Data-Driven Demand-Side Flexibility Quantification: Prediction and Approximation of Flexibility Envelopes

TL;DR

The paper tackles the challenge of real-time, scalable quantification of residential energy flexibility by introducing a data-driven workflow that replaces iterative optimization with machine learning-based prediction of the flexibility envelope and a probabilistic PDF-based envelope approximation. The approach leverages historical results from the existing optimization framework to train multiple predictors (Linear, KNN, SVM, AdaBoost) and to approximate envelopes via 2D Normal, 2D Skewed Normal, and 3D Gaussian Mixture Models, thereby reducing online computation by up to an order of magnitude and communication load by over a factor of six. Case-study data from the NEST building demonstrate that AdaBoost provides the best predictive accuracy, while 2D-SND and 3D-GMM offer favorable trade-offs between accuracy, parameter count, and speed for envelope communication. The proposed pipeline enables real-time, large-scale deployment of flexibility management, with significant practical impact for active distribution grids and DSOs, and points to future work on improving data efficiency and scalability.

Abstract

Real-time quantification of residential building energy flexibility is needed to enable a cost-efficient operation of active distribution grids. A promising means is to use the so-called flexibility envelope concept to represent the time-dependent and inter-temporally coupled flexibility potential. However, existing optimization-based quantification entails high computational burdens limiting flexibility utilization in real-time applications, and a more computationally efficient quantification approach is desired. Additionally, the communication of a flexibility envelope to system operators in its original form is data-intensive. In order to address the computational burdens, this paper first trains several machine learning models based on historical quantification results for online use. Subsequently, probability distribution functions are proposed to approximate the flexibility envelopes with significantly fewer parameters, which can be communicated to system operators instead of the original flexibility envelope. The results show that the most promising prediction and approximation approaches allow for a minimum reduction of the computational burden by a factor of 9 and of the communication load by a factor of 6.6, respectively.

Figures (8)

• Figure 1: Comparison between the existing optimization-based workflow (in dashed lines) and the proposed workflow (in solid lines).
• Figure 2: Calculation of maximal sustainable duration by (a) first identifying energy bounds and then (b) deriving sustainable duration of chosen power levels.
• Figure 3: Example of flexibility envelope.
• Figure 4: MAE of the flexibility prediction with respect to the power level.
• Figure 5: $\textrm{R}^2$ score of the flexibility prediction with respect to the power level.