基于时频域分析的LSTM电力系统报警方法研究
Research on LSTM Electric Power System Alarm Method Based on Time-frequency Domain Analysis
李慕轩1
作者信息
- 1. 国网天津市电力公司信息通信公司,天津 300140
- 折叠
摘要
为克服传统语音分析方法的局限性,文章采用基于时频域分析的长短时记忆网络模型,提出基于时频域分析的长短时记忆(Long Short-Term Memory,LSTM)电力系统报警方法.同时,在UrbanSound8K数据集上开展实验验证该方法的有效性.结果表明,该方法的准确性、精确度、召回率和F1分数等较高,表现出在正常和异常声音分类任务上的平衡性和稳定性.
Abstract
In order to overcome the limitations of traditional speech analysis methods,the LSTM electric power system alarm method based on time-frequency domain analysis is proposed by using time-frequency domain memory network model.At the same time,experiments are carried out on UrbanSound8K dataset to verify the effectiveness of the proposed method.The results show that the method has high accuracy,precision,recall rate and F1 score,which shows the balance and stability of normal and abnormal sound classification tasks.
关键词
电力系统/时频域分析/长短期记忆网络/报警方法Key words
electric power system/time-frequency domain analysis/long short-term memory network/alarm method引用本文复制引用
出版年
2024
NETL
NSTL
万方数据