A method for diagnosing the performance of a gas path of a gas turbine

Abstract

The application discloses a kind of gas path performance diagnosis methods of gas turbine, comprising: in the preset environment and under the preset working condition, the actual gas path measurable parameter of the gas turbine to be diagnosed is obtained, and the standard gas path measurable parameter under the healthy state is obtained;With the two as the basis, the relative deviation amount of the gas path measurable parameter of the gas turbine to be diagnosed is determined;Determine the similarity degree value between the relative deviation amount of the gas path measurable parameter and the standard gas path measurable parameter deviation amount set of each of the N preset gas path performance degradation modes, and all similarity degree values are arranged in descending order, to obtain the ordered array consisting of N preset gas path performance degradation modes;From the ordered array, obtain target preset gas path performance degradation mode and target health factor group, target preset gas path performance degradation mode refers to multiple undetermined health factors in undetermined health factor group should satisfy rationality judgment condition.The application improves the accuracy of diagnosis, and the diagnosis speed.

Description

Technical Field

[0001] This invention relates to the field of gas turbine technology, and in particular to a method for diagnosing the gas path performance of a gas turbine. Background Technology

[0002] During the operation of a gas turbine, the main gas path components, such as the compressor, combustion chamber, and turbine, operate under high temperature, high pressure, high speed, and high flow rate conditions. Unfiltered dust in the air, corrosive components in combustion products, friction between rotating parts and the combustion chamber, and carbon deposits on the combustion chamber nozzles will inevitably lead to problems such as fouling, corrosion, wear, and ablation on the compressor or turbine blades. These problems can range from reducing the gas turbine's thermal cycle efficiency and output power, negatively impacting component lifespan, to causing component damage and unexpected shutdowns, ultimately increasing maintenance costs. Therefore, diagnosing the gas path of a gas turbine is a necessary means of maintenance.

[0003] In related technologies, the health status of the gas turbine's gas path is mainly diagnosed through gas path performance diagnostic models. However, under normal circumstances, the number of measurable parameters in the gas path is less than the number of component health factors, leading to inaccurate diagnosis of the degree of gas path performance degradation in each component. Although assuming the range of changes in some health factors or the relationships between them when a specific mode of gas path performance degradation occurs, or pre-identifying components that have experienced gas path performance degradation, can reduce the impact of "the number of measurable parameters in the gas path being less than the number of component health factors" on gas path performance diagnosis, the effect is not ideal, resulting in the accuracy of diagnosing the gas path performance degradation of each component still being low. Summary of the Invention

[0004] This application provides a gas turbine gas path performance diagnosis method, which solves the technical problem of low accuracy in diagnosing the performance degradation of various components in the prior art, and achieves the technical effect of improving the accuracy of diagnosing the performance degradation of various components in the gas path.

[0005] This application provides a method for diagnosing the gas path performance of a gas turbine, the method comprising:

[0006] Under preset environment and preset operating conditions, obtain the actual measurable parameters of the gas path of the gas turbine to be diagnosed, as well as the standard measurable parameters of the gas path of the gas turbine to be diagnosed when there is no gas path performance degradation.

[0007] Based on the actual measurable parameters of the gas path and the standard measurable parameters of the gas path, determine the relative deviation of the measurable parameters of the gas path of the gas turbine to be diagnosed;

[0008] Determine the similarity value between the relative deviation of the gas path measurable parameters and the standard gas path measurable parameter deviation set of each of the N preset gas path performance degradation modes. Based on the order of all similarity values ​​from largest to smallest, obtain an ordered array consisting of the N preset gas path performance degradation modes, where N is a positive integer.

[0009] The target preset gas path performance degradation mode is obtained from the ordered array, and the undetermined health factor group of the target preset gas path performance degradation mode is taken as the target health factor group. The target preset gas path performance degradation mode refers to the preset gas path performance degradation mode in which multiple undetermined health factors in the undetermined health factor group meet the rationality judgment conditions. The multiple undetermined health factors in the target health factor group are used to characterize the performance degradation degree of the corresponding component in the gas turbine to be diagnosed.

[0010] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0011] The technical solution provided in this application determines the relative deviation of the measurable parameters of the gas path of the gas turbine to be diagnosed based on the actual measurable parameters of the gas path and the standard measurable parameters of the gas path. Based on the similarity value between the relative deviation of the measurable parameters of the gas path and the standard measurable parameter deviation set of each of the N preset gas path performance degradation modes, the N preset gas path performance degradation modes are sorted. The target preset gas path performance degradation mode in the group of undetermined health factors that all meet the reasonableness judgment conditions is determined. The target preset gas path performance degradation mode is taken as the degradation mode of the gas turbine to be diagnosed, and the corresponding undetermined health factor is taken as the target health factor.

[0012] This demonstrates that this embodiment did not perform extensive "trial" diagnoses across all gas path performance degradation modes, nor did it add more control equations based on experience. Instead, this embodiment relies entirely on the detected measurable parameters of the actual gas path, selecting a small number of highly similar modes from N gas path performance degradation modes. This reduces computational load and improves diagnostic speed to some extent. Furthermore, this embodiment sequentially performs quantitative calculations and rationality checks on the corresponding undetermined health factors under the highly similar gas path performance degradation modes. From a small number of pre-ordered gas path performance degradation modes, it determines the target gas path performance degradation mode and target health factor of the gas turbine to be diagnosed, improving the accuracy of pattern recognition and the diagnostic precision of the undetermined health factor. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 A flowchart of a gas turbine gas path performance diagnosis method provided in this application;

[0015] Figure 2 This is a schematic diagram of the gas circuit structure of a gas turbine;

[0016] Figure 3 A flowchart of the simulation model for the gas turbine gas path performance;

[0017] Figure 4a and Figure 4b A comparison chart showing the identification of gas path performance degradation using BP neural networks and extreme learning machines;

[0018] Figure 5 This is a flowchart of a gas turbine gas path performance diagnostic model.

[0019] Figure 6 A flowchart of another gas turbine gas path performance diagnosis method provided in this application;

[0020] Figure 7 A comparison chart showing the false recognition rates of gas path performance degradation mode identification using existing technologies and the gas path performance diagnosis method for a gas turbine provided in this application;

[0021] Figure 8 A comparison chart showing the misidentification rates of gas path performance degradation pattern recognition using existing technologies and a gas path performance diagnosis method for gas turbines provided in this application, both when the hyperparameter settings of the pattern recognition tool are unreasonable and reasonable.

[0022] Figure 9 This is a comparison chart showing the misidentification rates of gas path performance degradation pattern recognition using existing technologies and a gas path performance diagnosis method for a gas turbine provided in this application when the noise level of the test sample is higher than that of the training sample.

[0023] Figure 10a , Figure 10b , Figure 11a , Figure 11b This is a comparison chart of health factors obtained under four conditions: "implantation value when generating test samples", "calculated value under underdetermined conditions of diagnostic equation set", "calculated value based on the identification result of erroneous gas path performance degradation mode" and "calculated value obtained by the method provided in this embodiment". Detailed Implementation

[0024] This application provides a gas turbine gas path performance diagnosis method, which solves the technical problem of low accuracy in diagnosing the performance degradation of various components in the prior art.

[0025] To better understand the technical solutions provided in this application, the above technical solutions will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0026] First, it should be clarified that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0027] This embodiment provides, for example Figure 1 The method shown is a gas path performance diagnosis method for a gas turbine, the method comprising:

[0028] Step S11: Under preset environment and preset operating conditions, obtain the actual measurable parameters of the gas path of the gas turbine to be diagnosed, as well as the standard measurable parameters of the gas path of the gas turbine to be diagnosed when there is no gas path performance degradation.

[0029] In this embodiment, "measurable parameters of the gas path" refers to parameters that can be directly measured by gas path sensors on the gas turbine, such as gas path temperature, gas path pressure, fuel flow rate, and output power.

[0030] Actual measurable parameters of the gas path refer to the parameters directly obtained from the gas path sensors on the gas turbine under diagnosis. In other words, actual measurable parameters of the gas path are obtained when the gas turbine under diagnosis is at a certain level of gas path performance degradation.

[0031] Standard gas path measurable parameters refer to the parameters of the gas turbine under diagnosis when there is no gas path performance degradation. In other words, standard gas path measurable parameters refer to the parameters when all components of the gas turbine are in a healthy state.

[0032] Standard gas path measurement parameters are typically generated from a gas turbine gas path performance simulation model. Specifically, a gas turbine gas path performance simulation model is constructed based on the gas path structure of the gas turbine to be diagnosed. Based on the gas turbine gas path performance simulation model, standard measurable gas path parameters are obtained for the gas turbine to be diagnosed under preset environment and preset operating conditions, without gas path performance degradation.

[0033] If the preset environment, preset operating conditions, and the health status of various components of the gas turbine change, the actual measurable parameters of the gas path will also change. In order to ensure that the actual measurable parameters of the gas path correspond one-to-one with the standard measurable parameters of the gas path, when obtaining the standard measurable parameters of the gas path through the simulation model, the set values ​​of the environmental parameters and operating condition parameters in the simulation model also need to be consistent with the preset environment and preset operating conditions.

[0034] The preset environment includes atmospheric temperature, atmospheric pressure, and atmospheric relative humidity. The preset operating conditions include the rotor speed of the gas generator in the gas turbine, or a measurable parameter of the gas path that can characterize the operating load of the gas turbine, such as the mass flow rate of fuel.

[0035] A gas turbine gas path performance simulation model is used to simulate the operating conditions of a gas turbine. The simulation outputs are the gas turbine's operating data, i.e., measurable gas path parameters. Additionally, the gas path performance parameters of each component under its operating conditions can also be obtained. Figure 2 The structure of the gas turbine is described as follows:

[0036] like Figure 2 The diagram shows the gas path structure of a dual-rotor gas turbine, including a high-pressure compressor, combustion chamber, high-pressure turbine, and power turbine. Section 0 represents the atmosphere, section 1 is the high-pressure compressor inlet, section 2 is the high-pressure compressor outlet (which is also the combustion chamber air inlet), section 3 is the combustion chamber fuel inlet, section 4 is the high-pressure turbine inlet (which is also the combustion chamber outlet), section 5 is the high-pressure turbine outlet (which is also the power turbine inlet), and section 6 is the power turbine outlet. Section 7 represents the high-pressure rotor of the gas generator, and section 8 represents the power turbine rotor.

[0037] Figure 2 The working process of the gas turbine shown is as follows: Atmosphere is purified by an air filter, and the purified air is continuously drawn into a high-pressure compressor. The high-pressure compressor compresses the purified air to obtain pressure energy. The compressed gas enters the combustion chamber from the outlet of the high-pressure compressor, mixes with fuel, and is fully combusted, forming high-temperature and high-pressure gas at the outlet of the combustion chamber. Subsequently, the gas enters the high-pressure turbine, converting part of its heat energy into the rotational kinetic energy of the high-pressure turbine, i.e., mechanical work. The mechanical work output by the high-pressure turbine is consumed by the high-pressure compressor on the same rotor to maintain the stable operation of the gas generator (the gas generator is a device consisting of a high-pressure compressor, a combustion chamber, and a high-pressure turbine). Although the temperature and pressure of the gas flowing out of the high-pressure turbine have decreased, it still has a strong ability to do work. After flowing into the power turbine, part of the residual heat energy it carries is converted into mechanical work by the power turbine and consumed in the process of driving the load.

[0038] Gas turbine simulation actually involves simulating the operating data of various components of the gas turbine. Figure 2 The gas turbine with the shown gas path structure has a modular simulation model consisting of one compressor simulation module, one combustion chamber simulation module, and two turbine simulation modules. The gas path structure and mechanical connections between the modules are identical to those of a real gas turbine. After assembling the simulation modules of each component into a complete gas turbine gas path performance simulation model, and further combining the operating characteristic lines of each component, the model can simulate and output the measurable gas path parameters of the gas turbine under any environmental parameters, any operating conditions, and any healthy state. It can also provide the corresponding gas path performance parameters (which will be explained later).

[0039] Combination Figure 3 This paper explains the process of obtaining measurable parameters of the gas path from the gas path performance simulation model of the gas turbine (the measurable parameters of the gas path here can be the gas path measurable parameters of the gas turbine in any environment, under any operating condition, and under any gas path performance degradation mode and degree of each component).

[0040] First of all, Figure 3 The meanings of the various parameters appearing in the text are explained below:

[0041] T, p, m and These are the gas's temperature, pressure, mass flow rate, and composition (vectors), respectively, and RH is the atmospheric relative humidity. This represents the gas equivalent mass flow rate obtained by querying the component characteristic line. The gas equivalent mass flow rate is calculated based on the principle of mass balance between components; π, ε, and η are the pressure ratio of the (high-pressure) compressor, the expansion ratio of the high-pressure turbine (or power turbine), and the isentropic efficiency (or combustion efficiency), respectively; P is the power output or power consumption of the component. To represent the rotor speed, β is the auxiliary coordinate of the (high-pressure) compressor characteristic curve.

[0042] The subscripts HPC, CC, HPT, PT, and Fuel represent the high-pressure compressor, combustion chamber, high-pressure turbine, power turbine, and fuel, respectively. The subscripts In and Out represent the inlet and outlet sections of the components, respectively.

[0043] HF stands for Health Factor, and its subscripts IE, CMF, and CE represent isentropic efficiency, reduced mass flow rate, and combustion efficiency, respectively. It represents the residual vector composed of the residuals of each equilibrium equation during the k-th simulation iteration; Iter represents the iteration number. and These represent the maximum permissible residual and the maximum permissible number of iterations during the simulation iteration process, respectively.

[0044] Figure 3 Parameters marked "Prediction Parameters" are independent variable parameters; parameters marked "Specified Parameters" are parameters that remain unchanged during the simulation model's operation. Specified parameters are user-specified parameters, such as the health factors of various component simulation modules (health factors will be explained later), or more specifically, the converted mass flow rate health factor HF of the high-pressure compressor simulation module. HPC,CMF .

[0045] Secondly, for Figure 3 The workflow is explained as follows:

[0046] The first step is to set the values ​​of various parameters to obtain the gas path performance parameters and measurable parameters.

[0047] For the high-pressure compressor module: in T HPC,In p HPC,In Under RH import conditions, the specified health factor for the high-pressure compressor module is HF. HPC,CMF and HF HPC,IE The independent variable parameter of the high-pressure compressor module includes N. GG and β. According to N GG By querying the high-pressure compressor characteristic line using β, the standard value for the corresponding specified operating condition can be obtained. π HPC η HPC (standard π HPC η HPC This refers to the value of the high-pressure compressor in a healthy state under the specified operating conditions mentioned above. Furthermore, based on the mass flow balance principle between the inlet and outlet sections of the high-pressure compressor module, and... π HPC and η HPC HF HPC,CMF and HF HPC,IE The high-pressure compressor module can be used to calculate... Figure 3 The measurable parameter T of the gas path shown below the words "High Pressure Compressor Module" is... HPC,Out p HPC,Out And can obtain and And the gas path performance parameters shown to the right of the words "High Pressure Compressor Module" π HPC and P HPC .

[0048] For the combustion chamber module: Specify the health factor of the combustion chamber simulation module as HF. CC,CE And specify the combustion efficiency of the combustion chamber as η. HPC The independent variable parameters of the combustion chamber module include Under the high-pressure compressor outlet conditions given by the high-pressure compressor module (i.e., T) HPC,Out p HPC,Out , These are the gas temperature, gas pressure, gas mass flow rate, and gas component ratio (vector) at the high-pressure compressor outlet, respectively. Based on the mass flow rate balance principle at each inlet and outlet section of the combustion chamber module, and... and η CC HF CC,CE This can be obtained through the combustion chamber module. Figure 3 The measurable parameter T of the gas path shown below the words "combustion chamber module" in the text. CC,Out p CC,Out And can obtain and In addition, the combustion chamber module will also output the measurable parameters of the gas path shown to the right of the words "combustion chamber module".

[0049] For the high-pressure turbine module: Specify the health factor of the high-pressure turbine module as HF. HPT,CMF and HF HPT,IE The independent variable parameters of the high-pressure turbine module include N. GG and ε HPT According to N GG and ε HPT By consulting the high-pressure turbine characteristic curve, the standard ε under a specified operating condition can be obtained. HPT η HPT and (Standard ε) HPT η HPT and This refers to the value of the high-pressure turbine in a healthy state under the specified operating conditions mentioned above. Therefore, under the combustion chamber outlet conditions given by the combustion chamber module (i.e., T...), CC,Out p CC,Out , These are the gas temperature, gas pressure, gas mass flow rate, and gas component ratio (vector) at the combustion chamber outlet. Based on the mass flow rate balance principle at each inlet and outlet section of the high-pressure turbine module, and ε... HPT η HPT and HF HPT,CMF and HF HPT,IE It can be calculated using the high-pressure turbine module. Figure 3 The measurable parameter T in the gas path shown below the words "High Pressure Turbine Module" is... HPT,Out p HPT,Out And can obtain and And the gas path performance parameters shown to the right of the words "High Pressure Turbine Module" ε HPT and P HPT .

[0050] For the power turbine module: Specify the health factor of the power turbine module as HF. PT,CMF and HF PT,IE The independent variable parameters of the power turbine module include N. PT and ε PT According to N PT and ε PT By consulting the power turbine characteristic line, the standard ε under a specified operating condition can be obtained. PT η PT and (Standard ε) PT η PT and This refers to the value of the power turbine in a healthy state under the specified operating conditions mentioned above. Therefore, under the high-pressure turbine outlet conditions given by the high-pressure turbine module (i.e., T...), HPT,Out p HPT,Out , These are the gas temperature, gas pressure, gas mass flow rate, and gas component ratio (vector) at the high-pressure turbine outlet, respectively. Based on the mass flow rate balance principle at each inlet and outlet section of the power turbine module and ε... PT η PT and HF PT,CMF and HF PT,IE You can get Figure 3 The measurable air path parameter T shown below the words "Power Turbine Module" in the text is as follows: PT,Out p PT,Out And can obtain and And the airflow performance parameters shown to the right of the words "Power Turbine Module" ε PT and P PT .

[0051] The second step is to verify the gas path performance parameters obtained in the first step.

[0052] Calculate the residuals of the power balance equation, pressure ratio (expansion ratio) balance equation, equivalent mass flow balance equation, and operating condition characteristic quantity balance equation of the gas generator. For the specific residual formulas, see equations (1)-(5). Adjust the independent variable parameter terms in the first step according to the residuals, and repeat the first and second steps until the residuals are obtained. If the residual value is below the preset residual threshold or the number of times the first and second steps are executed exceeds the preset number, the third step is executed.

[0053]

[0054]

[0055]

[0056]

[0057]

[0058] Equation (1) is the power balance equation of the gas generator; Equation (2) is the pressure ratio balance equation of the gas turbine; Equation (3) is the equivalent mass flow balance equation of the high-pressure turbine inlet; Equation (4) is the equivalent mass flow balance equation of the power turbine inlet; Equation (5) is the balance equation of the characteristic parameters of the gas turbine operating conditions.

[0059] In the formula, η is the residual of the power balance equation of the gas generator; mech The mechanical efficiency of the gas generator; The residual of the pressure ratio balance equation for the entire gas turbine; α NP The negative pressure coefficient of the exhaust pipe of the power turbine is α. NP =p PT,Out / p0, where p0 is atmospheric pressure; These are the residuals of the reduced mass flow balance equations for the high-pressure turbine and the power turbine, respectively. The residuals of the equilibrium equations representing the operating conditions; x sim and x targ These are the simulated and target values ​​of the gas turbine's operating conditions, respectively. Generally, the rotor speed of the gas generator is chosen as the operating condition characterization of the gas turbine. For the k-th iteration result, the residuals of the equilibrium equations in equations (1)-(5) are... The vector formed.

[0060] The third step is to end the simulation process and output the results. The gas path measurable parameters and gas path performance parameters are calculated in the last iteration with a residual value less than the preset residual threshold, and the obtained gas path measurable parameters are used as standard gas path measurable parameters.

[0061] After obtaining the measurable parameters of the standard gas path based on the simulation model, proceed to step S12.

[0062] Step S12: Determine the relative deviation of the gas path measurable parameters of the gas turbine to be diagnosed based on the actual gas path measurable parameters and the standard gas path measurable parameters.

[0063] By acquiring the actual and standard measurable parameters of the gas path under preset environment and preset operating conditions, and comparing the actual and standard measurable parameters, the relative deviation of the gas path measurable parameters of the gas turbine to be diagnosed can be obtained.

[0064] The relative deviation of measurable parameters in the gas path cannot directly reflect the degree of performance degradation of individual components. The purpose of obtaining the relative deviation of measurable parameters in the gas path is to determine the performance degradation mode of the gas turbine to be diagnosed.

[0065] To determine the degree of performance degradation in the gas path of each component, the actual measurable parameters of the gas path must be input into the gas path performance diagnostic model for quantitative calculation of each health factor. However, there is a problem that "the number of measurable parameters is less than the number of health factors." In this case, the gas path performance diagnostic model cannot provide accurate calculation results. Therefore, this embodiment qualitatively determines which component has experienced gas path performance degradation by using the relative deviation of the measurable parameters. Then, when calling the gas path performance diagnostic model, only the health factors of those components "identified as having gas path performance degradation" are quantitatively calculated, thus ensuring that the number of "undetermined health factors" is less than the number of measurable parameters.

[0066] Step S13: Determine the similarity value between the relative deviation of the gas path measurable parameters and the set of standard gas path measurable parameter deviations for each of the N preset gas path performance degradation modes. Based on the order of all similarity values ​​from largest to smallest, obtain an ordered array consisting of the N preset gas path performance degradation modes, where N is a positive integer.

[0067] The preset gas flow performance degradation mode refers to the gas flow performance degradation mode corresponding to a specific combination of health factors. To explain the health factors and gas flow performance degradation modes, the following explanation is provided:

[0068] As described above, measurable gas path parameters of a gas turbine under preset environments and operating conditions can be directly obtained using gas path sensors installed on the gas turbine. Based on the preset environment, preset operating conditions, and measurable gas path parameters, gas path performance parameters can be determined. These performance parameters reflect the performance status of the gas turbine components or the entire turbine. Gas path performance parameters may include the equivalent mass flow rate, equivalent rotational speed, pressure ratio (expansion ratio), isentropic efficiency, and overall thermal efficiency of each component.

[0069] The health factor is the ratio of gas path performance parameters under deterioration to those under healthy conditions. Each component of a gas turbine has a corresponding health factor to characterize the degree of gas path performance degradation. For example, for compressors (including high-pressure compressors) and turbines (including high-pressure turbines and power turbines), the health factor consists of two parts: the reduced mass flow rate factor and the isentropic efficiency factor.

[0070]

[0071] η x,Deg =HF x,IE ·η x,Health (7)

[0072] In the formula, HF x,CMF HF represents the reduced mass flow rate factor of a compressor or turbine. x,IE The isentropic efficiency factor represents the compressor or turbine. and These represent the reduced mass flow rates of the compressor or turbine when it is in a state of degraded gas path performance and a healthy state, respectively. Similarly, η x,Deg and η x,Health These represent the isentropic efficiency of the compressor or turbine when it is in a state of degraded gas path performance and a state of good performance, respectively.

[0073] Furthermore, it is generally assumed that when the health status of a compressor changes, the change in pressure ratio in its operating characteristics is consistent with the change in equivalent mass flow rate, i.e., the equivalent mass flow rate factor HF. Comp,CMF and pressure ratio factor HF Comp,PR They are numerically equal. Therefore, the compressor pressure ratio factor is not considered an independent health factor. Therefore, the change in the compressor's pressure ratio characteristic when the reduced mass flow rate changes can be given by equation (8):

[0074] π Comp,Deg =HF Comp,CMF ·(π Comp,Health -1)+1 (8)

[0075] In the formula, π Comp,Deg and π Comp,Health These represent the pressure ratios when the compressor is in a state of deteriorated gas path performance and when it is in a healthy state, respectively.

[0076] For the combustion chamber, there are two health factors: combustion efficiency factor and pressure recovery coefficient factor. Their application methods are shown in equations (9) and (10):

[0077] η CC,Deg =HF CC,Eff ·η CC,Health (9)

[0078] δ CC,Deg =HF CC,Rec ·δ CC,Health (10)

[0079] In the formula, HF CC,Eff The combustion efficiency factor representing the combustion chamber; HF CC,Rec η represents the pressure recovery coefficient factor of the combustion chamber. CC,Deg and η CC,Health These represent the combustion efficiency when the combustion chamber is in a state of degraded airflow performance and a healthy state, respectively. Similarly, δ CC,Deg and δ CC,Health These represent the pressure recovery coefficients when the combustion chamber is in a state of degraded fuel system performance and a healthy state, respectively. It should be noted that the pressure recovery coefficient factor only changes when the combustion chamber undergoes torsional deformation. Therefore, HF CC,Rec It is usually considered to be a constant of 1.

[0080] After understanding the health factors of each component, we will now explain the gas path performance degradation modes. Each gas path performance degradation mode represents a specific set of components that have experienced gas path performance degradation, and therefore also includes a corresponding set of changed health factors.

[0081] Returning to step S13, N preset gas path performance degradation modes are pre-defined. For each preset gas path performance degradation mode, a corresponding set of standard gas path measurable parameter deviations can be found. The relative deviations of the gas path measurable parameters obtained in step S12 are compared one by one with the set of standard gas path measurable parameter deviations of the N preset gas path performance degradation modes to obtain N similarity values. The N preset gas path performance degradation modes are sorted in descending order of the N similarity values ​​to obtain an ordered array composed of the N preset gas path performance degradation modes.

[0082] Step S13 can be achieved using a pattern recognition tool (which may be a neural network, extreme learning machine, grey relational analysis, etc.), as detailed below:

[0083] Step S21: Train the pattern recognition tool to be trained to obtain a standard pattern recognition tool.

[0084] The training process for the pattern recognition tool is as follows:

[0085] Step S31: Use the gas turbine simulation model to generate multiple measurable parameters of the first gas path of the gas turbine to be diagnosed under standard environment, different operating conditions and different gas path performance degradation modes.

[0086] Step S32: Use the gas turbine simulation model to generate multiple samples of measurable parameters of the second gas path of the gas turbine to be diagnosed under standard environment and different operating conditions, without gas path performance degradation.

[0087] Step S33: Based on the relative deviation between the first gas path measurable parameter sample and the second gas path measurable parameter sample under each of the multiple operating conditions, obtain multiple gas path measurable parameter relative deviation training samples.

[0088] Step S34: Based on the correspondence between the training samples of relative deviations of multiple measurable parameters of the gas path and different gas path performance degradation modes, train the training pattern recognition tool to obtain the standard pattern recognition tool.

[0089] Step S22: Using a standard pattern recognition tool, determine the similarity value between the relative deviation of the gas path measurable parameters and the set of standard gas path measurable parameter deviations for each of the N preset gas path performance degradation modes. Based on the order of all similarity values ​​from largest to smallest, obtain an ordered array composed of the N preset gas path performance degradation modes.

[0090] Traditionally, a higher similarity value indicates a greater similarity between the gas path performance degradation mode of the gas turbine under diagnosis and the mode corresponding to that similarity value. Related technologies often rely solely on pattern recognition tools, directly using the highest-scoring mode as the actual gas path performance degradation mode of the gas turbine. However, because sets of relative deviations of measurable gas path parameters belonging to different gas path performance degradation modes overlap in vector space, the gas path performance degradation mode determined by the pattern recognition tool may not be the actual gas path performance degradation mode of the gas turbine under diagnosis. Therefore, a higher similarity value only indicates a greater probability that the gas path performance degradation mode of the gas turbine under diagnosis is the mode corresponding to that similarity value; it still leaves the possibility that the gas path performance degradation mode of the gas turbine under diagnosis is not the mode with the highest similarity value. To improve accuracy, after obtaining the ordered array, step S14 is performed to filter a preset number of preset gas path performance degradation modes in the ordered array, determining the final target health factor and its corresponding target preset gas path performance degradation mode.

[0091] Specifically, an example illustrates the shortcomings of using pattern recognition tools to identify gas path performance degradation patterns in related technologies. Taking BP neural networks and extreme learning machines, common in pattern recognition tools, as examples, both have multiple output layer neurons when used for pattern recognition. The relative deviation of the measurable parameters of the gas path of the gas turbine to be diagnosed belongs to the gas path performance degradation pattern corresponding to the neuron with the highest score.

[0092] With a certain Figure 2 Taking a dual-rotor gas turbine with the shown gas path structure (e.g., the GE LM2500+ gas turbine) as an example, Figure 4 (including...) Figure 4a and Figure 4b This paper demonstrates the error recognition results obtained when using a BP neural network and an extreme learning machine to perform pattern recognition on the relative deviation of two different gas path measurable parameters of this type of gas turbine. The specific gas path performance degradation components corresponding to the gas path performance degradation mode numbers in Figure 4 are as follows: (1) HPC; (2) CC; (3) HPT; (4) PT; (5) HPC+CC; (6) HPC+HPT; (7) HPC+PT; (8) CC+HPT; (9) CC+PT; (10) HPT+PT; (11) HPC+CC+HPT; (12) HPC+CC+PT; (13) HPC+HPT+PT; (14) CC+HPT+PT.

[0093] As shown in Figure 4, this type of gas turbine has 14 gas path performance degradation modes. Therefore, both mode recognition tools have 14 output layer neurons. When simulating the gas turbine gas path performance using the gas turbine gas path performance simulation model, the input mode is the 11th gas path performance degradation mode. See Figure 4a In the BP neural network identification results, the relative deviations of the measurable parameters of the airway corresponding to the sample to be diagnosed showed a high degree of similarity to the 13th and 11th airway performance degradation patterns, respectively. In the extreme learning machine identification results, the relative deviations of the measurable parameters of the airway corresponding to the sample to be diagnosed showed a high degree of similarity to the 13th, 11th, 6th, and 10th airway performance degradation patterns, respectively. However, as mentioned above, the degradation category simulated in this study was the 11th category, but based on the "maximum score" principle, both tools incorrectly identified it as the 13th category, which had the highest similarity. Similarly, see... Figure 4b Both the BP neural network and the extreme learning machine incorrectly identified the gas path performance degradation mode, which should have belonged to category 14, as category 11.

[0094] It is worth noting that although both tools gave incorrect identification results, the relative deviation of the gas path measurable parameters of the sample to be diagnosed showed a high degree of similarity to the actual gas path performance degradation mode. Therefore, in this embodiment, the pattern recognition tool is more suitable as a similarity discrimination tool. A BP neural network or extreme learning machine is used to determine the similarity value between the relative deviation of the gas path measurable parameters of the sample to be diagnosed and the standard gas path measurable parameter deviation set of each of the N preset gas path performance degradation modes. According to the order of all similarity values ​​from largest to smallest, an ordered array composed of N preset gas path performance degradation modes is obtained. After determining the ordered array, step S14 is continued to verify which gas path performance degradation mode in the ordered array the actual gas path performance degradation mode of the gas turbine to be diagnosed is.

[0095] Furthermore, as can be seen from the working principle of the Extreme Learning Machine (ELM), it does not involve iterative updates of hidden layer weights and biases, thus exhibiting extremely fast training speed. Moreover, as shown in Figure 4, compared to the Backpropagation Neural Network (BPNN), the scores output by the neurons in the ELM output layer more comprehensively reflect the similarity between the sample to be diagnosed and various gas path performance degradation patterns. Therefore, preferably, in the embodiments of the present invention, the ELM is selected as the pattern recognition tool.

[0096] Step S14: Obtain the target preset gas path performance degradation mode from the ordered array, and take the undetermined health factor group of the target preset gas path performance degradation mode as the target health factor group. The target preset gas path performance degradation mode refers to the preset gas path performance degradation mode in which multiple undetermined health factors in the undetermined health factor group meet the rationality judgment conditions. The multiple undetermined health factors in the target health factor group are used to characterize the degree of gas path performance degradation of the corresponding component in the gas turbine to be diagnosed.

[0097] Select a preset gas path performance degradation mode from the ordered array where multiple undetermined health factors in the undetermined health factor group all meet the rationality judgment conditions, and use it as the target preset gas path performance degradation mode. Use the corresponding undetermined health factor group as the target health factor group of the gas turbine to be diagnosed. The health factors in the target health factor group quantitatively characterize the degree of gas path performance degradation of the gas turbine to be diagnosed.

[0098] Retrieve the target preset gas path performance degradation mode from the ordered array, and use the undetermined health factors of the target preset gas path performance degradation mode as the target health factors, including:

[0099] Step S41: Sequentially determine whether multiple undetermined health factors in the undetermined health factor group of the preset gas path performance degradation mode in the ordered array all meet the rationality judgment conditions.

[0100] Step S42: If it is determined that multiple undetermined health factors in the undetermined health factor group of a certain preset gas path performance degradation mode meet the rationality judgment conditions, then the preset gas path performance degradation mode corresponding to the undetermined health factor group that meets the rationality judgment conditions is taken as the target preset gas path performance degradation mode, and the undetermined health factor group of the target preset gas path performance degradation mode is taken as the target health factor group.

[0101] The calculation method for each undetermined health factor in the undetermined health factor group of each preset gas path performance degradation mode includes:

[0102] Step S51: Based on the relative deviation of the measurable parameters of the gas path, determine the target component in the gas turbine to be diagnosed that has gas path performance degradation;

[0103] Step S52: Determine the undetermined health factors of the target component;

[0104] Step S53: Construct a group of undetermined health factors based on all undetermined health factors of all target components in the gas turbine to be diagnosed.

[0105] For example, for compressors and turbines, the health factors are two: the reduced mass flow rate factor and the isentropic efficiency factor. For combustion chambers, there are also two health factors: the combustion efficiency factor and the pressure recovery coefficient factor. However, the pressure recovery coefficient factor only changes when the combustion chamber undergoes torsional deformation, which is usually not the case. Therefore, the pressure recovery coefficient factor of the combustion chamber can be considered a constant of 1, meaning it is not used for quantitative calculation as a pending health factor. In summary, the only pending health factor for the combustion chamber is typically the combustion efficiency factor.

[0106] Therefore, this embodiment can first screen out components that have experienced gas path performance degradation and exclude components that have not experienced gas path performance degradation, thereby reducing the number of undetermined health factors. This can alleviate the ambiguity effect to a certain extent, and in optimal conditions, it can even prevent the ambiguity effect from occurring.

[0107] If the current preset gas path performance degradation mode is not the gas path performance degradation mode of the gas turbine to be diagnosed, then the next preset gas path performance degradation mode will be verified until the obtained pending health factor meets the reasonableness judgment conditions. Then the pending health factor will be used as the target health factor and the corresponding preset gas path performance degradation mode will be used as the target preset gas path performance degradation mode.

[0108] Preferably, for the gas turbine to be diagnosed, when the first judgment is made on whether the undetermined health factors of the preset gas path performance degradation mode in the ordered array meet the reasonableness judgment conditions, for the preset gas path performance degradation mode with the largest similarity value in the ordered array, the judgment is made on whether multiple undetermined health factors in the undetermined health factor group of the preset gas path performance degradation mode in the ordered array all meet the reasonableness judgment conditions. That is, in step S14, the preset gas path performance degradation modes in the ordered array are sorted from largest to smallest according to the similarity value. The first preset gas path performance degradation mode in the ordered array is verified to determine whether the first preset gas path performance degradation mode is the gas path performance degradation mode of the gas turbine to be diagnosed. If not, the second preset gas path performance degradation mode in the ordered array is verified, and so on, to determine the gas path performance degradation mode of the gas turbine to be diagnosed.

[0109] However, typically, the gas path performance degradation mode of the gas turbine to be diagnosed exists in the first few preset gas path performance degradation modes of an ordered array (e.g., the first three preset gas path performance degradation modes). Therefore, in order to save computational resources and improve computational efficiency, the following technical means are provided:

[0110] After obtaining an ordered array consisting of N preset gas path performance degradation modes, the method includes:

[0111] Step S61: Select a preset number of candidate preset gas path performance degradation modes from the ordered array to form an ordered subarray. The candidate preset gas path performance degradation modes refer to the preset number of preset gas path performance degradation modes selected from the preset gas path performance degradation mode with the largest similarity value after sorting N preset gas path performance degradation modes in descending order of similarity value.

[0112] Step S62: Obtain the target preset gas path performance degradation mode from the ordered array, and use the undetermined health factors of the target preset gas path performance degradation mode as the target health factors, including:

[0113] Obtain the target preset gas path performance degradation mode from the ordered subarray, and use the undetermined health factor of the target preset gas path performance degradation mode as the target health factor.

[0114] Specifically, a preset number of similarity values ​​are selected from an ordered array to determine the top few preset gas path performance degradation modes. These preset gas path performance degradation modes form an ordered subarray. The preset gas path performance degradation modes in this subarray are then verified to determine if they correspond to the degradation modes of the gas turbine to be diagnosed. If a preset gas path performance degradation mode is determined to be the target preset gas path performance degradation mode, the target health factor group for the gas turbine to be diagnosed can be identified. Typically, the preset number is 3.

[0115] In this embodiment, step S14 is implemented using a gas path performance diagnostic model, that is, the model calculates the undetermined health factors for a preset number of preset gas path performance degradation modes. The working principle of the gas path performance diagnostic model is based on the working principle of the gas turbine gas path performance simulation model. The specific working principle of the gas path performance diagnostic model is as follows:

[0116] With Figure 2 Taking the dual-rotor gas turbine with the shown gas path structure as an example, Figure 5 A workflow for a gas path performance diagnostic model is presented. (Comparison) Figure 3 It can be seen that the gas path performance diagnostic model is based on the gas path performance simulation model, and further adds "gas path diagnostic iteration" for calculating the health factor values ​​of each component. Therefore, the complete workflow of the gas path performance diagnostic model can be divided into two parts: "performance simulation iteration (inner layer iteration)" and "gas path diagnostic iteration (outer layer iteration)".

[0117] [Performance Simulation Iteration (Inner Layer Iteration)]

[0118] Performance simulation iteration (inner layer iteration) refers to the simulation process of the gas turbine gas path performance simulation model. This involves inputting environmental parameters, specified parameters, operating condition characterization parameters, and a second predicted parameter into the gas turbine gas path performance simulation model. By adjusting the first predicted parameter, the residual of the performance simulation iteration is adjusted. If the values ​​are less than a preset threshold, the measurable parameters and performance parameters of the gas path under the corresponding conditions are obtained.

[0119] [Gas Pathway Diagnostic Iteration (Outer Layer Iteration)]

[0120] The measurable parameters of the gas path obtained from the simulation model are compared with the actual measurable parameters of the gas path measured by sensors on the gas turbine to be diagnosed. If the difference between the two exceeds a preset threshold, it means that the measurable parameters of the gas path obtained from the simulation model do not match the actual parameters, and the simulation needs to be repeated. That is, the second prediction parameter is adjusted according to the difference between the two, the inner iteration is repeated, the outer iteration is repeated, and so on, until the difference between the two is less than the preset threshold. The second prediction parameter at this time is then used as the target health factor.

[0121] The difference between the two can be the residual, as shown in equation (11).

[0122]

[0123] In the formula, HF j MP represents the j-th health factor, and there are n health factors in total; i This represents the i-th gas path measurement parameter, and there are a total of m gas path measurement parameters. It is the residual of the i-th governing equation.

[0124] Equation 11, while representing the difference between the two, also serves as the control equation for each term in the iterative process of gas path diagnosis. There are a total of m control equations, which constitute the diagnostic equation set.

[0125] Obviously, when the gas path diagnosis iteration (outer iteration) process ends, the simulated values ​​of the gas path measurable parameters at each cross section output by the performance simulation iteration (inner iteration) process are consistent with the actual values ​​of the gas path measurable parameters of the gas turbine to be diagnosed (i.e., the difference is less than the preset threshold).

[0126] exist Figure 5 middle, This represents the residual vector composed of the residuals of each governing equation during the k-th diagnostic iteration, and it has... Iter represents the number of iterations. and These represent the maximum permissible residual and the maximum permissible number of iterations during the diagnostic iteration process, respectively.

[0127] Typically, the number of measurable parameters in the gas path is less than the number of undetermined health factors, thus creating a fuzzy effect. This embodiment can first eliminate components that have not experienced gas path performance degradation, thereby reducing the number of undetermined health factors and improving the fuzzy effect. In optimal conditions, it can even prevent the fuzzy effect altogether. To illustrate that the technology provided in this embodiment can improve the accuracy of health factors and reduce the computational load of the entire diagnostic process while avoiding the "fuzzy effect," the "fuzzy effect" will first be explained, then the solutions provided in related technologies to address the "fuzzy effect" will be briefly described, and finally, the principle by which this embodiment solves the "fuzzy effect" and improves the calculation accuracy of health factors will be explained.

[0128] Fuzzy Effect

[0129] In the field of gas path diagnostics, it is well known that when using a gas path performance diagnostic model to quantitatively calculate the health factors of each component, the number of control equations in the diagnostic equation set (i.e., the diagnostic equation set expressed by Equation 11, which has m control equations, with n health factors) must be greater than or equal to the number of health factors. In other words, the diagnostic equation set must be either over-determined or well-determined during the gas path diagnostic iteration process. Otherwise, the diagnostic results will exhibit a significant "fuzzy effect." The "fuzzy effect" refers to the fact that regardless of whether each component is in a healthy state, the calculated health factors using the gas path performance diagnostic model will not equal 1; that is, all components will be diagnosed as having gas path performance degradation. Therefore, when the number of measurable parameters in the gas path is less than the number of health factors, the obtained diagnostic results are unreliable.

[0130] According to equation (11), the number of measurable parameters in the gas path is equal to the number of control equations in the diagnostic equation set. Therefore, if the number of measurable parameters in the gas path is less than the number of health factors, the diagnostic equation set is underdetermined and a unique solution cannot be obtained. Under this condition, the health factor values ​​obtained through iterative calculation of gas path diagnosis do not match the performance degradation of various components of the gas turbine, which is specifically manifested as the "fuzzy effect".

[0131] [Solutions to address fuzziness effects using relevant technologies]

[0132] In practice, the number of measurable parameters in a gas turbine's gas path is usually less than the number of component health factors. When it's impossible to add new gas path sensors, there are generally two methods to make the diagnostic equations fit or over-stressed to eliminate the "fuzzy effect": Method one is to increase the constraints during gas path diagnosis based on the experience of technicians, i.e., increase the number of control equations in the diagnostic equation set; Method two is to select a subset of all health factors as undetermined health factors to reduce the number of undetermined variables (i.e., undetermined health factors) in the diagnostic equation set.

[0133] [Method 1]

[0134] The constraints in Method 1 refer to the range of values ​​for various health factors and the proportional relationship between them when the compressor or turbine experiences a certain mode of gas path performance degradation. The range of values ​​and the proportional relationship in the constraints are generally set based on experience. Tables 1 and 2 show two empirically based proportional relationships, respectively.

[0135] Based on Method 1, when eliminating the "fuzzy effect" in the diagnostic results, it is necessary to first determine the gas path performance degradation mode of the component, and then add corresponding control equations according to the ratios of various health factor variables given in Table 1 or Table 2 (or other empirical basis), so as to achieve over-determination or proper determination of the diagnostic equation set during the gas path diagnostic iteration process.

[0136] Table 1

[0137]

[0138] Table 2

[0139]

[0140] For example, when it is first determined that fouling has occurred in the compressor, an additional governing equation is added to the workflow of gas path diagnosis iteration (outer iteration), as shown in Equation 12:

[0141]

[0142] It can be seen from the comparison between Table 1 and Table 2 that in this field, there is no uniform standard for the ratio of each health factor corresponding to the fault characteristics of gas turbines, that is, when a certain mode of gas path performance degradation occurs in a gas turbine; moreover, for gas turbines with different aerodynamic designs, different gas path structures and used in different application scenarios, this ratio of variation amounts of each health factor set based on experience does not have wide applicability.

[0143] [Method 2]

[0144] It is assumed that among all p components of the gas turbine to be diagnosed, there are at most q (q<p) components with gas path performance degradation. All k health factors corresponding to these q components are undetermined health factors. For the remaining p-q components, they are considered to have no gas path performance degradation, and the corresponding n-k health factors (assuming the total number of health factors of p components is n) are set as a constant 1. Further, if the condition k≤m (m is the number of measurable gas path parameters) is satisfied, the diagnostic equations have a unique solution. Therefore, before calculating the value of each undetermined health factor using the gas path performance diagnosis model, it is necessary to first screen out the components with gas path performance degradation.

[0145] Method 2 involves the following multiple situations, which are now illustrated by specific examples.

[0146] The first example of Method 2 is as follows: first, from n health factors corresponding to all components of the gas turbine to be diagnosed, m undetermined health factors that need to be determined by solving the diagnostic equations are selected, and the remaining n-m health factors are set to a fixed value 1, and there are a total of C combinations of health factors that meet this requirement k m . Then, using the actually measured gas path measurable parameters, the gas path performance diagnosis model is used to respectively analyze the above-mentioned C kThe process of solving for m combinations of undetermined health factors can be viewed as a "pre-diagnosis." Based on this, the mean and standard deviation of each health factor are calculated across all "pre-diagnosis" results. Furthermore, by introducing the concept of a "diagnostic index," components exhibiting gas path performance degradation are screened. Finally, the gas path performance diagnostic model is used again to perform a "final diagnosis" only on components identified as having gas path performance degradation.

[0147] It should be noted that in the first embodiment of Method Two: ① Each combination of undetermined health factors represents a gas path performance degradation mode; ② The "diagnostic index" is equal to the absolute value of the mean of a certain undetermined health factor divided by its standard deviation; ③ Components corresponding to health factors with high "diagnostic indices" will be determined to have gas path performance degradation; ④ During the "final diagnosis" process, the health factors of components not determined to have gas path performance degradation will all be set to a constant of 1.

[0148] To more clearly illustrate the implementation process of the above embodiments, the following will combine... Figure 2 The gas turbine with the gas path structure shown will be further explained.

[0149] A certain having Figure 2 The dual-rotor gas turbine with the shown gas path structure (e.g., the GE LM2500+ gas turbine) has four gas path components: high-pressure compressor, combustion chamber, high-pressure turbine, and power turbine. There are eight corresponding health factors, including: high-pressure compressor equivalent mass flow rate factor HF. HPC,CMF High-pressure compressor isentropic efficiency factor HF HPC,IE Combustion efficiency factor HF CC,Eff Combustion chamber pressure recovery coefficient factor HF CC,Rec High-pressure turbine equivalent mass flow factor HF HPT,CMF High-pressure turbine isentropic efficiency factor HF HPT,IE HF, the equivalent mass flow factor of the power turbine PT,CMF And the isentropic efficiency factor HF of the power turbine PT,IE Among these eight health factors, the combustion chamber pressure recovery coefficient factor HF... CC,Rec They usually do not change. Therefore, the remaining seven health factors are generally considered to be the focus of gastrointestinal performance diagnosis.

[0150] Meanwhile, in this type of gas turbine, there are a total of 10 parameters that can be measured by the gas path sensors, namely:

[0151] (1) Parameters used to characterize the environmental conditions of the gas turbine, namely the high-pressure compressor inlet temperature T HPC,In High-pressure compressor inlet pressure p HPC,InAnd atmospheric relative humidity (RH), a total of 3 items;

[0152] (2) Parameters used to characterize the operating conditions of the gas turbine, namely the gas generator rotor speed N GG (i.e., the rotor speed of the high-pressure compressor and the high-pressure turbine), totaling 1 item;

[0153] (3) Measurable parameters of the gas path, namely the high-pressure compressor outlet temperature T HPC,Out High-pressure compressor outlet pressure p HPC,Out Fuel gas mass flow rate High-pressure turbine outlet temperature T HPT,Out High-pressure turbine outlet pressure p HPT,Out and power turbine outlet temperature T PT,Out There are a total of 6 items.

[0154] In summary, when performing gas path diagnostics on this type of gas turbine, there are seven health factors that require close attention, but only six of them are measurable gas path parameters that can be used to calculate the values ​​of these health factors. Therefore, to ensure that the diagnostic equations have a unique solution, it is necessary to assume that some components do not experience gas path performance degradation; that is, the health factors of these components will be set to a constant of 1.

[0155] Preferably, under the condition that the diagnostic equation set has a unique solution, it should be assumed that a larger number of components experience gas path performance degradation; that is, more health factors should be set as undetermined health factors. This allows for a more comprehensive and accurate diagnosis of the overall health status of the gas turbine. Referring to the steps in the embodiment: First, in the diagnostic equation set, six undetermined health factors are set, thus resulting in seven possible gas path performance degradation modes with unique solutions (k=7, m=6, ...). Then, using the obtained measurable parameters of the gas turbine gas path, the values ​​of each undetermined health factor were calculated under the seven gas path performance degradation modes mentioned above. Based on this, the mean value (HF) of each undetermined health factor was calculated in all diagnostic results. Average,i and standard deviation σ Average,i By introducing the "diagnostic index" DI (DI = |HF) Average,i | / σ Average,i The concept of a diagnostic model is used to achieve a "pre-diagnosis" of the health status of all components. During this process, components with high DI (Digital Indicator) values ​​are identified as having gas path performance degradation. Finally, the diagnostic model is used again to perform a "final diagnosis" only on the components identified as having gas path performance degradation. The results represent the health status of each component, and the diagnosis ends. In the "final diagnosis" process, the health factor corresponding to components not identified as having gas path performance degradation is set to a constant of 1.

[0156] Based on the examples above using Method 2, it can be seen that the quantitative relationship between health factors and measurable parameters of the airway determines the number of "pre-diagnoses".

[0157] The second example of Method Two involves first collecting multiple sets of samples from a gas turbine under the same health condition. Next, the gas path performance diagnostic model is invoked, treating all health factors as undetermined health factors. Under the condition that the diagnostic equations are underdetermined, each of the aforementioned samples is diagnosed. Then, the mean and standard deviation of each health factor in all "pre-diagnosis" results are calculated, and the concept of a "diagnostic index" (explained previously and not repeated here) is introduced to screen components exhibiting gas path performance degradation. Finally, the gas path performance diagnostic model is used again to perform a "final diagnosis" only on components determined to have gas path performance degradation.

[0158] Compared to the two embodiments of Method 2, the similarity lies in that the "pre-diagnosis" process, that is, the process of screening components with deteriorating gas path performance, requires a large number of quantitative diagnoses using the gas path performance diagnostic model, thus failing to meet the requirements of real-time diagnosis.

[0159] Compared to the two embodiments of Method 2 described above, the differences are as follows: ① In the second embodiment, the "pre-diagnosis" process targets multiple groups of samples to be diagnosed; while in the first embodiment, the "pre-diagnosis" process targets only one group of samples to be diagnosed. ② In the second embodiment, all health factors of all components are undetermined health factors, that is, the diagnostic equation set during "pre-diagnosis" is underdetermined; while in the first embodiment, only some health factors are treated as undetermined health factors, and the diagnostic equation set is determined.

[0160] The third embodiment of Method Two involves the following steps: First, using a gas path performance simulation model, training samples of relative deviations of multiple measurable gas path parameters are generated for the gas turbine under the following conditions: in a healthy state and under different gas path performance degradation modes, both in a standard environment and under different operating conditions. Second, based on the correspondence between the training samples of relative deviations of multiple measurable gas path parameters and the corresponding multiple different gas path performance degradation modes, a training pattern recognition tool is developed to obtain a standard pattern recognition tool. Third, using the actual measurable gas path parameters of the gas turbine under the following conditions collected in a preset environment and under preset operating conditions, and the standard measurable gas path parameters of the gas turbine under the following conditions generated by the gas turbine gas path performance simulation model in the same preset environment and under the same preset operating conditions, the relative deviations of the measurable gas path parameters of the gas turbine under the following conditions are calculated. Finally, the relative deviations of the measurable gas path parameters of the gas turbine under the following conditions are input into the standard pattern recognition tool, which then determines which gas path performance degradation mode the gas turbine under the following conditions is in. Next, based on the gas path performance degradation mode of the gas turbine under diagnosis determined by the standard identification tool, the health factors corresponding to the components whose gas path performance degradation has been determined are taken as potential health factors. Finally, the actual measurable gas path parameters of the gas turbine under diagnosis, collected under preset environment and preset operating conditions, are input into the gas path performance diagnosis model to calculate the values ​​of the potential health factors, thus completing the gas path performance diagnosis.

[0161] In other words, to address the ambiguity effect, related technologies offer solutions such as increasing the number of control equations based on the diagnostic experience of technicians, and conducting numerous quantitative diagnoses (i.e., "pre-diagnostics") using gas path performance diagnostic models to screen components exhibiting gas path performance degradation, thereby reducing the number of undetermined health factors. Alternatively, pattern recognition tools can be used alone to identify degradation patterns in the gas turbine to be diagnosed, further reducing the number of undetermined health factors. However, in these technologies, increasing the number of control equations requires technicians' understanding of the mechanisms affecting different gas path performance degradation modes and a large accumulation of data. Diagnostic experience, on the other hand, is a subjective summary by staff, containing uncertainties and lacking transferability, which can lead to low accuracy in the calculated health factors. Furthermore, conducting numerous quantitative diagnoses using gas path performance diagnostic models to screen components exhibiting gas path performance degradation cannot guarantee real-time diagnostics. If a pattern recognition tool is used alone to identify the gas path performance degradation patterns of a gas turbine to be diagnosed, the recognition results will be affected by the recognition capability of the pattern recognition tool. If the relevant parameters of the pattern recognition tool are not set properly, or if the training samples used to train the pattern recognition tool cannot accurately reflect the noise and other factors that the gas turbine to be diagnosed is subjected to in actual operation, or if the training samples corresponding to several different gas path performance degradation patterns overlap in the vector space, the recognition effect will decrease.

[0162] [The principle behind this embodiment of solving the "ambiguity effect" and improving the accuracy of health factors]

[0163] The technical solution provided in this embodiment determines the relative deviation of the measurable parameters of the gas path of the gas turbine to be diagnosed based on the actual measurable parameters and the standard measurable parameters of the gas path. Based on the similarity between the relative deviation of the gas path measurable parameters and the set of standard measurable parameter deviations for N preset gas path performance degradation modes, the N preset gas path performance degradation modes are sorted in descending order of similarity. Starting with the preset gas path performance degradation mode with the highest similarity, a preset number of preset gas path performance degradation modes are selected. Then, the preset gas path performance degradation mode that meets the reasonableness judgment criteria is taken as the target preset gas path performance degradation mode, and the target preset gas path performance degradation mode is taken as the gas path performance degradation mode of the gas turbine to be diagnosed. The corresponding undetermined health factor group is taken as the target health factor group. It can be seen that this embodiment does not attempt to diagnose all combinations of health factors, nor does it add more control equations based on experience. As can be seen, this embodiment is entirely based on the detected actual measurable parameters of the gas path. From N gas path performance degradation modes, a small number of highly similar modes are selected, which reduces the amount of calculation and improves the diagnostic speed to a certain extent. Furthermore, this embodiment sequentially performs quantitative calculations and rationality checks on the corresponding undetermined health factors under the highly similar gas path performance degradation modes. From a small number of sorted preset gas path performance degradation modes, the target gas path performance degradation mode and target health factor of the gas turbine to be diagnosed are determined, which improves the accuracy of pattern recognition and the diagnostic precision of the undetermined health factors.

[0164] When performing step S14, it is determined whether the health factor to be determined meets the rationality judgment conditions (the rationality judgment conditions include condition 1 and condition 2), including steps S71 and S72, wherein step S71 corresponds to condition 1 and step S72 corresponds to condition 2.

[0165] Step S71: Input the actual measurable parameters of the gas path into the gas turbine gas path performance diagnostic model. By adjusting the values ​​of multiple undetermined health factor parameters in multiple undetermined health factor groups in the gas turbine gas path performance diagnostic model, the residual between the predicted measurable parameters of the gas path output by the gas turbine gas path performance diagnostic model and the actual measurable parameters of the gas path is lower than a preset threshold. In other words, the value of the iterative residual after each diagnostic iteration gradually decreases and stabilizes is lower than the preset threshold. Generally speaking, the lower the preset threshold is set, the better. However, in actual operation, the specific value of the preset threshold can be determined according to the accuracy requirements of the project. For example, the preset threshold can be 0.3%.

[0166] Step S72: Determine whether the change characteristics of each undetermined health factor in the undetermined health factor group meet the rationality judgment conditions.

[0167] Steps S71 and S72 involve two conditions, meaning that the reasonableness judgment condition consists of two conditions.

[0168] The iterative residual in step S71 can be expressed by equation (15):

[0169]

[0170] In the formula, These represent the various measurable parameters of the actual gas path of the gas turbine to be diagnosed; These represent the measurable parameters of the gas path of the gas turbine to be diagnosed, predicted and output by the diagnostic model. This represents the residual vector.

[0171] exist Figure 5 middle, This represents the iterative residual generated during the k-th diagnostic iteration, i.e. 2-norm.

[0172] It should be noted that in practical applications, the measurable parameters of the gas turbine's gas path contain noise. Therefore, the calculated values ​​of the undetermined health factors based on the measurable gas path parameters may have slight deviations from the actual health status of each component. To avoid this deviation from adversely affecting the judgment of the reasonableness of the calculated results of the undetermined health factors, preferably, when the diagnosed change in the component's health factor falls within a certain set threshold (δ... Health When the value is within the range of δ, the health factor is considered unchanged, i.e., its change characteristic is considered to be "0". Preferably, δ Health It can be set to 0.004 (that is, the threshold is 0.4% when a certain health factor is considered to have changed).

[0173] Regarding step S72, the judgment criterion is whether the change characteristics of the undetermined health factor in the obtained calculation results are the same as those when a typical gas path fault or its combination occurs. For example, as shown in Table 3, when the high-pressure compressor blades accumulate fouling, the reduced mass flow rate and isentropic efficiency of the high-pressure compressor decrease, therefore both its reduced flow rate factor and isentropic efficiency factor decrease, and their change characteristics are represented by "-1". When all components are fault-free, the reduced flow rate and isentropic efficiency of each component hardly decrease, so the corresponding change characteristics of the health factor are represented by "0". When the high-pressure turbine blades wear, the reduced flow rate of the high-pressure turbine increases, so its reduced mass flow rate factor change characteristics are represented by "+1"; at the same time, the isentropic efficiency of the high-pressure turbine decreases, so its isentropic efficiency factor change characteristics are represented by "-1". In Table 3, a change characteristic of "0" represents that the value of the health factor hardly changes; a change characteristic of "+1" represents that the value of the health factor is higher than 1; and a change characteristic of "-1" represents that the value of the health factor is lower than 1.

[0174] Table 3

[0175]

[0176] In summary, this embodiment proposes a gas turbine gas path performance diagnosis method, which can be implemented based on a gas turbine gas path performance simulation model, a gas turbine gas path performance diagnosis model, and a pattern recognition tool. The gas turbine gas path performance simulation model can simulate the measurable gas path parameters of the gas turbine under healthy conditions and typical gas path performance degradation modes. The gas turbine gas path performance diagnosis model is used to quantitatively calculate the relative deviations of the performance parameters of each component, i.e., the values ​​of the health factors of each component. The pattern recognition tool is used to determine the similarity between the deviations of the measurable gas path parameters of the gas turbine to be diagnosed and the standard sets of deviations of the measurable gas path parameters for various preset gas path performance degradation modes, and to rank all preset gas path performance degradation modes according to the similarity, thereby determining the gas path performance degradation mode of the gas turbine to be diagnosed.

[0177] In other words, this embodiment first determines the relative deviation of the measurable parameters of the gas path of the gas turbine to be diagnosed. Based on the similarity between the relative deviation of the measurable parameters of the gas path and the set of standard measurable parameter deviations corresponding to various preset gas path performance degradation modes, the various preset gas path performance degradation modes are sorted. Second, from the sorted various preset gas path performance degradation modes, the degradation mode in which each of the pending health factors in the pending health factor group meets the reasonableness judgment condition is selected as the target preset gas path performance degradation mode, and this pending health factor group is used as the target health factor group of the gas turbine to be diagnosed, so as to characterize the degree of degradation of each component of the gas turbine to be diagnosed.

[0178] This embodiment improves the accuracy of identifying performance degradation modes in gas turbines and the calculation precision of target health factors, making it effectively applicable to real-time gas path diagnosis of gas turbines in the field where measurement noise exists. Furthermore, this embodiment addresses the problem of inaccurate diagnostic results due to the limited number of measurable parameters in the gas path; it reduces the computational load of the diagnostic process while improving diagnostic accuracy.

[0179] This embodiment improves the accuracy of identifying gas path performance degradation patterns; it can suppress the problem of a significant drop in recognition accuracy caused by unreasonable parameter settings in pattern recognition tools; it can suppress the negative impact of measurement noise on recognition accuracy; it avoids the problem of inaccurate calculation results of undetermined health factors caused by gas path performance diagnosis of gas turbines under incorrect gas path performance degradation pattern classification; and it avoids the problem of increased computational load and diagnosis time caused by quantitative diagnosis of all undetermined health factor combinations or a large number of samples to be diagnosed during the screening of components with gas path performance degradation.

[0180] Combination Figure 6 The technical solution provided in this embodiment will now be comprehensively explained with a specific example.

[0181] Train the pattern recognition tool. Input all performance degradation modes and the corresponding relative deviations of measurable gas path parameters into the pattern recognition tool for training.

[0182] Based on the correspondence between the training samples of relative deviations of multiple measurable parameters of the gas path and different gas path performance degradation modes, the training pattern recognition tool is trained to obtain a standard pattern recognition tool.

[0183] Obtain the actual measurable parameters of the gas path of the gas turbine to be diagnosed. Based on the environmental parameters and operating conditions corresponding to the actual measurable parameters of the gas path, output the standard measurable parameters of the gas path of the gas turbine under healthy conditions through the gas path performance simulation model.

[0184] The relative deviation of the gas path measurable parameters is obtained based on the actual gas path measurable parameters and the standard gas path measurable parameters. The relative deviation of the gas path measurable parameters is input into the extreme learning machine to calculate the similarity between the relative deviation of the gas path measurable parameters and all gas path performance degradation modes. Based on the score corresponding to the similarity, the gas path performance degradation mode to which the diagnostic sample belongs is ranked according to similarity.

[0185] The i-th gas path performance degradation mode is determined from the sorted gas path performance degradation modes, and the corresponding undetermined health factor is determined. The actual measurable parameters of the gas path of the gas turbine to be diagnosed are input into the gas path performance diagnosis model. The determined undetermined health factor is quantitatively calculated, and it is determined whether the undetermined health factor calculated by the gas path performance diagnosis model meets the rationality judgment condition. It is then determined whether the i-th gas path performance degradation mode is the gas path performance degradation mode of the gas turbine to be diagnosed.

[0186] If the pending health factor meets the rationality judgment condition and the i-th gas path performance degradation mode is determined to be the gas path performance degradation mode of the gas turbine to be diagnosed, then the pending health factor is taken as the target health factor.

[0187] If the pending health factors do not meet the reasonableness judgment criteria, the gas path performance degradation mode of the gas turbine to be diagnosed is re-determined from the remaining gas path performance degradation modes. If the number of repeated judgments exceeds the maximum number of attempts, the diagnosis is stopped and a diagnosis failure is reported.

[0188] based on Figure 6 The diagnostic method implementation flow shown in this embodiment preferably sets the maximum number of attempts to 3, meaning that a maximum of 3 trial calculations are allowed before the target health factor is obtained. Therefore, this avoids the increased computational load and diagnostic time caused by performing trial calculations with all possible combinations of undetermined health factors, or by screening for components with deteriorating gas path performance through numerous "pre-diagnosis" processes. Moreover, under this setting, the gas path performance diagnostic method for gas turbines provided in this embodiment still has a significantly higher pattern recognition accuracy than BP neural networks, extreme learning machines, and support vector machines.

[0189] The progress of the technical solution provided in this embodiment will now be illustrated with some specific comparative data.

[0190] [Comparison 1]

[0191] like Figure 7 As shown, the false recognition rates are displayed when using BP neural network (BPNN), extreme learning machine (ELM), support vector machine (SVM) and the method provided in this embodiment to identify the gas path performance degradation mode of a gas turbine.

[0192] like Figure 7 As shown, the method provided in this embodiment demonstrates better performance in identifying gas path performance degradation modes for single-component, dual-component, and triple-component systems. Specifically, the false recognition rate for single-component gas path performance degradation modes is approximately 5.5%; the false recognition rate for dual-component gas path performance degradation modes is approximately 1.9%; and the false recognition rate for triple-component gas path performance degradation modes is only 0.64%.

[0193] It should be further clarified that the misidentification results in almost all single-component and dual-component gas path performance degradation modes are "overall" misidentifications. That is, among all components considered to have gas path performance degradation, not only are all components that actually have gas path performance degradation included, but also components that do not actually have gas path performance degradation. For example, if a high-pressure compressor experiences a certain mode of gas path performance degradation, the components identified as having gas path performance degradation are the high-pressure compressor and the high-pressure turbine.

[0194] The "overlapping" misidentification can lead to an increase in the number of undetermined health factors when using the gas path diagnostic model for subsequent quantitative calculations. However, these undetermined health factors originating from components without gas path performance degradation will have calculated values ​​very close to 1. This indicates that although the gas path performance degradation pattern is incorrectly identified, subsequent quantitative calculations of each undetermined health factor using the gas path performance diagnostic model can eliminate the adverse effects of the "overlapping" misidentification on the final diagnostic results.

[0195] Another type of misidentification is the "omission type," which means that among all components considered to have deteriorated gas path performance, not all components actually exhibiting such deterioration are included. Clearly, the misidentification results for the gas path performance degradation modes of all three components fall under the "omission type." For example, the high-pressure compressor, combustion chamber, and high-pressure turbine may have experienced gas path performance degradation, but the identified deteriorated components are the high-pressure turbine and the power turbine. The occurrence of "omission type" misidentification results will cause the calculated results of all pending health factors to be inconsistent with the actual health status of the corresponding components.

[0196] In summary, the false recognition rate of the three-component gas path performance degradation mode directly determines the quality of the gas path performance diagnosis result. The method provided in this embodiment has a false recognition rate of only 0.55%, which is far lower than the 4% of other methods. This result fully demonstrates that the method provided in this embodiment improves the accuracy of gas path performance degradation mode recognition.

[0197] [Comparison 2]

[0198] like Figure 8 As shown, Figure 8 Taking the performance degradation mode of the three-component gas path as an example, this paper demonstrates the impact of unreasonable parameter (hyperparameter) settings in the pattern recognition tools (Extreme Learning Machine and Support Vector Machine) on the pattern recognition results.

[0199] like Figure 8As shown, with reasonable hyperparameters, the false recognition rate of Extreme Learning Machine and Support Vector Machine is about 4%; with unreasonable hyperparameters, the false recognition rate of both will rise to about 5.5%. However, for the method provided in this embodiment, even with the above-mentioned unreasonable hyperparameter settings, its false recognition rate is only about 0.8%.

[0200] This result fully demonstrates that the method provided in this embodiment can suppress the problem of a significant drop in recognition accuracy caused by unreasonable parameter settings in pattern recognition tools.

[0201] It should be noted that in the Extreme Learning Machine (ELM) and the ELM used in this embodiment, the reasonable hyperparameters are: penalty factor C = 1024, and Gaussian kernel kernel parameter σ = 0.0156. In the Support Vector Machine (SVM), the reasonable hyperparameters are: penalty factor C = 512, and Gaussian kernel kernel parameter σ = 2. In the ELM, SVM, and the ELM used in this embodiment, an unreasonable set of hyperparameters is: penalty factor C = 2, and Gaussian kernel kernel parameter σ = 0.5.

[0202] [Comparison 3]

[0203] like Figure 9 As shown, Figure 9 Taking the three-component gas path performance degradation mode as an example, the impact of measurement noise on the recognition accuracy of different pattern recognition tools is demonstrated.

[0204] like Figure 9 As shown, when the training samples and the samples to be diagnosed (test samples) contain the same level of random measurement noise, the false recognition rate of the three pattern recognition tools—BP neural network, extreme learning machine, and support vector machine—is approximately 4%. If the level of random measurement noise in the diagnostic samples (test samples) is increased to twice the original level, the false recognition rate of the above three pattern recognition tools increases to approximately 6%. Under the same conditions, the false recognition rate of the method provided in this embodiment only slightly increases from 1% to 1.6%.

[0205] This result fully demonstrates that the method provided in this embodiment can suppress the negative impact of measurement noise on recognition accuracy. The greater application value of this beneficial effect lies in the fact that when the pattern recognition tool (preferably an Extreme Learning Machine in this embodiment) is trained using idealized simulation data and applied to the recognition of gas path performance degradation patterns in on-site gas turbines, the recognition accuracy will not significantly decrease due to the presence of more unknown interference in the measurable parameters of the gas path of the on-site gas turbine.

[0206] It should be noted that random measurement noise is unavoidable. In order to obtain more accurate gas path measurable parameters, during the process of simulating the generation of training samples and collecting samples to be diagnosed from the gas turbine to be diagnosed, a multi-sampling-point averaging method can be used to reduce the noise of the gas path measurable parameters. In this embodiment, both the training samples and the samples to be diagnosed are obtained by averaging 10 sampling points, as shown in equation (16):

[0207]

[0208] In the formula, This represents the measured value of the i-th measurable parameter of the gas path at the j-th sampling point; The mean of 10 sampling points represents the measured value of the i-th gas path measurable parameter.

[0209] [Comparison 4]

[0210] As shown in Figure 10 (including) Figure 10a and Figure 10b ) and Figure 11 (including Figure 11a and Figure 11b As shown in the figure, Figure 10a and Figure 11a Taking the two-component and three-component performance degradation modes as examples, the changes of various health factors relative to the value "1" are compared and shown in the following situations: ① Implanted value when the test sample is generated by the gas path performance simulation model (hereinafter referred to as "implanted value when the test sample is generated" in the figure); ② Calculated value obtained by directly using the gas path performance diagnostic model (diagnostic equation set underdetermined) (hereinafter referred to as "calculated value under the condition of underdetermined diagnostic equation set" in the figure); ③ Calculated value obtained by using the gas path performance diagnostic model based on the erroneous pattern recognition result (hereinafter referred to as "calculated value based on the erroneous gas path performance degradation pattern recognition result" in the figure); ④ Calculated value obtained by using the method provided in this embodiment. Correspondingly, Figures 10(b) and 11(b) show the absolute error between each calculated value and the implanted value. Health factors corresponding to the letter abbreviations in Figures 10 and 11: (1) HF HPC,CMF (2) HF represents the health factor of the high-pressure compressor converted mass flow rate; HPC,IE Represents the isentropic efficiency health factor of high-pressure compressors; (3) HF CC,CE Representative combustion efficiency health factor of the combustion chamber; (4) HF HPT,CMF Represents the health factor of the high-pressure turbine equivalent mass flow rate; (5) HF HPT,IE Represents the health factor of high-pressure turbine isentropic efficiency; (6) HF PT,CMF Represents the health factor of the equivalent mass flow rate of the power turbine; (7) HF PT,IE This represents the health factor of the power turbine's isentropic efficiency.

[0211] like Figure 10b and Figure 11b As shown, when the diagnostic equations of the gas path performance diagnostic model are under-determined, the root mean square error (RMSE) between the calculated and implanted values ​​of various health factors is 0.23%. Figure 10b ) and 1.09% ( Figure 11b If airway diagnosis is performed under an incorrect airway performance degradation mode, the root mean square error between the calculated and implanted values ​​of various health factors not only fails to decrease, but actually increases to 0.43%. Figure 10b ) and 1.20% ( Figure 11b In stark contrast, the method provided in this embodiment, with its more rational diagnostic process, first ensures that the identified gas path performance degradation pattern matches the actual situation, thereby significantly improving the accuracy of health factor calculation results. Quantitative comparison shows that the root mean square error (RMSE) between the implanted and calculated values ​​of each health factor is only 0.06% (…). Figure 10b ) and 0.28% Figure 11b ).

[0212] This result fully demonstrates that the gas path performance diagnosis method of this embodiment avoids the problem of unreasonable selection of pending health factors caused by gas path diagnosis under the wrong gas path performance degradation mode classification, which in turn leads to inaccurate calculation results of pending health factors.

[0213] Since the electronic device described in this embodiment is an electronic device used to implement the information processing method in the embodiments of this application, those skilled in the art can understand the specific implementation methods and various variations of the electronic device in this embodiment based on the information processing method described in the embodiments of this application. Therefore, how the electronic device implements the method in the embodiments of this application will not be described in detail here. Any electronic device used by those skilled in the art to implement the information processing method in the embodiments of this application falls within the scope of protection of this application.

[0214] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0215] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions.

[0216] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0217] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for diagnosing the gas path performance of a gas turbine, characterized in that, The method includes: Under preset environment and preset operating conditions, the actual measurable parameters of the gas path of the gas turbine to be diagnosed, as well as the standard measurable parameters of the gas path of the gas turbine to be diagnosed when there is no gas path performance degradation, are obtained. Based on the actual measurable parameters of the gas path and the standard measurable parameters of the gas path, determine the relative deviation of the measurable parameters of the gas path of the gas turbine to be diagnosed. Determine the similarity value between the relative deviation of the gas path measurable parameters and the standard gas path measurable parameter deviation set of each of the N preset gas path performance degradation modes. Based on the order of all similarity values ​​from largest to smallest, obtain an ordered array composed of the N preset gas path performance degradation modes, where N is a positive integer. The target preset gas path performance degradation mode is obtained from the ordered array, and the undetermined health factor group of the target preset gas path performance degradation mode is taken as the target health factor group. The target preset gas path performance degradation mode refers to the preset gas path performance degradation mode in which multiple undetermined health factors in the undetermined health factor group meet the rationality judgment conditions. The multiple undetermined health factors in the target health factor group are used to characterize the degree of gas path performance degradation of the corresponding component in the gas turbine to be diagnosed. Determining whether multiple undetermined health factors in the undetermined health factor group all meet the reasonableness judgment conditions includes: The actual measurable parameters of the gas path are input into the gas turbine gas path performance diagnosis model. By adjusting the values ​​of multiple undetermined health factor parameter items in the undetermined health factor group in the gas turbine gas path performance diagnosis model, the residual between the predicted gas path measurable parameters output by the gas turbine gas path performance diagnosis model and the actual gas path measurable parameters is less than a preset threshold. It is also determined whether the change characteristics of each undetermined health factor in the undetermined health factor group meet the reasonableness judgment conditions. After obtaining an ordered array consisting of the N preset gas path performance degradation modes, the method includes: A preset number of candidate preset gas path performance degradation modes are selected from the ordered array to form an ordered subarray. The candidate preset gas path performance degradation modes refer to the preset number of preset gas path performance degradation modes selected starting from the preset gas path performance degradation mode with the largest similarity value after sorting N preset gas path performance degradation modes in descending order of similarity value. Obtain the target preset gas path performance degradation mode from the ordered array, and use the undetermined health factor group of the target preset gas path performance degradation mode as the target health factor group, including: The first preset gas path performance degradation mode in the ordered subarray is verified to determine whether the first preset gas path performance degradation mode is the gas path performance degradation mode of the gas turbine to be diagnosed. If not, the second preset gas path performance degradation mode in the ordered subarray is verified, and so on, to obtain the target preset gas path performance degradation mode from the ordered subarray, and the undetermined health factor group of the target preset gas path performance degradation mode is taken as the target health factor group.

2. The method as described in claim 1, characterized in that, The step of obtaining the target preset gas path performance degradation mode from the ordered array and using the undetermined health factor group of the target preset gas path performance degradation mode as the target health factor group includes: Sequentially determine whether multiple undetermined health factors in the undetermined health factor group of the preset gas path performance degradation mode in the ordered array all meet the reasonableness judgment condition; If it is determined that multiple undetermined health factors in a group of undetermined health factors for a certain preset gas path performance degradation mode meet the rationality judgment condition, then the preset gas path performance degradation mode corresponding to the group of undetermined health factors that meets the rationality judgment condition is taken as the target preset gas path performance degradation mode, and the undetermined health factor group of the target preset gas path performance degradation mode is taken as the target health factor group.

3. The method as described in claim 2, characterized in that, When the method first executes the judgment on whether multiple undetermined health factors in the undetermined health factor group of the preset gas path performance degradation mode in the ordered array all meet the reasonableness judgment condition, the method includes: For the preset gas path performance degradation mode with the highest similarity value in the ordered array, it is determined whether multiple undetermined health factors in the undetermined health factor group of the preset gas path performance degradation mode in the ordered array all meet the reasonableness judgment condition.

4. The method as described in claim 1, characterized in that, Obtain the standard measurable parameters of the gas turbine to be diagnosed when there is no gas path performance degradation, including: Based on the gas path structure of the gas turbine to be diagnosed, a gas path performance simulation model of the gas turbine is constructed. Based on the gas turbine gas path performance simulation model, the standard measurable parameters of the gas turbine to be diagnosed are obtained under the preset environment and the preset operating conditions, and when no gas path performance degradation occurs.

5. The method as described in claim 1, characterized in that, The method involves determining the similarity value between the relative deviation of the gas path measurable parameters and the set of standard gas path measurable parameter deviations for each of the N preset gas path performance degradation modes. Based on the descending order of all similarity values, an ordered array consisting of the N preset gas path performance degradation modes is obtained, including: The pattern recognition tool to be trained is trained to obtain a standard pattern recognition tool; Using the standard pattern recognition tool, the similarity value between the relative deviation of the gas path measurable parameters and the set of standard gas path measurable parameter deviations for each of the N preset gas path performance degradation modes is determined. Based on the order of all similarity values ​​from largest to smallest, an ordered array composed of the N preset gas path performance degradation modes is obtained.

6. The method as described in claim 5, characterized in that, The process of training the pattern recognition tool to obtain a standard pattern recognition tool includes: Using a gas turbine gas path performance simulation model, multiple measurable parameters of the first gas path of the gas turbine to be diagnosed are generated under standard environment, different operating conditions and different gas path performance degradation modes. Using the gas turbine gas path performance simulation model, multiple samples of second gas path measurable parameters of the gas turbine to be diagnosed are generated under standard environment and different operating conditions, without gas path performance degradation. Based on the relative deviation between the first gas path measurable parameter sample and the second gas path measurable parameter sample under each of the various operating conditions, multiple gas path measurable parameter relative deviation training samples are obtained. Based on the correspondence between the training samples of the relative deviations of the multiple measurable parameters of the gas path and the different gas path performance degradation modes, the training mode recognition tool is trained to obtain the standard mode recognition tool.

7. The method as described in claim 1, characterized in that, Calculate each undetermined health factor in the undetermined health factor group of the preset gas path performance degradation mode in the ordered array, including: Based on the relative deviation of the measurable parameters of the gas path, the target component in the gas turbine to be diagnosed that has gas path performance degradation is determined; Determine the undetermined health factors of the target component; Based on all the stated undetermined health factors, construct the undetermined health factor group.

Images