Avatar

Renato G. Nascimento

AI Applied Scientist

Intel

Biography

I’m an Applied Scientist and AI Engineer with over 15 years of R&D experience in software for consumer and industrial applications. I’ve developed and deployed AI solutions, empowering data science teams to build and visualize multiple analytics models and techniques at scale, deployed to diverse cloud services. With a BSc in Computer Science and Ph.D. in Mechanical Engineering, I’ve created novel methods combining physics-based knowledge with Deep Learning frameworks creating hybrid AI solutions. In my research in collaboration with the Diagnostics and Prognostics lab at NASA Ames, we enabled fast and accurate monitoring of Lithium-ion batteries with probabilistic hybrid machine learning.

Interests

  • Machine Learning
  • Physics-Informed Neural Networks
  • Data Visualization

Education

  • PhD in Mechanical Engineering, 2022

    University of Central Florida

  • MSc in Aerospace Engineering, 2020

    University of Central Florida

  • BSc in Computer Science, 2014

    UNESP - SP, Brazil

Featured Publications

Renato G. Nascimento, Felipe A. C. Viana, Matteo Corbetta, Chetan S. Kulkarni
September 2021 Journal of Power Sources

Hybrid physics-informed neural networks for lithium-ion battery modeling and prognosis

Lithium-ion batteries are commonly used to power unmanned aircraft vehicles (UAVs). The ability to model and forecast remaining useful life of these batteries enables UAV reliability assurance. Building principled accurate models is challenging due to the complex electrochemistry that governs battery operation. Alternatively, reduced order models have the advantage of capturing the overall behavior of battery discharge, although they suffer from simplifications and residual discrepancy. This paper presents a hybrid modeling approach that directly implements physics within deep neural networks. While most of the input–output relationship is captured by reduced-order models, data-driven kernels reduce the gap between predictions and observations. A reduced-order model based on Nernst and Butler–Volmer equations represents the overall battery discharge, and a multilayer perceptron models the battery non-ideal voltage. Battery aging is characterized by time-dependent internal resistance and the amount of available Li-ions, which are modeled through an ensemble of variational Bayesian multilayer perceptrons. The approach is validated using data publicly available through the NASA Prognostics Center of Excellence website. Results showed that our hybrid battery prognosis model can be successfully calibrated, even with a limited number of observations. Moreover, the model can help optimizing battery operation by offering long-term forecast of battery capacity.

PDF

Experience

 
 
 
 
 

AI Applied Scientist

Intel

May 2022 – Present Santa Clara, CA
AI Emerging Tecnologies
 
 
 
 
 

Solutions Architect I Inter

Amazon

Jun 2020 – Aug 2020 Cupertino, CA
Developed deep learning solutions for the Neuron SDK and AWS Inferentia
 
 
 
 
 

Intern with KBR

NASA Ames Research Center

May 2020 – Aug 2020 Moffett Field, CA
Diagnostics and prognostics modeling of drone powertrain systems with physics-informed machine learning
 
 
 
 
 

Graduate Research Assistant

Probabilistic Mechanics Laboratory @ UCF

Aug 2018 – May 2022 Orlando, FL
Key Projects:

  • Physics-informed neural networks for cumulative damage modeling applied in fatigue estimation of aircraft fuselage panels.
  • Reverse engineering quadrotor drone dynamics with physics-informed neural networks.
 
 
 
 
 

Chief consultant and Owner

Renato Giorgiani Software ME

Jan 2015 – Aug 2018 SP, Brazil

Performed software design, research, and development for GE Global Research (Niskayuna, NY) and BHGE Digital (San Ramon, CA).

Key Projects:

  • AI Framework to build and visualize multiple analytics models and techniques in large scale deployed to cloud services. Demos presented in Google Next (2018, 2019) and Microsoft Azure AI 2019 with interactive cluster visualization.
  • Probabilistic cumulative damage model for aircraft engines application: real-time processing and visualization of execution data (simulated engine operation) in a Raspberry PI board.
  • Big data visualization tools: unique and customs web visualization tools using the advanced components and visualization (e.g., d3.js and Plot.ly).
  • User-defined model application: NodeJs embed web application (self-runnable app) to run custom R and Python scripts with data loading and visualization over the web browser.
  • Platform frontend remodeling and development. Convert plain javascript code to structured AngularJs application, creating reusable components. NodeJs embed web application, including session management and security layers, and optimal build and deployment scripts.
 
 
 
 
 

R&D Intern

Probabilistics Design Laboratory @ GE Global Research

May 2014 – Aug 2014 Niskayuna, NY

Supported key research and development activities for Usage Based Lifing project

Key Projects:

  • Big data visualization tools: development of an “in-browser” data analytics and visualization application. Parallel data analysis with web workers and fast and optimal visualization charts to process and visualize big data files directly on the browser.
  • HTLM5/JavaScript advanced Information Visualization tools.

Curriculum Vitae

Full Curriculum Vitae

Download

Publications

Renato G. Nascimento, Felipe A. C. Viana, Matteo Corbetta, Chetan S. Kulkarni (2021). Hybrid physics-informed neural networks for lithium-ion battery modeling and prognosis. Journal of Power Sources.

PDF

Renato G. Nascimento, Felipe A. C. Viana, Matteo Corbetta, Chetan S. Kulkarni (2021). Usage-based Lifing of Lithium-Ion Battery with Hybrid Physics-Informed Neural Networks. AIAA AVIATION 2021 FORUM.

Kajetan Fricke, Renato G. Nascimento, Felipe A. C. Viana (2021). Quadcopter Soft Vertical Landing Control with Hybrid Physics-informed Machine Learning. AIAA SciTech Forum.

Felipe A. C. Viana, Renato G. Nascimento, Arinan Dourado, Yigit Yucesan (2020). Estimating model inadequacy in ordinary differential equations with physics-informed neural networks. Computers & Structures.

PDF Code

Andre V. Zuben, Renato G. Nascimento, Felipe A. C. Viana (2020). Visualizing corrosion in automobiles using generative adversarial networks. Annual Conference of the PHM Society.

PDF

Renato G. Nascimento, Kajetan Fricke, Felipe A. C. Viana (2020). A tutorial on solving ordinary differential equations using Python and hybrid physics-informed neural network. Engineering Applications of Artificial Intelligence Journal.

PDF Code

Renato G. Nascimento, Felipe A. C. Viana (2020). Cumulative Damage Modeling with Recurrent Neural Networks. AIAA Journal.

PDF Code

Renato G. Nascimento, Kajetan Fricke, Felipe A. C. Viana (2020). Quadcopter Control Optimization through Machine Learning. AIAA SciTech Forum.

Renato G. Nascimento, Felipe A. C. Viana (2020). Satellite Image Classification and Segmentation with Transfer Learning. AIAA SciTech Forum.

Renato G. Nascimento, Felipe A. C. Viana (2019). Fleet prognosis with physics-informed recurrent neural networks. The 12th International Workshop on Structural Health Monitoring.

Patents

Autonomous Hybrid Analytics Modeling Platform
Integrated development environment for analytic authoring
Systems and methods for optimizing graphics processing for rapid large data visualization
Self-aware and self-registering software & analytics platform components
System architecture for secure and rapid development, deployment and management of analytics and software systems

Contact