Research Of Location And Pattern Recognition On Distributed Optical Fiber Perimeter Security System

Fiber fence is a kind of system using optic-fiber as the sensing medium, canrealize a large range of real-time monitoring, which is getting more and moreattention in the field of security. Because of the advantages of anti-electromagneticinterference, corrosion resistance，stable chemical property, high sensitivity and so on,fiber sensors are especially suitable to be applied in the intelligent perimeter securitysystem.Location technology is the main research direction of the optical fiber perimetersecurity system. In this paper, a new algorithm based on "zero passage rate" isdesigned on the basis of thorough understanding the fiber optic signal, Selectingappropriate signal segment, not only reduce the calculation amount, but also greatlyenhance the positioning accuracy.In the aspect of pattern recognition, because of the high complexity ofcharacteristic vector extraction wavelet method, the paper put forward the methods ofusing all-phase filter groups to realize feature extraction, take advantages of all-phasefilter, making the algorithm highly portable and can be used on multiple environments,extending the value of engineering application.The detailed work is listed below:1) Apply all-phase filter into the security prevention system, enrich the technicalcontent, realize the tightly bind of academic research and engineeringapplication.2) According to the problem of location, a more quickly and accurately locationalgorithm based on "zero passage rate" is put forward on the basis of previousresearch.3) Use All-phase filter to realize the endpoint detection of signal, whichimproved time resolution to ms level, reduce the influence of undesired signalin the process, and improve the recognition accuracy.4) Realize the feature extraction using the all-phase filter bank, greatly reducedthe complexity relative to. Different from the fixed of wavelet analysis,All-phase filter bank is easy to realize the band decomposition of interestedfrequency domain, which has good adapting. 5) Understand in depth of the neural network structure and simplified thetraining method of neural network, make it more suitable for the engineeringapplication, reduce the programming work of system realize using Labview.6) Conducting simulation use field signal, proved the reliability of the algorithmin the paper.