github davidusb-geek/emhass v0.17.3
EMHASS version v0.17.3
on GitHub

4 hours ago

Thermal Upgrades, AI Integrations, and Smarter Control

Version 0.17.3 is a massive release packed with advanced thermal modeling, smarter battery management, and a huge leap forward in developer and AI-agent tooling. Thanks to the hard work of our contributors, EMHASS is now more robust, customizable, and easier to integrate than ever.

Here are the major highlights for this release:

🌡️ Thermal & HVAC Overhaul

We’ve vastly expanded the capabilities of the thermal battery model to support a wider array of real-world heating systems:

  • Advanced Weather Modeling: Added weather-compensated supply temperatures, buffer floors, and surface solar gain calculations.
  • Broader System Support: Introduced a new constant-efficiency mode to support gas, oil, and district heating systems, plus dedicated support for heat pump group hot water.
  • Finer Cost Control: Added per-load cost forecasts for thermal batteries.

🔋 Smarter Battery & Load Logic

Optimization gets a major boost with new constraints and grouping logic:

  • Deferrable Load Groups: You can now group deferrable loads together for synchronized optimization.
  • SOC Protection: Added a tunable penalty for allowing the battery State of Charge (SOC) to drop below a specific threshold.
  • Runtime Accuracy: Currently-running single-constant loads are now strictly pinned to the start of the horizon.
  • Constraint Fixes: Resolved a logic flaw in the grid/battery interaction constraints and clamped adjusted PV forecasts so they can never drop below zero.

🤖 The "AI & API" Update

We are laying the groundwork for the future of automated coding and custom integrations:

  • New API Endpoint: Easily fetch the latest optimization data with the new GET /api/v1/last-run JSON endpoint.
  • AI-Ready Documentation: EMHASS now officially supports AI coding agents! We've added llms.txt, a comprehensive AGENTS.md ruleset, and specific behavioral guardrails to help AI assistants safely contribute to the project.
  • Logging & Schemas: Added a runtime banner with per-stage execution timings and updated our parameter schemas with a new structured unit field for better machine readability.

🛠 UI, Fixes, & Stability

  • Web UI Improvements: Optimization result tables now feature sticky first rows and columns for much easier reading on large datasets.
  • Concurrency Caching Fixes: Resolved pickle.UnpicklingError and EOFError crashes that occurred during concurrent reads of cached data files by the web server.
  • API Protection: Implemented daily call caps and stale cache reuse to prevent exhausting Solcast API rate limits.
  • Internationalization: Swept the documentation to replace hardcoded €/kWh strings with a more welcoming, globally applicable currency/kWh.

A huge thank you to the contributors who made this release possible: @snauwaertc, @OptimalNothing90, @hossamnagy, @kallegrens, @carposio, @lutorm, @sokorn, @m8ram, @torsteinelv, @JosephSalisbury, @rdeknijf, and @dependabot!

Improvement

  • feat: weather-compensated supply T and buffer floor for thermal_battery (@snauwaertc)
  • feat: per-load cost forecasts and surface solar gain on thermal_battery (@snauwaertc)
  • feat: constant-efficiency mode for thermal_battery (gas / oil / district) (@snauwaertc)
  • Add a penalty to the battery SOC being below a threshold. (@lutorm)
  • feat(api): add GET /api/v1/last-run JSON endpoint (@OptimalNothing90)
  • feat(schema): add unit field to param_definitions.json (#826) (@OptimalNothing90)
  • chore: add pre-commit hooks mirroring CI ruff checks (@kallegrens)
  • build(deps): update skforecast requirement from <0.21.0,>=0.19.1 to >=0.19.1,<0.23.0 (@dependabot[bot])
  • Feature: heatpump group hotwater (@sokorn)
  • Generalize Load cost forecast value (@m8ram)
  • feat(logging): runtime banner + per-stage timings (closes #793) (@OptimalNothing90)
  • Pin currently-running single-constant loads to start of horizon (@carposio)
  • Add sticky first row and column to optimization results table (@carposio)
  • Update optimization.py Fix logic flaw in grid/battery interaction constraints. (@torsteinelv)
  • Add llms.txt support to documentation (@OptimalNothing90)
  • New version v0.17.2 (@davidusb-geek)
  • Adds support for deferrable load groups (@JosephSalisbury)

Documentation

Fix

  • fix(ci): skip SonarQube Scan on fork PRs (#857) (@OptimalNothing90)
  • Fix pickle.UnpicklingError crash on concurrent reads of cached data files (@hossamnagy)
  • fix(test): pin load_forecast_method in test_load_deactivation_zero_operating_timesteps (#856) (@OptimalNothing90)
  • fix: handle EOFError when reading pickle cache in web dashboard routes (@hossamnagy)
  • fix(config): align config_defaults.json with param_definitions.json per #830 Option-B decision (@OptimalNothing90)
  • fix: deduplicate DataFrame index before asfreq() in _apply_df_freq_horizon (@hossamnagy)
  • fix: normalize None timesteps and use .get() for safe optim_conf reads (@hossamnagy)
  • fix(schema): correct 5 default mismatches in param_definitions.json (@OptimalNothing90)
  • fix(utils): wire ignore_pv_feedback_during_curtailment runtime flag (#818) (@OptimalNothing90)
  • fix: regression_model default typo (AdaBoostRegression -> AdaBoostRegressor) (@OptimalNothing90)
  • Fix sticky header horizontal scroll position on initial show (@carposio)
  • fix(forecast): clamp adjusted PV forecast to >= 0 (fixes #521) (@rdeknijf)
  • fix: prevent thermal battery infeasibility when q_input_start is zero (@OptimalNothing90)
  • fix: add state_class to published sensors and prevent ML crash on short history (@kallegrens)
  • fix: prevent Solcast rate limit exhaustion with daily call cap and stale cache reuse (@kallegrens)

What's Changed

New Contributors

Full Changelog: v0.17.2...v0.17.3

Check out latest releases or
releases around davidusb-geek/emhass v0.17.3

Don't miss a new emhass release

NewReleases is sending notifications on new releases.

Get notifications