CN101714969B - Method for demodulating railway communication signals - Google Patents

Abstract

The invention discloses a method for demodulating railway communication signals and comprises the steps of demodulating the signal sample by fast Fourier transform (FFT) after obtaining the signal to obtain a carrier frequency signal; performing quadrature modulation and filtration on the sampled signal by the obtained carrier frequency signal, re-sampling according to the number of the sampling points, performing FFT to the data after re-sampling, performing energy gravity center frequency spectrum rectification to the spectral line after algorithm and demodulating out a low frequency signal. The inventive method adopts FFT and ZFFT to demodulate low frequency signals and carrier frequency signals with accurate modulation, low algorithm amount and small time delay, and can adequately satisfy the requirements of current railway communication.

Description

A kind of method of demodulating railway communication signals

Technical field

The present invention relates to communication technical field, refer to especially a kind of method of demodulating railway communication signals.

Background technology

Source of Railway Communication and Signalling is subjected to the impact of Railway Environment, and the friction with track in the process of train operation produces the very large impulsive noise of instantaneous amplitude, coloured noise and power frequency interference easily, and signal spectrum also can be subject to the impact of nonlinear distortion.The recognition technology of Railway Signal Information is by the time domain of utilizing signal and frequency domain character, under the environment of very noisy low signal-to-noise ratio, utilizes the spectral interference technology for eliminating, extracts carrier frequency information and the low frequency information of signal.

Present signal message recognizer, the methods that adopt sequential detection more, frequency information by time domain zero crossing detection computations signal, this method is subject to the particularly interference of impulsive noise of noise easily, need the complicated analogue filter circuit of design, and identifying is longer, and the accuracy of identification of information is lower problem also.

Summary of the invention

In view of this, the invention reside in the method that a kind of demodulating railway communication signals is provided, when solving above-mentioned sequential detection, identifying is longer, and the accuracy of identification of information is lower problem also.

For addressing the above problem, the invention provides a kind of method of demodulating railway communication signals, comprising: behind the picked up signal, demodulate CF signal to described signal sampling and by fast Flourier FFT; By the described CF signal that obtains, the quantity of sampled point is sampled to the signal that obtains according to the prearranged multiple amplification is rear, after data quadrature modulation and filtering after the sampling, quantity according to described sampled point is sampled again, data after the described again sampling are done the FFT computing, and the spectral line after the computing done the energy barycenter Spectrum Correction, demodulate low frequency signal.

Preferably, during to described signal sampling and by FFT conversion demodulation CF signal, proportion extraction method DIF is to the sampled point computing.

Preferably, described predetermined multiplication factor is 4.

Preferably, described Source of Railway Communication and Signalling is amplitude frequency shift keying signal ASK or frequency shift keying fsk signal.

Method of the present invention adopts FFT and ZFFT demodulation low frequency, CF signal, and demodulation is accurate, and operand is low, and time delay is little, can fully satisfy the needs of present railway communication.

Description of drawings

Fig. 1 is principle schematic of the present invention;

Fig. 2 is the flow chart of demodulation low frequency signal of the present invention;

Fig. 3 is the design sketch of emulation.

Embodiment

For clearly demonstrating the scheme among the present invention, the below provides preferred embodiment and is described with reference to the accompanying drawings.

Referring to Fig. 1, the applicable signal of method of the present invention comprises ASK (amplitude frequency shift keying) signal that adopts in the railway communication, also can be used for FSK (frequency shift keying) signal.After at first signal, the noise that receives being carried out filtering, again carrier frequency and low frequency are identified.

Now ATP (driverless train protection) signal in the train communication carries out modeling as example to railway signal.

Table 1 is the signal message tabulation of ATP, carrier frequency wherein is the frequency of train uplink and downlink, to maximum speed restriction that should running section, 12 kinds of low frequencies represent respectively different rate signals to each frequency of low frequency part respectively, and lower Frequency represents higher train running speed.The duty ratio of low frequency code is i.e. 0 and 1 alternately appearance in 1: 1, and the identical duration is arranged.

Table 1

Signal is by the generation of multiplying each other of a square wave sequence and a sinusoidal carrier, can be expressed as a bipolarity rectangle sequence and a sinusoidal carrier multiplies each other.

$s\left(t\right)=\left[\underset{n}{\mathrm{\Σ}}{a}_{n}g(t-n{T}_s)\right]{\cos w}_{c}t$ Formula (1)

G (t) is to be T the duration herein _sRectangular pulse, and a _nBe-1 and 1 Serial No. that replaces, ω _cBe the angular frequency of carrier wave, n is constant, order

$c\left(t\right)=\underset{n}{\mathrm{\Σ}}{a}_{n}g(t-\mathrm{nT})$ Formula (2)

S (t)=c (t) cosw then _cT, wherein c (t) is the bipolarity square pulse.

Ambipolar square-wave signal is also deployable to be the Fourier series expression formula:

$v\left(t\right)=\frac{2V_s}{\mathrm{\π}}(\sin w_{0}t-\frac{1}{3}\sin 3w_{0}t+\frac{1}{5}\sin 5w_{0}t\·\·\·)$ Formula (3)

Wherein, ${\mathrm{<figure-callout\; id="0"\; label="w"\; filenames="H2009102235301E00013.png"\; state="\{\{state\}\}">w</figure-callout>}}_0=\frac{2\mathrm{\&pi;}}{T},$

The first-harmonic that is called this square wave signal, its cycle

Identical with the cycle of square wave itself.In the formula (2-3) all the other every all be higher harmonic components, their angular frequency is the integral multiple of first-harmonic angular frequency.

This shows that square wave can resolve into the sine wave of odd, wherein the amplitude of first-harmonic is the highest, and the amplitude of its each harmonic reduces successively.So the frequency spectrum of square wave also can resolve into each odd sine wave frequency spectrum and.

By to above-mentioned signal analysis because the signal that receives is square wave and sinusoidal wave modulation signal, therefore, separating timing, signal is carried out filtering after, demodulate wherein CF signal and low frequency signal.

For CF signal, can adopt fast Flourier (FFT) demodulation; When adopting FFT to separate timing, can be by decimation in frequency method DIF to the sampled point computing.

For signal after the carrier frequency identification, for low frequency signal, can adopt amplification fast Flourier ZFFT to demodulate signal frequency point, because the interval of Frequency is less, need higher frequency resolution, if improve the length that resolution just must improve data, amount of calculation also increases thereupon.Consider that it is unwanted that whole frequency range is adopted high-resolution, only need that the frequency range at signal message place is done local the amplification and get final product, and the data multiple modulation after will amplifying and resampling computing, can obtain correct result, and computation reduction.

Adopt the process of ZFFT demodulation at first the signal that receives to be sampled referring to Fig. 2, the signal after the sampling is carried out multiple modulation and filtering after, to the data resampling that amplifies, and calculate frequency values by FFT, the below describes demodulating process in detail.

To signal sampling the time, after at first enlarging with sampled point and according to predetermined multiplication factor D, to signal sampling;

For example: for signal x ₀(t), the highest frequency of signal is f ' _m, at time span t ' _pThe interior M point sampling of doing, the sampling period is T _s, calculate x with M point FFT ₀(t) frequency resolution is Δ f ', will observe x with doubly resolution ax/f=Δ f ' of D/D now ₀(t), the centre frequency of its frequency spectrum is f ₀=l ₀Δ f, bandwidth is B=M Δ f.If keep sample rate f this moment _sConstant, then the time record length of signal should be original D times, sampling number N=Dt ' _p/ T _s=DM.

To the signal shift frequency after the sampling, namely to x ₀(t) doing quadrature modulation gets:

$x\left(n\right)=x_0\left(n\right)\cos (2\mathrm{\&pi;<figure-callout\; id="0"\; label="n\; l"\; filenames="H2009102235301E00013.png"\; state="\{\{state\}\}">n</figure-callout>}{l}_{}{}_0/N)-j{x}_0\left(n\right)\sin (2\mathrm{\&pi;<figure-callout\; id="0"\; label="n\; l"\; filenames="H2009102235301E00013.png"\; state="\{\{state\}\}">n</figure-callout>}{l}_{}{}_0/N)$

$=x_0\left(n\right)e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="H2009102235301E00013.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;n}{l}_0/N}=x_0\left(n\right)W_N^{{\mathrm{nl}}_0}$ Formula (4-1)

4-1 gets by formula, x ₀(n) and the frequency spectrum of x (n) be:

$X_0\left(k\right)=\underset{n=0}{\overset{N-1}{\mathrm{\&Sigma;}}}{x}_0\left(n\right)W_N^{\mathrm{nk}}$ K=0,1,2 ... N-1 formula (4-2)

$X\left(k\right)=\underset{n=0}{\overset{N-1}{\mathrm{\&Sigma;}}}x\left(n\right)W_N^{\mathrm{nk}}=\underset{n=0}{\overset{N-1}{\mathrm{\&Sigma;}}}{x}_0\left(n\right)W_N^{n(k+l_0)}=X_0(k+l_0)$ Formula (4-3)

Can be seen by formula (4-2) and formula (4-3), after the quadrature modulation, the signal spectrum l that moves to left ₀Δ f, then f ₀Be moved to the zero-frequency place.The highest frequency of signal is M Δ f/2, is f ' to the sample rate of this section resampling therefore _s=M Δ f has reduced N/M doubly than the sample rate of original signal.Because time span is constant, the frequency resolution that calculates with M point FFT is identical with the frequency resolution that the N point calculates.

Sample through the signal after the quadrature modulation, the data after the sampling are got continuously at 1024 make FFT, signal spectrum is proofreaied and correct to be obtained the estimation of base band low frequency again.

Factor D=15 are extracted in order, and then the frequency resolution of FFT is

$\mathrm{\&Delta;f}=\frac{f_s}{N_{1}D}=\frac{40000}{1024*15}=2.6042\mathrm{HZ}$ Formula (4)

Obviously, the frequency resolution that is calculated by following formula can not satisfy specification requirement, for improving frequency detection accuracy, can realize by the Spectrum Correction technology, adopts the energy barycenter method here.

Energy centrobaric spectrum correcting method will be found out " center " of all FFT power spectral lines exactly, and the corresponding frequency in this center is exactly the estimated value of signal frequency.If sample frequency is f _s, to do spectrum and count and be N, the spectrum wire size of peak value is m in the main lobe, Y _iBe power spectrum i bar spectral line value, x ₀Be the main lobe center of gravity, the formula of energy correction method emending frequency is:

$x_0=\frac{\underset{i=-n}{\overset{n}{\mathrm{\&Sigma;}}}{Y}_i(m+i)f_s/N}{\underset{i=-n}{\overset{n}{\mathrm{\&Sigma;}}}{Y}_i}$ Formula (5)

And to asking again the FFT may be several on several spectral lines energy behind the list entries " windowing " that resamples, thereby utilize main spectral line and about each two totally 5 be the frequency that recoverable goes out signal.Obtain the amplitude maximum if suppose FFT spectral line X (k) at the k=m place, then its Frequency Estimation expression formula is

$f=\frac{\underset{k=m-2}{\overset{m+2}{\mathrm{\&Sigma;}}}k{\left|X\left(k\right)\right|}^2}{\underset{k=m-2}{\overset{m+2}{\mathrm{\&Sigma;}}}{\left|X\left(k\right)\right|}^2}\mathrm{\&Delta;f}$ Formula (6)

As shown in Figure 3, be that carrier frequency is 20000HZ, low frequency is the ATP signal of 16HZ, it is 16384 points that original samples is counted, multiplication factor D=4, and what mark with circle among the figure is signal spectrum after shift frequency resamples, can see, near the zero-frequency and-near the 15HZ two spectral lines are maximum.The point that marks with x is that these two spectral lines are proofreaied and correct Frequency point afterwards through energy barycenter.After the energy barycenter correction, the precision of extracting low frequency information further improves.

The present invention in use, according to the definition of present Source of Railway Communication and Signalling, signal carrier frequency highest frequency is 21KHZ, according to the nyquist sampling law, determines that sample rate should be more than the 45KHZ.Take the ATP signal as example, its different low frequencies differ that the interval is minimum to be 3HZ, if frequency resolution has satisfied the ATP signal, other signal also one satisfies surely to the requirement of frequency resolution.Low frequency differs 3HZ, and its frequency resolution necessarily can meet the demands less than 1.5HZ.The time domain brachymemma time is T=1/df=1/1.5s=0.667s so, and sample frequency is 48KHZ.The sampling number that needs is at least 32000.

In the situation of known ASK signal carrier frequency, need to carry out ZFFT to the signal spectrum around the carrier frequency and amplify analysis, be 150HZ for maximum low frequency, the frequency range that needs to amplify is f0 ± 250HZ, the frequency range that total like this needs amplify is 500HZ, being converted into digital angular frequency is 500/48000rad=0.0104rad, according to ω ₂-ω ₁≤ π/D, ω ₁〉=k π/D, ω ₂≤ (k+1) π/D can obtain the scope of D.In order to obtain the relation of noise resisting ability and multiplication factor D, through test, preferred D=4; For the signal beyond the low frequency, preferably D=16.

Before proofreading and correct and the simulation result after proofreading and correct as shown in Figure 3, that carrier frequency is 20KHZ, low frequency is the ATP signal of 16HZ, it is 16384 points that original samples is counted, multiplication factor D=4, what mark with circle among the figure is signal spectrum after shift frequency resamples, can see, near the zero-frequency and-near the 15HZ two spectral lines are maximum.The point that marks with x is that these two spectral lines are proofreaied and correct Frequency point afterwards through energy barycenter.Simulation results show, after the energy barycenter correction, the precision of extracting low frequency information further improves.

Method of the present invention adopts FFT and ZFFT demodulation low frequency, CF signal, and demodulation is accurate, and operand is low, and time delay is little, can fully satisfy the needs of present railway communication.

For the method for setting forth among each embodiment of the present invention, within the spirit and principles in the present invention all, any modification of doing, be equal to replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (3)

1. the method for a demodulating railway communication signals is characterized in that, comprising: behind the picked up signal, demodulate CF signal to described signal sampling and by fast Flourier FFT; By the described CF signal that obtains, the quantity of sampled point is sampled to the signal that obtains according to the prearranged multiple amplification is rear, after the modulation of the signal in orthogonal after the sampling and filtering, quantity according to described sampled point is sampled again, data after the described again sampling are done the FFT computing, and the spectral line after the computing done the energy barycenter Spectrum Correction, demodulate low frequency signal;

Described predetermined multiplication factor is 4.

2. method according to claim 1 is characterized in that, during to described signal sampling and by FFT conversion demodulation CF signal, proportion extraction method DIF carries out computing to sampled point.

3. method according to claim 1 is characterized in that, described Source of Railway Communication and Signalling is amplitude frequency shift keying signal ASK or frequency shift keying fsk signal.

Images