Bayesian Uncertainty Quantification with Anchored Ensembles for Robust EV Power Consumption Prediction

2025-11-09

AI Summary

To address the challenge of jointly quantifying model and data uncertainties in electric vehicle energy consumption forecasting, this paper proposes an anchored ensemble LSTM framework that integrates a Student’s t-distributed output likelihood with Gaussian weight priors. Under maximum a posteriori (MAP) estimation, it enables closed-form prediction interval computation. By anchoring priors to enhance posterior diversity and leveraging the heavy-tailed Student’s t distribution for robustness against outliers, the method yields calibrated uncertainty intervals via a single forward pass—eliminating the need for Monte Carlo sampling. Experiments demonstrate strong point prediction performance (RMSE = 3.36, R² = 0.93) and precise coverage matching nominal levels (e.g., 90% predicted intervals achieve ≈90% empirical coverage). It significantly outperforms MC Dropout and quantile regression: at equivalent coverage, its prediction intervals are narrower, thus balancing theoretical rigor, predictive reliability, and real-time inference requirements.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Intelligent Robots: State Estimation
• Reasoning under Uncertainty: Uncertainty Representations

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Accurate EV power estimation underpins range prediction and energy management, yet practitioners need both point accuracy and trustworthy uncertainty. We propose an anchored-ensemble Long Short-Term Memory (LSTM) with a Student-t likelihood that jointly captures epistemic (model) and aleatoric (data) uncertainty. Anchoring imposes a Gaussian weight prior (MAP training), yielding posterior-like diversity without test-time sampling, while the t-head provides heavy-tailed robustness and closed-form prediction intervals. Using vehicle-kinematic time series (e.g., speed, motor RPM), our model attains strong accuracy: RMSE 3.36 +/- 1.10, MAE 2.21 +/- 0.89, R-squared = 0.93 +/- 0.02, explained variance 0.93 +/- 0.02, and delivers well-calibrated uncertainty bands with near-nominal coverage. Against competitive baselines (Student-t MC dropout; quantile regression with/without anchoring), our method matches or improves log-scores while producing sharper intervals at the same coverage. Crucially for real-time deployment, inference is a single deterministic pass per ensemble member (or a weight-averaged collapse), eliminating Monte Carlo latency. The result is a compact, theoretically grounded estimator that couples accuracy, calibration, and systems efficiency, enabling reliable range estimation and decision-making for production EV energy management.

Problem

• Accurately predict EV power consumption with uncertainty quantification
• Jointly capture model and data uncertainty using anchored LSTM ensembles
• Provide robust prediction intervals for real-time EV energy management

Innovation

• Anchored-ensemble LSTM with Student-t likelihood
• Gaussian weight prior for posterior-like diversity
• Single deterministic pass for efficient inference