Unlocking the Full Potential of Neural NILM: On Automation, Hyperparameters & Modular Pipelines

Non-Intrusive Load Monitoring (NILM) is a technique used to monitor the energy usage of individual appliances and devices in a home or building, without the need to physically measure each appliance or device. This allows energy managers to more accurately understand how energy is being used in the building. The basic principle behind NILM is to measure the overall energy usage of the building, and then identify patterns in the usage that can be attributed to specific appliances or devices. By analyzing the total energy usage, NILM can identify the type of appliance and its energy consumption. This information can then be used to make informed decisions about energy management, such as identifying energy-efficient appliances and optimizing energy usage. NILM is important for energy management applications because it provides a more comprehensive view of energy use. By understanding the energy usage of individual devices, energy managers can make better decisions about how to optimize energy usage and reduce energy costs. Furthermore, NILM can identify potential problems in the system, such as inefficient appliances, which can be addressed in order to improve efficiency.

Overview of the NILM pipeline in Deep-NILMTK

In recent years, Non-Intrusive Load Monitoring (NILM) has become an important tool for identifying the power consumption of individual appliances from a single metering point. Deep learning models are gaining traction in this area, however, there are still many challenges surrounding NILM datasets and the lack of common experimental guidelines. This lack of features and best practices guidelines has limited the adoption of efficient research instruments and made it difficult to compare, replicate, and share results.

To address this problem, we have proposed a novel open-source toolkit, Deep-NILMTK, which leverages the best practices for Deep Learning and offers a common testing bed for NILM algorithms. This toolkit includes a modular NILM pipeline that can be easily customised and introduces the concept of Experiment Templating to improve research efficiency. To demonstrate the effectiveness of the tool, we have created an online NILM benchmark repository and conducted a case-study with eight of the most popular deep NILM algorithms. All sources for the tool are available on Github, along with the accompanying documentation.

Leveraging this concept and DL best practices, a case-study of creating an online NILM benchmark repository is provided at https://github.com/BHafsa/DNN-NILM-benchmark considering eight of the most popular deep NILM algorithms. All sources relative to the tool are publicly available on Github https://github.com/BHafsa/deep-nilmtk-v1 along with the corresponding documentation.

Further information can be found in the paper

Hafsa Bousbiat, Anthony Faustine, Christoph Klemenjak, Lucas Pereira, and Wilfried Elmenreich. Unlocking the full potential of neural NILM: On automation, hyperparameters & modular pipelines. IEEE Transactions on Industrial Informatics, pages 1–9, 9 2022. (doi:10.1109/TII.2022.3206322)

 

Adaptive Weighted Recurrence Graph blocks for event-based NILM

To this day, hyperparameter tuning remains a cumbersome task in Non-Intrusive Load Monitoring (NILM) research, as researchers and practitioners are forced to invest a considerable amount of time in this task.

This paper proposes adaptive weighted recurrence graph blocks (AWRG) for appliance feature representation in event-based NILM. An AWRG block can be combined with traditional deep neural network architectures such as Convolutional Neural Networks for appliance recognition. Our approach transforms one cycle per activation current into a weighted recurrence graph and treats the associated hyper-parameters as learn-able parameters.

We evaluate our technique on two energy datasets, the industrial dataset LILACD and the residential PLAID dataset. The outcome of our experiments shows that transforming current waveforms into weighted recurrence graphs provides a better feature representation and thus, improved classification results. It is concluded that our approach can guarantee uniqueness of appliance features, leading to enhanced generalisation abilities when compared to the widely researched V-I image features. Furthermore, we show that the initialisation parameters of the AWRG’s have a significant impact on the performance and training convergence.

We provide the implementation of AWRG on Github.  If you find this tool useful and use it (or parts of it), we ask you to cite the following work in your publications:

A. Faustine, L. Pereira and C. Klemenjak, "Adaptive Weighted Recurrence Graphs for Appliance Recognition in Non-Intrusive Load Monitoring," in IEEE Transactions on Smart Grid, doi: 10.1109/TSG.2020.3010621.

Learn more about the authors Anthony Faustine, Lucas Pereira and Christoph Klemenjak.

Modeling Solar Radiation

In our tutorial at the 2014 IEEE Innovative Smart Grid Technologies - Asia conference, Tamer Khatib and I covered the application of machine learning techniques for energy applications, in particular for modeling solar radiation. In the first part we explored meta-heuristic search algorithms and envisioned their application for designing distributed, self-organizing control systems using evolutionary algorithms. We provide an open-source software tool, FREVO, to conveniently apply this approach of finding the proper configuration of a local agent.

In the second part, we targeted the problem of solar radiation modeling. After stepping through different classical modeling approaches, we presented the possibility of using artificial neural networks to learn the correlation of input parameters such as latitude, longitude, temperature, humidity, month, day, hour to predict global and diffuse solar radiation. Experiments show that this method can achieve a high accuracy compared to existing models.

Links:

• Download FREVO (open source, platform-independent)
• Video: 6 minute introduction to FREVO
• A. Sobe, I. Fehérvári, and W. Elmenreich. FREVO: A tool for evolving and evaluating self-organizing systems. In Proceedings of the 1st International Workshop on Evaluation for Self-Adaptive and Self-Organizing Systems, Lyon, France, September 2012.
• I. Fehervari and W. Elmenreich. Evolution as a tool to design self-organizing systems. In Self-Organizing Systems, volume LNCS 8221, pages 139–144. Springer Verlag, 2014.
• T. Khatib, A Mohamed, K Sopian. A review of solar energy modeling techniques. J. of Renewable & Sustainable Energy Reviews. 2012.16(5): 2864-2869.
• T. Khatib, A. Mohamed, K. Sopian, M. Mahmoud. Assessment of Artificial Neural Networks for Hourly Solar Radiation Prediction. J. of Photoenergy. 2012. 2012(ID 946890):1-7.
• IEEE Innovative Smart Grid Technologies - Asia conference site