Python-Based MPPT Control System

I need a compact Python application that takes real-time irradiance (G) and temperature (T) readings from my field sensors, runs them through a scikit-learn linear-regression model to generate a first-pass duty-cycle estimate (D_pred), then refines that value with a Perturb & Observe routine to lock onto the exact maximum-power point before sending the final duty cycle to a DC-DC boost converter.

Your job is to:

• Build and train the scikit-learn regression model (I have sample data; the script must also keep learning online as new sensor values stream in).
• Code the P&O micro-adjustment so it works seamlessly with the model output and avoids traditional drift issues.
• Interface the whole chain with the sensor feed and the converter’s PWM control pin in real time; the solution has to be fast enough for sub-second updates.
• Hand over clean, well-commented Python files plus a short README that explains setup, required libraries, and how to deploy on a Pi-class SBC.

I’ll test by running the script on live hardware; acceptance is achieved when the converter consistently tracks the MPP under rapidly changing light and temperature.