snazzybean/roommind

What it solves

RoomMind is an intelligent climate control system for Home Assistant that replaces simple on/off thermostats with a self-learning thermal model. It prevents temperature overshoot, reduces unnecessary heating by accounting for solar gain, and optimizes energy use across multiple heating and cooling devices.

How it works

The system uses an Extended Kalman Filter (EKF) to create a per-room thermal model that learns heat loss rates and solar responsiveness over time. Once calibrated, a Model Predictive Control (MPC) optimizer calculates proportional power outputs rather than binary on/off commands. It integrates with Home Assistant entities to handle presence detection, vacation modes, and window/door pauses, and can even estimate surface humidity to prevent mold growth.

Who it’s for

Homeowners using Home Assistant who want more precise, energy-efficient climate control and a centralized management panel for their smart thermostats and AC units.

Highlights

• Self-Learning MPC: Automatically learns room thermal behavior to provide smoother temperature curves.
• Solar Gain Awareness: Estimates solar irradiance to reduce heating when the sun provides warmth.
• Mold Prevention: Uses the DIN 4108-2 method to estimate surface humidity and raise temperatures to prevent mold.
• Proportional Valve Control: Sends calculated setpoints to TRVs to avoid the abrupt changes of on/off control.
• Heat Source Orchestration: Automatically chooses the most efficient device (e.g., TRV vs. AC) based on outdoor conditions.
• Integrated Analytics: Provides temperature charts and model predictions over 24h to 90 days.