| When a high-speed aircraft is flying in the atmosphere,its head is affected by the aerodynamic optical flow field,the image received by the aircraft’s airborne optical system is shifted,which leads to errors or even failures in the control and decision of the aircraft.In military applications,this effect is often fatal.In this paper,the aerooptical imaging deviation for a typical blunt-headed aircraft is analyzed and predicted.The research contents are as follows:(1)For a typical blunt-headed aircraft,the relationship between the aero-optical imaging deviation and the LOS angle of the target is given: with the LOS angle increasing,the aero-optical imaging deviation monotonically decreases.The imaging deviation slope is introduced to characterize the sensitivity of the imaging deviation to the change of the LOS angle: When the LOS angle is between 5° and 40°,the imaging deviation is sensitive to the change of the LOS angle,and the imaging deviation slope is steep;When the LOS angle is 40° to 85°,the imaging deviation decreases slowly and the imaging deviation slope is relatively flat.(2)In terms of aero-optical imaging deviation prediction.The improved particle swarm optimization(IPSO)algorithm is used to optimize the weight and threshold of BP neural network,and the optimized IPSO-BP model is applied to the prediction of aero-optical imaging deviation.In IPSO algorithm,the diversity of particles in the population is judged according to the aggregation degree.When the aggregation degree is detected to exceed the set threshold,the concept of similarity is used to measure the similarity between particles and the global optimal particles,and the particles with higher similarity are discretized by the strategy of variation to increase the diversity of particle swarm,so as to make the global search ability and local exploration of the population Ability tends to balance.(3)Based on bat algorithm in optimization algorithm,an improved bat algorithm(IBA)optimization limit learning machine(ELM)method for aero-optical imaging deviation prediction is implemented.Use the global optimization ability of the improved bat algorithm to obtain the optimal input layer weight matrix and hidden layer bias of the ELM model,which overcomes the disadvantage of the traditional elm that the input layer weight matrix and the hidden layer bias reduce the generalization ability of the network.The IBA-ELM model has high prediction accuracy and fast operation speed,which provides a way for the fast estimation of aero-optical imaging migration. |
