You can deploy this model and start earning money today!

Dijo-404/
mhd-nanofluid-ev-thermal-surrogate

Name: mhd-nanofluid-ev-thermal-surrogate
Availability: InStock

particle-swarm-optimization

Views

No views yet

MHD Hybrid Nanofluid EV Battery Thermal Management — Neural Surrogate + PSO

A physics-calibrated neural network surrogate model for predicting thermal performance of MHD hybrid nanofluid cooling systems in electric vehicle battery thermal management, coupled with Particle Swarm Optimization (PSO) for intelligent parameter optimization.

🔬 Overview

This model replaces expensive CFD simulations with a fast neural network surrogate that predicts:

Output	Description	Units
`T_max`	Maximum battery surface temperature	°C
`Nu`	Nusselt number (heat transfer coefficient)	—
`S_gen`	Total entropy generation (normalized)	—
`delta_T`	Cell-to-cell temperature difference	°C
`BL_suppression`	Boundary layer suppression	%
`k_ratio`	Thermal conductivity ratio (k_hnf/k_bf)	—

From 3 input parameters:

Input	Range	Description
`Ha`	0–60	Hartmann number (magnetic field strength)
`phi`	0.01–0.05	Nanoparticle volume fraction
`u_in`	0.05–0.30 m/s	Inlet flow velocity

📊 Model Performance

Metric	R² Score	MAE	MAPE (%)
T_max	0.979	0.76°C	1.76
Nu	0.960	0.54	2.53
S_gen	0.991	0.012	3.27
delta_T	0.980	0.21°C	1.67
BL_suppression	0.999	0.20%	6.96
k_ratio	0.999	0.002	0.17
Overall	0.985	—	—

🏗️ Architecture

Type: Multi-output MLP with residual connections
Layers: [64, 128, 128, 64] hidden units
Activation: Tanh (physics-smooth)
Training: 2000 epochs, Adam optimizer, physics-informed loss
Dataset: 5000 Latin Hypercube samples from governing equations

⚡ PSO Optimization Results

The PSO optimizer finds optimal cooling parameters:

Parameter	PSO Optimal	Paper Reference
Ha	~22–32	32.4
φ	~0.04–0.05	0.038
u₀	~0.19–0.29 m/s	0.187 m/s

Key Results:

🌡️ Peak temperature reduction: 25–35% vs conventional cooling
📉 Entropy minimization: ~31.5% reduction
✅ Battery temperature maintained within safe range (< 40°C)

🚀 Quick Start

python

import torch
import numpy as np
import json

# Load model
from model import ThermalSurrogateModel, DataNormalizer, get_model_config

config = get_model_config()
model = ThermalSurrogateModel(
    input_dim=3, hidden_dims=[64, 128, 128, 64], 
    output_dim=6, dropout=0.0
)
model.load_state_dict(torch.load('model.pt', weights_only=True))
model.eval()

normalizer = DataNormalizer.load('normalizer.json')

# Predict: [Ha=32.4, phi=0.038, u_in=0.187]
X = np.array([[32.4, 0.038, 0.187]], dtype=np.float32)
X_norm = normalizer.transform_input(X)
with torch.no_grad():
    pred = model(torch.tensor(X_norm)).numpy()
result = normalizer.inverse_transform_output(pred)
print(f"T_max: {result[0,0]:.1f}°C, Nu: {result[0,1]:.1f}")

📂 Files

File	Description
`model.pt`	Trained PyTorch model weights
`normalizer.json`	Input/output normalization parameters
`config.json`	Model architecture configuration
`model.py`	Model class definition
`data_generator.py`	Physics-based synthetic data generator
`pso_optimizer.py`	PSO optimization module
`predict.py`	High-level prediction interface
`train.py`	Training script
`evaluation.json`	Evaluation metrics
`pso_results.json`	PSO optimization results + Pareto front

📄 Based On

Paper: "AI-Assisted Thermodynamic Optimization of MHD Hybrid Nanofluid Flow for Electric Vehicle Battery Thermal Management Using Particle Swarm Optimization"

Governing Equations

Continuity: ∂u/∂x + ∂v/∂y = 0

Momentum (NS + MHD): ρ_hnf(u·∂u/∂x + v·∂u/∂y) = -∂p/∂x + μ_hnf·∇²u - σ_hnf·B₀²·u

Energy: (ρCₚ)_hnf(u·∂T/∂x + v·∂T/∂y) = k_hnf·∇²T + μ_hnf·Φ + σ_hnf·B₀²·u²

Entropy: S_gen = k_hnf/T₀²·|∇T|² + μ_hnf/T₀·Φ + σ_hnf·B₀²·u²/T₀

📜 License

Apache 2.0

Dijo-404/mhd-nanofluid-ev-thermal-surrogate