dc110123/pinn_repo

How dc110123/pinn_repo works

This repository develops a universal physics-informed neural network for solar-panel I–V and maximum-power prediction. Its documented workflow builds a filtered panel dataset, generates synthetic curves, trains and tunes a PINN, then checks curve shape, population accuracy, and dynamic-weather behavior. The README describes the training-module separation, while the sampled main.py is only a greeting, so the documented training entry point is shown rather than an edge from main.py. Validation code references model/scaler files not present in the tree; those missing assets are not represented as repository nodes.

Main entry points: data_loader.py Data preparation · panel_shape_validation.py Curve shape checks · panel_accuracy_validation.py Population accuracy · panel_simulation.py Dynamic simulation

Components 15

Panel data

Data preparation src/data_loader.py
Synthetic I–V data src/data_loader.py

PINN training

Bayesian training src/bayesian_optimization_and_training.py
Optuna search src/bayesian_optimization_and_training.py
PINN training engine src/network_and_train_step.py
PINN model src/network_and_train_step.py

Model validation

Trained weights model/universal_pinn_weights.weights.h5
Curve shape checks simulation/Test_A/panel_shape_validation.py
Population accuracy simulation/Test_B/panel_accuracy_validation.py
Dynamic simulation simulation/Test_C/panel_simulation.py

Panel analysis

Cell-count analysis src/eda_cell_count.py
Materials analysis src/eda_panel_materials.py

Other components

Engineer
CEC database
CLI stub main.py

Connections 23

Engineer → Bayesian training: starts documented search README.md
CEC database → Data preparation: provides modules README.md
Data preparation → Synthetic I–V data: generates curves src/data_loader.py
Synthetic I–V data → Bayesian training: supplies training data README.md
Bayesian training → Optuna search: searches parameters src/bayesian_optimization_and_training.py
Bayesian training → PINN training engine: calls train_model README.md
PINN training engine → PINN model: builds and trains src/network_and_train_step.py
Engineer → Curve shape checks: runs checks README.md
Engineer → Population accuracy: runs benchmark README.md
Engineer → Dynamic simulation: runs simulation README.md
Curve shape checks → Data preparation: loads panel database simulation/Test_A/panel_shape_validation.py
Population accuracy → Data preparation: loads panel database simulation/Test_B/panel_accuracy_validation.py
Dynamic simulation → Data preparation: loads panel database simulation/Test_C/panel_simulation.py
Curve shape checks → Trained weights: loads weights simulation/Test_A/panel_shape_validation.py
Population accuracy → Trained weights: loads weights simulation/Test_B/panel_accuracy_validation.py
Dynamic simulation → Trained weights: loads weights simulation/Test_C/panel_simulation.py
Curve shape checks → PINN model: predicts curves simulation/Test_A/panel_shape_validation.py
Population accuracy → PINN model: predicts energy simulation/Test_B/panel_accuracy_validation.py
Dynamic simulation → PINN model: predicts power simulation/Test_C/panel_simulation.py
Engineer → Cell-count analysis: runs analysis src/eda_cell_count.py
Engineer → Materials analysis: runs analysis src/eda_panel_materials.py
CEC database → Cell-count analysis: provides modules src/eda_cell_count.py
CEC database → Materials analysis: provides modules src/eda_panel_materials.py

More Jupyter Notebook diagrams: roboflow/notebooks · blue-yonder/tsfresh · mitdeeplearning/introtodeeplearning · graviraja/mlops-basics · openbmb/minicpm

Drawn by GitDiagram on October 1, 2026 from the repository's file tree, README and source files. Read as Markdown ·