# Voltage-Based Load Recognition in Low Voltage Distribution Grids with Deep Learning

^{*}

## Abstract

**:**

## 1. Introduction

- (1)
- The proposed method can be used in an online manner to recognize large loads in the local grid environment within a time range of a few hundreds of seconds while only using the measured voltage level. Instead of observing households, the surrounding grid environment can be analyzed and grid participants from other grid connection points can be recognized.
- (2)
- A concept how to generate a dataset to train a recognition algorithm without a need for extensive simulation effort is provided.
- (3)
- For demonstration, the developed method is applied to simplified grid situations with only two EVs charging in a test grid structure, which represents a proof of concept for this method.
- (4)
- It is shown that in low voltage distribution grids the relative location of the active loads, the transformer, and the inverter influences the load recognition accuracy in a significant manner.
- (5)
- The results of the investigations in this study indicate that in comparison to multi-layer perceptrons the convolutional neural networks are the better choice to use for load recognition in time series data.

## 2. Materials and Methods

#### 2.1. General Concept

#### 2.1.1. General Task Formulation

#### 2.1.2. Example Task Formulation

#### 2.2. Algorithm Selection

#### 2.2.1. Multi-Layer Perceptron Neural Network (MLP)

#### 2.2.2. Convolutional Neural Network (CNN)

#### 2.3. Training Environment

#### 2.3.1. Phase 1—Power Profiles

#### 2.3.2. Phase 2—Voltage Profiles

**Figure 2.**Scheme of reference grid number 8 from [45] with labeled grid nodes.

#### 2.3.3. Phase 3—Training Data Generation for Different Grid Connection Points

#### 2.3.4. Phase 4—Training Data Generation for Overlapping Periods

#### 2.3.5. Phase 5—Training Phase

#### 2.4. Validation

#### 2.5. Evaluation

#### 2.6. Hyper-Parameter Optimization

## 3. Results

#### 3.1. Analysis of Recognition Accuracy inside the Reference Grid Model MONA 8

#### 3.1.1. Evaluation of Hyper-Parameter Selections

#### 3.1.2. Evaluation of Validation Results for Best CNN

#### 3.2. Analysis of Line Length Influence

#### 3.2.1. Variation of Transformer Line Length

#### 3.2.2. Variation of Measurement Point Location between Load and Transformer

## 4. Discussion

## 5. Conclusions

## Author Contributions

## Funding

## Institutional Review Board Statement

## Informed Consent Statement

## Data Availability Statement

## Acknowledgments

## Conflicts of Interest

## Abbreviations

API | Application programming interface |

CNN | Convolutional neural network |

DER | Distributed energy resource |

EV | Electric vehicle |

GMM | Gaussian mixture model |

MLP | Multi-layer perceptron |

MONA | Merit Order Netzausbau |

NAYY | Classification of a power grid cable: N—norm line; A—aluminum core; Y—insulation of the cores made of polyvinyl chloride; Y—cable sheathing made of polyvinyl chloride |

NILM | Non-intrusive load monitoring |

ReLU | Rectifier linear unit |

TPE | Tree-structured parzen estimator |

## References

- Arbeitsgruppe Erneuerbare Energien—Statistik. Erneuerbare Energien 2020; Technical Report; Bundesministerium für Wirtschaft und Energie (BMWi): Berlin, Germany, 2021. [Google Scholar]
- Federal Ministry for the Environment, Nature Conservation and Nuclear Safety (BMU). Climate Action Programme 2030: Measures to Achieve the 2030 Climate Protection Goals; Technical Report; Federal Ministry for the Environment, Nature Conservation and Nuclear Safety (BMU): Berlin, Germany, 2019. [Google Scholar]
- Deutsche Energie-Agentur GmbH; Technische Universität Dortmund. Dena-Studie Systemdienstleistungen 2030; Endbericht, Deutsche Energie-Agentur GmbH: Berlin, Germany, 2014. [Google Scholar]
- Woyte, A.; Van Thong, V.; Belmans, R.; Nijs, J. Voltage fluctuations on distribution level introduced by photovoltaic systems. IEEE Trans. Energy Convers.
**2006**, 21, 202–209. [Google Scholar] [CrossRef] - Tavakoli, A.; Saha, S.; Arif, M.T.; Haque, M.E.; Mendis, N.; Oo, A.M. Impacts of grid integration of solar PV and electric vehicle on grid stability, power quality and energy economics: A review. IET Energy Syst. Integr.
**2020**, 2, 243–260. [Google Scholar] [CrossRef] - Deutsche Kommission Elektrotechnik Elektronik Informationstechnik im DIN und VDE. DIN EN 50160: Merkmale der Spannung in öffentlichen Elektrizitätsversorgunsnetzwerken; DIN-Norm, Deutsches Institut für Normung e.V.: Berlin, Germany, 2011. [Google Scholar]
- Mahmud, N.; Zahedi, A. Review of control strategies for voltage regulation of the smart distribution network with high penetration of renewable distributed generation. Renew. Sustain. Energy Rev.
**2016**, 64, 582–595. [Google Scholar] [CrossRef] - Verband der Elektrotechnik Elektronik Informationstechnik e.V. VDE-AR-N 4105—Erzeugungsanlagen am Niederspannungsnetz; Technical Report; Deutsches Institut für Normung e.V.: Berlin, Germany, 2018. [Google Scholar]
- Li, C.; Jin, C.; Sharma, R. Coordination of PV Smart Inverters Using Deep Reinforcement Learning for Grid Voltage Regulation. In Proceedings of the 2019 18th IEEE International Conference on Machine Learning and Applications (ICMLA), Boca Raton, FL, USA, 16–19 December 2019; pp. 1930–1937. [Google Scholar] [CrossRef] [Green Version]
- Duan, J.; Shi, D.; Diao, R.; Li, H.; Wang, Z.; Zhang, B.; Bian, D.; Yi, Z. Deep-Reinforcement-Learning-Based Autonomous Voltage Control for Power Grid Operations. IEEE Trans. Power Syst.
**2020**, 35, 814–817. [Google Scholar] [CrossRef] - Beyer, K.; Beckmann, R.; Geißendörfer, S.; von Maydell, K.; Agert, C. Adaptive Online-Learning Volt-Var Control for Smart Inverters Using Deep Reinforcement Learning. Energies
**2021**, 14, 1991. [Google Scholar] [CrossRef] - Liu, H.; Wu, W. Online Multi-Agent Reinforcement Learning for Decentralized Inverter-Based Volt-VAR Control. IEEE Trans. Smart Grid
**2021**, 12, 2980–2990. [Google Scholar] [CrossRef] - Yang, Q.; Wang, G.; Sadeghi, A.; Giannakis, G.B.; Sun, J. Two-Timescale Voltage Control in Distribution Grids Using Deep Reinforcement Learning. IEEE Trans. Smart Grid
**2020**, 11, 2313–2323. [Google Scholar] [CrossRef] [Green Version] - Hart, G. Nonintrusive appliance load monitoring. Proc. IEEE
**1992**, 80, 1870–1891. [Google Scholar] [CrossRef] - Aladesanmi, E.J.; Folly, K.A. Overview of non-intrusive load monitoring and identification techniques. IFAC-PapersOnLine
**2015**, 48, 415–420. [Google Scholar] [CrossRef] - Ghosh, S.; Chatterjee, A.; Chatterjee, D. Load monitoring of residential elecrical loads based on switching transient analysis. In Proceedings of the 2017 IEEE Calcutta Conference (CALCON), Kolkata, India, 2–3 December 2017; pp. 428–432. [Google Scholar] [CrossRef]
- Athanasiadis, C.; Doukas, D.; Papadopoulos, T.; Chrysopoulos, A. A Scalable Real-Time Non-Intrusive Load Monitoring System for the Estimation of Household Appliance Power Consumption. Energies
**2021**, 14, 767. [Google Scholar] [CrossRef] - Faustine, A.; Mvungi, N.H.; Kaijage, S.; Michael, K. A Survey on Non-Intrusive Load Monitoring Methodies and Techniques for Energy Disaggregation Problem. arXiv
**2017**, arXiv:1703.00785. [Google Scholar] - Huber, P.; Calatroni, A.; Rumsch, A.; Paice, A. Review on Deep Neural Networks Applied to Low-Frequency NILM. Energies
**2021**, 14, 2390. [Google Scholar] [CrossRef] - Bernard, T. Non-Intrusive Load Monitoring (NILM): Combining Multiple Distinct Electrical Features and Unsupervised Machine Learning Techniques. Ph.D. Thesis, Universität Duisburg-Essen, Duisburg, Germany, 2018. [Google Scholar]
- Brucke, K.; Arens, S.; Telle, J.S.; Steens, T.; Hanke, B.; von Maydell, K.; Agert, C. A Non-Intrusive Load Monitoring Approach for Very Short Term Power Predictions in Commercial Buildings. arXiv
**2020**, arXiv:2007.11819. [Google Scholar] [CrossRef] - Parson, O.; Ghosh, S.; Weal, M.; Rogers, A. An Unsupervised Training Method for Non-Intrusive Appliance Load Monitoring. Artif. Intell.
**2014**, 217, 1–19. [Google Scholar] [CrossRef] - Zufferey, D.; Gisler, C.; Khaled, O.A.; Hennebert, J. Machine learning approaches for electric appliance classification. In Proceedings of the 2012 11th International Conference on Information Science, Signal Processing and their Applications (ISSPA), Montreal, QC, Canada, 2–5 July 2012; pp. 740–745. [Google Scholar] [CrossRef]
- de Souza, W.A.; Garcia, F.D.; Marafão, F.P.; da Silva, L.C.P.; Simões, M.G. Load Disaggregation Using Microscopic Power Features and Pattern Recognition. Energies
**2019**, 12, 2641. [Google Scholar] [CrossRef] [Green Version] - Basu, K.; Debusschere, V.; Bacha, S.; Maulik, U.; Bondyopadhyay, S. Nonintrusive Load Monitoring: A Temporal Multilabel Classification Approach. IEEE Trans. Ind. Inform.
**2015**, 11, 262–270. [Google Scholar] [CrossRef] - Singh, S.; Majumdar, A. Non-Intrusive Load Monitoring via Multi-Label Sparse Representation-Based Classification. IEEE Trans. Smart Grid
**2020**, 11, 1799–1801. [Google Scholar] [CrossRef] [Green Version] - Ruzzelli, A.G.; Nicolas, C.; Schoofs, A.; O’Hare, G.M.P. Real-Time Recognition and Profiling of Appliances through a Single Electricity Sensor. In Proceedings of the 2010 7th Annual IEEE Communications Society Conference on Sensor, Mesh and Ad Hoc Communications and Networks (SECON), Boston, MA, USA, 21–25 June 2010; pp. 1–9. [Google Scholar] [CrossRef] [Green Version]
- Dharmakeerthi, C.H.; Mithulananthan, N.; Saha, T.K. Impact of electric vehicle fast charging on power system voltage stability. Int. J. Electr. Power Energy Syst.
**2014**, 57, 241–249. [Google Scholar] [CrossRef] - Krystalakos, O.; Nalmpantis, C.; Vrakas, D. Sliding Window Approach for Online Energy Disaggregation Using Artificial Neural Networks. In Proceedings of the 10th Hellenic Conference on Artificial Intelligence, Patras, Greece, 9–12 July 2018; Association for Computing Machinery: New York, NY, USA; pp. 1–6. [Google Scholar] [CrossRef]
- Wang, J.; Yang, Y.; Mao, J.; Huang, Z.; Huang, C.; Xu, W. CNN-RNN: A Unified Framework for Multi-label Image Classification. In Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, USA, 27–30 June 2016; pp. 2285–2294. [Google Scholar] [CrossRef] [Green Version]
- Li, Y.; Wang, Y. A Multi-label Image Classification Algorithm Based on Attention Model. In Proceedings of the 2018 IEEE/ACIS 17th International Conference on Computer and Information Science (ICIS), Singapore, 6–8 June 2018; pp. 728–731. [Google Scholar] [CrossRef]
- Song, L.; Liu, J.; Qian, B.; Sun, M.; Yang, K.; Sun, M.; Abbas, S. A Deep Multi-Modal CNN for Multi-Instance Multi-Label Image Classification. IEEE Trans. Image Process.
**2018**, 27, 6025–6038. [Google Scholar] [CrossRef] - Massidda, L.; Marrocu, M.; Manca, S. Non-Intrusive Load Disaggregation by Convolutional Neural Network and Multilabel Classification. Appl. Sci.
**2020**, 10, 1454. [Google Scholar] [CrossRef] [Green Version] - Krizhevsky, A.; Sutskever, I.; Hinton, G.E. ImageNet classification with deep convolutional neural networks. Commun. ACM
**2017**, 60, 84–90. [Google Scholar] [CrossRef] - He, K.; Zhang, X.; Ren, S.; Sun, J. Deep Residual Learning for Image Recognition. In Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, USA, 27–30 June 2016; pp. 770–778. [Google Scholar] [CrossRef] [Green Version]
- Ashiquzzaman, A.; Tushar, A.K. Handwritten Arabic numeral recognition using deep learning neural networks. In Proceedings of the 2017 IEEE International Conference on Imaging, Vision Pattern Recognition (icIVPR), Dhaka, Bangladesh, 13–14 February 2017; pp. 1–4. [Google Scholar] [CrossRef] [Green Version]
- Schroff, F.; Kalenichenko, D.; Philbin, J. FaceNet: A unified embedding for face recognition and clustering. In Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Boston, MA, USA, 7–12 June 2015; pp. 815–823. [Google Scholar] [CrossRef] [Green Version]
- Zeiler, M.; Ranzato, M.; Monga, R.; Mao, M.; Yang, K.; Le, Q.; Nguyen, P.; Senior, A.; Vanhoucke, V.; Dean, J.; et al. On rectified linear units for speech processing. In Proceedings of the 2013 IEEE International Conference on Acoustics, Speech and Signal Processing, Vancouver, BC, Canada, 26–31 May 2013; pp. 3517–3521. [Google Scholar] [CrossRef] [Green Version]
- Qian, Y.; Bi, M.; Tan, T.; Yu, K. Very Deep Convolutional Neural Networks for Noise Robust Speech Recognition. IEEE/ACM Trans. Audio Speech Lang. Process.
**2016**, 24, 2263–2276. [Google Scholar] [CrossRef] - Silver, D.; Huang, A.; Maddison, C.J.; Guez, A.; Sifre, L.; van den Driessche, G.; Schrittwieser, J.; Antonoglou, I.; Panneershelvam, V.; Lanctot, M.; et al. Mastering the game of Go with deep neural networks and tree search. Nature
**2016**, 529, 484–489. [Google Scholar] [CrossRef] - Fawaz, H.I.; Forestier, G.; Weber, J.; Idoumghar, L.; Muller, P.A. Deep learning for time series classification: A review. Data Min. Knowl. Discov.
**2019**, 33, 917–963. [Google Scholar] [CrossRef] [Green Version] - Dzieia, M.; Hübscher, H.; Jagla, D.; Klaue, J.; Petersen, H.J.; Wickert, H. Elektronik Tabellen: Betriebs- und Automatisierungstechnik, 2nd ed.; Westermann: Braunschweig, Germany, 2016. [Google Scholar]
- Chollet, F.; Gibson, A.; Allaire, J.J.; Rahman, F.; Branchaud-Charron, F.; Lee, T.; de Marmiesse, G.; Jin, H.; Watson, M.; Zhu, S. Keras. 2015. Available online: https://keras.io (accessed on 14 November 2021).
- Janitza Electronics. Power Quality Analyser UMG 604-PRO. 2019. Available online: https://www.janitza.com/us/datasheets.html (accessed on 14 November 2021).
- Projekt MONA 2030: Grundlage für die Bewertung von Netzoptimierenden Maßnahmen: Teilbericht Basisdaten; Technical Report; FfE Forschungsstelle für Energiewirtschaft e.V.: München, Germany, 2017.
- The MathWorks, Inc. Simscape Documentation. 2021. Available online: https://de.mathworks.com/help/physmod/simscape/index.html (accessed on 14 November 2021).
- The MathWorks, Inc. Simscape Electrical Documentation. 2021. Available online: https://de.mathworks.com/help/physmod/sps/index.html (accessed on 14 November 2021).
- Abadi, M.; Agarwal, A.; Barham, P.; Brevdo, E.; Chen, Z.; Citro, C.; Corrado, G.S.; Davis, A.; Dean, J.; Devin, M.; et al. TensorFlow: Large-Scale Machine Learning on Heterogeneous Systems. 2015. Available online: tensorflow.org (accessed on 14 November 2021).
- Akiba, T.; Sano, S.; Yanase, T.; Ohta, T.; Koyama, M. Optuna: A Next-generation Hyperparameter Optimization Framework. In Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, KDD ’19, Anchorage, AK, USA, 4–8 August 2019; Association for Computing Machinery: New York, NY, USA, 2019; pp. 2623–2631. [Google Scholar] [CrossRef]
- Duchi, J.; Hazan, E.; Singer, Y. Adaptive Subgradient Methods for Online Learning and Stochastic Optimization. J. Mach. Learn. Res.
**2011**, 12, 2121–2359. [Google Scholar] - Zeiler, M.D. ADADELTA: An Adaptive Learning Rate Method. arXiv
**2012**, arXiv:1212.5701. [Google Scholar] - Kingma, D.P.; Ba, J. Adam: A Method for Stochastic Optimization. arXiv
**2017**, arXiv:1412.6980. [Google Scholar] - Luo, Y.; Yin, L.; Bai, W.; Mao, K. An Appraisal of Incremental Learning Methods. Entropy
**2020**, 22, 1190. [Google Scholar] [CrossRef] - Siddiqui, Z.A.; Park, U. Progressive Convolutional Neural Network for Incremental Learning. Electronics
**2021**, 10, 1879. [Google Scholar] [CrossRef] - Sarwar, S.; Ankit, A.; Roy, K. Incremental Learning in Deep Convolutional Neural Networks Using Partial Network Sharing. IEEE Access
**2017**, 8, 4615–4628. [Google Scholar] [CrossRef]

**Figure 1.**Active power curves during a charging process of both electric vehicles used in this paper in 1 s-resolution.

**Figure 4.**Comparison of the MLP- and the CNN-based hyper-parameter optimization concentrating on specific hyper-parameters. Every cross represents a single trial. The crosses are drawn partially transparent, such that points which occurred more than once during optimization appear darker. In (

**a**,

**b**) the advantage of the optimizer choices of Adagrad and Adadelta is shown. In (

**c**,

**d**) the impact of the number of layers on the objective value is presented. In (

**e**,

**f**) the influence of the window length parameter on the objective values is shown.

**Figure 5.**Comparison of hyper-parameter optimizations with focus on window length. Every cross represents a single trial. The crosses are drawn partially transparent, such that points which occurred more than once during optimization appear darker. In (

**a**) the impact of the window length on the MLP-based optimization is shown, while in (

**b**) the less influence of this parameter for the CNN-based optimization is presented.

**Figure 6.**Comparison of MLP and CNN optimization regarding the history of the objective values for 100 trials. In (

**a**) the less successful MLP optimization history is shown, while in (

**b**) the objective values obtained in the CNN optimization are presented.

**Figure 7.**Recognition accuracies for different validation scenarios achieved by the best CNN found in hyper-parameter optimization. A scenario containing a single load activation of electric vehicle i at grid node $Nx$ is named as $Nx$-i. A scenario containing an overlapping activation with electric vehicle i at node $Nx$ and electric vehicle j at node $Ny$ is named as $NxNy$-$ij$. Within these scenarios, the voltage was measured at grid node $N10$.

**Figure 8.**Recognition accuracies achieved by the best CNN found in hyper-parameter optimization in two activation scenarios at grid node $N1$, which is the farthest node to the measurement point at $N10$. The first scenario with relatively inaccurate classification of the first electric vehicle’s status contains an activation of electric vehicle 1 at node $N1$ and no activation of electric vehicle 2 ($N1-1$), and the second one vice versa ($N1-2$).

**Figure 10.**Relation between the line length from the measurement point/load to the transformer and the recognition accuracy using the best CNN and the best MLP found during the hyper-parameter optimization. Each panel shows the results of one neural network based on a scenario containing an activation of one electric vehicle (EV), while the other one was inactive. (

**a**) MLP, data: active EV${}_{1}$ and inactive EV${}_{2}$. (

**b**) CNN, data: active EV${}_{1}$ and inactive EV${}_{2}$. (

**c**) MLP, data: inactive EV${}_{1}$ and active EV${}_{2}$. (

**d**) CNN, data: inactive EV${}_{1}$ and active EV${}_{2}$.

**Figure 11.**Relation between line length from measurement point to the transformer and to the load bus and the recognition accuracy using the best CNN and the best MLP found during the hyper-parameter optimization. Each panel shows decreasing accuracies with increasing line length achieved by either the MLP or the CNN. The results in (

**a**–

**d**) are based on a dataset containing an activation of one electric vehicle (EV), while the other one was inactive. In (

**e**–

**f**) an overlapping situation is tested. (

**a**) MLP, data: active EV${}_{1}$ and inactive EV${}_{2}$. (

**b**) CNN, data: active EV${}_{1}$ and inactive EV${}_{2}$. (

**c**) MLP, data: inactive EV${}_{1}$ and active EV${}_{2}$. (

**d**) CNN, data: inactive EV${}_{1}$ and active EV${}_{2}$. (

**e**) MLP, data: active EV${}_{1}$ and active EV${}_{2}$. (

**f**) CNN, data: active EV${}_{1}$ and active EV${}_{2}$.

Step | Class 1 | Class 2 | Training |
---|---|---|---|

1 | 1 | 0 | Scaling factors: |

0.92, 0.94, 0.96, 0.98, 1.0 | |||

31,268 samples ≈ 13.0% | |||

2 | 0 | 1 | Scaling factors: |

0.92, 0.94, 0.96, 0.98, 1.0 | |||

174,173 samples ≈ 72.6% | |||

3 | 1 | 1 | Number of shifts with fixed EV${}_{1}$: 5 |

Number of shifts with fixed EV${}_{2}$: 5 | |||

19,507 samples ≈ 8.1% | |||

4 | 0 | 0 | In final dataset: |

15,000 samples ≈ 6.3% |

Parameter | Description |
---|---|

Model type | The type of the neural network architecture (MLP or CNN). |

Window length | The number of historical data points given into the neural network as one sample. |

Optimizer | The optimization algorithm used for training of the neural network. |

Batch size | The number of samples used for one update of the network’s weights. |

Learning rate | The step-size of the weights’ updates. |

Number of layers | The number of connected layers of the neural network (only dense or convolutional layers counted if applicable). |

Number of neurons | The number of neurons in a dense layer of an MLP. |

Filters | The number of filters used in a convolutional layers (same for all layers). |

Kernel sizes | The size of the filters used in the convolutional layers. |

Parameter Model Type | MLP-Optimization MLP | CNN-Optimization CNN |
---|---|---|

Window length | 10 s–190 s, step-size 20 | 10 s–190 s, step-size 20 |

Optimizer | Adam, Adadelta, Adagrad | Adam, Adadelta, Adagrad [50,51,52] |

Batch size | 64–768, step-size 64 | 64–768, step-size 64 |

Learning rate | Log-uniformly distributed | Log-uniformly distributed |

in $[0.01,0.1]$ | in $[0.01,0.1]$ | |

Number of layers | 1–3 (dense layers) | 1–3 (convolutional layers) |

Number of neurons | 16–128 per layer | - |

Filters | - | 64–128, step-size 32 |

Kernel sizes | - | 3–9, step-size 2 |

Parameter | MLP | CNN |
---|---|---|

Window length | 150 | 90 |

Batch size | 320 | 64 |

Learning rate | ≈0.034 | ≈0.019 |

Number of layers | 3 dense layers | 3 convolutional layers |

Number of neurons | $[64,56,40]$ | - |

Filters | - | $[96,96,96]$ |

Kernel sizes | - | $[5,7,5]$ |

Optimizer | Adadelta | Adagrad |

Optimal value | 2.88 | 2.98 |

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |

© 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Schlachter, H.; Geißendörfer, S.; von Maydell, K.; Agert, C.
Voltage-Based Load Recognition in Low Voltage Distribution Grids with Deep Learning. *Energies* **2022**, *15*, 104.
https://doi.org/10.3390/en15010104

**AMA Style**

Schlachter H, Geißendörfer S, von Maydell K, Agert C.
Voltage-Based Load Recognition in Low Voltage Distribution Grids with Deep Learning. *Energies*. 2022; 15(1):104.
https://doi.org/10.3390/en15010104

**Chicago/Turabian Style**

Schlachter, Henning, Stefan Geißendörfer, Karsten von Maydell, and Carsten Agert.
2022. "Voltage-Based Load Recognition in Low Voltage Distribution Grids with Deep Learning" *Energies* 15, no. 1: 104.
https://doi.org/10.3390/en15010104