Computer Vision Solution for OPF Flare Control

Supporting operations and safety with Machine Learning

Computer Vision Solution for OPF Flare Control

Supporting operations and safety with Machine Learning

Computer Vision Solution for OPF Flare Control

Supporting operations and safety with Machine Learning
The Client
Sakhalin Energy, Russia
The Company's activities include production, transportation, processing, and marketing of oil and natural gas.
The Client
Sakhalin Energy, Russia
The Company's activities include production, transportation, processing, and marketing of oil and natural gas.
Python, SciKitLearn/Pandas, Pytorch, VAE (1D-CNN/LSTM/GRU), TensorBoard, Random Forest, Decision Tree, K-nearest neighbours, Support Vector Machines
Python, SciKitLearn/Pandas, Pytorch, VAE (1D-CNN/LSTM/GRU), TensorBoard, Random Forest, Decision Tree, K-nearest neighbours, Support Vector Machines
Python, SciKitLearn/Pandas, Pytorch, VAE (1D-CNN/LSTM/GRU), TensorBoard, Random Forest, Decision Tree, K-nearest neighbours, Support Vector Machines

Challenge

Challenge

Challenge


The task was to develop an Machine Learning model to assist an operator who visually controls the performance of a gas flare. Missing a certain behavior of the flare can result in flare tip damage and plant shut down, which leads to tremendous losses in idle time.

The task was to develop an Machine Learning model to assist an operator who visually controls the performance of a gas flare. Missing a certain behavior of the flare can result in flare tip damage and plant shut down, which leads to tremendous losses in idle time.

The task was to develop an Machine Learning model to assist an operator who visually controls the performance of a gas flare. Missing a certain behavior of the flare can result in flare tip damage and plant shut down, which leads to tremendous losses in idle time.

Approach

The R&D process was performed on the Client's premises, as most of their data are restricted. After training the Client's team on Machine Learning fundamentals, we started working on the project. As the result, we've built several models based on decision trees.

Approach

The R&D process was performed on the Client's premises, as most of their data are restricted. After training the Client's team on Machine Learning fundamentals, we started working on the project. As the result, we've built several models based on decision trees.

Approach

The R&D process was performed on the Client's premises, as most of their data are restricted. After training the Client's team on Machine Learning fundamentals, we started working on the project. As the result, we've built several models based on decision trees.
Stabilize image
Assess if the frame is suitable for analysis
Localize burning in upper and lower parts of the image
Stabilize image
Assess if the frame is suitable for analysis
Localize burning in upper and lower parts of the image
Stabilize image
Assess if the frame is suitable for analysis
Localize burning in upper and lower parts of the image
Due to Client's restrictions we were limited to a virtual machine without GPU
Due to Client's restrictions we were limited to a virtual machine without GPU
Due to Client's restrictions we were limited to a virtual machine without GPU

Solution

As a result of joint effort the Computer Vision driven tools were developed that help the operator monitor the flare behaviour. Now he can identify unwanted flare burning and take timely action.

An option with an infrared camera was offered to the Client, who is now considering it.

Solution

As a result of joint effort the Computer Vision driven tools were developed that help the operator monitor the flare behaviour. Now he can identify unwanted flare burning and take timely action.

An option with an infrared camera was offered to the Client, who is now considering it.

Solution

As a result of joint effort the Computer Vision driven tools were developed that help the operator monitor the flare behaviour. Now he can identify unwanted flare burning and take timely action.

An option with an infrared camera was offered to the Client, who is now considering it.
They successfully achieved most of the goals for the project and offered the most suitable solutions to problems
They successfully achieved most of the goals for the project and offered the most suitable solutions to problems
They successfully achieved most of the goals for the project and offered the most suitable solutions to problems
Client's Testimonial
Initially, we hired ENBISYS to set up and train our in-house Machine Learning team. They are very result-oriented and outstanding engineers. The workflow was collaborative and effective. Their expertise convinced us in their ability to handle even the most sensitive data properly and get the results we expected.
Client's Testimonial
Initially, we hired ENBISYS to set up and train our in-house Machine Learning team. They are very result-oriented and outstanding engineers. The workflow was collaborative and effective. Their expertise convinced us in their ability to handle even the most sensitive data properly and get the results we expected.
Client's Testimonial
Initially, we hired ENBISYS to set up and train our in-house Machine Learning team. They are very result-oriented and outstanding engineers. The workflow was collaborative and effective. Their expertise convinced us in their ability to handle even the most sensitive data properly and get the results we expected.
Let's discuss your Case!
You agree to our Privacy Policy and Terms of Service
Let's discuss your Case!
You agree to our Privacy Policy and Terms of Service
Let's discuss your Case!
You agree to our Privacy Policy and Terms of Service