Research Of Interference Source Identification Technology Under Strong Noise Background And Software Development

With the recognition of seabed interference signal widely used in the field of national defense construction, each country pays more and more attention to the development of this field. Submarine cable is an important channel for information exchange between countries,but it had occurred repeatedly that the cable was damaged and the information was stolen in recent years. In practical applications, most objects can be seen as objects with active signals, so we can obtain the specific properties of them by analyzing the interference signal in order to determine what kind of interference object or a signal it is. For this demand, the thesis studies the interference source recognition technology and designs the corresponding software platform. The technology includes signal preprocessing, extracting characteristic parameters and design for classification.And the software is mainly used to to observe all aspects of the processing result more convenient and intuitive.Signal preprocessing is used to reduce the noise for the collected original signal. Because of the complexity of the marine environment, noise intensity is very large, and the SNR of original signal is relatively low, referring to the existing wavelet threshold noise reduction methods, this thesis presents a noise-reduction method under strong noise background, namely improved semi-soft threshold noise reduction algorithm, and it is more desirable to achieve the noise reduction effect by simulation comparison.Extracting characteristics is the most critical aspect of the interference identification system. This thesis completes the extracting process from time domain, frequency domain and time-frequency domain, including normalized spectral bandwidth, power spectrum estimation, and presents a new characteristic parameter "Time Domain SAR", In addition, cycles Power Spectrum Estimation of different interference signals is studied.Classification is the final step, and the results show the performance of the system. Firstly, this thesis studies classification methods commonly used in object recognition systems; Secondly, analyzes requirements of this system for classification; Finally, designs the classification method based on data matching. The method matches and classifies the feature information according to established samples library.Software platform is based on the waterfall model and modular thinking, using software integration, C# and Matlab programming technology, multi-threading technology, firstly, designs the overall framework, control, status check and other functions in Visual Stdio 2010, then designs the algorithm of every aspect in Matlab software, finally builds a characteristic sample library in SQL Sever and realizes the generation of interference source. Software completes the noise reduction and feature extraction by calling dll generated from Matlab, displays graphics processed by various algorithms, matches data in the database, finally obtains the classification results.This thesis studies, designs and verifies various algorithms through analysis and simulation in software platform, has successfully distinguished certain signals. But it remains to be tested, made improvements and amendments in practical applications to monitor information in cable better.