Hypersonic vehicle is a new concept vehicle developed by many countries in recent years.Aiming at the problem that it is difficult to track its flight trajectory with high precision,combined with the strong adaptive and self-learning ability of neural network,an online feedback filtering algorithm for hypersonic vehicle is proposed.Based on the current statistical model,BP neural network and Kalman filter are used to design the filter to realize the high-precision tracking of hypersonic vehicle.Finally,the simulation results show that the online feedback filtering algorithm is effective in tracking hypersonic vehicle.

The difference in the structure of radar transmitters and the use of electronic devices determine the uniqueness of the radiation source. The emitter individual identification method based on new intelligent feature set is proposed. Firstly, the received signals are transformed in time-domain, frequency-domain, time-frequency domain, polarization domain, and the features are extracted, secondly, the new intelligent feature set of each radiation source is constructed. In order to improve the computing speed and remove redundant information, master feature of the new intelligent feature set are extracted using principal component analysis, and then by choosing the optimal RBF kernel function with support vector machine (SVM), the individual identification is realized. By simulating and verifying the signals of three radiation sources, the radiation source can be characterized uniquely with the new intelligent feature set. Under low SNR, the radiation source individual can be identified quickly and effectively, by simulating the CW,BPSK,LFM signal respectively, more than 85% recognition rate can be reached in more than 3 db SNR, which verifies the validity and feasibility of the method.

To improve the accuracy and stability of the BDS/GPS combined positioning system, the high-degree cubature Kalman filter (CKF) is applied to the estimation of the positioning parameters, and the diagonalization of matrix (DM) transformation is used to replace the Cholesky decomposition in the standard high degree CKF. Through the DM transformation, the square root matrix of the covariance matrix has more accurate state statistics and retains the characteristics of the original feature space information, and improves the filtering accuracy. The covariance matrix of DM transformation does not require positive definite, so the stability of filtering is enhanced. The test results show that the filtering algorithm is effective and feasible for improving the positioning accuracy and stability.