[Paper Review] Architecture and Algorithms for Privacy Preserving Thermal Inertial Load Management by A Load Serving Entity

Abhishek Halder, Xinbo Geng+2 more|arXiv (Cornell University)|Jun 30, 2016

Smart Grid Energy Management12 references3 citations

TL;DR

This paper proposes a privacy-preserving architecture and algorithms for a Load Serving Entity (LSE) to manage residential air conditioners via thermal inertial load control, ensuring individual comfort while preserving user privacy. By using a contract-based model with differential privacy and setpoint velocity control, the LSE optimizes energy procurement and achieves demand response without accessing individual AC states, enabling cost-effective, scalable, and private load management.

ABSTRACT

Motivated by the growing importance of demand response in modern power system's operations, we propose an architecture and supporting algorithms for privacy preserving thermal inertial load management as a service provided by the load serving entity (LSE). We focus on an LSE managing a population of its customers' air conditioners, and propose a contractual model where the LSE guarantees quality of service to each customer in terms of keeping their indoor temperature trajectories within respective bands around the desired individual comfort temperatures. We show how the LSE can price the contracts differentiated by the flexibility embodied by the width of the specified bands. We address architectural questions of (i) how the LSE can strategize its energy procurement based on price and ambient temperature forecasts, (ii) how an LSE can close the real time control loop at the aggregate level while providing individual comfort guarantees to loads, without ever measuring the states of an air conditioner for privacy reasons. Control algorithms to enable our proposed architecture are given, and their efficacy is demonstrated on real data.

Motivation & Objective

• To design an architecture enabling a Load Serving Entity (LSE) to manage a population of residential air conditioners while ensuring individual comfort and privacy.
• To address the challenge of maintaining individual comfort guarantees without measuring or accessing the internal states of individual air conditioners.
• To enable cost-effective energy procurement by optimizing aggregate load control based on price and temperature forecasts.
• To develop a pricing mechanism for contracts that reflects the flexibility offered by customers’ comfort tolerance bands.
• To ensure differential privacy in the aggregate control process, protecting individual user data from exposure.

Proposed method

• The LSE employs a model-free setpoint velocity control algorithm that adjusts air conditioner setpoints based on real-time feedback without requiring individual state measurements.
• It uses a contract-based model where each customer specifies a comfort band around their desired temperature, and the LSE prices contracts based on the width of this band.
• Differential privacy is enforced via Laplace noise injection on the aggregate power consumption estimate, ensuring that individual AC states remain private.
• The LSE formulates an open-loop optimal control problem to minimize energy procurement costs, using day-ahead price and ambient temperature forecasts.
• A statistical performance analysis quantifies the impact of real-time deviations from day-ahead plans on procurement costs.
• The architecture enables real-time control at the aggregate level while maintaining individual comfort constraints through a privacy-preserving estimation mechanism.

Experimental results

Research questions

• RQ1 How can an LSE manage a population of thermal inertial loads to ensure individual comfort without accessing individual AC states?
• RQ2 What architectural and algorithmic framework enables privacy-preserving demand response in thermal inertial load management?
• RQ3 How can the LSE optimize energy procurement costs using price and ambient temperature forecasts?
• RQ4 What is the relationship between comfort tolerance (band width) and contract pricing in a privacy-preserving framework?
• RQ5 How does the system perform under real-time deviations from day-ahead forecasts, and what is the statistical impact on cost?

Key findings

• The proposed setpoint velocity control algorithm successfully maintains individual comfort bounds without requiring measurement of individual AC states, ensuring privacy.
• The LSE achieves ε-differential privacy in aggregate power estimation by injecting Laplace noise with scale parameter $ P_e / ε $, protecting individual data.
• The contract pricing mechanism enables revenue differentiation based on comfort tolerance, with a linear fit between cost increase and removed ACs providing a basis for dynamic pricing.
• Simulation results on ERCOT data show that the LSE can reduce energy procurement costs by strategically managing load based on forecasts and privacy-preserving control.
• The computational complexity scales efficiently, with MATLAB linprog solving planning problems for 500 homes in under 10 seconds, indicating practical deployability.
• The system demonstrates robustness to real-time deviations, with statistical performance bounds quantifying cost impacts under forecast uncertainty.