Optimizing Internet-of-Things Energy Management: Integrating Theory of Inventive Problem Solving With Transfer Learning and Advanced Optimization Algorithms

This article introduces the Energy-Efficient Theory of Inventive Problem Solving (EETRIZ) approach, designed to reduce electrical energy waste resulting from human mistakes in IoT-enabled environments. EETRIZ utilizes a novel integration of transfer learning and an activity-dependent environmental management algorithm to adjust settings dynamically according to real-time occupancy and activity data, hence improving energy efficiency. This system efficiently utilizes the advantages of the artificial intelligence-based adaptive gradient algorithm (AdaGrad) and root mean squared propagation (RMSProp) optimization methods to improve prediction accuracy through enhanced weight determination. EETRIZ is developed in the C programming language and is underpinned by comprehensive platforms and libraries, such as MPLAB, Nuvoton 8051 Series microcontroller unit (MCU) programming, GNU’s Not Unix multi-precision library (GMPLibrary)-GMP-5.1.1, and Miracle Library. Thorough hardware testing verifies that EETRIZ surpasses current solutions in energy efficiency, cost-effectiveness, accuracy, and user-friendliness. The system’s capacity to simultaneously control numerous IoT devices enhances its utility in various environments, including residences, workplaces, and educational facilities, providing a scalable solution to mitigate excessive energy consumption resulting from human mistakes.