基于改进型神经网络的臭氧发生器模型研究  

Research on Ozone Generator Model Based on Improved Neural Network

作　　者：翟维枫[1] 黄理邮 孙德辉[1] 董哲[1] ZHAI Wei-feng;HUANG Li-you;SUN De-hui;DONG Zhe(School of Electrical and Control Engineering,North China University of Technology,Beijing 100144,China)

机构地区：[1]北方工业大学电气与控制工程学院,北京100144

出　　处：《计算机仿真》2022年第9期355-358,402,共5页Computer Simulation

摘　　要：介质阻挡放电(DBD)等离子体生产臭氧的过程复杂,是典型的非线性系统。为了更深入研究臭氧发生过程的输入输出关系,提出使用神经网络建立臭氧发生模型。通过200g/h型号的DBD板式臭氧发生器实验平台获取大量实验数据,以遗传算法与BP神经网络相结合的方式建立臭氧浓度预测模型,并提出了模型在线更新的方法。实验结果表明,GA-BP方法相比未经优化的BP神经网络能够更快速建立较准确的模型,在线更新机制有效改善了离线神经网络模型的不足,研究结果为进一步提高臭氧浓度的精确预测和系统闭环控制提供了理论基础。The process of ozone production by dielectric barrier discharge(DBD) plasma is a typical nonlinear system. In order to study the relationship between the input and output of the ozone generation process, a neural network model is proposed. A large number of experimental data are obtained by using the 200 g/h DBD plate ozone generator experimental platform. The ozone concentration prediction model is established by combining genetic algorithm with BP neural network, and the method of online updating of the model is proposed. The experimental results show that the GA-BP method can establish a model more accurately and quickly than the unoptimized BP neural network. The online update mechanism effectively improves the shortcomings of the offline neural network model. The research results provide a theoretical basis for further improving the accurate prediction of ozone concentration and closed-loop control of the system.

关 键 词：介质阻挡放电 臭氧发生模型 遗传算法 神经网络 在线更新 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]