CN101527585B - Device for achieving electric power system data self-adapting compression and method thereof - Google Patents

Abstract

The invention relates to a device for achieving electric power system data self-adapting compression and a method thereof, wherein, the device comprises an equiphase sampling unit and a compression unit, analog signals is connected with the input end of the equiphase sampling unit, and the output end of the equiphase sampling unit is connected the compression unit; and the method comprises the following steps: 1. a compressor input sequence is obtained by conducting synchronous equiphase sampling on the input signals; 2. and then the compressor input sequence is subjeced to equiphase point sequence compression. The Signal-to-Noise by adopting the equiphase sampling method of the invention is obviously higher than that adopting uniformly-spaced sampling, and with the increasing of the compression ratio, the Signal-to-Noise adopting uniformly-spaced sampling is minished quickly, but the Signal-to-Noise adopting the equiphase sampling method of the invention is minished relatively slowly, thereby the equiphase sampling data can realize big compression ratio and effectively and stably improve the communication bandwidth of the electric power communications system.

Description

A kind of device and method of realizing the electric power system data self-adapting compression

[technical field]

The present invention relates to the data sampling and the data compression technique of electric line communication system, relate in particular to a kind of device and method of realizing the electric power system data self-adapting compression.

[background technology]

Along with the raising of Automation of Electric Systems degree, supervisory information system, Distributed Control System constantly perfect, the storage of magnanimity process data more and more receives people's attention.The information that is richly stored with in these data, it is for analyzing the operation of power networks state, providing control and optimisation strategy, failure diagnosis and Knowledge Discovery and data mining significant.But because data volume is huge, it is unpractical that data are got off with the form long preservation of prototype, and therefore the actual data compression method of the suitable engineering of research reduces the active demand that the redundancy that exists in the mass data has become power industry.

Usually, when being handled, adopts power network signal equal interval sampling method (claiming the asynchronous-sampling method again), because mains frequency is by the common decision of all power plants and network load in the electric power system; Under effects such as power disturbance and equipment fault; The frequency of actual electric network signal usually can fluctuate near rated frequency, and this can cause between the circulation of equal interval sampling periodic data the coupling with the circulation internal information, thereby causes that numerical fluctuations sharply increases between circulation; In addition; Equal interval sampling caused non-integer-period sampledly makes that also numerical values recited appears periodically between circulation, the raising of the multiple that this will the restricting data compression, thereby influence is to the lifting of communication bandwidth.

[summary of the invention]

For solve exist in the prior art under effect such as power disturbance and equipment fault; The frequency of actual electric network signal can fluctuate near rated frequency usually; Can cause between equal interval sampling periodic data circulation coupling with the circulation internal information; Thereby cause that numerical fluctuations sharply increases between circulation, equal interval sampling caused non-integer-period sampledly makes that also numerical values recited appears periodically between circulation, the raising of the multiple that this will the restricting data compression; Thereby influence the invention provides a kind of device and method of realizing the electric power system data self-adapting compression to these technical problems of lifting of communication bandwidth.

The present invention solves the technical scheme that the prior art problem adopted: a kind of device of realizing the electric power system data self-adapting compression is provided, has it is characterized in that: the device of said realization electric power system data self-adapting compression comprises:

Be used for realizing the periodicity analog signal sampling is obtained compressing list entries, between the loop cycle of signal, implement the unequal interval sampling, in loop cycle, implement the count equiphase sampling unit of equal interval sampling of fixed sample in power line communication;

Be used for said compression list entries is carried out the compression unit of processed compressed;

Said analog signal is connected with the input of said equiphase sampling unit, and the output of said equiphase sampling unit is connected with said compression unit.

According to an optimal technical scheme of the present invention: the device of said realization electric power system data self-adapting compression also comprises: be used for said compression unit packed data is formed the decompress(ion) output sequence after decompression; To recover the decompression unit of original signal waveform, the input of said decompression unit is connected with the output of said compression unit.

According to an optimal technical scheme of the present invention: said equiphase sampling unit further comprises filter unit, comparator unit, sampling pulse generation unit, time mark generator unit, frequency overlapped-resistable filter unit and synchronized sampling unit; Wherein, Said filter unit is connected with said comparator unit; Said comparator unit is connected with said time mark generator unit with said sampling pulse generation unit respectively; Said sampling pulse generation unit and said synchronized sampling unit, said frequency overlapped-resistable filter unit is connected with said synchronized sampling unit.

According to an optimal technical scheme of the present invention: said filter unit is a band pass filter, and the passband central frequency of said band pass filter is 50Hz; Said comparator unit is to be used for analog signal and no-voltage after the harmonic carcellation interference are compared, to produce the comparator of zero crossing pulse signal; Said sampling pulse generation unit; Be used for according to the rising edge timing of internal clocking, produce the zero crossing time series, and carry out mathematical statistics according to said zero crossing time series in each zero crossing pulse; Eliminate and measure random error; Predict the moment of next zero crossing pulse, confirm the sample frequency of current period, realize the equiphase sampling according to predicted value; Said time mark generator unit is for being used for according to zero crossing pulse target time mark generator when accurately time signal produces zero crossing; Said frequency overlapped-resistable filter unit is that primary signal is implemented LPF, to satisfy the frequency overlapped-resistable filter of Shannon's sampling theorem; Said synchronized sampling unit is used for according to said sample frequency the analog signal after said frequency overlapped-resistable filter cell processing being carried out equiphase sampling and output sampled value.

According to an optimal technical scheme of the present invention: said equiphase sampling unit is identical with sampling number between the cycle in the cycle; Preserve a temporal information in each equiphase sampling period; And with this temporal information employing delta modulation code processing, said temporal information is multi-way shared.

The present invention also provides a kind of method that realizes the electric power system data self-adapting compression, and said electric power system data self-adapting compression method comprises step: the first step: input signal is carried out synchronous equiphase sampling, obtain compressing list entries; Second step: said compression list entries is carried out the compression of equiphase point sequence.

According to an optimal technical scheme of the present invention: said electric power system data self-adapting compression method further comprised for the 3rd step: said packed data is carried out decompress(ion), obtain the decompress(ion) output sequence.

According to an optimal technical scheme of the present invention: the said first step comprises substep: one, analog signal is carried out Filtering Processing through said filter unit, harmonic carcellation disturbs, and produces the zero crossing pulse signal through said comparator unit again; Two, utilize said zero crossing pulse signal to generate the zero crossing time series through said pulse generation unit, and handle according to said time series, statistics is eliminated and is measured random error; Three, the next zero crossing pulse of prediction and is confirmed the sampling instant of each sampled point in the current period to utilize said sampling instant to carry out the equiphase sampling according to predicted value constantly, obtains compressing list entries.

According to an optimal technical scheme of the present invention: said second step comprises substep: one, carry out data processing; Said compression list entries is transformed into two-dimensional space from the one-dimensional space; The matrix of forming a M * N size; Calculate the mean value of each row, deduct its corresponding average of being expert at each element in the matrix; Two, carry out wavelet decomposition; Utilize integer lifting wavelet transform each capable K layer wavelet decomposition of carrying out to this section gained matrix in the said step; The low frequency coefficient that only keeps K layer wavelet decomposition; Obtain the coefficient matrix of a M * A size, wherein, A is the number of the low frequency coefficient of K layer wavelet decomposition; Three, carry out compressed encoding, M mean value that obtains in going on foot this section said one and the big or small coefficient matrix of M * A that this section obtained in said two steps carry out entropy coding formation compressed file.

According to an optimal technical scheme of the present invention: said three steps comprise substep: one, carry out decoding processing, said compressed file is carried out the entropy decoding, obtain M mean value and a M * A coefficient matrix; Two, carry out wavelet reconstruction,, obtain M * N restructuring matrix with each this line data of line reconstruction in the coefficient matrix; Three, carry out data processing, each element of said restructuring matrix is added institute's corresponding average of being expert at, then restructuring matrix is carried out two dimension to one dimension conversion generation output sequence.

Beneficial effect of the present invention is: the signal to noise ratio of equiphase sampling that adopts the inventive method is apparently higher than adopting equal interval sampling; And along with the increase of compression ratio; Adopt the signal to noise ratio of equal interval sampling to reduce rapidly; And adopt equiphase of the present invention sampling signal to noise ratio reduce slow relatively many, thereby explanation equiphase sampled data can realize big compression ratio, can be effectively and stably improve the communication bandwidth of power communication system.

[description of drawings]

Fig. 1 realizes the device basic principle schematic of electric power system data self-adapting compression for the present invention;

Fig. 2 realizes the medium phase sample cellular construction of the device sketch map of electric power system data self-adapting compression for the present invention;

Fig. 3 realizes the method flow sketch map of electric power system data self-adapting compression for the present invention is a kind of;

Fig. 4 estimates zero crossing impulsive measurement markers and accurate markers graph of a relation for sampling pulse generation unit of the present invention adopts mathematical statistics;

Fig. 5. Synchronous Sampling Pulse production process sketch map;

Fig. 6 is the sequential relationship sketch map between zero crossing pulse among the present invention, Synchronous Sampling Pulse and the analog-to-digital conversion output;

Fig. 7 is the data compression and the decompression process sketch map of compression unit of the present invention and decompression unit;

Fig. 8. 200 * 32768 dot image sketch mapes of equal interval sampling;

Fig. 9. 200 * 32768 points of equiphase sampling;

Figure 10 is for adopting two groups of data compression reconstruct of two kinds of method of sampling gained comparing result sketch map among the present invention.

[embodiment]

Below in conjunction with accompanying drawing technical scheme of the present invention is elaborated.

See also the device basic principle schematic that Fig. 1 the present invention realizes the electric power system data self-adapting compression.As shown in Figure 1: the device of said realization electric power system data self-adapting compression comprises: be used for realizing the periodicity analog signal sampling in power line communication; Obtain compressing list entries; Between the loop cycle of signal, implement the unequal interval sampling, in loop cycle, implement the count equiphase sampling unit 101 of equal interval sampling of fixed sample; The compression unit 102 that is used for said compression list entries is carried out processed compressed be used for said compression unit 102 packed datas are formed the decompress(ion) output sequence after decompression; To recover the decompression unit 103 of original signal waveform; Said analog signal is connected with the input of said equiphase sampling unit 101; The output of said equiphase sampling unit 101 is connected with said compression unit 102, and the input of said decompression unit 103 is connected with the output of said compression unit 102.

Analog signal forms the compression list entries through said equiphase sampling unit 101 among Fig. 1; The compression list entries forms compressed file through said compression unit 102 again; Like this, the recorder data of electric power system is just preserved with the form of compressed file, has saved memory space; Said compressed file is handled the back through said decompression unit 103 again at receiving terminal and is formed the decompression output sequence, thereby is able to recover original signal waveform.

Here, said analog signal is periodic voltage or current signal.

Said equiphase sampling unit 101 is used for power line communication and realizes analog signal is carried out periodic samples, and be specially: the phase is gathered M point weekly, and the point of this M is equally spaced, and is integer-period sampled; The sampling interval of sampled point maybe be different between cycle and cycle, but sampling number is the same.

See also the medium phase sample of device unit 101 structural representations that Fig. 2 the present invention realizes the electric power system data self-adapting compression.As shown in Figure 2; Said equiphase sampling unit 101 further comprises filter unit 201, comparator unit 202, sampling pulse generation unit 203, time mark generator unit 204, frequency overlapped-resistable filter unit 205 and synchronized sampling unit 206; Wherein, Said filter unit 201 is connected with said comparator unit 202; Said comparator unit 202 is connected with said time mark generator unit 204 with said sampling pulse generation unit 203 respectively, said sampling pulse generation unit 203 and said synchronized sampling unit 206, and said frequency overlapped-resistable filter unit 205 is connected with said synchronized sampling unit 206.

In optimal technical scheme of the present invention: said filter unit 201 is a band pass filter, and the passband central frequency of said band pass filter is 50Hz; Said comparator unit 202 compares analog signal and no-voltage after the harmonic carcellation interference for being used for, to produce the comparator of zero crossing pulse signal; Said sampling pulse generation unit 203; Be used for according to the rising edge timing of internal clocking, produce the zero crossing time series, and carry out mathematical statistics according to said zero crossing time series in each zero crossing pulse; Eliminate and measure random error; Predict the moment of next zero crossing pulse, confirm the sample frequency of current period, realize the equiphase sampling according to predicted value; Said time mark generator unit 204 is for being used for according to zero crossing pulse target time mark generator when accurately time signal produces zero crossing; Said frequency overlapped-resistable filter unit 205 is that primary signal is implemented LPF, to satisfy the frequency overlapped-resistable filter of Shannon's sampling theorem; Said synchronized sampling unit 206 is used for according to said sample frequency the analog signal after handling through said frequency overlapped-resistable filter unit 205 being carried out equiphase sampling and output sampled value.

Here, said equiphase sampling is in loop cycle, to remain equal interval sampling, but between circulation, adopts the unequal interval sampling.So each cycle also need be preserved a temporal information, the coding of temporal information can adopt delta modulation code to reduce redundant information; In addition; In the practical application, owing to phase place between the multiple signals in the electrical network is consistent, so temporal information is multi-way shared.

See also a kind of method flow sketch map of realizing the electric power system data self-adapting compression of Fig. 3 the present invention.As shown in Figure 3, said electric power system data self-adapting compression method comprises step: the first step: input signal is carried out synchronous equiphase sampling, obtain compressing list entries; Second step: said compression list entries is carried out the compression of equiphase point sequence.

In optimal technical scheme of the present invention: said electric power system data self-adapting compression method further comprised for the 3rd step: said packed data is carried out decompress(ion), obtain the decompress(ion) output sequence.

In optimal technical scheme of the present invention: the said first step comprises substep: one, analog signal is carried out Filtering Processing through said filter unit 201, harmonic carcellation disturbs, and produces the zero crossing pulse signal through said comparator unit 202 again; Two, utilize said zero crossing pulse signal to generate the zero crossing time series through said pulse generation unit, and handle according to said time series, statistics is eliminated and is measured random error; Three, the next zero crossing pulse of prediction and is confirmed the sampling instant of each sampled point in the current period to utilize said sampling instant to carry out the equiphase sampling according to predicted value constantly, obtains compressing list entries.

In optimal technical scheme of the present invention: said second step comprises substep: one, carry out data processing; Said compression list entries is transformed into two-dimensional space from the one-dimensional space; The matrix of forming a M * N size; Calculate the mean value of each row, deduct its corresponding average of being expert at each element in the matrix; Two, carry out wavelet decomposition; Utilize integer lifting wavelet transform each capable K layer wavelet decomposition of carrying out to this section gained matrix in the said step; The low frequency coefficient that only keeps K layer wavelet decomposition; Obtain the coefficient matrix of a M * A size, wherein, A is the number of the low frequency coefficient of K layer wavelet decomposition; Three, carry out compressed encoding, M mean value that obtains in going on foot this section said one and the big or small coefficient matrix of M * A that this section obtained in said two steps carry out entropy coding formation compressed file.

In optimal technical scheme of the present invention: said three steps comprise substep: one, carry out decoding processing, said compressed file is carried out the entropy decoding, obtain M mean value and a M * A coefficient matrix; Two, carry out wavelet reconstruction,, obtain M * N restructuring matrix with each this line data of line reconstruction in the coefficient matrix; Three, carry out data processing, each element of said restructuring matrix is added institute's corresponding average of being expert at, then restructuring matrix is carried out two dimension to one dimension conversion generation output sequence.

Below to technical scheme of the present invention launch the explanation:

The power system frequency characteristic is by the common decision of all power plants in the electric power system and load, and under effects such as power disturbance and equipment fault, the frequency of actual electric network signal can fluctuate near rated frequency usually.The fluctuation of these numerical values recited that all can cause circulating in addition, non-integer-period sampledly makes that also numerical values recited appears periodically between circulation, the multiple that these all can the restricting data compression.

Synchronized sampling has weakened the information coupling between the numerical value in the numerical value and circulation between circulation greatly, makes that redundancy is eliminated more easily between circulation, thereby leads to the big compression ratio of acquisition easily.

In algorithm in the sampling pulse generation unit 203 and the concrete implementation procedure based on the frequency real-time estimate problem of binary linear regression method:

Line voltage under normal operation; There is variation slowly in mains frequency; In subrange, can be approximated to be linear change, but have big measure error, with internal clocking CLK real-time tracking zero crossing pulse signal; Adopt mathematical statistic method to estimate the zero crossing time accurately, and then realize high-precision equiphase sampling.

Seeing also Fig. 4 sampling pulse generation unit 203 of the present invention adopts mathematical statistics to estimate zero crossing impulsive measurement markers and accurate markers graph of a relation; Be defined as y (x) and u (x) to the measured value of zero crossing time and exact value respectively respectively, then there are relation as shown in Figure 4 in y (x) and u (x):

Represent the time with reference axis, y (1), y (2), y (3) ... Y (m) is the zero crossing time measured value; U (1), u (2), u (3) ... U (m) is an exact value; ε (1), ε (2), ε (3) ... ε (m) is a measure error, and they are relevant with sensor noise, clocking noise, comparator precision; T is the mains voltage signal cycle.As shown in Figure 4, get y (1), y (2), y (3) ... Y (m) can get as research object:

y(x)＝u(x)+ε(x) (1)

$u\left(x\right)=u\left(1\right)+\underset{i=1}{\overset{x-1}{\mathrm{\Σ}}}{T}_i---\left(2\right)$

In the formula, x ∈ N, ε (x)～N (0, σ ²), x is the sample sequence number.

The power network signal cycle is carried out local linearization to get:

T _i＝T ₁+ΔT·(i-1) (3)

Then wushu (3) substitution (2) formula gets:

u(x)＝b ₀+b ₁·x+b ₂·x ² (4)

In the formula:

b ₀＝u(1)-T ₁+ΔT

${\mathrm{<figure-callout\; id="1"\; label="b"\; filenames="H2009101066791E00031.png"\; state="\{\{state\}\}">b</figure-callout>}}_1=T_1-\frac{3\&CenterDot;\mathrm{\&Delta;<figure-callout\; id="2"\; label="T"\; filenames="H2009101066791E00031.png"\; state="\{\{state\}\}">T</figure-callout>}}{2}$

${\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="H2009101066791E00031.png"\; state="\{\{state\}\}">b</figure-callout>}}_2=\frac{\mathrm{\&Delta;<figure-callout\; id="2"\; label="T"\; filenames="H2009101066791E00031.png"\; state="\{\{state\}\}">T</figure-callout>}}{2}$

Wushu (4) substitution formula (1):

y(x)＝b ₀+b ₁·x+b ₂·x ²+ε(x) (5)

Make x ₁=x, x ₂=x ², substitution (5) formula gets:

y＝b ₀+b ₁·x ₁+b ₂·x ₂+ε (6)

Form shown in the formula (6) meets the binary linear regression model, therefore can carry out linear regression analysis to y.X1, x2 and y are observed, and their value is designated as x1i, x2i and yi respectively during the i time observation, and random error is ε i, then obtains equation:

y _i＝b ₀+b ₁·x _1i+b ₂·x _2i+ε _i (7)

In formula (7), carry out m time altogether and observe, use least square method that b0, b1, b2 are estimated then, the estimation that gets b0, b1, b2 is respectively:

$\left[\begin{array}{c}{\hat{b}}_1\\ {\hat{b}}_2\end{array}\right]{=L}^{-1}\mathrm{XY}---\left(8\right)$

${\hat{b}}_0=\stackrel{\&OverBar;}{y}-{\hat{\mathrm{<figure-callout\; id="1"\; label="b\; ^"\; filenames="H2009101066791E00031.png"\; state="\{\{state\}\}">b</figure-callout>}}}_1\&CenterDot;{\stackrel{\&OverBar;}{x}}_1-{\hat{\mathrm{<figure-callout\; id="2"\; label="b\; ^"\; filenames="H2009101066791E00031.png"\; state="\{\{state\}\}">b</figure-callout>}}}_2\&CenterDot;{\stackrel{\&OverBar;}{x}}_2$

In the formula:

$X=\left[\begin{array}{cccc}{x}_{11}^{*}& x_{12}^{*}& \&CenterDot;\&CenterDot;\&CenterDot;& x_{1m}^{*}\\ x_{21}^{*}& x_{22}^{*}& \&CenterDot;\&CenterDot;\&CenterDot;& x_{2m}^{*}\end{array}\right],$ $Y=\left[\begin{array}{c}{y}_1\\ {\mathrm{<figure-callout\; id="2"\; label="y"\; filenames="H2009101066791E00031.png"\; state="\{\{state\}\}">y</figure-callout>}}_2\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ y_m\end{array}\right]$

$L=\left[\begin{array}{cc}\underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}{x}_{1i}^{*}\&CenterDot;x_{1i}^{*}& \underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}{x}_{1i}^{*}\&CenterDot;x_{2i}^{*}\\ \underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}{x}_{2i}^{*}\&CenterDot;x_{1i}^{*}& \underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}{x}_{2i}^{*}\&CenterDot;x_{2i}^{*}\end{array}\right],$ $x_{\mathrm{ki}}^{*}=x_{\mathrm{ki}}-{\stackrel{\&OverBar;}{x}}_k$

Then get by formula (4):

$\hat{u}\left(x\right)={\hat{b}}_0+{\hat{b}}_1\&CenterDot;x+{\hat{b}}_2\&CenterDot;x^2---\left(9\right)$

Here, utilize the accurate measured value of above-mentioned gained again, and then the algorithmic procedure of generation Synchronous Sampling Pulse is following:

This algorithm use sliding window mode realizes; In said sliding window, preserve m up-to-date zero crossing all the time constantly; Whenever there being new zero crossing pulse to produce, said window just moves forward a step, gives up the oldest data; Add up-to-date data, and then calculate the binary linear regression model parameter to the new sample of this group.

Here, the sampling pulse strict synchronism is to realize the key condition of equiphase data acquisition in zero crossing moment predicted value.Through the zero crossing pulse signal is carried out synchronizing and frequency doubling; Can produce the sampling pulse synchronous with zero crossing; When last Synchronous Sampling Pulse in the one-period after U1 exports constantly; Sampling pulse generation unit 203 carries out trend extropolation by formula (9) and gets according to the next zero crossing time U2 of the parameter prediction of binary linear regression model:

${\mathrm{<figure-callout\; id="2"\; label="U"\; filenames="H2009101066791E00031.png"\; state="\{\{state\}\}">U</figure-callout>}}_2=\hat{u}(m+i)={\hat{b}}_0+{\hat{b}}_1(m+i)+{\hat{b}}_2{(m+i)}^2,(i>0)---\left(10\right)$

I is the step number of u (m) to U2 in the formula (10); Size is determined by the model parameter renewal frequency; After U2 confirms; Just can be according to linear relationship in the hope of the moment of each sampled point in phase this week, total sampling number of establishing each cycle is K, then can set up the relation between the sampling instant corresponding with it of each sampled point:

$y\left(x\right)=\mathrm{<figure-callout\; id="1"\; label="U"\; filenames="H2009101066791E00031.png"\; state="\{\{state\}\}">U</figure-callout>}1+\frac{U_2-U_1}{K}\&CenterDot;x,x=\mathrm{1,2},...K---\left(11\right)$

Here; The control of said period measurement and sampling instant realizes that by field programmable gate array (FPGA) timer owing to receive the restriction of crystal oscillator frequency, the resolution of timer is limited; So behind the frequency measurement, the sampling instant that is provided by timer is compared with calculated value and is also had truncated error.The advantage of formula (11) is the truncated error that only exists single sampled point to be produced by counter, and can not produce cumulative errors by truncated error, and we are referred to as the synchronous digital frequency multiplication method to this frequency-doubling method.At this moment, can adopt the method for rounding up farthest to reduce the measure error that produces because of phase error.

See also Fig. 5 Synchronous Sampling Pulse production process sketch map; As shown in Figure 5; The generation of Synchronous Sampling Pulse realizes based on two separate processes, and regression analysis and synchronizing and frequency doubling are as shown in Figure 5; What last reference axis was represented among Fig. 5 is regression analysis process, and what following reference axis was represented is the synchronizing and frequency doubling process.

Sum up above process, two separate processes of regression analysis and synchronizing and frequency doubling can specifically be represented as follows:

A. regression analysis

(1) waits for that new zero crossing produces;

(2) after new zero crossing produces, the sliding window position that moves forward;

(3),, formula (8) continues execution in step (1) then by calculating

according to the sample sequence in the window.

B. synchronizing and frequency doubling

(1) U1 begins constantly, and value is calculated U2 by formula (10) according to

;

(2) calculate s (1), s (2) according to U1, U2 value by formula (11) ... s (K);

(3) s (1) that calculates according to step (2), s (2) ... s (K) time series produces Synchronous Sampling Pulse successively;

(4) U1=U2 continues execution in step (1) then.

See also the sequential relationship sketch map between zero crossing pulse among Fig. 6 the present invention, Synchronous Sampling Pulse and the AD conversion output, ZPS is a zero crossing pulse prediction signal among the figure, and SYN is the Synchronous Sampling Pulse signal synchronous with ZPS; ADC is synchronous AD transformation result; It is the rising edge at each SYN that AD conversion starts constantly, as can beappreciated from fig. 4, the timing Design strict guarantee synchronism of AD conversion with the zero crossing pulse; Each cycle equal interval sampling K time has been realized the equiphase sampling.

See also the data compression and the decompression process sketch map of Fig. 7 compression unit 102 of the present invention and decompression unit 103.Realize the compression of equiphase point data, at first will signal be transformed into two-dimensional space from the one-dimensional space.For cyclical signal, so not only help the decomposition of cyclic stationary signal, and the variation of matrix on line direction will be very steady with the sampled data in one or more cycles as the row in the two-dimensional matrix.

For long-term stable state record ripple, numerical value change is slow between circulation, and numerical value change is very fast in the circulation, and data redundancy is far smaller than between circulation redundant in the circulation.

Data compression as shown in Figure 7 and decompression process sketch map, X={x ₁, x ₂X _LBe the compression list entries, $\hat{X}=\{{\hat{x}}_1,{\hat{x}}_2\&CenterDot;\&CenterDot;\&CenterDot;{\hat{x}}_L\}$ For the decompression output sequence, establish L=M * N, wherein M is the sampling number in each cycle, N is a periodicity.

Detailed process is:

A. compression algorithm:

Step (1) data processing: at first list entries is transformed into two-dimensional space from the one-dimensional space, forms the matrix of a M * N size; Calculate the mean value of each row then; Deduct its corresponding average of being expert at each element in the matrix at last.

Step (2) wavelet decomposition: utilize integer lifting wavelet transform that each row of gained matrix in the step (1) is carried out K layer wavelet decomposition; The low frequency coefficient that only keeps K layer wavelet decomposition; Obtain the coefficient matrix of a M * A size, wherein A is the number of the low frequency coefficient of K layer wavelet decomposition.

Step (3) coding: the coefficient matrix of M the mean value that obtains step (1) and M * A size that step (2) is obtained carries out entropy coding formation compressed file.

B. decompression algorithm:

(1) decoding: file is carried out the entropy decoding, obtain M mean value and a M * A coefficient matrix.

(2) wavelet reconstruction:, obtain M * N restructuring matrix with each this line data of line reconstruction in the coefficient matrix.

(3) data processing: at first, each element of restructuring matrix adds institute's corresponding average of being expert at; Then, carry out two dimension to restructuring matrix and produce output sequence to the one dimension conversion.

Above process can realize harmless or lossy compression method, works as K=0, i.e. the wavelet decomposition restructuring procedure is not carried out in expression, directly gets into step (3) after compression process step (1) is accomplished; In like manner, directly get into step (3) after decompression process step (1) is accomplished.

The equal interval sampling and the equiphase of the 50Hz frequency voltage signal that records with the electrical network same point are sampled two groups of data as research object.

First group of data carried out equal interval sampling with the sample frequency of 10000 point/seconds, 6553600 sampled points of recording occurring continuously, and the sampling number in the signal nominal period (0.02 second) is 200,32768 cycles are equivalent to sample.

Second group of data carried out the equiphase sampling according to the method for preceding text, and the phase sampling number is 200 weekly, and 32768 cycles of recording occurring continuously, sampled point also are totally 6553600.

Convert these two groups of data the form of two-dimensional matrix (size is 200 row 32768 row) into respectively, and show like Fig. 8, shown in Figure 9 with the mode of gray-scale map.Fig. 8 is equal interval sampling (being an asynchronous-sampling) design sketch, and visible because mains frequency fluctuates and be non-integer-period sampled, the data of its X direction present fluctuation.Fig. 9 is the design sketch that utilizes the equiphase sampling (being synchronized sampling) of the method for the invention realization, and its X direction data are consistent.

Definition compression measurement index compression ratio CR and signal to noise ratio snr:

$C_R=\frac{H_0}{H_C},$ $S_{\mathrm{NR}}=101g\frac{\underset{i=1}{\overset{\mathrm{<figure-callout\; id="2"\; label="L\; x\; i"\; filenames="H2009101066791E00031.png"\; state="\{\{state\}\}">L</figure-callout>}}{\mathrm{\&Sigma;}}}{x}_i^{}{}_2^{}}{\underset{i=1}{\overset{L}{\mathrm{\&Sigma;}}}{(x_i-{\hat{x}}_i)}^2}$

Ho is the size of compression preceding document in the formula, and Hc is file size after the compression.Above two groups of data are compressed the compression result that obtains see table 1 and table 2.

Table 1 equal interval sampling data compression tabulation

Table 2 equiphase sampled data compressed list

Carry out the data compression test according to the step of above-mentioned compression algorithm respectively with Fig. 8, two groups of data shown in Figure 9; The figure place of used digital to analog converter is 16 in the test; Gather 6553600 points altogether, compression preceding document size is 13107200 bytes, in the compression test; Wavelet transformation has adopted integer haar wavelet arithmetic, and entropy coding has adopted arithmetic coding.

Then with shown in the table 2, first group of number only carries out lossless compress with entropy coding to the result of two groups of data compressions in table 1 and the table 2 like table 1, only does the direct current translation before the entropy coding and handles.Compare and assessment in order two groups of data compression algorithms to be carried out performance; Shown in Figure 10 being shown of two groups of data by compression ratio and signal to noise ratio relation table, can see from Figure 10, under the identical compression ratio; The signal to noise ratio of equiphase sampling is apparently higher than equal interval sampling; And along with compression ratio increases, the signal to noise ratio of equal interval sampling reduces rapidly, and reducing of the signal to noise ratio of equiphase sampling is slow relatively many; This explanation equiphase sampled data can realize big compression ratio, and the dotted portion among Figure 10 is represented is the compression ratio during without entropy coding.Can see again that from table 1 table 2 comparing result equiphase sampling entropy coding compression ratio has still kept higher signal to noise ratio apparently higher than the equal interval sampling compression ratio when especially the equiphase sampling is compressed to 5347.7 times.

The signal to noise ratio of equiphase sampling that adopts the inventive method is apparently higher than adopting equal interval sampling; And along with the increase of compression ratio; Adopt the signal to noise ratio of equal interval sampling to reduce rapidly; And adopt equiphase of the present invention sampling signal to noise ratio reduce slow relatively many, thereby explanation equiphase sampled data can realize big compression ratio, can be effectively and stably improve the communication bandwidth of power communication system.

Above content is to combine concrete optimal technical scheme to the further explain that the present invention did, and can not assert that practical implementation of the present invention is confined to these explanations.For the those of ordinary skill of technical field under the present invention, under the prerequisite that does not break away from the present invention's design, can also make some simple deduction or replace, all should be regarded as belonging to protection scope of the present invention.

Claims (9)

1. device of realizing electric power system data self-adapting compression, it is characterized in that: the device of said realization electric power system data self-adapting compression comprises:

Be used for realizing the periodicity analog signal sampling in power line communication; Obtain compressing list entries; Between the loop cycle of signal, implement the unequal interval sampling, in loop cycle, implement the count equiphase sampling unit (101) of equal interval sampling of fixed sample;

Be used for said compression list entries is carried out the compression unit (102) of processed compressed;

Said analog signal is connected with the input of said equiphase sampling unit (101), and the output of said equiphase sampling unit (101) is connected with said compression unit (102).

2. according to the device of the said realization electric power system data self-adapting compression of claim 1, it is characterized in that: the device of said realization electric power system data self-adapting compression also comprises:

Be used for said compression unit (102) packed data is formed the decompress(ion) output sequence after decompression, to recover the decompression unit (103) of original signal waveform, the input of said decompression unit (103) is connected with the output of said compression unit (102).

3. the device that compresses according to the said realization electric power system data self-adapting of claim 1; It is characterized in that: said equiphase sampling unit (101) further comprises filter unit (201), comparator unit (202), sampling pulse generation unit (203), time mark generator unit (204), frequency overlapped-resistable filter unit (205) and synchronized sampling unit (206); Wherein, Said filter unit (201) is connected with said comparator unit (202); Said comparator unit (202) is connected with said time mark generator unit (204) with said sampling pulse generation unit (203) respectively; Said sampling pulse generation unit (203) is connected with said synchronized sampling unit (206), and said frequency overlapped-resistable filter unit (205) is connected with said synchronized sampling unit (206); Said filter unit (201) is a band pass filter, and the passband central frequency of said band pass filter is 50Hz;

Said comparator unit (202) compares analog signal and no-voltage after the harmonic carcellation interference for being used for, to produce the comparator of zero crossing pulse signal;

Said sampling pulse generation unit (203); Be used for according to the rising edge timing of internal clocking, produce the zero crossing time series, and carry out mathematical statistics according to said zero crossing time series in each zero crossing pulse; Eliminate and measure random error; Predict the moment of next zero crossing pulse, confirm the sample frequency of current period, realize the equiphase sampling according to predicted value;

Said time mark generator unit (204) is for being used for according to zero crossing pulse target time mark generator when accurately time signal produces zero crossing;

Said frequency overlapped-resistable filter unit (205) is that primary signal is implemented LPF, to satisfy the frequency overlapped-resistable filter of Shannon's sampling theorem;

Said synchronized sampling unit (206) is used for according to said sample frequency the analog signal after said frequency overlapped-resistable filter cell processing being carried out equiphase sampling and output sampled value.

4. the device that compresses according to the said realization electric power system data self-adapting of claim 1; It is characterized in that: said equiphase sampling unit (101) is identical with sampling number between the cycle in the cycle; Preserve a temporal information in each equiphase sampling period; And with this temporal information employing delta modulation code processing, said temporal information is multi-way shared.

5. method that realizes electric power system data self-adapting compression, it is characterized in that: said electric power system data self-adapting compression method comprises step:

A: input signal is carried out the sampling of synchronous equiphase, obtain compressing list entries, said synchronous equiphase is sampled as implements the unequal interval sampling between the loop cycle of signal, in loop cycle, implements the fixed sample equal interval sampling of counting;

B: said compression list entries is carried out the compression of equiphase point sequence.

6. according to the method for the said realization electric power system data self-adapting compression of claim 5, it is characterized in that: said electric power system data self-adapting compression method further comprises step:

C: packed data is carried out decompress(ion), obtain the decompress(ion) output sequence.

7. according to the method for the said realization electric power system data self-adapting compression of claim 5, it is characterized in that: said step a comprises substep:

A1: analog passband signal is crossed filter unit (201) carry out Filtering Processing, harmonic carcellation disturbs, and passes through comparator unit (202) again and produces the zero crossing pulse signal;

A2: utilize said zero crossing pulse signal to generate the zero crossing time series through the pulse generation unit, and handle according to said time series, statistics is eliminated and is measured random error;

A3: predict next zero crossing pulse constantly, and confirm the sampling instant of each sampled point in the current period to utilize said sampling instant to carry out the equiphase sampling, obtain compressing list entries according to predicted value.

8. according to the method for the said realization electric power system data self-adapting compression of claim 6, it is characterized in that: said step b comprises substep:

B1: carry out data processing; Said compression list entries is transformed into two-dimensional space from the one-dimensional space; The matrix of forming one

size; Calculate the mean value of each row, deduct its corresponding average of being expert at each element in the matrix;

B2: carry out wavelet decomposition; Utilize integer lifting wavelet transform that each row of gained matrix among the said step b1 is carried out K layer wavelet decomposition; The low frequency coefficient that only keeps K layer wavelet decomposition; Obtain the coefficient matrix of one

size; Wherein, A is the number of the low frequency coefficient of K layer wavelet decomposition;

B3: carry out compressed encoding, coefficient matrix of

size that M mean value that obtains said step b1 and said step b2 obtain carries out entropy coding and forms compressed file;

Said M is the sampling number in each cycle, and said N is a periodicity.

9. the method for said according to Claim 8 realization electric power system data self-adapting compression, it is characterized in that: said step c comprises substep:

C1: carry out decoding processing; Compressed file is carried out the entropy decoding, obtain M mean value and

coefficient matrix;

C2: carry out wavelet reconstruction; With each this line data of line reconstruction in the coefficient matrix, obtain

restructuring matrix;

C3: carry out data processing, each element of said restructuring matrix is added institute's corresponding average of being expert at, then restructuring matrix is carried out two dimension to one dimension conversion generation output sequence;

Said M is the sampling number in each cycle, and said N is a periodicity, and A is the number of the low frequency coefficient of K layer wavelet decomposition.

Images