基于LSTM时间序列模型的电力变压器温度预测方法研究

Research on Temperature Prediction Method of Power Transformer Based on LSTM Time Series Model

传统的电力变压器运行状态监控方式仅能做到实时监控,无法做出超前预测。笔者提出基于LSTM时间序列模型的电力变压器温度预测方法。通过分析LSTM时间序列预测模型原理,建立基于LSTM的变压器温度预测模型,采用仿真实验验证变压器温度预测的有效性。实验结果表明,LSTM模型比ANN模型具有更优秀的预测能力和拟合效果,为电力设备运行监控领域提供更多的量化辅助信息,具有较高的实用性。

The traditional power transformer operation condition monitoring mode can only achieve real-time monitoring and cannot make advanced prediction.The temperature prediction method of power transformer based on LSTM time series model is proposed.By analyzing the principle of LSTM time series prediction model,a transformer temperature prediction model based on LSTM is established,and the effectiveness of transformer temperature prediction is verified by simulation experiments.The experimental results show that LSTM model has better prediction ability and fitting effect than ANN model,which provides more quantitative auxiliary information for the field of power equipment operation monitoring and has high practicability.