的基础。

The work of this paper designs the logical strcture and training parameter of the NN for the complicated non-linear system which needs to control the temperature and humidity at the same time. Proceeding from the practical angle of project ,it considers the identified precision and identified learning time. Considering various kinds of factors that influence the temperature and humidity controls , trained Networks are checked using samples which checked by various conditions. And it must satisfy stated precision. The process of train and check is doing at intervals, then the NN parameter is obtained. This neural network can mapping input and output of the control system with expanse ranges accurately, which lays a reliable foundation for further research of temperature and humidity control system and the design of neural network controller.