Study Of Target Data Processing For High Frequency Radar

High frequency (HF) radar is an effective means to execute large-scale and long-range strategic surveillance due to its over-the-horizon alert ability and excellent detection ability to the low-observability targets, so the world pays more attention to it. As the key section in radar system, target data processing is used to accomplish multiple targets tracking realtimely, estimating the motion state parameters, data association and trace processing, etc, which can suppress effectively the false alarms caused by target-like disturbances, improve the continuous tracking ability of target sequence, and obtain the better situation display. Target data processing in HF radar is carried on under the bad conditions of low space resolution, low precision of position measurement, uncertain statistical characteristics of measurement error, low data rate, unknown target motion mode and target may be lost. As the observation environment gets more complex, target's maneuverability, diversity, density and low-observability are enhanced correspondingly. Therefore, it is necessary to do some research on adaptive and reliable target data processing algorithm. In this paper, the algorithm and technique are studied detailed to improve the abilities of making target trace processed stably, formed rapidly and effectively and maintained continuously in target data processing system of HF radar.Considering the special data processing environment in HF radar, a method of tracking system modeling driven by target measuring data is presented, which overcomes the limitations in the ideal prior knowledge models. The target state transformation model is built in radar coordinate through improved self-learning method. The statistical parameters of measuring noises are estimated adaptively by dual median smooth algorithm, and more accurate estimations of target state are acquired by limited memory adaptive filter algorithm. Then the effects of tracking filter are evaluated. In the preprocessing of measuring data, Doppler velocity which estimation is more precise is used to correct the false radial range measurement. A realtime outlier discrimination criterion based on fuzzy set theory is presented, which can verify the rationality and improve the veracity of those data to avoid the filter diverging. Through the results of simulation and processing of actual measuring data, it is shown that the improved tracking filter has less dependence on the prior knowledge information, can reflect the tracking system state adaptively and has the stable and credible performance.The spatial trace matching principle is introduced to restrict the smoothness of target movement trace. And the measuring precision of radical range is improved according the matching principle between the radical velocity and range. This method is not restricted by the radar system style, not influenced by the change of working parameters, the difference of data rate and the lost of measuring data. The target trace can be formed rapidly and effectively only using this method. Combining with the former improved tracking filter algorithm, the two-step tracking procedure is formed. Firstly, the rationality of target movement trace is ensured by spatial trace matching processing and the noise characteristics are estimated. Then the self-learning target motion model is built by the results of former step, and the limited memory self-learning adaptive filer is executed. The spatial trace matching principle can effectively control the accuracy of tracking model, weaken the jitters of the trace and preserve the stability of the data processing system. This two-step tracking algorithm can be used in the measurement point fusion processing of multiple radar networks. In the synchronous fusion and the in-sequence measurement asynchronous fusion, sequential processing is merely required during the second step of the time domain filter. While in the out-of-sequence measurement asynchronous fusion, target state estimation can be carried out through the spatial trace matching processing singly.For the reason of target temporary lost, the break-continue method in trace processing under the condition of the lack of measuring data is studied to protect the continuous tracking of the same target. In the environment of multiple targets and clutters, the data could be also associated even the target measurement appears discontinuously. The improved logic law is used in the track initiation. After the trace is formed stably, it is not canceled immediately when the target is lost temporarily, but adaptive blind tracking method is used to lock its motion pattern to keep tracking with opportune batches until the target redetection. Some problems involved in blind tracking, such as multi-step state prediction, times setting, association bound adjustment, trace smoothing after the target is recaptured and the trace termination logics, are studied in detail. To resolve the problem of trace interruption caused by traces intersection, an algorithm called multiple hypothesis cheap joint probability data association is proposed to allocate the data.The combination of tracking and detection can prevent the detected target from missing easily, suppress effectively the false alarms caused by the non-background noise, and improve the performance of target data continuous processing. Signal detection is firstly processed during the spectrum data of a single batch. Those targets with strong echoes can be detected in the basic signal detection. While for those weak target echoes, the zone of signal detection are adjusted according to the state prediction information and sheltered tracking signal detection is processed with low threshold. Considering the multiple target situations which appear frequently in the lower Doppler velocity region during the sea surveillance, a whole-peak-outlier- elimination criterion is proposed to depress the overvalued detection threshold for the sake of avoiding the weaken target is shielded. Target detection in tracking is applied for confirming the target by the temporary trace quality parameter, identifying the false trace and forming the clutter map.The target data processing system in HF Radar is designed and realized, which has been applied in the real project. This system is divided into several parts such as system parameter setting, target motion design, spectrum data generation, target detection and tracking and state display, etc. The processing results of experiment data and simulation of various target motion show that the synthesis target data processing methods proposed in this dissertation have perfect trace processing performance with continuity, stability, celerity and validity.