arxiv: 2604.24766 · v1 · submitted 2026-04-04 · 💻 cs.LG · cs.AI· eess.SP

Recognition: 2 theorem links

· Lean Theorem

GCA-BULF: A Bottom-Up Framework for Short-Term Load Forecasting Using Grouped Critical Appliances

Yunhao Yao , Jinwei Fang , Puhan Luo , Zhiqiang Wang , Jiahui Hou , Xiang-Yang Li

Authors on Pith no claims yet

Pith reviewed 2026-05-13 18:45 UTC · model grok-4.3

classification 💻 cs.LG cs.AIeess.SP

keywords short-term load forecastingbottom-up approachcritical appliancesappliance groupingelectricity consumptionenergy managementload predictionbuilding energy

0 comments

The pith

A framework selects critical appliances and groups them to forecast total electricity loads more accurately than previous methods.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents GCA-BULF, a bottom-up framework that forecasts short-term building electricity loads by focusing only on critical appliances. It ranks appliances according to power use, on-off frequency, and usage regularity, selects the important ones with iterative decomposition, groups them by shared spatial and temporal patterns, and combines their predictions to estimate total consumption. This design reduces the expense of full appliance monitoring while capturing more detail than simple aggregate forecasts. Experiments on residential and office data demonstrate clear accuracy improvements over existing approaches.

Core claim

GCA-BULF identifies critical appliances through a filtering process that ranks them by power consumption, switching frequency, and periodicity and applies iterative decomposition to select those that best represent total load; it then groups correlated appliances for joint prediction and merges the group outputs to produce the final total load forecast, yielding hourly accuracy gains of 20.85 to 57.88 percent over top-down methods and 33.03 to 92.48 percent over other bottom-up methods.

What carries the argument

The Critical Appliance Filtering module that ranks appliances by power, switching frequency and periodicity and selects a minimal set via iterative load decomposition for reliable total load reconstruction from group forecasts.

If this is right

Total load forecasts can be built from a small number of appliance groups without tracking every device.
Spatial and temporal correlations between appliances allow effective clustering for joint forecasting.
The method applies to both residential homes and office buildings with similar accuracy improvements.
Combining multiple group-level predictions refines the overall forecast beyond single-group results.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

This selection process might generalize to other time-series forecasting tasks where full data collection is expensive.
Real-world deployment could integrate with smart home systems to enable automated load shifting based on the forecasts.
Further work could test whether the filtering criteria remain effective across different climates or building types.

Load-bearing premise

The appliances chosen by ranking power, frequency, and periodicity, plus iterative decomposition, include enough information that their group forecasts can reconstruct total load without large errors from the excluded appliances.

What would settle it

A test dataset where using forecasts from the selected critical appliance groups yields average hourly errors exceeding those from full bottom-up monitoring by more than 20 percent would falsify the reliability of the selection.

Figures

Figures reproduced from arXiv: 2604.24766 by Jiahui Hou, Jinwei Fang, Puhan Luo, Xiang-Yang Li, Yunhao Yao, Zhiqiang Wang.

**Figure 1.** Figure 1: System Overview of GCA-BULF. form a set D. Since some appliances may be added, removed or relocated, real-time monitoring of all appliances’ loads is costly and impractical. Therefore, we define a subset D′ = {d1, d2, ..., dn} ⊆ D, representing the appliances whose loads can be monitored. Let L = [l1, l2, ..., lm] denote the total load at each time step for all appliances in D, and Li = [l i 1 , li 2 , ..… view at source ↗

**Figure 2.** Figure 2: Examples of Critical Appliance Filtering on the UK-DALE Dataset. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 4.** Figure 4: The Structure of Our Collaborative Load Forecasting Framework. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗

**Figure 4.** Figure 4: AggNet(·) provides a preliminary total load forecast, while GroupNet1(·), ..., GroupNetg(·) predict the future loads of all critical appliance groups significantly affecting total load fluctuations. By integrating the group-level forecasts with the preliminary total load prediction, CoP redictor(·) refines the total load estimation as follows: h (0) i = [l pre agg, lpre 1 , lpre 2 , ..., lpre g ], h (ι) i … view at source ↗

**Figure 5.** Figure 5: MAE (KW) and MAPE of GCA-BULF Under Varying Model [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗

**Figure 7.** Figure 7: Comparison with Existing Top-Down Methods on UK-DALE. [PITH_FULL_IMAGE:figures/full_fig_p008_7.png] view at source ↗

**Figure 9.** Figure 9: Case Study: GCA-BULF vs. Existing Top-Down Methods on UK-DALE. [PITH_FULL_IMAGE:figures/full_fig_p008_9.png] view at source ↗

**Figure 10.** Figure 10: Comparison with Existing Bottom-Up Methods on UK-DALE. [PITH_FULL_IMAGE:figures/full_fig_p009_10.png] view at source ↗

**Figure 12.** Figure 12: Case Study: GCA-BULF vs. Existing Bottom-Up Methods on UK-DALE. [PITH_FULL_IMAGE:figures/full_fig_p009_12.png] view at source ↗

read the original abstract

With the rise of time-of-use and tiered electricity pricing, energy consumers are encouraged to adopt peak-shifting strategies by automatically controlling high-power appliances. These help lower energy costs while enhancing the power grid's stability. To support such energy management with high resilience and responsiveness, reliable short-term load forecasting (STLF) plays a critical role. STLF predicts electricity consumption over time horizons ranging from minutes to days, using historical data, temporal patterns, and contextual factors. Traditional top-down forecasting methods struggle to capture the complex consumption patterns of diverse and mixed appliance loads. Although bottom-up methods improve forecasting accuracy by integrating appliance-level data, monitoring all appliances is costly, and many do not meaningfully impact total load prediction. Therefore, we propose GCA-BULF, a bottom-up short-term load forecasting framework based on grouped critical appliances, supported by three key designs. First, the Critical Appliance Filtering module ranks appliances according to their power consumption, switching frequency, and usage pattern periodicity, and identifies critical ones through iterative load decomposition. Next, the Related Appliance Grouping module clusters these appliances based on spatial and temporal correlations for group-level forecasting. Finally, the Collaborative Load Forecasting module refines the total load prediction by combining multiple group-level forecasts. We evaluate GCA-BULF on residential and office building load forecasting tasks. Experimental results reveal that GCA-BULF improves hourly total load forecasting by 20.85%-57.88% compared to existing top-down methods and by 33.03%-92.48% compared to bottom-up methods.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

GCA-BULF filters to critical appliances then groups them for collaborative forecasting, with reported gains that still need a residual-error check to confirm they survive reconstruction.

read the letter

The main point is a practical bottom-up pipeline that ranks appliances by power, switching frequency and periodicity, drops the rest via iterative decomposition, clusters the survivors by spatial-temporal correlation, and combines their forecasts. The specific three-module stack is not a standard extension of prior top-down or full bottom-up work, so the named framework is new on the surface. It targets a clear cost-accuracy trade-off in building load forecasting, where monitoring every appliance is expensive and many loads add little signal. The grouping and collaboration steps are a sensible way to recover some of the lost information without full coverage. The abstract claims 20-57% better hourly accuracy than top-down baselines and 33-92% better than other bottom-up ones on residential and office data, which would matter for time-of-use pricing and grid stability if the numbers hold. The filtering logic itself is straightforward and reproducible from the description. The soft spot is exactly the one flagged in the stress test: no quantitative bound or ablation on how much total load variance is left in the residual after critical-appliance selection. If low-power but high-variance devices are dropped, the collaborative stage cannot recover what was never measured, yet the reported gains are only on the final selected subset. The abstract also gives no baseline code details, cross-validation scheme, or error bars, so the size of the improvement is hard to judge from the given text alone. This paper is for applied researchers and engineers who already work on short-term load forecasting for buildings and want a cheaper monitoring setup. A reader who needs to implement or extend the method would get value from the pipeline description and the empirical numbers, even if they have to add their own validation. It is worth sending to peer review because the core engineering claim is testable and the problem is real; the referees can check the missing bounds and baseline rigor without the paper being incoherent on its own terms.

Referee Report

2 major / 2 minor

Summary. The paper proposes GCA-BULF, a bottom-up short-term load forecasting framework that first filters critical appliances via ranking on power, switching frequency and periodicity followed by iterative decomposition, then groups them by spatial-temporal correlations, and finally combines group-level forecasts to predict total load. It reports 20.85%-57.88% improvement over top-down methods and 33.03%-92.48% over bottom-up methods on residential and office building datasets for hourly forecasting.

Significance. If the empirical gains are shown to be robust under proper cross-validation and reconstruction-error controls, the approach could reduce the cost of appliance-level monitoring while preserving forecast accuracy for time-of-use pricing applications; the modular design (filtering + grouping + collaboration) is a practical contribution to bottom-up STLF.

major comments (2)

[Critical Appliance Filtering module] Critical Appliance Filtering module (described in the abstract and §3): the claim that the selected subset reconstructs total load without significant information loss is load-bearing for all reported gains, yet no quantitative bound (e.g., MAPE or variance of the residual load after iterative decomposition) or ablation on the number of retained appliances is provided; without this, the 20–92 % improvements cannot be distinguished from favorable subset selection.
[Experimental results] Experimental section: the headline percentage improvements are stated without reporting baseline implementations, dataset sizes, cross-validation procedure, statistical significance tests, or error bars; this makes it impossible to verify that the gains are not due to post-hoc tuning or dataset-specific effects.

minor comments (2)

Notation for the three modules is introduced only descriptively; explicit pseudocode or equations for the iterative decomposition step and the collaborative fusion would improve reproducibility.
The abstract and introduction should include at least one reference to the specific datasets used (e.g., REDD, Pecan Street) and the exact forecasting horizon (hourly) to allow immediate comparison with prior work.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on the Critical Appliance Filtering module and experimental reporting. We will revise the manuscript to strengthen these aspects while preserving the core contributions of GCA-BULF.

read point-by-point responses

Referee: [Critical Appliance Filtering module] Critical Appliance Filtering module (described in the abstract and §3): the claim that the selected subset reconstructs total load without significant information loss is load-bearing for all reported gains, yet no quantitative bound (e.g., MAPE or variance of the residual load after iterative decomposition) or ablation on the number of retained appliances is provided; without this, the 20–92 % improvements cannot be distinguished from favorable subset selection.

Authors: We agree that explicit reconstruction metrics are necessary to support the claim of minimal information loss. The manuscript describes the ranking by power, frequency, and periodicity plus iterative decomposition, but does not report residual error bounds or ablations. In the revision we will add MAPE and variance of the residual load after decomposition, together with an ablation table varying the number of retained appliances, to demonstrate that performance gains remain consistent across reasonable subset sizes. revision: yes
Referee: [Experimental results] Experimental section: the headline percentage improvements are stated without reporting baseline implementations, dataset sizes, cross-validation procedure, statistical significance tests, or error bars; this makes it impossible to verify that the gains are not due to post-hoc tuning or dataset-specific effects.

Authors: We acknowledge that the current experimental section lacks sufficient detail for independent verification. The revised manuscript will specify the exact baseline implementations (including hyper-parameters), dataset sizes and splits, the time-series cross-validation procedure (rolling-origin with 24-hour horizons), statistical significance tests (paired t-test and Wilcoxon signed-rank), and error bars or standard deviations for all reported metrics on both residential and office datasets. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical framework validated on external benchmarks

full rationale

The paper proposes an algorithmic framework (Critical Appliance Filtering by power/frequency/periodicity + iterative decomposition, Related Appliance Grouping by correlations, Collaborative Load Forecasting) whose performance claims rest entirely on experimental comparisons against top-down and bottom-up baselines on residential and office datasets. No equations, uniqueness theorems, or self-citations are invoked to derive the total-load reconstruction; the reported 20-92% gains are measured outcomes, not quantities forced by the selection criteria themselves. The filtering step is a heuristic whose residual error is not bounded mathematically but is instead assessed via end-to-end forecasting accuracy, keeping the derivation chain self-contained and non-circular.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The framework rests on standard time-series forecasting assumptions plus the domain assumption that a small subset of appliances can proxy total load after grouping; no new physical entities are postulated.

axioms (1)

domain assumption A small number of critical appliances identified by power, frequency and periodicity suffice to reconstruct total load via group forecasts
Invoked in the Critical Appliance Filtering and Related Appliance Grouping modules described in the abstract

pith-pipeline@v0.9.0 · 5605 in / 1285 out tokens · 44206 ms · 2026-05-13T18:45:30.771955+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

Critical Appliance Filtering ranks appliances by power consumption, state change frequency, and usage pattern periodicity... ctrb(di) = vola(di) + α·period(di)

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

32 extracted references · 32 canonical work pages

[1]

Electric vehicle route optimization considering time-of-use electricity price by learnable partheno-genetic algorithm,

H. Yang, S. Yang, Y . Xu, E. Cao, M. Lai, and Z. Dong, “Electric vehicle route optimization considering time-of-use electricity price by learnable partheno-genetic algorithm,”IEEE Transactions on smart grid, vol. 6, no. 2, pp. 657–666, 2015

work page 2015
[2]

Robustly coordinated operation of a multi-energy microgrid with flexible electric and thermal loads,

C. Zhang, Y . Xu, Z. Li, and Z. Y . Dong, “Robustly coordinated operation of a multi-energy microgrid with flexible electric and thermal loads,” IEEE Transactions on Smart Grid, vol. 10, no. 3, pp. 2765–2775, 2018

work page 2018
[3]

Traffic processing and fingerprint generation for smart home device event,

Y . Yao, J. Hou, S. Zhang, Z. Xu, and X.-Y . Li, “Traffic processing and fingerprint generation for smart home device event,” in2022 IEEE 28th International Conference on Parallel and Distributed Systems (ICPADS). IEEE, 2023, pp. 9–16

work page 2023
[4]

Secoinfer: Secure dnn end-edge collaborative inference framework optimizing privacy and latency,

Y . Yao, J. Hou, G. Wu, Y . Cheng, M. Yuan, P. Luo, Z. Wang, and X.-Y . Li, “Secoinfer: Secure dnn end-edge collaborative inference framework optimizing privacy and latency,”ACM Transactions on Sensor Networks, vol. 20, no. 6, pp. 1–29, 2024

work page 2024
[5]

Trafficdiary: User attribute inference based on smart home traffic traces,

Y . Yao, J. Hou, M. Yuan, H. Zhang, Z. Xu, and X.-Y . Li, “Trafficdiary: User attribute inference based on smart home traffic traces,”ACM Transactions on Internet Technology, 2025

work page 2025
[6]

A cost optimized-reliable- resilient-realtime-rule-based energy management scheme for a spv-bes- based microgrid for smart building applications,

S. Chakraborty, G. Modi, and B. Singh, “A cost optimized-reliable- resilient-realtime-rule-based energy management scheme for a spv-bes- based microgrid for smart building applications,”IEEE Transactions on Smart Grid, vol. 14, no. 4, pp. 2572–2581, 2022

work page 2022
[7]

Deciding when to use a personalized model for load forecasting,

D. Qin, Q. Wen, Z. Zhou, L. Sun, and Y . Wang, “Deciding when to use a personalized model for load forecasting,”IEEE Transactions on Smart Grid, 2024

work page 2024
[8]

Improving model general- ization for short-term customer load forecasting with causal inference,

Z. Wang, H. Zhang, R. Yang, and Y . Chen, “Improving model general- ization for short-term customer load forecasting with causal inference,” IEEE Transactions on Smart Grid, 2024

work page 2024
[9]

A short-term load forecasting method using integrated cnn and lstm network,

S. H. Rafi, S. R. Deeba, E. Hossainet al., “A short-term load forecasting method using integrated cnn and lstm network,”IEEE access, vol. 9, pp. 32 436–32 448, 2021

work page 2021
[10]

A novel temporal feature selection based lstm model for electrical short- term load forecasting,

K. Ijaz, Z. Hussain, J. Ahmad, S. F. Ali, M. Adnan, and I. Khosa, “A novel temporal feature selection based lstm model for electrical short- term load forecasting,”IEEE Access, vol. 10, pp. 82 596–82 613, 2022

work page 2022
[11]

Short-term load forecasting based on lstm networks considering attention mechanism,

J. Lin, J. Ma, J. Zhu, and Y . Cui, “Short-term load forecasting based on lstm networks considering attention mechanism,”International Journal of Electrical Power & Energy Systems, vol. 137, p. 107818, 2022

work page 2022
[12]

Robust framework based on hybrid deep learning approach for short term load forecasting of building electricity demand,

C. Sekhar and R. Dahiya, “Robust framework based on hybrid deep learning approach for short term load forecasting of building electricity demand,”Energy, vol. 268, p. 126660, 2023

work page 2023
[13]

Short-term load forecasting in smart grids using hybrid deep learning,

M. M. Asiri, G. Aldehim, F. A. Alotaibi, M. M. Alnfiai, M. Assiri, and A. Mahmud, “Short-term load forecasting in smart grids using hybrid deep learning,”IEEE Access, vol. 12, pp. 23 504–23 513, 2024

work page 2024
[14]

Sac-convlstm: A novel spatio-temporal deep learning-based approach for a short term power load forecasting,

R. Jalalifar, M. R. Delavar, and S. F. Ghaderi, “Sac-convlstm: A novel spatio-temporal deep learning-based approach for a short term power load forecasting,”Expert Systems with Applications, vol. 237, p. 121487, 2024

work page 2024
[15]

Houseec: Day-ahead household electrical energy consumption forecasting using deep learning,

I. Kiprijanovska, S. Stankoski, I. Ilievski, S. Jovanovski, M. Gams, and H. Gjoreski, “Houseec: Day-ahead household electrical energy consumption forecasting using deep learning,”Energies, vol. 13, no. 10, p. 2672, 2020

work page 2020
[16]

A combined deep learning load forecasting model of single household resident user considering multi-time scale electricity consumption behavior,

W. Yang, J. Shi, S. Li, Z. Song, Z. Zhang, and Z. Chen, “A combined deep learning load forecasting model of single household resident user considering multi-time scale electricity consumption behavior,”Applied Energy, vol. 307, p. 118197, 2022

work page 2022
[17]

Hybrid multitask multi-information fusion deep learning for household short-term load forecasting,

L. Jiang, X. Wang, W. Li, L. Wang, X. Yin, and L. Jia, “Hybrid multitask multi-information fusion deep learning for household short-term load forecasting,”IEEE Transactions on Smart Grid, vol. 12, no. 6, pp. 5362– 5372, 2021

work page 2021
[18]

Short-term residential load forecasting with baseline-refinement profiles and bi- attention mechanism,

J.-W. Xiao, P. Liu, H. Fang, X.-K. Liu, and Y .-W. Wang, “Short-term residential load forecasting with baseline-refinement profiles and bi- attention mechanism,”IEEE Transactions on Smart Grid, 2023

work page 2023
[19]

Residential appliance-level load fore- casting with deep learning,

M. Razghandi and D. Turgut, “Residential appliance-level load fore- casting with deep learning,” inGLOBECOM 2020-2020 IEEE global communications conference. IEEE, 2020, pp. 1–6

work page 2020
[20]

Ap- pliance level short-term load forecasting via recurrent neural network,

Y . Zhou, A. S. Nair, D. Ganger, A. Tripathi, C. Baone, and H. Zhu, “Ap- pliance level short-term load forecasting via recurrent neural network,” in2022 IEEE Power & Energy Society General Meeting (PESGM). IEEE, 2022, pp. 1–5

work page 2022
[21]

Short- term load forecasting for smart home appliances with sequence to sequence learning,

M. Razghandi, H. Zhou, M. Erol-Kantarci, and D. Turgut, “Short- term load forecasting for smart home appliances with sequence to sequence learning,” inICC 2021-IEEE International Conference on Communications. IEEE, 2021, pp. 1–6

work page 2021
[22]

Home appliance load forecasting based on improved informer,

B. Liu, Z. Li, Z. Li, and C. Chen, “Home appliance load forecasting based on improved informer,” in2023 3rd International Conference on Intelligent Communications and Computing (ICC). IEEE, 2023, pp. 74–79

work page 2023
[23]

A bottom-up model for household load profile based on the consumption behavior of residents,

B. Gao, X. Liu, and Z. Zhu, “A bottom-up model for household load profile based on the consumption behavior of residents,”Energies, vol. 11, no. 8, p. 2112, 2018

work page 2018
[24]

A kalman filter-based bottom-up approach for household short-term load forecast,

Z. Zheng, H. Chen, and X. Luo, “A kalman filter-based bottom-up approach for household short-term load forecast,”Applied Energy, vol. 250, pp. 882–894, 2019

work page 2019
[25]

A bottom-up short-term residential load forecasting approach based on appliance characteristic analysis and multi-task learning,

S. Wang, X. Deng, H. Chen, Q. Shi, and D. Xu, “A bottom-up short-term residential load forecasting approach based on appliance characteristic analysis and multi-task learning,”Electric Power Systems Research, vol. 196, p. 107233, 2021

work page 2021
[26]

Efficient deep generative model for short-term household load forecasting using non-intrusive load monitoring,

A. Langevin, M. Cheriet, and G. Gagnon, “Efficient deep generative model for short-term household load forecasting using non-intrusive load monitoring,”Sustainable Energy, Grids and Networks, vol. 34, p. 101006, 2023

work page 2023
[27]

The uk-dale dataset, domestic appliance- level electricity demand and whole-house demand from five uk homes,

J. Kelly and W. Knottenbelt, “The uk-dale dataset, domestic appliance- level electricity demand and whole-house demand from five uk homes,” Scientific data, vol. 2, no. 1, pp. 1–14, 2015

work page 2015
[28]

A comprehensive review on deep learning approaches for short-term load forecasting,

Y . Eren and ˙I. K¨uc ¸¨ukdemiral, “A comprehensive review on deep learning approaches for short-term load forecasting,”Renewable and Sustainable Energy Reviews, vol. 189, p. 114031, 2024

work page 2024
[29]

Short-term load forecasting: A comprehensive review and simulation study with cnn-lstm hybrids approach,

K. Ullah, M. Ahsan, S. M. Hasanat, M. Haris, H. Yousaf, S. F. Raza, R. Tandon, S. Abid, and Z. Ullah, “Short-term load forecasting: A comprehensive review and simulation study with cnn-lstm hybrids approach,”IEEE Access, 2024

work page 2024
[30]

Review of family- level short-term load forecasting and its application in household energy management system,

P. Ma, S. Cui, M. Chen, S. Zhou, and K. Wang, “Review of family- level short-term load forecasting and its application in household energy management system,”Energies, vol. 16, no. 15, p. 5809, 2023

work page 2023
[31]

Short-term load forecasting models: A review of challenges, progress, and the road ahead,

S. Akhtar, S. Shahzad, A. Zaheer, H. S. Ullah, H. Kilic, R. Gono, M. Jasi ´nski, and Z. Leonowicz, “Short-term load forecasting models: A review of challenges, progress, and the road ahead,”Energies, vol. 16, no. 10, p. 4060, 2023

work page 2023
[32]

Short-term load forecasting of electricity demand for the residential sector based on modelling techniques: a systematic review,

F. Rodrigues, C. Cardeira, J. M. Calado, and R. Melicio, “Short-term load forecasting of electricity demand for the residential sector based on modelling techniques: a systematic review,”Energies, vol. 16, no. 10, p. 4098, 2023

work page 2023