Recent literature (2021–2025) — compact table

Key contribution / finding

Quick note (strength / limitation)

Reviews advances in MPC, ML and hybrid controllers; highlights CO₂ & cost gains and shortage of long-term field studies.

Up-to-date algorithmic view; notes validation gap.

Demonstrated ML (SVR and variants) improves forecast accuracy and overall microgrid/HEMS performance.

Strong ML results; mostly simulation/bench datasets.

ML-based energy monitoring / NILM methods

Presents systematic ML procedure for reliable monitoring (useful for HEMS sensing & decision-making).

Useful for measurement & data-driven control; needs large-scale tests.

Global scan of residential HEMS products, definitions, and deployment barriers (interoperability, business models).

Practical industry-oriented; excellent for real-world constraints.

Systematic review of energy-management models (2018–2024)

PRISMA-based synthesis of model types used in smart homes; highlights trend toward hybrid ML+MPC approaches.

Systematic & recent; covers 2018–2024 literature window.

“MPC for residential demand flexibility”, 2023

MPC applied to residential DERs (case study)

Shows MPC yields significant demand flexibility and cost savings when coordinating PV, ESS and loads.

Case study with measured benefits; depends on forecast quality.

Cai et al. / SARLEM project (learning-based MPC), 2022–2023

RL + MPC for HEMS (thermal/hot-water control)

Proposes MPC-stabilized RL that balances comfort and energy objectives; improves robustness.

Novel control theory + simulations; limited field validation.

TODEREAN et al., 2025 / Demand response review

Recent DR strategies & optimization for buildings

Surveys DR optimization methods and integration with building energy systems.

Good for DR-centric HEMS design and aggregator roles.

Publishes datasets for NILM/HEMS evaluation (NILMforIHEM, HEMS2IEA) to support reproducible research.

Important data resource to validate algorithms.

MDPI / Sustainability 2024 (MPC + PV+EV+BESS)

MPC strategies for smart homes with PV, EV, BESS

Scenario-based MPC to mitigate prediction errors; shows economic benefits when uncertainty handled.

Practical control approach for real installations.

Advantages (recent-evidence)

Improved cost savings & tariff optimization

MPC/ML schedulers reduce bills by shifting loads/using PV+ESS.

Better forecast-driven decisions (lower mismatch)

ML forecasting reduces PV/load prediction errors, improving controller outcomes.

HEMS can provide flexibility to aggregators (recent DR optimization studies).

Growing availability of datasets & product maturity

New datasets and market scans accelerate validation and commercial deployment.

Disadvantages / challenges (recent-evidence)