Multi-Agent Reinforcement Learning Based User-Centric Demand Response via Non-Intrusive Load Monitoring
Bibliographic record
Abstract
In contrast to generation of electricity following demand of electricity, energy demand management at consumer end is termed as Demand Side Management (DSM).DSM encourages energy savings on the demand side by modification of the energy usage pattern.The reduction or shifting of electricity usage during high cost periods for financial incentives is called Demand response (DR).Benefits of DR policies are in terms of energy optimization, grid optimization, power system stability have been researched in detail.But the discomfort that it involves at customer end to change their consumption pattern is generally neglected.The work in this thesis proposes, day ahead user-centric demand response framework for residential house energy management based on multi-agent reinforcement learning to achieve user-centric demand response which also capitalizes on energy efficiency and financial incentives.An appropriate finite Markov decision process (FMDP) with discrete time steps is used to describe the hour-ahead energy consumption scheduling problem under consideration.For this, predicting the next day appliance level load demand, machine learning and non-intrusive load monitoring is used.The previous day is studied to forecast load demand and analysis of operation time of appliances to which optimal appliance scheduling is shared to the consumer.The disaggregation of total demand into appliance level is done by non-intrusive load monitoring (NILM).NILM is a technique for electrical power disaggregation that is affordable and only requires one point of measurement, as is the case with smart meters.The Python-based NILM Toolkit is used iii for case study and it provides various modern methods for disaggregation.Then, each appliance is categorized based on the user feedback into time-shiftable, non-shiftable, power-shiftable appliance.After integrating this appliance feedback, a cost effective algorithm based on the predicted time of operation of appliances, and scheduling of appliances is achieved by multi-agent reinforcement learning.This takes into account the comfort of users while reducing the cost of energy use.This framework is implemented on REFIT dataset which contains data for 2 years of 20 houses in the UK.The dataset contains meter reading both at appliance level and house level.The selected house is house 2 which has 9 appliances which are categorized and the proposed technique is implemented.The results demonstrate the effectiveness of the framework as it reduced energy costs by 15 percent by shifting appliance operation time intervals to off-peak period, while integrating the benefit of both the consumer and the electricity supplier.The results also reveal the reduction in peak demand which is beneficial to the power system stability.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".