Electricity Theft Detection Using Deep Reinforcement Learning in Smart Power Grids
Bibliographic record
Abstract
In smart power grids, smart meters (SMs) are deployed at the end side of customers to report fine-grained power consumption readings periodically to the utility for energy management and load monitoring. However, electricity theft cyber-attacks can be launched by fraudulent customers through compromising their SMs to report false readings to pay less for their electricity usage. These attacks harmfully affect the power sector since they cause substantial financial loss and degrade the grid performance because the readings are used for energy management. Supervised machine learning approaches have been used in the literature to detect the attacks, but to the best of our knowledge, the use of reinforcement learning (RL) has not been investigated yet. RL can be better than the existing approaches because it can adapt more efficiently with the dynamic nature of cyber-attacks and consumption patterns due to its capability to learn by exploration and exploitation mechanisms and deciding optimal actions. In this article, a deep reinforcement learning (DRL) approach is proposed as a promising solution to the electricity theft problem. The samples of real dataset are employed as an environment and rewards are given based on detection errors made during training. In particular, the proposed approach is presented in four different scenarios. First, a global detection model is constructed using a deep Q network (DQN) and a double deep Q network (DDQN) with different architectures of deep neural networks. Second, the global detector is used to build a customized detection model for new customers to achieve high detection accuracy while preventing zero-day attacks. Third, changing the consumption pattern of the existing customers is taken into consideration in the third scenario. Fourth, the challenges of defending against newly launched cyber-attacks are addressed in the fourth scenario. Extensive experiments have been conducted, and the results demonstrate that the proposed DRL approach can boost the detection of electricity theft cyberattacks, and it can efficiently learn new consumption patterns, changes in the consumption patterns of existing customers, and newly launched cyber-attacks.
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 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 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".