Real-Time Price-Based Demand Response Management for Residential Appliances via Stochastic Optimization and Robust Optimization

TLDR

Residential appliance operation tasks are classified as deferrable/non‑deferrable and interruptible/non‑interruptible based on their demand‑response preferences and spatial‑temporal characteristics. The study evaluates real‑time price‑based demand‑response management for residential appliances using stochastic and robust optimization techniques. The authors embed a real‑time price‑based DR application into smart meters that automatically schedules appliances in 5‑minute slots, employing a scenario‑based Monte Carlo stochastic optimization to minimize expected daily electricity payment while managing downside risk, and a robust optimization that minimizes worst‑case payment under price uncertainty intervals, with both approaches formulated as mixed‑integer linear programs solved by state‑of‑the‑art MILP solvers. Numerical results demonstrate the attributes of the two optimization approaches for solving the real‑time optimal DR management problem for residential appliances.

Abstract

This paper evaluates the real-time price-based demand response (DR) management for residential appliances via stochastic optimization and robust optimization approaches. The proposed real-time price-based DR management application can be imbedded into smart meters and automatically executed on-line for determining the optimal operation of residential appliances within 5-minute time slots while considering uncertainties in real-time electricity prices. Operation tasks of residential appliances are categorized into deferrable/non-deferrable and interruptible/non-interruptible ones based on appliances' DR preferences as well as their distinct spatial and temporal operation characteristics. The stochastic optimization adopts the scenario-based approach via Monte Carlo (MC) simulation for minimizing the expected electricity payment for the entire day, while controlling the financial risks associated with real-time electricity price uncertainties via the expected downside risks formulation. Price uncertainty intervals are considered in the robust optimization for minimizing the worst-case electricity payment while flexibly adjusting the solution robustness. Both approaches are formulated as mixed-integer linear programming (MILP) problems and solved by state-of-the-art MILP solvers. The numerical results show attributes of the two approaches for solving the real-time optimal DR management problem for residential appliances.