Scholarlink

In the smart manufacturing, automatic detection of anomalies is significant to maximize the productivity. The time series data from the process of production are generally the difficult sequence and its evaluation involved numerous variables. However, the anomaly detection-based Deep Learning (DL) algorithms is the effective method. In this research, the State value with Long Short-Term Memory (S-LSTM) method is proposed for the detection of anomalies in time series data. It is collected from the resource and is balanced by using the Synthetic Minority Oversampling Technique (SMOTE) method. Then the features are extracted and classified by using the S-LSTM method with high detection accuracy and less false detection of anomalies. The S-LSTM method is analyzed with metrics of recall, precision, accuracy and f1-score. The S-LSTM method obtained 99.83% accuracy, 99.80% precision, 99.54% recall, and 99.33% f1-score that is superior to existing methods like Measurement Intrusion Detection System (MIDS).

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

State Value with Long Short-Term Memory-Based Anomaly Detection in Time Series Data

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Haider Alabdeli,
  • Abbas Hameed Abdul Hussein,
  • Arelly Madhavi

摘要

In the smart manufacturing, automatic detection of anomalies is significant to maximize the productivity. The time series data from the process of production are generally the difficult sequence and its evaluation involved numerous variables. However, the anomaly detection-based Deep Learning (DL) algorithms is the effective method. In this research, the State value with Long Short-Term Memory (S-LSTM) method is proposed for the detection of anomalies in time series data. It is collected from the resource and is balanced by using the Synthetic Minority Oversampling Technique (SMOTE) method. Then the features are extracted and classified by using the S-LSTM method with high detection accuracy and less false detection of anomalies. The S-LSTM method is analyzed with metrics of recall, precision, accuracy and f1-score. The S-LSTM method obtained 99.83% accuracy, 99.80% precision, 99.54% recall, and 99.33% f1-score that is superior to existing methods like Measurement Intrusion Detection System (MIDS).