关键词: 时间序列模型, 故障上限评估, 不规则波动, 参数寻优, 经验模态分解

Abstract:

When a sudden malfunction occurs in gas insulated switchgear and is not dealt with in a timely manner， it not only leads to the stagnation of the production line， but also causes considerable economic losses to the enterprise. To avoid such situations， this study proposes a device fault upper limit evaluation algorithm based on time series models and deep learning. This algorithm combines the essence of modern data analysis， aiming to improve the accuracy and efficiency of fault detection. This method first applies empirical mode decomposition technology to handle the irregular fluctuation components in time series data. Through this method， the noise and redundant information in the fault signal are effectively removed， making the fault features clearer and easier to identify. Next， we introduced the long short-term memory network in deep learning algorithms. A simultaneous algorithm was constructed by combining LSTM network with EMD technology. This algorithm can simultaneously utilize the temporal and spatial characteristics of data to more accurately identify sensitive components in fault signals. After obtaining equipment fault data， empirical mode decomposition is performed using a simultaneous algorithm to obtain the intrinsic mode function components of the data signal. Then， extract the fault sensitive components from these components and analyze their relationship values and mutual information with the original data signal. Finally， in the process of evaluating the upper limit of equipment failure， a deep learning model is constructed and the refractive and reflection coefficients are determined to achieve accurate evaluation of the upper limit of equipment failure. Experimental tests have shown that the proposed algorithm has achieved ideal results in evaluating the upper limit of equipment failures. The obtained curve has a high degree of fit with the actual result curve， proving that the algorithm can scientifically evaluate the upper limit of equipment faults and has certain application value. This algorithm can not only be applied to the field of fault detection in gas insulated switchgear， but also be extended to other similar industrial equipment， contributing to the safety production and economic benefits of enterprises.

Key words: time series model, failure upper limit evaluation, irregular fluctuations, parameter optimization, empirical mode decomposition