用人工神经网络预测HL-2A等离子体放电破裂

Prediction of plasma disruption using artificial neural network on HL-2A

使用人工神经网络(ANN)对HL-2A装置破裂放电进行了离线预测研究。采用了两种方法训练网络,一种方法是采用原始实验数据作为网络输入训练网络,另一种是把训练样本中的Mirnov原始实验信号进行预处理,目的是突出Mirnov原始信号隐含的破裂信息。比较这两种方法,结果表明第二种方法获得的网络对破裂放电能够做出更加准确的预测。

Prediction of plasma disruption using artificial neural networks （ANN） has been investigated on HL-2A. Two methods have been adopted to train ANN. In the first one, the original experimental data are used directly as the ANN input. In the second one, in order to reduce the noise and enlarge the implicit disruption information in original Mirnov signal, the preprocessing is given in original Mirnov signal in the training subset. The prediction results indicate that the second method is more accurate and effective comparing with the first one.