Study On Key Technologies For Radar Target Detection And Tracking

The improvement of marine radar performance is of great practical significance for maritime safety. This dissertation focuses on radar target detection and tracking. The main contributions of the dissertation are summarized as follows:1. Statistical model of actual sea clutter data is studied. The independence of sea clutter is studied by correlation coefficient test method. And the relevance of sea clutter is studied by using auto-correlation function method. The prior information provides the important theoretical basis for subsequent signal processing.2. The offset time control (OTC) algorithm based on sea clutter prediction is proposed in view of the strong sea clutter intensity may cause signal distortion and blockage. Sea clutter model is established by using linear prediction and neural network prediction method, respectively. The radar sensitivity is controlled by the sea clutter prediction error. Experiment results are provided to verify the effectiveness of the proposed method.3. The method of target detection based on neural network ensembles is studied to reduce the effect of training sample selection on target detection. According to the performance of sub-network in validation set, fuzzy C-means clustering method is used to give the corresponding weights and improve the target detection ability. According to the real-time dynamic change of sea clutter, the parameters of the linear prediction model are updated online, and the sea clutter model is updated adaptively, which improves the target detection ability. The validity and reliability of the proposed method are verified by experimental research.4. Aiming at the difference of fractal characteristics between target and sea clutter, an improved fractal method for detecting small floating target is proposed. The de-trending fluctuation analysis method is used to calculate the sea clutter fractal dimension. In the slope-intercept decision space, small target detection result is getted by K-means clustering analysis method. Experimental results show that the improved algorithm improves the target detection probability.5. Aiming at the unknown statistical information of noise and continuous characteristics of target measurement, the smooth spline fitting filter algorithm is proposed. The smoothness and continuity of smoothing spline are used to constrain target trajectory. The least squares approximation value of measurements is compromised with bending and smoothness of spline. Meanwhile, the true states of target are reflected. The efficiency and reliability of the proposed approaches are illustrated from simulation results.