Method, device and equipment for predicting failure of through-flow part of steam turbine and storage medium

Abstract

This application discloses a method, apparatus, equipment, and storage medium for fault prediction of the flow passage of a steam turbine, relating to the field of steam turbine fault detection technology, which can improve the accuracy of the determined fault prediction results. The method includes: acquiring a target operating data set of the steam turbine and the target actual extraction point temperature of the high-pressure cylinder of the steam turbine; the target operating data set includes real-time operating data corresponding to the operating parameters of each candidate unit of the steam turbine; inputting the target operating data set into an extraction point temperature prediction model to obtain the target predicted extraction point temperature of the high-pressure cylinder output by the extraction point temperature prediction model; the extraction point temperature prediction model is pre-trained based on historical operating data sets arranged in chronological order and historical actual extraction point temperatures corresponding to each historical operating data set; and determining the fault prediction result for the flow passage of the steam turbine based on the target deviation value between the target actual extraction point temperature and the target predicted extraction point temperature.

Description

Technical Field

[0001] This application relates to the field of steam turbine fault detection technology, and in particular to a method, apparatus, equipment and storage medium for predicting faults in the flow passage of a steam turbine. Background Technology

[0002] A steam turbine is a device that converts the thermal energy of fuel into mechanical energy and can be widely used in various fields (such as thermal power generation). Among them, the flow path of the steam turbine is a particularly important component, and its operating status can directly affect the operating efficiency and performance of the steam turbine. Therefore, fault prediction of the flow path of the steam turbine is particularly important.

[0003] In existing technologies, the thermodynamic characteristic parameters of the steam turbine (such as the real-time power of the steam turbine unit) can generally be used as fault analysis indicators to predict whether a fault has occurred in the flow passage of the steam turbine.

[0004] However, since steam turbines operate in real time and the operation of steam turbine units has inherent dynamic characteristics, the thermodynamic characteristic parameters that are generally determined will have large temporal differences (that is, the thermodynamic characteristic parameters determined at a certain moment will have a large error compared with the actual thermodynamic characteristic parameters at that moment). As a result, when predicting faults in the steam turbine's flow passage, the collected thermodynamic characteristic parameters cannot characterize the real-time operating status of the steam turbine's flow passage, thus affecting the fault prediction results for the steam turbine's flow passage. Summary of the Invention

[0005] This application provides a method, apparatus, equipment, and storage medium for predicting faults in the flow path of a steam turbine, which can improve the accuracy of the determined fault prediction results.

[0006] To achieve the above objectives, this application adopts the following technical solution:

[0007] In a first aspect, this application provides a fault prediction method for the flow path of a steam turbine, comprising: acquiring a target operating data set of the steam turbine and a target actual extraction point temperature of the high-pressure cylinder of the steam turbine; wherein, the target operating data set includes real-time operating data corresponding to the operating parameters of each candidate unit of the steam turbine; inputting the target operating data set into an extraction point temperature prediction model to obtain the target predicted extraction point temperature of the high-pressure cylinder output by the extraction point temperature prediction model; the extraction point temperature prediction model is pre-trained based on each historical operating data set arranged in chronological order and the historical actual extraction point temperatures corresponding to each historical operating data set; and determining the fault prediction result for the flow path of the steam turbine based on the target deviation value between the target actual extraction point temperature and the target predicted extraction point temperature.

[0008] In the technical solution provided in this application, when predicting whether a fault has occurred in the flow passage of a steam turbine, the operating data set of the steam turbine at this time (i.e., the target operating data set in this application) and the actual extraction point temperature of the high-pressure cylinder of the steam turbine at this time (i.e., the target actual extraction point temperature in this application) can be obtained first. Then, the target operating data set can be input into a pre-trained extraction point temperature prediction model to obtain the predicted extraction point temperature of the high-pressure cylinder at this time (i.e., the target predicted extraction point temperature in this application) output by the extraction point temperature prediction model. Afterwards, the fault prediction result for the flow passage of the steam turbine can be determined based on the deviation value between the target actual extraction point temperature and the target predicted extraction point temperature (i.e., the target deviation value in this application). Because steam turbine units have inherent dynamic characteristics under normal operating conditions, that is, under normal operating conditions, the extraction point temperature of the high-pressure cylinder at a certain moment is closely related to the extraction point temperature of the high-pressure cylinder before that moment; and the acquired target operating data set includes real-time operating data corresponding to the operating parameters of each candidate unit of the steam turbine, and the operating parameters of each candidate unit can be the unit operating parameters related to the extraction point temperature of the high-pressure cylinder (that is, the magnitude of the operating parameters of each candidate unit will affect the extraction point temperature of the high-pressure cylinder). Based on this, when predicting whether a fault has occurred in the flow path of the steam turbine, this application can use the extraction point temperature prediction model based on the target operating data set to predict the extraction point temperature of the high-pressure cylinder under normal operating conditions (i.e., determine the target predicted extraction point temperature in this application). Since the extraction point temperature prediction model is pre-trained based on historical operating data sets arranged in chronological order and the historical actual extraction point temperatures corresponding to each historical operating data set, the target predicted extraction point temperature can be determined relatively accurately through the extraction point temperature prediction model. Therefore, when the difference between the target predicted extraction point temperature and the target actual extraction point temperature is large, it indicates that a fault may have occurred in the flow path of the steam turbine.

[0009] Furthermore, since the extraction port of the high-pressure cylinder is generally located very close to the flow path of the turbine, the error between the collected extraction point temperature of the high-pressure cylinder at a certain moment (i.e., the actual extraction point temperature) and the true extraction point temperature of the high-pressure cylinder at that moment is very small. In other words, the temporal difference of this parameter, the extraction point temperature of the high-pressure cylinder, is very small. Therefore, compared to using thermodynamic characteristic parameters as fault analysis indicators, this application uses the extraction point temperature of the high-pressure cylinder as a fault analysis indicator, which can reduce the impact of temporal differences on the fault prediction results. It can be seen that this application uses the extraction point temperature of the high-pressure cylinder, which has a smaller temporal difference, as a fault analysis indicator for the flow path of the turbine. By determining the fault prediction result for the flow path of the turbine through the deviation between the actual and predicted extraction point temperatures of the high-pressure cylinder, the accuracy of the determined fault prediction result can be improved.

[0010] Secondly, this application provides a fault prediction device for the flow path of a steam turbine, the device comprising:

[0011] The acquisition module is used to acquire the target operating data set of the steam turbine and the target actual extraction point temperature of the high-pressure cylinder of the steam turbine. The target operating data set includes real-time operating data corresponding to the operating parameters of each candidate unit of the steam turbine. The model processing module is used to input the target operating data set into the extraction point temperature prediction model to obtain the target predicted extraction point temperature of the high-pressure cylinder output by the extraction point temperature prediction model. The extraction point temperature prediction model is pre-trained based on each historical operating data set arranged in chronological order and the historical actual extraction point temperature corresponding to each historical operating data set. The prediction module is used to determine the fault prediction result for the flow passage of the steam turbine based on the target deviation value between the target actual extraction point temperature and the target predicted extraction point temperature.

[0012] Thirdly, this application provides a fault prediction device for the flow path of a steam turbine, including a memory, a processor, a bus, and a communication interface; the memory is used to store computer execution instructions, and the processor is connected to the memory via the bus; when the fault prediction device for the flow path of the steam turbine is running, the processor executes the computer execution instructions stored in the memory, so that the fault prediction device for the flow path of the steam turbine performs the fault prediction method for the flow path of the steam turbine as provided in the first aspect above.

[0013] Fourthly, this application provides a computer-readable storage medium storing instructions that, when executed by a computer, cause the computer to perform the fault prediction method for the steam turbine flow path provided in the first aspect.

[0014] Fifthly, this application provides a computer program product including computer instructions that, when executed on a computer, cause the computer to perform the fault prediction method for the steam turbine flow path provided in the first aspect.

[0015] It should be noted that the aforementioned computer instructions may be stored, in whole or in part, on a computer-readable storage medium. This computer-readable storage medium may be packaged together with the processor of the fault prediction device for the turbine's flow path, or it may be packaged separately from the processor of the fault prediction device for the turbine's flow path; this application does not impose any limitations on this.

[0016] The descriptions of the second, third, fourth, and fifth aspects in this application can be referenced to the detailed description of the first aspect; and the beneficial effects of the descriptions of the second, third, fourth, and fifth aspects can be referenced to the analysis of the beneficial effects of the first aspect, which will not be repeated here.

[0017] In this application, the names of the aforementioned devices or functional modules are not limited, and in actual implementation, these devices or functional modules may appear under other names. As long as the functions of each device or functional module are similar to those in this application, they all fall within the scope of this application and its equivalents.

[0018] These or other aspects of this application will become more readily apparent in the following description. Attached Figure Description

[0019] Figure 1 A flowchart illustrating a method for predicting faults in the flow path of a steam turbine, provided in an embodiment of this application;

[0020] Figure 2 A schematic diagram of the structure of a first GRU subnetwork provided in an embodiment of this application;

[0021] Figure 3 A flowchart illustrating another method for predicting faults in the flow path of a steam turbine, provided in an embodiment of this application;

[0022] Figure 4 A flowchart illustrating another method for predicting faults in the flow path of a steam turbine, provided in an embodiment of this application;

[0023] Figure 5 A schematic diagram of the structure of a fault prediction device for the flow passage of a steam turbine provided in an embodiment of this application;

[0024] Figure 6 This is a schematic diagram of the structure of a fault prediction device for the flow passage of a steam turbine, provided in an embodiment of this application. Detailed Implementation

[0025] The following description, in conjunction with the accompanying drawings, details the fault prediction method, apparatus, equipment, and storage medium for the steam turbine flow path provided in the embodiments of this application.

[0026] In this article, the term "and / or" is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone.

[0027] The terms "first" and "second," etc., used in the specification and drawings of this application are used to distinguish different objects or to distinguish different treatments of the same object, rather than to describe a specific order of objects.

[0028] Furthermore, the terms "comprising" and "having," and any variations thereof, used in the description of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the steps or units listed, but may optionally include other steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus.

[0029] It should be noted that in the embodiments of this application, the words "exemplary" or "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of the words "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0030] In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0031] Furthermore, the acquisition, storage, use, and processing of data in this application's technical solution all comply with relevant national laws and regulations.

[0032] In existing technologies, the thermodynamic characteristic parameters of a steam turbine (such as the real-time power of the turbine unit) are generally used as fault analysis indicators to predict whether a fault has occurred in the flow path of the steam turbine. However, because the steam turbine operates in real time and the turbine unit has inherent dynamic characteristics, the thermodynamic characteristic parameters that are generally determined will have significant temporal differences (that is, the thermodynamic characteristic parameters determined at a certain moment will have a large error compared to the actual thermodynamic characteristic parameters at that moment). As a result, when predicting faults in the flow path of the steam turbine, the collected thermodynamic characteristic parameters cannot characterize the real-time operating state of the flow path, thus affecting the fault prediction results for the flow path.

[0033] To address the problems existing in the prior art, this application provides a fault prediction method for the flow passage of a steam turbine. This method uses the extraction point temperature of the high-pressure cylinder with small time differences as the fault analysis index of the flow passage of the steam turbine. The fault prediction result for the flow passage of the steam turbine is determined by the deviation between the actual extraction point temperature of the high-pressure cylinder and the predicted extraction point temperature, which can improve the accuracy of the determined fault prediction result.

[0034] The fault prediction method for the turbine flow path provided in this application embodiment can be executed by the fault prediction device for the turbine flow path provided in this application embodiment. This device can be implemented through software and / or hardware and integrated into the fault prediction equipment for the turbine flow path that executes this method. Taking the thermal power generation field as an example, the fault prediction equipment for the turbine flow path can be the main control server of the thermal power generation system, or it can be the control server corresponding to the turbine.

[0035] The fault prediction method for the steam turbine flow path provided in this application will be described below with reference to the accompanying drawings.

[0036] Reference Figure 1 The fault prediction method for the steam turbine flow path provided in this application embodiment may include S101-S103:

[0037] S101. Obtain the target operating data set of the steam turbine and the target actual extraction point temperature of the high-pressure cylinder of the steam turbine.

[0038] The target operating data set includes real-time operating data collected at the current moment, corresponding to the operating parameters of each candidate turbine unit. Each candidate turbine unit's operating parameter can be one of multiple operating parameters related to the extraction steam temperature of the high-pressure cylinder; that is, at the same time, the magnitude of each candidate turbine unit's operating parameter will affect the extraction steam temperature of the high-pressure cylinder. The target actual extraction steam temperature can be the extraction steam temperature of the high-pressure cylinder collected at the current moment. For example, if the high-pressure cylinder has one extraction port, the extraction steam temperature can be the temperature collected at that extraction port; if the high-pressure cylinder has multiple extraction ports, the extraction steam temperature can be the temperature collected at the first extraction port of the high-pressure cylinder (generally represented by "1#").

[0039] In one possible implementation, the faulty equipment in the turbine's flow path can be communicatively connected to multiple sensors (including temperature sensors, pressure sensors, etc.) to acquire sensor data collected by each sensor. These sensors include multiple sensors used to collect the actual temperature of the target extraction point, as well as real-time operating data corresponding to the operating parameters of each candidate turbine unit.

[0040] Optionally, since at the same time, the unit load, high-pressure cylinder exhaust temperature, high-pressure cylinder exhaust pressure, regulating stage temperature, regulating stage pressure, high-pressure cylinder extraction point pressure, high-pressure heater inlet water temperature, high-pressure heater outlet water temperature, and high-pressure heater condensate temperature will all affect the high-pressure cylinder extraction point temperature, the operating parameters of each candidate unit may include at least one of the following: unit load, high-pressure cylinder exhaust temperature, high-pressure cylinder exhaust pressure, regulating stage temperature, regulating stage pressure, high-pressure cylinder extraction point pressure, high-pressure heater inlet water temperature, high-pressure heater outlet water temperature, and high-pressure heater condensate temperature.

[0041] In practical applications, the dimensions of the operating parameters of different candidate units are different. Therefore, in the embodiments of this application, the operating data in the obtained operating data group (historical operating data group or target operating data group) are all normalized data.

[0042] S102. Input the target operating data set into the extraction point temperature prediction model to obtain the target predicted extraction point temperature of the high-pressure cylinder output by the extraction point temperature prediction model.

[0043] The extraction steam point temperature prediction model is pre-trained based on historical operating data sets arranged in chronological order and the corresponding historical actual extraction steam point temperatures. Each historical operating data set can be historical operating data collected before the current time, based on a preset sampling interval (i.e., a predetermined sampling period), corresponding to the operating parameters of each candidate turbine unit. Each historical actual extraction steam point temperature can be the extraction steam point temperature of the high-pressure cylinder collected before the current time, based on a preset sampling interval. Each historical operating data set corresponds one-to-one with each historical actual extraction steam point temperature. For example, taking each historical operating data group as an example, if T represents the current time and T0 represents the preset sampling interval, then each historical operating data group includes historical operating data corresponding to the operating parameters of each candidate unit of the steam turbine, collected at T-3T0, T-2T0, and T-T0 respectively; correspondingly, each historical actual extraction steam point temperature includes the extraction steam point temperature of the high-pressure cylinder, collected at T-3T0, T-2T0, and T-T0 respectively.

[0044] In one possible implementation, the extraction point temperature prediction model can be trained as follows: Each historical operating data set is sequentially input into a pre-determined recurrent neural network (RNN) model according to time sequence, yielding the historical predicted extraction point temperature output by the RNN model corresponding to each historical operating data set. Then, the root mean square error (RMSE) between each historical actual extraction point temperature and its corresponding historical predicted extraction point temperature can be used as the objective function. The optimization direction is to minimize the objective function, thereby adjusting the RNN model parameters to obtain the extraction point temperature prediction model.

[0045] Optionally, during the training of the extraction point temperature prediction model, the Human Evolutionary Optimization Algorithm (HEOA) can be used to optimize the model parameters of the RNN model.

[0046] Optionally, the RNN model can be a Gated Recurrent Unit (GRU) model. Correspondingly, the extraction point temperature prediction model can include a first GRU subnetwork, a second GRU subnetwork, and a fully connected subnetwork. Inputting the target operating data set into the extraction point temperature prediction model to obtain the target predicted extraction point temperature of the high-pressure cylinder output by the model includes: inputting the target operating data set into the extraction point temperature prediction model; processing the target operating data set and the primary hidden state value corresponding to the previous operating data set through the first GRU subnetwork to obtain the primary hidden state value of the target operating data set; processing the primary hidden state value of the target operating data set and the secondary hidden state value corresponding to the previous operating data set through the second GRU subnetwork to obtain the secondary hidden state value of the target operating data set; and processing the secondary hidden state value of the target operating data set through the fully connected subnetwork to obtain the target predicted extraction point temperature.

[0047] In one possible implementation, the first GRU subnetwork includes one layer of GRUs, and the second GRU subnetwork has the same structure as the first GRU subnetwork, also including one layer of GRUs. The first GRU subnetwork, the second GRU subnetwork, and the fully connected subnetwork are connected sequentially. That is, the output of the first GRU subnetwork (i.e., the primary hidden state value in this embodiment) is used as the input of the second GRU subnetwork, and the output of the second GRU subnetwork (i.e., the secondary hidden state value in this embodiment) is used as the input of the fully connected subnetwork. The fully connected subnetwork is used to learn the correlation between the secondary hidden state value and the extraction point temperature of the high-pressure cylinder, thereby predicting the extraction point temperature of the high-pressure cylinder.

[0048] Taking the first GRU sub-network as an example, refer to Figure 2 This is a schematic diagram of the structure of a first GRU subnetwork provided in an embodiment of this application. Figure 2 As shown, the first GRU subnetwork includes a reset gate (output is...) Figure 2 r in t ) and update gate (output is Figure 2 z in t ), will target running data group x t After inputting the first GRU subnetwork, the output r of the reset gate can be obtained using the following expression. t and update the output z of the gate t :

[0049] r t =sigmoid(W r ·[h t-1 x t (1)

[0050] z t =sigmoid(W z ·[h t-1 x t (2)

[0051] Among them, h t-1 Indicates the target running data group x t The previous running data group x t-1 The corresponding primary hidden state value; [h t-1 x t ] indicates that h t-1 and x t (A one-dimensional column vector consisting of operating parameters of n-1 candidate units) is concatenated into column vectors; W r W is a predetermined 1*n weight matrix; r ·[h t-1 x t ] indicates W r and [h t-1 x t Perform matrix multiplication; sigmoid(W r ·[h t-1 x t ]) represents applying a pre-determined activation function to W r ·[h t-1 x t Normalization is performed. Taking the operating parameters of each candidate unit, which include the operating parameters of three candidate units, as an example, [h t-1 x t W is a 4x1 matrix. rIt is a 1x4 matrix.

[0052] In addition, W z It is also predetermined and different from W r The 1*n weight matrix, W z ·[h t-1 x t ] indicates W z and [h t-1 x t Perform matrix multiplication; sigmoid(W z ·[h t-1 x t ]) represents applying a pre-determined activation function to W z ·[h t-1 x t Normalization was performed.

[0053] After obtaining the output r of the reset gate t and update the output z of the gate t Then, the primary hidden state value h of the target running data group can be determined by the following expression. t :

[0054]

[0055]

[0056] Where, r t *h t-1 Indicates that for r t and h t-1 Perform element-wise multiplication, [r t *h t-1 x t ] indicates that for r t *h t-1 and x t Column vector concatenation, where W is a predetermined 1*n weight matrix, tanh(W·[r t *h t-1 x t ]) indicates that [r t *h t-1 x t ] Calculate the hyperbolic tangent function; (1-z t )*h t-1 Indicates the relationship between (1-z) t ) and h t-1 Perform element-wise multiplication. Indicates z t and Perform element-wise multiplication.

[0057] The second GRU subnetwork has the same structure as the first GRU subnetwork, except that W and W' are different in the two GRU subnetworks. z W r The values ​​of x differ. In the second GRU subnetwork, the primary hidden state value of the target running data group output by the first GRU subnetwork is used as x in the second GRU subnetwork. t Furthermore, W, W z W r All are 1*2 weight matrices. The process by which the second GRU subnetwork determines the secondary hidden state value of the target running data group can be referred to the process by which the first GRU subnetwork determines the primary hidden state value of the target running data group as described above, and will not be repeated here in the embodiments of this application.

[0058] In this embodiment, a GRU model with fewer parameters, faster training speed, and higher accuracy can be selected as the RNN model. This saves computational resources for the fault prediction equipment in the turbine flow path and improves model training efficiency and output accuracy. Furthermore, to further improve the model output accuracy, a two-layer GRU is selected in the GRU model.

[0059] Optionally, the first GRU subnetwork includes each neuron layer; the first GRU subnetwork processes the target running data group and the primary hidden state value corresponding to the previous running data group of the target running data group to obtain the primary hidden state value of the target running data group, including: determining the first neuron layer in each neuron layer as the current neuron layer; determining the current hidden state value output by the current neuron layer for the target running data group based on the hidden state value of the target running data group and the hidden state value output by the current neuron layer for the previous running data group of the target running data group; determining the next neuron layer from each neuron layer, determining the next hidden state value output by the next neuron layer for the target running data group based on the current hidden state value of the next neuron layer and the hidden state value output by the next neuron layer for the previous running data group of the target running data group, and determining the next hidden state value as the latest current hidden state value, until the next neuron layer is the last neuron layer in each neuron layer; and determining the latest current hidden state value as the primary hidden state value of the target running data group.

[0060] Specifically, the primary hidden state value corresponding to the previous running data group of the target running data group is the hidden state value of the previous running data group output by the last neuron layer for the target running data group. Furthermore, each time the current neuron layer outputs a hidden state value, it automatically records that hidden state value for future use. For example, the primary hidden state value corresponding to the previous running data group of the target running data group is the hidden state value recorded by the last neuron layer during the process of determining the predicted extraction point temperature corresponding to the previous running data group.

[0061] In one possible implementation, any neuron layer in the first GRU subnetwork can be Figure 2 The GRU structure shown includes multiple sequentially connected neuron layers in the first GRU subnetwork. The output of the current neuron layer serves as the input to the next neuron layer, and the output of the last neuron layer (i.e., the final neuron layer) is the primary hidden state value of the target running data set. The process of determining the hidden state value output by each neuron layer can be referred to the description above for determining the primary hidden state value h of the target running data set. t The process is not described in detail in the embodiments of this application.

[0062] To further improve the accuracy of the model output, in this embodiment, multiple neuron layers can be deployed in the first GRU subnetwork. Similarly, multiple neuron layers can be deployed in the second GRU subnetwork. For example, in this embodiment, the first GRU subnetwork includes 120 neuron layers, and the second GRU subnetwork includes 60 neuron layers.

[0063] S103. Based on the target deviation between the actual target extraction point temperature and the target predicted extraction point temperature, determine the fault prediction result for the flow path of the steam turbine.

[0064] The target deviation between the actual target extraction point temperature and the predicted target extraction point temperature can be the absolute value of the difference between the actual target extraction point temperature and the predicted target extraction point temperature.

[0065] In one possible implementation, when the target deviation value is greater than a predetermined value, the fault prediction result can be determined as a fault has occurred; otherwise, the fault prediction result is determined as no fault has occurred.

[0066] Optionally, based on the target deviation between the actual extraction point temperature and the predicted extraction point temperature, the fault prediction result for the flow path of the steam turbine is determined, including: determining a reference deviation value according to the historical actual extraction point temperature and the historical predicted extraction point temperature corresponding to each historical operating data group; if the difference between the target deviation value and the reference deviation value meets the first preset condition, the fault prediction result is determined to be a fault; if the difference between the target deviation value and the reference deviation value does not meet the first preset condition, the fault prediction result is determined to be no fault.

[0067] In one possible implementation, the absolute value of the difference between the actual historical extraction point temperature and the corresponding historical predicted extraction point temperature can be determined (i.e., each historical deviation value is determined), and then the standard deviation of each historical deviation value is determined as the benchmark deviation value.

[0068] The first preset condition can be a predetermined condition; for example, the first preset condition can be that the target deviation value is greater than three times the benchmark deviation value.

[0069] In this embodiment, to improve the accuracy of the determined fault prediction results, a benchmark deviation value can be determined based on the historical actual extraction steam point temperature corresponding to each historical operating data group and the historical predicted extraction steam point temperature corresponding to each historical operating data group. This benchmark deviation value can more accurately represent the difference between the predicted extraction steam point temperature and the actual extraction steam point temperature under normal operating conditions of the turbine unit. Therefore, by comparing the target deviation value with the benchmark deviation value, it is possible to more accurately predict whether the turbine unit is in normal operating condition, that is, to more accurately determine the fault prediction result for the flow path of the turbine.

[0070] Optionally, after determining that a fault prediction result indicates a fault has occurred, the fault prediction method for the turbine flow path provided in this application embodiment further includes: sending a load reduction signal to the turbine's control and regulation system and starting to execute a periodic fault clearance judgment task; during the execution of the fault clearance judgment task, if it is determined that the fault has been cleared, sending a load restoration signal to the control and regulation system and pausing the execution of the fault clearance judgment task; after a preset time period for executing the fault clearance judgment task, if it is determined that the fault has not been cleared, issuing a control failure warning signal.

[0071] The execution cycle of the fault resolution judgment task can be a preset sampling interval; the task content of the fault resolution judgment task includes: acquiring the real-time operating data set of the steam turbine and the real-time actual extraction point temperature of the high-pressure cylinder; inputting the real-time operating data set into the extraction point temperature prediction model to obtain the real-time predicted extraction point temperature of the high-pressure cylinder output by the extraction point temperature prediction model; and determining whether the fault has been resolved based on the real-time deviation value between the real-time actual extraction point temperature and the real-time predicted extraction point temperature, as well as the benchmark deviation value.

[0072] In one possible implementation, if the real-time deviation value is less than twice the reference deviation value, the fault is determined to be resolved; otherwise, the fault is determined to be not resolved. Furthermore, the process of specifically determining the real-time predicted extraction point temperature of the high-pressure cylinder output by the extraction point temperature prediction model is the same as the process of determining the target predicted extraction point temperature of the high-pressure cylinder output by the extraction point temperature prediction model, and will not be repeated here in the embodiments of this application.

[0073] In this embodiment, after a fault is detected, a load reduction signal can be sent to the turbine's control and regulation system to instruct the system to eliminate the fault by reducing the load. Furthermore, after a fault is detected, a periodic fault resolution judgment task can be performed to determine whether the fault has been resolved. If the fault has been resolved, a load restoration signal can be sent to the control and regulation system to restore the load to normal. Additionally, if the fault has not been resolved after a preset time period of the fault resolution judgment task, a control failure warning signal can be issued to remind maintenance personnel to inspect the equipment. Through this automatic adjustment process, some faults can be automatically eliminated, reducing the workload of maintenance personnel.

[0074] Optionally, after determining that no fault has occurred in the fault prediction result, the fault prediction method for the turbine flow passage provided in this application embodiment further includes: updating each historical operating data group based on the target operating data group; and updating the extraction point temperature prediction model based on the updated historical operating data group and the historical actual extraction point temperature corresponding to each updated historical operating data group.

[0075] The process of updating the extraction point temperature prediction model based on the updated historical operating data sets and the corresponding historical actual extraction point temperatures can be referred to the aforementioned process of training the extraction point temperature prediction model.

[0076] In practical applications, this embodiment of the application can perform a fault prediction task once at a preset sampling interval (i.e., acquire the target operating data set and the target actual extraction steam point temperature, and determine the fault prediction result for the flow path of the steam turbine). To improve the accuracy of the determined fault prediction result, this embodiment of the application can dynamically update the extraction steam point temperature prediction model after the fault prediction task is completed. In addition, while dynamically updating the extraction steam point temperature prediction model, the baseline standard deviation can also be dynamically updated.

[0077] In summary, the fault prediction method for the flow passage of a steam turbine provided in this application embodiment, when predicting whether a fault has occurred in the flow passage of the steam turbine, can first obtain the current operating data set of the steam turbine (i.e., the target operating data set in this application embodiment) and the current actual extraction point temperature of the high-pressure cylinder of the steam turbine (i.e., the target actual extraction point temperature in this application embodiment). Then, the target operating data set can be input into a pre-trained extraction point temperature prediction model to obtain the predicted extraction point temperature of the high-pressure cylinder output by the extraction point temperature prediction model (i.e., the target predicted extraction point temperature in this application embodiment). Afterwards, based on the current deviation value between the target actual extraction point temperature and the target predicted extraction point temperature (i.e., the target deviation value in this application embodiment), the fault prediction result for the flow passage of the steam turbine can be determined. Because steam turbine units have inherent dynamic characteristics under normal operating conditions, that is, under normal operating conditions, the extraction point temperature of the high-pressure cylinder at a certain moment is closely related to the extraction point temperature of the high-pressure cylinder before that moment; and the acquired target operating data set includes real-time operating data corresponding to the operating parameters of each candidate unit of the steam turbine, and the operating parameters of each candidate unit can be the unit operating parameters related to the extraction point temperature of the high-pressure cylinder (that is, the magnitude of the operating parameters of each candidate unit will affect the height of the extraction point temperature of the high-pressure cylinder). Based on this, in predicting whether a fault has occurred in the flow path of the steam turbine, this embodiment of the application can use the extraction point temperature prediction model based on the target operating data set to predict the extraction point temperature of the high-pressure cylinder under normal operating conditions (i.e., determine the target predicted extraction point temperature in this embodiment of the application). Since the extraction point temperature prediction model is pre-trained based on historical operating data sets arranged in chronological order and the historical actual extraction point temperatures corresponding to each historical operating data set, the target predicted extraction point temperature can be determined relatively accurately. Therefore, when the difference between the target predicted extraction point temperature and the target actual extraction point temperature is large, it indicates that a fault may have occurred in the flow path of the steam turbine.

[0078] Furthermore, since the extraction port of the high-pressure cylinder is generally located very close to the flow path of the turbine, the error between the collected extraction point temperature of the high-pressure cylinder at a certain moment (i.e., the actual extraction point temperature) and the true extraction point temperature of the high-pressure cylinder at that moment is very small. In other words, the temporal difference of this parameter, the extraction point temperature of the high-pressure cylinder, is very small. Therefore, compared to using thermodynamic characteristic parameters as fault analysis indicators, using the extraction point temperature of the high-pressure cylinder as a fault analysis indicator in this application can reduce the impact of temporal differences on the fault prediction results. It can be seen that the embodiments of this application use the extraction point temperature of the high-pressure cylinder, which has a small temporal difference, as the fault analysis indicator for the flow path of the turbine. By determining the fault prediction result for the flow path of the turbine through the deviation between the actual extraction point temperature and the predicted extraction point temperature of the high-pressure cylinder, the accuracy of the determined fault prediction result can be improved.

[0079] Optional, such as Figure 3 As shown in the embodiment of this application, a fault prediction method for the flow path of a steam turbine is also provided, which may include S301-S304:

[0080] S301. Based on the historical operating data groups arranged in chronological order and the historical actual extraction point temperatures corresponding to each historical operating data group, the pre-determined GRU model is trained to obtain the extraction point temperature prediction model.

[0081] S302, Obtain the target operating data set of the steam turbine and the target actual extraction point temperature of the high-pressure cylinder of the steam turbine.

[0082] S303. Input the target operating data set into the extraction point temperature prediction model. Process the primary hidden state value of the target operating data set and the primary hidden state value corresponding to the previous operating data set through the first GRU sub-network to obtain the primary hidden state value of the target operating data set. Process the primary hidden state value of the target operating data set and the secondary hidden state value corresponding to the previous operating data set through the second GRU sub-network to obtain the secondary hidden state value of the target operating data set. Process the secondary hidden state value of the target operating data set through the fully connected sub-network to obtain the target predicted extraction point temperature.

[0083] S304. Based on the target deviation between the actual target extraction point temperature and the target predicted extraction point temperature, determine the fault prediction result for the flow path of the steam turbine.

[0084] Optional, such as Figure 4 As shown in the embodiment of this application, a fault prediction method for the flow path of a steam turbine is also provided, which may include steps S401-S4011:

[0085] S401. Based on the historical operating data groups arranged in chronological order and the historical actual extraction steam point temperatures corresponding to each historical operating data group, train the pre-determined GRU model to obtain the extraction steam point temperature prediction model, and obtain the historical predicted extraction steam point temperatures corresponding to each historical operating data group output by the extraction steam point temperature prediction model.

[0086] S402, Obtain the target operating data set of the steam turbine and the target actual extraction point temperature of the high-pressure cylinder of the steam turbine.

[0087] S403. Input the target operating data set into the extraction point temperature prediction model to obtain the target predicted extraction point temperature of the high-pressure cylinder output by the extraction point temperature prediction model.

[0088] S404. Determine the baseline deviation value based on the historical actual extraction steam point temperature corresponding to each historical operating data group and the historical predicted extraction steam point temperature corresponding to each historical operating data group.

[0089] S405. Determine whether the target deviation between the actual target extraction point temperature and the target predicted extraction point temperature is greater than three times the reference deviation.

[0090] If the target deviation value is determined to be greater than three times the reference deviation value, proceed to step S406; if the target deviation value is determined to be less than or equal to three times the reference deviation value, proceed to step S4010.

[0091] S406. If the fault prediction result is determined to be a fault, a load reduction signal is sent to the turbine control and regulation system, and the periodic fault resolution judgment task is started.

[0092] S407. Determine the real-time deviation value based on the real-time actual extraction point temperature and the real-time predicted extraction point temperature.

[0093] S408. Determine whether the real-time deviation value is less than twice the reference deviation value.

[0094] If the real-time deviation value is determined to be less than twice the reference deviation value, proceed to step S409; if the real-time deviation value is determined to be greater than or equal to twice the reference deviation value, return to step S407.

[0095] S409. Once the fault has been determined, send a load recovery signal to the control and regulation system and suspend the fault clearance judgment task.

[0096] After step S409, return to step S402.

[0097] S4010. If the fault prediction result is determined to be no fault, update each historical operating data group based on the target operating data group.

[0098] S4011. Based on the updated historical operating data sets and the corresponding historical actual extraction point temperatures for each updated historical operating data set, the extraction point temperature prediction model is updated.

[0099] After step S4011, return to step S402.

[0100] like Figure 5 As shown in the figure, this application embodiment also provides a fault prediction device for the flow passage of a steam turbine, which may include: an acquisition module 11, a model processing module 21 and a prediction module 31.

[0101] Specifically, the acquisition module 11 executes S101 in the above method embodiment, the model processing module 21 executes S102 in the above method embodiment, and the prediction module 31 executes S103 in the above method embodiment.

[0102] Specifically, the acquisition module 11 is used to acquire the target operating data set of the steam turbine and the target actual extraction point temperature of the high-pressure cylinder of the steam turbine; wherein, the target operating data set includes real-time operating data corresponding to the operating parameters of each candidate unit of the steam turbine;

[0103] Model processing module 21 is used to input the target operating data set into the extraction point temperature prediction model to obtain the target predicted extraction point temperature of the high-pressure cylinder output by the extraction point temperature prediction model; the extraction point temperature prediction model is pre-trained based on each historical operating data set arranged in time sequence and the historical actual extraction point temperature corresponding to each historical operating data set.

[0104] The prediction module 31 is used to determine the fault prediction result for the flow passage of the steam turbine based on the target deviation value between the target actual extraction point temperature and the target predicted extraction point temperature.

[0105] Optionally, the extraction point temperature prediction model includes a first GRU subnetwork, a second GRU subnetwork, and a fully connected subnetwork; the model processing module 21 is specifically used for:

[0106] The target operating data set is input into the extraction point temperature prediction model. The first GRU subnetwork processes the primary hidden state value of the target operating data set and the primary hidden state value corresponding to the previous operating data set to obtain the primary hidden state value of the target operating data set. The second GRU subnetwork processes the primary hidden state value of the target operating data set and the secondary hidden state value corresponding to the previous operating data set to obtain the secondary hidden state value of the target operating data set. The fully connected subnetwork processes the secondary hidden state value of the target operating data set to obtain the target predicted extraction point temperature.

[0107] Optionally, the first GRU subnetwork includes layers of neurons; the model processing module 21 is further used for:

[0108] The first neuron layer in each neuron layer is determined as the current neuron layer;

[0109] The current hidden state value of the target running data group is determined by the current neuron layer's output for the target running data group and the hidden state value of the previous running data group output by the current neuron layer for the target running data group.

[0110] The next neuron layer is determined from each neuron layer. Using the current hidden state value of the next neuron layer and the hidden state value of the previous running data group output by the next neuron layer for the target running data group, the next hidden state value output by the next neuron layer for the target running data group is determined, and this next hidden state value is set as the latest current hidden state value. This process continues until the next neuron layer is the last neuron layer in the series. The primary hidden state value corresponding to the previous running data group of the target running data group is the hidden state value output by the last neuron layer for the previous running data group of the target running data group.

[0111] The latest current hidden state value is determined as the primary hidden state value of the target running data group.

[0112] Optionally, the prediction module 31 is specifically used for:

[0113] Based on the historical actual extraction steam point temperature corresponding to each historical operating data group and the historical predicted extraction steam point temperature corresponding to each historical operating data group, a benchmark deviation value is determined; if the difference between the target deviation value and the benchmark deviation value meets the first preset condition, the fault prediction result is determined to be a fault; if the difference between the target deviation value and the benchmark deviation value does not meet the first preset condition, the fault prediction result is determined to be no fault.

[0114] Optionally, the fault prediction device for the turbine flow path section also includes a sending module;

[0115] The sending module is used to send a load reduction signal to the turbine's control and regulation system after the prediction module 31 determines that a fault has occurred, and to begin executing a periodic fault clearance judgment task. The fault clearance judgment task includes: acquiring the turbine's real-time operating data set and the real-time actual extraction point temperature of the high-pressure cylinder; inputting the real-time operating data set into the extraction point temperature prediction model to obtain the real-time predicted extraction point temperature of the high-pressure cylinder output by the extraction point temperature prediction model; and determining whether the fault has been cleared based on the real-time deviation value between the real-time actual extraction point temperature and the real-time predicted extraction point temperature, as well as the reference deviation value.

[0116] The sending module is also used to send a load recovery signal to the control and regulation system and suspend the execution of the fault clearance judgment task if it is determined that the fault has been cleared during the execution of the fault clearance judgment task.

[0117] The sending module is also used to issue a control failure warning signal if it is determined that the fault has not been resolved after a preset time period for performing the fault resolution judgment task.

[0118] Optionally, the fault prediction device for the turbine flow path also includes an update module; the update module is used for:

[0119] After the prediction module 31 determines that the fault prediction result is no fault, it updates each historical operating data group based on the target operating data group; based on the updated historical operating data group and the historical actual extraction point temperature corresponding to each updated historical operating data group, it updates the extraction point temperature prediction model.

[0120] Optionally, the operating parameters of each candidate unit include at least one of the following: unit load, high-pressure cylinder exhaust temperature, high-pressure cylinder exhaust pressure, regulating stage temperature, regulating stage pressure, high-pressure cylinder extraction point pressure, high-pressure heater inlet water temperature, high-pressure heater outlet water temperature, and high-pressure heater condensate temperature.

[0121] Optionally, the fault prediction device for the turbine flow passage section may also include a storage module, which is used to store the program code of the fault prediction device for the turbine flow passage section.

[0122] like Figure 6 As shown, this application embodiment also provides a fault prediction device for the flow path of a steam turbine, including a memory 41 and a processor (e.g., Figure 6The processor includes 42-1 and 42-2, bus 43 and communication interface 44; memory 41 is used to store computer execution instructions, and the processor and memory 41 are connected through bus 43; when the fault prediction device of the turbine flow passage is running, the processor executes the computer execution instructions stored in memory 41 so that the fault prediction device of the turbine flow passage performs the fault prediction method of the turbine flow passage as provided in the above embodiments.

[0123] In a specific implementation, as one example, the processor may include one or more central processing units (CPUs), for example... Figure 6 CPU0 and CPU1 are shown in the diagram. Furthermore, as one embodiment, the fault prediction device for the turbine flow path may include multiple processors, such as... Figure 6 The processors 42-1 and 42-2 are shown in the diagram. Each of these processors (CPUs) can be a single-core processor or a multi-core processor. Here, "processor" can refer to one or more devices, circuits, and / or processing cores used to process data (e.g., computer program instructions).

[0124] The memory 41 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital versatile optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. The memory 41 may exist independently and be connected to the processor via bus 43. The memory 41 may also be integrated with the processor.

[0125] In a specific implementation, memory 41 is used to store data from this application and computer execution instructions corresponding to the software program of this application. The processor can perform various functions of the fault prediction device for the turbine flow passage by running or executing the software program stored in memory 41 and calling the data stored in memory 41.

[0126] Communication interface 44 uses any transceiver-like device for communicating with other devices or communication networks, such as control systems, radio access networks (RAN), wireless local area networks (WLAN), etc. Communication interface 44 may include a receiving unit to implement receiving functions and a transmitting unit to implement transmitting functions.

[0127] Bus 43 can be an industry standard architecture (ISA) bus, a peripheral component interconnect (PCI) bus, or an extended industry standard architecture (EISA) bus, etc. This bus 43 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 6 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0128] As an example, combined Figure 5 The acquisition module in the fault prediction device for the steam turbine flow path section performs the same function as... Figure 6 The receiving unit in the middle performs the same function, and the prediction module in the fault prediction device of the turbine flow passage section performs the same function. Figure 6 The processor in the memory performs the same function. When the fault prediction device for the turbine flow path includes a storage module, the storage module performs the same function as... Figure 6 The memory in them performs the same function.

[0129] The explanation of the relevant content in this embodiment can be found in the above method embodiment, and will not be repeated here.

[0130] Through the above description of the embodiments, those skilled in the art will clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0131] This application also provides a computer-readable storage medium storing instructions that, when executed by a computer, cause the computer to perform the fault prediction method for the steam turbine flow path provided in the above embodiments.

[0132] The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, RAM, ROM, an erasable programmable read-only memory (EPROM), a register, a hard disk, an optical fiber, a CD-ROM, an optical storage device, a magnetic storage device, or any suitable combination thereof, or any other form of computer-readable storage medium known in the art. An exemplary storage medium is coupled to a processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can reside in an application-specific integrated circuit (ASIC). In embodiments of this application, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0133] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for predicting faults in the flow path of a steam turbine, characterized in that, include: The target operating data set of the steam turbine and the target actual extraction point temperature of the high-pressure cylinder of the steam turbine are obtained; wherein, the target operating data set includes real-time operating data corresponding to the operating parameters of each candidate unit of the steam turbine; The target operating data set is input into the extraction point temperature prediction model to obtain the target predicted extraction point temperature of the high-pressure cylinder output by the extraction point temperature prediction model; the extraction point temperature prediction model is pre-trained based on each historical operating data set arranged in chronological order and the historical actual extraction point temperature corresponding to each historical operating data set. Based on the target deviation between the actual target extraction point temperature and the target predicted extraction point temperature, the fault prediction result for the flow passage section of the steam turbine is determined. The extraction point temperature prediction model includes a first gated cyclic unit (GRU) subnetwork, a second GRU subnetwork, and a fully connected subnetwork; the step of inputting the target operating data set into the extraction point temperature prediction model to obtain the target predicted extraction point temperature of the high-pressure cylinder output by the extraction point temperature prediction model includes: The target operating data set is input into the extraction point temperature prediction model. The primary hidden state value of the target operating data set and the previous operating data set of the target operating data set are processed through the first GRU sub-network to obtain the primary hidden state value of the target operating data set. The second GRU subnetwork processes the primary hidden state value of the target running data group and the secondary hidden state value corresponding to the previous running data group of the target running data group to obtain the secondary hidden state value of the target running data group. The secondary hidden state values ​​of the target running data group are processed through the fully connected subnetwork to obtain the target predicted extraction point temperature. The determination of fault prediction results for the flow path section of the steam turbine based on the target deviation value between the actual target extraction point temperature and the target predicted extraction point temperature includes: The baseline deviation value is determined based on the actual historical extraction steam point temperature corresponding to each historical operating data group and the predicted historical extraction steam point temperature corresponding to each historical operating data group. If the difference between the target deviation value and the benchmark deviation value satisfies a first preset condition, the fault prediction result is determined to be a fault. If the difference between the target deviation value and the benchmark deviation value does not meet the first preset condition, the fault prediction result is determined to be no fault.

2. The fault prediction method for the flow path of a steam turbine according to claim 1, characterized in that, The first GRU subnetwork includes layers of neurons; the step of processing the target running data group and the primary hidden state value corresponding to the previous running data group of the target running data group through the first GRU subnetwork to obtain the primary hidden state value of the target running data group includes: The first neuron layer in each neuron layer is determined as the current neuron layer; The current hidden state value output by the current neuron layer for the target running data group is determined by the current neuron layer for the target running data group and the hidden state value output by the current neuron layer for the previous running data group for the target running data group. The next neuron layer is determined from each neuron layer. Using the current hidden state value of the next neuron layer and the hidden state value of the previous running data group output by the next neuron layer for the target running data group, the next hidden state value output by the next neuron layer for the target running data group is determined, and this next hidden state value is set as the latest current hidden state value. This process continues until the next neuron layer is the last neuron layer among all the neuron layers. The primary hidden state value corresponding to the previous running data group of the target running data group is the hidden state value output by the last neuron layer for the previous running data group of the target running data group. The latest current hidden state value is determined as the primary hidden state value of the target running data group.

3. The fault prediction method for the flow path of a steam turbine according to claim 1, characterized in that, After determining that the fault prediction result indicates a fault has occurred, the method further includes: A load reduction signal is sent to the control and regulation system of the steam turbine, and a periodic fault resolution judgment task is initiated. The fault resolution judgment task includes: acquiring the real-time operating data set of the steam turbine and the real-time actual extraction steam point temperature of the high-pressure cylinder; inputting the real-time operating data set into the extraction steam point temperature prediction model to obtain the real-time predicted extraction steam point temperature of the high-pressure cylinder output by the extraction steam point temperature prediction model; and determining whether the fault has been resolved based on the real-time deviation between the real-time actual extraction steam point temperature and the real-time predicted extraction steam point temperature, and the reference deviation value. If, during the execution of the fault resolution judgment task, it is determined that the fault has been resolved, a load recovery signal is sent to the control and regulation system, and the execution of the fault resolution judgment task is suspended. If, after a preset time has elapsed since the fault resolution judgment task was performed, the fault is determined not to be resolved, a control failure warning signal is issued.

4. The fault prediction method for the flow path of a steam turbine according to claim 1, characterized in that, After determining that the fault prediction result is that no fault has occurred, the method further includes: Update each historical running data group based on the target running data group; The extraction point temperature prediction model is updated based on the updated historical operating data sets and the corresponding historical actual extraction point temperatures for each updated historical operating data set.

5. The fault prediction method for the flow passage of a steam turbine according to any one of claims 1-4, characterized in that, The operating parameters of each candidate unit include at least one of the following: unit load, high-pressure cylinder exhaust temperature, high-pressure cylinder exhaust pressure, regulating stage temperature, regulating stage pressure, high-pressure cylinder extraction point pressure, high-pressure heater inlet water temperature, high-pressure heater outlet water temperature, and high-pressure heater condensate temperature.

6. A fault prediction device for the flow path of a steam turbine, characterized in that, include: The acquisition module is used to acquire the target operating data set of the steam turbine and the target actual extraction point temperature of the high-pressure cylinder of the steam turbine; wherein, the target operating data set includes real-time operating data corresponding to the operating parameters of each candidate unit of the steam turbine; The model processing module is used to input the target operating data set into the extraction point temperature prediction model to obtain the target predicted extraction point temperature of the high-pressure cylinder output by the extraction point temperature prediction model; the extraction point temperature prediction model is pre-trained based on each historical operating data set arranged in chronological order and the historical actual extraction point temperature corresponding to each historical operating data set. The prediction module is used to determine the fault prediction result for the flow passage of the steam turbine based on the target deviation value between the actual target extraction point temperature and the target predicted extraction point temperature. The extraction point temperature prediction model includes a first GRU subnetwork, a second GRU subnetwork, and a fully connected subnetwork. The model processing module is specifically used for: inputting the target operating data set into the extraction point temperature prediction model; processing the target operating data set and the primary hidden state value corresponding to the previous operating data set through the first GRU sub-network to obtain the primary hidden state value of the target operating data set; processing the primary hidden state value of the target operating data set and the secondary hidden state value corresponding to the previous operating data set through the second GRU sub-network to obtain the secondary hidden state value of the target operating data set; and processing the secondary hidden state value of the target operating data set through the fully connected sub-network to obtain the target predicted extraction point temperature. The prediction module is specifically used to: determine a baseline deviation value based on the historical actual extraction point temperature corresponding to each historical operating data group and the historical predicted extraction point temperature corresponding to each historical operating data group; determine the fault prediction result as a fault occurring if the difference between the target deviation value and the baseline deviation value meets a first preset condition; and determine the fault prediction result as no fault occurring if the difference between the target deviation value and the baseline deviation value does not meet the first preset condition.

7. A fault prediction device for the flow path of a steam turbine, characterized in that, It includes a memory, a processor, a bus, and a communication interface; the memory is used to store computer-executed instructions, and the processor is connected to the memory via the bus; When the fault prediction device for the turbine flow passage is running, the processor executes the computer execution instructions stored in the memory to cause the fault prediction device for the turbine flow passage to perform the fault prediction method for the turbine flow passage as described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed by a computer, cause the computer to perform the fault prediction method for the steam turbine flow path as described in any one of claims 1-5.

Images