CN102680243B - Online judgment method for steam flow shock excitation fault of steam turbine generator unit - Google Patents

Abstract

The invention discloses an online judgment method for a steam flow shock excitation fault of a steam turbine generator unit in the technical field of state monitoring and fault diagnosis of rotary mechanical vibration. The online judgment method comprises the following steps of: acquiring relative vibration data of a shaft for supporting a bearing on one side of a high pressure rotor of the steam turbine generator unit, a rotating speed signal of a rotor, a key-phase signal of the rotor and power data of the unit; forming a unit power data sequence, a low-frequency vibration amplitude maximum sequence and a entropy sequence of the low-frequency vibration amplitude sequence; calculating an increasing trend parameter of the unit power data sequence, a maximum value of the low-frequency vibration amplitude maximum sequence, an entropy sequence skewness parameter of the low-frequency vibration amplitude sequence and Kendall relevant parameters of the entropy sequence of the low-frequency vibration amplitude sequence and the unit power data sequence; and finally, judging whether the steam flow shock excitation fault occurs or not by using the result. According to the online judgment method disclosed by the invention, the automatic real-time online monitoring and judgment of the steam flow shock excitation fault of the high pressure rotor are realized and the efficiency and accuracy of analysis and diagnosis are improved.

Description

Turbo-generator Set steam flow excitation On-line Fault method of discrimination

Technical field

The invention belongs to rotating machinery vibrating condition monitoring and fault diagnosis technical field, relate in particular to a kind of Turbo-generator Set steam flow excitation On-line Fault method of discrimination.

Background technology

Steam flow excitation be a kind of usually occur in large-size steam turbine high (in) press epitrochanterian, the low-frequency vibration phenomenon of being brought out by steam exciting force.Due to the raising of high pressure high temperature turbosets along with turbine steam condition, can cause the increase of high pressure cylinder admission density, flow velocity to improve, the tangential force of vapor action on high pressure rotor improves the sensitivity of dynamic and static gaps, hermetically-sealed construction and rotor and cylinder alignment degree, increase the exciting force that acts on high pressure rotor, directly affected the available rate of unit.

Unit generation steam flow excitation fault, often shows as low-frequency vibration and becomes suddenly the high load capacity stage that large and steam flow unit fault often occurs in unit.Judge whether unit steam flow excitation fault occurs, conventionally by the professional with certain field operation experiences and professional knowledge technical ability, completed, bring thus that analysis result is higher to personnel's subjectivity degree of dependence, the analytic process labor intensive resource time, analytical work high in cost of production problem, and cannot accomplish real-time automatic on-line monitoring, the analysis of steam flow excitation fault and differentiate.Therefore, propose a kind of large turbo-type generator group steam flow excitation On-line Fault method of discrimination and just seem very important.

Large turbo-type generator group steam flow excitation On-line Fault method of discrimination provided by the invention, the data such as unit operation rotor axle Relative Vibration, the power of the assembling unit are carried out to real-time automatic on-line monitoring, analysis and differentiation, judge whether high pressure rotor steam flow excitation fault occurs, improve efficiency and the accuracy of high pressure rotor steam flow excitation fault analysis and diagnosis work.

Summary of the invention

The object of the invention is to, propose a kind of Turbo-generator Set steam flow excitation On-line Fault method of discrimination, in order to realize real-time automatic on-line monitoring, the analysis of large turbo-type generator group steam flow excitation fault and to differentiate.

For achieving the above object, technical scheme provided by the invention is that a kind of Turbo-generator Set steam flow excitation On-line Fault method of discrimination, is characterized in that described method comprises:

Step 1: set initial time T _s, very first time stepping length t ₁, the second time stepping length t ₂, the first setting value D ₁with the second setting value D ₂;

Step 2: axle Relative Vibration data, the tach signal of rotor, the key signal of rotor and the power of the assembling unit data of Real-time Collection Turbo-generator Set high pressure rotor one side radial journal bearing;

Step 3: from initial time T _sstart, every very first time stepping length t ₁the power of the assembling unit data P of storage current time ^u; Every the second time stepping length t ₂, the maximal value A of calculating current time low-frequency vibration amplitude sequence ^fmax, the entropy E of current time low-frequency vibration amplitude sequence and the variation kurtosis parameter κ of current time low-frequency vibration amplitude sequence ^a, and store the maximal value A of current time low-frequency vibration amplitude sequence ^fmaxentropy E with current time low-frequency vibration amplitude sequence;

When meeting the variation kurtosis parameter κ of current time low-frequency vibration amplitude sequence ^abe greater than the first setting value D1 and current time and initial time T _sdifference be more than or equal to the second setting value D ₂with the first stepping length t ₁product time, using current time as stopping T constantly _n, by the power of the assembling unit data P of the current time of storage ^uas stopping T constantly _npower of the assembling unit data execution step 4;

Step 4: choose and stop T constantly _nin two moment before, be designated as respectively the first T constantly formerly ₁with the second T constantly formerly ₂, and meet

Step 5: will be from the first T constantly formerly ₁rise to stopping T constantly _nonly, every very first time stepping length t ₁the power of the assembling unit data P of storage ^uaccording to the sequencing of storage time, line up power of the assembling unit data sequence will be from the second T constantly formerly ₂rise to stopping T constantly _nonly, every the second stepping length t ₂the maximal value A of the low-frequency vibration amplitude sequence of storage ^fmaxaccording to the sequencing of storage time, line up low-frequency vibration amplitude maximal value sequence every the second stepping length t ₂the entropy E of the low-frequency vibration amplitude sequence of storage lines up the entropy sequence { E of low-frequency vibration amplitude sequence according to the sequencing of storage time _j, $j=\mathrm{1,2},...,\frac{T_N-T_2}{t_2};$

Step 6: calculate unit power parameter and low-frequency vibration parameter, comprising:

1) that calculates unit power data sequence increases progressively trend parameter I ^p;

2) calculate the maximal value of low-frequency vibration amplitude maximal value sequence

3) calculate the variation degree of bias parameter S of the entropy sequence of low-frequency vibration amplitude sequence ^e;

4) calculate the entropy sequence of low-frequency vibration amplitude sequence and the Kendall related coefficient τ of power of the assembling unit data sequence;

Step 7: according to stopping T constantly _npower of the assembling unit data power of the assembling unit data sequence increase progressively trend parameter I ^p, low-frequency vibration amplitude maximal value sequence maximal value the variation degree of bias parameter S of the entropy sequence of low-frequency vibration amplitude sequence ^ekendall related coefficient τ with entropy sequence and the power of the assembling unit data sequence of low-frequency vibration amplitude sequence, judges whether high pressure rotor steam flow excitation fault occurs.

The entropy E of described calculating current time low-frequency vibration amplitude sequence adopts formula

$E=\underset{k=1}{\overset{l}{\mathrm{\Σ}}}\left[{\left(A_k^{\mathrm{freq}}\right)}^2\ln \left({\left(A_k^{\mathrm{freq}}\right)}^2\right)\right],$

Wherein, for k data of low-frequency vibration amplitude sequence, k=1,2 ... l, l is the data amount check of predefined low-frequency vibration amplitude sequence, and regulation is worked as time,

The variation kurtosis parameter κ of described current time low-frequency vibration amplitude sequence ^aadopt formula

${\mathrm{\κ}}^A=1/l\underset{k=1}{\overset{l}{\mathrm{\Σ}}}{(A_k^{\mathrm{freq}}-{\mathrm{\μ}}^A)}^4/{\left({\mathrm{\σ}}^A\right)}^4,$

Wherein, for k data of low-frequency vibration amplitude sequence, k=1,2 ... l, l is the data amount check of predefined low-frequency vibration amplitude sequence, μ ^athe average of low-frequency vibration amplitude sequence, σ ^athe standard deviation of low-frequency vibration amplitude sequence,

Described calculating unit power data sequence increase progressively trend parameter I ^padopt formula

I ^P=S ^P/[1/2×n×(n-1)]，

Wherein, n is the data amount check of power of the assembling unit data sequence, S ^pit is the Ser.No. of power of the assembling unit data sequence; Described Ser.No. refers to the right sum of order in a data sequence; Described order is to referring to that the front and back position of a logarithm is identical with size order in a data sequence, and number is above less than number below.

The variation degree of bias parameter S of the entropy sequence of described calculating low-frequency vibration amplitude sequence ^eadopt formula

$S^E=1/n\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{(E_j-{\mathrm{\μ}}^E)}^3/{\left({\mathrm{\σ}}^E\right)}^3,$

Wherein, E _jj data of the entropy sequence of low-frequency vibration amplitude sequence, j=1,2 ..., n, μ ^ethe average of the entropy sequence of low-frequency vibration amplitude sequence, σ ^ethe standard deviation of the entropy sequence of low-frequency vibration amplitude sequence, n is the data amount check of the entropy sequence of low-frequency vibration amplitude sequence.

The entropy sequence of described calculating low-frequency vibration amplitude sequence and the Kendall related coefficient τ of power of the assembling unit data sequence adopt formula

τ=(n _s-n _d)/[1/2×(n ²-1)]，

Wherein, n is the data amount check of entropy sequence or the power of the assembling unit data sequence of low-frequency vibration amplitude sequence, n _sin the entropy sequence and power of the assembling unit data sequence of low-frequency vibration amplitude sequence, the right sum of sequence that mediation is arranged, n _din the entropy sequence and power of the assembling unit data sequence of low-frequency vibration amplitude sequence, the right sum of sequence of anharmonic arrangement;

The sequence that described mediation is arranged is to referring to, for 2 sequence { A _kand { B _kin any 2 sequences to (A _m, B _m) and (A _n, B _n), q>=m, n>=1, q is sequence { A _kand sequence { B _kdata amount check, if meet A simultaneously _m>A _nand B _m>B _nor meet A simultaneously _m<A _nand B _m<B _n, 2 sequences are to (A _m, B _m) and (A _n, B _n) be to be in harmonious proportion the sequence pair of arranging;

The sequence of described anharmonic arrangement is to referring to, for 2 sequence { A _kand { B _kin any 2 sequences to (A _m, B _m) and (A _n, B _n), q>=m, n>=1, q is sequence { A _kand sequence { B _kdata amount check, if meet A simultaneously _m>A _nand B _m<B _nor meet A simultaneously _m<A _nand B _m>B _n, 2 sequences are to (A _m, B _m) and (A _n, B _n) be the sequence pair of anharmonic arrangement.

Described step 7 specifically, meets following 5 conditions when simultaneously:

1) stop T constantly _npower of the assembling unit data be greater than the 3rd setting value;

2) power of the assembling unit data sequence increases progressively trend parameter I ^pbe greater than the 4th setting value;

3) maximal value of low-frequency vibration amplitude maximal value sequence be greater than the 5th setting value;

4) the variation degree of bias parameter of the entropy sequence of low-frequency vibration amplitude sequence is greater than S ^ebe greater than the 6th setting value;

5) the Kendall related coefficient τ of the entropy sequence of low-frequency vibration amplitude sequence and power of the assembling unit data sequence is greater than the 7th setting value;

Judge high pressure rotor generation steam flow excitation fault; Otherwise, judge that steam flow excitation fault does not occur high pressure rotor.

Method provided by the invention, utilizes the data such as unit operation rotor axle Relative Vibration, the power of the assembling unit, through computational discrimination high pressure rotor, whether steam flow excitation fault occurs, and has realized automatic real time on-line monitoring, analysis and the differentiation of steam flow excitation fault.

Accompanying drawing explanation

Fig. 1 is Turbo-generator Set steam flow excitation On-line Fault method of discrimination process flow diagram;

Fig. 2 is that Turbo-generator Set steam flow excitation On-line Fault is differentiated schematic diagram.

Embodiment

Below in conjunction with accompanying drawing, preferred embodiment is elaborated.Should be emphasized that, following explanation is only exemplary, rather than in order to limit the scope of the invention and to apply.

Embodiment

Fig. 1 is Turbo-generator Set steam flow excitation On-line Fault method of discrimination process flow diagram.In Fig. 1, Turbo-generator Set steam flow excitation On-line Fault method of discrimination comprises:

Step 101: set initial time T _s=0 second, very first time stepping length t ₁=3 seconds, the second time stepping length t ₂=0.1 second, the first setting value D ₁the=30 and second setting value D ₂=300.

Step 102: axle Relative Vibration data, the tach signal of rotor, the key signal of rotor and the power of the assembling unit data of Real-time Collection Turbo-generator Set high pressure rotor one side radial journal bearing.

The tach signal of armature spindle Relative Vibration data, rotor and key signal can obtain from the supervisory instrument (TSI) of configuration Turbo-generator Set, and power of the assembling unit data-signal can obtain from the dcs (DCS) of configuration Turbo-generator Set.In the present embodiment, the tach signal of armature spindle Relative Vibration data, rotor and key signal are to obtain from the supervisory instrument (TSI) of configuration Turbo-generator Set, and power of the assembling unit data-signal is to obtain from the dcs (DCS) of configuration Turbo-generator Set.Fig. 2 is that Turbo-generator Set steam flow excitation On-line Fault is differentiated schematic diagram, as shown in Figure 2, and in the slot that data collecting card insertion industrial microcomputer (IPC) provides.According to the requirement of data collecting card, data acquisition conditioning device is processed axle Relative Vibration signal, the tach signal of rotor, the key signal from Turbo-generator Set supervisory instrument (TSI), the vibration at high speed data collecting card in axle Relative Vibration signal after treatment, the tach signal of rotor, key signal input IPC.Each passage technology parameter of vibrating data collection card is 50ks/s, 24bit.Meanwhile, data acquisition conditioning device is processed the power of the assembling unit data-signal from Turbo-generator Set dcs (DCS), the data collecting card in bearing oil temperature data signal input IPC after treatment.Each passage technology parameter of data collecting card is 1ks/s, 16bit.

According to the method, design concrete Turbo-generator Set steam flow excitation On-line Fault discriminating program, online discriminating program is arranged in industrial microcomputer (IPC).Once diagnosis cyclic process in Turbo-generator Set steam flow excitation On-line Fault discriminating program, comprises that the real time data acquisition that relates in diagnostic method calculates that storage, real time discriminating, power of the assembling unit data parameters are calculated, low-frequency vibration parameter calculates in real time, related coefficient is calculated in real time and the series of computation such as fault verification is analyzed link.

Step 103: from initial time T _c=0 second starts, every very first time stepping length t ₁the power of the assembling unit data P of=3 seconds storage current times ^u.As shown in Figure 2, power of the assembling unit data obtain from the dcs (DCS) of configuration Turbo-generator Set.

Every the second time stepping length t ₂=0.1 second, calculate current time low-frequency vibration amplitude sequence.As shown in Figure 2, the routine analyzer in industrial microcomputer (IPC) is by adopting near the axle Relative Vibration data that record vibration at high speed data collecting card Real-time Collection Turbo-generator Set high pressure rotor A side radial journal bearing, tach signal and the key signal of rotor.Each passage technology parameter of vibrating data collection card is 50ks/s, 24bit.For the axle Relative Vibration data of unit high pressure rotor A side, utilize FFT frequency spectrum analysis method, calculate the corresponding vibration amplitude data sequence of the different vibration frequencies of current time from low frequency to high frequency (amplitude unit is μ m, micron).For vibration at high speed data collecting card, each constantly can collect the corresponding vibration amplitude data sequence of different vibration frequencies from low frequency to high frequency, therefrom intercepting obtains the current time low-frequency vibration amplitude sequence that frequency is less than unit working speed respective frequencies (50Hz), is designated as can set vibrating data collection frequency and image data amount, make low-frequency vibration amplitude sequence data number l=98.The axle Relative Vibration data, the tach signal of rotor and the key signal of rotor that utilize the Turbo-generator Set high pressure rotor one side radial journal bearing of current time collection, calculate current time low-frequency vibration amplitude sequence be the conventional technology of those skilled in the art, repeated no more in the present invention.

In the low-frequency vibration amplitude sequence that obtains current time after, can calculate the maximal value A of current time low-frequency vibration amplitude sequence ^fmax, its computing formula is in the present embodiment, l=98.

According to the low-frequency vibration amplitude sequence of current time can also calculate the entropy E of current time low-frequency vibration amplitude sequence and the variation kurtosis parameter κ of current time low-frequency vibration amplitude sequence ^a.

According to the low-frequency vibration amplitude sequence of current time the entropy E that calculates current time low-frequency vibration amplitude sequence adopts formula

$E=\underset{k=1}{\overset{l}{\mathrm{\&Sigma;}}}\left[{\left(A_k^{\mathrm{freq}}\right)}^2\ln \left({\left(A_k^{\mathrm{freq}}\right)}^2\right)\right]---\left(1\right)$

In formula, for low-frequency vibration amplitude sequence k data, k=1,2 ... l, l=98, and regulation is worked as ${\left(A_k^{\mathrm{freq}}\right)}^2=0$ Time, $\ln \left({\left(A_k^{\mathrm{freq}}\right)}^2\right)=0.$

According to the low-frequency vibration amplitude sequence of current time calculate the variation kurtosis parameter κ of current time low-frequency vibration amplitude sequence ^aadopt formula

${\mathrm{\&kappa;}}^A=1/l\underset{k=1}{\overset{l}{\mathrm{\&Sigma;}}}{(A_k^{\mathrm{freq}}-{\mathrm{\&mu;}}^A)}^4/{\left({\mathrm{\&sigma;}}^A\right)}^4---\left(2\right)$

In formula, for low-frequency vibration amplitude sequence k data, k=1,2 ... l, l=98, μ ^ait is low-frequency vibration amplitude sequence average, σ ^ait is low-frequency vibration amplitude sequence standard deviation, ${\mathrm{\&sigma;}}^A=\sqrt{1/l\underset{k=1}{\overset{l}{\mathrm{\&Sigma;}}}{(A_k^{\mathrm{freq}}-{\mathrm{\&mu;}}^A)}^2}.$

Every the second time stepping length t ₂=0.1 second, calculate the maximal value A of current time low-frequency vibration amplitude sequence ^fmaxafter the entropy E of current time low-frequency vibration amplitude sequence, all to store, to use in subsequent calculations process.

Judge whether to satisfy condition simultaneously:

1) the variation kurtosis parameter κ of current time low-frequency vibration amplitude sequence ^abe greater than the first setting value D ₁, i.e. κ ^a>D ₁=30;

2) current time and initial time T _sdifference be more than or equal to the second setting value D ₂with the first stepping length t ₁product, i.e. T _c-T _s>=D ₂* t ₁=10 * 30=300, T _cfor current time.

If satisfy condition 1 simultaneously) and 2), by current time T _cas stopping T constantly _n, by the current time T of storage _cpower of the assembling unit data P ^uas stopping T constantly _npower of the assembling unit data

In the present embodiment, suppose to work as T _cin the time of=300 seconds, κ ^a>30, and T _c-T _s=300-0>=300, stop T constantly _n=300 seconds, stop T constantly _nthe power of the assembling unit data of=300 seconds are

Step 104: choose and stop T constantly _nin two moment before, be designated as respectively the first T constantly formerly ₁with the second T constantly formerly ₂, and meet

In the present embodiment, choose the first T constantly formerly ₁=0 second, the second T constantly formerly ₂=290 seconds. $\frac{T_N-T_1}{t_1}=\frac{300-0}{3}=100,$ $\frac{T_N-T_2}{t_2}=\frac{300-290}{0.1}=100,$ Satisfy condition $\frac{T_N-T_1}{t_1}=\frac{T_N-T_2}{t_2}.$

Step 105: will be from the first T constantly formerly ₁within=0 second, rise to stopping T constantly _n=300 seconds only, every very first time stepping length t ₁the power of the assembling unit data P of storage in=3 seconds ^uaccording to the sequencing of storage time, line up power of the assembling unit data sequence owing to storing a secondary data from 0 second to 300 seconds every 3 seconds, therefore coexist and stored up 100 power of the assembling unit data, power of the assembling unit data sequence data amount check be 100, i.e. i=1,2 ... 100.

Will be from the second T constantly formerly ₂within=290 seconds, rise to stopping T constantly _n=300 seconds only, every the second stepping length t ₂the maximal value A of the low-frequency vibration amplitude sequence of storage in=0.1 second ^fmaxaccording to the sequencing of storage time, line up low-frequency vibration amplitude maximal value sequence every the second stepping length t ₂the entropy E of the low-frequency vibration amplitude sequence of storage in=0.1 second lines up the entropy sequence { E of low-frequency vibration amplitude sequence according to the sequencing of storage time _j, owing to storing a secondary data from 290 seconds to 300 seconds every 0.1 second, therefore coexist and stored up the entropy of 100 low-frequency vibration amplitude maximal values and 100 low-frequency vibration amplitude sequences, so low-frequency vibration amplitude maximal value sequence entropy sequence { E with low-frequency vibration amplitude sequence _jdata amount check be all 100, i.e. j=1,2 ... 100.

Step 106: calculate unit power parameter and low-frequency vibration parameter, comprising:

1) that calculates unit power data sequence increases progressively trend parameter I ^p.

That calculates unit power data sequence increases progressively trend parameter I ^padopt formula

I ^P=S ^P/[1/2×n×(n-1)] （3）

In formula, n is the data amount check of power of the assembling unit data sequence, in the present embodiment, and n=100.S ^pit is the Ser.No. of power of the assembling unit data sequence.Ser.No. refers to the right sum of order in a data sequence.Order is to referring to that the front and back position of a logarithm is identical with size order in a data sequence, and the number before is less than number below.

2) calculate the maximal value of low-frequency vibration amplitude maximal value sequence

The maximal value of low-frequency vibration amplitude maximal value sequence is utilized formula in the present embodiment, n=100.

3) calculate the variation degree of bias parameter S of the entropy sequence of low-frequency vibration amplitude sequence ^e.

Calculate the variation degree of bias parameter S of the entropy sequence of low-frequency vibration amplitude sequence ^eadopt formula

$S^E=1/n\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{(E_j-{\mathrm{\&mu;}}^E)}^3/{\left({\mathrm{\&sigma;}}^E\right)}^3---\left(4\right)$

In formula, E _jj data of the entropy sequence of low-frequency vibration amplitude sequence, j=1,2 ..., n, μ ^ethe average of the entropy sequence of low-frequency vibration amplitude sequence, σ ^ethe standard deviation of the entropy sequence of low-frequency vibration amplitude sequence, n is the data amount check of the entropy sequence of low-frequency vibration amplitude sequence.In the present embodiment, n=100.

4) calculate the entropy sequence of low-frequency vibration amplitude sequence and the Kendall related coefficient τ of power of the assembling unit data sequence.

Calculate the Kendall related coefficient τ employing formula of entropy sequence and the power of the assembling unit data sequence of low-frequency vibration amplitude sequence

τ=(n _s-n _d)/[1/2×(n ²-1)] （5）

In formula, n is the data amount check of entropy sequence or the power of the assembling unit data sequence of low-frequency vibration amplitude sequence, in the present embodiment, and n=100.N _sin the entropy sequence and power of the assembling unit data sequence of low-frequency vibration amplitude sequence, the right sum of sequence that mediation is arranged, n _din the entropy sequence and power of the assembling unit data sequence of low-frequency vibration amplitude sequence, the right sum of sequence of anharmonic arrangement.

The sequence that mediation is arranged is to referring to, for 2 sequence { A _kand { B _kin any 2 sequences to (A _m, B _m) and (A _n, B _n), q>=m, n>=1, q is sequence { A _kand sequence { B _kdata amount check, if meet A simultaneously _m>A _nand B _m>B _nor meet A simultaneously _m<A _nand B _m<B _n, 2 sequences are to (A _m, B _m) and (A _n, B _n) be to be in harmonious proportion the sequence pair of arranging.The sequence of anharmonic arrangement is to referring to, for 2 sequence { A _kand { B _kin any 2 sequences to (A _m, B _m) and (A _n, B _n), q>=m, n>=1, q is sequence { A _kand sequence { B _kdata amount check, if meet A simultaneously _m>A _nand B _m<B _nor meet A simultaneously _m<A _nand B _m>B _n, 2 sequences are to (A _m, B _m) and (A _n, B _n) be the sequence pair of anharmonic arrangement.For 2 sequence { A _kand { B _kin any 2 sequences to (A _m, B _m) and (A _n, B _n), Q>=m, n>=1, Q is sequence { A _kor sequence { B _kdata amount check, if meet A _m=A _nor B _m=B _n, 2 sequences are to (A _m, B _m) and (A _n, B _n) neither mediation collating sequence pair, neither anharmonic collating sequence pair.

In the present invention, for the entropy sequence { E of low-frequency vibration amplitude sequence _iand power of the assembling unit data sequence in any 2 sequences pair with 100>=m, n>=1, if meet E simultaneously _m>E _nand or meet E simultaneously _m<E _nand these 2 sequences pair with to be in harmonious proportion the sequence pair of arranging.Entropy sequence { E for low-frequency vibration amplitude sequence _iand power of the assembling unit data sequence in any 2 sequences pair with 100>=m, n>=1, if meet E simultaneously _m>E _nand or meet E simultaneously _m<E _nand these 2 sequences pair with it is the sequence pair of anharmonic arrangement.Entropy sequence { E for low-frequency vibration amplitude sequence _iand power of the assembling unit data sequence in any 2 sequences pair with 100>=m, n>=1, if meet E _m=E _nor these 2 sequences pair with neither mediation collating sequence pair, neither anharmonic collating sequence pair.

Step 107: according to stopping the power of the assembling unit data of TN constantly power of the assembling unit data sequence increase progressively trend parameter I ^p, low-frequency vibration amplitude maximal value sequence maximal value the variation degree of bias parameter S of the entropy sequence of low-frequency vibration amplitude sequence ^ekendall related coefficient τ with entropy sequence and the power of the assembling unit data sequence of low-frequency vibration amplitude sequence, judges whether high pressure rotor steam flow excitation fault occurs.

Set respectively the 3rd setting value D ₃=150MW(megawatt), the 4th setting value D ₄the=0.8, five setting value D ₅=30 μ m(microns), the 6th setting value D ₆the=1.5 and the 7th setting value D ₇=0.7.Above-mentioned setting value is used for assisting to judge whether high pressure rotor steam flow excitation fault occurs, and each setting value is according to high pressure rotor running requirements and the standard of Turbo-generator Set.

When simultaneously, meet following 5 conditions:

1) stop T constantly _npower of the assembling unit data be greater than the 3rd setting value,

2) power of the assembling unit data sequence increases progressively trend parameter I ^pbe greater than the 4th setting value, i.e. I ^p>D ₄=0.8;

3) maximal value of low-frequency vibration amplitude maximal value sequence be greater than the 5th setting value,

$A_{\max }^{f\max }>D_5=30\mathrm{\&mu;m};$

4) the variation degree of bias parameter of the entropy sequence of low-frequency vibration amplitude sequence is greater than S ^ebe greater than the 6th setting value, i.e. S ^e>D ₆=1.5;

5) the Kendall related coefficient τ of the entropy sequence of low-frequency vibration amplitude sequence and power of the assembling unit data sequence is greater than the 7th setting value, i.e. τ >D ₇=0.7;

Judge high pressure rotor generation steam flow excitation fault; Otherwise, judge that steam flow excitation fault does not occur high pressure rotor.

Turbo-generator Set steam flow excitation On-line Fault method of discrimination provided by the invention, the data such as unit operation rotor axle Relative Vibration, the power of the assembling unit are carried out to real-time automatic on-line monitoring, analysis and differentiation, judge whether high pressure rotor steam flow excitation fault occurs, improve efficiency and the accuracy of high pressure rotor steam flow excitation fault analysis and diagnosis work.

The above; be only the present invention's embodiment preferably, but protection scope of the present invention is not limited to this, is anyly familiar with in technical scope that those skilled in the art disclose in the present invention; the variation that can expect easily or replacement, within all should being encompassed in protection scope of the present invention.Therefore, protection scope of the present invention should be as the criterion with the protection domain of claim.

Claims (6)

1. a Turbo-generator Set steam flow excitation On-line Fault method of discrimination, is characterized in that described method comprises:

Step 1: set initial time T _s, very first time stepping length t ₁, the second time stepping length t ₂, the first setting value D ₁with the second setting value D ₂;

Step 2: axle Relative Vibration data, the tach signal of rotor, the key signal of rotor and the power of the assembling unit data of Real-time Collection Turbo-generator Set high pressure rotor one side radial journal bearing;

Step 3: from initial time T _sstart, every very first time stepping length t ₁the power of the assembling unit data P of storage current time ^u; Every the second time stepping length t ₂, the maximal value A of calculating current time low-frequency vibration amplitude sequence ^fmax, the entropy E of current time low-frequency vibration amplitude sequence and the variation kurtosis parameter κ of current time low-frequency vibration amplitude sequence ^a, and store the maximal value A of current time low-frequency vibration amplitude sequence ^fmaxentropy E with current time low-frequency vibration amplitude sequence;

When meeting the variation kurtosis parameter κ of current time low-frequency vibration amplitude sequence ^abe greater than the first setting value D ₁and current time and initial time T _sdifference be more than or equal to the second setting value D ₂with the first stepping length t ₁product time, using current time as stopping T constantly _n, by the power of the assembling unit data P of the current time of storage ^uas stopping T constantly _npower of the assembling unit data execution step 4;

Step 4: choose and stop T constantly _nin two moment before, be designated as respectively the first T constantly formerly ₁with the second T constantly formerly ₂, and meet

Step 5: will be from the first T constantly formerly ₁rise to stopping T constantly _nonly, every very first time stepping length t ₁the power of the assembling unit data P of storage ^uaccording to the sequencing of storage time, line up power of the assembling unit data sequence { P _i ^u, will be from the second T constantly formerly ₂rise to stopping T constantly _nonly, every the second stepping length t ₂the maximal value A of the low-frequency vibration amplitude sequence of storage ^fmaxaccording to the sequencing of storage time, line up low-frequency vibration amplitude maximal value sequence every the second stepping length t ₂the entropy E of the low-frequency vibration amplitude sequence of storage lines up the entropy sequence { E of low-frequency vibration amplitude sequence according to the sequencing of storage time _j,

Step 6: calculate unit power parameter and low-frequency vibration parameter, comprising:

1) that calculates unit power data sequence increases progressively trend parameter I ^p;

2) calculate the maximal value of low-frequency vibration amplitude maximal value sequence

3) calculate the variation degree of bias parameter S of the entropy sequence of low-frequency vibration amplitude sequence ^e;

4) calculate the entropy sequence of low-frequency vibration amplitude sequence and the Kendall related coefficient τ of power of the assembling unit data sequence;

Step 7: according to stopping T constantly _npower of the assembling unit data power of the assembling unit data sequence increase progressively trend parameter I ^p, low-frequency vibration amplitude maximal value sequence maximal value the variation degree of bias parameter S of the entropy sequence of low-frequency vibration amplitude sequence ^ekendall related coefficient τ with entropy sequence and the power of the assembling unit data sequence of low-frequency vibration amplitude sequence, judges whether high pressure rotor steam flow excitation fault occurs, and specifically, meets following 5 conditions when simultaneously:

1) stop T constantly _npower of the assembling unit data be greater than the 3rd setting value;

2) power of the assembling unit data sequence increases progressively trend parameter I ^pbe greater than the 4th setting value;

3) maximal value of low-frequency vibration amplitude maximal value sequence be greater than the 5th setting value;

4) the variation degree of bias parameter of the entropy sequence of low-frequency vibration amplitude sequence is greater than S ^ebe greater than the 6th setting value;

5) the Kendall related coefficient τ of the entropy sequence of low-frequency vibration amplitude sequence and power of the assembling unit data sequence is greater than the 7th setting value;

Judge high pressure rotor generation steam flow excitation fault; Otherwise, judge that steam flow excitation fault does not occur high pressure rotor.

2. Turbo-generator Set steam flow excitation On-line Fault method of discrimination according to claim 1, is characterized in that the entropy E of described calculating current time low-frequency vibration amplitude sequence adopts formula

$E=\underset{k=1}{\overset{l}{\mathrm{\&Sigma;}}}\left[{\left(A_k^{\mathrm{freq}}\right)}^2\ln \left({\left(A_k^{\mathrm{freq}}\right)}^2\right)\right],$

Wherein, for k data of low-frequency vibration amplitude sequence, k=1,2 ... l, l is the data amount check of predefined low-frequency vibration amplitude sequence, and regulation is worked as time,

3. Turbo-generator Set steam flow excitation On-line Fault method of discrimination according to claim 1, is characterized in that the variation kurtosis parameter κ of described current time low-frequency vibration amplitude sequence ^aadopt formula

${\mathrm{\&kappa;}}^A=1/l\underset{k=1}{\overset{l}{\mathrm{\&Sigma;}}}{(A_k^{\mathrm{freq}}-{\mathrm{\&mu;}}^A)}^4/{\left({\mathrm{\&sigma;}}^A\right)}^4,$

Wherein, for k data of low-frequency vibration amplitude sequence, k=1,2 ... l, l is the data amount check of predefined low-frequency vibration amplitude sequence, μ ^athe average of low-frequency vibration amplitude sequence, σ ^athe standard deviation of low-frequency vibration amplitude sequence,

4. Turbo-generator Set steam flow excitation On-line Fault method of discrimination according to claim 1, what it is characterized in that described calculating unit power data sequence increases progressively trend parameter I ^padopt formula

I ^P＝S ^P/[1/2×n×(n-1)]，

Wherein, n is the data amount check of power of the assembling unit data sequence, S ^pit is the Ser.No. of power of the assembling unit data sequence; Described Ser.No. refers to the right sum of order in a data sequence; Described order is to referring to that the front and back position of a logarithm is identical with size order in a data sequence, and number is above less than number below.

5. Turbo-generator Set steam flow excitation On-line Fault method of discrimination according to claim 1, is characterized in that the variation degree of bias parameter S of the entropy sequence of described calculating low-frequency vibration amplitude sequence ^eadopt formula

$S^E=1/n\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{(E_j-{\mathrm{\&mu;}}^E)}^3/{\left({\mathrm{\&sigma;}}^E\right)}^3,$

Wherein, E _jj data of the entropy sequence of low-frequency vibration amplitude sequence, j=1,2 ..., n, μ ^ethe average of the entropy sequence of low-frequency vibration amplitude sequence, σ ^ethe standard deviation of the entropy sequence of low-frequency vibration amplitude sequence, n is the data amount check of the entropy sequence of low-frequency vibration amplitude sequence.

6. Turbo-generator Set steam flow excitation On-line Fault method of discrimination according to claim 1, is characterized in that the entropy sequence of described calculating low-frequency vibration amplitude sequence and the Kendall related coefficient τ of power of the assembling unit data sequence adopt formula

τ＝(n _s-n _d)/[1/2×(n ²-1)]，

Wherein, n is the data amount check of entropy sequence or the power of the assembling unit data sequence of low-frequency vibration amplitude sequence, n _sin the entropy sequence and power of the assembling unit data sequence of low-frequency vibration amplitude sequence, the right sum of sequence that mediation is arranged, n _din the entropy sequence and power of the assembling unit data sequence of low-frequency vibration amplitude sequence, the right sum of sequence of anharmonic arrangement;

The sequence that described mediation is arranged is to referring to, for 2 sequence { A _kand { B _kin any 2 sequences to (A _m, B _m) and (A _n, B _n), q>=m, n>=1, q is sequence { A _kand sequence { B _kdata amount check, if meet A simultaneously _m>A _nand B _m>B _nor meet A simultaneously _m<A _nand B _m<B _n, 2 sequences are to (A _m, B _m) and (A _n, B _n) be to be in harmonious proportion the sequence pair of arranging;

The sequence of described anharmonic arrangement is to referring to, for 2 sequence { A _kand { B _kin any 2 sequences to (A _m, B _m) and (A _n, B _n), q>=m, n>=1, q is sequence { A _kand sequence { B _kdata amount check, if meet A simultaneously _m>A _nand B _m<B _nor meet A simultaneously _m<A _nand B _m>B _n, 2 sequences are to (A _m, B _m) and (A _n, B _n) be the sequence pair of anharmonic arrangement.