Behaviour-Conditioned Neural Processes for Adaptive Residential Short-Term Load Forecasting

2026-07-17

AI Summary

This study addresses the challenges posed by the heterogeneity and time-varying nature of residential electricity consumption behavior in short-term load forecasting. To this end, the authors propose a behavior-conditioned Attentive Neural Process (ANP) model that jointly models discrete behavioral categories and continuous functional uncertainty within the neural process framework. By leveraging clustering to generate weak supervision signals, the method enables context-adaptive prediction without requiring ground-truth behavioral labels. Experimental results on the SGSC dataset demonstrate that the proposed model achieves average reductions of 7.9% in MAE and 6.9% in CRPS compared to standard ANP and fixed-window baselines, with particularly pronounced improvements under limited context conditions.

Abstract

Residential short-term load forecasting (STLF) is challenging because household demand is heterogeneous, temporally variable, and shaped by diverse behavioural routines. This work investigates whether inferred behavioural structure can be embedded within the forecasting mechanism of a Neural Process-based probabilistic model, rather than used only as an external grouping signal, for context-conditioned residential STLF. We propose a behaviour-conditioned Attentive Neural Process framework that treats each load profile as a forecasting task. Behavioural structure is represented by a discrete latent variable inferred from the available context and used for behaviour-conditioned decoder conditioning, while a continuous latent variable captures shared functional uncertainty across heterogeneous profiles. To enable conditioning without ground-truth behavioural labels, clustering-derived information provides weak supervision during training, whereas test-time conditioning relies only on context-inferred class distributions. Experiments on the Smart Grid, Smart City (SGSC) dataset use user-disjoint train/validation/test splits, variable context lengths, and multi-step forecast horizons, with comparisons against a label-agnostic ANP baseline and fixed-window deterministic STLF baselines. The proposed variants improve MAE and CRPS over ANP across horizons and context settings, with the largest gains under limited context. The best-performing variant achieves average reductions of 7.9% in MAE and 6.9% in CRPS relative to ANP. Compared with fixed-window baselines, this variant achieves lower RMSE across all evaluated horizons while maintaining competitive MAE, suggesting fewer large prediction deviations under heterogeneous consumption patterns. These results support single-model, uncertainty-aware forecasting across heterogeneous households, contexts, and horizons.

Problem

• residential short-term load forecasting
• behavioural routines
• heterogeneous demand
• context-conditioned forecasting
• probabilistic forecasting

Innovation

• Behaviour-Conditioned Neural Processes
• Short-Term Load Forecasting
• Probabilistic Forecasting
• Latent Variable Modeling
• Weak Supervision