论文部分内容阅读
文中基于广义回归神经网络(GRNN)技术挖掘数据变化规律,构建了GRNN神经网络预测模型,运用该模型对油纸绝缘变压器进行寿命预测。将变压器绝缘纸老化过程中生成的特征产物如糠醛、CO2和CO的质量分数,以及相关时间参量作为模型输入。将所采集的多组油纸绝缘变压器的测试样本作为基础数据,运用该模型对相应的变压器进行寿命预测。结果表明,模型寿命预测的输出值与实际值基本一致,从而验证了模型的合理性,这对监测绝缘材料老化状态的进一步研究具有现实意义。
Based on GRNN technique, the variation regularity of data is explored, and the prediction model of GRNN neural network is constructed. The model is used to predict the life of oil-paper insulated transformer. Enter the mass fraction of the characteristic products such as furfural, CO2 and CO generated during the aging process of the transformer insulation paper, and the related time parameters. The collected samples of oil-paper insulation transformer test samples as the basic data, the use of the model of the corresponding transformer life prediction. The results show that the output value of model life prediction is basically consistent with the actual value, which verifies the rationality of the model, which is of practical significance for further study on monitoring the aging status of insulating materials.