Development of an Artificial Neural Network Control Allocation Algorithm for Small Tailless Aircraft Based on Dynamic Allocation Method

• Published in: International journal of aeronautical and space sciences Vol. 23; no. 2; pp. 363 - 378
• Main Authors: Kang, Jisoo, Choi, Keeyoung
• Format: Journal Article
• Language: English
• Published: Seoul The Korean Society for Aeronautical & Space Sciences (KSAS) 01.04.2022; 한국항공우주학회
• Subjects: Aerospace Technology and Astronautics; Engineering; Fluid- and Aerodynamics; Original Paper; 항공우주공학; Optimized multi-sine input; Tailless aircraft; Artificial neural network; Real time; Control allocation
• ISSN: 2093-274X; 2093-2480
• DOI: 10.1007/s42405-021-00425-4

This paper presents an artificial neural network (ANN) control allocation algorithm based on a dynamic allocation (DA) method that finds the optimal sets of control surfaces satisfying command requirements. The purpose of the research is to develop an algorithm with performance comparable to that of the DA and efficiency adequate for a small over-actuated tailless aircraft with limited computation power. Scheduling or optimization allocation methods are typically used, but each has disadvantages such as performance degradation and requirement of heavy computation power, respectively. The proposed algorithm is trained by the Bayesian regularization algorithm for improving generalization performance. Training sets are efficiently collected by an optimized multi-sine input and cover broad flight conditions. The performance of the ANN allocator is verified through four aspects, which are control allocation performance, computation efficiency, robustness, and real-time performance. The proposed control allocation method shows a remarkable performance compared with the dynamic allocation method. The ANN is well suited as a part of the embedded software for a small tailless aircraft which typically has limited computational resources.