Signal identification method of optical fiber perimeter defense system based on improved FB feature and GRNN network

An optical fiber perimeter, signal recognition technology, applied in the field of signal recognition, can solve the problems of few recognition types, too ideal recognition conditions, complex training process, etc., to meet real-time and accuracy, real-time efficient signal recognition, and simple model training. Effect

Pending Publication Date: 2022-03-11
SHANGHAI UNIV
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Some methods have a higher recognition rate but fewer recognition types, some methods do not consider the impact of environmental noise on the signal, the recognition conditions are too ideal, and some training processes are complicated, the recognition time is long, and the real-time performance is poor

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Signal identification method of optical fiber perimeter defense system based on improved FB feature and GRNN network
  • Signal identification method of optical fiber perimeter defense system based on improved FB feature and GRNN network
  • Signal identification method of optical fiber perimeter defense system based on improved FB feature and GRNN network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0035] In this instance, see figure 1 , a signal recognition method of an optical fiber perimeter defense system based on the improved FB feature and GRNN network, including the following steps:

[0036] Step 1, collecting several output signals of different types of sensing events of the fiber optic perimeter defense system in advance;

[0037] Step 2, using an endpoint detection algorithm to obtain effective signal segments of the collected signal;

[0038] Step 3, extract the improved FB feature for each signal segment, as a training sample, and use its signal type as the corresponding output label;

[0039] Step 4, train the GRNN through the training samples and their labels, adjust the parameters, and generate the optimal model;

[0040] Step 5, also perform endpoint detection on the newly collected signal to be identified, obtain its effective signal segment, and extract the improved FB feature of the signal segment as a test sample;

[0041] Step 6, use the generated...

Embodiment 2

[0044] This embodiment is basically the same as Embodiment 1, especially in that:

[0045] In this example, see figure 2 , the extraction of improved FB features described in step 3 needs to be achieved through the following steps:

[0046] 1) Normalize the signal fragments to eliminate the amplitude difference of different signals;

[0047] 2) Windowing the normalized signal segment frame by frame, so that both ends of each frame signal are attenuated to zero;

[0048] 3) Calculate the power spectrum of the signal segment;

[0049] 4) multiply the power spectrum of the signal segment with the band-pass filter bank formed by 26 triangular filters under the Mel scale, to obtain the signal power under the Mel filter bank;

[0050] 5) Calculate the root mean square of the signal power under each filter to obtain the effective value of the signal power, and form a one-dimensional vector containing 26 elements, which is the improved FB feature.

[0051] In this embodiment, by ...

Embodiment 3

[0053] This embodiment is basically the same as Embodiment 1, especially in that:

[0054] In this embodiment, in step 1, simulate and collect the output signals corresponding to the sensing events of no disturbance, shaking, knocking and running of the fiber optic perimeter defense system under three weather conditions: sunny day, rainy day and windy day.

[0055] In this embodiment, the output signals corresponding to the sensing events under the three weather conditions of sunny day, rainy day and windy day are collected as the original signal of the training data, so as to realize real-time and efficient signal recognition under noise interference.

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a signal identification method of an optical fiber perimeter defense system based on an improved FB feature and a GRNN network. The method comprises the following steps: acquiring a plurality of output signals of different types of sensing events in advance; obtaining effective signal segments of the signal by using an endpoint detection algorithm; extracting the improved FB feature of each signal segment as a training sample, and taking the signal type as a corresponding output label; training the GRNN through the training sample to generate an optimal model; performing end point detection on the to-be-identified signal to obtain an effective signal segment, and extracting an improved FB feature of the signal segment as a test sample; and identifying a test sample by using the generated GRNN optimal model. FB features are improved by adding effective value information of signal power under a Mel filter bank, so that differences of different types of signals are highlighted; the GRNN is used for identifying signals, parameters needing to be adjusted are few, and the model is easy to train. The method is high in signal recognition rate and good in real-time performance, and is expected to meet the actual requirements of an optical fiber perimeter defense system.

Description

technical field [0001] The present invention relates to a signal recognition method, in particular to a signal recognition method of an optical fiber perimeter defense system based on improved FB (Filter Bank, filter bank) features and GRNN (Generalized Regression Neural Network, generalized regression neural network) network . Background technique [0002] Due to the advantages of long-distance continuous monitoring and anti-electromagnetic interference, the fiber optic perimeter defense system has been rapidly developed in both national defense and civilian fields. The recognition rate and recognition time of detection signals are the main parameters used to evaluate the performance of the fiber optic perimeter defense system. In order to improve the accuracy of signal recognition and shorten the recognition time, relevant scholars have proposed a variety of signal recognition algorithms. At present, the most common is the signal recognition method that combines feature ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/00G06N3/04G06N3/08G08B13/24G08B21/18G10L21/0208
CPCG06N3/08G08B21/18G08B13/2497G10L21/0208G06N3/044G06F2218/02G06F2218/08
Inventor 方捻陆海南王陆唐
Owner SHANGHAI UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap