CN105134386A - On-line monitoring method for gas turbine combustion system based on measuring-point weighted value - Google Patents

Abstract

The invention discloses an on-line monitoring method for a gas turbine combustion system based on a measuring-point weighted value, belonging to the field of monitoring of the gas turbine combustion system. A current combustion monitoring system cannot judge the changing trend of a combustion state. According to the on-line monitoring method for the gas turbine combustion system based on the measuring-point weighted value, n temperature measuring points are uniformly and circumferentially arranged on the turbine outlet of a gas turbine, so that exhaust gas temperature data Ti during normal operation at a tm time period is obtained; the correlation factor alpha i,1 of Ti and T1 is added, and prediction values of Ti are separately obtained according to the relation function between Ti and T1; the exhaust gas temperature theoretic value T1<'> of a fault-free temperature measuring point 1 is calculated; the difference between the theoretical value T1<'> of the temperature measuring point 1 and the measured value T1 of the temperature measuring point 1 is defined as delta T1, and delta T1 satisfies normal distribution with the mean value 0 and standard deviation delta1; and monitoring on the temperature measuring point is performed through the relationship between delta T1 and the range from minus 3 delta1 to 3 delta1. By adopting the method disclosed by the invention, on-line monitoring of the exhaust gas temperature of the gas turbine is realized, and an abnormal evolution process can be accurately detected by fully utilizing the correlation among the measuring points of the exhaust gas temperature.

Description

Based on the gas turbine combustion system on-line monitoring method of measuring point weighted value

Technical field

The present invention relates to a kind of gas turbine combustion system on-line monitoring method based on measuring point weighted value.

Background technique

Gas turbine is as novel power equipment, there is compact structure, operate steadily, safe and reliable, can start fast and be with dynamic load, there is the advantages such as the higher thermal efficiency, be widely used in aviation, ground and naval vessel etc., therefore the abnormality detection of gas turbine is to the actual important in inhibiting of production.In gas turbine unit running, for gas turbine combustion system abnormality detection aspect, the fault of more than 50% is all relevant with firing chamber.Because the parts long-term works such as combustion chambers burn cylinder are at the high-temperature area of 1600 DEG C, work under bad environment, equipment is once occur that defect may constitute a threat to the nozzle in downstream and movable vane piece safety.Therefore, be necessary to monitor the working condition of firing chamber.

Combustion system, once break down, can make combustor exit temperature occur abnormal.Therefore we can monitor the operation conditions of combustion system by detecting combustor exit temperature.But conventional temperature-measuring element cannot in the region long-term work of high temperature like this, and therefore, in unit turbine exhaust passage, circumference has been evenly arranged several exhaust temperature thermocouples, and the temperature that thermocouple is surveyed is exactly row's temperature of gas turbine.Firing chamber and thermocouple deployment scenarios schematic diagram as shown in Figure 1.Judge whether the working condition of combustion barrel occurs exception by the situation of arranging temperature.

The MARKVI combustion monitoring system definition S that GE company develops is that warm dispersion degree is arranged in the permission of delivery temperature, thinks that S is the average exhaust T of gas turbine outlet ₄ ^*, compressor delivery temperature T ₄ ^*function, concrete function is empirical correlation:

$S=(60+0.145{T_4}^{*}-0.08{T_2}^{*}|\begin{array}{c}750\\ 50\end{array})|\begin{array}{c}150\\ 50\end{array}+\left(100\right)$

In this formula, temperature all Shi take ℉ as unit of measurement.Right of formula 100 with bracket, just add this under representing variable working condition condition.

In addition, MARKVI combustion monitoring system also defines: S1 is the most difference between high scale and minimum reading of delivery temperature thermocouple; S2 is the most difference between high scale and the 2nd low scale of delivery temperature thermocouple; S3 is the most difference between high scale and the 3rd low scale of delivery temperature thermocouple.

Based on above-mentioned formula and definition, the discrimination principles of MARK VI combustion monitoring protective system is shown in Fig. 2.In Fig. 2, K ₁, K ₂, K ₃three parameters empirically defined.In typical case:

K ₁＝1.0；K ₂＝5.0；K ₃＝0.8

The discrimination principles of combustion monitoring as shown in Figure 2; Find in actual applications, this kind of method detects exists serious " afterwards " diagnosis phenomenon, and namely when detection system sends warning, combustion system has been damaged more serious.

Summary of the invention

The object of the invention is cannot by the detection of abnormal evolution process in order to solve existing combustion monitoring system, to the problem that combustion regime variation tendency judges, occur can detect in early days in fault, and propose a kind of gas turbine combustion system on-line monitoring method based on measuring point weighted value.

Based on a gas turbine combustion system on-line monitoring method for measuring point weighted value, it is characterized in that: the described gas turbine combustion system on-line monitoring method based on measuring point weighted value is realized by following steps:

Step one, gas turbine turbine outlet circumference arrange n temperature point equably, gas turbine is at t _min period, normal operation, obtains t _mrow's temperature data T=[T in period ₁, T ₂..., T _i... T _n]; Wherein, T _irepresent that i-th temperature point is at t _mrow's temperature data that in period, each time point records, $T_1=\&lsqb;T_{1t_1},T_{1t_2},...,T_{1t_m}\&rsqb;,T_2=\&lsqb;T_{2t_1},T_{2t_2},...,T_{2t_m}\&rsqb;,$ $T_i=\&lsqb;T_{{\mathrm{it}}_1},T_{{\mathrm{it}}_2},...,T_{{\mathrm{it}}_m}\&rsqb;,T_n=\&lsqb;T_{{\mathrm{nt}}_1},T_{{\mathrm{nt}}_2},...,T_{{\mathrm{nt}}_m}\&rsqb;;$

Step 2, utilize the method for least squares warm data T of the row of obtaining respectively _iwith the warm data T of row ₁relation function, that is:

T ₁＝f _i,1(T _i)，i＝2,3,4,…,n

Wherein, f _{i, 1}represent the warm data T of row _iwith the warm data T of row ₁relation function; Meet: T ₁=k _{i, 1}t _i+ b _{i, 1}, i=2,3,4 ..., n, wherein k _{i, 1}and b _{i, 1}can be drawn by method of least squares, that is:

$k_{i,1}=\frac{\underset{a=t_1}{\overset{t_m}{\&Sigma;}}(T_{ia}-\stackrel{\&OverBar;}{T_i})(T_{1a}-\stackrel{\&OverBar;}{T_1})}{\underset{a=t_1}{\overset{t_m}{\&Sigma;}}{(T_{ia}-\stackrel{\&OverBar;}{T_i})}^2}$

$b_{i,1}=\stackrel{\&OverBar;}{T_1}-k_{i,1}\stackrel{\&OverBar;}{T_i};$

Step 3, utilize Pearson product-moment coefficient represent row warm data T _iwith the warm data T of row ₁between coherence, the warm data T of the row of obtaining _iwith the warm data T of row ₁correlation factor α _{i, 1}, and draw:

T ₁with T ₂between correlation factor be α _2,1,

T ₁with T ₃between correlation factor be α _3,1,

T ₁with T _nbetween correlation factor be α _{n, 1};

Step 4: the warm data T of the row of obtaining obtained according to step 2 _iwith the warm data T of row ₁relation function, the row's temperature data T respectively t recorded ₂to T _nbe brought into relation function: T ₁=f _{i, 1}(T _i) in, the warm data T of the row of obtaining respectively ₁predicted value T _2,1, T _3,1... T _{i, 1}, T _{n, 1}:

T _i,1＝f _i,1(T _i)，i＝2,3,4,…,n

Wherein, f _{i, 1}represent row's temperature data T that step 2 draws _iwith the warm data T of row ₁relation function; T refers to t ₁to t _min a moment;

Step 5: the row's temperature theoretical value T calculating temperature point 1 during t gas turbine fault-free ₁';

Step 6: the theoretical value T defining temperature point 1 during each moment unit fault-free ₁' with the measured value T of temperature point 1 ₁difference be △ T ₁: △ T ₁=T ₁'-T ₁; By gas turbine t _mrow's temperature data T=[T in period ₁, T ₂..., T _i... T _n] substitute into step 4 and step 5, obtain △ T ₁meet that average is 0, standard deviation is σ ₁normal distribution, i.e. △ T ₁~ N (0, σ ₁);

Step 7: monitoring gas turbine being entered to operation phase temperature point to be measured 1: row's temperature data of being recorded in each moment substitute into step 4 and step 5, if the theoretical value T of temperature point 1 when obtaining unit failure-free operation ₁'; Monitor the theoretical value T of temperature point 1 during each moment unit fault-free ₁' with the measured value T of temperature point 1 ₁difference be △ T ₁if, △ T ₁at [-3 σ ₁, 3 σ ₁] scope in, then unit fault-free is described, if exceed [-3 σ ₁, 3 σ ₁] scope, then illustrate unit break down;

Step 8: monitoring gas turbine being entered to operation phase temperature point to be measured 2 to n: repeat step 2 to step 6, if the theoretical value T of temperature point 2 to n when obtaining unit fault-free respectively ₂', T ₃' ..., T _n', and the difference △ T of theoretical value and measured value ₂, △ T ₃..., △ T _nstandard deviation △ σ ₂, △ σ ₃..., △ σ _n; Correspondingly, if △ is T ₂, △ T ₃..., △ T _nall respectively at [-3 σ _n, 3 σ _n] scope, then unit fault-free is described, if exceed [-3 σ ₁, 3 σ ₁] scope, then illustrate unit break down.

Beneficial effect of the present invention is:

When combustion barrel occurs abnormal time, the result of row's temperature also there will be exception, judge whether the working condition of combustion barrel occurs exception by the situation of arranging temperature, thus monitor that the delivery temperature of gas turbine carrys out the working condition in indirect monitor firing chamber, obtain the operation conditions of monitoring combustion system.

The present invention considers the coherence between different temperatures measuring point, the adding by correlation factor when calculating the coherence between different temperatures measuring point, enhance the weights of those measuring points strong with certain measuring point coherence, reduce the weights of those measuring points weak with this measuring point coherence.Compared with prior art, the inventive method realizes the on-line monitoring of gas turbine row temperature, and the coherence between the row's of making full use of each measuring point of temperature, accurately detects abnormal evolution process.The present invention can realize the exception monitoring of combustion engine row temperature better, finds fault discovery fault even comparatively early timely, thus reduce because gas turbine produce fault can not the possibility that causes of Timeliness coverage.

Accompanying drawing explanation

Fig. 1 is the firing chamber that relates to of background technique of the present invention and thermocouple deployment scenarios schematic diagram;

Fig. 2 is the discrimination principles figure of the combustion monitoring that background technique of the present invention relates to;

Fig. 3 is flow chart of the present invention;

Embodiment

Embodiment one:

The gas turbine combustion system on-line monitoring method based on measuring point weighted value of present embodiment, the gas turbine combustion system on-line monitoring method based on measuring point weighted value described in composition graphs 3 is realized by following steps:

Step one, gas turbine turbine outlet circumference arrange n temperature point equably, gas turbine is at t _min period, normal operation, obtains t _mrow's temperature data T=[T in period ₁, T ₂..., T _i... T _n]; Wherein, T _irepresent that i-th temperature point is at t _mrow's temperature data that in period, each time point records, $T_1=\&lsqb;T_{1t_1},T_{1t_2},...,T_{1t_m}\&rsqb;,T_2=\&lsqb;T_{2t_1},T_{2t_2},...,T_{2t_m}\&rsqb;,$ $T_i=\&lsqb;T_{{\mathrm{it}}_1},T_{{\mathrm{it}}_2},...,T_{{\mathrm{it}}_m}\&rsqb;,T_n=\&lsqb;T_{{\mathrm{nt}}_1},T_{{\mathrm{nt}}_2},...,T_{{\mathrm{nt}}_m}\&rsqb;;$

Step 2: utilize the method for least squares warm data T of the row of obtaining respectively ₂, T ₃... T _i..., T _nwith the warm data T of row ₁relation function, that is:

T ₁＝f _i,1(T _i)，i＝2,3,4,…,n

Wherein, f _{i, 1}represent the warm data T of row _iwith the warm data T of row ₁relation function; Meet, T ₁=k _{i, 1}t _i+ b _{i, 1}, i=2,3,4 ..., n, wherein k _{i, 1}and b _{i, 1}can be drawn by method of least squares, that is:

$k_{i,1}=\frac{\underset{a=t_1}{\overset{t_m}{\&Sigma;}}(T_{ia}-\stackrel{\&OverBar;}{T_i})(T_{1a}-\stackrel{\&OverBar;}{T_1})}{\underset{a=t_1}{\overset{t_m}{\&Sigma;}}{(T_{ia}-\stackrel{\&OverBar;}{T_i})}^2}$

$b_{i,1}=\stackrel{\&OverBar;}{T_1}-k_{i,1}\stackrel{\&OverBar;}{T_i};$

Step 3: utilize Pearson product-moment coefficient to represent the warm data T of row ₂, T ₃... T _i... T _nwith the warm data T of row ₁between coherence, the warm data T of the row of obtaining ₂, T ₃... T _i... T _nwith the warm data T of row ₁correlation factor α _{i, 1}, and draw:

T ₁with T ₂between correlation factor be α _2,1,

T ₁with T ₃between correlation factor be α _3,1,

T ₁with T _nbetween correlation factor be α _{n, 1};

Step 4: the warm data T of the row of obtaining obtained according to step 2 ₂, T ₃... T _i... T _nwith the warm data T of row ₁relation function, the row's temperature data T respectively t recorded ₂to T _nbe brought into relation function: T _{i, 1}=f _{i, 1}(T _i), i=2,3,4 ..., in n, the warm data T of the row of obtaining respectively ₂, T ₃... T _i... T _nto T ₁predicted value T _2,1, T _3,1... T _{i, 1}, T _{n, 1}:

T _2,1＝f _2,1(T ₁)，

T _3,1＝f _3,1(T ₁)，

…

T _i,1＝f _i,1(T ₁)，

T _n,1＝f _n,1(T ₁)；

Wherein, f _{i, 1}represent row's temperature data T that step 2 draws _iwith the warm data T of row ₁relation function; T refers to t ₁to t _min a moment;

Step 5: the row's temperature theoretical value T calculating temperature point 1 during t gas turbine fault-free ₁';

Step 6: the theoretical value T defining temperature point 1 during each moment unit fault-free ₁' with the measured value T of temperature point 1 ₁difference be △ T ₁: △ T ₁=T ₁'-T ₁; By gas turbine t _mrow's temperature data T=[T in period ₁, T ₂..., T _i... T _n] substitute into step 4 and step 5, obtain △ T ₁meet that average is 0, standard deviation is σ ₁normal distribution, i.e. △ T ₁~ N (0, σ ₁);

Step 7: monitoring gas turbine being entered to operation phase temperature point to be measured 1: row's temperature data of being recorded in each moment substitute into step 4 and step 5, if the theoretical value T of temperature point 1 when obtaining unit failure-free operation ₁'; Monitor the theoretical value T of temperature point 1 during each moment unit fault-free ₁' with the measured value T of temperature point 1 ₁difference be △ T ₁if, △ T ₁at [-3 σ ₁, 3 σ ₁] scope in, then unit fault-free is described, if exceed [-3 σ ₁, 3 σ ₁] scope, then illustrate unit break down;

Step 8: monitoring gas turbine being entered to operation phase temperature point to be measured 2 to n: repeat step 2 to step 6, if the theoretical value T of temperature point 2 to n when obtaining unit fault-free respectively ₂', T ₃' ..., T _n', and the difference △ T of theoretical value and measured value ₂, △ T ₃..., △ T _nstandard deviation △ σ ₂, △ σ ₃..., △ σ _n; Correspondingly, if △ is T ₂, △ T ₃..., △ T _nall respectively at [-3 σ _n, 3 σ _n] scope, then unit fault-free is described, if exceed [-3 σ ₁, 3 σ ₁] scope, then illustrate unit break down.

Embodiment two:

With embodiment one unlike, the gas turbine combustion system on-line monitoring method based on measuring point weighted value of present embodiment, arranges warm data (T described in step 3 ₂, T ₃...) T _i(..., T _n) and the warm data T of row ₁correlation factor α _{i, 1}computational methods as follows:

${\mathrm{\&alpha;}}_{i,1}=\frac{\underset{j=t_1}{\overset{t_m}{\&Sigma;}}(T_{ij}-{\stackrel{\&OverBar;}{T}}_{ij})(T_{1j}-{\stackrel{\&OverBar;}{T}}_{1j})}{\sqrt{\underset{j=t_1}{\overset{t_m}{\&Sigma;}}{(T_{ij}-{\stackrel{\&OverBar;}{T}}_{ij})}^2}\sqrt{\underset{j=t_1}{\overset{t_m}{\&Sigma;}}{(T_{1j}-{\stackrel{\&OverBar;}{T}}_{1j})}^2}}$

Wherein, j represents in time period t _min the 1st time point to m time point, j=t ₁... t _m.

Embodiment three:

With embodiment one or two unlike, the gas turbine combustion system on-line monitoring method based on measuring point weighted value of present embodiment,

Row's temperature theoretical value T of temperature point 1 during t gas turbine fault-free is calculated described in step 5 ₁' process be, according to row warm theoretical value formula:

$T_1^{\&prime;}=\frac{{\mathrm{\&Sigma;}}_{i=2}^n{\mathrm{\&alpha;}}_{i,1}{T}_{i,1}}{{\mathrm{\&Sigma;}}_{i=2}^n{\mathrm{\&alpha;}}_{i,1}}$

Calculate row's temperature theoretical value T of temperature point 1 during t gas turbine fault-free ₁'.

Claims (3)

1. based on a gas turbine combustion system on-line monitoring method for measuring point weighted value, it is characterized in that: the described gas turbine combustion system on-line monitoring method based on measuring point weighted value is realized by following steps:

Step one, gas turbine turbine outlet circumference arrange n temperature point equably, gas turbine is at t _min period, normal operation, obtains t _mrow's temperature data T=[T in period ₁, T ₂..., T _i... T _n]; Wherein, T _irepresent that i-th temperature point is at t _mrow's temperature data that in period, each time point records, $T_1=\&lsqb;T_{1t_1},T_{1t_2},...,T_{1t_m}\&rsqb;,T_2=\&lsqb;T_{2t_1},T_{2t_2},...,T_{2t_m}\&rsqb;,$ $T_i=\&lsqb;T_{{\mathrm{it}}_1},T_{{\mathrm{it}}_2},...,T_{{\mathrm{it}}_m}\&rsqb;,T_n=\&lsqb;T_{{\mathrm{nt}}_1},T_{{\mathrm{nt}}_2},...,T_{{\mathrm{nt}}_m}\&rsqb;;$

Step 2, utilize the method for least squares warm data T of the row of obtaining respectively _iwith the warm data T of row ₁relation function, that is:

T ₁＝f _i,1(T _i)，i＝2,3,4,…,n

Wherein, f _{i, 1}represent the warm data T of row _iwith the warm data T of row ₁relation function; Meet: T ₁=k _{i, 1}t _i+ b _{i, 1}, i=2,3,4 ..., n, wherein k _{i, 1}and b _{i, 1}can be drawn by method of least squares, that is:

$k_{i,1}=\frac{\underset{a=t_1}{\overset{t_m}{\&Sigma;}}(T_{ia}-\stackrel{\&OverBar;}{T_i})(T_{1a}-\stackrel{\&OverBar;}{T_1})}{\underset{a=t_1}{\overset{t_m}{\&Sigma;}}{(T_{ia}-\stackrel{\&OverBar;}{T_i})}^2}$

$b_{i,1}=\stackrel{\&OverBar;}{T_1}-k_{i,1}\stackrel{\&OverBar;}{T_i};$

Step 3, utilize Pearson product-moment coefficient represent row warm data T _iwith the warm data T of row ₁between coherence, the warm data T of the row of obtaining _iwith the warm data T of row ₁correlation factor α _{i, 1}, and draw:

T ₁with T ₂between correlation factor be α _2,1,

T ₁with T ₃between correlation factor be α _3,1,

T ₁with T _nbetween correlation factor be α _{n, 1};

Step 4: the warm data T of the row of obtaining obtained according to step 2 _iwith the warm data T of row ₁relation function, the row's temperature data T respectively t recorded ₂to T _nbe brought into relation function: T ₁=f _{i, 1}(T _i) in, the warm data T of the row of obtaining respectively ₁predicted value T _2,1, T _3,1... T _{i, 1}, T _{n, 1}:

T _i,1＝f _i,1(T _i)，i＝2,3,4,…,n

Wherein, f _{i, 1}represent row's temperature data T that step 2 draws _iwith the warm data T of row ₁relation function; T refers to t ₁to t _min a moment;

Step 5: the row's temperature theoretical value T calculating temperature point 1 during t gas turbine fault-free ₁';

Step 6: the theoretical value T defining temperature point 1 during each moment unit fault-free ₁' with the measured value T of temperature point 1 ₁difference be Δ T ₁: Δ T ₁=T ₁'-T ₁; By gas turbine t _mrow's temperature data T=[T in period ₁, T ₂..., T _i... T _n] substitute into step 4 and step 5, obtain Δ T ₁meet that average is 0, standard deviation is σ ₁normal distribution, i.e. Δ T ₁~ N (0, σ ₁);

Step 7: monitoring gas turbine being entered to operation phase temperature point to be measured 1: row's temperature data of being recorded in each moment substitute into step 4 and step 5, if the theoretical value T of temperature point 1 when obtaining unit failure-free operation ₁'; Monitor the theoretical value T of temperature point 1 during each moment unit fault-free ₁' with the measured value T of temperature point 1 ₁difference be Δ T ₁if, Δ T ₁at [-3 σ ₁, 3 σ ₁] scope in, then unit fault-free is described, if exceed [-3 σ ₁, 3 σ ₁] scope, then illustrate unit break down;

Step 8: monitoring gas turbine being entered to operation phase temperature point to be measured 2 to n: repeat step 2 to step 6, if the theoretical value T of temperature point 2 to n when obtaining unit fault-free respectively ₂', T ₃' ..., T _n', and the difference Δ T of theoretical value and measured value ₂, Δ T ₃..., Δ T _nstandard deviation Δ σ ₂, Δ σ ₃..., Δ σ _n; Correspondingly, if Δ T ₂, Δ T ₃..., Δ T _nall respectively at [-3 σ _n, 3 σ _n] scope, then unit fault-free is described, if exceed [-3 σ ₁, 3 σ ₁] scope, then illustrate unit break down.

2., according to claim 1 based on the gas turbine combustion system on-line monitoring method of measuring point weighted value, it is characterized in that: described in step 3, arrange warm data T _iwith the warm data T of row ₁correlation factor α _{i, 1}computational methods as follows:

${\mathrm{\&alpha;}}_{i,1}=\frac{\underset{j=t_1}{\overset{t_m}{\&Sigma;}}(T_{ij}-{\stackrel{\&OverBar;}{T}}_{ij})(T_{1j}-{\stackrel{\&OverBar;}{T}}_{1j})}{\sqrt{\underset{j=t_1}{\overset{t_m}{\&Sigma;}}{(T_{ij}-{\stackrel{\&OverBar;}{T}}_{ij})}^2}\sqrt{\underset{j=t_1}{\overset{t_m}{\&Sigma;}}{(T_{1j}-{\stackrel{\&OverBar;}{T}}_{1j})}^2}}$

Wherein, j represents in time period t _min the 1st time point to m time point, j=t ₁... t _m.

3. according to claim 1 or 2 based on the gas turbine combustion system on-line monitoring method of measuring point weighted value, it is characterized in that: the row's temperature theoretical value T calculating temperature point 1 during t gas turbine fault-free described in step 5 ₁' process be, according to row warm theoretical value formula:

$T_1^{\&prime;}=\frac{{\mathrm{\&Sigma;}}_{i=2}^n{\mathrm{\&alpha;}}_{i,1}{T}_{i,1}}{{\mathrm{\&Sigma;}}_{i=2}^n{\mathrm{\&alpha;}}_{i,1}}$

Calculate row's temperature theoretical value T of temperature point 1 during t gas turbine fault-free ₁'.