Application Research Of Extreme Learning Machine In Passive Positioning Technology

Today,the battlefield in the 21 st century has been separated from large-scale human struggle,and modern electronic countermeasure technology continues to develop and mature.Obtaining the enemy’s target position has become the key to mastering the initiative on the battlefield.Therefore,passive positioning has become the most important technology in the field of electronic countermeasures.one.Passive positioning gets rid of the limitation of electromagnetic signal transmission,and at the same time,it can avoid being detected by the enemy.It is more stable and reliable in the face of enemy interference,ensuring the security of the system,but the system requires complex calculations.In order to obtain the position of the enemy’s target.The extreme learning machine based on the feedforward neural network has a better effect in the face of nonlinear feature relationships,and is suitable for dealing with complex nonlinear fitting problems between the longitude and latitude position of the target and the data measured by the receiving device.In this paper,passive positioning is taken as the research object.In order to solve the problem of complex calculation and poor positioning accuracy of passive positioning,the target positioning is achieved by establishing Angle of Arrival(AOA)and Time-Difference of Arrival(TDOA),Time-Difference of Arrival/Angle of Arrival,(TDOA/AOA)and the nonlinear mapping relationship between the longitude and latitude of the detection target.The specific research contents are as follows:(1)In AOA-based passive positioning,aiming at the problem of large positioning errors and being easily affected by measurement errors in AOA passive positioning,a nuclear Extreme Learning Machine is introduced,and an optimization algorithm based on Butterfly Optimization Algorithm(BOA)is proposed.A localization model for Kernel Extreme Learning Machines.The model uses the improved Butterfly Optimization Algorithm to optimize the kernel parameters and regularization coefficients of the KELM.The optimized KELM can more efficiently utilize the kernel function to map the input space to the high-dimensional feature space.The results show that the target position predicted by KELM’s AOA passive positioning model is very close to the real position after optimized by the improved Butterfly Optimization Algorithm.Compared with the positioning model of Extreme Learning Machine(ELM)and other KELM positioning models,the positioning accuracy is higher.(2)In TDOA-based passive positioning,in order to solve the problems of traditional TDOA passive positioning calculation complexity,slow speed,and threshold effect,the idea of deep learning is introduced into the extreme learning machine,and a Chimp Optimization Algorithm is established(Ch OA)to optimize the localization model for Deep Extreme Learning Machine.The model uses the improved Chimp Optimization Algorithm to determine the number of network nodes and regularization factor of DELM,and then the optimized DELM uses the TDOA data obtained from multi-station measurement to locate the detection target.The experimental results show that the positioning accuracy of DELM’s TDOA passive positioning model after the optimization of the improved Chimp Optimization Algorithm is higher than that of the DELM model after the optimization of the Chimp Optimization Algorithm.Compared with other DELM positioning models and single-core extreme learning machine model positioning,the effect is higher.(3)In view of the large number of TDOA passive positioning receiving stations and the low accuracy of AOA passive positioning of long-distance targets,the TDOA positioning method and the AOA positioning method are combined to form a TDOA/AOA passive positioning method.An improved Salp Swarm Algorithm(SSA)optimized HKELM uses TDOA/AOA data to achieve target positioning.The model uses the improved Salp Swarm Algorithm to determine the network parameters of the Hybrid Kernel Extreme Learning Machine,and then the optimized HKELM locates the detected target.The experimental results show that the Hybrid Kernel Extreme Learning Machine has stronger generalization than the Extreme Learning Machine,and compared with other HKELM positioning models,the Hybrid Kernel Extreme Learning Machine has higher positioning accuracy after the optimization of the improved Salp Swarm Algorithm.