Using evolutionary algorithms to solve complicated optimisation problems- UJ’s Prof Yanxia Sun

Using evolutionary algorithms to solve complicated optimisation problems- UJ’s Prof Yanxia Sun


Publishing Date: 6/2/2021 9:00 AM

The field of evolutionary algorithms yields various theoretical approaches that help search for global optima within complex environments. These environments are often highly multidimensional, nonlinear, noisy, and with many constraints.

Yanxia Sun is serving as a Professor in the Department of Electrical and Electronic Engineering Science and Acting Vice- Dean of the Faculty of Engineering and the Built Environment at the University of Johannesburg (UJ). 

WATCH: Professorial Inauguration: Prof Yanxia Sun

 

In her professorial address, she emphasised that evolutionary algorithms (EA's) are motivated by optimisation processes that we observe in nature and can solve most of the complicated optimisation problems using model or data driven methods. Her presentation gave a brief introduction of several most famous evolutionary algorithms, which include genetic algorithm, differential evolution, multi-objective optimisation concepts and methods.

The findings reveal the algorithms applied to real-world mechanical, biological, and chemical problems, as well as to theoretical benchmarks. "The renewable energy systems include different energy sources and end users, and their complicated optimisation problem with different constraints. Two examples of using evolutionary algorithm to solve the renewable system optimisation problems do show the efficiency of the algorithms," explained Prof Sun.

Furthermore, because UJ is moving towards the Fourth Industrial Revolution (4IR) thinking, her aim is to advance the identification of combination of problem classes and effective optimisation methods, as well as the innovative and hybrid implementations of the algorithms.

Prof Sun stressed that EAs are definitely good tools to quickly develop a prototype optimisation method to solve a problem. "This can give you some baseline results and a significant insight into the problem structure. Especially if you are developing a solution for an industry partner, quickly obtaining a small prototype piece of software can be helpful. This allows you and your partner to verify whether the problem is defined correctly and whether some constraints are missing."



Prof Yanxia Sun