A PROACTIVE ARCHITECTURE FOR SMART HOME CONTROL BASED ON CONTEXTUAL USER INTENTS AND MULTI-OBJECTIVE OPTIMIZATION

DOI: 10.31673/2412-4338.2025.038716

Abstract

This paper addresses the problem of low adaptability and flexibility in existing smart home control systems. An analysis of the existing literature reveals that most current approaches rely on static rules, failing to account for dynamic, context-dependent user preferences. To overcome these limitations, this work proposes and investigates a novel proactive control architecture. Its key elements include predictive modeling to evaluate future states, multi-objective optimization based on high-level user intents (e.g., the balance between comfort and economy), and an adaptive learning mechanism. The novelty of the approach lies in the system's ability to learn contextual intents (e.g., for different times of day) by analyzing manual user overrides as implicit feedback, rather than learning a single set of global preferences.

The architecture's effectiveness was validated through a 60-day simulation with seasonal changes and dynamic tariffs, comparing its performance against two rule-based systems. The results demonstrate that the proactive agent achieved the highest level of comfort (41.5% of the time in the target zone) and, most importantly, reduced the number of manual interventions by over 97% compared to the baseline model. It is proven that the proposed approach enables the creation of flexible, human-centric systems that autonomously adapt to inhabitants' behavior, thereby minimizing their cognitive load.

Keywords: smart home, proactive control, adaptive learning, intent-based control, contextual intents, multiobjective optimization, simulation modeling.