Research On Continuous Tumbling Self-calibration For Inertial Navigation Platform System

In order to realize the inertial platform self-calibration process faster,more convenient,higher precision,continuous tumbling self-calibration method is studied.Firstly,the main error parameters of the inertial platform and the inertial instrument are analyzed,including 47 error parameters such as the perpendicularity of each frame axis of the inertial platform,the installation errors of the inertial instrument and the output error model coefficients of the inertial instrument.Then,a suitable coordinate system is established,and the transfer process of various error sources between the coordinate systems is derived,and the observation models of accelerometer and gyroscope are given.The attitude error equation expressed by the dynamic misalignment angle is taken as the state equation,while the output of the accelerometer is used as the observation equation,so that the state space model of continuous tumbling of inertial platform is established.Two continuous tumbling paths are designed using observability analysis and optimization algorithm,which can fully excite various error parameters.The PWCS theory and SVD theory are introduced firstly and the error parameters and misalignment angles of the system are taken together as state variables.The observable matrix of the system based on PWCS is used to judge the observability of the system state under a certain path.According to the parameter coupling or non-observable state,the path is redesigned to ensure that all parameters can be excited.SVD is used to calculate the observability of all kinds of system error parameters and judge the ability of error parameter identification in this path.Through the analysis of PWCS and SVD,an appropriate path of rotation is given,and the simulation analysis is carried out to obtain the output curves of the accelerometers under this path.Then,the basic theory of PSO is introduced and the appropriate index function is designed with the information matrix of the state space model as a reference.For the constraints of the index function,penalty function method and direct method are used to carry out simulation analysis respectively.Simulation results show that the direct method can ensure better optimization effect of each particle in the global scope,and realize the goal of minimum index function and zero constraint quantity.By analyzing the parameters of PSO,the values of the learning factor and the inertial factor are given during the iteration of the algorithm,so as to ensure the fast convergence and optimization of the algorithm in the global scope.Finally,a suitable continuous tumbling path is designed by taking the torque angular velocity as a reference.Based on the above rotating path design,Kalman filtering is used to identify the error parameters.In the filtering process,the misalignment angle and 47 error parameters are also used as state variables to establish a suitable filtering model.After the proper filtering model is established and the error parameter units are given,parameter identification is realized by giving the error variation curves in the process of parameter identification,and the final identification error and identification accuracy of all parameters are listed.The identification results show that the platform can also realize the self-calibration of many error parameters and the parameter identification accuracy is high.