Engineering · Machine learning

Predictive CFD with Neural Networks

A compact surrogate model for predicting indoor temperature from simulation data.

Period
2022
Status
Course research project
Perspectives
Engineering · Research

30-second case brief

Question, responsibility, method, result.

Question
A compact surrogate model for predicting indoor temperature from simulation data.
Responsibility
Data preparation, ANN development, evaluation, and technical reporting within a three-person team
Methods
20,000+ hourly observations · 13 inputs
Result
20,000+ hourly records · 13 simulation inputs
The surrogate maps 13 simulation inputs to indoor ambient temperature and is evaluated against held-out records.

Project scope and responsibility

Context
Drexel University
Role
Data preparation, ANN development, evaluation, and technical reporting within a three-person team
Dataset
20,000+ hourly observations · 13 inputs
Model
Feed-forward ANN · ADAM optimization
Team
V. Edske · J. Moussa · R. Tajik
01 / 02

Overview

Detailed building simulations can be slow to rerun during iterative analysis. This project tested whether an artificial neural network could learn the relationship between 13 simulation inputs and indoor ambient temperature, creating a faster surrogate for a defined case and dataset.

A compact architecture was trained on normalized simulation data.
02 / 02

Evaluation

More than 20,000 hourly records were separated into 70 percent training and 30 percent testing subsets. The selected configuration used sigmoid activation, ADAM optimization, a batch size of 36, and 100 epochs. Error definitions and validation scope remain visible alongside the result.

Held-out predictions follow the source simulation pattern across test observations.
Correlation checks complement the summary error metric.
Back to project navigation ↑

Contact

Interested in hiring me?