Neural network approach estimates RC thermal building model parameters for MPC

A new research paper introduces the "Estimator from Scratch" approach for estimating parameters in RC thermal building models, which are crucial for energy-efficient model predictive control (MPC). This method embeds physical equations into a neural network's training process to overcome limitations of traditional algorithms, such as local minima and high computational costs. An enhanced version, the "Pretrained Estimator," further improves accuracy and removes the need for initial guesses by pretraining on data from multiple buildings, demonstrating superior prediction performance and lower, more consistent MPC costs across various benchmarks. AI

IMPACT This research offers a more efficient and accurate method for controlling building energy systems, potentially reducing energy consumption and operational costs.

RANK_REASON Research paper published on arXiv detailing a new method for parameter estimation in thermal building models. [lever_c_demoted from research: ic=1 ai=0.7]

Read on arXiv cs.LG →

• artificial neural network
• arXiv
• Estimator from Scratch
• Fabian Raisch
• fully black-box neural network
• genetic-algorithm-based RC estimator
• model predictive control
• Pretrained Estimator
• RC models

• paper
• infra

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →