Spectral analysis technology is widely used in agriculture， food， medicine， chemical industry， environment and other fields due to its advantages of simplicity， speed and non-destructiveness. However， spectral data suffer from high variable dimensionality， multicollinearity， inclusion of redundant information and high-frequency noise， etc. Variable selection is a crucial step in reducing data dimensionality and improving the efficiency and predictive performance of chemometric modeling. Meta-heuristic algorithms have been widely used for spectral variable selection in recent years due to their powerful global optimization capability， good adaptability， parallel potential and robustness. In this paper， we systematically review the progress of the application of four types of metaheuristic algorithms， namely， evolutionary algorithms， swarm intelligence optimization algorithms， physics or chemistry-based algorithms， and human-based algorithms， to the selection of spectral variables， and summarize the strengths and weaknesses of metaheuristic algorithms. Finally， the challenges and future prospects of metaheuristic algorithms for selecting spectral variables are discussed to provide a theoretical basis and technical reference for further promoting the application of metaheuristic algorithms in spectral variable selection.

Contents

1 Introduction

2 Spectral variable selection methods based on evolutionary algorithms

2.1 Genetic algorithm

2.2 Differential evolution

2.3 Evolutionary programming

2.4 Evolutionary strategies

2.5 Genetic programming

3 Spectral variable selection methods based on physics or chemistry algorithms

3.1 Simulated annealing

3.2 Gravitational search algorithm

3.3 Sine cosine algorithm

3.4 Harmony search algorithm

3.5 Equilibrium optimizer

4 Spectral variable selection methods based on swarm intelligence optimization algorithms

4.1 Ant colony optimization

4.2 Particle swarm optimization

4.3 Firefly algorithm

4.4 Grey wolf optimizer

4.5 Whale optimization algorithm

4.6 Other swarm intelligence optimization algorithms

5 Spectral variable selection methods based on human algorithms

5.1 Tabu search

5.2 Team progress algorithm

5.3 Battle royale optimization algorithm

5.4 Hiking optimization algorithm

6 Applications

6.1 Hyperspectral image

6.2 Near-infrared spectroscopy

6.3 Ultraviolet spectroscopy

6.4 Mid-infrared spectroscopy

6.5 Raman spectroscopy

6.6 Other spectroscopy

7 Conclusion and outlook