🤖 AI Summary
To address the challenges of large point prediction errors and difficulty in uncertainty quantification for small-scale aggregated loads (e.g., microgrids), this paper proposes a clustering-based residual bootstrap interval forecasting method. The approach innovatively couples unsupervised K-means clustering with residual bootstrap sampling: historical load patterns are first clustered to partition residuals dynamically; then, cluster-specific bootstrap resampling and quantile regression modeling are performed within similar operational contexts. This enhances both calibration accuracy and contextual adaptability of prediction intervals. Experiments on the real-world EULR dataset demonstrate that, compared to conventional bootstrap methods, the proposed method reduces average prediction interval width by 12.7% at 90%–99% confidence levels, while maintaining coverage deviation within ±1.5%. These results validate its effectiveness and robustness in low-aggregation, high-stochasticity scenarios.
📝 Abstract
Accurate predictions of electricity demands are necessary for managing operations in a small aggregation load setting like a Microgrid. Due to low aggregation, the electricity demands can be highly stochastic and point estimates would lead to inflated errors. Interval estimation in this scenario, would provide a range of values within which the future values might lie and helps quantify the errors around the point estimates. This paper introduces a residual bootstrap algorithm to generate interval estimates of day-ahead electricity demand. A machine learning algorithm is used to obtain the point estimates of electricity demand and respective residuals on the training set. The obtained residuals are stored in memory and the memory is further partitioned. Days with similar demand patterns are grouped in clusters using an unsupervised learning algorithm and these clusters are used to partition the memory. The point estimates for test day are used to find the closest cluster of similar days and the residuals are bootstrapped from the chosen cluster. This algorithm is evaluated on the real electricity demand data from EULR(End Use Load Research) and is compared to other bootstrapping methods for varying confidence intervals.
