The oil and gas sector is a complex industry that relies on a vast array of processes and equipment operating continuously to ensure stable, safe, and reliable operations. Despite rigorous monitoring, anomalies such as reactor fouling, leakage due to extended corrosion, and blower fouling with dust can occur. These anomalies can have severe impacts on the refining process and are often not identifiable through classical monitoring techniques.

Detecting these anomalies early is crucial for several reasons:

1. Safety: Early detection of anomalies can prevent hazardous situations, ensuring the safety of personnel and the environment.
2. Operational Stability: Identifying issues before they escalate helps maintain stable operations, reducing the risk of unexpected shutdowns.
3. Reduced Downtime: Proactive anomaly detection allows for timely maintenance, minimizing downtime and ensuring continuous production.
4. Cost Savings: By addressing issues early, we can reduce maintenance costs and avoid expensive repairs or replacements.
5. Efficiency: Maintaining equipment in optimal condition ensures efficient operation, leading to better resource utilization and energy savings.

Thanks to the growth of digital transformation at TotalEnergies and the development of advanced machine learning models like Deep Neural Networks, Convolutional Neural Networks (CNN), and Long Short-Term Memory (LSTM) networks, it is now possible to raise alerts in advance or at the right time to take necessary actions. This internship is intended to benchmark different algorithms with various use cases and create a cartography of applications pertinent to the type of anomaly detected, their frequency of occurrence, and the type of alerts to be raised.

As a Machine Learning for Early Anomaly Detection in Oil and Gas: Benchmarking and Cartographic Preparation M/F trainee, your missions will be : 

• Bibliography Search: Conduct a comprehensive literature review to identify various techniques used for detecting different types of anomalies, not only in the oil and gas sector but also in other industries. This will help us draw inspiration and apply best practices from diverse fields.
• Benchmarking Algorithms: Evaluate different machine learning algorithms, including Deep Neural Networks, Convolutional Neural Networks (CNN), and Long Short-Term Memory (LSTM) models, to determine their effectiveness in detecting anomalies.
• Model Adjustment and Transformation: Adapt and fine-tune machine learning models to apply them to our specific use cases, ensuring they are optimized for our operational environment.
• Cartography of Applications: Develop a detailed mapping of how different models perform across various anomaly detection use cases. This exercise will provide deep insights into the strengths and particularities of each model, identifying the most suitable applications based on the type of anomaly, its frequency, and the nature of alerts required. Successfully completing this task will significantly enhance our understanding and application of machine learning in anomaly detection, potentially contributing to advancements in the field.
• Collaboration: Work closely with a multidisciplinary team, including data scientists, process engineers, and operators from both the central team and site locations. This collaboration will ensure that the models are practically applicable and effective in real-world settings.

You will evolve within a team of experienced professionals and with a tutor-coach, the reference for your future profession. Individualized support will help you develop your autonomy and lead you to your diploma!

Currently enrolled in an engineering school or Master's program in the Data Sciences and R&D field, are you looking for an 6-month end-of-study internship starting in March 2025?

Are you pursuing your master's in Data Science? We're looking for enthusiastic individuals who are eager to apply their skills in a real-world setting. A background or basic knowledge in Chemical or Process Engineering is nice to have but not required.

Technical Skills: Strong understanding of machine learning algorithms and their applications. Experience with Python and machine learning libraries such as TensorFlow or PyTorch. Familiarity with data mining, data cleaning, data analysis, and model development, testing, and fine-tuning.

Analytical Abilities: Do you enjoy data analysis and creating stories from operational data? Are you passionate about data cleaning and transforming raw data into valuable insights? If so, we want you!

Communication Skills: Good communication skills are required for effective collaboration, presentations, and report writing. Proficiency in English is a must.

Team Player: Ability to work independently and as part of a multidisciplinary team, including data scientists, process engineers, and operators.

So don't wait any longer, apply to join our team!