TY - JOUR
T1 - Neural Network Based Hysteresis Compensation of Piezoelectric Stack Actuator Driven Active Control of Helicopter Vibration
AU - Meng, De
AU - Xia, Pinqi
AU - Lang, Kai
AU - Smith, Edward C.
AU - Rahn, Christopher D.
N1 - Funding Information:
This work was supported by the National Natural Science Foundation of China (Grant No. 11572150) and the Funding of Jiangsu Innovation Program for Graduate Education (No. KYLX15_0229, 2015).
Funding Information:
This work was supported by the National Natural Science Foundation of China (Grant No. 11572150 ) and the Funding of Jiangsu Innovation Program for Graduate Education (No. KYLX15_0229, 2015 ). Appendix A
PY - 2020/2/1
Y1 - 2020/2/1
N2 - Piezoelectric Stack Actuators (PSAs) have become promising actuators for active control of helicopter fuselage vibration owing to their large output force, rapid response speed, wide working frequency and light weight. However, the hysteresis nonlinearity of PSA has negative influence on the performance of PSA-driven active vibration control. In this paper, to improve the performance of PSA-driven active control of helicopter vibration, the hysteresis nonlinear neural network and hysteresis compensation neural network of PSA have been established based on the Nonlinear Auto Regression eXogenous (NARX) model. The PSA's voltage-displacement relations under two-harmonic actuations were used for training the neural networks of hysteresis nonlinearity and hysteresis compensation. Then the compensation neural network for PSA's hysteresis was integrated into an active control system for helicopter vibration. The results of experimental investigation performed on a scale model of a representative helicopter fuselage floor structure indicate that the neural network based hysteresis compensation of PSA for active control of helicopter vibration can effectively reduce vibration of the scaled frame structure and can reduce more vibration than without compensation.
AB - Piezoelectric Stack Actuators (PSAs) have become promising actuators for active control of helicopter fuselage vibration owing to their large output force, rapid response speed, wide working frequency and light weight. However, the hysteresis nonlinearity of PSA has negative influence on the performance of PSA-driven active vibration control. In this paper, to improve the performance of PSA-driven active control of helicopter vibration, the hysteresis nonlinear neural network and hysteresis compensation neural network of PSA have been established based on the Nonlinear Auto Regression eXogenous (NARX) model. The PSA's voltage-displacement relations under two-harmonic actuations were used for training the neural networks of hysteresis nonlinearity and hysteresis compensation. Then the compensation neural network for PSA's hysteresis was integrated into an active control system for helicopter vibration. The results of experimental investigation performed on a scale model of a representative helicopter fuselage floor structure indicate that the neural network based hysteresis compensation of PSA for active control of helicopter vibration can effectively reduce vibration of the scaled frame structure and can reduce more vibration than without compensation.
UR - http://www.scopus.com/inward/record.url?scp=85077734762&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85077734762&partnerID=8YFLogxK
U2 - 10.1016/j.sna.2019.111809
DO - 10.1016/j.sna.2019.111809
M3 - Article
AN - SCOPUS:85077734762
VL - 302
JO - Sensors and Actuators A: Physical
JF - Sensors and Actuators A: Physical
SN - 0924-4247
M1 - 111809
ER -