CN105912878A - Gas turbine adaptive gas circuit component performance diagnostic method based on combination of thermal model and particle swarm optimization - Google Patents

Abstract

The invention aims at providing a gas turbine adaptive gas circuit component performance diagnostic method based on the combination of a thermal model and particle swarm optimization. The method comprises the steps that the gas turbine nonlinear thermal model is established, gas circuit health indexes of a gas compressor and a turbine are redefined with similarity reduced parameters, gar circuit measurement parameters in a certain time frame in the stable running process of a current object gas turbine are collected and subjected to noise reduction processing and then serve as measurement parameters of a gas circuit to be subjected to offline diagnosis, and gas circuit health indexes of current components are obtained through iteration optimizing calculation of the particle swarm optimization and used for evaluating the practical performance health condition of the object gas turbine. By means of the method, the problem that the diagnostic precision of a traditional gas turbine gas circuit component performance diagnostic method is likely to be affected by changes of environment conditions and operation conditions is solved, the local optimization feature of the traditional diagnostic algorithm is changed, the accuracy of a diagnostic result is improved, the diagnostic process is simplified, and the method can be effectively applied to performance diagnostic situations with measurement noise and for complex gas turbine units.

Description

Gas turbine self-adaptive gas circuit component performance diagnosis method based on combination of thermal model and particle swarm optimization algorithm

Technical Field

The invention relates to a diagnostic method, in particular to a method for diagnosing the performance of a gas turbine.

Background

In the second half of the 20 th century, gas turbines have received increasing attention in the industrial power plant field, oil and gas pipeline transportation, and ship industry as they are widely used in the aviation industry. In operation, various types of gas turbines are subject to gradual performance degradation due to severe operating conditions of high temperature, high pressure, high rotational speed, and high stress, and environmental pollution. The main gas path components of a gas turbine include a compressor, a combustor, and a turbine. These major components are subject to various degradation phenomena over time, such as fouling, leakage, corrosion, heat distortion, foreign object damage, etc., which can cause performance degradation and can lead to serious failure safety issues. Therefore, the performance health of current gas turbines is very important information for gas turbine users.

Most gas turbine maintenance strategies today are preventative maintenance, usually considering whether minor, medium, major repairs are required in terms of the Equivalent Operating Hours (EOH) as indicated by the engine manufacturer. The shut-down of the combustion engine, whether planned or unplanned, always represents a costly expense. In order to save maintenance costs, the user needs to adopt a maintenance strategy, i.e. predictive maintenance, based on the actual performance health of the combustion engine.

The gas path component performance diagnosis method based on the thermodynamic model decision is widely applied to the performance and health state monitoring of the gas turbine and becomes one of the key technologies for supporting the innovation of the maintenance strategy. The general gas circuit component performance diagnosis method based on thermal model decision uses the performance parameters (absolute parameters) of the gas circuit component to define the component health index, so that the gas circuit measurement parameters need to be subjected to data preprocessing before diagnosis to eliminate the influence of the change of the operating performance of the gas turbine caused by the change of environmental conditions or operating conditions and working medium components. And because the performance parameters of the components are taken as adaptive variables, the diagnosis is generally carried out in two steps: the method comprises the steps that firstly, component performance parameters such as air flow, pressure ratio of a gas compressor, isentropic efficiency of the gas compressor, turbine front temperature, turbine isentropic efficiency and other absolute performance parameters are obtained through calculation by a linear or non-linear Newton-Lepyson iterative algorithm according to gas path measurement parameters; and the second step is to compare the component operating point under the actual performance decline condition with the component operating point under the health reference condition on the same component characteristic diagram, so as to observe the degree of deviation of the characteristic line on the component characteristic diagram (namely the gas circuit component health index) at the moment, and further evaluate the performance health condition of the current component. As the number of components participating in the diagnosis increases in the gas turbine, the dimension of the failure coefficient matrix increases, and is additionally disturbed by measurement noise, the blurring effect (i.e., although some components are not actually degraded, the diagnosed degradation is distributed over almost all component gas path health indices) may be strong, and the components with actual degradation are not identified.

Disclosure of Invention

The invention aims to provide a gas turbine self-adaptive gas circuit component performance diagnosis method based on the combination of a thermal model and a particle swarm optimization algorithm, which is effectively suitable for complex gas turbine units with measurement noise.

The purpose of the invention is realized as follows:

the invention relates to a gas turbine self-adaptive gas circuit component performance diagnosis method based on the combination of a thermal model and a particle swarm optimization algorithm, which is characterized in that:

(1) establishing a non-linear thermal model of the gas turbine based on gas path measurement parameters of the object gas turbine when the gas turbine is newly put into operation or is healthy, wherein the gas compressor and the turbine are represented in a similar reduced parameter form;

(2) redefining the gas path health indexes of the gas compressor and the turbine by using similar reduced parameters, and eliminating the influence of the change of the running performance of the gas turbine caused by the change of environmental conditions;

(3) collecting gas path measurement parameters of a time interval when the current object gas turbine stably operates, and taking the gas path measurement parameters as gas path measurement parameters to be diagnosed off line after noise reduction treatment;

(4) setting the environment input condition and the operation input condition of the established thermal model of the gas turbine to be consistent with the operation condition of the object gas turbine during sampling, and eliminating the influence of the change of the operation performance of the gas turbine caused by the change of the environment condition and the operation condition;

(5) and taking the root mean square error between the gas path measurement parameters to be diagnosed off line and the gas path parameter data calculated by the thermodynamic model as a target function, and obtaining the current gas path health index of each part through iterative optimization calculation by a particle swarm optimization algorithm so as to evaluate the actual performance health condition of the target gas turbine.

The present invention may further comprise:

1. the specific steps of establishing the nonlinear thermal model of the gas turbine in the step (1) are as follows:

(a) establishing a thermal model of a gas turbine component level by utilizing the relative reduced parameters of the components, wherein the data of the gas compressor and the turbine characteristic line are arranged into a universal relative reduced parameter form:

the characteristic line of the air compressor is arranged into a general similar folding parameter form as follows:

G_C,cor,rel＝f(n_cor,rel,π_C,rel)

η_C,rel＝f(n_cor,rel,π_C,rel)

whereinIs the relative reduced rotation speed, n is the actual rotation speed,for compressor inlet stagnation temperature, R_gThe lower corner mark 0 represents a design point for a gas constant of a working medium flowing through the gas compressor;

to a relative reduced flow, G_CIn order to realize the actual inlet flow of the compressor,is the stagnation pressure of the inlet of the compressor,is a relative pressure ratio of [ pi ]_Cη for actual compressor pressure ratio_C,rel＝η_C/η_C0For relative isentropic efficiency, η_CThe isentropic efficiency of the actual compressor;

the turbine characteristic line is arranged into a general similar reduced parameter form as follows:

G_T,cor,rel＝f(n_cor,rel,π_T,rel)

η_T,rel＝f(n_cor,rel,π_T,rel)

in the formula:is the relative reduced rotation speed, n is the actual rotation speed,for turbine inlet stagnation temperature, R_gIs the constant of the gas flowing through the turbine working medium,to a relative reduced flow, G_TFor the actual turbine inlet flow rate,η for relative pressure ratio_T,rel＝η_T/η_T0For relative isentropic efficiency, η_TFor the isentropic efficiency of the actual turbine, the lower corner mark 0 represents a design point;

(b) according to the collected gas circuit measurement parameters of the target gas turbine when the gas turbine is newly put into operation or is healthy, the characteristic line data of each part is gradually corrected, the calculated value of the established gas turbine thermal model is matched with the gas circuit measurement parameters, and therefore the negative influence of the calculation error of the thermal model on the diagnosis result is eliminated.

2. The concrete steps of step (2) redefining the gas path health indexes of the gas compressor and the turbine by using similar reduced parameters are as follows:

(a) the overall performance health condition of the gas turbine is represented by the gas circuit health indexes of all parts, and the gas circuit health indexes of the gas compressor and the turbine are redefined by relative reduced parameters so as to eliminate negative influence on a diagnosis result caused by environmental condition change;

(b) the compressor gas circuit health index is defined as follows:

SF_C,FC＝G_{C,cor,rel,deg}/G_C,cor,rel

ΔSF_C,FC＝(G_{C,cor,rel,deg}-G_C,cor,rel)/G_C,cor,rel

SF_C,Eff＝η_C,deg/η_C

ΔSF_C,Eff＝(η_C,deg-η_C)/η_C

wherein SF_C,FCThe flow characteristic index of the compressor is obtained; g_{C,cor,rel,deg}The relative reduced flow is obtained when the performance of the gas compressor is declined; g_C,cor,relThe relative flow rate of the compressor is reduced when the compressor is healthy; SF_C,Effη for compressor efficiency characteristic index_C,degη isentropic efficiency for compressor performance degradation_CThe isentropic efficiency of the compressor when the compressor is healthy;

the combustor gas path health index is defined as follows:

SF_B,Eff＝η_B,deg/η_B

ΔSF_B,Eff＝(η_B,deg-η_B)/η_B

wherein SF_B,Effη is the combustion efficiency performance index of the combustion chamber_B,degη for combustion efficiency in combustion chamber performance degradation_BThe combustion efficiency is the combustion efficiency when the combustion chamber is healthy;

the turbine gas path health index is defined as follows:

SF_T,FC＝G_T,cor,deg/G_T,cor

ΔSF_T,FC＝(G_T,cor,deg-G_T,cor)/G_T,cor

SF_T,Eff＝η_T,deg/η_T

ΔSF_T,Eff＝(η_T,deg-η_T)/η_T

wherein SF_T,FCIs a turbine flow performance index; g_T,cor,degThe flow is reduced when the turbine performance is reduced; g_T,corThe flow is reduced for the health of the turbine; SF_T,Effη for turbine efficiency performance index_T,degFor isentropic efficiency in turbine performance degradation η_TThe constant entropy efficiency of the turbine is healthy.

3. And (5) taking the root mean square error between the gas path measurement parameters to be diagnosed off-line and the gas path parameter data calculated by the thermodynamic model as a target function, obtaining the current gas path health index of each component through iterative optimization calculation of a particle swarm optimization algorithm, and specifically evaluating the actual performance health condition of the target gas turbine as follows:

(a) by gas path measurement data to be diagnosed off-lineGas path parameter data calculated by thermal modelThe root mean square error between is the objective function Fitness:

$Fitness=\sqrt{\frac{\underset{i=1}{\overset{M}{\Σ}}{\[(z_{i,predicted}-z_{i,actual})/z_{i,actual}\]}^2}{M}}$

(b) iterative optimization calculation is carried out through a particle swarm optimization algorithm to obtain the current gas path health index of each component so as to evaluate the actual component performance health condition of the object gas turbine,

$\hat{z}=\[z_{1,predicted},...,z_{i,predicted},...,z_{M,predicted}\]$

$\stackrel{\→}{z}=\[z_{1,actual},...,z_{i,actual},...,z_{M,actual}\]$

whereinMeasuring parameter vectors of a gas path calculated for a thermal model of the gas turbine;locating the actually measured gas path parameter vector; m is the number of gas path measurement parameters;

$\mathrm{\Δ}\stackrel{\→}{SF}=\[{\mathrm{\ΔSF}}_{C,FC},{\mathrm{\ΔSF}}_{C,Eff},{\mathrm{\ΔSF}}_{B,Eff},{\mathrm{\ΔSF}}_{T,FC},{\mathrm{\ΔSF}}_{T,Eff}\]$

whereinThe gas path component health index;

taking a root mean square error as an objective function, the following formula is given:

$Fitness=\sqrt{\frac{\underset{i=1}{\overset{M}{\Σ}}{\[(z_{i,predicted}-z_{i,actual})/z_{i,actual}\]}^2}{M}}$

in the formula, Fitness is an optimization target and is optimized along with iterationWhen Fitness approaches 0, the calculated gas path measurement parameterAnd measured gas path parametersMatching, and outputting the final global optimal solution

The invention has the advantages that:

(1) according to the invention, a thermal model of a gas turbine component level is established by utilizing the relative reduced parameters of the components according to the aerodynamic thermodynamic characteristics of the gas turbine, wherein the data of the characteristic lines of the gas compressor and the turbine are arranged into a universal relative reduced parameter form.

(2) Compared with the prior art, the method can more accurately represent the change of the gas path health index caused by the performance degradation of the components, and eliminate the negative influence on the diagnosis result caused by the change of environmental conditions (atmospheric pressure, temperature and relative humidity).

(3) The invention improves the local optimization characteristic of the traditional diagnosis algorithm (Newton-Laprison iterative algorithm) from the perspective of global optimization (adopting a particle swarm optimization algorithm), improves the accuracy of diagnosis results, simplifies the diagnosis process, and can be effectively suitable for the performance diagnosis conditions of the complex gas turbine unit with measurement noise.

Drawings

FIG. 1 is a flow chart of the present invention;

FIG. 2 is a schematic diagram of the diagnostic process of the present invention;

FIG. 3 is a gas turbine gas circuit working cross-section marking diagram for a certain type of three-axis ship;

FIG. 4 is a diagnosis result of a gas turbine diagnosis case for a triaxial vessel;

fig. 5 is an algorithm iterative computational search process of the present invention for the diagnostic case.

Detailed Description

The invention will now be described in more detail by way of example with reference to the accompanying drawings in which:

with reference to fig. 1-5, the method for diagnosing the performance of the self-adaptive gas circuit component of the gas turbine based on the combination of the thermal model and the particle swarm optimization algorithm comprises the following steps:

step 1), establishing a gas turbine nonlinear thermal model capable of completely reflecting the characteristics of each part based on gas path measurement parameters when a target gas turbine is newly put into operation (or is healthy), wherein a gas compressor and a turbine are represented in a similar reduced parameter form;

step 2), redefining the gas path health indexes of the gas compressor and the turbine by using similar reduced parameters, and eliminating the influence of the change of the operation performance of the gas turbine caused by the change of environmental conditions (atmospheric pressure, temperature and relative humidity);

step 3), collecting gas path measurement parameters of a certain time period when the current object gas turbine stably operates, and taking the gas path measurement parameters as gas path measurement parameters to be diagnosed off line after noise reduction treatment;

and 4), setting the environmental input conditions (atmospheric pressure, temperature and relative humidity) and the operation input conditions of the established thermodynamic model of the gas turbine to be consistent with the operation conditions of the target gas turbine during sampling, and eliminating the influence of the change of the operation performance of the gas turbine caused by the change of the environmental conditions (atmospheric pressure, temperature and relative humidity) and the operation conditions.

And 5) taking the root mean square error between the gas path measurement data to be diagnosed off line and the gas path parameter data calculated by the thermodynamic model as a target function, and obtaining the current gas path health indexes of each part (a gas compressor, a turbine and a combustion chamber) through iterative optimization calculation by a particle swarm optimization algorithm so as to evaluate the actual performance health condition of the target gas turbine.

As a further optimization scheme of the gas turbine adaptive gas circuit component performance diagnosis method based on the combination of the thermal model and the particle swarm optimization algorithm, the specific steps of establishing the gas turbine nonlinear thermal model capable of completely reflecting the characteristics of each component based on the gas circuit measurement parameters when the object gas turbine is newly put into operation (or is healthy) in the step 1) are as follows:

step 1.1), establishing a gas turbine component-level thermal model by utilizing component relative reduced parameters according to the aerodynamic and thermodynamic characteristics of the gas turbine, wherein the data of a gas compressor and a turbine characteristic line are arranged into a universal relative reduced parameter form.

The characteristic line of the gas compressor is arranged into a universal similar folding parameter form as follows:

G_C,cor,rel＝f(n_cor,rel,π_C,rel)

η_C,rel＝f(n_cor,rel,π_C,rel)

whereinIs the relative reduced rotation speed, n is the actual rotation speed,for compressor inlet stagnation temperature, R_gThe lower corner mark 0 represents a design point for a gas constant of a working medium flowing through the gas compressor;

to a relative reduced flow, G_CIn order to realize the actual inlet flow of the compressor,is the stagnation pressure of the inlet of the compressor,is a relative pressure ratio of [ pi ]_Cη for actual compressor pressure ratio_C,rel＝η_C/η_C0For relative isentropic efficiency, η_CThe isentropic efficiency of the actual compressor is obtained.

The turbine characteristic line is arranged into a general similar reduced parameter form as follows:

G_T,cor,rel＝f(n_cor,rel,π_T,rel)

η_T,rel＝f(n_cor,rel,π_T,rel)

in the formula:is the relative reduced rotation speed, n is the actual rotation speed,for turbine inlet stagnation temperature, R_gIs the constant of the gas flowing through the turbine working medium,to a relative reduced flow, G_TFor the actual turbine inlet flow rate,η for relative pressure ratio_T,rel＝η_T/η_T0For relative isentropic efficiency, η_TFor the actual turbine isentropic efficiency, the lower corner 0 represents the design point.

And 1.2) gradually correcting characteristic line data (including design working conditions and variable working conditions) of each part according to collected gas path measurement parameters (after noise reduction treatment) of the target gas turbine when the gas turbine is newly put into operation (or is healthy), so that a calculated value of the established gas turbine thermal model is matched with the gas path measurement parameters, and negative influence on a diagnosis result caused by a calculation error of the thermal model is eliminated.

As a further optimization scheme of the gas turbine self-adaptive gas circuit component performance diagnosis method based on the combination of the thermal model and the particle swarm optimization algorithm, the specific step of redefining the gas circuit health indexes of the gas compressor and the turbine by using similar reduced parameters in the step 2) is as follows:

and 2.1), the overall performance health condition of the gas turbine can be represented by gas circuit health indexes of all main parts, such as flow characteristic indexes and efficiency characteristic indexes of a compressor and a turbine and efficiency characteristic indexes of a combustion chamber. The gas path health indices (representing the deviations of the component characteristic lines) of the compressor and the turbine are redefined using relative reduced parameters to eliminate negative effects on the diagnostic results due to changes in environmental conditions (atmospheric pressure, temperature and relative humidity).

Step 2.2), the gas path health index of the gas compressor is defined as follows:

SF_C,FC＝G_{C,cor,rel,deg}/G_C,cor,rel

ΔSF_C,FC＝(G_{C,cor,rel,deg}-G_C,cor,rel)/G_C,cor,rel

SF_C,Eff＝η_C,deg/η_C

ΔSF_C,Eff＝(η_C,deg-η_C)/η_C

wherein SF_C,FCThe flow characteristic index of the compressor is obtained; g_{C,cor,rel,deg}The relative reduced flow of the compressor (when the performance is degraded); g_C,cor,relThe relative reduced flow of the compressor (in the healthy state); SF_C,Effη for compressor efficiency characteristic index_C,degFor isentropic efficiency of compressor (in performance degradation) η_CThe isentropic efficiency of the air compressor (in the healthy state).

The combustor gas path health index is defined as follows:

SF_B,Eff＝η_B,deg/η_B

ΔSF_B,Eff＝(η_B,deg-η_B)/η_B

wherein SF_B,Effη is the combustion efficiency performance index of the combustion chamber_B,degη for combustion efficiency of combustion chamber (performance degradation)_BThe combustion efficiency of the combustion chamber (in the healthy state).

The turbine gas path health index is defined as follows:

SF_T,FC＝G_T,cor,deg/G_T,cor

ΔSF_T,FC＝(G_T,cor,deg-G_T,cor)/G_T,cor

SF_T,Eff＝η_T,deg/η_T

ΔSF_T,Eff＝(η_T,deg-η_T)/η_T

wherein SF_T,FCIs a turbine flow performance index; g_T,cor,degThe flow rate is reduced for the turbine (when the performance is degraded); g_T,corThe flow is reduced for the turbine (in the health state); SF_T,Effη for turbine efficiency performance index_T,degFor turbine (performance degradation) isentropic efficiency η_TIs the turbine (healthy) isentropic efficiency.

As a further optimization scheme of the gas turbine adaptive gas circuit component performance diagnosis method based on the combination of the thermal model and the particle swarm optimization algorithm, in the step 5), the root mean square error between the gas circuit measurement data to be diagnosed off-line and the gas circuit parameter data calculated by the thermal model is taken as a target function, the iterative optimization calculation is performed through the particle swarm optimization algorithm to obtain the current gas circuit health index of each component (a gas compressor, a turbine and a combustion chamber), and the specific steps for evaluating the actual performance health condition of the target gas turbine are as follows:

step 5.1), using the gas path measurement data to be diagnosed off lineGas path parameter data calculated by thermal modelThe root mean square error between is the objective function Fitness:

$Fitness=\sqrt{\frac{\underset{i=1}{\overset{M}{\Σ}}{\[(z_{i,predicted}-z_{i,actual})/z_{i,actual}\]}^2}{M}}$

and 5.2) obtaining the current gas path health index of each component (a gas compressor, a turbine and a combustion chamber) through iterative optimization calculation of a particle swarm optimization algorithm so as to evaluate the actual component performance health condition of the target gas turbine. A schematic diagram of a gas turbine adaptive gas circuit component performance diagnosis method based on combination of a thermal model and a particle swarm optimization algorithm is shown in FIG. 2.

Wherein,

$\hat{z}=\[z_{1,predicted},...,z_{i,predicted},...,z_{M,predicted}\]$

$\stackrel{\→}{z}=\[z_{1,actual},...,z_{i,actual},...,z_{M,actual}\]$

whereinMeasuring parameter vectors of a gas path calculated for a thermal model of the gas turbine;locating the actually measured gas path parameter vector; m is the number of gas path measurement parameters;

$\mathrm{\Δ}\stackrel{\→}{SF}=\[{\mathrm{\ΔSF}}_{C,FC},{\mathrm{\ΔSF}}_{C,Eff},{\mathrm{\ΔSF}}_{B,Eff},{\mathrm{\ΔSF}}_{T,FC},{\mathrm{\ΔSF}}_{T,Eff}\]$

whereinGas path component health index (as a particle in a particle swarm optimization algorithm).

Here, the root mean square error is taken as an objective function, as shown in the formula:

$Fitness=\sqrt{\frac{\underset{i=1}{\overset{M}{\Σ}}{\[(z_{i,predicted}-z_{i,actual})/z_{i,actual}\]}^2}{M}}$

wherein Fitness is an optimization target, and along with iterative optimization, when the Fitness is gradually close to 0, the calculated gas path measurement parametersAnd measured gas path parametersMatching, and outputting the final global optimal solution

The core technical content of the invention is to solve the problem that the diagnosis precision of the traditional gas turbine gas path component performance diagnosis method is easily influenced by environmental conditions (atmospheric pressure, temperature and relative humidity) and operation condition changes, improve the local optimization characteristic of the traditional diagnosis algorithm (Newton-Lapton iterative algorithm) from the perspective of global optimization, improve the accuracy of diagnosis results, simplify the diagnosis process and be effectively suitable for the performance diagnosis conditions of the complex gas turbine unit with measurement noise. According to the aerodynamic and thermodynamic characteristics of the gas turbine, a part-level thermodynamic model of the gas turbine is established by utilizing part relative reduced parameters, wherein data of a gas compressor and turbine characteristic lines are arranged into a universal relative reduced parameter form, and compared with the prior art, the method can be used for setting health indexes of all gas path parts during gas path diagnosis more simply and conveniently; redefining the gas path health indexes (representing the deviation of component characteristic lines) of the gas compressor and the turbine by using relative reduced parameters, and compared with the prior art, more accurately representing the change of the gas path health indexes caused by the performance degradation of components and eliminating the negative influence on a diagnosis result caused by the change of environmental conditions (atmospheric pressure, temperature and relative humidity); from the perspective of global optimization (adopting a particle swarm optimization algorithm), the local optimization characteristic of the traditional diagnosis algorithm (Newton-Lapton iterative algorithm) is improved, the accuracy of a diagnosis result is improved, the diagnosis process is simplified, and the method can be effectively applied to the performance diagnosis conditions of the complex gas turbine unit with measurement noise.

Taking the health diagnosis of the gas path component of a gas turbine for a triaxial ship as an example, a gas path working cross-section marking diagram of the gas turbine for the triaxial ship is shown in fig. 3. This type of three-shaft gas turbine comprises two compressors (i.e. a low-pressure compressor (LC) and a high-pressure compressor (HC)), a combustion chamber (B) and three turbines (i.e. a high-pressure turbine (HT), a low-pressure turbine (LT) and a Power Turbine (PT)), wherein the generator is connected to the Power Turbine (PT) through a reduction gearbox. The output work of the low-pressure turbine (LT) drives the low-pressure compressor (LC) through a low-pressure shaft to compress the air coming out of the air inlet, and the output work of the high-pressure turbine (HT) drives the high-pressure compressor (HC) through a high-pressure shaft to continue to compress the air coming out of the low-pressure compressor (LC). High-pressure air from a high-pressure compressor (HC) enters a combustion chamber (B) to perform combustion chemical reaction with fuel to generate high-temperature and high-pressure fuel gas, and the fuel gas sequentially enters a high-pressure turbine (HT), a low-pressure turbine (LT) and a Power Turbine (PT) to drive the turbines to output work. Finally, a Power Turbine (PT) drives a generator through a reduction gearbox to produce electrical power. Meanwhile, cooling air extracted from the compressor flows into a hot-end airflow channel to cool the static blades, the movable blades and the wheel discs of the first stages of the turbines. When the gas turbine is operating steadily, the electrical power of the generator and the rotational speed of the Power Turbine (PT) are usually maintained constant as the main control parameters. The gas circuit measurement parameters of the unit are shown in table 1, and the health indexes of all gas circuit components are shown in table 2.

TABLE 1 gas path measurement parameters of this type of gas turbine unit

TABLE 2 health index of major gas path components

A nonlinear thermal model of the gas turbine which can completely reflect the characteristics of each part is established based on gas path measurement parameters when the gas turbine is put into operation (or is healthy), wherein a compressor and a turbine are represented in a relative reduced parameter form.

The characteristic line of the gas compressor is arranged into a universal similar folding parameter form as follows:

G_C,cor,rel＝f(n_cor,rel,π_C,rel)

η_C,rel＝f(n_cor,rel,π_C,rel)

whereinIs the relative reduced rotation speed, n is the actual rotation speed,for compressor inlet stagnation temperature, R_gThe lower corner mark 0 represents a design point for a gas constant of a working medium flowing through the gas compressor;

to a relative reduced flow, G_CIn order to realize the actual inlet flow of the compressor,is the stagnation pressure of the inlet of the compressor,is a relative pressure ratio of [ pi ]_Cη for actual compressor pressure ratio_C,rel＝η_C/η_C0For relative isentropic efficiency, η_CThe isentropic efficiency of the actual compressor is obtained.

The turbine characteristic line is arranged into a general similar reduced parameter form as follows:

G_T,cor,rel＝f(n_cor,rel,π_T,rel)

η_T,rel＝f(n_cor,rel,π_T,rel)

in the formula:is the relative reduced rotation speed, n is the actual rotation speed,for turbine inlet stagnation temperature, R_gIs the constant of the gas flowing through the turbine working medium,to a relative reduced flow, G_TFor actual turbine admissionThe flow rate of the port is controlled,η for relative pressure ratio_T,rel＝η_T/η_T0For relative isentropic efficiency, η_TFor the actual turbine isentropic efficiency, the lower corner 0 represents the design point.

And establishing a thermal model of the object gas turbine on a Matlab simulation platform. The input condition of the thermodynamic model is an environmental condition (atmospheric temperature t)₀Pressure P₀Relative humidity RH), generator output power Ne (as operating conditions), fuel composition, fuel lower heating value, gas path component health index(for a new commissioning group of units,). The calculated output of the thermodynamic model is the fuel flow G_fThermal parameters (such as total pressure and total temperature) and rotating speed at the cross sections of the inlet and outlet gas paths of each part.

The gas path health indexes (representing the deviation of component characteristic lines) of the compressor and the turbine are redefined by using relative reduced parameters, and the influence of the change of the operation performance of the gas turbine caused by the change of environmental conditions (atmospheric pressure, temperature and relative humidity) is eliminated.

The compressor gas circuit health index is defined as follows:

SF_C,FC＝G_{C,cor,rel,deg}/G_C,cor,rel

ΔSF_C,FC＝(G_{C,cor,rel,deg}-G_C,cor,rel)/G_C,cor,rel

SF_C,Eff＝η_C,deg/η_C

ΔSF_C,Eff＝(η_C,deg-η_C)/η_C

whereinSF_C,FCThe flow characteristic index of the compressor is obtained; g_{C,cor,rel,deg}The relative reduced flow of the compressor (when the performance is degraded); g_C,cor,relThe relative reduced flow of the compressor (in the healthy state); SF_C,Effη for compressor efficiency characteristic index_C,degFor isentropic efficiency of compressor (in performance degradation) η_CThe isentropic efficiency of the air compressor (in the healthy state).

The combustor gas path health index is defined as follows:

SF_B,Eff＝η_B,deg/η_B

ΔSF_B,Eff＝(η_B,deg-η_B)/η_B

wherein SF_B,Effη is the combustion efficiency performance index of the combustion chamber_B,degη for combustion efficiency of combustion chamber (performance degradation)_BThe combustion efficiency of the combustion chamber (in the healthy state).

The turbine gas path health index is defined as follows:

SF_T,FC＝G_T,cor,deg/G_T,cor

ΔSF_T,FC＝(G_T,cor,deg-G_T,cor)/G_T,cor

SF_T,Eff＝η_T,deg/η_T

ΔSF_T,Eff＝(η_T,deg-η_T)/η_T

wherein SF_T,FCIs a turbine flow performance index; g_T,cor,degThe flow rate is reduced for the turbine (when the performance is degraded); g_T,corThe flow is reduced for the turbine (in the health state); SF_T,Effη for turbine efficiency performance index_T,degFor turbine (performance degradation) isentropic efficiency η_TIs the turbine (healthy) isentropic efficiency.

And acquiring gas path measurement parameters of a certain time period when the current object gas turbine stably operates, and performing noise reduction treatment to obtain the gas path measurement parameters to be subjected to offline diagnosis.

Setting the environmental input conditions (atmospheric temperature t) of the established thermodynamic model of the gas turbine₀Pressure P₀Relative humidity RH) and the operation input condition (generator output power Ne) are consistent with the operation condition of the subject gas turbine at the time of sampling, eliminating the influence of the variation in the operation performance of the gas turbine due to the variation in the environmental condition and the operation condition.

By gas path measurement data to be diagnosed off-lineGas path parameter data calculated by thermal modelThe root mean square error of the components is an objective function Fitness, and the gas path health indexes (of the components for evaluating the actual component performance health conditions of the target gas turbine) of the current components (a compressor, a turbine and a combustion chamber) are obtained through iterative optimization calculation by a particle swarm optimization algorithm, wherein a schematic diagram of a diagnosis process of the gas turbine self-adaptive gas path component performance diagnosis method based on the combination of a thermal model and the particle swarm optimization algorithm is shown in FIG. 1.

Wherein,

$\hat{z}=\[z_{1,predicted},...,z_{i,predicted},...,z_{M,predicted}\]$

$\stackrel{\→}{z}=\[z_{1,actual},...,z_{i,actual},...,z_{M,actual}\]$

whereinMeasuring parameter vectors of a gas path calculated for a thermal model of the gas turbine;locating the actually measured gas path parameter vector; m is the number of gas path measurement parameters;

$\mathrm{\Δ}\stackrel{\→}{SF}=\[{\mathrm{\ΔSF}}_{C,FC},{\mathrm{\ΔSF}}_{C,Eff},{\mathrm{\ΔSF}}_{B,Eff},{\mathrm{\ΔSF}}_{T,FC},{\mathrm{\ΔSF}}_{T,Eff}\]$

whereinGas path component health index (as a particle in a particle swarm optimization algorithm).

Here, the root mean square error is taken as an objective function, as shown in the formula:

$Fitness=\sqrt{\frac{\underset{i=1}{\overset{M}{\Σ}}{\[(z_{i,predicted}-z_{i,actual})/z_{i,actual}\]}^2}{M}}$

wherein Fitness is an optimization target, and along with iterative optimization, when the Fitness is gradually close to 0, the calculated gas path measurement parametersAnd measured gas path parametersMatching, and outputting the final global optimal solution

The parameters related to the PSO algorithm are set as shown in Table 3, where the evolution generation number is 80 and the population size is 60, and used for searching the optimal health index of the component(as shown in table 3).

TABLE 3 selection of parameters associated with particle swarm optimization

Parameter(s)	Value of
		Population size	60
Evolution algebra	80

After the diagnosis step of the gas turbine adaptive gas circuit component performance diagnosis method based on the combination of the thermal model and the particle swarm optimization algorithm is performed, a final diagnosis result can be obtained and is shown in fig. 4.

The PSO-GPA is a performance diagnosis method of the self-adaptive gas circuit component of the gas turbine based on the combination of a thermal model and a particle swarm optimization algorithm, and the GPA is a traditional diagnosis method. The iterative computation search process of the gas turbine self-adaptive gas circuit component performance diagnosis method based on the combination of the thermal model and the particle swarm optimization algorithm in the diagnosis case is shown in FIG. 5.

As can be seen from fig. 4, since the core algorithm (newton-raphson iteration algorithm) of the conventional diagnosis method is essentially a local iteration optimization method, and the core algorithm of the gas turbine adaptive gas circuit component performance diagnosis method based on the combination of the thermal model and the particle swarm optimization algorithm is essentially a global iteration optimization method, the gas turbine adaptive gas circuit component performance diagnosis method based on the combination of the thermal model and the particle swarm optimization algorithm can more effectively eliminate the fuzzy effect than the conventional diagnosis method, and accurately identify and isolate the components with degraded performance.

Claims (4)

1. The gas turbine self-adaptive gas path component performance diagnosis method based on the combination of the thermal model and the particle swarm optimization algorithm is characterized in that:

(1) establishing a non-linear thermal model of the gas turbine based on gas path measurement parameters of the object gas turbine when the gas turbine is newly put into operation or is healthy, wherein the gas compressor and the turbine are represented in a similar reduced parameter form;

(2) redefining the gas path health indexes of the gas compressor and the turbine by using similar reduced parameters, and eliminating the influence of the change of the running performance of the gas turbine caused by the change of environmental conditions;

(3) collecting gas path measurement parameters of a time interval when the current object gas turbine stably operates, and taking the gas path measurement parameters as gas path measurement parameters to be diagnosed off line after noise reduction treatment;

(4) setting the environment input condition and the operation input condition of the established thermal model of the gas turbine to be consistent with the operation condition of the object gas turbine during sampling, and eliminating the influence of the change of the operation performance of the gas turbine caused by the change of the environment condition and the operation condition;

(5) and taking the root mean square error between the gas path measurement parameters to be diagnosed off line and the gas path parameter data calculated by the thermodynamic model as a target function, and obtaining the current gas path health index of each part through iterative optimization calculation by a particle swarm optimization algorithm so as to evaluate the actual performance health condition of the target gas turbine.

2. The method for diagnosing the performance of the self-adaptive gas path component of the gas turbine based on the combination of the thermodynamic model and the particle swarm optimization algorithm according to claim 1, wherein the method comprises the following steps: the specific steps of establishing the nonlinear thermal model of the gas turbine in the step (1) are as follows:

(a) establishing a thermal model of a gas turbine component level by utilizing the relative reduced parameters of the components, wherein the data of the gas compressor and the turbine characteristic line are arranged into a universal relative reduced parameter form:

the characteristic line of the air compressor is arranged into a general similar folding parameter form as follows:

G_C,cor,rel＝f(n_cor,rel,π_C,rel)

η_C,rel＝f(n_cor,rel,π_C,rel)

whereinIs the relative reduced rotation speed, n is the actual rotation speed,for compressor inlet stagnation temperature, R_gFor the gas constant of the working medium flowing through the compressor, the lower corner mark 0 represents the design point；

To a relative reduced flow, G_CIn order to realize the actual inlet flow of the compressor,is the stagnation pressure of the inlet of the compressor,is a relative pressure ratio of [ pi ]_Cη for actual compressor pressure ratio_C,rel＝η_C/η_C0For relative isentropic efficiency, η_CThe isentropic efficiency of the actual compressor;

the turbine characteristic line is arranged into a general similar reduced parameter form as follows:

G_T,cor,rel＝f(n_cor,rel,π_T,rel)

η_T,rel＝f(n_cor,rel,π_T,rel)

in the formula:is the relative reduced rotation speed, n is the actual rotation speed,for turbine inlet stagnation temperature, R_gIs the constant of the gas flowing through the turbine working medium,to a relative reduced flow, G_TFor the actual turbine inlet flow rate,η for relative pressure ratio_T,rel＝η_T/η_T0For relative isentropic efficiency, η_TFor the isentropic efficiency of the actual turbine, the lower corner mark 0 represents a design point;

(b) according to the collected gas circuit measurement parameters of the target gas turbine when the gas turbine is newly put into operation or is healthy, the characteristic line data of each part is gradually corrected, the calculated value of the established gas turbine thermal model is matched with the gas circuit measurement parameters, and therefore the negative influence of the calculation error of the thermal model on the diagnosis result is eliminated.

3. The method for diagnosing the performance of the self-adaptive gas path component of the gas turbine based on the combination of the thermodynamic model and the particle swarm optimization algorithm according to claim 2, wherein the method comprises the following steps: the concrete steps of step (2) redefining the gas path health indexes of the gas compressor and the turbine by using similar reduced parameters are as follows:

(a) the overall performance health condition of the gas turbine is represented by the gas circuit health indexes of all parts, and the gas circuit health indexes of the gas compressor and the turbine are redefined by relative reduced parameters so as to eliminate negative influence on a diagnosis result caused by environmental condition change;

(b) the compressor gas circuit health index is defined as follows:

SF_C,FC＝G_{C,cor,rel,deg}/G_C,cor,rel

ΔSF_C,FC＝(G_{C,cor,rel,deg}-G_C,cor,rel)/G_C,cor,rel

SF_C,Eff＝η_C,deg/η_C

ΔSF_C,Eff＝(η_C,deg-η_C)/η_C

wherein SF_C,FCThe flow characteristic index of the compressor is obtained; g_{C,cor,rel,deg}The relative reduced flow is obtained when the performance of the gas compressor is declined; g_C,cor,relThe relative flow rate of the compressor is reduced when the compressor is healthy; SF_C,Effη for compressor efficiency characteristic index_C,degη isentropic efficiency for compressor performance degradation_CThe isentropic efficiency of the compressor when the compressor is healthy;

the combustor gas path health index is defined as follows:

SF_B,Eff＝η_B,deg/η_B

ΔSF_B,Eff＝(η_B,deg-η_B)/η_B

wherein SF_B,Effη is the combustion efficiency performance index of the combustion chamber_B,degη for combustion efficiency in combustion chamber performance degradation_BThe combustion efficiency is the combustion efficiency when the combustion chamber is healthy;

the turbine gas path health index is defined as follows:

SF_T,FC＝G_T,cor,deg/G_T,cor

ΔSF_T,FC＝(G_T,cor,deg-G_T,cor)/G_T,cor

SF_T,Eff＝η_T,deg/η_T

ΔSF_T,Eff＝(η_T,deg-η_T)/η_T

wherein SF_T,FCIs a turbine flow performance index; g_T,cor,degThe flow is reduced when the turbine performance is reduced; g_T,corThe flow is reduced for the health of the turbine; SF_T,Effη for turbine efficiency performance index_T,degFor isentropic efficiency in turbine performance degradation η_TThe constant entropy efficiency of the turbine is healthy.

4. The method for diagnosing the performance of the self-adaptive gas path component of the gas turbine based on the combination of the thermodynamic model and the particle swarm optimization algorithm according to claim 3, wherein the method comprises the following steps: and (5) taking the root mean square error between the gas path measurement parameters to be diagnosed off-line and the gas path parameter data calculated by the thermodynamic model as a target function, obtaining the current gas path health index of each component through iterative optimization calculation of a particle swarm optimization algorithm, and specifically evaluating the actual performance health condition of the target gas turbine as follows:

(a) by gas path measurement data to be diagnosed off-lineGas path parameter data calculated by thermal modelThe root mean square error between is the objective function Fitness:

$Fitness=\sqrt{\frac{\underset{i=1}{\overset{M}{\Σ}}{\[(z_{i,predicted}-z_{i,actual})/z_{i,actual}\]}^2}{M}}$

(b) iterative optimization calculation is carried out through a particle swarm optimization algorithm to obtain the current gas path health index of each component so as to evaluate the actual component performance health condition of the object gas turbine,

$\hat{z}=\[z_{1,predicted},...,z_{i,predicted},...,z_{M,predicted}\]$

$\stackrel{\→}{z}=\[z_{1,actual},...,z_{i,actual},...,z_{M,actual}\]$

whereinMeasuring parameter vectors of a gas path calculated for a thermal model of the gas turbine;locating the actually measured gas path parameter vector; m is the number of gas path measurement parameters;

$\mathrm{\Δ}\stackrel{\→}{SF}=\[{\mathrm{\ΔSF}}_{C,FC},{\mathrm{\ΔSF}}_{C,Eff},{\mathrm{\ΔSF}}_{B,Eff},{\mathrm{\ΔSF}}_{T,FC},{\mathrm{\ΔSF}}_{T,Eff}\]$

whereinThe gas path component health index;

taking a root mean square error as an objective function, the following formula is given:

$Fitness=\sqrt{\frac{\underset{i=1}{\overset{M}{\Σ}}{\[(z_{i,predicted}-z_{i,actual})/z_{i,actual}\]}^2}{M}}$

in the formula, Fitness is an optimization target, and as iteration optimization is carried out, when the Fitness approaches to 0, the calculated gas path measurement parametersAnd measured gas path parametersMatching, and outputting the final global optimal solution