Research On Target Recognition Simulation Technology Of Anti-infrared Bait

Infrared bait bombs have been widely used in aircraft and ships because of their good combat effectiveness,high interference and low cost-effectiveness.The rapid development of infrared decoy bombs has promoted the upgrade of infrared guided missile technology.The traditional target recognition algorithm is no longer suitable for future infrared countermeasures.The target recognition algorithm based on deep learning has important research significance in infrared confrontation.Since the true infrared image of the stealth aircraft is highly confidential,and the deep learning requires a large number of infrared image samples,the method of obtaining the infrared image by simulation has been recognized by researchers in various countries.In this paper,the F22 skin,tail flame and decoy bombs of stealth aircraft are modeled and simulated.The related technologies of detector imaging are studied,which provides theoretical and data basis for target recognition of infrared bait.The specific research work is as follows:Firstly,based on the simulation of fluid mechanics,this paper establishes a geometric model for stealth aircraft tail flame,skin and bait bombs,and uses CFD and FLUENT to simulate the geometric model of stealth aircraft,using Curtis-Goldsell(CG)method to approximate Calculate the radiation intensity of the infrared mid-wave(3~5 m?)and infrared long-wave(8~12 m?)of the tail flame;the radiation intensity of the infrared mid-wave and infrared long-wave of the skin is calculated by superimposing the surface energy method;the radiant energy of the burning agent in the decoy bomb is analyzed and calculated.The variation of the radiation intensity of the decoy bomb with the emission time;the influence of carbon dioxide,water vapor and particles in the atmosphere on the imaging is analyzed,and the atmospheric transmittance in the infrared band is proposed.Infrared radiation intensity simulation provides data support for detector imaging simulation.Secondly,for the lack of real infrared image data of stealth aircraft,this paper uses digital simulation to solve this problem.In this paper,Unity3 D is used to establish a 3D simulation environment.The detector model is built in 3D environment.The effect of electrical crosstalk and automatic gain circuit(AGC)on imaging is considered in imaging.The output infrared simulation image sequence is used to construct the deep learning target recognition data set.The YOLOv3 algorithm provides a large amount of infrared image data.Thirdly,this paper uses YOLOv3 as the deep learning target recognition algorithm,and the simulation will get the image establishment data set,70% of the data set as the deep learning training set,and 30% as the test set.The YOLOv3 algorithm is used to learn the characteristics of stealth aircraft and decoy bombs in the image,and the output weight file is used for testing,and the training results are evaluated by corresponding indicators.Finally,in order to simplify the operation steps and facilitate the use,a set of target recognition simulation system software for anti-infrared bait is developed by using C++ and Matlab programming.In summary,the model obtained by deep learning has a detection accuracy of 87% and a detection speed of 29 FPS on the test set.