Clustering Rooftop PV Systems via Probabilistic Embeddings
This work addresses the problem of managing rooftop PV systems for aggregators and system operators, but it is incremental as it builds on existing clustering and embedding techniques.
The paper tackled the challenge of monitoring and analyzing large, spatially distributed rooftop PV systems with high-dimensional, missing-value-affected time-series data by proposing a probabilistic entity embedding-based clustering framework. The result was that this method produced concise, uncertainty-aware cluster profiles that outperformed a physics-based baseline in representativeness and robustness, and supported reliable missing-value imputation, as demonstrated on a multi-year residential PV dataset.
As the number of rooftop photovoltaic (PV) installations increases, aggregators and system operators are required to monitor and analyze these systems, raising the challenge of integration and management of large, spatially distributed time-series data that are both high-dimensional and affected by missing values. In this work, a probabilistic entity embedding-based clustering framework is proposed to address these problems. This method encodes each PV system's characteristic power generation patterns and uncertainty as a probability distribution, then groups systems by their statistical distances and agglomerative clustering. Applied to a multi-year residential PV dataset, it produces concise, uncertainty-aware cluster profiles that outperform a physics-based baseline in representativeness and robustness, and support reliable missing-value imputation. A systematic hyperparameter study further offers practical guidance for balancing model performance and robustness.