Scheduling of home appliances based on adaptive user optimization and diverse forecasting models.
Author
Rodriguez, Roberto Daniel
Term
4. term
Education
Publication year
2021
Submitted on
2021-05-17
Pages
52
Abstract
In the United States, homes and businesses use about 73% of all electricity. As smart grids expand, ideas like Demand Side Management (DSM), Advanced Metering (AM), Demand Response (DR), and improved scheduling and forecasting are becoming more common. Renewable sources such as wind turbines and solar panels (photovoltaics) are variable and hard to predict, which can create mismatches between supply and demand. One way to reduce these imbalances is to shift when electricity is used. At the household level, this requires flexibility – being able to run some devices earlier or later without losing comfort. This project identifies household loads that can be shifted and represents them as flexible consumption descriptions (flex-offers). We focus on wet appliances — washing machines and dishwashers — because they can change their operating times to better match renewable production and account for about 30% of household electricity use. In DSM, pricing mechanisms encourage people to move their use to cheaper times; for example, time-of-use tariffs set different prices across the day, nudging demand to off-peak hours. In this context, we apply machine learning models to schedule appliance operation so that household demand better aligns with prices and renewable generation.
I USA står bolig- og erhvervsbyggeri for cirka 73% af elforbruget. Når de intelligente elnet (Smart Grid) udbredes, bliver begreber som styring af efterspørgslen (DSM), avanceret måling (AM), efterspørgselsrespons (DR) samt planlægning og prognoser mere udbredte. Vedvarende energikilder som vindmøller og solceller (fotovoltaik) er variable og svære at forudsige, hvilket kan skabe ubalance mellem produktion og forbrug. En måde at mindske disse ubalancer på er at flytte, hvornår strømmen bruges. I husholdninger kræver det fleksibilitet – at nogle apparater kan køre tidligere eller senere uden at gå på kompromis med komforten. Dette projekt udpeger de husholdningsbelastninger, der kan flyttes, og beskriver dem som fleksible forbrugsbeskrivelser (flex-offers). Vi fokuserer på våde apparater – vaskemaskine og opvaskemaskine – fordi de kan ændre kørselstidspunkt for bedre at passe til produktionen fra vedvarende energi og udgør omkring 30% af husholdningens elforbrug. Inden for DSM bruges prissignaler til at få forbrugere til at ændre adfærd; for eksempel sætter tidsdifferentierede elpriser forskellige takster hen over døgnet og skubber forbruget mod lavlastperioder. I denne sammenhæng anvender vi maskinlæringsmodeller til at planlægge apparaters drift, så husholdningens forbrug bedre matcher priser og produktion fra vedvarende energi.
[This apstract has been rewritten with the help of AI based on the project's original abstract]
Keywords