CN101042059A - Method and system for on-line monitoring steam turbine roter low-cycle fatigue life consumption - Google Patents

Abstract

It relates to a turbine rotor low cycle fatigue durability loss on line monitoring system, featuring in its computation, application server and software, database server, outside system interface, turbine control system DEH and parameter testing point, web server and user browser that connects with the computation and application server and each end item user browser respectively, with the computation and application server connecting with the data base server which connects with the turbine control system DEH and parameter testing point through the outside system interface. It is made of two processes, with the no.1 using artificial nerve network to determine turbine rotor equivalent stress coefficient, the no.2 process being on line computation, monitoring and control of the turbine rotor cycle fatigue loss. It realizes the working times of the turbine rotor of 10000 times.

Description

A kind of method and system of on-line monitoring steam turbine roter low-cycle fatigue life consumption

Technical field

The present invention relates to a kind of method and system of on-line monitoring steam turbine roter low-cycle fatigue life consumption, relate in particular to a kind of method of steam turbine roter low-cycle fatigue life consumption in line computation and in-service monitoring and control, be applied to the on line control of steam turbine roter low-cycle fatigue life consumption and the life-span online management of turbine rotor, belong to the steam turbine technology field.

Background technique

In the process of steam turbine startup, shutdown and load change; steam turbine steam inlet condition and flow passage component parameter change in time; inhomogeneous heating of turbine rotor or cooling; turbine rotor is the bigger thermal stress of the even generation of temperature distributing disproportionation radially, causes that the equivalent stress of turbine rotor and low-cycle fatigue life loss increase.If steam turbine roter low-cycle fatigue life consumption is controlled inappropriate, will cause shorten the working life of turbine rotor.In order to prolong the working life of turbine rotor, need be at the low-cycle fatigue life loss of line computation, supervision and control turbine rotor.It is 200610030243.5 as 200610030244.X and " equivalent stress safety margin coefficient " application number that prior art is used " at the relative stress of line computation and control steam turbine rotator equivalent stress " application number, control the low cycle fatigue loss of turbine rotor, owing in the relative stress of steam turbine rotator equivalent stress or the calculating of equivalent stress safety margin coefficient, used YIELD STRENGTH σ 0.2 and rupture strength σ _FrAnd do not adopt the low cycle fatigue curve of rotor material, thereby make the relative stress that adopts steam turbine rotator equivalent stress or equivalent stress safety margin coefficient become indirectly a kind of rather than directly control the service life management method of the low-cycle fatigue life loss of turbine rotor.

Summary of the invention

The method and system of on-line monitoring steam turbine roter low-cycle fatigue life consumption that the purpose of this invention is to provide the low-cycle fatigue life loss of a kind of online real-time calculating and in-service monitoring and control turbine rotor.

For realizing above purpose, technological scheme of the present invention provides a kind of on-line monitoring steam turbine roter low-cycle fatigue life consumption system, it is characterized in that, calculated/applied server and software by low-cycle fatigue life loss, database server, external system interface, steam turbine control system DEH and parameter measuring point, web page server and user side browser are formed, web page server is connected with each user side browser with the calculated/applied server respectively, the calculated/applied server is connected with database server, and database server is connected with the parameter measuring point with steam turbine control system DEH by external system interface.

Described steam turbine control system DEH and parameter measuring point, control system DEH are the supporting automatic control system of steam turbine, and the parameter measuring point is 9 parameters of steam turbine control system DEH on-line monitoring, comprises main steam pressure P ₀, main steam temperature t ₀, inner cylinder metal monitor temperature θ ₀, turbine speed n, steam turbine load N, preceding 1 minute to 5 minutes steam turbine main steam temperature variance ratio δ t ₀The variance ratio Δ t of the load changing rate δ N (MW/ branch) of (℃/minute), preceding 1 minute to 5 minutes steam turbine, preceding 6 minutes to 14 minutes steam turbine main steam temperature ₀The load changing rate Δ N (MW/ branch) of (℃/minute), preceding 6 minutes to 14 minutes steam turbine.

A kind of on-line monitoring steam turbine roter low-cycle fatigue life consumption method, it is characterized in that, adopt the computer software of the steam turbine roter low-cycle fatigue life consumption of C language compilation to operate on the calculated/applied server, be applied to the on-line monitoring of steam turbine roter low-cycle fatigue life consumption, the artificial neural network technology of 10 input layers of employing calculates the correction factor of the equivalent stress of turbine rotor, adopted the low cycle fatigue curve of rotor material and low-cycle fatigue life and the low-cycle fatigue life loss that the symmetrical cycle method is calculated turbine rotor, its method is made up of two flow processs:

First pass: adopt artificial neural network technology to determine the correction factor of steam turbine rotator equivalent stress

The first step: adopt the simplified model of prior art to calculate rotor name compound stress

Adopt the temperature field of the simplified model calculated off-line turbine rotor of prior art, the turbine rotor that center hole is arranged such as is reduced at the heavy wall cylindrical model, the turbine rotor of no center hole is reduced to the equal diameter cylinder model, the temperature field of adopting method of difference to calculate this simplified model; Adopt the nominal centrifugal stress and the nominal thermal stress of any radius of analytic formula calculating rotor of prior art, at the compound radial stress σ of any radius name _r, nominal compound tangential stress σ _θWith nominal compound axial stress σ _zFormula be expressed as respectively:;

σ _r＝σ _rth+σ _rt；

σ _θ＝σ _θth+σ _θt；

σ _z＝σ _zth+σ _zt；

In the formula:

σ _Rth---name is thermal stress radially

σ _Rt---name is centrifugal stress radially

σ _{θ th}---the tangential thermal stress of name

σ _{θ t}---the tangential centrifugal stress of name

σ _Zth---name is thermal stress axially

σ _Zt---name is centrifugal stress axially;

Second step: calculate nominal equivalent stress

The formula of any radius name of turbine rotor equivalent stress is expressed as:

${\mathrm{\σ}}_{\mathrm{ne}}=\frac{1}{\sqrt{2}}{[{({\mathrm{\σ}}_z-{\mathrm{\σ}}_r)}^2+{({\mathrm{\σ}}_r-{\mathrm{\σ}}_{\mathrm{\θ}})}^2+{({\mathrm{\σ}}_{\mathrm{\θ}}-{\mathrm{\σ}}_z)}^2]}^{\frac{1}{2}};$

The 3rd step: determine transient state equivalent stress correction factor y _i

Same roots rotor for the same model steam turbine, set up bidimensional or three-dimensional finite element model according to the practical structures size, the change condition of vapor (steam) temperature, pressure, rotating speed and the load etc. of the given steam turbine identical with the first step, adopt the FEM (finite element) model calculated off-line of prior art, draw the equivalent stress σ that the turbine rotor same area comprises thermal stress and centrifugal stress _Eqa, for the transient state moment t of steam turbine start-up course, stopping process or load change process _i, the rotor outer surface equivalent stress σ that FEM (finite element) model calculates _EqaThe rotor surface name equivalent stress σ that calculates divided by simplified model _Ne, calculate the equivalent stress correction factor y of rotor outer surface _i:

y _i＝σ _eqa/σ _ne；

The 4th step: set up the artificial neural network that calculates equivalent stress correction factor y

Set up three-layer artificial neural network's model, be made up of input layer, hidden layer and output layer, each node of input layer is connected with each node of hidden layer, and each node of hidden layer is connected with output layer;

Described input layer is made up of 10 nodes, and 10 physical quantitys of 10 input layer correspondences are: steam turbine main steam pressure P ₀, steam turbine main steam temperature t ₀, steam turbine working speed n, steam turbine load N, preceding 1 minute to 5 minutes steam turbine main steam temperature variance ratio δ t ₀The variance ratio Δ t of the load changing rate δ N (MW/ branch) of (℃/minute), preceding 1 minute to 5 minutes steam turbine, preceding 6 minutes to 14 minutes steam turbine main steam temperature ₀Load changing rate Δ N (MW/ branch), the turbine rotor volume averaging temperature θ of (℃/minute), preceding 6 minutes to 14 minutes steam turbine _mInitial moment inner cylinder metal monitor temperature θ with steam turbine transient (being start-up course, stopping process and load change process) ₀

Described hidden layer is made up of 20 to 30 nodes;

Described output layer has only a node, the correction factor y of expression steam turbine rotator equivalent stress;

The 5th step: connection weights and the threshold value of determining artificial neural network

The correction factor y of each turbine rotor transient state equivalent stress _i, all to one group of steam turbine main steam pressure P should be arranged _0i, steam turbine main steam temperature t _0i, steam turbine working speed n _i, steam turbine load N _i, preceding 1 minute to 5 minutes steam turbine main steam temperature variance ratio δ t _0iThe load changing rate δ N of (℃/minute), preceding 1 minute to 5 minutes steam turbine _iThe variance ratio Δ t of (MW/ branch), preceding 6 minutes to 14 minutes steam turbine main steam temperature _0iThe load changing rate Δ N of (℃/minute), preceding 6 minutes to 14 minutes steam turbine _i(MW/ branch), turbine rotor volume averaging temperature θ _MiInitial moment inner cylinder metal monitor temperature θ with the steam turbine transient _0i, the correction factor y of Q the steam turbine rotator equivalent stress that calculates for a large amount of start-up courses, stopping process and load change process _iWith Q group input value, constitute Q group training sample, to Q group input sample data P _0i, t _0i, n _i, N _i, δ t _0i, δ N _i, Δ t _0i, Δ N _i, θ _Mi, θ _0iWith Q output value y _iAdopt batch algorithms, for artificial nerve network model shown in Figure 1, relative error quadratic sum with network export target value and actual value output is trained as performance function, train the performance function of network to reach given 0.01%, realize the Nonlinear Mapping between input output, just can finish training.Adopt prior art,, draw and be connected weights, hidden node and the internodal threshold value of weights, hidden node and the threshold value of output layer node of being connected of output layer between this artificial neural network input layer and hidden node through the learning training of artificial neural network;

The 6th step: write the artificial neural network software that calculates the equivalent stress correction factor

By the threshold value that is connected weights and node between the artificial nerve network model that calculates steam turbine rotator equivalent stress correction factor y and its node, adopt the C language compilation to calculate the software for calculation of steam turbine rotator equivalent stress correction factor y, as the subroutine of steam turbine roter low-cycle fatigue life consumption in line computation;

Second flow process: steam turbine roter low-cycle fatigue life consumption is in line computation, supervision and control

The 7th step: adopt the nominal equivalent stress σ of the online real-time calculating of simplified model _Ne

Adopt the simplified model that waits heavy wall cylinder or equal diameter cylinder of prior art, the volume averaging temperature θ of online real-time calculating turbine rotor _mNominal equivalent stress σ with the typical radius place _Ne

The 8th step: use the online real-time calculating equivalent stress correction factor y of artificial neural network technology

The steam turbine main steam pressure P of Input Online monitoring ₀, steam turbine main steam temperature t ₀, steam turbine working speed n, steam turbine load N, preceding 1 minute to 5 minutes steam turbine main steam temperature variance ratio δ t ₀The variance ratio Δ t of the load changing rate δ N (MW/ branch) of (℃/minute), preceding 1 minute to 5 minutes steam turbine, preceding 6 minutes to 14 minutes steam turbine main steam temperature ₀Load changing rate Δ N (MW/ branch), the inner casing metal monitor temperature θ of the initial moment of steam turbine transient of (℃/minute), preceding 6 minutes to 14 minutes steam turbine ₀And the online turbine rotor volume averaging temperature θ that calculates _m, use the artificial neural network subroutine of calculating equivalent stress correction factor y, calculate steam turbine rotator equivalent stress correction factor y in real time;

The 9th step: online real-time calculating equivalent stress monitoring value σ _Eq

The monitoring value σ of online real-time calculating turbine rotor typical radius place equivalent stress _Eq:

σ _eq＝y×σ _ne；

The tenth step: the low cycle fatigue strain amplitude ε that calculates symmetrical cycle _a

The low cycle fatigue strain amplitude ε of turbine rotor life-span weak part symmetrical cycle _aFormula be expressed as:

${\mathrm{\ϵ}}_a=\frac{(1+\mathrm{\μ}){\mathrm{\σ}}_{\mathrm{eq}}}{1.5E}$

In the formula:

The Poisson's ratio of μ---rotor material

The Young's modulus of E---rotor material;

The 11 step: the low-cycle fatigue life N that determines symmetrical cycle

According to rotor material low cycle fatigue curve ε _a=f (N) can determine the low-cycle fatigue life N of turbine rotor symmetrical cycle;

The 12 step: calculate low-cycle fatigue life loss d

Steam turbine start once shut down once or load up once or the formula of the low-cycle fatigue life loss d that once produces of load down be expressed as:

$d=\frac{1}{2N}\×100\%;$

The 13 step: the threshold value [d] of determining low-cycle fatigue life loss

Threshold value [d] according to the experience definition steam turbine roter low-cycle fatigue life consumption of being engaged in life Prediction of Steam Turbine Rotor technical research work accumulation for many years is illustrated in table 1;

[table 1]

The method of operation	Life consumption threshold value [d] (%)	Access times	The accumulation life consumption
				Cold start	0.0150	200	3.00
Warm starting	0.0100	1200	12.00
				Hot starting, hot start	0.0045	8100	36.45
Very hot startup	0.0045	300	1.35
				The load change load up	0.0003	12000	3.60
The load change load down	0.0003	12000	3.60
				Shutdown at sliding parameters	0.0150	200	3.00
Orderly closedown	0.0010	9500	9.50
				Emergency shutdown	0.0015	300	0.45
Add up to			72.95

The 14 step: the in-service monitoring and the control of the damage of turbine rotor low-cycle fatigue life

Online real-time calculating steam turbine roter low-cycle fatigue life consumption d, and the threshold value [d] of the low-cycle fatigue life loss of the turbine rotor that provides with table 1 is relatively, the monitoring of steam turbine roter low-cycle fatigue life consumption and control are divided into following three kinds of situations:

1. if d≤0.8[d], the variance ratio of steam turbine main steam temperature and load (power) variance ratio is by the stated number Value Operations of " steam turbine operation rules ";

2. if 0.8[d]＜d≤[d], reduce the main steam temperature variance ratio and the load changing rate of steam turbine, to reduce the low-cycle fatigue life loss of steam turbine high pressure rotor;

3. [if d]≤d＜1.05[d], the main steam temperature variance ratio and the load change rate of control steam turbine are 0, with the low-cycle fatigue life loss of further reduction turbine rotor;

The 15 step: the online warning of steam turbine roter low-cycle fatigue life consumption and beat gate stop-start

If the low-cycle fatigue life loss d of online real-time calculating turbine rotor is 1.05[d]≤d＜1.25[d], send warning, tripping grinder after 30 minutes; If d 〉=1.25[d], give the alarm tripping grinder after 1 minute.

The present invention has following characteristics

1. the computer software with the steam turbine roter low-cycle fatigue life consumption of C language compilation is installed on the calculated/applied server of steam turbine roter low-cycle fatigue life consumption in line computation, according to the time lag of software set, the turbine parameters of the on-line monitoring that from database server, reads, the equivalent stress of online real-time calculating turbine rotor and low-cycle fatigue life loss, the result that computational analysis draws, deliver to database server again and preserve, call for web page server.

2. database server is deposited two class data:

Primary sources are the data of turbine parameters on-line monitoring, comprise main steam pressure P ₀, main steam temperature t ₀, inner cylinder metal monitor temperature θ ₀, turbine speed n, steam turbine load N, preceding 1 minute to 5 minutes steam turbine main steam temperature variance ratio δ t ₀The variance ratio Δ t of the load changing rate δ N (MW/ branch) of (℃/minute), preceding 1 minute to 5 minutes steam turbine, preceding 6 minutes to 14 minutes steam turbine main steam temperature ₀The load changing rate Δ N (MW/ branch) of (℃/minute), preceding 6 minutes to 14 minutes steam turbine;

Secondary sources are the result of calculation of steam turbine roter low-cycle fatigue life consumption, comprise the equivalent stress σ of turbine rotor _Eq, low cycle fatigue strain amplitude ε _a, symmetrical cycle low-cycle fatigue life N, low-cycle fatigue life loss d, low-cycle fatigue life loss report to the police and data, the steam turbine operation control measure of making lock;

3. external system interface has two kinds of functions:

The one, the monitor value of turbine parameters is deposited in database;

The 2nd, the steam turbine operation control measure are transferred to steam turbine control system (DEH);

4. steam turbine automatic control system (DEH) has two kinds of functions with the parameter measuring point:

One provides the parameter of steam turbine on-line monitoring, comprises main steam pressure P ₀, main steam temperature t ₀, inner cylinder metal monitor temperature θ ₀, turbine speed n, steam turbine load N, preceding 1 minute to 5 minutes steam turbine main steam temperature variance ratio δ t ₀The variance ratio Δ t of the load changing rate δ N (MW/ branch) of (℃/minute), preceding 1 minute to 5 minutes steam turbine, preceding 6 minutes to 14 minutes steam turbine main steam temperature ₀The load changing rate Δ N (MW/ branch) of (℃/minute), preceding 6 minutes to 14 minutes steam turbine;

The 2nd, instruct the operation of steam turbine according to the result of calculation of steam turbine roter low-cycle fatigue life consumption, make the low-cycle fatigue life loss of turbine rotor be in slave mode;

5. the online result of calculation of steam turbine roter low-cycle fatigue life consumption is issued on web page server, according to the browser end user is that the technician of power plant sends request, real-time result of calculation by steam turbine roter low-cycle fatigue life consumption in the calculated/applied server calls database server, on web page server, form the result of calculation page of steam turbine roter low-cycle fatigue life consumption, return to the browser end user, instruct steam turbine operation;

6. the user side browser is used for checking the predicting the outcome of low-cycle fatigue life loss of the rotor of different steam turbine.

The in-service monitoring of the steam turbine roter low-cycle fatigue life consumption that the present invention provides and the method and system of control can realize online real-time calculating and the LINE REAL TIME MONITORING and the control of steam turbine roter low-cycle fatigue life consumption.If steam turbine roter low-cycle fatigue life consumption is greater than threshold value, reduce steam turbine roter low-cycle fatigue life consumption by the rate of temperature change of online real-time control steam turbine main steam and the load changing rate of steam turbine, make the low-cycle fatigue life loss of turbine rotor be in slave mode, to prolong the low-cycle fatigue life of turbine rotor.When if steam turbine roter low-cycle fatigue life consumption reaches alarming value; send warning; and in the time of setting, shut down, to reduce the low-cycle fatigue life loss of turbine rotor, reached the technique effect that monitors and control steam turbine roter low-cycle fatigue life consumption.

Advantage of the present invention is to realize quick online real-time calculating and the in-service monitoring and the control of steam turbine roter low-cycle fatigue life consumption, realized that turbine rotor starts (IEC IEC is defined as and starts 3830 times) 10000 times in the phase under arms, reached to prolong the turbine rotor technique effect in working life.

Description of drawings

Fig. 1 adopts the skeleton diagram of low-cycle fatigue life loss online monitoring system for the present invention;

Fig. 2 is an artificial nerve network model of the present invention;

Fig. 3 is the flow chart of method that the present invention adopts;

Fig. 4 is the computer software block diagram of method that the present invention adopts;

Fig. 5 is the Changing Pattern of outer surface equivalent stress behind the high pressure rotor governing stage impeller of cold start process;

Fig. 6 is the Changing Pattern of outer surface symmetrical cycle low cycle fatigue strain amplitude ε a behind the high pressure rotor governing stage impeller of cold start process;

Fig. 7 is the Changing Pattern of outer surface low-cycle fatigue life loss behind the high pressure rotor governing stage impeller of cold start process.

Embodiment

As shown in Figure 1, be steam turbine roter low-cycle fatigue life consumption in-service monitoring and control system, calculated/applied server 1 and software by low-cycle fatigue life loss, database server 2, external system interface 3, steam turbine control system (DEH) and parameter measuring point 4, web page server 5 and user side browser 6 are formed, web page server 5 is connected with each user side browser 6 with calculated/applied server 1 respectively, calculated/applied server 1 is connected with database server 2, and database server 2 is connected with parameter measuring point 4 with steam turbine control system DEH by external system interface 3.

As shown in Figure 2, be artificial nerve network model of the present invention, as shown in Figure 3, flow chart for method that the present invention adopts, as shown in Figure 4, for the steam turbine roter low-cycle fatigue life consumption that adopts the C language compilation in line computation, monitor and the calculated/applied computer software block diagram of controlling, this software is installed in steam turbine roter low-cycle fatigue life consumption on the calculated/applied server of line computation, online real-time calculating, monitor and the low-cycle fatigue life loss of controlling turbine rotor, be applied to the online management of steam turbine roter low-cycle fatigue life consumption.

The invention will be further described below in conjunction with drawings and Examples.

Embodiment

For certain model throttle (steam) temperature is the high pressure rotor of 566 ℃ overcritical 600MW steam turbine, adopt the flow chart of low-cycle fatigue life loss online monitoring system shown in Figure 1, artificial nerve network model that the present invention shown in Figure 2 adopts, the method that the invention provides shown in Figure 3 and the software for calculation block diagram of the inventive method shown in Figure 4, this turbine high-pressure rotor equivalent stress σ in the cold start process that calculates _Eq, low cycle fatigue strain amplitude ε _aList in Fig. 5, Fig. 6 and Fig. 7 respectively with the result of calculation of low-cycle fatigue life loss d.

The first step, second step and the 7th step: the online nominal equivalent stress σ that calculates this fillet position, turbine high-pressure rotor governing stage impeller outer surface in the cold start process of simplified model that adopts prior art _NeIn Fig. 5, represent with curve 1.

The 3rd step: adopt the FEM (finite element) model of prior art, calculated off-line draws the equivalent stress σ of this fillet position, turbine high-pressure rotor governing stage impeller outer surface in the cold start process _EqaIn Fig. 5, represent with curve 2, in Fig. 5, the equivalent stress value σ of the corresponding curve 2 of same abscissa _EqaNominal equivalent stress value σ with curve 1 _NeCompare, draw a series of equivalent stress correction factor y _i

The 4th step: set up three-layer artificial neural network's model as shown in Figure 2, comprise input layer, hidden layer and output layer.Input layer is made up of 10 nodes, and 10 physical quantitys of 10 input layer correspondences are: steam turbine main steam pressure P ₀, steam turbine main steam temperature t ₀, steam turbine working speed n, steam turbine load N, preceding 3 minutes steam turbine main steam temperature variance ratio δ t ₀The variance ratio Δ t of the load changing rate δ N (MW/ branch) of (℃/minute), preceding 3 minutes steam turbine, preceding 10 minutes steam turbine main steam temperature ₀Load changing rate Δ N (MW/ branch), the turbine rotor volume averaging temperature θ of (℃/minute), preceding 10 minutes steam turbine _mInitial moment inner cylinder metal monitor temperature θ with steam turbine transient (start-up course, stopping process and load change process) ₀Hidden layer is made up of 20 to 30 nodes; Output layer has only a node, the correction factor y of expression steam turbine rotator equivalent stress.

The 5th step and the 6th step: to artificial nerve network model shown in Figure 2, adopt batch algorithms, study and training through artificial neural network, determine the connection weights and the threshold value of artificial neural network, and write the artificial neural network software that calculates equivalent stress correction factor y, as the subroutine of steam turbine roter low-cycle fatigue life consumption software for calculation.

The 8th step and the 9th step: use artificial neural network technology at line computation equivalent stress correction factor y, by formula σ _Eq=y * σ _NeThe equivalent stress correction value σ at online this fillet position, steam turbine high pressure rotor governing stage impeller outer surface that calculates _EqCurve 3 expressions in Fig. 5.

The tenth step: this steam turbine high pressure rotor adopts the CrMoV material, and the pass of its Poisson's ratio μ and operating temperature t is: μ=0.2876+3.8316 * 10 ^-5T-1.4636 * 10 ^-7t ²+ 2.7588 * 10 ^-10t ³, the pass of its elastic modulus E and operating temperature t is: E=2.1663 * 10 ⁵-0.8487t+0.1554t ²-2.7826 * 10 ^-4t ³The symmetrical cycle low cycle fatigue strain amplitude of cold start process ${\mathrm{\ϵ}}_a=\frac{(1+\mathrm{\μ}){\mathrm{\σ}}_{\mathrm{eq}}}{1.5E}$ Result of calculation in Fig. 6, use curve representation.

The 11 step and the 12 step: this turbine rotor adopts the CrMoV material, and the test curve of the low cycle fatigue of its ε a=f (N) is: ε _a=k ₁(2N) ^-k2+ k ₃(2N) ^-k4, k here ₁, k ₂, k ₃And k ₄Be testing of materials constant.

After adopting above formula to calculate low-cycle fatigue life N, the low-cycle fatigue life loss d that calculates this turbine rotor by d=1/ (2N) is illustrated in Fig. 7 again.

The 13 step: the threshold value of this steam turbine roter low-cycle fatigue life consumption [d] is illustrated in table 2, at cold start process [d]=0.015%.

The 14 step and the 15 step: in Fig. 7 of the Changing Pattern of this turbine high-pressure roter low-cycle fatigue life consumption d, A district d≤0.8[d], the main steam temperature variance ratio of steam turbine and load changing rate are pressed the stated number Value Operations of " steam turbine operation rules "; B district 0.8[d]＜d＜[d], reduce the main steam temperature variance ratio and the load changing rate of steam turbine, to reduce steam turbine roter low-cycle fatigue life consumption d; C district [d]≤d＜1.05[d] the main steam temperature variance ratio and the load change rate of control steam turbine be 0, to reduce steam turbine roter low-cycle fatigue life consumption d; D district 1.05[d]≤d＜1.25[d], send warning, tripping grinder after 30 minutes; E district d 〉=1.25[d], give the alarm tripping grinder after 1 minute.

Adopt the on-line monitoring method and system of turbine rotor low cycle fatigue loss provided by the invention, the employing artificial neural network technology calculates the equivalent stress of this model steam turbine high pressure rotor, equivalent stress calculation accuracy height; Adopt material low cycle fatigue test curve; the low-cycle fatigue life N and the low-cycle fatigue life loss d of this model steam turbine high pressure rotor of online real-time calculating; by controlling the start-up course of this model steam turbine; the main steam temperature variance ratio and the load changing rate of stopping process and load change process; can be implemented in line real time monitoring and the low-cycle fatigue life loss of controlling this model steam turbine high pressure rotor; realized this model steam turbine high pressure roter low-cycle fatigue life consumption d is controlled in the allowed band of low-cycle fatigue life loss threshold value [d]; make the low-cycle fatigue life of this model steam turbine high pressure rotor be in slave mode; reached in the military service phase of steam turbine and to have started 10000 times, realized prolonging this model steam turbine high pressure rotor technique effect in working life.

Claims (3)

1. on-line monitoring steam turbine roter low-cycle fatigue life consumption system, it is characterized in that, calculated/applied server (1) and software by low-cycle fatigue life loss, database server (2), external system interface (3), steam turbine control system (DEH) and parameter measuring point (4), web page server (5) and user side browser (6) are formed, web page server (5) is connected with each user side browser (6) with calculated/applied server (1) respectively, calculated/applied server (1) is connected with database server (2), and database server (2) is connected with parameter measuring point (4) with steam turbine control system DEH by external system interface (3).

2. a kind of on-line monitoring steam turbine roter low-cycle fatigue life consumption according to claim 1 system, it is characterized in that, described steam turbine control system DEH and parameter measuring point, control system DEH is the supporting automatic control system of steam turbine, the parameter measuring point is 9 parameters of steam turbine control system DEH on-line monitoring, comprise main steam pressure P0, main steam temperature t0, inner cylinder metal monitor temperature θ 0, turbine speed n, steam turbine load N, preceding 1 minute to 5 minutes steam turbine main steam temperature variance ratio δ t0, the load changing rate δ N of preceding 1 minute to 5 minutes steam turbine, the variance ratio Δ t0 of preceding 6 minutes to 14 minutes steam turbine main steam temperature, the load changing rate Δ N of preceding 6 minutes to 14 minutes steam turbine.

3. a kind of on-line monitoring steam turbine roter low-cycle fatigue life consumption method according to claim 1, it is characterized in that, adopt the computer software of the steam turbine roter low-cycle fatigue life consumption of C language compilation to operate on the calculated/applied server (1), be applied to the on-line monitoring of steam turbine roter low-cycle fatigue life consumption, the artificial neural network technology of 10 input layers of employing calculates the correction factor of the equivalent stress of turbine rotor, adopt the low cycle fatigue curve of rotor material and low-cycle fatigue life and the low-cycle fatigue life loss that the symmetrical cycle method is calculated turbine rotor, its method is made up of two flow processs:

First pass: adopt artificial neural network technology to determine the correction factor of steam turbine rotator equivalent stress

The first step: adopt the simplified model of prior art to calculate rotor name compound stress

Adopt the temperature field of the simplified model calculated off-line turbine rotor of prior art, the turbine rotor that center hole is arranged such as is reduced at the heavy wall cylindrical model, the turbine rotor of no center hole is reduced to the equal diameter cylinder model, the temperature field of adopting method of difference to calculate this simplified model; Adopt the nominal centrifugal stress and the nominal thermal stress of any radius of analytic formula calculating rotor of prior art, at the compound radial stress σ of any radius name _r, nominal compound tangential stress σ _θWith nominal compound axial stress σ _zFormula be expressed as respectively:

σ _r＝σ _rth+σ _rt；

σ _θ＝σ _θth+σ _θt；

σ _z＝σ _zth+σ _zt；

In the formula:

σ _Rth---name is thermal stress radially

σ _Rt---name is centrifugal stress radially

σ _{θ th}---the tangential thermal stress of name

σ _{θ t}---the tangential centrifugal stress of name

σ _Zth---name is thermal stress axially

σ _Zt---name is centrifugal stress axially;

Second step: calculate nominal equivalent stress

The formula of any radius name of turbine rotor equivalent stress is expressed as:

${\mathrm{\σ}}_{\mathrm{ne}}=\frac{1}{\sqrt{2}}{[{({\mathrm{\σ}}_z-{\mathrm{\σ}}_r)}^2+{({\mathrm{\σ}}_r-{\mathrm{\σ}}_{\mathrm{\θ}})}^2+{({\mathrm{\σ}}_{\mathrm{\θ}}-{\mathrm{\σ}}_z)}^2]}^{\frac{1}{2}};$

The 3rd step: determine transient state equivalent stress correction factor y _i

Same roots rotor for the same model steam turbine, set up bidimensional or three-dimensional finite element model according to the practical structures size, the change condition of vapor (steam) temperature, pressure, rotating speed and the load etc. of the given steam turbine identical with the first step, adopt the FEM (finite element) model calculated off-line of prior art, draw the equivalent stress σ that the turbine rotor same area comprises thermal stress and centrifugal stress _Eqa, for the transient state moment t of steam turbine start-up course, stopping process or load change process _i, the rotor outer surface equivalent stress σ that FEM (finite element) model calculates _EqaThe rotor surface name equivalent stress σ that calculates divided by simplified model _Ne, calculate the equivalent stress correction factor y of rotor outer surface _i:

y _i＝σ _eqa/σ _ne；

The 4th step: set up the artificial neural network that calculates equivalent stress correction factor y

Set up three-layer artificial neural network's model, be made up of input layer, hidden layer and output layer, each node of input layer is connected with each node of hidden layer, and each node of hidden layer is connected with output layer;

Described input layer is made up of 10 nodes, and 10 physical quantitys of 10 input layer correspondences are: steam turbine main steam pressure P ₀, steam turbine main steam temperature t ₀, steam turbine working speed n, steam turbine load N, preceding 1 minute to 5 minutes steam turbine main steam temperature variance ratio δ t ₀The variance ratio Δ t of the load changing rate δ N (MW/ branch) of (℃/minute), preceding 1 minute to 5 minutes steam turbine, preceding 6 minutes to 14 minutes steam turbine main steam temperature ₀Load changing rate Δ N (MW/ branch), the turbine rotor volume averaging temperature θ of (℃/minute), preceding 6 minutes to 14 minutes steam turbine _mInitial moment inner cylinder metal monitor temperature θ with steam turbine transient (being start-up course, stopping process and load change process) ₀

Described hidden layer is made up of 20 to 30 nodes;

Described output layer has only a node, the correction factor y of expression steam turbine rotator equivalent stress;

The 5th step: connection weights and the threshold value of determining artificial neural network

The correction factor y of each turbine rotor transient state equivalent stress _i, all to one group of steam turbine main steam pressure P should be arranged _0i, steam turbine main steam temperature t _0i, steam turbine working speed n _i, steam turbine load N _i, preceding 1 minute to 5 minutes steam turbine main steam temperature variance ratio δ t _0i, preceding 1 minute to 5 minutes steam turbine load changing rate δ N _i, preceding 6 minutes to 14 minutes steam turbine main steam temperature variance ratio Δ t _0i, preceding 6 minutes to 14 minutes steam turbine load changing rate Δ N _i, turbine rotor volume averaging temperature θ _MiInitial moment inner cylinder metal monitor temperature θ with the steam turbine transient _0i, the correction factor y of Q the steam turbine rotator equivalent stress that calculates for a large amount of start-up courses, stopping process and load change process _iWith Q group input value, constitute Q group training sample, to Q group input sample data P _0i, t _0i, n _i, N _i, δ t _0i, δ N _i, Δ t _0i, Δ N _i, θ _Mi, θ _0iWith Q output value y _iAdopt batch algorithms, relative error quadratic sum with network export target value and actual value output is trained as performance function, train the performance function of network to reach given 0.01%, realize the Nonlinear Mapping between input output, just can finish training, draw and be connected weights, hidden node and the internodal threshold value of weights, hidden node and the threshold value of output layer node of being connected of output layer between this artificial neural network input layer and hidden node.

The 6th step: write the artificial neural network software that calculates the equivalent stress correction factor

By the threshold value that is connected weights and node between the artificial nerve network model that calculates steam turbine rotator equivalent stress correction factor y and its node, adopt the C language compilation to calculate the software for calculation of steam turbine rotator equivalent stress correction factor y, as the subroutine of steam turbine roter low-cycle fatigue life consumption in line computation;

Second flow process: steam turbine roter low-cycle fatigue life consumption is in line computation, supervision and control

The 7th step: adopt the nominal equivalent stress σ of the online real-time calculating of simplified model _Ne

Adopt the simplified model that waits heavy wall cylinder or equal diameter cylinder of prior art, the volume averaging temperature θ of online real-time calculating turbine rotor _mNominal equivalent stress σ with the typical radius place _Ne

The 8th step: use the online real-time calculating equivalent stress correction factor y of artificial neural network technology

The steam turbine main steam pressure P of Input Online monitoring ₀, steam turbine main steam temperature t ₀, steam turbine working speed n, steam turbine load N, preceding 1 minute to 5 minutes steam turbine main steam temperature variance ratio δ t ₀, the load changing rate δ N of preceding 1 minute to 5 minutes steam turbine, the variance ratio Δ t of preceding 6 minutes to 14 minutes steam turbine main steam temperature ₀, preceding 6 minutes to 14 minutes steam turbine load changing rate Δ N, inner casing metal monitor temperature θ of the initial moment of steam turbine transient ₀And the online turbine rotor volume averaging temperature θ that calculates _m, use the artificial neural network subroutine of calculating equivalent stress correction factor y, calculate steam turbine rotator equivalent stress correction factor y in real time;

The 9th step: online real-time calculating equivalent stress monitoring value σ _Eq

The monitoring value σ of online real-time calculating turbine rotor typical radius place equivalent stress _Eq:

σ _eq＝y×σ _ne；

The tenth step: the low cycle fatigue strain amplitude ε that calculates symmetrical cycle _a

The low cycle fatigue strain amplitude ε of turbine rotor life-span weak part symmetrical cycle _aFormula be expressed as:

${\mathrm{\ϵ}}_a=\frac{(1+\mathrm{\μ}){\mathrm{\σ}}_{\mathrm{eq}}}{1.5E}$

In the formula:

The Poisson's ratio of μ---rotor material

The Young's modulus of E---rotor material;

The 11 step: the low-cycle fatigue life N that determines symmetrical cycle

According to rotor material low cycle fatigue curve ε _a=f (N) can determine the low-cycle fatigue life N of turbine rotor symmetrical cycle;

The 12 step: calculate low-cycle fatigue life loss d

Steam turbine start once shut down once or load up once or the formula of the low-cycle fatigue life loss d that once produces of load down be expressed as:

$d=\frac{1}{2N}\×100\%;$

The 13 step: the threshold value [d] of determining low-cycle fatigue life loss

Threshold value [d] according to the experience definition steam turbine roter low-cycle fatigue life consumption of being engaged in life Prediction of Steam Turbine Rotor technical research work accumulation for many years is illustrated in table 1;

[table 1] The method of operation Life consumption threshold value [d] (%) Access times The accumulation life consumption Cold start 0.0150 200 3.00 Warm starting 0.0100 1200 12.00 Hot starting, hot start 0.0045 8100 36.45 Very hot startup 0.0045 300 1.35 The load change load up 0.0003 12000 3.60 The load change load down 0.0003 12000 3.60 Shutdown at sliding parameters 0.0150 200 3.00 Orderly closedown 0.0010 9500 9.50 Emergency shutdown 0.0015 300 0.45 Add up to 72.95

The 14 step: the in-service monitoring and the control of the damage of turbine rotor low-cycle fatigue life

Online real-time calculating steam turbine roter low-cycle fatigue life consumption d, and the threshold value [d] of the low-cycle fatigue life loss of the turbine rotor that provides with table 1 is relatively, the monitoring of steam turbine roter low-cycle fatigue life consumption and control are divided into following three kinds of situations:

1. if d≤0.8[d], the variance ratio of steam turbine main steam temperature and load changing rate are by the stated number Value Operations of " steam turbine operation rules ";

2. if 0.8[d]＜d≤[d], reduce the main steam temperature variance ratio and the load changing rate of steam turbine, to reduce the low-cycle fatigue life loss of steam turbine high pressure rotor;

3. [if d]≤d＜1.05[d], the main steam temperature variance ratio and the load change rate of control steam turbine are 0, with the low-cycle fatigue life loss of further reduction turbine rotor;

The 15 step: the online warning of steam turbine roter low-cycle fatigue life consumption and beat gate stop-start

If the low-cycle fatigue life loss d of online real-time calculating turbine rotor is 1.05[d]≤d＜1.25[d], send warning, tripping grinder after 30 minutes; If d 〉=1.25[d], give the alarm tripping grinder after 1 minute.

Images