Sorry, you need to enable JavaScript to visit this website.
.pdf asset-performance field_planning_color fit_for_purpose_color production_optimization_recovery_color kpi-box kpi-money kpi-time ::before
NVIDIA has done a proof of concept of deep learning for many oil and gas customers. The company is working with BHGE on pumps, turbines, and compressors to increase production, reduced maintenance intervals, and costs.

Traditional computing starts with an equation and collect data, compares with actual observations, and modifies simulations to refine calculation to better reflect observations.

Deep learning follows the opposite process: starts with data, creates a neural network equation, collects data, compares with observed data, and tweaks to increase accuracy over time. Essentially, deep learning is software writing software.

Predictive maintenance is the intersection between physics and deep learning, delivering 93% prediction accuracy, and reducing false negatives and false positives by 5X. Training of neural networks is now within the bounds of reality. Innovation doesn’t stop.

Let’s team together and move forward with energy to achieve the amazing benefits of AI by implementing it at-scale across all of your operating assets.

By Shanker Trivedi, Senior Vice President of Enterprise Businesses, NVIDIA

Share with Friends & Colleagues