基于时间卷积网络的GIS设备振动信号特征预测OA

The variation of vibration signal of GIS equipment can reflect the mechanical condition inside the equip-ment.For improving the prediction accuracy of vibration signal characteristics of GIS equipment,in this paper a combined group measurement model based on decomposition-forecasting-reconstruction is proposed.First,based on historical vibration signals of GIS,vibration characteristic parameters are extracted in frequency domain by Fourier transform.Then,in order to eliminate as much as possible the influence due to the non-stationary characteristics of the vibration characteristic parameter sequence,the normalized sequence is decomposed by the variational mode decomposition(VMD)optimized by particle swarm optimization(PSO).Finally,the time convolution network(TCN)is used to predict a set of stationary modal components obtained by decomposition.The experimental results show that the root mean square error and the average absolute percentage error of the combined prediction model based on PSO-VMD-TCN proposed in this paper are 1.79%and 0.13%,respectively,which are superior to other methods in prediction accuracy and are conducive to the early fault diagnosis of GIS equipment.