Introduction

With the popularity of Supervisory Information System (SIS), Supervisory Control and Data Acquisition (SCADA) system and Internet of Things (IoT) sensors, we can easily obtain abundant sensor data in manufacturing.

We could save manufacturing maintenance costs and prevent further damages if we can analyze the possible anomaly casualties among the IoT sensor data.

This project applies different data science approaches on anomaly detection/prediction and causality analysis.

How we built it

We created the project in Jupyter Notebooks using Google Colaboratory, a cloud-based platform where the entire team can work on the project in real-time. By leveraging the ability of Python’s VAR libraries, we implemented the statistical method of Granger Causality to aid the prediction and detection of anomalies within the data.

Challenges we ran into

Indexing the time and date for the required syntax of the Python VAR library was one of the largest challenges encountered in this project.

The .CSV file that held our real-world dataset was actually delimited by semicolons (;), not commas (,) as expected. Thus, for a while, we received a number of errors when running our code.

Accomplishments that we're proud of

The collaboration of our all-female team was incredible! Our team members had never met nor worked together before. Our greatest accomplishments have been identifying with the purpose of Granger Causal analysis and incorporating it towards an interoperable program with boundless applications.

What we learned

Statistical analysis with programming libraries and data visualizations can relay pertinent information for innovating and preventing system failure across industries.

What's next for Causality Analytics for Smart Manufacturing

Applying this knowledge to save the world!

Whether in manufacturing, critical mission systems, healthcare, and more, predicting anomalies in data helps anticipate and identify issues before they happen. This work has the ability to not only optimize resources by identifying disparities in the needs and operations of systems, but especially in the case of healthcare, has the potential to save lives by identifying anomalies under time-sensitive conditions.

Built With

  • data-analysis
  • granger-causality
  • python
  • var
Share this project:

Updates