CN-121559158-B - Electric power data acquisition method for submarine optical cable

CN121559158BCN 121559158 BCN121559158 BCN 121559158BCN-121559158-B

Abstract

The invention belongs to the field of data acquisition, in particular to a power data acquisition method for a submarine optical cable, which aims to solve the technical problems that the traditional filtering method is difficult to cope with complex noise and has poor data purification effect, and comprises the following steps of S1, obtaining a frequency domain input signal; S2, obtaining a prediction error, calculating a frequency domain gain by using the enhanced frequency-variant forgetting factor matrix, S3, performing cross-band correction on filter coefficients of all frequency points when updating correction is performed, applying attenuation weights to the correction values in a known interference harmonic frequency band, S4, obtaining a frequency domain of purified data by using the corrected filter coefficients, and obtaining time domain purified data by inverse fast Fourier transform. The invention can better extract the effective power data of the submarine optical cable under the strong noise background, and improves the signal-to-noise ratio and the reliability of the acquired data.

Inventors

LIU SHIJUN
XIE XIAOZHEN
LIU HUAHAN
JIA MENGTING
Jin Danyang
YE JING

Assignees

烽火海洋网络设备有限公司
烽华海洋工程设备有限责任公司

Dates

Publication Date: 20260505
Application Date: 20260123

Claims (10)

1. The electric power data acquisition method for the submarine optical cable is characterized by comprising the following steps of: S1, acquiring original power data, framing, acquiring an asymmetric hybrid window function according to spectral characteristic parameters of each data frame, and windowing and fast Fourier transforming each data frame by adopting the asymmetric hybrid window function to acquire a frequency domain input signal; S2, a state space prediction model is established for the filter coefficient of each frequency point to obtain a priori predicted value of each frequency point, and the predicted value of the frequency domain input signal is calculated according to the priori predicted value to obtain a predicted error; s3, triggering updating correction of the filter coefficients when the statistical distance between the statistical distribution of the prediction errors and the preset baseline noise model exceeds a preset threshold value, performing cross-band correction on the filter coefficients of all frequency points by using the frequency domain gain and the prediction errors when the updating correction is performed, and applying attenuation weights related to interference characteristics to the correction values in a known interference harmonic frequency band; s4, obtaining the frequency domain of the purified data by using the corrected filter coefficients, and obtaining the time domain purified data by inverse fast Fourier transform.
2. The method for collecting power data of an undersea optical fiber cable according to claim 1, wherein a power supply voltage signal or a current signal of the undersea optical fiber cable is collected to obtain a discrete digital signal sequence, the digital signal sequence is divided into a plurality of data frames with set lengths, and a plurality of overlapped sampling points are arranged between adjacent data frames.
3. The method for collecting power data for a submarine optical cable according to claim 1, wherein in S1, an asymmetric hybrid window function is obtained according to a spectral feature parameter of each data frame, and each data frame is windowed and fast fourier transformed by using the asymmetric hybrid window function, so as to obtain a frequency domain input signal, which includes: Calculating the spectral kurtosis of the data frame, calculating the total harmonic distortion degree by taking the spectral kurtosis as a transient characteristic index, and taking the total harmonic distortion degree as a harmonic pollution index; obtaining standard deviation parameter sigma of a Gaussian window function according to the transient characteristic index, and determining parameters of the Blackman-Harris window function according to the harmonic pollution index; And splicing the front half window of the Gaussian window function and the rear half window of the Blackman-Harris window function to obtain an asymmetric mixed window function.
4. A method for power data acquisition for submarine optical cable according to claim 3, wherein use is made of Calculating standard deviation of Gaussian window function , wherein, Is that Is set at the maximum value of (c), Is that Is set to be a minimum value of (c), Is the reference midpoint of the spectral kurtosis, Is a constant value, and is used for the treatment of the skin, At a maximum value And minimum value In the middle of the process, the intermediate part, Is the transient characteristic index of the data frame.
5. The method for collecting power data of an undersea optical cable according to claim 1, wherein in S2, a frequency-dependent forgetting factor matrix is calculated according to a signal-to-noise ratio of each frequency point and its neighboring frequency points, including: calculating the signal-to-noise ratio of frequency point k by using the current signal power and the estimated noise power ; By S-shaped function Calculating an initial forgetting factor of each frequency point, wherein, Is the forgetting factor of frequency bin k, At the minimum value of the forgetting factor, At the maximum value of the forgetting factor, For the signal-to-noise ratio reference to the demarcation value, Slope control parameters for the S-shaped function; And carrying out moving average filtering on the sequence formed by the initial forgetting factors of all the frequency points along a frequency axis to obtain diagonal elements of the frequency-dependent forgetting factor matrix.
6. The method for collecting power data for an undersea optical fiber cable according to claim 1, wherein in S3, data is collected in advance in a period of no signal or only background noise, a gaussian mixture model representing the amplitude distribution of the noise signal is established, and the gaussian mixture model is used as a baseline noise model.
7. The method for collecting power data for an undersea optical cable according to claim 6, wherein in S3, a set of prediction errors e (k, n) of all frequency points of a current data frame is collected, a probability density function of the set is calculated to form a real-time error distribution, and a distance between the real-time error distribution and a preset baseline noise model is calculated by using KL divergence.
8. The method for power data collection for a submarine optical cable according to any one of claims 1-7, wherein in S3, triggering an update correction of the filter coefficients when the statistical distance between the statistical distribution of the prediction error and the preset baseline noise model exceeds a preset threshold value comprises: the baseline noise model is a Gaussian distribution model Variance of The noise power average value of a plurality of continuous data frames is determined under the initial state without electric load through an acquisition system; At each moment, extracting the prediction errors of a plurality of nearest sampling points, and calculating a probability density function P (e) of the prediction errors; Calculation of P (e) and baseline noise model using KL divergence Statistical distance between ; When (when) When the value is larger than a preset threshold value, updating and correcting the filter coefficient are triggered.
9. The power data collection method for a submarine optical cable according to claim 8, wherein performing update correction by using a frequency domain gain and a prediction error, performing cross-band correction on filter coefficients of all frequency points, and applying attenuation weights related to interference characteristics to the correction amounts in a known interference harmonic band, comprises: acquiring tri-diagonal coupling matrix representing correlation between frequency points The main diagonal elements are 1, and the secondary diagonal elements are adjacent frequency point energy correlation coefficients obtained according to historical data statistics; Initial correction vector to be calculated using frequency domain gain and prediction error And the tri-diagonal coupling matrix Multiplying to obtain coupling correction ; Aiming at known power frequency harmonic interference, an attenuation weight vector is constructed Setting a weight value smaller than 1 at the position corresponding to the harmonic frequency point and a plurality of frequency points on the left and right sides of the harmonic frequency point, and setting the weight value to be 1 at the rest frequency points; Will couple the correction amount And decaying weight vector Element-wise multiplication is performed to obtain the correction amount applied to the filter coefficient.
10. The method for collecting power data for an undersea optical fiber cable according to claim 1, wherein in S4, the corrected filter coefficients are used to multiply the frequency domain input signal to obtain a posterior prediction error, which is the frequency domain of the cleaned signal.

Description

Electric power data acquisition method for submarine optical cable Technical Field The invention belongs to the field of data acquisition, and particularly relates to a power data acquisition method for an undersea optical cable. Background The submarine optical cable is an artery for global information communication, and the working state of key equipment such as a repeater can be effectively monitored and potential faults can be timely early-warned through real-time acquisition and analysis of electric power data of the submarine optical cable. The submarine environment is complex, the power data is often influenced by strong electromagnetic interference, ocean environment noise and the like in the transmission process, the signal to noise ratio is extremely low, and useful signals are seriously submerged. In order to extract an effective power characteristic signal from a strong noise background, conventional methods such as Least Mean Square (LMS) and Recursive Least Square (RLS) in a time domain, and frequency domain adaptive filtering are adopted, but the effect is limited although noise can be suppressed to a certain extent. Specifically, the traditional method adopts a fixed window function to carry out framing treatment, cannot adapt to the change of signal spectrum characteristics, and is easy to cause serious spectrum leakage and frequency resolution degradation. The traditional frequency domain RLS (FRLS) algorithm cannot be matched with the characteristic that the energy distribution of noise on different frequency bands is uneven, unnecessary calculation overhead and coefficient imbalance risks are caused when the signal-to-noise ratio is high, and the internal relevance of broadband interference or harmonic interference among the frequency bands is ignored. For the known strong periodic interference of power frequency harmonic waves and the like, a targeted suppression strategy matched with the interference characteristics is lacked, and obvious interference components still remain in the purified data, so that the accuracy of subsequent state evaluation is affected. Disclosure of Invention The invention provides a power data acquisition method for a submarine optical cable, which aims to solve the technical problems that the traditional filtering method is difficult to cope with complex noise and has poor data purification effect. A method of power data acquisition for an undersea optical fiber cable, comprising the steps of: S1, acquiring original power data, framing, acquiring an asymmetric hybrid window function according to spectral characteristic parameters of each data frame, and windowing and fast Fourier transforming each data frame by adopting the asymmetric hybrid window function to acquire a frequency domain input signal; S2, a state space prediction model is established for the filter coefficient of each frequency point to obtain a priori predicted value of each frequency point, and the predicted value of the frequency domain input signal is calculated according to the priori predicted value to obtain a predicted error; s3, triggering updating correction of the filter coefficients when the statistical distance between the statistical distribution of the prediction errors and the preset baseline noise model exceeds a preset threshold value, performing cross-band correction on the filter coefficients of all frequency points by using the frequency domain gain and the prediction errors when the updating correction is performed, and applying attenuation weights related to interference characteristics to the correction values in a known interference harmonic frequency band; s4, obtaining the frequency domain of the purified data by using the corrected filter coefficients, and obtaining the time domain purified data by inverse fast Fourier transform. Further, a power supply voltage signal or a current signal of the submarine optical cable is collected to obtain a discrete digital signal sequence, the digital signal sequence is divided into a plurality of data frames with set lengths, and a plurality of overlapped sampling points are arranged between adjacent data frames. Further, in S1, an asymmetric hybrid window function is obtained according to the spectral feature parameter of each data frame, and windowing and fast fourier transformation are performed on each data frame by using the asymmetric hybrid window function, so as to obtain a frequency domain input signal, which includes: Calculating the spectral kurtosis of the data frame, calculating the total harmonic distortion degree by taking the spectral kurtosis as a transient characteristic index, and taking the total harmonic distortion degree as a harmonic pollution index; obtaining standard deviation parameter sigma of a Gaussian window function according to the transient characteristic index, and determining parameters of the Blackman-Harris window function according to the harmonic pollution index; And splicing the front half