/Oil-Gas-Leakage-Analysis-and-Equipment-Failure-Detection

Examine the all the leakages happened from 2010-2017 and apply machine learning to detect equipment failure

Primary LanguageJupyter Notebook

Rolling in the Deep

Description: A journey from oil leakage to applying AI to predict equipment failures

** Project status: ** Completed

Dataset

Oil Pipeline Accidents, 2010-Present

Location: database.csv

Description: This database includes a record for each oil pipeline leak or spills reported to the Pipeline and Hazardous Materials Safety Administration since 2010. These records include the incident date and time, operator and pipeline, cause of the incident, type of hazardous liquid and quantity lost, injuries and fatalities, and associated costs.

Predictive Equipment Failures

Location: equipfails

Description: Predict downhole equipment failures using sensor data!

Features summary: There are two types of sensor columns in these data sets :

  • measure columns - These columns are a single measurement for the sensor.

  • histogram bin columns - These columns are set of 10 columns that are different bins of a sensor that show its distribution over time.

Technologies:

python, pandas, matplotlib, plotly, scikit-learn

Result

All of the results can be found in the medium post

However, the visualizations are all static. For interactive graphs, go to Accident_visulization.ipynb and execute the file.