An Intelligent Water Temperature Control System Using Deep Q-Learning for Personalized Comfort and Energy Efficiency

View at publisher Summarise with VIRAT
Authors: Vaibhav Raaj Singh, J Arun Pandian
Cited by: 0
References: 5

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Domestic hot water is important to domestic comfort and is an important source of energy. Manual mixing, and fixed thermostat methods are not capable of accommodating dynamic conditions at the inlet side, seasonal changes and personal preferences leading to deterioration of comfort and wastage of energy. The following paper suggests a proposal of an intelligent and fully-adaptive water temperature control system learning optimal mixing strategies in real-time using Deep Q-Learning (DQN). Our developed physics-based digital twin included the thermal mixing dynamics with valve response characteristics including dead-band and hysteresis, transport delays, and heat losses in our design. This environment is engaged by the reinforcement learning agent to be in a position to make policies that reduce tracking error, actual energy consumption (kWh) and control oscillations. A habit-learning module anticipates the preferred temperature of the user during all day hours and adjusts the system proactively. Compared analysis indicates that DQN controller is much more precise than PID and rule-based controllers (error by 83.5% lower) and more stable (reduction of kWh by 17.2%), which makes the reinforcement learning a valid paradigm of the next-generation smart water heating.

Fields of Study & Subjects

Building Energy and Comfort Optimization IoT-based Smart Home Systems Reinforcement Learning in Robotics

Publication details

Journal / source
International Conference on Advances in Intelligent Computing and Applications (AICAPS)
Pages
1–6
Publication date
11 February 2026
Language
EN