Abstract: When Infrared Search and Track (IRST) System works in double station cooperative detection and positioning in operational use, it is difficult to match a maneuvering target model with the actual target motion, which worses position estimation error. Therefore, an adaptive filtering algorithm based on curvilinear model is proposed. The method here shifts the thinking that turning angular velocity is estimated by direction angle and tangential acceleration is estimated by inserted acceleration in traditional method. In this paper, turning angular velocity and target acceleration is both treated as augmented state variable and its corresponding process noise covariance expression is derived, which greatly decreases the calculation burden of traditional two-layer filtering structure, as well as improving the estimation accuracy to tangential acceleration. In the meanwhile, the paper optimizes the heading angle calculation method based on library function "atan2()" in embedded software and its transfer relationship between four quadrants. It demonstrates that the proposed method has the better adaptation to maneuvering target and has better positioning accuracy through double IRST cooperative simulation instance.

Key words: Infrared Search and Track (IRST) System, double IRST cooperative positioning, curvilinear model, maneuvering target positioning