COMPARISON OF THE OPTIMIZATION OF DECACOPTER LONGITUDINAL FLIGHT UNDER THE EFFECT OF MORPHING WITH SPSA, SGD AND YSA

Abstract

Control studies on unmanned aerial vehicles (UAVs) have gained popularity in recent years. Maximising UAV performance is the main point of the control field. Rotary wing UAVs in the UAV group are named according to the number of rotors. In this study, the control of decacopter, a UAV with 10 rotors, is considered. The morphing effect was obtained by changing the arm lengths of the decacopter. Arm lengths are estimated by simultaneous perturbation stochastic approximation (SPSA) and control parameters for longitudinal flight are obtained by estimating proportional integral derivative (PID) coefficients according to the estimated arm lengths. The moments of inertia were estimated separately with stochastic gradient descent (SGD) and artificial neural networks (ANN) and longitudinal flight simulations were performed. The data set for training with SGD and ANN was obtained from the models drawn in Solidworks drawing programme. Longitudinal flight simulations were performed in Matlab/Simulink environment according to both SGD and ANN predictions and the results were compared according to the design performance criteria. According to the design performance criteria, it was observed that the parameters estimated by ANN gave better results in longitudinal flight under the effect of morphnig compared to SGD.

Keywords

• decacopter
• UAV
• optimization
• SGD
• ANN
• SPSA

DECACOPTER BAŞKALAŞIM ETKİSİ ALTINDA BOYLAMASINA UÇUŞUNUN SPSA, SGD VE YSA İLE OPTİMİZASYONUN KARŞILAŞTIRILMASI

Abstract

İnsansız hava raçları (İHA) üzerine kontrol çalışmaları son yıllarda popülerlik kazanan bir konudur. İHA performansının maksimize edilmesi kontrol alanın temel noktasıdır. İHA grubunda yer alan döner kanatlı İHA’lar rotor sayısına göre isimlendirilir. Bu çalışmada 10 rotora sahip İHA olan decacopter kontrolü ele alınmıştır. Decacopter’in kol uzunlukları değiştirilerek başkalaşım etkisi elde edilmiştir. Simultaneous perturbation stochastic approximation (SPSA) ile kol uzunlukları tahmin edilmiş ve tahmin edilen kol uzunluklarına göre oransal integral türev (PID) katsayıları da tahmin edilerek boylamasına uçuş için kontrol parametreleri elde edilmiştir. Değişen kol uzunluklarına göre atalet momentlerinin tahmini ise stochastic gradient descent (SGD) ve yapay sinir ağları (YSA) ile ayrı ayrı tahmin edilerek boylamasına uçuş simülasyonları yapılmıştır. SGD ve YSA ile eğitim yapılabilmesi veri seti Solidworks çizim programında çizilen modellerden elde edilmiştir. Matlab/Simulink ortamında hem SGD hem de YSA tahminlerine göre boylamasına uçuş simülasyonları yapılmış ve sonuçlar tasarım performans kriterlerine göre karşılaştırılmıştır. Tasarım performans kriterlerine göre YSA ile tahmin edilen parametrelerin SGD’ye göre başkalaşım etkisi altında boylamasına uçuşta daha iyi sonuçlar verdiği gözlemlenmiştir.

Keywords

• decacopter
• İHA
• optimizasyon
• SGD
• YSA
• SPSA

References

1. [1]İ. ATA, “Hibrit-Elektrikli İtki Sistemine Sahip Cessna 172S Uçağında Uçuş Parametrelerinin Gri İlişkisel Analiz Yöntemi ile Optimizasyonu,” Gazi Üniversitesi Fen Bilim. Derg. Part C Tasarım ve Teknol., vol. 11, no. 4, pp. 1223–1231, 2023, doi: 10.29109/gujsc.1394062.
2. [2] M. M. Ferdaus, S. G. Anavatti, M. Pratama, and M. A. Garratt, “Towards the use of fuzzy logic systems in rotary wing unmanned aerial vehicle: a review,” Artif. Intell. Rev., vol. 53, no. 1, pp. 257–290, 2020, doi: 10.1007/s10462-018-9653-z.
3. [3] S. D. Hanford, L. N. Long, and J. F. Horn, “A small semi-autonomous rotary-wing Unmanned Air Vehicle (UAV),” Collect. Tech. Pap. - InfoTech Aerosp. Adv. Contemp. Aerosp. Technol. Their Integr., vol. 3, no. September, pp. 1539–1548, 2005, doi: 10.2514/6.2005-7077.
4. [4] T. Zhao and W. Li, “Design Configuration and Technical Application of Rotary-Wing Unmanned Aerial Vehicles,” Mechatronics Intell. Transp. Syst., vol. 1, no. 1, pp. 69–85, 2022, doi: 10.56578/mits010108.
5. [5] C. Lee, S. Kim, and B. Chu, “A Survey: Flight Mechanism and Mechanical Structure of the UAV,” Int. J. Precis. Eng. Manuf., vol. 22, no. 4, pp. 719–743, 2021, doi: 10.1007/s12541-021-00489-y.
6. [6] T. Oktay and O. Kose, “Simultaneous quadrotor autopilot system and collective morphing system design,” Aircr. Eng. Aerosp. Technol., vol. 92, no. 7, pp. 1093–1100, 2020, doi: 10.1108/AEAT-01-2020-0026.
7. [7] N. Bucki and M. W. Mueller, “Design and Control of a Passively Morphing quadcopter,” in 2019 International Conference on Robotics and Automation (ICRA), IEEE, 2019, pp. 9116–9122.
8. [8] O. Kose and T. Oktay, “Simultaneous design of morphing hexarotor and autopilot system by using deep neural network and SPSA,” Aircr. Eng. Aerosp. Technol., vol. 95, no. 6, pp. 939–949, 2023, doi: 10.1108/AEAT-07-2022-0178.

9. [9] O. KOSE, “Başkalaşımın Octorotor Boylamasına Uçuşuna Etkisi,” Black Sea J. Eng. Sci., vol. 6, no. 3, pp. 185–192, 2023, doi: 10.34248/bsengineering.1273089.
10. [10] T. Oktay, F. Sal, O. Kose, and E. Ozen, “Stochastic Longitudinal Autopilot Tuning for Best Autonomous Flight Performance of a Morphing Decacopter,” in ICRETS 2023: International Conference on Research in Engineering, Technology and Science, Budapest: ISRES Publishing, 2023, pp. 50–58.
11. [11] L. CHU, Q. LI, F. GU, X. DU, Y. HE, and Y. DENG, “Design, modeling, and control of morphing aircraft: A review,” Chinese J. Aeronaut., vol. 35, no. 5, pp. 220–246, 2022, doi: 10.1016/j.cja.2021.09.013.
12. [12] J. C. Spall, “An Overview of the Simultaneous Perturbation Method for Efficient Optimization,” 1998. [13] Yang WANG, “Modified Simultaneous Perturbation Stochastic Approximation Method for Power Capture Maximization of Wind Turbines,” Kansas State University, 2013.
13. [14] J. C. Spall and D. C. Chin, “Traffic-responsive signal timing for system-wide traffic control,” Transp. Res. Part C Emerg. Technol., vol. 5, no. 3–4, pp. 153–163, 1997, doi: 10.1016/S0968-090X(97)00012-0.
14. [15] D. Heydon, B and D. Hill, S, “Maximizing Target Damage Through Optimal Aimpoint Patterning,” in A/AA 3rd Biennial National Forum on Weapon System Effectiveness, Seal Beach, 2003, pp. 18–20.
15. [16] T. Oktay and S. Coban, “Simultaneous longitudinal and lateral flight control systems design for both passive and active morphing TUAVs,” Elektron. ir Elektrotechnika, vol. 23, no. 5, pp. 15–20, 2017.
16. [17] X. Cui, W. Zhang, Z. Tüske, and M. Picheny, “Evolutionary stochastic gradient descent for optimization of deep neural networks,” Adv. Neural Inf. Process. Syst., vol. 2018-Decem, no. NeurIPS, pp. 6048–6058, 2018.
17. [18] B. Gess and S. Kassing, “Stochastic Modified Flows , Mean-Field Limits and Dynamics of Stochastic Gradient Descent,” vol. 25, pp. 1–27, 2024.
18. [19] A. Arı and M. Erşen Berberler, “Yapay Sinir Ağları ile Tahmin ve Sınıflandırma Problemlerinin Çözümü İçin Arayüz Tasarımı,” Acta Infologica, vol. 1, no. 2, pp. 55–72, 2017.