A new method for spinning projectile aerodynamic estimation: Extreme learning machine optimized by adaptive particle swarm

Jun Guan,Wenjun Yi,Youran Xia

doi:10.1063/5.0076103

Abstract

Aiming at the problem of aerodynamic parameter identification of a spinning projectile, an adaptive particle swarm optimization for the extreme learning machine algorithm is proposed in this paper. The algorithm uses the adaptive particle swarm optimization algorithm to optimize the hidden layer weight and threshold of the extreme learning machine to avoid the problem of unstable identification results caused by the random weight and threshold of the traditional extreme learning machine. The free flight data of the projectile are generated by numerical simulation, and the aerodynamic parameters of a projectile are identified by the proposed algorithm. Simulation results show that the proposed algorithm can effectively identify the aerodynamic parameters of the projectile, and it has high identification accuracy and fast convergence speed. The proposed algorithm is suitable for engineering applications.

Highlights

To acquire the proper aerodynamic parameters of an uncontrolled projectile is the critical technology to reduce the impact point dispersion and improve the hit accuracy.1 Typically, the main ways to obtain projectile aerodynamic parameters are theoretical calculation, wind tunnel measuring, actual flight data identification method, etc
The adaptive particle swarm optimization extreme learning machine (APSO-Extreme learning machine (ELM)) algorithm is used for the aerodynamic identification of a spinning projectile
The structure of the APSO-ELM model is the simplest, and the identification accuracy is much higher than ELM and PSO-ELM

Summary

INTRODUCTION

To acquire the proper aerodynamic parameters of an uncontrolled projectile is the critical technology to reduce the impact point dispersion and improve the hit accuracy. Typically, the main ways to obtain projectile aerodynamic parameters are theoretical calculation, wind tunnel measuring, actual flight data identification method, etc. To acquire the proper aerodynamic parameters of an uncontrolled projectile is the critical technology to reduce the impact point dispersion and improve the hit accuracy.. The study of the theory and technology of projectile aerodynamic parameter identification is an essential subfield in that of the aerocraft. Developing the projectile aerodynamic parameter identification theory and methods can more effectively know the projectile dynamic properties, which has crucial theoretical research and engineering application values. Using swarm intelligence algorithms to optimize the extreme learning machine can enhance the convergence speed and network accuracy. An adaptive particle swarm optimization (APSO) algorithm is proposed to improve the optimization efficiency of the algorithm. The adaptive particle swarm optimization extreme learning machine (APSO-ELM) algorithm is used for the aerodynamic identification of a spinning projectile

MODELING

Extreme learning machine

Adaptive inertial weight

Ballistic data preprocessing

APSO-ELM algorithm steps

CONCLUSIONS