CN101446605A - Low-frequency oscillation monitoring method for power system - Google Patents

Abstract

The invention discloses a low-frequency oscillation monitoring method for a power system, and the method applies the oscillation feature identification method to carry out the preliminary identification of whether sampling data contains oscillation information or not, introduces the general digital signal processing method-digital filter method to extract related oscillation components, applies the wave peak indirect identification method to calculate oscillation amplitude through phase angle and amplitude value which are corresponding to each sampling value in a period wave and applies multi-level alarm thresholds to highlight important alarm information. The method can effectively improve the pertinence and the accuracy of the low-frequency oscillation on-line monitoring.

Description

A kind of low-frequency oscillation monitoring method for power system

Technical field

The present invention proposes a kind of low-frequency oscillation monitoring method for power system, belong to the theory and the analysis field of dynamic power system.

Background technology

1, low-frequency oscillation brief introduction

Whether system stability problem, i.e. electric system keep stable operation after being subjected to various disturbances, be the key problem that the dynamic power system need research and solve.The system stability problem comprises that transient stability, dynamic stability and three big classes of steady stability, the lasting low-frequency oscillation that causes owing to underdamping between generator amature when electric system is subjected to microvariations belong to the category that power system dynamic stability is analyzed.

The oscillation frequency of this oscillatory occurences is generally 0.2～2.5Hz.Because when low-frequency oscillation took place, corresponding vibration also can take place in the performance number on the transmission line of electricity, so be referred to as oscillation of power or electromechanical oscillations again.

Low-frequency oscillation has two kinds of mode of oscillation: vibration (plant mode) in vibration (interarea mode) and the zone between system domain.When vibrating between the generation systems territory, electrical distance is bigger between the unit that participates in vibrating, and oscillation frequency is lower (as 0.2～0.5Hz); When vibrating in the generation area, electrical distance is less between the unit that participates in vibrating, oscillation frequency higher (more than 1Hz).Vibration has bigger harm than vibration in the zone between system domain.

2, the necessity of China's electrical network feature and enforcement low-frequency oscillation monitoring analysis

China's electrical network area coverage is big, structural weak, and the distribution of various primary energy and the density of load are extremely inhomogeneous, and power supply is often away from load center, the standard power transmission line length that unit installed capacity is shared is than the much less of developed country.

Because low-frequency oscillation or oscillation of power appear on long distance, the heavy load power transmission line usually, adopt modern fast, more be easy to generate under the condition of high limited value multiple excitation system, so it is very big that the possibility of low-frequency oscillation takes place on China's electrical network, the safe and stable operation of electric system is faced with great challenge.

From the low-frequency oscillation accident several times that has taken place both at home and abroad, this accident is serious to electrical network harm, has restricted the ability to transmit electricity of electrical network greatly.By the long-term accumulation of operating experience, when find producing between the territory low-frequency oscillation, the phase angular oscillation of each node has a process that grows from weak to strong in the grid, is the process of accident development.By low-frequency oscillation is effectively monitored, can it be caught at the initial stage that accident takes place, real-time follow-up is analyzed its state of development, in time discerns oscillation center, the excision disturbing source, thus drop to minimum to the harmful effect of electrical network low-frequency oscillation.But for the low-frequency oscillation that takes place in the zone,, continue shortly, calm down fast characteristics,, and have little time to take counter-measure so this oscillatory occurences only can be monitored to because that it often has a starting of oscillation is fast.For this vibration, should carry out record to real time data, analyze the oscillation information in the recognition data then, and the information to extract, improve the system damping characteristic, the final inhibition that realizes low-frequency oscillation.

3, existing low-frequency oscillation on-line monitoring method

The on-line monitoring method of low-frequency oscillation at present mainly contains: 1) adopting to improve the Prony algorithm is that the low-frequency oscillation line model recognizer of core is monitored low-frequency oscillation; 2) according to monitored amount crude sampling value, adopt " cycle-the amplitude detecting method " low-frequency oscillation is monitored.In the first method,, generally need in algorithm, add filtration module because the noiseproof feature of Prony algorithm is relatively poor.Continuously certain monitoring variable is carried out filtering and a large amount of cpu resource of Prony analysis needs consumption, when the monitored signal number increases, the system resource that this algorithm takies will reach unacceptable degree.In the second method, the cycle-amplitude detecting method principle is simple, fast operation, but this method analyze at raw data, be difficult to avoid the mistake alarm that causes by the meaningless spike burr that comprises in the crude sampling value.

Summary of the invention

The present invention proposes a kind of low-frequency oscillation monitoring method for power system, it is characterized in that: may further comprise the steps:

1, earlier the variation track of one section active power data in the data buffer zone is analyzed, amplitude, the cycle of preliminary identification oscillating curve, recognition result and The Characteristics of Low Frequency Oscillations are compared, if the identification amount does not obviously meet oscillation characteristics, then upgrade the content in the data buffer, one section new active power data are repeated above-mentioned analytic process,, then carry out subsequent analysis if the identification amount meets the fundamental oscillation feature;

2, described active power data segment utilizes the digital filter algorithm, extracts the oscillating component of oscillation frequency between 0.2～2.5Hz in the described active power data segment;

3, the above-mentioned oscillating component of just extracting is used crest indirect identification method identification oscillation amplitude, when oscillation amplitude surpasses the alarming threshold value, proposes alarm;

4, described alarming threshold value is divided into two sections, when satisfying low alarming threshold value, and prompting " system fluctuation " alarm, when satisfying higher alarming threshold value continuously several times, prompting " low-frequency oscillation " alarm.

When the on-line monitoring algorithm is judged monitored line power generation low-frequency oscillation, with the current time is center time point, amount to 1 minute data before and after accumulating this moment, off-line low frequency oscillation mode analysis module is delivered in packing, calls the off-line analysis algorithm this low-frequency oscillation is analyzed.

The present invention has the following advantages:

1,, can evade effectively because the low-frequency oscillation monitoring algorithm that insignificant spike burr and the normal saltus step of sampled value (as: branchs/making process) cause in the original sampled signal misses the alarm problem based on " the oscillation characteristics recognition methods " of data variation Trajectory Design;

2, the use digital filter extracts the low-frequency oscillation component accurate and effective in the crude sampling value;

3, use the oscillation amplitude of " crest indirect identification method " identification low-frequency oscillation component, the amplitude of identification is more accurate, and She Ji low-frequency oscillation criterion is more reliable in view of the above;

4, the mode of criterion segmentation can be alarmed according to the different dangerous situation of grid, outstanding major accident;

5, adopt multiple mode to monitor low-frequency oscillation, the monitoring blind area of each method is supplied mutually; Classification progressively detects oscillation characteristics, has improved the operational efficiency and the speed of low-frequency oscillation on-line monitoring algorithm;

Description of drawings

The present invention is further described below in conjunction with accompanying drawing.

Fig. 1 is a low-frequency oscillation of electric power system on-line monitoring method flow diagram;

Fig. 2 is the periodic convolution synoptic diagram;

Fig. 3 is digital filter realization flow figure;

Fig. 4 is an oscillation characteristics recognizer flow process;

Fig. 5 is by peak-valley location determination oscillation frequency principle schematic;

Fig. 6 is by the effectively full number of oscillation algorithm flow of peak/valley location recognition;

Fig. 7 is a digital filter filtering synoptic diagram;

Fig. 8 is the signal of crest indirect identification method principle;

Fig. 9 is a crest indirect identification method flow diagram;

Figure 10 is the alarm detection flow process of low-frequency oscillation on-line monitoring algorithm;

Embodiment

Below more specifically introduce content of the present invention.

The low-frequency oscillation on-line monitoring not only can be used for the supervision and the protection of electric power networks, also can be used for discerning unstable circuit simultaneously, differentiates the validity of vibration braking measure; Mode of oscillation identification not only can be understood the built-in oscillation characteristic of the monitored circuit of wall scroll, also has positive effect to analyzing the oscillation source position with the vibration reason.

The objective of the invention is: by improving traditional low-frequency oscillation on-line monitoring method, introduce the on-line monitoring that general digital signal processing method is effectively realized low-frequency oscillation, improve the specific aim and the accuracy of monitoring.

According to electric system on-line monitoring algorithm flow of the present invention as shown in Figure 1.

Step 1: initialization.

Step1: design digital filter.

Design digital filter based on window technique among the present invention.

Basic design philosophy is: choose a kind of suitable ideal frequency selective filter, then its impulse response is blocked (or windowing) to obtain the FIR wave filter of a linear phase and cause and effect.Therefore, this method focuses on selecting appropriate " window function " and " ideal frequency selective filter ".

Use H _d(e ^{J ω}) represent one " ideal frequency selective filter ", it has unit amplitude gain and linear phase characteristic in whole passband, and has zero response in stopband.One optimum wideband is ω _cThe LPF of＜π is provided by following formula

$H_d\left(e^{\mathrm{j\&omega;}}\right)=\left\{\begin{array}{cc}1*e^{-\mathrm{j\&alpha;\&omega;}},& \left|\mathrm{\&omega;}\right|\&le;{\mathrm{\&omega;}}_{\mathrm{<figure-callout\; id="0"\; label="c"\; filenames="A200810227843E00191.png,A200810227843E00201.png"\; state="\{\{state\}\}">c</figure-callout>}}\\ 0,& {\mathrm{\&omega;}}_c\&le;\left|\mathrm{\&omega;}\right|\&le;\mathrm{\&pi;}\end{array}\right.---\left(1\right)$

ω wherein _cBe also referred to as cutoff frequency, α is called sample delay.The impulse response of this wave filter is shown below:

$h_d\left(n\right)=F^{-1}\left[H_d\left(e^{\mathrm{j\&omega;}}\right)\right]=\frac{1}{2\mathrm{\&pi;}}{\&Integral;}_{-\mathrm{\&pi;}}^{\mathrm{\&pi;}}{H}_d\left(e^{\mathrm{j\&omega;}}\right)e^{\mathrm{j\&omega;n}}\mathrm{d\&omega;}$

$=\frac{1}{2\mathrm{\&pi;}}\underset{-{\mathrm{\&omega;}}_c}{\overset{{\mathrm{\&omega;}}_{\mathrm{<figure-callout\; id="1"\; label="c"\; filenames="A200810227843E00171.png,A200810227843E00191.png"\; state="\{\{state\}\}">c</figure-callout>}}}{\&Integral;}}1*e^{-\mathrm{j\&alpha;\&omega;}}{e}^{\mathrm{j\&omega;n}}\mathrm{d\&omega;}---\left(2\right)$

$=\frac{\sin \left[{\mathrm{\&omega;}}_c(n-\mathrm{\&alpha;})\right]}{\mathrm{\&pi;}(n-\mathrm{\&alpha;})}$

The windowing computing is exactly with h _d(n) both sides are blocked, thereby to obtain a length be the cause and effect of M and possess linear phase FIR filter h (n), is shown below:

$h\left(n\right)=\left\{\begin{array}{cc}{h}_d\left(n\right),& 0\&le;n\&le;M-1\\ 0,& \mathrm{others}\end{array}\right.---\left(3\right)$

With

$\mathrm{\&alpha;}=\frac{M-1}{2}---\left(4\right)$

H (n) can be by h _d(n) and a certain window function ω (n) multiply each other and obtain, promptly

h(n)＝h _d(n)ω(n) (5)

Wherein, ω (n) function is certain function about the α symmetry in 0≤n≤M-1, and when n was worth for other, functional value was 0.

This cause and effect FIR filter response H (e in frequency domain ^{J ω}) be by H _d(e ^{J ω}) and window response W (e ^{J ω}) periodic convolution obtain, promptly

$H\left(e^{\mathrm{j\&omega;}}\right)=\frac{1}{2\mathrm{\&pi;}}\underset{-\mathrm{\&pi;}}{\overset{\mathrm{\&pi;}}{\&Integral;}}W\left(e^{\mathrm{j\&lambda;}}\right)H_d\left(e^{j(\mathrm{\&omega;}-\mathrm{\&lambda;})}\right)\mathrm{d\&lambda;}---\left(6\right)$

For typical window response, accompanying drawing 2 shows the result of periodic convolution.

For given wave filter technology requirement, need selective filter length M and certain window function ω (n) with the narrowest main lobe width and as far as possible little secondary lobe.

The cardinal rule of window design method is:

1) as far as possible little M---reduce calculated amount;

2) narrow as far as possible main lobe width---dwindle transitional zone, strengthen the frequency-selecting energy;

3) as far as possible little maximum secondary lobe height---reduce ripple, strengthen the frequency-selecting energy;

The window function of Xuan Zeing is herein: kaiser window.It has two parameters, by regulating this two parameters, main lobe width and maximum secondary lobe height is just met design requirement simultaneously.

The time domain form of kaiser window function can be expressed as:

$\begin{array}{cc}w\left(k\right)=\frac{I_0\left[\mathrm{\&beta;}\sqrt{1-{(1-\frac{2k}{N-1})}^2}\right]}{I_0\left(\mathrm{\&beta;}\right)}& 0\&le;k\&le;N-1\end{array}---\left(7\right)$

I wherein ₀Be first kind distortion zeroth order Bei Saier function (β), β is the form parameter of window function, is determined by following formula:

$\mathrm{\&beta;}=\left\{\begin{array}{cc}0,& \mathrm{\&alpha;}<13.26\\ 0.76609{(\mathrm{\&alpha;}-13.26)}^{0.4}+0.09834(\mathrm{\&alpha;}-13.26),& 13.26<\mathrm{\&alpha;}<60\\ 0.12438(\mathrm{\&alpha;}+6.3),& 60<\mathrm{\&alpha;}<120\end{array}\right.---\left(8\right)$

Wherein, α is the main lobe value of kaiser window function and the difference (dB) between the side lobe levels.Change the value of β, can carry out freely selecting main lobe width and side lobe attenuation.The value of β is big more, and the side lobe levels of window function frequency spectrum is just more little, and its main lobe width is just wide more.

Calculate first kind zeroth order distortion Bei Saier function I ₀(x) be equal to the J that calculates pure imaginary argument ₀(x), the pass between them is:

I ₀(x)＝(-1) ⁿJ ₀(ix) (9)

The formula that calculates them all is the approximation polynomial that is provided by Abramowitz and Stegun, and is specific as follows:

When | x|＜3.75,

I ₀(x)＝p ₁+p ₂y+p ₃y ²+p ₄y ³+p ₅y ⁴+p ₆y ⁵+p ₇y ⁶ (10)

When | x| 〉=3.75,

$I_0\left(x\right)\&ap;e^{\left|x\right|}/\sqrt{\left|x\right|}*({\mathrm{<figure-callout\; id="1"\; label="q"\; filenames="A200810227843E00171.png,A200810227843E00191.png"\; state="\{\{state\}\}">q</figure-callout>}}_1+q_{2}z+q_3{\mathrm{<figure-callout\; id="2"\; label="z"\; filenames="A200810227843E00191.png,A200810227843E00201.png"\; state="\{\{state\}\}">z</figure-callout>}}^2+q_4{\mathrm{<figure-callout\; id="3"\; label="z"\; filenames="A200810227843E00191.png"\; state="\{\{state\}\}">z</figure-callout>}}^3+q_5{\mathrm{<figure-callout\; id="4"\; label="z"\; filenames="A200810227843E00191.png,A200810227843E00201.png"\; state="\{\{state\}\}">z</figure-callout>}}^4+q_6{\mathrm{<figure-callout\; id="5"\; label="z"\; filenames="A200810227843E00191.png"\; state="\{\{state\}\}">z</figure-callout>}}^5+q_7{\mathrm{<figure-callout\; id="6"\; label="z"\; filenames="A200810227843E00191.png,A200810227843E00201.png"\; state="\{\{state\}\}">z</figure-callout>}}^6+q_8{\mathrm{<figure-callout\; id="7"\; label="z"\; filenames="A200810227843E00191.png"\; state="\{\{state\}\}">z</figure-callout>}}^7+q_9{z}^8)---\left(11\right)$

Wherein: y=(x/3.75) ², z=3.75/|x|, coefficient p _iAnd q _iAs shown in the table.

Table 1 coefficient table

i	p i	q i
			1	1	0.39894228
2	3.5156229	0.01328592
			3	3.0899424	0.00225319
4	1.2067492	-0.00157565
			5	0.2659732	0.00916281
6	0.0360768	-0.02057706
			7	0.0045813	0.02635537
8		-0.01647633
			9		0.00392377

The kaiser window parameter that needs to determine is: window width L and parameter beta.

By the discussion of front as can be known: strengthen window width, can dwindle main lobe width, dwindled the transitional zone of wave filter simultaneously.Therefore wave filter window width L should select according to " transitional zone " this technical indicator that provides in the design conditions.

The first secondary lobe relative height (dB) has corresponding relation in the β parameter of kaiser window and the window function spectrum curve.Because the first secondary lobe relative height (dB) is more little, the stopband ripple of wave filter is just more little.So select suitable β parameter according to " stopband ripple " that provide in the design conditions this technical indicator.

Filter Design constraint condition generally includes: the stopband edge frequencies omega _s, the passband edge frequencies omega _pWith peak error δ (use the THE DESIGN OF WINDOW FUNCTION wave filter, identical error arranged) at passband and stopband.Definition

A＝-20log ₁₀δ (12)

Can calculate by following formula by the β value of A value regulation so

$\mathrm{\&beta;}=\left\{\begin{array}{cc}0.1102(A-8.7)& A>50\\ 0.5842(A-21)0.4+0.07886(A-21)& 21\&le;A\&le;50\\ 0.0& A<21\end{array}\right.---\left(13\right)$

If the definition transitional zone is

Δω＝ω _s-ω _p (14)

Unite A so and Δ ω can determine window width

$L=\frac{A-8}{2.285\mathrm{\&Delta;\&omega;}}---\left(15\right)$

Shown in the wave filter realization flow accompanying drawing 3.

Step2: determine raw data buffering storehouse length.

When raw data buffering storehouse is identical with the wave filter window width, only can access a filtering output value.So the length of data buffering storehouse should be at least greater than filter width.When only several filtering output value being arranged, can't so should can hold the full oscillatory process that two frequencies are 0.2Hz at least in the data buffering storehouse, promptly can hold 10 seconds real-time sampling data from wherein differentiating oscillation information.The raw data sampling rate is high more, and the raw data buffering storehouse that needs is long more.

Step3: determine the alarm thresholding.

Should set the alarm thresholding respectively according to monitored amount sampled value Changing Pattern.The alarm thresholding of setting according to low-frequency oscillation monitoring criterion needs has:

Max-Ever: oscillation characteristics method of identification thresholding 1 (more current maximum sampled value and history samples mean value);

Max-Min-Ever: oscillation characteristics method of identification thresholding 2 (difference of more current sampling maximin and history samples mean value);

Low-Alarm-Level: system fluctuation alarm thresholding;

High-Alarm-Level: low-frequency oscillation alarm thresholding;

High-Alarm-Delay: low-frequency oscillation alarm time-delay.

Step 2: obtain new sampled value, be pressed into raw data buffering storehouse.

In some cases, the new sampled value of acquisition possibly can't once be filled up raw data buffering storehouse.The buffer zone storehouse just can carry out subsequent operation after must pressing and expiring, otherwise will can't obtain correct operation result because data volume is not enough.Step 3: use " oscillation characteristics method of identification " and discern whether comprise oscillation information in the original sampling data roughly,, do not carry out follow-up monitoring algorithm process for the raw data that does not obviously comprise oscillation information.

The oscillation characteristics method of identification is divided into and is following three parts, and algorithm flow as shown in Figure 4.

Step1: more current maximum sampled value and history samples mean value.

Before calculating new sampled value and being pressed into storehouse, the mean value of sampled data in the storehouse, then be pressed into new sampled value, the traversal storehouse, obtain maximum sampled value, maximum sampled value and calculating averaging of income value are compared, if the absolute value of two numerical difference between, judges then that data can not comprise low-frequency oscillation information in the current stack less than default thresholding Max-Ever.

Step2: the difference of more current sampling maximin and history samples mean value.

Before calculating new sampled value and being pressed into storehouse, the mean value of sampled data in the storehouse, then be pressed into new sampled value, the traversal storehouse, obtain maximum, minimum sampled value, calculate the absolute value of maximum, minimum sampling value difference, calculate the absolute value of the difference of this absolute value and mean value again, if the absolute value that calculate this moment, judges then that data can not comprise low-frequency oscillation information in the current stack less than default thresholding Max-Min-Ever.

Step3: identification vibration equilibrium position, based on equilibrium position identification peak value and valley,, judge in the current data sequence it is to have the effective vibration that comprises The Characteristics of Low Frequency Oscillations according to frequency numerical value according to peak-to-peak value range estimation oscillation frequency.

Algorithm principle sees that shown in the accompanying drawing 5, algorithm flow as shown in Figure 6.

1) identification vibration equilibrium position

Utilize dichotomy, all sampled values in the buffer zone are sorted, the equilibrium position of sequence intermediate value as current vibration.

2) positive-negative half-cycle of division waveform

Try to achieve the relative amplitude sequence of sample sequence, determine several zero crossings with respect to the vibration equilibrium position.If the sampled value between two zero crossings, is then confirmed as this section sampled value the positive half cycle of vibration, otherwise then it is confirmed as the negative half period of vibration for just with respect to the vibration equilibrium position.

3) identification is based on the peak value and the valley of equilibrium position

Travel through each positive half cycle that vibrates successively, obtain maximum sampled value, deposit its absolute position in sample sequence in the peak sequence; Travel through each vibration negative half period successively, obtain minimum sampled value, deposit its absolute position in sample sequence in the valley position sequence.

4) calculate the peak-to-peak value distance, paddy-valley distance is judged oscillation frequency

Calculate in the peak sequence, the location interval of two consecutive values can be extrapolated two time intervals between the adjacent peak value by sample frequency and location interval, the institute's time-consuming that promptly once vibrates entirely, and can extrapolate the oscillation frequency of full vibration this time thus.

5) discern effective number of oscillation, judge whether vibration has The Characteristics of Low Frequency Oscillations

If oscillation frequency is positioned at 0.2Hz ~ 2.5Hz, think that then vibration this time is effectively vibration.Travel through whole peak sequence and valley position sequence, obtain effective number of oscillation of storehouse sample sequence.

Judge that according to the effective number of oscillation criterion that comprises The Characteristics of Low Frequency Oscillations in the sequence is: effective number of oscillation that current sample sequence is calculated is greater than setting value; Before new sampled value is pressed into effective number of oscillation NUM1 of calculating of data sequence and to new sampled value be pressed into that the back data sequence calculates effective number of oscillation NUM2's and greater than setting value.Be about to effective number of oscillation and default thresholding relatively,, think that then sample sequence may comprise oscillation information, need to continue it is carried out subsequent analysis if greater than this thresholding; If less than this thresholding, then the effective number of oscillation a that obtains this time analyzed in record.After the sampled value storehouse is pressed into new sampled value, will reanalyse effective number of oscillation b and a addition that obtains to sample sequence in the storehouse, and if greater than this thresholding, can think that also sample sequence may comprise oscillation information; If still less than this value, then replace a with b, repeat above-mentioned analytic process.

Step 4: oscillation frequency is the oscillating component of 0.2～2.5Hz in the NEURAL DISCHARGE BY DIGITAL FILTER extraction raw data.

As shown in Figure 7, when using digital filter that the sampled value in the raw data buffering storehouse is carried out filtering, use filtering window from front to back successively frame get one section continuous data in the storehouse, thereby obtain continuous filtering output value successively.But because filtering window has certain-length, so can not carry out filtering to all raw data.If filtering window length is L, L is an odd number, and the sampled value of each (L-1)/2 of raw data buffering storehouse head and afterbody can not obtain filtering so.The filtering data result will be pressed into " data buffering storehouse after the filtering " successively.

Step 5: use " crest indirect identification method " amplitude of oscillation of identification filter outputting data signals.

The basic thought of crest indirect identification method is to utilize sinusoidal features of shape, by the current amplitude of oscillation scope of the numerical estimation of each sampled point in several cycles on the curve.

As shown in Figure 8: the equal interval sampling that is 10 cosine curve to an amplitude in half or complete oscillation period, can see that 1/3 sampled point numerical value drops in the interval of [5,5], shown in gray area among the figure.Can obtain as drawing a conclusion: if current sampling point all is positioned at half or a complete oscillation period of cosine curve, and the absolute value that 1/3 sampled value is only arranged is less than A, then deducibility, and the amplitude of current vibration is about 2A.

Algorithm flow is as shown in Figure 9:

Step1: identification vibration equilibrium position

Utilize dichotomy, all numerical value in the data buffering storehouse after the filtering are sorted, the equilibrium position of sequence intermediate value as current vibration.

Step2: the truncated data sequence makes data sequence length comprise several complete vibration half cycles

Try to achieve the relative amplitude sequence of data sequence, determine several zero crossings with respect to the vibration equilibrium position.With first zero crossing is starting point, and last zero crossing is a terminal point, and the truncated data sequence makes data sequence length comprise several complete vibration half cycles.

Step3: according to the ultimate principle identification amplitude of oscillation of crest indirect identification method

Use the data sequence ordering of dichotomy to intercepting, find a sampled value A, making only has 1/3 numerical value to be worth less than this in the sequence, and the amplitude of oscillation of identification is 2A.

Step 6:, judge whether prompt alarm with gained " amplitude " and " alarm thresholding " comparison.

The amplitude of identification is compared with alarm thresholding Low-Alarm-Level, if amplitude less than this thresholding, prompt alarm not then; If amplitude greater than this thresholding, is then pointed out " system fluctuation alarm ".

When prompting " system fluctuation alarm ", with the amplitude and alarm thresholding High-Alarm-Level comparison of identification, if amplitude then returns less than this thresholding, be written into new sampled value, the analysis of a beginning new round again; If amplitude greater than this numerical value, then adds 1 with " low-frequency oscillation alarm delay accumulation device ".In back to back lower whorl is analyzed, if the amplitude of identification then with the totalizer zero clearing, returns less than alarm thresholding High-Alarm-Level, be written into new sampled value, the analysis of a beginning new round; If the amplitude of identification greater than alarm thresholding High-Alarm-Level, then adds 1 with " low-frequency oscillation alarm delay accumulation device " again.If accumulator value equals low-frequency oscillation alarm time-delay thresholding High-Alarm-Delay, then prompting " low-frequency oscillation alarm ".

Algorithm flow is referring to accompanying drawing 10.

Step 7: when satisfying alarm conditions, the data that amount to 1 minute before and after the alarm are constantly recorded ripple, packing data calls data in the off-line analysis Algorithm Analysis bag automatically automatically.

Invention has been described according to specific exemplary embodiment herein.It will be conspicuous carrying out suitable replacement to one skilled in the art or revise under not departing from the scope of the present invention.Exemplary embodiment only is illustrative, rather than to the restriction of scope of the present invention, scope of the present invention is by appended claim definition.

Claims (1)

1, a kind of low-frequency oscillation monitoring method for power system is characterized in that: may further comprise the steps:

1) at first the variation track of one section active power data in the data buffer zone is analyzed, amplitude, the cycle of preliminary identification oscillating curve, recognition result and The Characteristics of Low Frequency Oscillations are compared, if the identification amount does not obviously meet oscillation characteristics, then upgrade the content in the data buffer, one section new active power data are repeated above-mentioned analytic process,, then carry out subsequent analysis if the identification amount meets the fundamental oscillation feature;

2) utilize the digital filter algorithm with regard to described active power data segment, extract the oscillating component of oscillation frequency between 0.2～2.5Hz in the described active power data segment;

3) the above-mentioned oscillating component of just extracting is used crest indirect identification method identification oscillation amplitude, when oscillation amplitude surpasses the alarming threshold value, proposes alarm;

4) described alarming threshold value is divided into two sections, when satisfying low alarming threshold value, and prompting " system fluctuation " alarm, when satisfying higher alarming threshold value continuously several times, prompting " low-frequency oscillation " alarm.

Images