Intelligent fault diagnosis method, system, equipment and storage medium for steam turbine generator unit

Abstract

The application provides a steam turbine generator unit intelligent fault diagnosis method, system, device and storage medium, comprising: acquiring the operation state data and external factor data of the steam turbine generator unit in real time through a multi-channel data acquisition system; performing multi-source feature extraction on the original data obtained after data preprocessing, and performing feature fusion on the extracted multi-source features; monitoring and diagnosing the operation state of the steam turbine generator unit according to a pre-constructed intelligent fault diagnosis model, wherein the intelligent fault diagnosis model is a deep learning neural network model trained based on the fused multi-source feature data. The application solves the problems of data missing, sparsity and scarcity when the traditional deep learning method is applied to steam turbine generator unit fault detection, and can realize high-precision monitoring, intelligent diagnosis and early warning of multiple component and multiple type coupling faults of the unit, effectively ensuring stable and reliable operation of the unit.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of fault diagnosis of steam turbine generator units, and particularly relates to an intelligent fault diagnosis method, system, device and storage medium for a steam turbine generator unit. BACKGROUND

[0002] The steam turbine generator unit is a core device in the power system, and the safety and stability of its operation are of great significance to the entire system. In the existing fault monitoring process of the steam turbine generator unit, data loss or sparsity is caused by sensor failure, data uploading delay and other reasons, and in addition, there are problems such as non-uniformity of dimensions and large differences in importance among multi-source state parameters, resulting in low accuracy of fault monitoring and affecting the safe and stable operation of the unit. SUMMARY

[0003] Therefore, the application aims to provide an intelligent fault diagnosis method, system, device and storage medium for a steam turbine generator unit to solve the problem of inaccurate fault monitoring of the steam turbine generator unit and affect the stable operation of the unit due to data loss, sparsity and scarcity when the traditional deep learning method is applied to fault detection of the steam turbine generator unit.

[0004] To achieve the above-mentioned purpose, the technical scheme of the application is as follows:

[0005] In a first aspect, the application provides an intelligent fault diagnosis method for a steam turbine generator unit, comprising:

[0006] obtaining the running state data and external factor data of the steam turbine generator unit in real time through a multi-channel data acquisition system; wherein the running state data includes mechanical state data, driving state data and process state data of the main machine and auxiliary machine;

[0007] performing multi-source feature extraction on the original data obtained after data preprocessing, and performing feature fusion on the extracted multi-source features;

[0008] monitoring and diagnosing the running state of the steam turbine generator unit according to a pre-constructed intelligent fault diagnosis model, wherein the intelligent fault diagnosis model is a deep learning neural network model trained based on the fused multi-source feature data.

[0009] In a second aspect, based on the same inventive concept, the application further provides an intelligent fault diagnosis system for a steam turbine generator unit, comprising:

[0010] a data acquisition module configured to obtain the running state data and external factor data of the steam turbine generator unit in real time through a multi-channel data acquisition system; wherein the running state data includes mechanical state data, driving state data and process state data of the main machine and auxiliary machine;

[0011] The feature extraction module is configured to perform multi-source feature extraction on the original data obtained after data preprocessing, and perform feature fusion on the extracted multi-source features.

[0012] The fault diagnosis module is configured to monitor and diagnose the operation state of the steam turbine generator unit according to a pre-constructed intelligent fault diagnosis model, wherein the intelligent fault diagnosis model is a deep learning neural network model trained based on the fused multi-source feature data.

[0013] In a third aspect, based on the same inventive concept, the present application also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to realize the method of the first aspect.

[0014] In a fourth aspect, based on the same inventive concept, the present application also provides a non-transitory computer readable storage medium, wherein the non-transitory computer readable storage medium stores computer instructions for causing the computer to execute the method of the first aspect.

[0015] Compared with the prior art, the steam turbine generator unit intelligent fault diagnosis method, system, device and storage medium of the present application have the following beneficial effects:

[0016] The steam turbine generator unit intelligent fault diagnosis method, system, device and storage medium of the present application, the method extracts the correlation features between different measuring points and different monitoring quantities based on the full state parameter monitoring of the monitored object full process, realizes high-precision detection of multi-component and multi-type coupling faults of the unit based on the constructed intelligent fault diagnosis model, and finally realizes closed-loop control from monitoring to maintenance, improves the reliability and maintenance efficiency of the equipment. BRIEF DESCRIPTION OF DRAWINGS

[0017] The accompanying drawings, which form a part of this application, are intended to provide further understanding of the application and are incorporated herein for explanation of the application. The schematic embodiments of the present application and their descriptions are used to explain the present application, and do not constitute improper limitation on the present application. In the drawings:

[0018] Figure 1 A steam turbine generator unit intelligent fault diagnosis method flow chart is described in the embodiments of the present application;

[0019] Figure 2 A steam turbine generator unit intelligent fault diagnosis system flow chart is described in the embodiments of the present application;

[0020] Figure 3 An electronic device hardware structure schematic diagram is described in the embodiments of the present application.

[0021] Explanation of reference signs:

[0022] 11 - data acquisition module; 12 - feature extraction module; 13 - fault diagnosis module; 1010 - processor; 1020 - memory; 1030 - input / output interface; 1040 - communication interface; 1050 - bus. DETAILED DESCRIPTION

[0023] For the purpose, technical solutions and advantages of the present application to be more clear, the present application is further described in detail below in combination with specific embodiments and with reference to the drawings.

[0024] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present application should be understood as the usual meaning understood by those skilled in the art to which the embodiments of the present application belong. The terms "first", "second" and similar terms used in the embodiments of the present application do not represent any order, quantity or importance, but are only used to distinguish different components. The terms "include" or "contain" and similar terms mean that the elements or objects before the terms cover the elements or objects listed after the terms and their equivalents, and do not exclude other elements or objects. The terms "connect" or "connected" and similar terms are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. The terms "up", "down", "left", "right" and the like only represent relative positional relationships, and when the absolute positions of the described objects change, the relative positional relationships may also change accordingly.

[0025] The embodiments of the present application are described in detail below in combination with the drawings.

[0026] Please refer to Figure 1 As shown in the drawings, the embodiments of the present application provide an intelligent fault diagnosis method for a steam turbine generator unit, which specifically includes the following steps:

[0027] Step S101, real-time acquisition of operation state data and external factor data of the steam turbine generator unit through a multi-channel data acquisition system; wherein the operation state data includes mechanical state data, driving state data and process state data of the main engine and auxiliary engine.

[0028] Specifically, in the present embodiment, the steam turbine generator set includes: 8 main machines, i.e. 6 main machines with a power of 350 MW and 2 main machines with a power of 600 MW, 6 auxiliary machines (a total of 12 feed water pump turbine machines, i.e. covering auxiliary machine main equipment such as feed water pumps, condensate water pumps, circulating pumps, fans, compressors, etc.). The vibration signals, phase signals, shaft displacement signals, etc. of the main machines and auxiliary machines come from the vibration signal buffer output port of the TSI system (turbine safety monitoring system), and are connected to the data collector through the interface module in a hard-wired manner to ensure the safety of the TSI system; the auxiliary machines and other equipment not connected to the online data collector are connected to the intelligent platform through the multi-channel portable data acquisition and analysis instrument.

[0029] The process signals (including temperature, flow, valve position, pressure, power) are output through the DCSAO module to connect to the process quantity signal acquisition module of the data collector to realize hard signal connection and ensure the safety of the system.

[0030] The MD4008 intelligent data collector used in the present scheme can work with the front-end system, and there are two types of 4 modules and 8 modules to choose from, both of which have independent main control modules, complete module classification, strong signal processing pertinence, and can be flexibly configured according to the signal type. The modules and the main control module communicate through a high-speed communication bus, and the maximum communication rate can reach 40 Gbps.

[0031] The modules of the MD4008 collector can collect, calculate, analyze and store the output signals of the relevant monitoring instruments / transducers and the output signals of the vibration sensors (such as ICP acceleration sensors, eddy current displacement sensors, etc.), photoelectric / eddy current / magnetoelectric / rotational speed sensors, etc. in real time, so as to monitor and diagnose the running state of the equipment. It can transmit the collected data to the upper computer or central server through various ways such as wired network (network port and optical port), 4G, etc., support custom protocol and MODBUS standard protocol, and can communicate with the data of other systems of the user.

[0032] The collector can complete more than 80% of the data processing load, greatly reducing the work load of the system server, and greatly improving the smoothness of the system operation.

[0033] The data sampling strategy executed by the collector is:

[0034] (1) Collect 1 set of dynamic data for every 10 sets of static data. The sampling frequency is generally 32x rotational speed frequency, a total of 32 rotations, 1024 points of data are obtained, the frequency spectrum resolution is 400 lines, and the base frequency resolution is 1 / 32x rotational speed frequency;

[0035] The default setting of dynamic data collection is: 128 points / rotationx16 rotations, i.e.

[0036] ①The highest analysis frequency is 50x rotational frequency;

[0037] ②The frequency spectrum resolution is 400 lines;

[0038] ③The greenish resolution is 1 / 16x rotational frequency.

[0039] ⑵During the start-stop process, 320 groups of static data and 32 groups of dynamic data are collected at the highest rotational speed, and the lowest collection rotational speed is not less than 200 rpm. Among them, rpm represents the rotational speed per minute, and the highest rotational speed is the rated highest continuous working rotational speed of the unit. During the start-stop process, 8 cycles of dynamic data are collected, and the sampling frequency is increased to 128x rotational frequency;

[0040] ⑶When an alarm event occurs, 50 groups of dynamic data before and after the first alarm event and 500 groups of static data are collected and "frozen" in the "first-in, first-out" data "black box". The black box data need to be specially marked. The black box mode is a sampling frequency of 32x rotational frequency and a sampling length of 32 cycles;

[0041] ⑷At about 200 rpm, 1 group of data is collected as "initial deviation" data for storage, which can be used for compensation in subsequent signal analysis. The initial deviation data collection method is exactly the same as the steady-state data collection method (using synchronous whole-cycle collection, the same sampling frequency and sampling length);

[0042] ⑸When a new unit is tested or tested after maintenance, 1 group of dynamic data (including 10 groups of static data) is stored as "initial state" data for subsequent comparative analysis;

[0043] ⑹The capacity of the edge side embedded database can be set according to actual needs. Generally, if 8 channels of input are used, the embedded database memory capacity should be not less than 16 GB (actual calculation is subject to actual calculation);

[0044] ⑺The speed-up curve of the unit and the performance curve of the compressor can be stored in the database as reference data for monitoring during the start-stop process and unit load adjustment.

[0045] Step S102, multi-source feature extraction is performed on the original data obtained after data preprocessing, and feature fusion is performed on the extracted multi-source features.

[0046] Specifically, in the present embodiment, after the original data is acquired, the data is subjected to three-level data processing and cleaning, the purpose being to clean noise (for vibration, temperature, etc. data, noise is removed by a low-pass, high-pass filter to retain useful signals), fill in missing values (missing sensor data is filled in using an interpolation method, such as linear interpolation, Lagrange interpolation, etc.), normalize different data scales (due to differences in data dimensions and units, the data needs to be normalized or standardized to make the scales uniform, facilitating model learning), etc.

[0047] The collection module in the present embodiment implements first-level data processing and collation, the main control module implements second-level data processing and cleaning, and the server implements third-level output processing and cleaning.

[0048] Features that are helpful for fault diagnosis are extracted from the original data through a feature extraction strategy, and these features will serve as input features for the intelligent fault diagnosis model, providing necessary information for subsequent modeling.

[0049] In some embodiments, mechanical state features, driving state features, process state features, and external factor features are respectively extracted according to a preset feature extraction strategy;

[0050] The multiple-source features extracted are subjected to weighted feature fusion through an attention mechanism to dynamically learn the importance of each type of feature for fault prediction.

[0051] Specifically, in the present embodiment, for the mechanical state features, the present embodiment extracts by analyzing local trend changes in device vibration or operation data, and captures the evolving trend of the device state based on a local mean change rate.

[0052] Suppose the device vibration signal is Local trend features are extracted through the following steps:

[0053] S201, calculate the local mean: calculate the local mean of the signal in a sliding window:

[0054]

[0055] In the formula, denotes the window width, denotes the local mean at position .

[0056] S202, calculate the local mean change rate from the local mean, which can reflect the state change of the mechanical device:

[0057]

[0058] In the formula, denotes the time interval.

[0059] For the driving state feature, the driving state feature directly reflects the influence of the driving system on the mechanical equipment. By analyzing the interaction between the driving signal and the equipment response, the driving state can be captured more accurately. The present embodiment quantifies the relationship between the driving signal and the equipment response signal based on mutual information.

[0060] Suppose the driving signal is and the equipment response signal is , the mutual information is calculated as follows:

[0061]

[0062] wherein, is the entropy of the driving signal , indicating the uncertainty of the signal; is the entropy of the response signal ; and is the joint entropy of the driving signal and the response signal .

[0063] By calculating , the influence of the driving signal on the equipment state can be quantified, thereby extracting the driving feature. If the mutual information is large, it indicates that the driving signal and the equipment response have strong correlation, which may indicate an abnormal state.

[0064] For the process state feature, the process state is usually determined by factors such as operating conditions and processing environment. In order to effectively capture the process state, the present embodiment analyzes the influence of process parameters on equipment performance by constructing a dynamic constraint model, and combines a machine learning model to identify features under different process states.

[0065] Suppose the process parameter is , which affects the equipment output . A dynamic constraint model (such as a state space model) is constructed to extract features.

[0066] wherein, the dynamic constraint equation is as follows:

[0067]

[0068] In the formula, is the model parameter, indicating the influence of the process condition on the equipment.

[0069] Through historical data, the model parameters are updated using the recursive least squares method, and the process state feature is calculated in real-time monitoring:

[0070]

[0071] The real-time features of the process state are obtained through changes in the model parameters.

[0072] For the environmental factor features, the embodiment is based on spatial-temporal correlation analysis to extract external factor features. By analyzing the time variation and spatial distribution of the external environment, the influence of the external factors on the equipment state can be obtained.

[0073] Suppose the external factors are Each factor varies with time, and spatial-temporal cross-correlation is used to measure the relationship between the external factors and the equipment state:

[0074]

[0075] In the formula, denotes the cross-correlation number of the external factors and the equipment state, and denote the mean values of the external factors and the equipment state, respectively.

[0076] By analyzing the spatial-temporal correlation in different time windows, the influence features of the external factors on the equipment state are extracted.

[0077] According to the features obtained above, an attention network is constructed to learn the importance of each feature vector (specifically through an attention score function). According to the obtained attention weight, all features are weighted and summed to obtain the fused features.

[0078] Step S103, according to the pre-constructed intelligent fault diagnosis model, the running state of the steam turbine generator unit is monitored and fault diagnosed, wherein the intelligent fault diagnosis model is a deep learning neural network model trained based on the fused multi-source feature data.

[0079] In some embodiments, a data set is constructed based on the fused multi-source feature vector, and the first network model, the second network model and the third network model are trained through the data set respectively, to obtain the class prediction probability corresponding to each model;

[0080] A plurality of weighting coefficients are defined, and weighted voting fusion is performed according to the class prediction probability. According to the weighted probability, a final classification decision is made, and the fault class and confidence of the steam turbine generator unit are output.

[0081] Specifically, in the present embodiment, a data set is constructed based on the fused multi-source feature vector, the data set is randomly divided into a training set, a test set and a validation set, and a first network model (i.e., a CNN model), a second network model (i.e., an LSTM model) and a third network model (i.e., an XGBoost) are trained respectively to obtain the prediction probability corresponding to each model. After obtaining the prediction probability of each model, the final prediction probability is calculated by weighted summation according to the weighting coefficients, and the final classification decision is made (in the present embodiment, the class with the maximum probability is taken as the final prediction result), and thus the fault class and the confidence of the turbogenerator unit are obtained. The specific formula is as follows:

[0082]

[0083] wherein, , , the weighting coefficients satisfy the condition + + =1, , , are the prediction probabilities of the CNN, LSTM and XGoost models for the class .

[0084] In some embodiments, the life prediction of the turbogenerator unit is further performed by a Weibull distribution algorithm, including:

[0085] collecting historical operation data of the equipment, fitting a Weibull distribution by a maximum likelihood estimation method to estimate the parameters of the Weibull distribution, and predicting the remaining life of the equipment by the estimated Weibull distribution parameters;

[0086] setting a trigger threshold according to a preset value of the remaining life, and performing equipment fault monitoring and early warning according to the trigger threshold.

[0087] Specifically, in the present embodiment, the historical operation data of the equipment, in particular the time of fault occurrence, are collected, the parameters of the Weibull distribution are estimated by a maximum likelihood estimation method (MLE), the remaining life of the equipment is predicted by the estimated Weibull distribution parameters, and a trigger threshold is set for the equipment according to the predicted remaining life.

[0088] For example, it is assumed that the equipment has been running for 200 hours, and the remaining life is predicted, for example, it is set that the equipment needs to be repaired when the remaining life is less than 10%.

[0089] wherein, the probability density function of the Weibull distribution is specifically as follows:

[0090] ;

[0091] wherein, represents the life of the device; represents a size parameter, controlling the width of the distribution, determining the scale of the life of the device; represents a shape parameter, controlling the failure mode of the device. Specifically:

[0092] When <1, the failure rate of the device decreases with time, which is manifested as early failure;

[0093] When =1, the failure rate of the device is constant, which is consistent with the exponential distribution, which is manifested as random failure;

[0094] When >1, the failure rate of the device increases with time, which is manifested as aging failure.

[0095] In this embodiment, the intelligent fault diagnosis and safety evaluation of the main components of the steam turbine are important contents of the smart power plant. The Weibull distribution algorithm is used to predict the reliability of the equipment, which is an important means to solve the quantization of the prediction result. The prediction result can be used as part of the trigger threshold of the equipment management system, so as to achieve closed-loop control from monitoring to maintenance.

[0096] The embodiment can automatically diagnose the following faults: rotor imbalance, initial imbalance, sudden imbalance, thermal bending imbalance, rotor bending, large radial load, misalignment, friction, local rubbing, full rubbing, severe friction, looseness, mechanical fastening looseness, low rigidity of the machine base, fluid excitation instability, steam turbine water hammer, oil film vortex and oil film oscillation, shaft crack, and other faults.

[0097] In some embodiments, the multi-channel data acquisition system performs a patrol on the sensor state of all input signals, and in response to determining that the sensor working state is abnormal, the system no longer acquires invalid data and only transmits "sensor state abnormal" information to the system.

[0098] Specifically, in this embodiment, the data authenticity detection function system performs a patrol on the sensor state of all input signals. When it is determined that the sensor working state is abnormal, the collector no longer acquires invalid data and only transmits "sensor state abnormal" information to the system, ensuring the authenticity of the collected data.

[0099] The intelligent fault diagnosis method of the steam turbine generator unit described in the embodiment realizes high-precision detection of multiple-component and multiple-type coupling faults of the unit by collecting multi-source feature data, based on full-state parameter monitoring of the monitored object, correlation feature extraction between different measuring points and different monitoring quantities, and based on the constructed intelligent fault diagnosis model. Finally, closed-loop control from monitoring to maintenance is realized, and the reliability and maintenance efficiency of the equipment are improved.

[0100] It should be noted that the above describes some embodiments of the present application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in an order different than that described in the above embodiments and still achieve the desired result. In addition, the processes depicted in the accompanying drawings do not necessarily require the particular order shown or sequential order to achieve the desired results. In certain implementations, multitasking and parallel processing are also possible or advantageous.

[0101] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the embodiments of the present application also provide an intelligent fault diagnosis system for a steam turbine generator unit.

[0102] As shown in Figure 2 , the intelligent fault diagnosis system for a steam turbine generator unit comprises:

[0103] The data acquisition module 11 is configured to acquire the running state data and external factor data of the steam turbine generator unit in real time through a multi-channel data acquisition system; wherein the running state data includes mechanical state data, driving state data and process state data of the main engine and auxiliary engine;

[0104] The feature extraction module 12 is configured to perform multi-source feature extraction on the original data obtained after data preprocessing, and perform feature fusion on the extracted multi-source features.

[0105] The fault diagnosis module 13 is configured to monitor and diagnose the running state of the steam turbine generator unit according to the pre-constructed intelligent fault diagnosis model, wherein the intelligent fault diagnosis model is a deep learning neural network model trained based on the fused multi-source feature data.

[0106] The embodiment establishes a comprehensive intelligent platform integrating a full-range unit state real-time online monitoring system, a real-time intelligent diagnosis system and a management system of a main unit and related auxiliary units of a steam turbine generator unit, and realizes intelligent operation and maintenance of the main unit and the auxiliary units. The system adopts a hardware platform based on distribution, modularization and edge intelligence, and a software based on an industrial big data data platform. Meanwhile, the absolute safety of a field TSI monitoring system and the high safety layout of a field network system are taken into account. The big data platform can realize intelligent prediction of unit operation data and faults, intelligent alarm and other functions. Meanwhile, the system can provide comprehensive conventional monitoring and analysis tools for special technical personnel.

[0107] For the convenience of description, the above system is described in various modules in terms of functions. Of course, the functions of the modules can be implemented in one or more software and / or hardware when implementing the embodiments of the present application.

[0108] The system of the above embodiments is used to implement the corresponding method in any of the foregoing embodiments, and has the beneficial effects of the corresponding method embodiments, which are not described here again.

[0109] Based on the same inventive concept, corresponding to any of the above method embodiments, the embodiments of the present application also provide an electronic device, including a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method of any one of the above embodiments.

[0110] Figure 3 A more specific hardware structure schematic diagram of an electronic device provided by the embodiment is shown, which can include a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040 and a bus 1050. The processor 1010, the memory 1020, the input / output interface 1030 and the communication interface 1040 are connected to each other through the bus 1050 for communication within the device.

[0111] The processor 1010 can be implemented in the form of a general-purpose CPU (Central Processing Unit, central processor), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC) or one or more integrated circuits, etc., for executing related programs to implement the technical solutions provided by the embodiments of the present application.

[0112] The memory 1020 can be implemented in the form of a ROM (Read Only Memory), a RAM (Random Access Memory), a static storage device, a dynamic storage device, etc. The memory 1020 can store an operating system and other application programs, and when the technical solutions provided in the embodiments of the present specification are implemented by software or firmware, the related program codes are stored in the memory 1020 and are called and executed by the processor 1010.

[0113] The input / output interface 1030 is configured to connect an input / output module to realize information input and output. The input / output module can be configured as a component in the device (not shown in the figure) or can be externally connected to the device to provide corresponding functions. The input device can include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output device can include a display, a speaker, a vibrator, an indicator light, etc.

[0114] The communication interface 1040 is configured to connect a communication module (not shown in the figure) to realize communication interaction between the device and other devices. The communication module can realize communication through a wired manner (such as a USB, a network cable, etc.) or through a wireless manner (such as a mobile network, WIFI, Bluetooth, etc.).

[0115] The bus 1050 includes a channel for transmitting information between various components (such as the processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040) of the device.

[0116] It should be noted that although the above device only shows the processor 1010, the memory 1020, the input / output interface 1030, the communication interface 1040, and the bus 1050, in the specific implementation process, the device can also include other components necessary for normal operation. In addition, those skilled in the art can understand that the above device can also only include components necessary for implementing the embodiments of the present specification, and does not have to include all the components shown in the figure.

[0117] The electronic device of the above embodiments is used to implement the corresponding method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiments, which are not described here again.

[0118] Based on the same inventive concept, corresponding to any of the above method embodiments, the present application also provides a non-transitory computer readable storage medium storing computer instructions for causing the computer to execute the method of any of the above embodiments.

[0119] The computer readable media of the present embodiments includes permanent and non-permanent, removable and non-removable media, which can be implemented by any method or technology to store information. The information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette, magnetic tape disk storage or other magnetic storage device, or any other non-transmission medium that can be used to store information accessible by a computing device.

[0120] The storage medium of the above embodiments stores computer instructions for causing the computer to perform the method as described in any of the above embodiments, and has the beneficial effects of the corresponding method embodiments, which are not described here again.

[0121] Those skilled in the art should understand that the discussion of any of the above embodiments is only exemplary, and is not intended to imply that the scope (including claims) of the present application is limited to these examples; the above embodiments or technical features between different embodiments can also be combined, the steps can be implemented in any order, and there are many other changes of the different aspects of the embodiments of the present application as described above, which are not provided in detail for the sake of brevity.

[0122] In addition, in order to simplify the description and discussion, and so as not to make the embodiments of the present application difficult to understand, the well-known power / ground connections of integrated circuit (IC) chips and other components can or can not be shown in the provided drawings. In addition, the devices can be shown in the form of block diagrams in order to avoid making the embodiments of the present application difficult to understand, and this also takes into account the fact that the details of the implementation of these block diagram devices are highly dependent on the platform to be implemented the embodiments of the present application (i.e. these details should be fully within the understanding of those skilled in the art). Where specific details (e.g. circuits) are set forth in order to describe the exemplary embodiments of the present application, it will be apparent to those skilled in the art that the embodiments of the present application can be implemented without these specific details or with variations of these specific details. Therefore, these descriptions should be considered illustrative rather than limiting.

[0123] While the present application has been described in connection with certain embodiments thereof, many modifications, substitutions, changes, and of forms will be apparent to those of ordinary skill in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) can use the embodiments discussed.

[0124] Embodiments of the present application are intended to cover all such alterations, modifications, and variations as they can come within the scope of the appended claims. Accordingly, although specific embodiments have been furthered in connection with the present application, any omission, substitution, or change, in principle and in form, made to the present application should be included in the scope of the present application.

Claims

1. A method for intelligent fault diagnosis of a steam turbine generator set, characterized in that: include: Acquire the operating status data and external factor data of the steam turbine generator set in real time through a multi-channel data acquisition system; wherein the operating status data includes the mechanical status data, drive status data and process status data of the main engine and auxiliary engine; Perform multi-source feature extraction on the raw data obtained after data preprocessing, and perform feature fusion on the extracted multi-source features, including: Mechanical status features are extracted by analyzing local trend changes in equipment vibration or operation data, and the evolution trend of equipment status is captured based on the local mean change rate; Assume that the device vibration signal is , extract local trend features through the following steps: S201. Calculate the local mean: Calculate the local mean of the signal within a sliding window: ； Where, Indicates the window width, Indicates location The local mean at ; S202. Calculate the local mean change rate based on the local mean. The local mean change rate at this time reflects the state change of the mechanical equipment: ； Where, Indicates a time interval; The driving state characteristics directly reflect the impact of the driving system on the mechanical equipment. The driving state is accurately captured by analyzing the interactive relationship between the driving signal and the equipment response, and the relationship between the driving signal and the equipment response signal is quantified based on the mutual information. Assume that the driving signal is And the device response signal is , mutual information The calculation formula is as follows: ； in, Is the driving signal The entropy of represents the uncertainty of the signal; In response to the signal Entropy; Is the driving signal and response signal The joint entropy of By calculation , quantify the impact of the driving signal on the device state, and thus extract the driving characteristics; As for process state characteristics, the process state is determined by operating conditions and processing environment factors. By building a dynamic constraint model, analyzing the impact of process parameters on equipment performance, and combining it with a machine learning model, we can identify characteristics under different process states. Assume that the process parameters are , which affects the device output ,features are extracted by building a dynamic constraint model; The dynamic constraint equation is as follows: ； Where, is the model parameter, which represents the influence of process conditions on the equipment; Update model parameters using recursive least squares method based on historical data , and calculate process status characteristics in real-time monitoring: ； Obtain real-time characteristics of process status through changes in model parameters; For environmental factor characteristics, external factor characteristics are extracted based on space-time correlation analysis. By analyzing the temporal changes and spatial distribution of the external environment, the impact of external factors on the equipment status is obtained; Assume that the external factors are , each factor changes over time, and space-time cross-correlation is used to measure the relationship between external factors and device status: ； Where, Represents the cross-correlation coefficient between external factors and device status, and represent the means of external factors and equipment status respectively; By analyzing the spatial-temporal correlation within different time windows, the impact characteristics of external factors on equipment status are extracted; Based on the obtained features, an attention network is constructed to learn the importance of each feature vector. According to the obtained attention weight, all features are weighted summed to obtain the fusion feature. The operating status of the steam turbine generator set is monitored and faults are diagnosed based on a pre-built intelligent fault diagnosis model, wherein the intelligent fault diagnosis model is a deep learning neural network model trained based on the fused multi-source feature data, including: A data set is constructed based on the fused multi-source feature vectors. The data set is randomly divided into a training set, a test set, and a validation set. The first network model, the second network model, and the third network model are trained separately to obtain the prediction probability corresponding to each model. After obtaining the prediction probability of each model, a weighted sum is performed according to the weighting coefficient to calculate the final prediction probability for the final classification decision, thereby obtaining the fault category and confidence of the steam turbine generator set. The specific formula is as follows: ； in, 、 、 Represents the weighting coefficient, satisfying the condition + + =1, 、 、 CNN, LSTM and XGoost models for categories The predicted probability of .

2. The method according to claim 1, wherein: The mechanical state data includes at least mechanical vibration, noise, axial displacement, bearing temperature, eccentricity and thermal expansion; The process status data at least includes flow rate, pressure and valve position; The driving state data at least includes a driving medium state and a driving output state; The external factor data at least includes pipeline network impact, lubrication status and environmental factors.

3. The method according to claim 1, wherein: The multi-channel data acquisition system inspects the sensor status of all input signals. In response to determining that the sensor working status is abnormal, the system no longer collects invalid data and only transmits the "sensor status abnormality" information to the system.

4. The method according to claim 1, wherein It also includes the life prediction of steam turbine generator equipment using the Weibull distribution algorithm, including: Collect historical operating data of the equipment and fit the Weibull distribution using the maximum likelihood estimation method to estimate the parameters of the Weibull distribution. Use the estimated Weibull distribution parameters to predict the remaining life of the equipment. A trigger threshold is set according to a preset value of the remaining life, and equipment failure monitoring and early warning are performed according to the trigger threshold.

5. The method according to claim 4, characterized in that The probability density function of the Weibull distribution is as follows: ； Where, Indicates the life of the equipment. Represents the size parameter, Represents the shape parameter.

6. An intelligent fault diagnosis system for a steam turbine generator set, characterized in that: include: a data acquisition module configured to acquire, in real time, operating status data and external factor data of the steam turbine generator set through a multi-channel data acquisition system; wherein the operating status data includes mechanical status data, drive status data, and process status data of the main engine and auxiliary engines; The feature extraction module is configured to extract multi-source features from the raw data obtained after data preprocessing and perform feature fusion on the extracted multi-source features, including: Mechanical status features are extracted by analyzing local trend changes in equipment vibration or operation data, and the evolution trend of equipment status is captured based on the local mean change rate; Assume that the device vibration signal is , extract local trend features through the following steps: S201. Calculate the local mean: Calculate the local mean of the signal within a sliding window: ； Where, Indicates the window width, Indicates location The local mean at ; S202. Calculate the local mean change rate based on the local mean. The local mean change rate at this time reflects the state change of the mechanical equipment: ； Where, Indicates a time interval; The driving state characteristics directly reflect the impact of the driving system on the mechanical equipment. The driving state is accurately captured by analyzing the interactive relationship between the driving signal and the equipment response, and the relationship between the driving signal and the equipment response signal is quantified based on the mutual information. Assume that the driving signal is And the device response signal is , mutual information The calculation formula is as follows: ； in, Is the driving signal The entropy of represents the uncertainty of the signal; In response to the signal Entropy; Is the driving signal and response signal The joint entropy of By calculation , quantify the impact of the driving signal on the device state, and thus extract the driving characteristics; As for process state characteristics, the process state is determined by operating conditions and processing environment factors. By building a dynamic constraint model, analyzing the impact of process parameters on equipment performance, and combining it with a machine learning model, we can identify characteristics under different process states. Assume that the process parameters are , which affects the device output ,features are extracted by building a dynamic constraint model; The dynamic constraint equation is as follows: ； Where, is the model parameter, which represents the influence of process conditions on the equipment; Update model parameters using recursive least squares method based on historical data , and calculate process status characteristics in real-time monitoring: ； Obtain real-time characteristics of process status through changes in model parameters; For environmental factor characteristics, external factor characteristics are extracted based on space-time correlation analysis. By analyzing the temporal changes and spatial distribution of the external environment, the impact of external factors on the equipment status is obtained; Assume that the external factors are , each factor changes over time, and space-time cross-correlation is used to measure the relationship between external factors and device status: ； Where, Represents the cross-correlation coefficient between external factors and device status, and represent the means of external factors and equipment status respectively; By analyzing the spatial-temporal correlation within different time windows, the impact characteristics of external factors on equipment status are extracted; Based on the obtained features, an attention network is constructed to learn the importance of each feature vector. According to the obtained attention weight, all features are weighted summed to obtain the fusion feature. The operating status of the steam turbine generator set is monitored and faults are diagnosed based on a pre-built intelligent fault diagnosis model, wherein the intelligent fault diagnosis model is a deep learning neural network model trained based on the fused multi-source feature data, including: A data set is constructed based on the fused multi-source feature vectors. The data set is randomly divided into a training set, a test set, and a validation set. The first network model, the second network model, and the third network model are trained separately to obtain the prediction probability corresponding to each model. After obtaining the prediction probability of each model, a weighted sum is performed according to the weighting coefficient to calculate the final prediction probability for the final classification decision, thereby obtaining the fault category and confidence of the steam turbine generator set. The specific formula is as follows: ； in, 、 、 Represents the weighting coefficient, satisfying the condition + + =1, 、 、 CNN, LSTM and XGoost models for categories The predicted probability of The fault diagnosis module is configured to monitor the operating status and diagnose faults of the steam turbine generator set based on a pre-built intelligent fault diagnosis model, wherein the intelligent fault diagnosis model is a deep learning neural network model trained based on the fused multi-source feature data.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method according to any one of claims 1 to 5 when executing the program.

8. A non-transitory computer-readable storage medium, characterized in that in, The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the method according to any one of claims 1 to 5.

Images