TY - GEN
T1 - Multivariable PID neural network based flight control system for Small-scale unmanned helicopter
AU - Qi, Guangping
AU - Song, Ping
AU - Li, Kejie
PY - 2009
Y1 - 2009
N2 - To design the flight control system (FCS) of small-scale unmanned helicopter is still a difficult challenge today. The hardware and software architecture of FCS was designed in this paper. And one novel control approach based on Multivariable PID Neural Network (MPIDNN) was firstly used to design the FCS of small-scale unmanned helicopter on the hardware platform. MPIDNN is suitable for controlling the multi-input multi-output (MIMO), nonlinear, highly coupled, uncertain and dynamic system such as helicopter. Both the training and study algorithm based on target function and MPIDNN forwards algorithm were designed in this control system. The result of simulation indicates that the training algorithm can solve the offline training and study problem of small-scale unmanned helicopter. The forwards algorithm can control the flight of helicopter well and its maximum magnitude of error is about 1%. Simulation shows that the performance of our control approach is perfect.
AB - To design the flight control system (FCS) of small-scale unmanned helicopter is still a difficult challenge today. The hardware and software architecture of FCS was designed in this paper. And one novel control approach based on Multivariable PID Neural Network (MPIDNN) was firstly used to design the FCS of small-scale unmanned helicopter on the hardware platform. MPIDNN is suitable for controlling the multi-input multi-output (MIMO), nonlinear, highly coupled, uncertain and dynamic system such as helicopter. Both the training and study algorithm based on target function and MPIDNN forwards algorithm were designed in this control system. The result of simulation indicates that the training algorithm can solve the offline training and study problem of small-scale unmanned helicopter. The forwards algorithm can control the flight of helicopter well and its maximum magnitude of error is about 1%. Simulation shows that the performance of our control approach is perfect.
UR - http://www.scopus.com/inward/record.url?scp=70449652598&partnerID=8YFLogxK
U2 - 10.1109/ICINFA.2009.5205123
DO - 10.1109/ICINFA.2009.5205123
M3 - Conference contribution
AN - SCOPUS:70449652598
SN - 9781424436088
T3 - 2009 IEEE International Conference on Information and Automation, ICIA 2009
SP - 1331
EP - 1335
BT - 2009 IEEE International Conference on Information and Automation, ICIA 2009
T2 - 2009 IEEE International Conference on Information and Automation, ICIA 2009
Y2 - 22 June 2009 through 25 June 2009
ER -