Methodology for the disaggregation and forecast of demand flexibility in large consumers with the application of non-intrusive load monitoring techniques  

作　　者：Marco Toledo-Orozco C.Celi F.Guartan Arturo Peralta Carlos Alvarez-Bel D.Morales 

机构地区：[1]Institute for Energy Engineering,Universitat Politecnica de Val`encia,Camino de Vera,Valencia,46022,Spain [2]Electrical Engineering Career,Universidad Politecnica Salesiana,Sede Cuenca,010103,Ecuador [3]Electrical Engineering Career,Circular Economy Laboratory-CIITT,Universidad Catolica de Cuenca,Sede Cuenca,010107,Ecuador

出　　处：《Energy and AI》2023年第3期88-103,共16页能源与人工智能（英文）

摘　　要：Technological advances,innovation and the new industry 4.0 paradigm guide Distribution System Operators towards a competitive market that requires the articulation of flexible demand response systems.The lack of measurement and standardization systems in the industry process chain in developing countries prevents the penetration of demand management models,generating inefficiency in the analysis and processing of informa-tion to validate the flexibility potential that large consumers can contribute to the network operator.In this sense,the research uses as input variables the energy and power of the load profile provided by the utility energy meter to obtain the disaggregated forecast in quarter-hour intervals in 4-time windows validated through metrics and its results evaluated by the RMS error to get the total error generated by the methodology with the appli-cation of Machine Learning and Big Data techniques in the Python computational tool through Combinatorial Disaggregation Optimization and Factorial Hidden Markov models.

关 键 词：Big data Combinatorial optimization Factorial hidden Markov model Machine learning Non-intrusive load monitoring Time of use tariffs 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]