Predict electrical submersible pump failures using ML models
Usage and Configuration
To train the models contained in this package, run the workflow ESP Failure Prediction: Train Model.
To perform predictions run the workflow ESP Failure Prediction: Predict. The schedules may be adapted to use different contexts.
Follow the steps given in How do I configure packages pulled from Datagration's package repository to change the configuration of the package.
ESP Failure Prediction using Binary Classification
Binary classification is a supervised learning technique that involves predicting one of two possible classes for a given data point. In the context of failure classification, binary classification can be used to predict whether a given data point represents a failure or a non-failure.
To perform binary classification for failure classification, a collection of data for various parameters related to the system or machine being monitored is needed. This data may include variables such as temperature, pressure, vibration, and so on. The process of binary classification typically involves training the algorithm on a labeled dataset, where each data point is labeled as either a failure or a non-failure. The algorithm then uses this labeled dataset to learn the relationship between the input signals and the output class. Once the algorithm has been trained, it can be used to classify new, unseen data points as either a failure or a non-failure. This can help to identify potential issues before they result in costly downtime or equipment failure.
Data Requirements
The Binary Classification model in this package requires the following data:
- Pump Failure (tag entries)
- downhole current
- motor temperature
- pump outlet pressure
- pump intake pressure
- pump intake temperature
- runtime between failures