Skip to content

kaushikmupadhya/Anomaly-Detection-in-Time-Series-Data

Repository files navigation

Anomaly-Detection-in-Time-Series-Data

Anomaly Detection in Time Series Data using Autoencoders approach.

A Research and Development Project to find anomalies in a time series data using deep learning auto encoder approach and finding Remaining Useful Life Characteristics of the data, as described in the paper "Wavelet Filter-based Weak Signature Detection Method and its Application on Roller Bearing Prognostics" by Hai Qiu, Jay Lee, and Jing Lin. The dataset used for this project is the NASA Prognostics Center of Excellence Data Set Repository.

Dataset

The NASA Prognostics Center of Excellence Data Set Repository contains several datasets related to prognostics and health management. For this project, we will be using the "Rolling element bearings dataset" which contains vibration signals from bearings with normal and faulty conditions. The dataset can be downloaded from the following link: https://www.nasa.gov/content/prognostics-center-of-excellence-data-set-repository

The paper by Hai Qiu, Jay Lee, and Jing Lin describes the application of this method to roller bearing prognostics. In this project, we will apply this method to the NASA Prognostics Center of Excellence Data Set Repository to detect anomalies in the vibration signals of the rolling element bearings.

Requirements

To run this project, you will need the following software installed on your computer:

  • Python (version 3 or later)
  • Jupyter Notebook
  • NumPy
  • SciPy
  • Matplotlib

Conclusion

The detection of anomalies in time series data is a crucial task in many industries, from predictive maintenance to finance. The autencoders detection method provides a tool for detecting anomalies in noisy data. By applying this method to the NASA Prognostics Center of Excellence Data Set Repository, we can gain insights into the health of rolling element bearings and potentially prevent catastrophic failures.

Reference Paper: Hai Qiu, Jay Lee, Jing Lin. “Wavelet Filter-based Weak Signature Detection Method and its Application on Roller Bearing Prognostics.” Journal of Sound and Vibration 289 (2006) 1066-1090

⭐ PS: Please leave a star to support.