TY - GEN
T1 - Study on Projectile Impact Point Prediction Based on BP Neural Network
AU - Wu, Nanqi
AU - Liang, Xinyu
AU - Deng, Zhihong
N1 - Publisher Copyright:
© 2022 IEEE.
PY - 2022
Y1 - 2022
N2 - Impact point prediction is the basis of trajectory correction and projectile hit accuracy promotion. Using the artificial neural network to predict impact points has the advantages of timely information output and avoiding error accumulation. In this paper, projectile impact point prediction is studied based on the BP neural network. Projectile flight dynamic phenomena are described through the six-degree-of-freedom rigid body trajectory equation set. The Levenberg-Marquardt algorithm is used to train the BP neural network. Projectile flight state parameters are set as network input, and the impact point position is set as network output. Horizontal components of projectile centroid acceleration are added as network input nodes, which is verified by experiments to be able to improve prediction performance effectively. The traditional method of exerting constant wind disturbance is improved, program structure simplified and data size reduced, which is verified by experiments to be able to meet the requirements of prediction accuracy. Experiments are designed to analyze the effect of data normalization on network performance, which shows that cancelling data normalization is helpful to improve prediction accuracy.
AB - Impact point prediction is the basis of trajectory correction and projectile hit accuracy promotion. Using the artificial neural network to predict impact points has the advantages of timely information output and avoiding error accumulation. In this paper, projectile impact point prediction is studied based on the BP neural network. Projectile flight dynamic phenomena are described through the six-degree-of-freedom rigid body trajectory equation set. The Levenberg-Marquardt algorithm is used to train the BP neural network. Projectile flight state parameters are set as network input, and the impact point position is set as network output. Horizontal components of projectile centroid acceleration are added as network input nodes, which is verified by experiments to be able to improve prediction performance effectively. The traditional method of exerting constant wind disturbance is improved, program structure simplified and data size reduced, which is verified by experiments to be able to meet the requirements of prediction accuracy. Experiments are designed to analyze the effect of data normalization on network performance, which shows that cancelling data normalization is helpful to improve prediction accuracy.
KW - BP neural network
KW - impact point prediction
KW - normalization
UR - http://www.scopus.com/inward/record.url?scp=85151128178&partnerID=8YFLogxK
U2 - 10.1109/CAC57257.2022.10056024
DO - 10.1109/CAC57257.2022.10056024
M3 - Conference contribution
AN - SCOPUS:85151128178
T3 - Proceedings - 2022 Chinese Automation Congress, CAC 2022
SP - 3683
EP - 3688
BT - Proceedings - 2022 Chinese Automation Congress, CAC 2022
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2022 Chinese Automation Congress, CAC 2022
Y2 - 25 November 2022 through 27 November 2022
ER -