Turbine control valve fault diagnosis method and device, storage medium and computer equipment

Abstract

The invention relates to the technical field of fault diagnosis, and provides a turbine control valve fault diagnosis method and device, a storage medium and computer equipment, and the method comprises the steps: firstly, obtaining a historical working condition data set and a simulation fault working condition data set of a turbine control valve; constructing a training set based on the historical working condition data set and the simulation fault working condition data set; secondly, constructing a steam turbine control valve fault diagnosis model, and training the steam turbine control valve fault diagnosis model based on the training set to obtain a trained steam turbine control valve fault diagnosis model; and finally, acquiring real-time working condition data of the steam turbine control valve, and determining a fault type of the steam turbine control valve based on the trained steam turbine control valve fault diagnosis model and the real-time working condition data. According to the embodiment, fault diagnosis is carried out on the steam turbine control valve by using the trained fault diagnosis model, so that fault identification and classification of the steam turbine control valve can be realized, and the safety and reliability of a steam turbine are improved.

Description

Technical Field

[0001] The present disclosure relates to the technical field of fault diagnosis, and in particular, to a steam turbine governing valve fault diagnosis method, device, storage medium, and computer device. Background Art

[0002] During the operation of a steam turbine, the governing valve system plays a crucial role. It is responsible for regulating the flow rate and pressure of the steam flow to ensure the stability of the steam turbine under different load conditions. However, the steam turbine governing valve system is prone to failures, affecting the normal operation of the system. Traditional fault diagnosis methods mainly rely on experienced maintenance personnel to judge faults through on-site detection, manual analysis, and the use of simple monitoring means. These methods are usually inefficient and vulnerable to human factors, resulting in poor accuracy and timeliness of fault diagnosis. Especially when facing complex and non-linear system faults, traditional methods often struggle to effectively capture the deep-level information of the faults.

[0003] In related technologies, the main fault diagnosis methods for steam turbine governing valves include those based on generative adversarial networks and CCD detection, etc. The method based on generative adversarial networks can generate high-quality simulation data, effectively alleviating the problem of insufficient samples. However, this method ignores the dependence relationship between time series data, resulting in poor performance when dealing with continuous time series data. The CCD detection method mainly relies on image and signal processing technologies. Although it can perform certain analysis on faults, its real-time performance is poor, and delays are likely to occur in scenarios that require quick response, affecting the efficiency of fault diagnosis. Summary of the Invention

[0004] The embodiments of the present disclosure at least provide a steam turbine governing valve fault diagnosis method, device, storage medium, and computer device. By using a trained fault diagnosis model to diagnose the faults of the steam turbine governing valve, the fault identification and classification of the steam turbine governing valve can be realized, thereby improving the safety and reliability of the steam turbine.

[0005] The embodiments of the present disclosure provide a steam turbine governing valve fault diagnosis method, including: Obtain the historical operating condition data set and the simulated fault operating condition data set of the steam turbine governing valve; and construct a training set based on the historical operating condition data set and the simulated fault operating condition data set; Construct a steam turbine governing valve fault diagnosis model, and train the steam turbine governing valve fault diagnosis model based on the training set to obtain a trained steam turbine governing valve fault diagnosis model; Obtain the real-time operating condition data of the steam turbine governing valve, and determine the fault type of the steam turbine governing valve based on the trained steam turbine governing valve fault diagnosis model and the real-time operating condition data.

[0006] In some possible embodiments, the simulated fault condition data set is obtained through the following steps: Construct a three-dimensional geometric model of the steam turbine governing valve; Based on the three-dimensional geometric model, the turbulence model, and the steam turbine operation simulation model, simulate the operating conditions inside the steam turbine governing valve, and introduce key parameter perturbations to simulate different fault types, thereby obtaining the simulated fault condition data set; wherein, the key parameter perturbations include pressure perturbation, steam flow perturbation, and load perturbation; the fault types include jamming faults, wear and corrosion faults, leakage faults, and fracture and detachment faults; The turbulence model is expressed as: ; Wherein, represents the fluid density; represents the fluid component; represents the turbulent viscosity coefficient; represents the Prandtl number of the turbulent kinetic energy; represents the turbulent kinetic energy generation term; represents the turbulent kinetic energy generated by buoyancy; represents the turbulent expansion term; represents the turbulent kinetic energy dissipation rate; The steam turbine operation simulation model includes a particle erosion wear model, a heat transfer model, a leakage flow model, and a force balance and jamming model; The particle erosion wear model is expressed as: ; Wherein, represents the material wear amount per unit time; C represents an empirical constant; represents the particle density; represents the particle erosion velocity; n, m represent empirical exponents; represents the particle impact angle; The heat transfer model is expressed as: ; ; ; Wherein, q represents the heat transfer amount; h represents the convective heat transfer coefficient; A represents the heat transfer area; represents the fluid temperature; represents the wall temperature; represents the wall heat; k represents the fluid thermal conductivity; represents the wall normal temperature gradient; represents the radiation heat; Denoted as the Stefan-Boltzmann constant; Denoted as the emissivity of the material surface; Denoted as the ambient temperature; The leakage flow rate model is expressed as: ; Where Q is denoted as the leakage flow rate; Denoted as the flow coefficient; B is denoted as the leakage area; Denoted as the upstream pressure; Denoted as the downstream pressure; The force balance and jamming model is expressed as: ; ; Where Denoted as the net acting force; Denoted as the fluid impact force; Denoted as the frictional force; Denoted as the spring force; m is denoted as the spool mass; x is denoted as the spool displacement; t is denoted as time.

[0007] In some possible embodiments, the historical operating condition data set includes multiple historical operating condition data; constructing a training set based on the historical operating condition data set and the simulated fault operating condition data set includes: Determining parameter thresholds based on the historical operating condition data set; where the parameter thresholds include a steam flow rate threshold, a pressure change rate threshold, and a load threshold; Judging whether the parameters corresponding to the historical operating condition data in the historical operating condition data set meet the parameter thresholds, and if not, determining the historical operating condition data as historical fault data; Constructing the training set based on the historical fault data and the simulated fault operating condition data set; where the training set includes a training data set and a training sample set; the training data set includes multiple training data; the training sample set includes training sample data corresponding to each training data, and the sample label corresponding to each training sample data is the fault type corresponding to the training data.

[0008] In some possible embodiments, constructing the steam turbine valve fault diagnosis model includes: Constructing an initial steam turbine valve fault diagnosis model, where the initial steam turbine valve fault diagnosis model includes a local feature extraction module, a global feature extraction module, and a fault diagnosis module; Replacing the local feature extraction module with an LSTM module and replacing the global feature extraction module with a Transformer module; Determine the steam turbine governing valve fault diagnosis model based on the initial steam turbine governing valve fault diagnosis model and the gating network after module replacement.

[0009] In some possible embodiments, training the steam turbine governing valve fault diagnosis model based on the training set includes: For each training data, perform local feature extraction on the training data based on the LSTM module to obtain a local feature vector; perform global feature extraction on the training data based on the Transformer module to obtain a global feature vector; and, based on the gating network, determine a weighting ratio according to the data characteristics of the training data, perform weighting processing on the local feature vector and the global feature vector according to the weighting ratio, and determine the corresponding to the training data based on the weighting processing result and the fault diagnosis module; Determine the loss value between the fault diagnosis information corresponding to the training data and the sample label corresponding to the training sample data based on a preset loss function, and adjust the model parameters of the steam turbine governing valve fault diagnosis model based on the loss value; Repeat the above steps until the training result meets the preset requirements to obtain the trained steam turbine governing valve fault diagnosis model.

[0010] In some possible embodiments, before determining the fault type of the steam turbine governing valve based on the trained steam turbine governing valve fault diagnosis model and the real-time operating condition data, it includes: Judge whether the parameters corresponding to the real-time operating condition data meet the parameter thresholds; When the parameters corresponding to the real-time operating condition data meet the parameter thresholds, determine that the steam turbine governing valve has no fault; When the parameters corresponding to the real-time operating condition data do not meet the parameter thresholds, input the real-time operating condition data into the trained steam turbine governing valve fault diagnosis model.

[0011] In some possible embodiments, after determining the fault type of the steam turbine governing valve based on the trained steam turbine governing valve fault diagnosis model and the real-time operating condition data, it further includes: Determine the fault cause and repair suggestions according to a preset fault knowledge base, the fault type, and the real-time operating condition data.

[0012] An embodiment of the present disclosure provides a steam turbine governing valve fault diagnosis device, including: A data acquisition module, configured to acquire the historical operating condition data set and the simulated fault operating condition data set of the steam turbine governing valve; and construct a training set based on the historical operating condition data set and the simulated fault operating condition data set; A model training module for constructing a steam turbine governing valve fault diagnosis model and training the steam turbine governing valve fault diagnosis model based on the training set to obtain a trained steam turbine governing valve fault diagnosis model; A fault diagnosis module for obtaining real-time operating condition data of a steam turbine governing valve and determining the fault type of the steam turbine governing valve based on the trained steam turbine governing valve fault diagnosis model and the real-time operating condition data.

[0013] In some possible embodiments, the data acquisition module is further configured to: Construct a three-dimensional geometric model of the steam turbine governing valve; Simulate the operating conditions inside the steam turbine governing valve based on the three-dimensional geometric model, the turbulence model, and the steam turbine working simulation model, and introduce key parameter perturbations to simulate different fault types to obtain the simulated fault condition data set; wherein, the key parameter perturbations include pressure perturbation, steam flow perturbation, and load perturbation; the fault types include jamming fault, wear and corrosion fault, leakage fault, and fracture and detachment fault; The turbulence model is expressed as: ; Wherein, Represents the fluid density; Represents the fluid component; Represents the turbulent viscosity coefficient; Represents the Prandtl number of the turbulent kinetic energy; Represents the turbulent kinetic energy generation term; Represents the turbulent kinetic energy generated by buoyancy; Represents the turbulent expansion term; Represents the turbulent kinetic energy dissipation rate; The steam turbine working simulation model includes a particle erosion wear model, a heat transfer model, a leakage flow model, and a force balance and jamming model; The particle erosion wear model is expressed as: ; Wherein, Represents the material wear amount per unit time; C represents an empirical constant; Represents the particle density; Represents the particle erosion velocity; n, m represent empirical exponents; Represents the particle impact angle; The heat transfer model is expressed as: ; ; ; Among them, q represents the heat transfer amount; h represents the convective heat transfer coefficient; A represents the heat transfer area; represents the fluid temperature; represents the wall temperature; represents the wall heat; k represents the fluid thermal conductivity; represents the normal temperature gradient of the wall; represents the radiative heat; represents the Stefan-Boltzmann constant; represents the emissivity of the material surface; represents the ambient temperature; The leakage flow model is expressed as: ; Among them, Q represents the leakage flow; represents the flow coefficient; B represents the leakage area; represents the upstream pressure; represents the downstream pressure; The force balance and jamming model is expressed as: ; ; Among them, represents the net acting force; represents the fluid impact force; represents the frictional force; represents the spring force; m represents the spool mass; x represents the spool displacement; t represents the time.

[0014] In some possible embodiments, the historical operating condition data set includes a plurality of historical operating condition data; the data acquisition module is specifically configured to: Determine parameter thresholds based on the historical operating condition data set; among them, the parameter thresholds include a steam flow threshold, a pressure change rate threshold, and a load threshold; Judge whether the parameters corresponding to each historical operating condition data in the historical operating condition data set meet the parameter thresholds, and if not, determine the historical operating condition data as historical fault data; Construct the training set based on the historical fault data and the simulated fault condition data set; among them, the training set includes a training data set and a training sample set; the training data set includes a plurality of training data; the training sample set includes training sample data corresponding to each training data, and the sample label corresponding to each training sample data is the fault type corresponding to the training data.

[0015] In some possible embodiments, the model training module is specifically configured to: Construct an initial steam turbine governing valve fault diagnosis model, where the initial steam turbine governing valve fault diagnosis model includes a local feature extraction module, a global feature extraction module, and a fault diagnosis module; Replace the local feature extraction module with an LSTM module, and replace the global feature extraction module with a Transformer module; Determine the steam turbine governing valve fault diagnosis model based on the initial steam turbine governing valve fault diagnosis model after module replacement and a gated network.

[0016] In some possible embodiments, the model training module is specifically configured to: For each training data, perform local feature extraction on the training data based on the LSTM module to obtain a local feature vector; perform global feature extraction on the training data based on the Transformer module to obtain a global feature vector; and, based on the gated network, determine a weighting ratio according to the data characteristics of the training data, perform weighted processing on the local feature vector and the global feature vector according to the weighting ratio, and determine the corresponding to the training data based on the weighted processing result and the fault diagnosis module; Determine the loss value between the fault diagnosis information corresponding to the training data and the sample label corresponding to the training sample data based on a preset loss function, and adjust the model parameters of the steam turbine governing valve fault diagnosis model based on the loss value; Repeat the above steps until the training result meets the preset requirements to obtain the trained steam turbine governing valve fault diagnosis model.

[0017] In some possible embodiments, the fault diagnosis module is further configured to: Judge whether the parameters corresponding to the real-time working condition data meet the parameter thresholds; When the parameters corresponding to the real-time working condition data meet the parameter thresholds, determine that the steam turbine governing valve has no fault; When the parameters corresponding to the real-time working condition data do not meet the parameter thresholds, input the real-time working condition data into the trained steam turbine governing valve fault diagnosis model.

[0018] In some possible embodiments, the fault diagnosis module is further configured to: Determine the fault cause and repair suggestions according to a preset fault knowledge base, the fault type, and the real-time working condition data.

[0019] An embodiment of the present disclosure provides a computer device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the computer device runs, the processor communicates with the memory through the bus. When the machine-readable instructions are executed by the processor, the steam turbine throttle fault diagnosis method described in any of the above possible embodiments is executed.

[0020] An embodiment of the present disclosure provides a computer-readable storage medium, on which a computer program is stored. When the computer program is run by a processor, the steam turbine throttle fault diagnosis method described in any of the above possible embodiments is implemented.

[0021] In the steam turbine throttle fault diagnosis method, device, storage medium, and computer device provided in the embodiments of the present disclosure, first, a historical operating condition data set and a simulated fault operating condition data set of the steam turbine throttle are obtained; and a training set is constructed based on the historical operating condition data set and the simulated fault operating condition data set. Secondly, a steam turbine throttle fault diagnosis model is constructed, and the steam turbine throttle fault diagnosis model is trained based on the training set to obtain a trained steam turbine throttle fault diagnosis model. Finally, real-time operating condition data of the steam turbine throttle is obtained, and the fault type of the steam turbine throttle is determined based on the trained steam turbine throttle fault diagnosis model and the real-time operating condition data.

[0022] In this way, by using the trained fault diagnosis model to diagnose the faults of the steam turbine throttle, the present disclosure can realize the fault identification and classification of the steam turbine throttle, thereby improving the safety and reliability of the steam turbine.

[0023] To make the above objects, features, and advantages of the present disclosure more obvious and understandable, the following specific preferred embodiments are given, and in conjunction with the accompanying drawings, the detailed description is as follows. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] To more clearly illustrate the technical solutions of the embodiments of the present disclosure, the accompanying drawings required to be cited in the embodiments will be briefly introduced below. The accompanying drawings herein are incorporated into the specification and form a part of this specification. These drawings show embodiments consistent with the present disclosure and are used together with the specification to explain the technical solutions of the present disclosure. It should be understood that the following drawings only show some embodiments of the present disclosure and should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0025] Figure 1 Shows a flowchart of a steam turbine throttle fault diagnosis method provided by an embodiment of the present disclosure; Figure 2Shows a flowchart of a fault condition data simulation method provided by an embodiment of the present disclosure; Figure 3 Shows a flowchart of a turbine governing valve fault diagnosis model training method provided by an embodiment of the present disclosure; Figure 4 Shows a flowchart of a threshold judgment method for real-time condition data provided by an embodiment of the present disclosure; Figure 5 Shows a structural schematic diagram of a turbine governing valve fault diagnosis device provided by an embodiment of the present disclosure; Figure 6 Shows a structural schematic diagram of a computer device provided by an embodiment of the present disclosure. Detailed implementation manners

[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present disclosure clearer, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only some of the embodiments of the present disclosure, rather than all of the embodiments. Usually, the components of the embodiments of the present disclosure described and illustrated in the accompanying drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present disclosure provided in the accompanying drawings is not intended to limit the scope of the present disclosure to be protected, but merely represents selected embodiments of the present disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of the present disclosure without creative efforts shall fall within the scope of protection of the present disclosure.

[0027] It should be noted that: similar reference numerals and letters denote similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0028] The term "and / or" in this article merely describes an association relationship and indicates that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the term "at least one" in this article means any one of a plurality or any combination of at least two of a plurality. For example, including at least one of A, B, and C may represent including any one or more elements selected from the set composed of A, B, and C.

[0029] At present, the traditional method of diagnosing turbine valve faults mainly relies on experienced maintenance personnel through on-site inspection and manual analysis. This method not only consumes a lot of manpower and time, but also has limited ability to capture and analyze fault information. Especially when facing complex turbine valve faults, due to the high coupling and dynamic nonlinear characteristics of the internal signals of the system, it is difficult for traditional methods to obtain deep fault information in a timely and accurate manner. In addition, the traditional monitoring system is mainly based on rule algorithms and simple threshold judgments, which has great limitations for complex multidimensional data feature extraction and fault trend prediction, and often requires subsequent manual intervention.

[0030] During the operation of the steam turbine, the failure of the steam turbine valve will not only affect the stability of the steam turbine operation, but also may cause problems such as abnormal system control and reduced thermodynamic efficiency. In severe cases, it may even lead to unplanned equipment shutdown, resulting in huge economic losses. In modern smart power plants, how to use advanced fault diagnosis technology to achieve real-time monitoring, fault location and trend prediction of steam turbine valves has become a key issue that needs to be solved urgently.

[0031] The research found that among the related technologies, the diagnostic methods for turbine valve faults mainly include those based on generative adversarial networks and CCD detection. The generative adversarial network-based method can generate high-quality simulation data and effectively alleviate the problem of insufficient samples, but this method ignores the dependency between time series data, resulting in poor performance when processing continuous time series data; while the CCD detection method mainly relies on image and signal processing technology. Although it can analyze the fault to a certain extent, its real-time performance is poor and it is prone to delays in scenarios that require rapid response, affecting the efficiency of fault diagnosis.

[0032] Based on the above research, a method, device, storage medium and computer equipment for diagnosing turbine regulating valve faults are provided in the embodiments of the present invention. First, a historical operating condition data set and a simulated fault operating condition data set of the turbine regulating valve are obtained; and a training set is constructed based on the historical operating condition data set and the simulated fault operating condition data set; secondly, a turbine regulating valve fault diagnosis model is constructed, and the turbine regulating valve fault diagnosis model is trained based on the training set to obtain a trained turbine regulating valve fault diagnosis model; finally, the real-time operating condition data of the turbine regulating valve is obtained, and the fault type of the turbine regulating valve is determined based on the trained turbine regulating valve fault diagnosis model and the real-time operating condition data.

[0033] In the disclosed embodiment, by using a trained fault diagnosis model to perform fault diagnosis on a steam turbine regulating valve, fault identification and classification of the steam turbine regulating valve can be achieved, thereby improving the safety and reliability of the steam turbine.

[0034] To facilitate the understanding of this embodiment, the execution subject of the steam turbine governing valve fault diagnosis method provided by the embodiments of the present disclosure will be introduced in detail first. The execution subject of the steam turbine governing valve fault diagnosis method provided by the embodiments of the present disclosure is a computer device. This computer device can be a server. Among them, the server can be an independent physical server, a server cluster or a distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud storage, big data, and artificial intelligence platforms.

[0035] The following will detail the steam turbine governing valve fault diagnosis method provided by the embodiments of the present application with reference to the accompanying drawings. Refer to Figure 1 As shown, it is a flowchart of a steam turbine governing valve fault diagnosis method provided by the embodiments of the present disclosure. The method includes the following S101 to S103: S101, obtain the historical operating condition data set and the simulated fault condition data set of the steam turbine governing valve; and construct a training set based on the historical operating condition data set and the simulated fault condition data set.

[0036] It can be understood that the steam turbine governing valve is an important device for controlling the speed of the steam turbine, ensuring that the steam turbine maintains a stable speed under various load changes. Its working principle is based on the regulation of steam flow, thereby controlling the output power of the steam turbine. The historical operating condition data set includes multiple historical operating condition data, and the historical operating condition data refers to multi-dimensional operating condition data collected during the past operation of the steam turbine governing valve, which usually can include indicators such as temperature, pressure, steam flow, and valve opening. The simulated fault condition data set is a data set generated by simulating various faults of the steam turbine governing valve through a simulation model. The simulation data includes various different fault modes, such as jamming faults, wear and corrosion faults, leakage faults, and fracture and detachment faults, etc.

[0037] Specifically, collecting the historical operating condition data set of the steam turbine governing valve reflects the state of the valve during historical operation, and the simulated fault condition data set simulates and covers the possible fault conditions of the steam turbine governing valve. Then, these two parts of the data set are combined to construct a training set for model training, further improving the diversity and coverage of the data, and at the same time avoiding the problems of insufficient data or distribution imbalance in traditional methods.

[0038] Exemplarily, referring to Figure 2 As shown, in order to accurately simulate and analyze the performance of the steam turbine governing valve under various fault conditions, the present disclosure has established multiple fault condition simulation models to obtain the simulated fault condition data set. Specifically, the simulated fault condition data of the present disclosure is obtained through the following steps S201 to S202: S201, construct a three-dimensional geometric model of the steam turbine governing valve.

[0039] Here, in order to make the model accurately reflect the actual structure and physical characteristics inside the throttle valve as much as possible, the present disclosure uses Solidwork tool to establish a three-dimensional geometric model of the throttle valve, accurately depicting the physical characteristics of the internal flow field and mechanical components of the throttle valve. This model not only includes key features of the fluid part such as the geometric shape of the throttle valve passage, the accurate position of the sealing surface, and the size of the throttle hole, but also covers the mechanical structures of dynamic components such as the valve core and spring assembly. In this way, it can be ensured that the subsequent simulation can accurately capture the interaction between the internal flow field and mechanical components of the throttle valve, providing a solid foundation for the simulation of fault conditions.

[0040] S202, perform a simulation on the operating conditions inside the steam turbine throttle valve based on the three-dimensional geometric model, the turbulence model, and the steam turbine operation simulation model, and introduce key parameter perturbations to simulate different fault types, obtaining the simulation fault condition dataset.

[0041] Among them, the key parameter perturbations include pressure perturbation, steam flow perturbation, and load perturbation; by adjusting key parameters such as pressure, steam flow, and load, various fault conditions that may be encountered during the actual operation of the throttle valve can be simulated. In this regard, the faults of the steam turbine throttle valve in the present disclosure are mainly divided into four categories, namely, jamming fault, wear and corrosion fault, leakage fault, and fracture and detachment fault. Specifically, the jamming fault mainly causes a decrease in the pressure after the regulating stage, while the wear and corrosion faults lead to changes in steam flow; the leakage fault is manifested as steam pressure fluctuations, and fracture and detachment will cause load out of control and a decrease in the expansion efficiency of the high-pressure cylinder and the intermediate-pressure cylinder. Although wear and corrosion may lead to leakage, the reasons for leakage are not limited to this, and an incompletely closed throttle valve may also be the cause of leakage. Therefore, the present disclosure independently classifies leakage as a category in the fault dataset to ensure the accuracy and comprehensiveness of fault classification.

[0042] Exemplarily, the turbulence model is used to describe the complex flow state of the fluid inside the throttle valve. It takes into account multiple factors such as fluid density, fluid components, and turbulent viscosity coefficient, and simulates the processes of generation and dissipation of turbulent kinetic energy through a series of equations. Among them, the turbulence model can be expressed as: ; Among them, is expressed as fluid density; is expressed as fluid components; is expressed as turbulent viscosity coefficient; is expressed as the Prandtl number of turbulent kinetic energy; is expressed as the generation term of turbulent kinetic energy; is expressed as the turbulent kinetic energy generated by buoyancy; is expressed as the turbulent expansion term; is expressed as the dissipation rate of turbulent kinetic energy; It is understandable that the steam turbine working simulation model covers multiple aspects such as the particle erosion wear model, heat transfer model, leakage flow model, and force balance and jamming model to comprehensively simulate the performance of the throttle valve under different working conditions. For example, the particle erosion wear model can predict the wear amount of the internal material of the throttle valve due to particle erosion; the heat transfer model is used to simulate the heat transfer phenomenon between the fluid and the solid wall inside the throttle valve and calculate the heat exchange between the fluid and the wall inside the throttle valve; the leakage flow model can estimate the leakage amount of the throttle valve under poor sealing conditions; and the force balance and jamming model is used to analyze the dynamic response of the throttle valve under unbalanced force, jamming, or lag conditions.

[0043] Here, when there is particle erosion inside the throttle valve (such as impurity particles or solid particles in steam), the particle force equation is: ; Where, represents the particle velocity vector; represents the dynamic viscosity of the fluid; represents the particle density; represents the particle diameter; represents the particle Reynolds number; represents the fluid density; represents the fluid velocity vector; represents the gravitational acceleration vector; represents other external forces.

[0044] The particle erosion wear model can be expressed as: ; Where, represents the material wear amount per unit time; C represents the empirical constant; represents the particle density; represents the particle erosion velocity; n, m represent the empirical exponents; represents the particle impact angle.

[0045] Here, the heat transfer model consists of the heat transfer equation, the calculation formula for the heat flux density on the wall surface, and the radiation heat flux equation; the heat transfer equation can be expressed as: ; The calculation formula for the heat flux density on the wall surface can be expressed as: ; The radiation heat flux equation can be expressed as: ; Where, q represents the heat transfer amount; h represents the convective heat transfer coefficient; A represents the heat transfer area; represents the fluid temperature; is represented as the wall temperature; is represented as the wall heat; k is represented as the fluid thermal conductivity; is represented as the normal temperature gradient of the wall; is represented as the radiative heat; is represented as the Stefan-Boltzmann constant; is represented as the emissivity of the material surface; is represented as the ambient temperature.

[0046] Here, wear or deformation of the throttle valve sealing surface may cause leakage, and the leakage flow rate can be calculated by the Bernoulli equation. Therefore, the leakage flow rate model can be expressed as: ; where Q is represented as the leakage flow rate; is represented as the flow coefficient; B is represented as the leakage area; is represented as the upstream pressure; is represented as the downstream pressure.

[0047] Here, during the movement of the throttle valve spool, it may be affected by the combined action of fluid impact force and friction force, resulting in jamming or lag. The force balance and jamming model can be expressed as: ; ; where is represented as the net acting force; is represented as the fluid impact force; is represented as the friction force; is represented as the spring force; m is represented as the spool mass; x is represented as the spool displacement; t is represented as time.

[0048] Specifically, after obtaining the simulation fault condition data set, a training set can be constructed based on the historical condition data set and the simulation fault condition data set, which may include the following steps (1) to (3): (1) Determine the parameter threshold based on the historical condition data set; (2) Judge whether the parameters corresponding to each historical condition data in the historical condition data set meet the parameter threshold. If not, determine the historical condition data as historical fault data; (3) Construct the training set based on the historical fault data and the simulation fault condition data set.

[0049] Specifically, a series of key parameter thresholds can be determined based on the historical operating condition dataset first. The parameter thresholds are crucial for identifying normal operating conditions and potential fault conditions, and can include but are not limited to steam flow threshold, pressure change rate threshold, load threshold, etc. The steam flow threshold is used to measure whether the flow rate of the steam system fluctuates within a normal range; the pressure change rate threshold is used to monitor the speed of pressure change to determine whether there are abnormal fluctuations; the load threshold reflects whether the working load of the system is within a safe and efficient range. By setting these thresholds, a clear basis for judgment can be provided for subsequent data screening and fault identification.

[0050] It can be understood that the simulated fault condition dataset is obtained by simulation based on each fault type, and there is no need to conduct further screening of fault data. Therefore, in this disclosure, only each piece of data in the historical operating condition dataset needs to be reviewed one by one to determine whether the corresponding parameters meet the previously determined parameter thresholds, aiming to screen out potential fault data from a large amount of historical data. If one or more parameters corresponding to a certain historical operating condition data exceed the set threshold range, then this piece of data is regarded as historical fault data and will be incorporated into the analysis and training process of the subsequent model.

[0051] After the screening of each historical operating condition data in the historical operating condition dataset is completed, the screened historical fault data can be combined with the simulated fault condition dataset to jointly construct a training set for training the model. This training set not only contains fault cases in real history but also incorporates fault condition data generated through simulation means, enriching the diversity and complexity of training samples. The training set is further divided into a training data set and a training sample set: the training data set contains multiple training data; the training sample set includes training sample data corresponding to each training data, and these training sample data are assigned sample labels, that is, the fault types corresponding to the training data.

[0052] In some other embodiments, the sample label corresponding to each training sample data can also include fault repair suggestions. These repair suggestions are obtained based on expert experience and fault analysis, providing targeted treatment measures for each fault type. In this way, not only the practicality of the model in fault prediction is improved, but also the model can provide effective diagnostic support while predicting faults.

[0053] In the embodiments of this disclosure, by dynamically comparing and intelligently calibrating the simulation data and the actual operation data, the accuracy and generalization ability of the data are significantly improved, enabling the generated dataset to better meet the training requirements of the LSTM deep learning model for complex operating condition identification and fault prediction.

[0054] S102. Construct a fault diagnosis model for the steam turbine governing valve, and train the fault diagnosis model for the steam turbine governing valve based on the training set to obtain a trained fault diagnosis model for the steam turbine governing valve.

[0055] It can be understood that in the related art, the fusion of Transformer and LSTM is such that one party serves as the encoder and the other as the decoder. However, this method relies on a fixed encoding-decoding structure and cannot be adjusted according to the characteristics of the data. For example, in practical applications, some tasks may require more attention to local features, while some tasks may require a greater weight on global information. Therefore, this fixed structural design may not provide optimal performance in some cases. In response, the present disclosure proposes to construct the fault diagnosis model for the steam turbine governing valve through the following steps (a) to (c): (a) Construct an initial fault diagnosis model for the steam turbine governing valve, where the initial fault diagnosis model for the steam turbine governing valve includes a local feature extraction module, a global feature extraction module, and a fault diagnosis module; (b) Replace the local feature extraction module with an LSTM module and replace the global feature extraction module with a Transformer module; (c) Determine the fault diagnosis model for the steam turbine governing valve based on the initial fault diagnosis model for the steam turbine governing valve after module replacement and a gated network.

[0056] Specifically, the initial fault diagnosis model for the steam turbine governing valve includes a local feature extraction module, a global feature extraction module, and a fault diagnosis module. The local feature extraction module is responsible for capturing local, temporal change patterns from the input data; the global feature extraction module focuses on identifying global change trends or complex correlations. Finally, the fault diagnosis module combines local and global information to classify or predict the model to diagnose whether there is a fault in the steam turbine governing valve.

[0057] Next, the present disclosure further optimizes the initial model by replacing the module, because LSTM has unique advantages in processing time series data, can capture dependencies within a long time span, and through its built-in memory unit, maintains sensitivity to historical data, accurately identifies the time series change pattern in the data, and is conducive to detecting sudden failures (stuck, broken and detached) of the turbine valve, so the present disclosure proposes to replace the local feature extraction module with the LSTM module. At the same time, the global feature extraction module is replaced with the Transformer module. The Transformer can flexibly capture global features through the self-attention mechanism, process sequence data in parallel and focus on key time points, and mine the global dependencies of the input data. Especially when processing large-scale data, the Transformer can effectively avoid the problem of LSTM training difficulties on long sequence data, which is conducive to detecting progressive faults (wear and corrosion, leakage). Through such module replacement, the model can better adapt to the characteristics of different data, thereby improving diagnostic accuracy and efficiency.

[0058] Furthermore, based on the replaced initial model and gating network, the final turbine valve regulating fault diagnosis model is determined. By introducing appropriate control mechanisms, the gating network can dynamically adjust the information flow between different modules, so that the model can adaptively select appropriate features for fault diagnosis according to the different characteristics of the input data. This flexibility makes the model not only highly accurate, but also able to cope with a variety of different application scenarios and data changes.

[0059] Specifically, the present disclosure adopts a gated recurrent unit (GRU) as a gating network. The GRU gating network mainly uses the feature values ​​output by LSTM and Transformer to dynamically evaluate the adaptability of LSTM and Transformer in the current task, and automatically adjusts their outputs by calculating the contribution ratio of LSTM and Transformer to generate a weighted feature representation.

[0060] Among them, GRU can effectively control the flow of information by introducing the "update gate" and "reset gate" mechanisms. The update gate determines the degree of integration of the hidden state at the current moment with the hidden state at the previous moment, thereby adjusting the model's memory strength of historical information; the reset gate controls the degree of forgetting of the hidden state at the previous moment. Based on this, GRU can dynamically adjust the weighted ratio of LSTM and Transformer outputs according to the characteristics of the input data, so that the model can flexibly select the most suitable feature extraction method when processing different types of sequence data, thereby improving the performance of classification tasks.

[0061] For example, refer to Figure 3As shown, when training the steam turbine governing valve fault diagnosis model based on the training data set and the training sample set, the following steps S301 to S303 can be included: S301. For each training data, perform local feature extraction on the training data based on the LSTM module to obtain a local feature vector; perform global feature extraction on the training data based on the Transformer module to obtain a global feature vector; and, based on the gating network, determine a weighting ratio according to the data characteristics of the training data, perform weighting processing on the local feature vector and the global feature vector according to the weighting ratio, and determine the fault diagnosis information corresponding to the training data based on the weighting processing result and the fault diagnosis module.

[0062] Specifically, for each training data, the LSTM module is used for local feature extraction. As a time series data processing tool, the LSTM module can extract local time-dependent features from the training data, and these features help to capture the instantaneous changes or fault symptoms that may occur during the operation of the steam turbine governing valve. At the same time, the Transformer module is used to perform global feature extraction on the same training data, and it can capture the correlation between long-distance data in the sequence. Since the occurrence of a fault is often not instantaneous but develops and evolves gradually, through global feature extraction, the Transformer module can identify the long-term trends or implicit global patterns that affect the operation of the governing valve, thereby further enhancing the accuracy of fault diagnosis.

[0063] It can be understood that after the extraction of local features and global features, the gating network dynamically adjusts the weighting ratio of local features and global features by analyzing the data characteristics of the training data. For example, the gating network may output a weighting vector [0.8, 0.2], indicating that the output of the LSTM accounts for 80% and the output of the Transformer accounts for 20%. Specifically, the gating network will adaptively determine the weights of local features and global features in the final diagnosis according to different data characteristics, enabling the model to flexibly adapt to different fault situations and improving the accuracy and robustness of the diagnosis. Among them, the data characteristics may include sequence length, data complexity, etc. Based on the weighting ratio calculated by the gating network, the outputs of the LSTM and the Transformer are weighted and fused to obtain the final feature representation, which is then used as the input and further processed in combination with the fault diagnosis module to finally obtain the fault diagnosis information corresponding to the training data.

[0064] Here, in the fault diagnosis model, first, the high-dimensional eigenvalues are dimensionally reduced through a fully connected layer to reduce the dimension of the feature space. That is, through the fully connected layer, the model compresses the multi-dimensional features into fewer dimensions. Then, the Softmax function is used to transform the dimensionally reduced features into a probability distribution. The Softmax function maps the scores of each fault category to the interval [0,1], and the sum of the probabilities of all fault categories is 1. Finally, the occurrence probability of each fault is output, and then the fault diagnosis information corresponding to the training data is determined based on the occurrence probabilities of each fault category, which may include the fault type corresponding to the training data.

[0065] In some other embodiments, the fault diagnosis information may further include the potential causes, solutions, etc. of the fault of the fault type corresponding to the training data, which are not specifically limited herein.

[0066] S302, determining the loss value between the fault diagnosis information corresponding to the training data and the sample label corresponding to the training sample data based on a preset loss function, and adjusting the model parameters of the steam turbine governing valve fault diagnosis model based on the loss value.

[0067] Here, after obtaining the fault diagnosis information corresponding to the training data, the difference between the fault diagnosis information output by the model and the actual training sample label is calculated based on a preset loss function. This loss value reflects the deviation of the current model in predicting the fault type. Through the feedback of the loss value, the model will adjust its parameters to better fit the training data, thereby improving the accuracy of fault diagnosis.

[0068] Exemplarily, the present invention uses the cross-entropy loss function for backpropagation to update the weight values. Assuming that the classification result of the model is y1 and the target value is y, the loss function L(y1, y) can be calculated. Through backpropagation, the loss function calculates the gradients for the parameters of each layer (including the weights of the LSTM, Transformer, and gated network), and uses the Adam optimization algorithm for updating. For the LSTM module and the Transformer module, backpropagation gradually updates the weights so that they can better extract temporal features or global features. For the gated network, the weights of the gated network will gradually adjust the weighted ratio of the outputs of the LSTM module and the Transformer module according to the feedback of the loss function, so that the final weighted features can better represent the complexity of the input data, and finally improve the classification accuracy.

[0069] In some other embodiments, the loss function of the model may also use the mean square error (MSE), mean absolute error (MAE), custom loss function, etc., which are not specifically limited herein.

[0070] S303. Repeat the above steps until the training result meets the preset requirements, and obtain the trained steam turbine governing valve fault diagnosis model.

[0071] Here, by continuously repeating the above process and performing multiple iterative trainings, the model parameters are adjusted according to the loss value of the previous training each time until the diagnostic result of the model reaches the preset accuracy requirement or other performance indicators. Through this cyclic optimization method, the model gradually converges, and finally a trained steam turbine governing valve fault diagnosis model that has been fully trained and can effectively perform the steam turbine governing valve fault diagnosis task is obtained.

[0072] The steam turbine governing valve fault diagnosis model proposed in the embodiments of the present disclosure combines the advantages of LSTM and Transformer, and dynamically adjusts feature weighting through a gating network, ensuring that the model can capture the key features of the data from multiple dimensions. Through the continuously optimized training strategy, the finally generated fault diagnosis model can provide efficient and accurate fault prediction and diagnosis results in practical applications, thereby improving the operation safety and reliability of the steam turbine governing valve.

[0073] S103. Obtain the real-time operating condition data of the steam turbine governing valve, and determine the fault type of the steam turbine governing valve based on the trained steam turbine governing valve fault diagnosis model and the real-time operating condition data.

[0074] Here, the real-time operating condition data of the steam turbine governing valve can be collected through a Distributed Control System (DCS), which can include data such as steam flow rate, pressure change rate, and load, and can also include key thermal parameters such as post-stage steam pressure, reheat steam pressure, and main steam pressure. Then, it is input into the trained steam turbine governing valve fault diagnosis model, and this model can determine whether there is a fault in the steam turbine governing valve and identify the type of the fault through the analysis of these data.

[0075] In this way, by using the trained steam turbine governing valve fault diagnosis model to monitor the operating state of the steam turbine governing valve in real time, not only can single-condition data be analyzed, but also comprehensive analysis of multiple parameters can be carried out, thereby improving the accuracy and timeliness of fault diagnosis. For example, when abnormal fluctuations occur in steam pressure or load, potential faults can be quickly detected, and based on the analysis results feedback by the model, it can be judged whether there are problems such as governing valve leakage. In this way, the system can issue an alarm in advance to prompt the operation and maintenance personnel to perform timely preventive maintenance or intervention. Thus, the lag of traditional manual detection methods can be overcome, and downtime and production losses caused by failure to detect faults in a timely manner can be avoided. Especially for steam turbine governing valves, timely identification and repair of faults such as leakage can effectively ensure the stability and efficient operation of the equipment, reduce the risk of damage, extend the service life of the equipment, and improve the safety and reliability of the system.

[0076] Specifically, referring to Figure 4 as shown, before inputting the real-time operating condition data into the steam turbine governing valve fault diagnosis model, S401 - S403 may further be included: S401, determining whether the parameters corresponding to the real-time operating condition data satisfy the parameter thresholds.

[0077] Here, by determining whether the parameters corresponding to the real-time operating condition data satisfy the parameter thresholds determined in advance, it can be quickly judged whether the current steam turbine governing valve is in a fault state, and further judged whether the current steam turbine governing valve needs to further analyze its corresponding fault type.

[0078] S402, when the parameters corresponding to the real-time operating condition data satisfy the parameter thresholds, determining that the steam turbine governing valve has no fault.

[0079] If the parameters in the real-time operating condition data all satisfy the corresponding parameter thresholds, it will be determined that the steam turbine governing valve has no fault. In this way, it helps to save resources and time without the need for complex fault diagnosis, and can reduce unnecessary intervention or maintenance. At this time, the system will mark the relevant working state data as normal and enter the daily monitoring and maintenance process.

[0080] S403, when the parameters corresponding to the real-time operating condition data do not satisfy the parameter thresholds, inputting the real-time operating condition data into the trained steam turbine governing valve fault diagnosis model.

[0081] If the parameters in the real-time operating condition data do not satisfy the set parameter thresholds, the system will input the real-time operating condition data into the trained steam turbine governing valve fault diagnosis model. The model will conduct in-depth analysis on the real-time operating condition data that does not meet the thresholds to help identify the fault type of the current steam turbine governing valve, and take corresponding measures according to the diagnosis results.

[0082] Here, after the steam turbine governing valve fault diagnosis model diagnoses the fault type of the steam turbine governing valve, the fault cause and repair suggestions can also be determined according to a preset fault knowledge base, the fault type, and real-time operating condition data. The preset fault knowledge base contains a large number of historical fault cases, the conditions under which the faults occur, possible causes, and countermeasures. This knowledge can help understand the potential mechanisms behind specific faults. For example, some faults may be caused by excessive wear of a key component in the governing valve system or by incorrect setting of certain parameters in the control system.

[0083] Specifically, through the identification of the fault type, relevant fault modes and their potential causes can be quickly retrieved from the knowledge base. Then, combined with on-site real-time data such as temperature, pressure, and vibration, the cause of the fault can be further confirmed, and detailed maintenance suggestions can be given, which may include specific repair steps, required tools and materials, repair time limits, and the impact on the long-term use of the equipment. In addition, the present disclosure also proposes to optimize the maintenance strategy, that is, to provide the best maintenance timing and methods among multiple fault types. According to the severity of the fault, the time required for maintenance, and the possible impact on production, an appropriate maintenance plan is preferentially selected to ensure the minimization of downtime and the reduction of maintenance costs.

[0084] In the present disclosure, through the data-driven optimized maintenance strategy, the fault diagnosis and maintenance work of the steam turbine become more efficient. The combination of real-time data and the preset fault knowledge base makes the fault diagnosis and maintenance plan not only limited to simple repair, but also a dynamic and continuously optimized process. This can not only improve the accuracy of fault diagnosis and maintenance efficiency, but also significantly improve the overall operation efficiency and reliability of the equipment by reducing unnecessary downtime.

[0085] In the embodiments of the present disclosure, the steam turbine governing valve fault diagnosis method, device, storage medium, and computer equipment provided can realize the fault identification and classification of the steam turbine governing valve by using the trained fault diagnosis model, thereby improving the safety and reliability of the steam turbine.

[0086] Those skilled in the art can understand that in the above method of the specific embodiment, the writing order of each step does not mean a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of each step should be determined according to its function and possible internal logic.

[0087] Based on the same inventive concept, the embodiments of the present disclosure also provide a steam turbine governing valve fault diagnosis device corresponding to the steam turbine governing valve fault diagnosis method. Since the principle of solving problems by the device in the embodiments of the present disclosure is similar to the above steam turbine governing valve fault diagnosis method in the embodiments of the present disclosure, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.

[0088] Referring to Figure 5 as shown, it is a schematic diagram of a steam turbine governing valve fault diagnosis device 500 provided by an embodiment of the present disclosure. The device includes: A data acquisition module 501, configured to acquire a historical operating condition data set and a simulated fault operating condition data set of the steam turbine governing valve; and construct a training set based on the historical operating condition data set and the simulated fault operating condition data set; A model training module 502, configured to construct a steam turbine governing valve fault diagnosis model, and train the steam turbine governing valve fault diagnosis model based on the training set to obtain a trained steam turbine governing valve fault diagnosis model; A fault diagnosis module 503, configured to acquire real-time operating condition data of the steam turbine governing valve, and determine the fault type of the steam turbine governing valve based on the trained steam turbine governing valve fault diagnosis model and the real-time operating condition data.

[0089] In some possible embodiments, the data acquisition module 501 is further configured to: Construct a three-dimensional geometric model of the steam turbine governing valve; Simulate the operating conditions inside the steam turbine governing valve based on the three-dimensional geometric model, a turbulence model, and a steam turbine working simulation model, and introduce key parameter perturbations to simulate different fault types, thereby obtaining the simulated fault operating condition data set; wherein, the key parameter perturbations include pressure perturbation, steam flow perturbation, and load perturbation; the fault types include jamming fault, wear and corrosion fault, leakage fault, and fracture and detachment fault; The turbulence model is expressed as: ; wherein, represents the fluid density; represents the fluid component; represents the turbulent viscosity coefficient; represents the Prandtl number of the turbulent kinetic energy; represents the turbulent kinetic energy generation term; represents the turbulent kinetic energy generated by buoyancy; represents the turbulent expansion term; represents the turbulent kinetic energy dissipation rate; The steam turbine working simulation model includes a particle erosion wear model, a heat transfer model, a leakage flow model, and a force balance and jamming model; The particle erosion wear model is expressed as: ; wherein, represents the material wear amount per unit time; C represents an empirical constant; represents the particle density; is expressed as the particle erosion velocity; n and m are expressed as empirical exponents; is expressed as the particle impact angle; The heat transfer model is expressed as: ; ; ; wherein, q is expressed as the heat transfer quantity; h is expressed as the convective heat transfer coefficient; A is expressed as the heat transfer area; is expressed as the fluid temperature; is expressed as the wall temperature; is expressed as the wall heat quantity; k is expressed as the fluid thermal conductivity; is expressed as the wall normal temperature gradient; is expressed as the radiation heat quantity; is expressed as the Stefan - Boltzmann constant; is expressed as the emissivity of the material surface; is expressed as the ambient temperature; The leakage flow rate model is expressed as: ; wherein, Q is expressed as the leakage flow rate; is expressed as the flow coefficient; B is expressed as the leakage area; is expressed as the upstream pressure; is expressed as the downstream pressure; The force balance and jamming model is expressed as: ; ; wherein, is expressed as the net acting force; is expressed as the fluid impact force; is expressed as the frictional force; is expressed as the spring force; m is expressed as the spool mass; x is expressed as the spool displacement; t is expressed as time.

[0090] In some possible embodiments, the historical operating condition data set includes multiple historical operating condition data; the data acquisition module 501 is specifically configured to: Determine a parameter threshold based on the historical operating condition data set; wherein, the parameter threshold includes a steam flow rate threshold, a pressure change rate threshold, and a load threshold; Judge whether the parameters corresponding to each historical operating condition data in the historical operating condition data set meet the parameter threshold, and if not, determine the historical operating condition data as historical fault data; Construct the training set based on the historical fault data and the simulated fault condition data set; wherein, the training set includes a training data set and a training sample set; the training data set includes a plurality of training data; the training sample set includes training sample data corresponding to each of the training data, and the sample label corresponding to each training sample data is the fault type corresponding to the training data.

[0091] In some possible embodiments, the model training module 502 is specifically configured to: Construct an initial steam turbine governing valve fault diagnosis model, wherein the initial steam turbine governing valve fault diagnosis model includes a local feature extraction module, a global feature extraction module, and a fault diagnosis module; Replace the local feature extraction module with an LSTM module, and replace the global feature extraction module with a Transformer module; Determine the steam turbine governing valve fault diagnosis model based on the initial steam turbine governing valve fault diagnosis model after module replacement and a gated network.

[0092] In some possible embodiments, the model training module 502 is specifically configured to: For each training data, perform local feature extraction on the training data based on the LSTM module to obtain a local feature vector; perform global feature extraction on the training data based on the Transformer module to obtain a global feature vector; and, based on the gated network, determine a weighting ratio according to the data characteristics of the training data, perform weighted processing on the local feature vector and the global feature vector according to the weighting ratio, and determine the corresponding to the training data based on the weighted processing result and the fault diagnosis module; Determine the loss value between the fault diagnosis information corresponding to the training data and the sample label corresponding to the training sample data based on a preset loss function, and adjust the model parameters of the steam turbine governing valve fault diagnosis model based on the loss value; Repeat the above steps until the training result meets the preset requirements to obtain the trained steam turbine governing valve fault diagnosis model.

[0093] In some possible embodiments, the fault diagnosis module 503 is further configured to: Judge whether the parameters corresponding to the real-time condition data meet the parameter threshold; When the parameters corresponding to the real-time condition data meet the parameter threshold, determine that the steam turbine governing valve has no fault; When the parameters corresponding to the real-time condition data do not meet the parameter threshold, input the real-time condition data into the trained steam turbine governing valve fault diagnosis model.

[0094] In some possible embodiments, the fault diagnosis module 503 is further configured to: Determine the cause of the fault and the repair suggestion according to a preset fault knowledge base, the type of the fault, and the real-time working condition data.

[0095] Based on the same inventive concept, an embodiment of the present disclosure further provides a computer device. Referring to Figure 6 As shown, it is a schematic structural diagram of a computer device 600 provided by an embodiment of the present disclosure, including a processor 601, a memory 602, and a bus 603. Among them, the memory 602 is used to store execution instructions, including an internal memory 6021 and an external memory 6022; here, the internal memory 6021 is also called the main memory, which is used to temporarily store the operation data in the processor 601 and the data exchanged with the external memory 6022 such as a hard disk, and the processor 601 exchanges data with the external memory 6022 through the internal memory 6021.

[0096] In an embodiment of the present application, the memory 602 is specifically configured to store the application program code for executing the solution of the present application, and is controlled by the processor 601 to execute. That is, when the computer device 600 runs, the processor 601 communicates with the memory 602 through the bus 603, so that the processor 601 executes the application program code stored in the memory 602, and further executes the method described in any of the foregoing embodiments.

[0097] Among them, the memory 602 may be, but is not limited to, a random access memory (RAM), a read only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), etc.

[0098] The processor 601 may be an integrated circuit chip with the ability to process signals. The above-mentioned processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. It can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0099] It can be understood that the structure schematically shown in the embodiments of the present application does not constitute a specific limitation on the computer device 600. In other embodiments of the present application, the computer device 600 may include more or fewer components than shown in the figure, or combine certain components, or split certain components, or have different component arrangements. The components shown in the figure can be implemented in hardware, software, or a combination of software and hardware.

[0100] The embodiments of the present disclosure also provide a computer-readable storage medium, on which a computer program is stored. When the computer program is run by a processor, it executes the steps of the steam turbine governing valve fault diagnosis method described in the above method embodiments. Among them, the storage medium may be a volatile or non-volatile computer-readable storage medium.

[0101] The embodiments of the present disclosure also provide a computer program product, which carries program codes. The instructions included in the program codes can be used to execute the steps of the steam turbine governing valve fault diagnosis method described in the above method embodiments. For specific details, please refer to the above method embodiments and will not be elaborated here.

[0102] Among them, the above computer program product can be specifically implemented in a manner of hardware, software, or a combination thereof. In an alternative embodiment, the computer program product is specifically embodied as a computer storage medium. In another alternative embodiment, the computer program product is specifically embodied as a software product, such as a software development kit (SDK), etc.

[0103] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems and devices described above can refer to the corresponding processes in the foregoing method embodiments, and will not be elaborated herein. In several embodiments provided in the present disclosure, it should be understood that the disclosed systems and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. For another example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the couplings or direct couplings or communication connections shown or discussed with each other can be through some communication interfaces. The indirect couplings or communication connections of the devices or units can be in electrical, mechanical or other forms.

[0104] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0105] In addition, in each embodiment of the present disclosure, the functional units can be integrated in one processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0106] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium executable by a processor. Based on such an understanding, the technical solution of the present disclosure, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of the present disclosure. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks, or optical discs that can store program codes.

[0107] Finally, it should be noted that the above-described embodiments are only specific embodiments of the present disclosure, used to illustrate the technical solutions of the present disclosure, rather than limiting them. The protection scope of the present disclosure is not limited thereto. Although the present disclosure has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that any person skilled in the art within the technical scope disclosed by the present disclosure can still modify the technical solutions described in the foregoing embodiments, or can easily think of changes, or perform equivalent replacements on some of the technical features; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present disclosure, and should all be covered within the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be subject to the protection scope of the claims.

Claims

1. A method for diagnosing faults of a steam turbine governing valve, characterized in that, Including: Obtain the historical operating condition dataset and the simulated fault condition dataset of the steam turbine governing valve; and construct a training set based on the historical operating condition dataset and the simulated fault condition dataset; Construct a steam turbine governing valve fault diagnosis model, and train the steam turbine governing valve fault diagnosis model based on the training set to obtain a trained steam turbine governing valve fault diagnosis model; Obtain the real-time operating condition data of the steam turbine governing valve, and determine the fault type of the steam turbine governing valve based on the trained steam turbine governing valve fault diagnosis model and the real-time operating condition data.

2. The method according to claim 1, wherein The simulated fault condition dataset is obtained through the following steps: Construct a three-dimensional geometric model of the steam turbine governing valve; Simulate the operating conditions inside the steam turbine governing valve based on the three-dimensional geometric model, the turbulence model, and the steam turbine working simulation model, and introduce key parameter perturbations to simulate different fault types to obtain the simulated fault condition dataset; wherein, the key parameter perturbations include pressure perturbation, steam flow perturbation, and load perturbation; the fault types include jamming fault, wear and corrosion fault, leakage fault, and fracture and detachment fault; The turbulence model is expressed as: ； wherein, is represented as the fluid density; is represented as the fluid component; is represented as the turbulent viscosity coefficient; is represented as the Prandtl number of the turbulent kinetic energy; is represented as the turbulent kinetic energy generation term; is represented as the turbulent kinetic energy generated by buoyancy; is represented as the turbulent expansion term; is represented as the turbulent kinetic energy dissipation rate; The steam turbine working simulation model includes a particle erosion wear model, a heat transfer model, a leakage flow model, and a force balance and jamming model; The particle erosion wear model is expressed as: ； wherein, represents the material wear amount per unit time; C represents an empirical constant; represents the particle density; represents the particle erosion velocity; n, m represent empirical exponents; represents the particle impact angle; The heat transfer model is expressed as: ； ； ； Among them, q represents the heat transfer amount; h represents the convective heat transfer coefficient; A represents the heat transfer area; represents the fluid temperature; represents the wall temperature; represents the wall heat; k represents the fluid thermal conductivity; represents the normal temperature gradient of the wall; represents the radiation heat; represents the Stefan-Boltzmann constant; represents the emissivity of the material surface; represents the ambient temperature; The leakage flow model is expressed as: ； Among them, Q represents the leakage flow rate; represents the flow coefficient; B represents the leakage area; represents the upstream pressure; represents the downstream pressure; The force balance and jamming model is expressed as: ； ； wherein, is represented as the net force; is represented as the fluid impact force; is represented as the frictional force; is represented as the spring force; m represents the spool mass; x represents the spool displacement; t represents the time.

3. The method according to claim 1, wherein The historical operating condition dataset includes multiple historical operating condition data; the construction of the training set based on the historical operating condition dataset and the simulated fault condition dataset includes: Determine parameter thresholds based on the historical operating condition dataset; wherein, the parameter thresholds include a steam flow threshold, a pressure change rate threshold, and a load threshold; Judge whether the parameters corresponding to each historical operating condition data in the historical operating condition dataset meet the parameter thresholds. If not, determine the historical operating condition data as historical fault data; Construct the training set based on the historical fault data and the simulated fault condition dataset; wherein, the training set includes a training dataset and a training sample set; the training dataset includes multiple training data; the training sample set includes training sample data corresponding to each training data, and the sample label corresponding to each training sample data is the fault type corresponding to the training data.

4. The method according to claim 3, characterized in that, The construction of the steam turbine governing valve fault diagnosis model includes: Construct an initial steam turbine governing valve fault diagnosis model, wherein the initial steam turbine governing valve fault diagnosis model includes a local feature extraction module, a global feature extraction module, and a fault diagnosis module; Replace the local feature extraction module with an LSTM module, and replace the global feature extraction module with a Transformer module; Determine the steam turbine governing valve fault diagnosis model based on the initial steam turbine governing valve fault diagnosis model after module replacement and the gated network.

5. The method according to claim 4, wherein The training of the steam turbine governing valve fault diagnosis model based on the training set includes: For each training data, local feature extraction is performed on the training data based on the LSTM module to obtain local feature vectors; global feature extraction is performed on the training data based on the Transformer module to obtain global feature vectors; and, based on the gating network, a weighting ratio is determined according to the data characteristics of the training data, the local feature vectors and the global feature vectors are weighted according to the weighting ratio, and based on the weighted processing result and the fault diagnosis module, fault diagnosis information corresponding to the training data is determined; Based on a preset loss function, a loss value between the fault diagnosis information corresponding to the training data and the sample label corresponding to the training sample data is determined, and based on the loss value, the model parameters of the steam turbine throttle fault diagnosis model are adjusted; Repeat the above steps until the training result meets the preset requirements to obtain the trained steam turbine throttle fault diagnosis model.

6. The method according to claim 3, characterized in that, Before determining the fault type of the steam turbine throttle based on the trained steam turbine throttle fault diagnosis model and the real-time operating condition data, it includes: Judging whether the parameters corresponding to the real-time operating condition data meet the parameter thresholds; When the parameters corresponding to the real-time operating condition data meet the parameter thresholds, it is determined that the steam turbine throttle has no fault; When the parameters corresponding to the real-time operating condition data do not meet the parameter thresholds, the real-time operating condition data is input into the trained steam turbine throttle fault diagnosis model.

7. The method according to claim 6, characterized in that After determining the fault type of the steam turbine throttle based on the trained steam turbine throttle fault diagnosis model and the real-time operating condition data, it further includes: Determining the fault cause and repair suggestions according to a preset fault knowledge base, the fault type and the real-time operating condition data.

8. A steam turbine governing valve fault diagnosis device, characterized in that, It includes: A data acquisition module, configured to acquire the historical operating condition data set and the simulated fault operating condition data set of the steam turbine throttle; and construct a training set based on the historical operating condition data set and the simulated fault operating condition data set; A model training module, configured to construct a steam turbine throttle fault diagnosis model, and train the steam turbine throttle fault diagnosis model based on the training set to obtain a trained steam turbine throttle fault diagnosis model; A fault diagnosis module, configured to acquire the real-time operating condition data of the steam turbine throttle, and determine the fault type of the steam turbine throttle based on the trained steam turbine throttle fault diagnosis model and the real-time operating condition data.

9. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

10. A computer device, comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, characterized in that, When the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

Images