Architecture Design And Implementation Of High Speed And Low Power Kalman Filter Based On FPGA

The Kalman filter algorithm not only has good filtering characteristics,but also adopts a recursive structure,so it is suitable for implementation with a computer or a programmable chip.The excellent characteristics of Kalman filtering algorithm make it have a wide range of applications in dynamic positioning,communication and signal processing,target tracking and many other fields.In engineering practice,in order to meet the real-time requirements of the system,a Field Programmable Gate Array(FPGA)is often used to implement the Kalman filter.With the continuous development of FPGA technology and people's requirements for FPGA design system performance,how to improve the system performance of FPGA-based Kalman filter is a major challenge for Kalman filter engineering applications.For this reason,this article has carried on the thorough research,has proposed two kind of high-performance Kalman filter framework based on FPGA,and has carried on the design analysis and the feasibility verification to it.First,this paper reviews the development of the Kalman filter theory,introduces the classical Kalman filter algorithm and its system mathematical model.Then,three commonly used non-linear Kalman filter algorithms,Extended Kalman Filter(EKF),Unscented Kalman Filter(UKF)and Cubature Kalman Filter(CKF),are introduced.Then the application of Kalman filter in helicopter rotor blockage prediction and maneuvering target tracking is studied,and the algorithm simulation analysis is performed.Then,from the perspective of FPGA design speed and power consumption,this paper proposes two FPGA-based high-speed low-power Kalman filter design architecture.One is an architecture based on the combination of Faddeev algorithm and oscillating array structure proposed for the EKF algorithm.The EKF implemented using this architecture can greatly increase the parallelism of the algorithm and improve the real-time performance of the system;Another is a soft-hardware joint architecture proposed for a nonlinear Kalman filter,Under this architecture,the software plays a role in system control and the nonlinear model part of the computing system,and implement the Kalman filter core algorithm part in hardware.The Kalman filter implemented by this architecture not only guarantees the performance,but also reasonably allocates system resources,and the system flexibility is greatly enhanced.Finally,FPGA implementation and system functional verification are performed for the Kalman filter of the helicopter rotor block model and target tracking model.Firstly,the Kalman filter of the helicopter rotor block model was implemented by FPGA,and the system was verified by the function.The result showed that the Kalman filter can predict the occlusion condition of the rotor well,thus realizing the burst communication of the helicopter return link;then the UKF design of the target tracking model is implemented using a hardware-software joint architecture.The functional verification of the system FPGA module is performed,and the performance of the FPGA implementation system in terms of speed,resources,and power consumption is analyzed.Then the software-hardware joint test was conducted on the ZYNQ7000 platform.The comparison with the simulation results verified the correctness of the system's functions.