基于经验模态分解和长短期记忆神经网络的变压器油中溶解气体浓度预测方法

Concentration Prediction of Dissolved Gases in Transformer Oil Based on Empirical Mode Decomposition and Long Short-term Memory Neural Networks

对油中溶解气体浓度进行有效预测,可为电力变压器故障诊断及预警提供重要依据。提出一种基于经验模态分解与长短期记忆神经网络的变压器油中溶解气体浓度预测方法。首先,运用经验模态分解将气体浓度序列分解为一组相对平稳的子序列分量,以减少不同趋势信息间的相互影响;然后,针对各子序列分别构建基于长短期记忆神经网络的时序预测模型,并利用贝叶斯理论对网络相关超参数进行优化,以提高单个模型的预测精度;最后,叠加各子序列预测结果得到气体浓度预测值。算例研究结果表明,相较于传统预测算法,所提方法能更好地追踪油中溶解气体的浓度变化趋势,具有更高的预测精度。

Accurate prediction of development trend of gas concentration dissolved in transformer oil can provide an important basis for fault diagnosis and early warning of the power transformer. Empirical mode decomposition(EMD) and long short-term memory(LSTM) neural networks were introduced into the prediction method of dissolved gases in oil.Firstly, the gas concentration sequence was decomposed into a group of relatively smooth components by using the EMD to reduce the mutual influences among diverse trend information.After that, forecasting models based on the LSTM neural networks were constructed respectively for each subsequence,and then the Bayesian theory was used for the hyperparameters optimization of neural networks to improve the forecasting accuracy. Finally, the prediction results of each subsequence were superimposed to obtain the final gas concentration forecasting results. The simulation study shows that the proposed prediction method can reflect the trend of dissolved gas content in power transformer oil, and outperform traditional prediction algorithms with respect to forecasting accuracy.