Water turbine real-time fault early warning method and system

Abstract

The invention discloses a water turbine real-time fault early warning method and system, and the method comprises the steps: obtaining the operation data of a water turbine, and carrying out the preprocessing of the operation data of the water turbine; analyzing the preprocessed operation data of the water turbine by using an improved gradient enhanced decision tree model, and establishing and updating a fault prediction model; and when the fault prediction model predicts a potential fault, an early warning signal is automatically sent out through the Internet of Things technology. The improved gradient enhancement decision tree model is utilized, the fault mode can be learned from historical and real-time data, the prediction model can be continuously optimized through adaptive feature selection and an online and incremental learning mechanism, fault prediction is more accurate, the method can automatically adapt to different operation conditions and environment changes, and the fault prediction efficiency is improved. The universality and the effectiveness of the early warning system are greatly improved; when a potential fault is detected, the system automatically triggers a preset emergency measure according to the severity of the fault.

Description

Technical Field

[0001] The present invention relates to the technical field of fault early warning, and in particular to a real-time fault early warning method and system for a water turbine. Background Art

[0002] In modern industrial production, turbines are an important energy conversion equipment and are widely used in hydropower stations, manufacturing and other industries. The stable operation of turbines is crucial to ensuring energy supply and production safety. However, due to the complexity of turbines and the variability of the operating environment, the types of failures that may occur are diverse and the consequences are serious. Therefore, it is particularly important to monitor and warn potential failures of turbines in real time and accurately.

[0003] Traditional turbine fault detection and early warning systems mostly rely on central processing systems or need to send data to remote servers for processing and analysis. This approach not only has delays in data transmission and processing, but also often lacks sufficient real-time performance and is difficult to respond quickly to faults. In addition, traditional systems also have shortcomings in fault prediction accuracy and adaptability. Because they usually use fixed algorithms to handle fault detection, they are difficult to adapt to complex and changing industrial environments, which limits their application effects under different operating conditions. Data processing in traditional methods usually takes a long time, resulting in delayed response time for fault warnings, which is unacceptable in fault warnings for key equipment such as turbines, because even a small delay may lead to serious mechanical failures or safety accidents. In addition, traditional systems often use a set of general algorithms to handle all types of mechanical equipment. This method is difficult to cover all potential failure modes, and it is impossible to optimize the prediction model based on the specific operating status and historical maintenance records of the equipment. Therefore, there is a lot of room for improvement in the accuracy and adaptability of fault prediction. Summary of the invention

[0004] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0005] In view of the above existing problems, the present invention is proposed. Therefore, the present invention provides a real-time fault warning method for a hydraulic turbine to solve the problems of data transmission and processing delay, insufficient prediction accuracy and insufficient adaptability existing in the traditional hydraulic turbine fault warning system.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a real-time fault warning method for a hydraulic turbine, comprising:

[0008] Acquiring turbine operation data, and preprocessing the turbine operation data;

[0009] The improved gradient boosted decision tree model is used to analyze the preprocessed turbine operation data, and the fault prediction model is established and updated;

[0010] When the fault prediction model predicts a potential fault, an early warning signal is automatically issued through the Internet of Things technology.

[0011] As a preferred solution of the real-time fault warning method for a hydraulic turbine according to the present invention, data preprocessing includes:

[0012] Data preprocessing includes eliminating outliers and missing data in the data set, applying threshold detection and interpolation techniques, identifying and replacing or deleting data points that exceed a predetermined threshold;

[0013] The pre-processed turbine operation data is classified and processed, and the data is divided into mechanical fatigue data, structural stress data and chemical property data according to the type and source of the data;

[0014] Use statistical analysis methods and pattern recognition techniques to conduct preliminary analysis on the classified data and extract key features and patterns in the data.

[0015] As a preferred solution of the real-time fault warning method for a hydraulic turbine described in the present invention, the improved gradient enhancement decision tree model includes:

[0016] The construction of the improved gradient enhanced decision tree model introduces dynamic learning rate, adaptive feature selection, and online and incremental learning mechanisms;

[0017] Initialize the improved gradient boosting decision tree model with a periodic adjustment factor to adapt to periodic maintenance activities or seasonal changes.

[0018] As a preferred solution of the real-time fault warning method for a hydraulic turbine according to the present invention, the dynamic learning rate, adaptive feature selection and online and incremental learning mechanisms include:

[0019] Apply dynamic learning rate to adapt to improved gradient boosting decision tree model performance and environment changes;

[0020] Evaluate the correlation between features and the current residual to select the most relevant features;

[0021] The improved gradient boosting decision tree model is updated online and incrementally to adapt to new data.

[0022] As a preferred solution of the real-time fault warning method for a hydraulic turbine according to the present invention, the method further comprises:

[0023] The residual error of the improved gradient boosted decision tree model is calculated, incorporating the effects of environmental and operating conditions, expressed as:

[0024]

[0025] Among them, r it represents the residual of the i-th data point at the t-th iteration, reflecting the deviation between the model prediction and the actual observation, y i is the true value, F(x i ) is the predicted value of the gradient boosting decision tree model, λ is the amplitude coefficient of periodic adjustment, is the phase shift and ω is the angular frequency.

[0026] As a preferred solution of the real-time fault warning method for a hydraulic turbine according to the present invention, when the fault prediction model predicts a potential fault, it includes:

[0027] Calculate the probability of failure occurrence, expressed as:

[0028]

[0029] Among them, P f (x) represents the probability of failure, F t,j (x) represents the fault prediction model, F t The eigenvalue of the jth output, β j is the weight parameter of feature j, and β0 is a constant term.

[0030] As a preferred solution of the real-time fault warning method for a water turbine according to the present invention, the method of automatically issuing a warning signal through the Internet of Things technology includes:

[0031] The warning signal is transmitted in real time to the operation control center and the receiving equipment of relevant maintenance personnel through the Internet of Things technology. The warning information includes the fault type, possible cause, recommended response measures, and the time and location of the fault.

[0032] Depending on the type and severity of the fault, preset emergency response measures are automatically executed, including adjusting the turbine's operating parameters, starting backup systems, or completely shutting down the turbine for maintenance.

[0033] In a second aspect, the present invention provides a system for real-time fault warning of a hydraulic turbine, comprising:

[0034] An acquisition module, used for acquiring turbine operation data and preprocessing the turbine operation data;

[0035] A prediction module is used to analyze the pre-processed turbine operation data using an improved gradient enhancement decision tree model to establish and update a fault prediction model;

[0036] The alarm module is used to automatically send out a warning signal through the Internet of Things technology when the fault prediction model predicts a potential fault.

[0037] In a third aspect, the present invention provides a computing device, comprising:

[0038] Memory and processor;

[0039] The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the real-time fault warning method for a hydraulic turbine are implemented.

[0040] In a fourth aspect, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the real-time fault warning method for a water turbine.

[0041] Compared with the prior art, the present invention has the following beneficial effects: the present invention utilizes an improved gradient enhanced decision tree model, which can not only learn fault modes from historical and real-time data, but also continuously optimize the prediction model through adaptive feature selection and online and incremental learning mechanisms, so that fault prediction is not only more accurate, but also can automatically adapt to different operating conditions and environmental changes, greatly improving the universality and effectiveness of the early warning system; when the present invention detects a potential fault, the system can automatically issue a warning signal and convey the warning information to the operation control center and maintenance personnel in real time through the Internet of Things technology. In addition, according to the severity of the fault, the system can also automatically trigger preset emergency measures, such as adjusting operating parameters, starting a backup system, or shutting down for maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0043] Figure 1 The present invention is a schematic diagram of the overall process of a real-time fault warning method for a hydraulic turbine according to an embodiment of the present invention. DETAILED DESCRIPTION

[0044] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0045] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0046] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0047] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0048] At the same time, in the description of the present invention, it should be noted that the orientations or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the orientations or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the system or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0049] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0050] Example 1

[0051] Reference Figure 1, as an embodiment of the present invention, provides a real-time fault warning method for a hydraulic turbine, comprising:

[0052] S100: acquiring turbine operation data and preprocessing the turbine operation data;

[0053] In the embodiment of the present application, a variety of sensors are deployed at key parts of the turbine to obtain turbine operation data in real time;

[0054] Specifically, acoustic emission sensors are used to detect mechanical fatigue and crack development; optical fiber sensors are used to measure structural deformation and internal stress; and electrochemical sensors are used to monitor water quality changes and corrosion rates. IoT technology is used to transmit turbine operating status data sets collected from multiple sensors in real time to edge computing devices located at or near the turbine site.

[0055] In the embodiment of the present application, acoustic emission sensors are deployed at the turbine blades, bearings and sealing rings of the turbine, and the crack development of the turbine is detected using acoustic emission sensors within a preset frequency response range. The acoustic emission sensors are configured to continuously monitor the sound waveform of the machine and convert the sound waveform data V se Real-time transmission, identifying and locating tiny cracks caused by mechanical fatigue using the following formula:

[0056]

[0057] Among them, V se (x) represents the intensity of the acoustic emission signal at time t, A is the signal amplitude, τ is the time of signal arrival, σ is the time diffusion parameter of the signal, and f is the signal frequency;

[0058] In the embodiment of the present application, a fiber optic sensor based on a Bragg grating of optical time domain reflection technology is installed at a key node of the turbine structure. The fiber optic sensor measures and records the deformation and internal stress of the turbine structure. The optical time domain reflection technology calculates the stress and deformation by measuring the change in the wavelength of light reflected in the optical fiber, and generates structural stress data V os , using the expression for the relationship between wavelength drift and stress and deformation in optical fiber:

[0059]

[0060] Where Δλ is the change in wavelength, λ0 is the wavelength without deformation, and P e is the photoelastic coefficient, n is the refractive index, and ∈ is the strain caused by stress;

[0061] In the embodiment of the present application, electrochemical sensors are installed at the water inlet and outlet of the turbine, and ion selective electrodes are used to monitor the conductivity, pH value and other chemical properties of the water quality, and the chemical data V is automatically recorded. se, to assess corrosion rates and their impact on turbine components:

[0062]

[0063] pH = -log[H + ];

[0064] Where G is the conductivity of the conductive element, L and A are the length and cross-sectional area of ​​the water flow, respectively, [H + ] is the hydrogen ion concentration;

[0065] In the embodiment of the present application, the sound waveform data, structural stress data and chemical data are integrated, and a turbine operation status data set D is constructed in real time on the local edge computing device. int The turbine operation status data set includes the time-synchronized measurement values ​​of each sensor and the corresponding sensor location information L pos :

[0066] D int (t)={(t,L pos ,v se (t),v os (t),v ec (t))|t∈T,L pos ∈Locations};

[0067] Among them, T represents the time set, and Locations represents the sensor location set.

[0068] Preferably, data preprocessing includes eliminating outliers and missing data in the data set, applying threshold detection and interpolation techniques, identifying and replacing or deleting data points that exceed a predetermined threshold;

[0069] In the embodiment of the present application, a linear or polynomial interpolation method is used to fill in the missing data points, and the outlier detection and replacement formula used in the data cleaning process is:

[0070]

[0071] Where interp(V(t-1),V(t+1)) means that for the maximum value V max or minimum value V min For the data point V(t), linear interpolation is performed using the previous and next data points, otherwise the original data is retained;

[0072] Preferably, the pre-processed turbine operation data is classified and processed, and the data is divided into a mechanical fatigue data class, a structural stress data class and a chemical property data class according to the type and source of the data;

[0073] In the embodiment of the present application, the classification is expressed as:

[0074] C i = {V(t)|V(t)∈D cln ,classify(V(t))=i},i∈{mf,ss,ch}

[0075] Where classify(V(t)) is a classification function that classifies the data into the corresponding category according to its signal type and source. Mechanical fatigue data category C mf , Structural stress data type C ss and chemical property data class C ch ;

[0076] Preferably, statistical analysis methods and pattern recognition techniques are used to perform preliminary analysis on the classified data to extract key features and patterns in the data, expressed as:

[0077]

[0078] Among them, f i Represents the classification data C i The feature functions extracted from , which include statistical parameters, spectrum analysis results or pattern recognition output based on machine learning, w i is the weight, indicating the importance of different features in fault prediction;

[0079] Finally, the preprocessed turbine operation data is generated as:

[0080]

[0081] in, represents a data fusion operation, combining the processing results from different classifications into a single feature vector, FFT (C i ) represents the data C for each category i i Fast Fourier transform is used to extract features in the frequency domain, PCA (C i ) means to perform principal component analysis on the data of each category, Stat(C i ) means extracting basic statistics from each category of data.

[0082] It should be noted that the present invention deploys sensors at key locations of the turbine to monitor various operating parameters of the equipment in real time, and uses edge computing technology to instantly process the data at the location where the data is generated, thereby reducing the time delay in data transmission and processing, enabling the system to quickly detect and respond in the early stages of fault formation, thereby preventing equipment damage or more serious safety accidents that may be caused by delayed processing.

[0083] S102: Analyze the preprocessed turbine operation data using an improved gradient enhancement decision tree model to establish and update a fault prediction model;

[0084] Preferably, the construction of the improved gradient boosting decision tree model introduces dynamic learning rate, adaptive feature selection, and online and incremental learning mechanisms;

[0085] Preferably, the improved gradient boosting decision tree model is initialized, and a periodic adjustment factor is used to adapt to periodic maintenance activities or seasonal changes, which is expressed as:

[0086]

[0087] Among them, γ represents a constant bias term that provides a baseline prediction for the model, and α k represents the amplitude coefficient of periodic adjustment, which controls the influence strength of the periodic function. ω0 represents the fundamental frequency, which defines the frequency of the periodic function. represents the phase shift, which determines the starting point of the periodic change, and k represents an integer multiple of the frequency, which is used to construct a multi-frequency periodic model;

[0088] Preferably, the residual of the improved gradient boosted decision tree model is calculated, taking into account the influence of environmental and operating conditions, and is expressed as:

[0089]

[0090] Among them, r it represents the residual of the i-th data point at the t-th iteration, reflecting the deviation between the model prediction and the actual observation, y i is the true value, F(x i ) is the predicted value of the gradient boosting decision tree model, λ is the amplitude coefficient of periodic adjustment, is the phase shift and ω is the angular frequency.

[0091] Preferably, a dynamic learning rate is applied to adapt the performance of the improved gradient boosting decision tree model and environmental changes; the correlation between the features and the current residual is evaluated to select the most relevant features; the improved gradient boosting decision tree model is updated online and incrementally to adapt to new data;

[0092] In the embodiment of the present application, adaptive feature selection evaluates the correlation between the feature and the current residual, and is used to select the most relevant feature as:

[0093]

[0094] Among them, MI represents the feature f i and residual r it The mutual information between

[0095] In the embodiment of the present application, a regularization method is used to prevent overfitting, and important features are selected to establish a decision tree represented as:

[0096] R jm , γ jm =tree(r it ,F t-1 ,X selected )+η∑||θ j || 2 ;

[0097] Among them, R jm represents the region defined by the decision tree at the jth node, γ jm Indicates that in region R jm The corresponding output adjustment amount, θ j represents the parameters of the jth decision tree, and η represents the regularization coefficient, which is used to prevent the model from overfitting;

[0098] In the embodiment of the present application, a dynamic learning rate is applied to adapt to the improved gradient enhancement decision tree model performance and environmental changes as expressed as:

[0099]

[0100] v t =v0e -δt ;

[0101] Among them, v t represents the dynamic learning rate of the tth iteration, v0 represents the initial learning rate, and δ represents the decay parameter, which is used to adjust the speed at which the learning rate decreases over time;

[0102] In the embodiment of the present application, the improved gradient boosting decision tree model is updated online and incrementally to adapt to new data as follows:

[0103] F new (x) = F t (x)+v·ΔF+κ·tanh(ΔF);

[0104] Where ΔF is the model update based on new data, and k is the nonlinear response strength used to adjust the impact of new information.

[0105] Specifically, the preprocessed turbine operation data is received for input into the fault prediction model; an improved gradient boosting decision tree model is initialized on the edge computing device, and the model is adapted to the changing turbine operation conditions and environment by integrating dynamic learning rate, adaptive feature selection, and online and incremental learning mechanisms; the improved gradient boosting decision tree model is applied to perform fault mode recognition and classification on the data in the turbine operation status data set; adaptive feature selection is implemented to dynamically select the features that are most relevant to the current residual; the fault prediction model is updated through online and incremental learning methods, so that the fault prediction model continuously adapts to new data and responds to changes in real-time environment and operating conditions.

[0106] It should be noted that, by introducing dynamic learning rate, adaptive feature selection, and online and incremental learning mechanisms, the improved gradient enhanced decision tree model of the present invention can more effectively analyze the preprocessed turbine operation data, accurately establish and dynamically update the fault prediction model, and this model not only adapts to periodic maintenance activities or seasonal changes, but also improves the accuracy of prediction by calculating residuals and considering the influence of environmental and operating conditions; at the same time, the adaptive feature selection mechanism ensures that the model only uses the most relevant features for prediction, the regularization method effectively prevents overfitting, and the application of dynamic learning rate enables the model to flexibly adapt to performance and environmental changes; through the online and incremental update mechanism, the model can continuously learn new data, thereby continuously optimizing the fault prediction capability and improving the reliability and maintenance efficiency of turbine operation.

[0107] S104: When the fault prediction model predicts a potential fault, an early warning signal is automatically issued through the Internet of Things technology;

[0108] Preferably, the probability of a failure occurring is calculated and expressed as:

[0109]

[0110] Among them, P f (x) represents the probability of failure, F t,j (x) represents the fault prediction model, F t The eigenvalue of the jth output, β j is the weight parameter of feature j, and β0 is a constant term.

[0111] Preferably, the issued warning signal is transmitted in real time to the operation control center and the receiving equipment of relevant maintenance personnel through the Internet of Things technology, and the warning information includes the type of fault, possible cause, recommended response measures, and the time and location of the fault;

[0112] Preferably, according to the type and severity of the fault, the preset emergency response measures are automatically executed, including adjusting the operating parameters of the turbine, starting the backup system or completely shutting down for maintenance, which is expressed as:

[0113]

[0114] Among them, A(x) represents the emergency response measures that are automatically executed, S severity (x) Assess the severity of the fault, T low and T med is the severity threshold, R param (x), R backup (x), R shutdown (x) Response measures, including adjusting turbine operating parameters, activating backup systems, or shutting down the turbine completely for maintenance;

[0115] In an embodiment of the present application, the system records the processing results and response time of each fault warning for future fault response optimization and model adjustment.

[0116] It should be noted that when the fault prediction model uses the Internet of Things technology to accurately predict the potential fault of the turbine, it can calculate the probability of the fault and immediately send a signal containing detailed early warning information to the operation control center and relevant maintenance personnel through the Internet of Things in real time; these early warning information not only clarifies the type of fault, possible causes and recommended response measures, but also points out the time and location of the fault, ensuring that relevant personnel can quickly understand and take action; according to the type and severity of the fault, the system will automatically trigger preset emergency response measures, such as adjusting operating parameters, starting backup systems or completely shutting down for maintenance, to minimize the impact of the fault on production; the system will also record the processing results and response time of each fault warning, providing valuable data for future fault response optimization and model adjustment, and continuously improving fault prevention and response capabilities.

[0117] The above is a schematic scheme of a real-time fault warning method for a water turbine in this embodiment. It should be noted that the technical scheme of the real-time fault warning system for a water turbine and the technical scheme of the real-time fault warning method for a water turbine are of the same concept, and the details of the technical scheme of the real-time fault warning system for a water turbine in this embodiment that are not described in detail can all be referred to the description of the technical scheme of the real-time fault warning method for a water turbine.

[0118] The real-time fault warning system for a hydraulic turbine in this embodiment includes:

[0119] An acquisition module is used to acquire the turbine operation data and pre-process the turbine operation data;

[0120] A prediction module is used to analyze the pre-processed turbine operation data using an improved gradient enhancement decision tree model to establish and update a fault prediction model;

[0121] The alarm module is used to automatically send out early warning signals through the Internet of Things technology when the fault prediction model predicts a potential fault.

[0122] This embodiment further provides a computing device, which is applicable to the case of real-time fault warning of a hydraulic turbine, and includes:

[0123] Memory and processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the real-time fault warning method for a hydraulic turbine as proposed in the above embodiment.

[0124] This embodiment further provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for realizing real-time fault warning of a hydraulic turbine as proposed in the above embodiment is implemented.

[0125] The storage medium proposed in this embodiment and the method for realizing real-time fault warning of a turbine proposed in the above embodiment belong to the same inventive concept. The technical details not described in detail in this embodiment can be referred to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0126] Through the above description of the implementation methods, the technicians in the relevant field can clearly understand that the present invention can be implemented by means of software and necessary general hardware, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ReadOnly, Memory, ROM), random access memory (RandomAccess Memory, RAM), flash memory (FLASH), hard disk or optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform the methods of various embodiments of the present invention.

Claims

1. A real-time fault warning method for a hydraulic turbine, characterized in that: include: Acquiring turbine operation data, and preprocessing the turbine operation data; The improved gradient boosted decision tree model is used to analyze the preprocessed turbine operation data, and the fault prediction model is established and updated; When the fault prediction model predicts a potential fault, an early warning signal is automatically issued through the Internet of Things technology.

2. The real-time fault warning method for a hydraulic turbine according to claim 1, characterized in that: Data preprocessing includes: Data preprocessing includes eliminating outliers and missing data in the data set, applying threshold detection and interpolation techniques, identifying and replacing or deleting data points that exceed a predetermined threshold; The pre-processed turbine operation data is classified and processed, and the data is divided into mechanical fatigue data, structural stress data and chemical property data according to the type and source of the data; Use statistical analysis methods and pattern recognition techniques to conduct preliminary analysis on the classified data and extract key features and patterns in the data.

3. The real-time fault warning method for a hydraulic turbine according to claim 1 or 2, characterized in that: The improved gradient boosting decision tree model includes, The construction of the improved gradient enhanced decision tree model introduces dynamic learning rate, adaptive feature selection, and online and incremental learning mechanisms; Initialize the improved gradient boosting decision tree model with a periodic adjustment factor to adapt to periodic maintenance activities or seasonal changes.

4. The real-time fault warning method for a hydraulic turbine according to claim 3, characterized in that: Dynamic learning rates, adaptive feature selection, and online and incremental learning mechanisms include, Apply dynamic learning rate to adapt to improved gradient boosting decision tree model performance and environment changes; Evaluate the correlation between features and the current residual to select the most relevant features; The improved gradient boosting decision tree model is updated online and incrementally to adapt to new data.

5. The real-time fault warning method for a hydraulic turbine according to claim 4, characterized in that: Also includes, The residual error of the improved gradient boosted decision tree model is calculated, incorporating the effects of environmental and operating conditions, expressed as: Among them, r pt represents the residual of the i-th data point at the t-th iteration, reflecting the deviation between the model prediction and the actual observation, y i is the true value, F(x p ) is the predicted value of the gradient boosting decision tree model, λ is the amplitude coefficient of periodic adjustment, is the phase shift and ω is the angular frequency.

6. The real-time fault warning method for a hydraulic turbine according to claim 5, characterized in that: When the fault prediction model predicts a potential fault, it includes: Calculate the probability of failure occurrence, expressed as: Among them, P f (x) represents the probability of failure, F t,j (x) represents the fault prediction model, F t The eigenvalue of the jth output, β j is the weight parameter of feature j, and β0 is a constant term.

7. The real-time fault warning method for a hydraulic turbine according to claim 6, characterized in that: Automatically sending out early warning signals through IoT technology includes: The warning signal is transmitted in real time to the operation control center and the receiving equipment of relevant maintenance personnel through the Internet of Things technology. The warning information includes the fault type, possible cause, recommended response measures, and the time and location of the fault. Depending on the type and severity of the fault, pre-set emergency response measures are automatically executed, including adjusting the turbine's operating parameters, activating backup systems, or completely shutting down the turbine for maintenance.

8. A system for real-time fault warning of a hydraulic turbine, characterized in that: include, An acquisition module, used for acquiring turbine operation data and preprocessing the turbine operation data; A prediction module is used to analyze the pre-processed turbine operation data using an improved gradient enhancement decision tree model to establish and update a fault prediction model; The alarm module is used to automatically send out a warning signal through the Internet of Things technology when the fault prediction model predicts a potential fault.

9. An electronic device, comprising: Memory and processor; The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the real-time fault warning method for a hydraulic turbine described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the method for real-time fault warning of a hydraulic turbine according to any one of claims 1 to 7.

Images