Seven models from 10 kW to 250 kW class — what the HMQ-EMS-300 family does, how the AI energy engine works, and which version fits which project.
The HMQ-EMS-300 series is our proprietary energy management platform — developed in-house, produced on our own SMT line, and deployed in residential, commercial and microgrid projects. Here is a technical walkthrough.
Hardware architecture. Every model combines an industrial-grade compute unit, hardware-level anti-backflow circuitry, and comprehensive interface coverage: 8x digital inputs/outputs, dual CAN bus, 8x RS485, Ethernet and 4G/5G. This means inverters, meters, heat pumps, chargers and BMS systems connect natively — no gateway cascade.
The AI energy engine. At the core is a forecasting and optimisation stack: it learns household or facility load patterns over 2-4 weeks, pulls weather data for PV generation forecasts, and combines both with dynamic electricity prices to schedule battery operation 24 hours ahead. The optimisation runs locally — no internet dependency for core functions, privacy by design.
Model lineup:
– 300S (Essential): core control and backup for compact residential systems
– 300M (Standard): full feature set including dynamic tariffs — the workhorse
– 300L / LB: display variant and large-storage integration for big residential
– 300P / PL (Professional AI, 2026 flagship): enhanced AI computing for complex C&I sites
– 300G: international variant for 60 Hz markets
– 300H / HL (AI flagship): maximum compute for microgrids, VPP aggregation and compute-power projects
Quality and compliance. The series is developed and pre-tested on our own EMC pre-compliance platform, and every unit runs automated AOI-inspected production. Firmware updates are OTA, backwards compatible.
Which model for which project? As a rule of thumb: single-family homes start at 300S/300M; multi-unit buildings and small commercial use 300M/300L; factories, microgrids and VPP sites go to 300P/300H. For a specific recommendation, contact our engineering team with your site data — we respond with a configuration proposal, not a generic datasheet.