CN104573321A - Recognition and processing method of bad data of dissolved gas in transformer oil - Google Patents

Abstract

The invention discloses a recognition and processing method of bad data of dissolved gas in transformer oil. The bad data mainly comprises dead data and abnormal jump data; the recognition and processing method of the bad data comprises the following steps: (a) recognizing and processing the dead data; (b) recognizing the abnormal jump data, wherein a condition set for data jump is that the change of single-point data is large, the data after jump is returned to a normal level, and the calculation of the data jump is mainly to confirm a data change amplitude threshold value; the processing of the abnormal jump data is to perform filtering processing on a gas concentration value based on a wavelet denoising principle in signal processing, and comprises the following steps: (1) selecting a wavelet and determining decomposition levels, and then performing wavelet decomposition calculation on a signal; (2) selecting a proper threshold value for a high-frequency coefficient under each decomposition size to perform soft threshold value quantitative treatment; (3) denoising by using a wavelet multi-size decomposition denoising principle aiming at the jump data, namely noise-containing data.

Description

A kind of identification of Gases Dissolved in Transformer Oil bad data and disposal route

Technical field

What the present invention relates to is a kind of identification and disposal route of filling transformer insulating oil gas concentration bad data, is mainly used in, in transformer oil chromatographic gas on-line monitoring system, belonging to transformer gas fault detect and warning technology field.

Background technology

In existing transformer gas fault detect and warning technology on-line monitoring system, cause due to the measuring error of sensor some data not meet the data of the Changing Pattern of electrical equipment own, or do not meet the data of the precision of surveying instrument own and scope; These data comprise the bad data such as doomed dead certificate, abnormal saltus step data; As these bad datas being used for the analytical calculation of transformer gas fault detect and warning system, the correctness of testing result will be had influence on, therefore needing to be processed.

Summary of the invention

The object of the invention is to the deficiency overcoming prior art existence, and provide a kind of method simple and reliable, the correctness of testing result can be improved, ensure identification and the disposal route of the Gases Dissolved in Transformer Oil bad data that power equipment safety runs.

The object of the invention is to have come by following technical solution, the identification of described Gases Dissolved in Transformer Oil bad data and disposal route, described bad data mainly comprises doomed dead certificate and abnormal saltus step data, wherein said doomed dead certificate refers in time series, normally should the data of time to time change, do not change within a period of time, these point data are called doomed dead certificate, determine doomed dead according to time, closely related with the data characteristic of measuring object;

Described abnormal saltus step data refer in time series, at T _i-1time data is owing to being interfered, and numerical value produces jumping characteristic change, at T _imoment falls after rise, and the amplitude of variation of numerical value has surmounted T _i-1fluctuating range before moment; It is characterized in that identification and the disposal route of described bad data comprise:

A) identification of doomed dead certificate and process: when detect doomed dead according to time, illustrate that sensor exists measurement problem, need artificial treatment;

B) identification of abnormal saltus step data and process, wherein the identification of abnormal saltus step data is: the condition of data jump setting is that one point data changes greatly, after saltus step, data can revert to normal level, for the calculating of data jump, mainly confirm data variation amplitude threshold;

The process of described abnormal saltus step data is: based on Wavelet Denoising Method principle in signal transacting, carry out filtering process to gas concentration value, its step is as follows:

(1) select a small echo and determine decompose level, then wavelet decomposition calculating is carried out to signal;

(2) a suitable threshold value is selected to carry out soft-threshold quantification treatment to the high frequency coefficient under each decomposition scale;

(3) according to the high frequency coefficient of the bottom low frequency coefficient of wavelet decomposition and each layer after quantification treatment, the reconstruct of one-dimensional signal is carried out, the estimated value of the original signal be restored.

For saltus step data and noisy data, use Multiscale Wavelet Decomposition denoising away from carrying out denoising.

In transformer oil chromatographic gas on-line monitoring system of the present invention, following data think doomed dead certificate:

(1) in master system, the data markers of reception is constant, and numerical value is also constant, and this kind of point is doomed dead certificate;

(2) hydrogen (H2), methane (CH4), ethane (C2H6), ethene (C2H4), carbon monoxide (CO), carbon dioxide (CO2) six kinds of gases and total hydrocarbon, wherein certain class gas values continues to be the point of null value, is doomed dead certificate;

(3) numerical value of data point is negative value;

In the identification of abnormal saltus step data, for the fluctuation of transformer oil chromatographic online monitoring data, for finding out rule, by N number of time series data, the consequent preceding paragraph that subtracts makes difference processing, forms { (V _i+1-V _i) data sequence, in fact this data sequence reflects the amplitude of data fluctuations, approximate Normal Distribution rule.According to 3 σ principles statistically, 3 σ upper control limit UCL and 3 σ lower control limit LCL are made to these fluctuation amplitude.Computing method:

Note mean value is C, and standard deviation is σ

$\begin{array}{cc}C=\frac{1}{n-1}\underset{i=1}{\overset{n-1}{\mathrm{\Σ}}}(V_{i+1}-V_i)& \mathrm{\σ}=\sqrt{\underset{i=1}{\overset{n-1}{\mathrm{\Σ}}}\frac{{(V_{i+1}-V_i-V)}^2}{n-2}}\end{array}$

UCL﹦C+3σ；LCL﹦C－3σ

As long as meet design conditions:

IF

(V _i+1－V _i)≥UCL OR (V _i+1－V _i)≤LCL

(V _i－V _i-1)≥UCL OR (V _i－V _i-1)≤LCL

│(V _i+1－V _i)+(V _i－V _i-1)│≤σ

Then

T _i﹦ jumppoint trip point.

It is simple and reliable that the present invention has method, can improve the correctness of testing result, ensures the features such as power equipment safety operation.

Accompanying drawing explanation

Fig. 1 is abnormal saltus step data variation schematic diagram of the present invention.

Fig. 2 is the trip point schematic diagram of abnormal saltus step data of the present invention.

Fig. 3 is little Bo Valve value denoise algorithm FB(flow block) belonging to the present invention.

Fig. 4 is noisy data and curves figure of the present invention.

Fig. 5 is denoising data and curves figure of the present invention.

Embodiment

Below in conjunction with accompanying drawing, the present invention will be described in detail: the identification of Gases Dissolved in Transformer Oil bad data of the present invention and disposal route, described bad data mainly comprises doomed dead certificate and abnormal saltus step data, wherein said doomed dead certificate refers in time series, normally should the data of time to time change, do not change within a period of time, these point data are called doomed dead certificate, determine doomed dead according to time, closely related with the data characteristic of measuring object;

In transformer oil chromatographic gas on-line monitoring system of the present invention, following data think doomed dead certificate:

(1) in master system, the data markers of reception is constant, and numerical value is also constant, and this kind of point is doomed dead certificate;

(2) hydrogen (H2), methane (CH4), ethane (C2H6), ethene (C2H4), carbon monoxide (CO), carbon dioxide (CO2) six kinds of gases and total hydrocarbon, wherein certain class gas values continues to be the point of null value, is doomed dead certificate;

(3) numerical value of data point is negative value;

Described abnormal saltus step data refer in time series, at T _i-1time data is owing to being interfered, and numerical value produces jumping characteristic change, at T _imoment falls after rise, and the amplitude of variation of numerical value has surmounted T _i-1fluctuating range before moment; As shown in Figure 1.

Identification and the disposal route of bad data of the present invention comprise:

A) identification of doomed dead certificate and process: when detect doomed dead according to time, illustrate that sensor exists measurement problem, need artificial treatment;

B) identification of abnormal saltus step data and process, wherein the identification of abnormal saltus step data is: the condition of data jump setting is that one point data changes greatly, and after saltus step, data can revert to normal level, T as shown in Figure 2 _kpoint; For the calculating of data jump, mainly confirm data variation amplitude threshold;

In the identification of abnormal saltus step data, for the fluctuation of transformer oil chromatographic online monitoring data, for finding out rule, by N number of time series data, the consequent preceding paragraph that subtracts makes difference processing, forms { (V _i+1-V _i) data sequence, in fact this data sequence reflects the amplitude of data fluctuations, approximate Normal Distribution rule; According to 3 σ principles statistically, 3 σ upper control limit UCL and 3 σ lower control limit LCL are made to these fluctuation amplitude; Computing method:

Note mean value is C, and standard deviation is σ

$\begin{array}{cc}C=\frac{1}{n-1}\underset{i=1}{\overset{n-1}{\mathrm{\Σ}}}(V_{i+1}-V_i)& \mathrm{\σ}=\sqrt{\underset{i=1}{\overset{n-1}{\mathrm{\Σ}}}\frac{{(V_{i+1}-V_i-V)}^2}{n-2}}\end{array}$

UCL﹦C+3σ；LCL﹦C－3σ

As long as meet design conditions:

IF

(V _i+1－V _i)≥UCL OR (V _i+1－V _i)≤LCL

(V _i－V _i-1)≥UCL OR (V _i－V _i-1)≤LCL

│(V _i+1－V _i)+(V _i－V _i-1)│≤σ

Then

T _i﹦ jumppoint trip point.

The process of described abnormal saltus step data is: based on Wavelet Denoising Method principle in signal transacting, carry out filtering process to gas concentration value, its step is as follows:

(1) select a small echo and determine decompose level, then wavelet decomposition calculating is carried out to signal;

(2) a suitable threshold value is selected to carry out soft-threshold quantification treatment to the high frequency coefficient under each decomposition scale;

(3) according to the high frequency coefficient of the bottom low frequency coefficient of wavelet decomposition and each layer after quantification treatment, the reconstruct of one-dimensional signal is carried out, the estimated value of the original signal be restored; Flow process as shown in Figure 3;

For saltus step data, i.e. noisy data, as shown in Figure 4, using Multiscale Wavelet Decomposition denoising away from carrying out denoising, the results are shown in Figure shown in 5.

As can be seen from Figure 4, due to the impact of surveying instrument and environmental noise, measurement data engrail is made

As can be seen from Figure 5, the data after denoising are smoother, and monotone increasing, meet the objective law of gas build, denoising effect is better.

Claims (2)

1. the identification of a Gases Dissolved in Transformer Oil bad data and disposal route, described bad data mainly comprises doomed dead certificate and abnormal saltus step data, wherein said doomed dead certificate refers in time series, normally should the data of time to time change, do not change within a period of time, these point data are called doomed dead certificate, determine doomed dead according to time, closely related with the data characteristic of measuring object; Described abnormal saltus step data refer in time series, at T _i-1time data is owing to being interfered, and numerical value produces jumping characteristic change, at T _imoment falls after rise, and the amplitude of variation of numerical value has surmounted T _i-1fluctuating range before moment; It is characterized in that identification and the disposal route of described bad data comprise:

A) identification of doomed dead certificate and process: when detect doomed dead according to time, illustrate that sensor exists measurement problem, need artificial treatment;

B) identification of abnormal saltus step data and process, wherein the identification of abnormal saltus step data is: the condition of data jump setting is that one point data changes greatly, after saltus step, data can revert to normal level, for the calculating of data jump, mainly confirm data variation amplitude threshold;

The process of described abnormal saltus step data is: based on Wavelet Denoising Method principle in signal transacting, carry out filtering process to gas concentration value, its step is as follows:

(1) select a small echo and determine decompose level, then wavelet decomposition calculating is carried out to signal;

(2) a suitable threshold value is selected to carry out soft-threshold quantification treatment to the high frequency coefficient under each decomposition scale;

(3) according to the high frequency coefficient of the bottom low frequency coefficient of wavelet decomposition and each layer after quantification treatment, the reconstruct of one-dimensional signal is carried out, the estimated value of the original signal be restored;

For saltus step data and noisy data, use Multiscale Wavelet Decomposition denoising away from carrying out denoising.

2. the identification of Gases Dissolved in Transformer Oil bad data according to claim 1 and disposal route, is characterized in that in described transformer oil chromatographic gas on-line monitoring system, following data think doomed dead certificate:

(1) in master system, the data markers of reception is constant, and numerical value is also constant, and this kind of point is doomed dead certificate;

(2) hydrogen (H2), methane (CH4), ethane (C2H6), ethene (C2H4), carbon monoxide (CO), carbon dioxide (CO2) six kinds of gases and total hydrocarbon, wherein certain class gas values continues to be the point of null value, is doomed dead certificate;

(3) numerical value of data point is negative value;

In the identification of abnormal saltus step data, for the fluctuation of transformer oil chromatographic online monitoring data, for finding out rule, by N number of time series data, the consequent preceding paragraph that subtracts makes difference processing, forms { (V _i+1-V _i) data sequence, in fact this data sequence reflects the amplitude of data fluctuations, approximate Normal Distribution rule.According to 3 σ principles statistically, 3 σ upper control limit UCL and 3 σ lower control limit LCL are made to these fluctuation amplitude.Computing method:

Note mean value is C, and standard deviation is σ

$C=\frac{1}{n-1}\underset{i=1}{\overset{n-1}{\mathrm{\Σ}}}(V_{i+1}-V_i)$ $\mathrm{\σ}=\sqrt{\underset{i=1}{\overset{n-1}{\mathrm{\Σ}}}\frac{{(V_{i+1}-V_i-C)}^2}{n-2}}$

UCL﹦C+3σ；LCL﹦C－3σ

As long as meet design conditions:

IF

(V _i+1－V _i)≥UCL OR(V _i+1－V _i)≤LCL

(V _i－V _i-1)≥UCL OR(V _i－V _i-1)≤LCL

│(V _i+1－V _i)+(V _i－V _i-1)│≤σ

Then

T _i﹦ jumppoint trip point.