eduzhai > Applied Sciences > Engineering >

Simulating Tariff Impact in Electrical Energy Consumption Profiles with Conditional Variational Autoencoders

  • Save

... pages left unread,continue reading

Document pages: 27 pages

Abstract: The implementation of efficient demand response (DR) programs for householdelectricity consumption would benefit from data-driven methods capable ofsimulating the impact of different tariffs schemes. This paper proposes a novelmethod based on conditional variational autoencoders (CVAE) to generate, froman electricity tariff profile combined with exogenous weather and calendarvariables, daily consumption profiles of consumers segmented in differentclusters. First, a large set of consumers is gathered into clusters accordingto their consumption behavior and price-responsiveness. The clustering methodis based on a causality model that measures the effect of a specific tariff onthe consumption level. Then, daily electrical energy consumption profiles aregenerated for each cluster with CVAE. This non-parametric approach is comparedto a semi-parametric data generator based on generalized additive models andthat uses prior knowledge of energy consumption. Experiments in a publiclyavailable data set show that, the proposed method presents comparableperformance to the semi-parametric one when it comes to generating the averagevalue of the original data. The main contribution from this new method is thecapacity to reproduce rebound and side effects in the generated consumptionprofiles. Indeed, the application of a special electricity tariff over a timewindow may also affect consumption outside this time window. Anothercontribution is that the clustering approach segments consumers according totheir daily consumption profile and elasticity to tariff changes. These tworesults combined are very relevant for an ex-ante testing of future DR policiesby system operators, retailers and energy regulators.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×