TY - CHAP A1 - Yadav, Akhilesh A1 - Sinha, Anuj A1 - Saidi, Abdessamad A1 - Trinkl, Christoph A1 - Zörner, Wilfried T1 - NILM based energy disaggregation algorithm for dairy farms T2 - NILM'20: Proceedings of the 5th International Workshop on Non-Intrusive Load Monitoring UR - https://doi.org/10.1145/3427771.3427846 KW - NILM KW - energy disaggregation KW - smart meter KW - dairy farms KW - deep neural networks Y1 - 2020 UR - https://doi.org/10.1145/3427771.3427846 SN - 978-1-4503-8191-8 SP - 16 EP - 19 PB - ACM CY - New York ER - TY - THES A1 - Sinha, Anuj T1 - Multi-feature based development of a power-disaggregation algorithm for dairy farms N2 - Smart meter technology implementation in the last decade had initiated many data collection processes, which have provided a strong foundation for the development of Artificial Intelligence (AI) based load monitoring systems. It is easier to identify the energy-saving potential with the help of advanced load monitoring systems. Since 2015, deep-learning-based Nonintrusive load monitoring (NILM) is being focused in the research community. It requires minimal hardware, which can justify its development and maintenance cost. Several AI-based models and tools are available for load monitoring, but it is challenging to identify a suitable model for the specific application. There is still a domain-specific transformation, and considerations are usually required. The residential sector has been the focus area due to the market size, but the industrial sector still has massive potential for research and development. Thus, in the presented thesis, dairy farms in Germany are targeted for developing a power disaggregation algorithm based on deep learning, which can identify the on/off state of individual appliances in the farm from the aggregated load profile data. Mainly four appliances named milk cooling (MK), milk pump (MP), vacuum pump (VP), and cleaning automatic machine (SA) are targeted for disaggregation. NILM is a promising approach to identify individual operating times of appliances. Thus, deep neural networkbased algorithms are developed, focusing mainly on one-dimensional convolution neural network (1D-CNN) and recurrent neural network (RNN). Literature research was carried out to determine the state-of-the-art of deep-learningbased NILM and understand AI technology. Data acquisition for model development and testing was made from four dairy farms based out of Bavaria, Germany. The presented work provides a detailed discussion about data pre-processing and development of models. The result shows that deep-learning-based disaggregation algorithms outperform for this application area, and the proposed model successfully identifies the states of individual appliances. The presented work provides a foundation for modifying the proposed algorithm or developing a new algorithm for real-time power disaggregation. Y1 - 2021 UR - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:573-8568 CY - Ingolstadt ER -