Development of evolutionary algorithms for intelligent real-time control systems using field-programmable technology

Almabrok, Abdoalnasir

Ανάπτυξη εξελικτικών αλγορίθμων για ευφυή συστήματα πραγματικού χρόνου με χρήση προγραμματιζόμενης τεχνολογίας πεδίου

Doctoral Thesis

Author

Almabrok, Abdoalnasir

Date

2026-06

Abstract

Ensuring optimal performance of control units, such as Proportional–Integral–Derivative (PID) controllers, is essential for maintaining efficiency, stability, and fast response time in industrial automation systems, such as Heating, Ventilation, and Air Conditioning (HVAC) systems. However, conventional controllers often struggle with the complex, non-linear dy-namics and disturbances present in real-time industrial applications. These challenges have created a growing need for advanced, robust, and adaptive optimization techniques. This PhD thesis focuses on a robust optimization algorithm, the Big Bang Big Crunch (BB-BC), as a fast, reliable solution for auto-tuning control parameters in practical applications. Motivated by the increasing demand for high-speed, energy-efficient, and real-time control in industrial systems, this work explores hardware-accelerated implementations of optimi-zation algorithms, focusing on Field-Programmable Gate Arrays (FPGAs). Although the BB-BC algorithm is well-suited for real-time applications due to its rapid convergence and low computational complexity, software-based implementations running on conventional pro-cessors often fail to meet strict timing constraints. To overcome this limitation, the thesis proposes a novel FPGA-based architecture for the BB-BC algorithm. Field programmable technology offers a realistic solution to improve response time and save power consumption, and presents an efficient way to implement complex algorithms, improving real-time re-sponse in control systems. In other words, the field programmable technology provides a flexible and parallel computing environment for executing optimization algorithms in real-time industrial applications. In this thesis, we show that the BB-BC algorithm is well-suited for FPGA-based acceleration; we propose hardware performance mechanisms that over-come the design limitations observed in classical genetic evolutionary algorithms, which of-ten hinder the performance of their hardware-based accelerators and exhibit limitations, particularly in terms of scalability and response time. We present an FPGA-based accelerator for the BB-BC algorithm, which incorporates a fully pipelined design of both BB-BC phases, Big Bang Phase (BBP), and Big Crunch Phase (BCP). A robust methodological framework for this FPGA-based accelerator has been developed and validated on a typical optimization problem using standard mathematical benchmarks. The experimental results demonstrate significant speedup over software counterparts. Despite the growing interest in robust, real-time optimization for industrial control systems typically characterized by disturbances, nonlinearities, and time delays, existing literature offers few solutions that combine fast-convergent algorithms with hardware acceleration. This thesis addresses this gap by studying the integration of BB-BC into multiple control strategies, including PID, Fractional-Order PID (FOPID), and the Learning Algorithm for Mul-tivariate Data Analysis (LAMDA), and by evaluating their FPGA-based implementations. In particular, the thesis examines the use of BB-BC to tune PID and FOPID controllers, fol-lowed by hardware deployment on FPGA platforms. The approach is validated on an HVAC system, using various modeling strategies such as Second-Order Plus Time Delay (SOPTD) models. Additionally, BB-BC-powered PID, FOPID, and LAMDA controllers are designed and analyzed for robustness under different disturbance profiles, enabling comparative evalua-tion across diverse industrial scenarios. MATLAB/Simulink simulations of the combined op-timization–control framework further highlight the superior timing behavior, scalability, and energy efficiency of the FPGA-accelerated BB-BC solutions compared to traditional tech-niques. The key objectives of this PhD thesis are to: ▪Develop a novel FPGA-based implementation of the BB-BC optimization algorithm and demonstrate its performance through representative case studies. ▪Integrate the BB-BC algorithm into PID and FOPID controllers for parameter optimiza-tion, and validate the proposed approach on HVAC systems. ▪Extend the BB-BC optimization approach to LAMDA controllers to explore its applica-bility in more complex industrial environments. ▪Enhance real-time performance and energy efficiency through hardware-accelerated optimization using FPGA technology.

Department

Σχολή Τεχνολογιών Πληροφορικής και Επικοινωνιών. Τμήμα Πληροφορικής

Number of pages

157

Language

English

URI

https://dione.lib.unipi.gr/xmlui/handle/unipi/19577

Collections

Τμήμα Πληροφορικής

Show full item record

Except where otherwise noted, this item's license is described as
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα