A social spider inspired metaheuristic for global numerical optimization and its applications

Abstract

The growing complexity of real-world problems has motivated computer scientists to search for efficient problem solving methods. Meta-heuristics based on evolutionary computation and swarm intelligence are outstanding examples of nature-inspired solution techniques. According to the No-Free- Lunch Theorem, all meta-heuristics shall perform equally when all possible objective functions are consider, but superior performing algorithms exist if particular classes of functions or general but real-world problems are considered. This motivates researchers to develop new searching methodologies with superior performance on new particular classes of problems.

This thesis focuses on developing a new meta-heuristic for solve global numerical optimization problems. Firstly, inspired by the social spiders, we propose a novel Social Spider Algorithm (SSA) to solve global optimization problems. This algorithm is mainly based on the foraging strategy of social spiders, utilizing the vibrations on the spider web to determine the positions of preys. Different from the previously proposed swarm intelligence algorithms, we introduce a new social animal foraging strategy model, i.e., information sharing model, to solve optimization problems. With the parameter settings generated from a preliminary test, the performance of SSA is assessed with a complete suite of benchmark functions, and compared with the state-of-the-art meta-heuristics. SSA attains a significant performance improvement over the compared algorithms. In addition, we investigate the parameter sensitivity thoroughly. We systematically evaluate SSA on a benchmark function suite with different control parameters. We employ an advanced non-parametric statistical test to generate statistically significant conclusion on the best performing parameter settings. The conclusion can be adopted in future work to reduce the effort in parameter tuning. In addition, we perform a success rate test to reveal the impact of the control parameters on the convergence speed of the algorithm.

Secondly, we identify two real-world practical optimization problems in different fields of research that can be solved with SSA efficiently. A new Optimal Cell zooming and Renewable energy utilization Problem (OCRP) is formulated and solved using SSA. This problem aims to reduce the total on-grid energy consumption by controlling the operating pattern of the base stations in a green cellular network. The optimization problem is decomposed and solved using SSA. The simulation results suggest significant energy savings compared with previous algorithms.

We also employ SSA to solve a classical power system operation and control problem, named Economic Load Dispatch (ELD) problem. As modern power system introduces new models of the power units, the problem is non-convex, non-differentiable, and non-continuous, making meta-heuristics good candidates as problem solvers. To solve such non-convex ELD problems, we propose a new approach by manipulating SSA and introducing new optimization schemes. The algorithm is modified and enhanced to adapt to the unique characteristics of ELD problems, e.g., valve-point effects, multifuel operations, prohibited operating zones, and line losses. To demonstrate the superiority of our proposed approach, five widely-adopted test systems are employed and the simulation results are compared with the state-of-the-art algorithms. The simulation results indicate that SSA can solve ELD problems effectively and efficiently.