Turbine fault detection method and device for heavy-duty gas turbine and medium

Abstract

The invention relates to a turbine fault detection method and device for a heavy-duty gas turbine and a medium. The method comprises the following steps that environment condition parameters and control condition parameters of all gas paths during operation of the heavy-duty gas turbine are obtained; inputting the environmental condition parameters and the control condition parameters into each pre-trained gas path measurable parameter prediction model to obtain a real-time theoretical health value of each gas path measurable parameter; and acquiring actual measurement values of measurable parameters of each gas path, and calculating turbine health characteristic parameters of each gas path in combination with the theoretical health values of the measurable parameters of each gas path to obtain a turbine fault detection result. Compared with the prior art, the method has the advantages that the turbine fault condition of the heavy-duty gas turbine can be accurately and quantitatively diagnosed under the complex variable working conditions, and the like.

Description

Technical Field

[0001] The invention relates to the technical field of energy and power engineering, and in particular to a method, equipment and medium for detecting turbine faults of a heavy-duty gas turbine. Background Art

[0002] Heavy-duty gas turbines have been widely used in the power generation field due to their flexible start and shutdown, high thermal efficiency and low environmental impact. With the further tightening of energy conservation, emission reduction and environmental protection requirements, the proportion of clean fuels such as natural gas in primary energy has gradually increased, and the proportion of heavy-duty gas turbines in power stations has become increasingly larger.

[0003] The turbine of a heavy-duty gas turbine is one of the important high-temperature flow-through components in a steam-gas combined cycle power plant. The turbine of a heavy-duty gas turbine is most susceptible to erosion and corrosion. Turbine erosion refers to the gradual wear of the surface material of the component due to the inhalation of pollutants such as dust, dirt, and carbon particles, which causes the surface roughness of the blades to increase, resulting in increased turbine flow capacity, reduced operating efficiency, reduced turbine expansion ratio, and increased turbine outlet temperature. Turbine corrosion refers to thermal corrosion caused by the presence of certain pollutants or molten salts in the combustion gas, which results in increased turbine flow capacity and reduced operating efficiency.

[0004] As the core power equipment for efficient energy conversion and clean utilization, heavy-duty gas turbines are a kind of thermal system with strong nonlinear coupling of various components. Changes in the environment and control conditions will cause significant changes in the internal state of the thermal system of the gas turbine. This brings great difficulties to how to diagnose and predict the performance degradation, aging, damage and failure of the components of this strong nonlinear thermal system (especially the high-temperature component of the turbine) through effective methods. So far, neither domestic and foreign manufacturers nor third-party technical service providers have developed mature diagnostic products. The operating conditions of the high-temperature flow-through components of gas turbines are complex and harsh, which are prone to various performance degradations or damages, and lead to serious failures. And due to the frequent frequency and peak regulation requirements of power plants, gas turbines need to operate more flexibly, which accelerates the loss of the design life of the high-temperature components of the unit.

[0005] However, the current users of gas turbines in steam-gas combined cycle power plants strictly follow the equivalent operating hours (EOH) required by the technical documents and relevant specifications provided by the gas turbine manufacturers to determine whether the gas turbine flow components (including compressors, combustion chambers and turbines) need minor repairs, medium repairs or major repairs. There are over-repairs and disrepairs, and the maintenance cost is extremely high. The calculation of the equivalent operating hours of the gas turbine is to determine when the gas turbine should be minor, medium and major repairs. Minor repairs focus on inspections, mainly covering the combustion chamber part, and inspecting the visible parts such as the compressor and turbine. Medium repairs focus on the inspection and repair of the hot channel part, which requires opening the combustion chamber cylinder and turbine section, inspecting and repairing the high-temperature blades, and cleaning the burner; the compressor part requires visual inspection, and endoscope inspection can be added as needed. Major overhauls focus on the inspection and repair of the entire channel, opening the upper half of the gas turbine cylinder, inspecting and repairing the burner, high-temperature blades, etc.; inspecting and cleaning the compressor. The disadvantage of the above-mentioned regular maintenance is that there is a high probability of over-maintenance, premature failure cannot be avoided, and the maintenance cost is extremely high. Summary of the invention

[0006] The purpose of the present invention is to provide a heavy-duty gas turbine turbine fault detection method, equipment and medium that can realize quantification of turbine fault detection under complex and variable operating conditions.

[0007] The purpose of the present invention can be achieved by the following technical solutions:

[0008] A method for detecting a turbine fault of a heavy-duty gas turbine comprises the following steps:

[0009] Obtain environmental condition parameters and control condition parameters of each gas path when the heavy-duty gas turbine is running;

[0010] Inputting the environmental condition parameters and control condition parameters into pre-trained prediction models of various gas path measurable parameters to obtain real-time theoretical health values ​​of various gas path measurable parameters;

[0011] The measured values ​​of the measurable parameters of each gas path are obtained, and the turbine health characteristic parameters of each gas path are calculated in combination with the theoretical health values ​​of the measurable parameters of each gas path to obtain the turbine fault detection result.

[0012] Furthermore, the environmental condition parameters include environmental pressure, environmental temperature, and relative humidity.

[0013] Furthermore, the control condition parameters include compressor inlet guide vane position feedback IGV, fuel temperature, engine speed, and fuel mass flow rate.

[0014] Furthermore, the measurable parameters of each gas circuit include compressor inlet pressure, compressor outlet pressure, compressor outlet temperature, turbine outlet pressure, turbine outlet temperature, and output power.

[0015] Furthermore, the calculation process of the turbine health characteristic parameters of each gas path includes:

[0016] Calculating a relative deviation vector based on the measured values ​​of the gas path measurable parameters and the theoretical health values ​​of the gas path measurable parameters;

[0017] The skewness and median of the relative deviation vector are calculated, and the skewness and median are used as turbine health characteristic parameters.

[0018] Furthermore, the calculation expression of the relative deviation vector is:

[0019]

[0020] In the formula, is the measured value of each gas path measurable parameter, is the theoretical health value of each gas path measurable parameter, P 1,act , P 1,pre are the measured value and theoretical healthy value of the compressor inlet pressure, t 2,act ,t 2,pre are the measured value and theoretical healthy value of the compressor outlet temperature, P 2,act , P 2,pre are the measured value and theoretical health value of the compressor outlet pressure, P 5,act , P 5,pre are the measured value and theoretical healthy value of turbine outlet pressure, t 5,act ,t 5,pre are the measured value and theoretical healthy value of turbine outlet temperature, N e,act 、N e,pre They are the measured value and theoretical healthy value of output power respectively.

[0021] Furthermore, the training steps of the prediction models of the measurable parameters of the gas path include:

[0022] Acquire historical gas path measurable parameters when the heavy-duty gas turbine is operating normally and healthily, wherein the historical gas path measurable parameters include historical environmental condition parameters, historical control condition parameters, and other historical gas path measurable parameters, wherein the other historical gas path measurable parameters include compressor inlet pressure, compressor inlet temperature, compressor outlet pressure, compressor outlet temperature, turbine exhaust pressure, turbine outlet temperature, and gas turbine power;

[0023] Construct a one-dimensional convolutional neural network;

[0024] The historical gas path measurable parameters are input into the one-dimensional convolutional neural network for training until the training is completed, and each gas path measurable parameter prediction model is obtained, wherein the historical operating state of the heavy-duty gas turbine is determined by the historical environmental condition parameters and the historical control condition parameters. During the training process, the remaining historical gas path measurable parameters are used as prediction targets, so that each gas path measurable parameter prediction model learns the mapping relationship between the historical environmental condition parameters, the historical control condition parameters and the remaining historical gas path measurable parameters.

[0025] Furthermore, the one-dimensional convolutional neural network includes an input layer, a one-dimensional convolutional layer, a one-dimensional maximum pooling layer, a fully connected layer and an output layer.

[0026] The present invention also provides an electronic device, comprising: one or more processors; a memory; and one or more programs stored in the memory, wherein the one or more programs include instructions for executing the heavy-duty gas turbine turbine fault detection method as described above.

[0027] The present invention also provides a computer-readable storage medium, comprising one or more programs for execution by one or more processors of an electronic device, wherein the one or more programs include instructions for executing the heavy-duty gas turbine turbine fault detection method as described above.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] (1) For an actually healthy heavy-duty gas turbine, its operating performance status can be uniquely determined by environmental conditions and control conditions. The present invention only needs to obtain the required environmental conditions and control conditions under complex operating conditions, and predict the theoretical health value of each gas path measurable parameter through each gas path measurable parameter prediction model, and calculate the turbine health characteristic parameters of each gas path by combining the actual measured values, thereby realizing the quantification of turbine fault detection under complex and variable operating conditions.

[0030] (2) The present invention calculates the skewness and median of the relative deviation vector through the theoretical health value and the measured value. These two statistical features are used as turbine health characteristic parameters. The turbine health characteristic parameters, as two-dimensional dimensionless data, can detect whether the turbine fails and the severity of the failure from the physical essence. On the one hand, the skewness and median two-dimensional health characteristic parameters extracted when the turbine fails can maintain good inter-class classification and intra-class aggregation on the plane with the skewness and median two-dimensional health characteristic parameters extracted when other flow-through components of the heavy-duty gas turbine fail, that is, it can effectively detect whether the turbine fails; on the other hand, the median value extracted when the turbine fails will decrease as the severity of the turbine failure increases, that is, it can effectively quantify the severity of the turbine failure.

[0031] (3) The present invention can accurately quantify the fault conditions of heavy-duty gas turbines under complex and variable operating conditions, providing a scientific basis and technical support for formulating appropriate and reasonable maintenance strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a schematic diagram of the method flow of the present invention;

[0033] Figure 2 A schematic diagram of a process for establishing a prediction model for each gas path measurable parameter of the present invention;

[0034] Figure 3 It is a flowchart of the overall implementation process of an embodiment of the present invention;

[0035] Figure 4 The atmospheric temperature change of the embodiment of the present invention;

[0036] Figure 5 The IGV opening degree change of the embodiment of the present invention;

[0037] Figure 6 The fuel mass flow rate change of the embodiment of the present invention;

[0038] Figure 7 is the diagnosis result of the embodiment of the present invention;

[0039] Figure 8 is a relative deviation vector of the measurable parameters of the air path when the performance of the intake system is degraded or damaged in the embodiment of the present invention;

[0040] Fig. 9 is a relative deviation vector of the measurable parameters of the gas path when the exhaust system performance is degraded or damaged in the embodiment of the present invention;

[0041] Fig.10 is a relative deviation vector of the measurable parameters of the gas path when the compressor performance is degraded or damaged in the embodiment of the present invention;

[0042] Fig.11 is a relative deviation vector of the measurable parameters of the gas path when the turbine performance is degraded or damaged in the embodiment of the present invention;

[0043] Fig.12 is a relative deviation vector of the measurable parameters of the gas path when the combustion chamber performance is degraded or damaged in the embodiment of the present invention;

[0044] Fig.13 These are two-dimensional dimensionless health characteristic parameters of each flow-through component diagnosed in the embodiment of the present invention. DETAILED DESCRIPTION

[0045] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0046] Example 1

[0047] This embodiment provides a method for detecting a turbine fault of a heavy-duty gas turbine. Figure 1 As shown, the method comprises the following steps:

[0048] S1. Collect historical gas path measurable parameters of the heavy-duty gas turbine to be diagnosed when it is running normally and healthily, and establish prediction models for each gas path measurable parameter through a deep neural network.

[0049] The historical gas path measurable parameters include environmental condition parameters, control condition parameters and other gas path measurable parameters, among which the gas path measurable parameters include compressor inlet pressure, compressor inlet temperature, compressor outlet pressure, compressor outlet temperature, turbine exhaust pressure, turbine outlet temperature and gas turbine power.

[0050] For an actual healthy heavy-duty gas turbine, its operating performance status can be determined solely by the environmental conditions (including the ambient pressure P 0 , ambient temperature t 0 , relative humidity RH) and control conditions (including compressor inlet guide vane position feedback IGV, fuel temperature t f , engine speed n, fuel mass flow rate G f ) is determined, and the remaining gas path measurable parameters are taken as prediction targets, so that each gas path measurable parameter prediction model learns the mapping relationship between the environmental condition parameters, the control condition parameters and the remaining gas path measurable parameters. Through the sensitivity analysis of the environmental condition parameters and the control condition parameters on the prediction accuracy of each gas path measurable parameter, the final input parameters of the deep neural network are the environmental conditions (including the ambient pressure P 0 , ambient temperature t 0 , relative humidity RH) and control conditions (including compressor inlet guide vane position feedback IGV, fuel temperature t f , engine speed n, fuel mass flow rate G f ), the output parameters of the deep neural network are the measurable parameters of each gas path (including the compressor inlet pressure P 1 , compressor outlet pressure P 2 , compressor outlet temperature t 2 , turbine outlet pressure P 5 , turbine outlet temperature t 5 , output power N e ).

[0051] The gas path measurable parameter prediction model is used to calculate the theoretical health value of each gas path measurable parameter of the actual gas turbine unit under complex variable operating conditions in real time. 1,pre ,t 2,pre ,P 2,pre ,P 5,pre ,t 5,pre ,N e,pre ],like Figure 3 shown.

[0052] Since the input of the deep neural network model is 7 parameters and the output is 6 parameters, a convolutional neural network can be used to achieve good training and prediction effects. The embodiment of the present invention adopts a one-dimensional convolutional neural network, and the structure of the deep convolutional neural network model is shown in Table 1 below.

[0053] Table 1 Deep convolutional neural network model structure

[0054]

[0055] S2. Design turbine health characteristic parameters as quantitative indicators to characterize the erosion and corrosion failure conditions of the turbine.

[0056] The measured values ​​of each gas path measurable parameter [P 1,act ,t 2,act ,P 2,act ,P 5,act ,t 5,act ,N e,act ] and theoretical health value [P 1,pre ,t 2,pre ,P 2,pre ,P 5,pre ,t 5,pre ,N e,pre ] divided by the theoretical health value to obtain the relative deviation vector:

[0057]

[0058] Then the skewness and median of the vector are calculated, and these two statistical characteristics are used as turbine health characteristic parameters.

[0059] The two-dimensional dimensionless health characteristic parameters obtained by this diagnosis can detect whether the turbine fails and how severe the fault is from the physical essence. On the one hand, the skewness and median two-dimensional health characteristic parameters extracted when the turbine fails can maintain good inter-class classification and intra-class aggregation on the plane with the skewness and median two-dimensional health characteristic parameters extracted when other flow-through components of the heavy-duty gas turbine fail, that is, it can effectively detect whether the turbine fails; on the other hand, the median value extracted when the turbine fails will become smaller as the severity of the turbine fault increases, that is, it can effectively quantify the severity of the turbine fault and provide a scientific basis for formulating appropriate and reasonable maintenance strategies.

[0060] In order to test the effectiveness of the detection method proposed in the present invention, the AE94.3AF heavy-duty gas turbine produced by Ansaldo was selected as the diagnostic object. This AE94.3A heavy-duty gas turbine has been widely used in domestic power plants. Its compressor is a 15-stage large-flow, axial-flow compressor and is equipped with 2-stage inlet adjustable guide vanes. The combustion chamber is an annular combustion chamber and is equipped with 24 low-NOx burners. The turbine is a 4-stage large-flow, axial-flow turbine. The detailed gas path measurable parameters of this heavy-duty gas turbine are shown in Table 2.

[0061] Table 2 Environmental and control parameter conditions and other measurable parameters of the gas path

[0062]

[0063] like Figure 2 As shown in the figure, firstly, the historical gas path measurable parameters of the heavy-duty gas turbine to be diagnosed are collected when it is running normally and healthily, and then the prediction model of each gas path measurable parameter is established through the deep neural network. For an actual healthy heavy-duty gas turbine, its operating performance status can be uniquely determined by the environmental conditions (including the environmental pressure P 0 , ambient temperature t 0 , relative humidity RH) and control conditions (including compressor inlet guide vane position feedback IGV, fuel temperature t f , engine speed n, fuel mass flow rate G f ) is determined, and the remaining gas path measurable parameters are taken as prediction targets, so that each gas path measurable parameter prediction model learns the mapping relationship between the environmental condition parameters, the control condition parameters and the remaining gas path measurable parameters. Through the sensitivity analysis of the environmental condition parameters and the control condition parameters on the prediction accuracy of each gas path measurable parameter, the final input parameters of the deep neural network are the environmental conditions (including the ambient pressure P 0 , ambient temperature t 0 , relative humidity RH) and control conditions (including compressor inlet guide vane position feedback IGV, fuel temperature t f , engine speed n, fuel mass flow rate G f ), the output parameters of the deep neural network are the measurable parameters of each gas path (including the compressor inlet pressure P 1 , compressor outlet pressure P 2 , compressor outlet temperature t 2 , turbine outlet pressure P 5 , turbine outlet temperature t 5 , output power N e The gas path measurable parameter prediction model is used to calculate the theoretical health value of each gas path measurable parameter of the actual gas turbine unit under complex variable operating conditions in real time [P 1,pre ,t 2,pre ,P2,pre ,P 5,pre ,t 5,pre ,N e,pre ].

[0064] The turbine health characteristic parameters are then designed as quantitative indicators to characterize the erosion and corrosion failure conditions of the turbine. 1,act ,t 2,act ,P 2,act ,P 5,act ,t 5,act ,N e,act ] and theoretical health value [P 1,pre ,t 2,pre ,P 2,pre ,P 5,pre ,t 5,pre ,N e,pre ] divided by the theoretical health value to obtain the relative deviation vector Then calculate the skewness and median of the vector, and these two statistical features are used as turbine health characteristic parameters. The two-dimensional dimensionless health characteristic parameters obtained by this diagnosis can detect whether the turbine fails and the severity of the failure from the physical essence. On the one hand, the skewness and median two-dimensional health characteristic parameters extracted when the turbine fails can maintain good inter-class classification and intra-class aggregation on the plane with the skewness and median two-dimensional health characteristic parameters extracted when other flow-through components of the heavy-duty gas turbine fail, that is, it can effectively detect whether the turbine fails; on the other hand, the median value extracted when the turbine fails will decrease as the severity of the turbine failure increases, that is, it can effectively quantify the severity of the turbine failure.

[0065] In order to test the effectiveness of the proposed method in the present invention in accurately quantifying the fault condition of a heavy-duty gas turbine under complex variable operating conditions (from the minimum opening load of the IGV after grid connection to the basic load of the IGV fully opened), the heavy-duty gas turbine unit was selected under complex variable operating conditions (such as Figures 4 to 6 Diagnostic testing is performed on samples of measurable gas path parameters when fault conditions of different severity (mild, medium, severe) occur in the turbine (as shown).

[0066] Depend on Figure 7As shown, the two-dimensional dimensionless health characteristic parameters obtained by the diagnosis can detect whether the turbine fails and the severity of the failure from the physical essence. On the one hand, the skewness and median two-dimensional health characteristic parameters extracted when the turbine fails can maintain good inter-class classification and intra-class aggregation on the plane with the skewness and median two-dimensional health characteristic parameters extracted when the heavy-duty gas turbine is operating normally and healthily, that is, it can effectively detect whether the turbine fails; on the other hand, the median value extracted when the turbine fails will become smaller as the severity of the turbine failure increases, that is, it can effectively quantify the severity of the turbine failure, and the extracted median value maintains good robustness and accuracy at different variable operating points, and is not significantly disturbed by changes in the operating conditions of the unit, which illustrates the good diagnostic accuracy and anti-interference of the method proposed in the present invention.

[0067] In addition, the anti-interference performance of the method proposed in the present invention on the influence of failures of other flow-through components of heavy-duty gas turbine units is analyzed. When failures of different severity occur in the intake system, compressor, combustion chamber, and turbine, the measured values ​​of the measurable parameters of each gas path can be obtained [P 1,act ,t 2,act ,P 2,act ,P 5,act ,t 5,act ,N e,act ] and theoretical health value [P 1,pre ,t 2,pre ,P 2,pre ,P 5,pre ,t 5,pre ,N e,pre ] relative deviation vector like Figures 8 to 12 shown.

[0068] Depend on Figures 8 to 12 As shown in the figure, when the environmental conditions and control parameter conditions are known, when the failure of different flow components of the heavy-duty gas turbine occurs, the measurable gas path parameters (i.e. [P 1 ,t 2 ,P 2 ,P 5 ,t 5 ,N e ]) show different magnitudes and different positive and negative change relationships. When the same component fails with different fault severity, the measurable gas path parameters (i.e. [P 1 ,t 2 ,P 2 ,P 5 ,t 5 ,N e]) present different sizes and the same positive and negative change relationship. Therefore, the skewness and median two-dimensional health characteristic parameters extracted when the turbine fails can maintain good inter-class classification and intra-class aggregation with the skewness and median two-dimensional health characteristic parameters extracted when other flow-through components of heavy-duty gas turbines fail on the plane, that is, it can effectively detect whether the turbine fails, such as Fig.13 As shown. Among them, Fig.13 The labels 0 to 15 represent the faults and severity of each flow component of the heavy-duty gas turbine unit, as shown in Table 3.

[0069] Table 3 Description of the failures and severity of each flow component of a heavy-duty gas turbine

[0070]

[0071] Depend on Fig.12 As shown in the figure, the two-dimensional dimensionless health characteristic parameters obtained by the diagnosis can detect whether the turbine fails and how severe the fault is from the physical essence. On the one hand, the skewness and median two-dimensional health characteristic parameters extracted when the turbine fails can maintain good inter-class classification and intra-class aggregation on the plane with the skewness and median two-dimensional health characteristic parameters extracted when other flow-through components of the heavy-duty gas turbine fail, that is, it can effectively detect whether the turbine fails; on the other hand, the median value extracted when the turbine fails will become smaller as the severity of the turbine fault deepens, that is, it can effectively quantify the severity of the turbine fault, and the extracted median value maintains good robustness and accuracy at different variable operating points, and is not significantly disturbed by changes in the unit's operating conditions.

[0072] In summary, when the environmental conditions and control parameter conditions are known, when the failure of different flow components of the heavy-duty gas turbine occurs, the measurable gas path parameters (i.e. [P 1 ,t 2 ,P 2 ,P 5 ,t 5 ,N e ]) show different magnitudes and different positive and negative change relationships. When the same component fails with different fault severity, the measurable gas path parameters (i.e. [P 1 ,t 2 ,P 2 ,P 5 ,t 5 ,N e]) present different sizes and the same positive and negative change relationship. Therefore, on the one hand, the skewness and median two-dimensional health characteristic parameters extracted when the turbine fails can maintain good inter-class classification and intra-class aggregation on the plane with the skewness and median two-dimensional health characteristic parameters extracted when other flow-through components of the heavy-duty gas turbine fail, that is, it can effectively detect whether the turbine fails; on the other hand, the median value extracted when the turbine fails will become smaller as the severity of the turbine failure increases, that is, it can effectively quantify the severity of the turbine failure, and the extracted median value maintains good robustness and accuracy at different variable operating points, and is not significantly disturbed by changes in the operating conditions of the unit. Therefore, the good turbine fault detection accuracy and anti-interference performance of the method proposed in the present invention can provide a scientific basis and technical support for the formulation of appropriate and reasonable maintenance strategies.

[0073] Example 2

[0074] This embodiment provides an electronic device, comprising: one or more processors; a memory; and one or more programs stored in the memory, wherein the one or more programs include instructions for executing the heavy-duty gas turbine turbine fault detection method as described in the above-mentioned embodiment 1.

[0075] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program code.

[0076] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes. The schemes in the embodiments of the present invention may be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.

[0077] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0078] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0079] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0080] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0081] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A heavy-duty gas turbine turbine fault detection method, characterized in that: The following steps are involved: Obtain environmental condition parameters and control condition parameters of each gas path when the heavy-duty gas turbine is running; Inputting the environmental condition parameters and control condition parameters into pre-trained prediction models of various gas path measurable parameters to obtain real-time theoretical health values ​​of various gas path measurable parameters; The measured values ​​of the measurable parameters of each gas path are obtained, and the turbine health characteristic parameters of each gas path are calculated in combination with the theoretical health values ​​of the measurable parameters of each gas path to obtain the turbine fault detection result.

2. A heavy-duty gas turbine turbine fault detection method according to claim 1, characterized in that: The environmental condition parameters include environmental pressure, environmental temperature, and relative humidity.

3. A heavy-duty gas turbine turbine fault detection method according to claim 1, characterized in that: The control condition parameters include compressor inlet guide vane position feedback IGV, fuel temperature, engine speed, and fuel mass flow rate.

4. A heavy-duty gas turbine turbine fault detection method according to claim 1, characterized in that: The measurable parameters of each gas circuit include compressor inlet pressure, compressor outlet pressure, compressor outlet temperature, turbine outlet pressure, turbine outlet temperature, and output power.

5. A heavy-duty gas turbine turbine fault detection method according to claim 1, characterized in that: The calculation process of the turbine health characteristic parameters of each gas path includes: Calculating a relative deviation vector based on the measured values ​​of the gas path measurable parameters and the theoretical health values ​​of the gas path measurable parameters; The skewness and median of the relative deviation vector are calculated, and the skewness and median are used as turbine health characteristic parameters.

6. A heavy-duty gas turbine turbine fault detection method according to claim 5, characterized in that: The calculation expression of the relative deviation vector is: In the formula, is the measured value of each gas path measurable parameter, is the theoretical health value of each gas path measurable parameter, P 1,act , P 1,pre are the measured value and theoretical healthy value of the compressor inlet pressure, t 2,act ,t 2,pre are the measured value and theoretical healthy value of the compressor outlet temperature, P 2,act , P 2,pre are the measured value and theoretical health value of the compressor outlet pressure, P 5,act , P 5,pre are the measured value and theoretical healthy value of turbine outlet pressure, t 5,act ,t 5,pre are the measured value and theoretical healthy value of turbine outlet temperature, N e,act 、N e,pre They are the measured value and theoretical healthy value of output power respectively.

7. A heavy-duty gas turbine turbine fault detection method according to claim 1, characterized in that: The training steps of each gas path measurable parameter prediction model include: Acquire historical gas path measurable parameters when the heavy-duty gas turbine is operating normally and healthily, wherein the historical gas path measurable parameters include historical environmental condition parameters, historical control condition parameters, and other historical gas path measurable parameters, wherein the other historical gas path measurable parameters include compressor inlet pressure, compressor inlet temperature, compressor outlet pressure, compressor outlet temperature, turbine exhaust pressure, turbine outlet temperature, and gas turbine power; Construct a one-dimensional convolutional neural network; The historical gas path measurable parameters are input into the one-dimensional convolutional neural network for training until the training is completed, and each gas path measurable parameter prediction model is obtained, wherein the historical operating state of the heavy-duty gas turbine is determined by the historical environmental condition parameters and the historical control condition parameters. During the training process, the remaining historical gas path measurable parameters are used as prediction targets, so that each gas path measurable parameter prediction model learns the mapping relationship between the historical environmental condition parameters, the historical control condition parameters and the remaining historical gas path measurable parameters.

8. A heavy-duty gas turbine turbine fault detection method according to claim 7, characterized in that: The one-dimensional convolutional neural network includes an input layer, a one-dimensional convolutional layer, a one-dimensional maximum pooling layer, a fully connected layer and an output layer.

9. An electronic device, characterized in that: include: one or more processors; Memory; and One or more programs stored in a memory, wherein the one or more programs include instructions for executing the heavy-duty gas turbine turbine fault detection method according to any one of claims 1-8.

10. A computer-readable storage medium, characterized in that: It comprises one or more programs for execution by one or more processors of an electronic device, wherein the one or more programs comprise instructions for executing the heavy-duty gas turbine turbine fault detection method as claimed in any one of claims 1 to 8.

Images