静电放电电磁脉冲耦合的非线性优化建模

Nonlinear Optimized Modeling of Electromagnetic Pulse Coupling in Electrostatic Discharge

针对现有机理建模算法普遍存在计算电磁脉冲响应过程过于复杂的问题,为能够给电子设备静电放电电磁脉冲响应计算提供一种简便有效的能量耦合建模方法,设计了脉冲场强测试仪的静电放电辐射实验。用NARX神经网络代替传统NARX网络,依靠遗传算法对网络的初始权值、阈值进行优化,以3.5 kV静电放电实验数据作为建模数据对系统进行非线性辨识,并对4.5 kV静电放电电磁脉冲响应进行预测。建模结果表明,两种模型均能准确预测响应波形,但优化后的NARX神经网络模型精度更高。该建模方法计算过程简单。该方法同样适用于其他电磁脉冲响应建模。

For the problem that the computing process of electromagnetic pulse response in the existing mechanism modeling algorithms is very complex, a radiation experiment of electrostatic discharge for pulse field sensor is designecl in order to offer a simple but effective method for computing electromagnetic pulse response of electronic devices. In the modeling, NARX Neural Network （NN NARX） is substituted for the conventional NARX network, and the layer weight matrices and bias vectors are optimized by genetic algorithm （GA）. Two models are built and trained on the basis of the 3.5 kV electrostatic discharge experiment data to identify the dynamic characteristics of the system. The electromagnetic pulse response of 4.5 kV electrostatic discharge is predicted by using the models, and the results show that both optimized NN NARX model and conventional NARX model can predict the response wave accurately through comparing the predicted response with the measured data, but the model optimized with NN NARX performs better. The proposed modeling method is easy in use. and also suitable for modeling of other electromagnetic pulse responses.