基于神经网络算法的等离子体负载动态参数模型
Dynamic Parameter Model of Plasma Loading Based on Neural Network Algorithm
包涵春 1郭亚逢 1关银霞 1李超 1唐诗雅 1杜宇1
作者信息
- 1. 化学品安全全国重点实验室,山东青岛 266104;中石化安全工程研究院有限公司, 山东青岛 266104
- 折叠
摘要
等离子体负载电学模型大多基于固定参数模型,忽略了负载等效参数变化对模型的影响,容易产生较大误差.为了改善因等效参数变化带来的误差,首先探究了等效电容、等效电阻等负载参数随外加电压幅值、频率的变化情况,据此训练了BP神经网络参数调整模块,建立了等离子体负载动态参数模型,实现了外施激励变化下负载等效参数的更新.结果表明,采用神经网络动态参数模型仿真精度为 95.70%,而采用固定参数模型仿真精度为 82.89%,仿真精度提高了15.45%.对简化实验工作量、指导等离子体反应器设计有着重要意义.
Abstract
Most plasma loading electrical models are based on fixed parameter models,ignoring the im-pact of changes in equivalent loading parameters on the model,which can easily lead to significant er-rors.In order to improve the errors caused by chan-ges in equivalent parameters,the variation of load parameters such as equivalent capacitance and e-quivalent resistance with the amplitude and frequen-cy of applied voltage was first investigated.Based on this,a BP neural network parameter adjustment module was trained,and a dynamic parameter mod-el of plasma load was established,achieving the up-date of load equivalent parameters under external excitation changes.The results showed that the sim-ulation accuracy using the neural network dynamic parameter model was 95.70%,while the simulation accuracy using the fixed parameter model was 82.89%,which improved the simulation accuracy by 15.45%,which was of great significance to sim-plify experimental workload and guide the design of plasma reactors.
关键词
介质阻挡放电/负载等效参数/等离子体电学模型/BP神经网络/动态参数模型Key words
dielectric barrier discharge/load equiv-alent parameters/plasma electrical model/BP neu-ral network/dynamic parameter model引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据