Monitoring method, system and equipment for flat plate sealing state maintenance and medium

Abstract

The invention discloses a monitoring method, system and equipment for flat plate sealing state maintenance and a medium, and relates to the technical field of sealing state monitoring, and the method comprises the steps: obtaining the current sealing ring wear degree of a to-be-detected unit, the cumulative load of the unit and the operated time of the unit; respectively inputting the wear degree of the sealing ring, the accumulated load of the unit and the operated time of the unit into a first prediction model, a second prediction model and a third prediction model for processing to respectively obtain a first prediction time length, a second prediction time length and a third prediction time length; performing multi-source fusion on the first prediction duration, the second prediction duration and the third prediction duration to obtain a fusion result, and determining the fusion result as a target prediction duration; the maintenance and replacement time and the sealing operation state of the flat plate seal are predicted according to the accumulated load of a unit and the thickness of a sealing ring in the same time in the history, and the maintenance period is accurately arranged according to the state of the sealing ring, so that the purpose of performing state maintenance on the flat plate seal is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of sealing state monitoring, and more specifically, to a monitoring method, system, equipment and medium for flat plate sealing state inspection and maintenance. Background Art

[0002] Most flat plate seals of turbines use polyurethane composite materials with outstanding wear resistance as sealing plates, and stainless steel as sealing rings. During operation, the water flow of the unit squeezes the sealing plate into the sealing ring, thereby playing a sealing role; in actual operation, the degree of wear of the sealing ring is much greater than that of the sealing plate. Because the wear of the sealing ring leads to seal failure, excessive leakage, and rapid rise in the water level of the top cover, the unit is forced to shut down for maintenance. At present, flat plate seals are mostly replaced by planned maintenance, but after inspection, most of the sealing rings replaced in this way can still meet the normal operation of the unit for a period of time, which results in excessive equipment maintenance and power loss, forming a waste of resources. Summary of the invention

[0003] The object of the present invention is to provide a monitoring method, system, equipment and medium for flat plate sealing status inspection to solve the problems existing in the above-mentioned background technology.

[0004] The above technical objectives of the present invention are achieved through the following technical solutions:

[0005] In a first aspect, the present application provides a method for monitoring the sealing status of a flat plate, comprising the following specific steps:

[0006] A training data set is obtained from the historical operation data of the turbine unit in the same period of time, and the operation time from the time point of each data in the training data set to the replacement time of the sealing ring in the period of time, and the training data set includes the sealing ring wear degree data, the unit cumulative load data and the unit operation time data;

[0007] The preset first model is trained using the seal ring wear degree data and the corresponding operating time, the preset second model is trained using the unit cumulative load data and the corresponding operating time, and the preset third model is trained using the unit running time data and the corresponding operating time;

[0008] Until the preset first model, the preset second model, and the preset third model respectively reach the corresponding training end conditions, the preset first model, the preset second model, and the preset third model that respectively reach the corresponding training end conditions are determined as the first prediction model, the second prediction model, and the third prediction model respectively;

[0009] Obtain the current degree of seal ring wear, cumulative load of the unit and running time of the unit for the unit to be tested, and input the degree of seal ring wear, cumulative load of the unit and running time of the unit into the first prediction model, the second prediction model and the third prediction model for processing, respectively, to obtain the first prediction duration, the second prediction duration and the third prediction duration;

[0010] The first prediction duration, the second prediction duration and the third prediction duration are subjected to multi-source fusion to obtain a fusion result, and the fusion result is determined as a target prediction duration, which represents the duration that the unit under test can still operate.

[0011] Based on the above technical solution, the present invention can also be improved as follows.

[0012] Furthermore, the above training end condition is that the loss functions corresponding to the preset first model, the preset second model and the preset third model do not exceed the threshold value, and the loss function of the preset first model, the preset second model or the preset third model is specifically:

[0013]

[0014] Where Loss represents the loss function value, N represents the total amount of data in the seal ring wear degree data, or the unit cumulative load data, or the unit running time data, a represents the sequence number of the data in the seal ring wear degree data, or the unit cumulative load data, or the unit running time data, θ a Indicates the running time of the ath data in the seal ring wear degree data, or the unit cumulative load data, or the unit running time data. It indicates the predicted value outputted after the a-th data is input into the preset first model, the preset second model, or the preset third model.

[0015] Furthermore, the above target prediction duration is specifically:

[0016]

[0017] Wherein, m represents the fusion result, m1 represents the first prediction duration, m2 represents the second prediction duration, m3 represents the third prediction duration, w1 represents the exponential weight of the first prediction duration, w2 represents the exponential weight of the second prediction duration, and w3 represents the exponential weight of the third prediction duration.

[0018] Furthermore, the exponential weight of the first predicted duration, the second predicted duration, or the third predicted duration is obtained by:

[0019] Based on the preset initial values ​​of the respective exponential weights, the output values ​​of the preset first model, the preset second model, and the preset third model during the training process are preliminarily integrated to obtain a first result;

[0020] Using the first result, calculate the loss values ​​corresponding to the exponential weights of the first predicted duration, the second predicted duration, and the third predicted duration respectively;

[0021] Based on each loss value, the corresponding exponential weight is iteratively updated until the loss value corresponding to each exponential weight reaches the iteration end condition, and the exponential weight that reaches the iteration end condition is used to perform multi-source fusion on the first prediction duration, the second prediction duration and the third prediction duration to obtain a fusion result.

[0022] Furthermore, the loss values ​​corresponding to the exponential weights of the first prediction duration, the second prediction duration, and the third prediction duration are specifically:

[0023]

[0024] In the formula, A represents the loss value corresponding to the current exponential weight, m(i) represents the first result, and y i Indicates the running time corresponding to the first result.

[0025] Furthermore, the corresponding exponential weights are iteratively updated based on each loss value, specifically:

[0026]

[0027] In the formula, w represents the updated exponential weight, w a represents the exponential weight before updating or the preset initial value of the exponential weight, α represents the current learning rate of the preset first model, the preset second model, or the preset third model, represents the current model weight of the preset first model, the preset second model, or the preset third model, and A represents the exponential weight w a The corresponding loss value.

[0028] In a second aspect, the present application provides a monitoring system for flat plate sealing state maintenance, which is applied to a monitoring method for flat plate sealing state maintenance in any one of the first aspects, comprising:

[0029] The first module is used to obtain a training data set from the historical operation data of the turbine unit in the same period of time, and the operation time from the time point of each data in the training data set to the replacement time of the sealing ring in the period of time, and the training data set includes the sealing ring wear degree data, the unit cumulative load data and the unit operation time data;

[0030] The second module is used to train the preset first model using the seal ring wear degree data and the corresponding operating time, to train the preset second model using the unit cumulative load data and the corresponding operating time, and to train the preset third model using the unit running time data and the corresponding operating time;

[0031] The third module is used to determine the preset first model, the preset second model and the preset third model that respectively reach the corresponding training end conditions as the first prediction model, the second prediction model and the third prediction model respectively until the preset first model, the preset second model and the preset third model respectively reach the corresponding training end conditions;

[0032] The fourth module is used to obtain the current seal ring wear degree, the cumulative load of the unit and the running time of the unit to be tested, and input the seal ring wear degree, the cumulative load of the unit and the running time of the unit into the first prediction model, the second prediction model and the third prediction model for processing, respectively, to obtain the first prediction time, the second prediction time and the third prediction time;

[0033] The fifth module is used to perform multi-source fusion of the first prediction duration, the second prediction duration and the third prediction duration to obtain a fusion result, and determine the fusion result as a target prediction duration, which represents the duration that the unit under test can still operate.

[0034] Furthermore, the fifth module includes:

[0035] A first submodule is used to preliminarily fuse the output values ​​of the preset first model, the preset second model and the preset third model during the training process based on the preset initial values ​​of each exponential weight and obtain a first result;

[0036] The second submodule is used to calculate the loss values ​​corresponding to the exponential weights of the first predicted duration, the second predicted duration, and the third predicted duration using the first result;

[0037] The third submodule is used to iteratively update the corresponding exponential weights based on each loss value until the loss value corresponding to each exponential weight reaches the iteration end condition, and use the exponential weights that reach the iteration end condition to perform multi-source fusion on the first prediction duration, the second prediction duration and the third prediction duration to obtain a fusion result.

[0038] In a third aspect, the present application 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 implements any one of the methods in the first aspect when executing the computer program.

[0039] In a fourth aspect, the present application provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions enable a computer to execute any one of the methods in the first aspect.

[0040] Compared with the prior art, the present invention has at least the following beneficial effects:

[0041] In the present application, firstly, a training data set for training a model is obtained from the historical operation data of the turbine unit, wherein the training data set includes the sealing ring wear degree data, the unit cumulative load data and the unit running time data arranged in time, and the running time corresponding to the data at each time point in the sealing ring wear degree data, the unit cumulative load data and the unit running time data, that is, the actual running time recorded in the historical operation data; secondly, the training data set is used to train the preset first model, the preset second model and the preset third model respectively, that is, the three data of the sealing ring wear degree data, the unit cumulative load data and the unit running time data are respectively trained for the corresponding models, until each model completes the training process; finally, the current sealing ring wear degree, the unit cumulative load and the unit running time of the unit to be tested are respectively processed using the trained models, so as to obtain the first prediction time, the second prediction time and the third prediction time, and then the three prediction results are used for multi-source fusion, and the fusion result is determined as the target prediction time, which indicates the time that the unit to be tested can still run.

[0042] In the present application, the thickness data of the flat plate seal is measured in real time by the sensor. The thickness data indicates the degree of wear of the sealing ring. With reference to many years of historical data, a model of the seal ring wear thickness, the cumulative load of the unit, and the predicted service life of the seal ring is established. Based on the above model, the time for the maintenance and replacement of the flat plate seal and the operating status of the seal are predicted according to the cumulative load of the unit and the thickness of the seal ring in the same period of history. According to the status of the seal ring, the maintenance cycle is accurately arranged, thereby achieving the purpose of condition maintenance of the flat plate seal. And the analysis and calculation of each model is used to effectively increase the service life of the seal ring, reduce the maintenance cost of the power station, and increase the power generation of the unit. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0044] Figure 1 A method flow chart of a monitoring method in an embodiment of the present invention;

[0045] Figure 2 A connection diagram of a monitoring system in an embodiment of the present invention;

[0046] Figure 3 A schematic diagram of the connection of electronic equipment in an embodiment of the present invention;

[0047] Figure 4 Schematic diagram of flat plate seal wear in an embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0049] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0051] In the description of the embodiments of the present invention, "plurality" means at least 2.

[0052] Embodiment 1:

[0053] In order to solve the problem that currently, the flat plate seals are mostly replaced by planned maintenance, but most of the seal rings replaced by this method can still meet the normal operation of the unit for a period of time after inspection, resulting in excessive equipment maintenance and power loss, forming a waste of resources, this embodiment provides a monitoring method for the maintenance of the flat plate seal state, such as Figure 1 As shown, the specific steps include:

[0054] S1, obtain a training data set from the historical operation data of the turbine unit in the same period of time, and the operating time from the time point of each data in the training data set during the period of time to the sealing ring replacement time, the training data set includes the sealing ring wear degree data, the unit cumulative load data and the unit operation time data.

[0055] Among them, three types of data can be selected from the historical operation data of the turbine unit for model training, which can be the sealing ring wear degree data, the unit cumulative load data and the unit running time data. Among these three types of data, the sealing ring wear degree data is the thickness of the sealing ring detected by the sensor, and the unit cumulative load data and the unit running time data are the routine and necessary data of the unit and are easier to obtain; in addition, it is also necessary to obtain the actual running time recorded at each time point in these three data during this period. For example, the sealing ring wear degree data recorded on the first day of this period is 22.7mm (the thickness of the sealing ring), and the sealing ring wear degree data recorded one month apart is 22.5mm. The two data correspond to the actual 11 months and 12 months recorded respectively.

[0056] Specifically, when quantifying the wear of the flat plate seal, it is necessary to find the maximum wear thickness. By measuring the sealing rings that have been replaced in the past, the corresponding weakest position of erosion can be found, and the maximum wear point can be calculated based on the data; taking the measurement data of the flat plate seal thickness replaced in Unit 2 of Yingzhan in 2019 as an example, as shown in Table 1.

[0057] Table 1 Flat plate seal thickness of Yingzhan Unit 2 (base thickness: 22.7 mm) Unit: mm

[0058]

[0059] From the above data, it can be judged that the joint of the two sealing rings is prone to wear; to verify this idea, multiple groups of scrapped sealing rings were measured and analyzed, and finally it was verified that the most prone to serious wear is the joint of the two sealing rings, and the point with the most serious wear is 10.22mm near the joint. At the same time, the maximum wear thickness of the sealing ring must be determined. The area A1 enclosed by the curve function at the maximum wear point is compared with the area B1 of the top cover drainage hole. When A1=B1, it is calculated as the maximum allowable wear area of ​​the seal, thereby determining the position of the maximum wear curve, such as Figure 4 As shown; among them, the area of ​​the top cover pressure-equalizing drainage hole is measured, and multiple groups of scrapped sealing ring data are measured at the same time. The functional relationship is established based on the theory provided by the patent to determine the maximum wear thickness. The maximum wear thickness is 3.39mmmm after reverse calculation. Since the scrapped sealing ring has never reached this thickness, the maximum warning thickness is set to 3.05mm to ensure a certain safety margin. The maximum warning thickness will be used as the maximum wear thickness allowed in the model calculation.

[0060] S2, use the sealing ring wear degree data and the corresponding operating time to train the preset first model, use the unit cumulative load data and the corresponding operating time to train the preset second model, and use the unit operating time data and the corresponding operating time to train the preset third model.

[0061] Among them, three models are trained respectively using three types of data, and when performing running time prediction, predictions are made respectively using the three types of data; specifically, the above-mentioned preset first model, preset second model and preset third model can be bp network models or LSTM models; the training process is: each type of data is input into the corresponding model for processing, and each model outputs the output of the corresponding type of data, which is the running time data, and the input and output of each model are used to calculate the corresponding loss function respectively, until the loss function of each model does not exceed the threshold.

[0062] S3, until the preset first model, the preset second model and the preset third model respectively reach the corresponding training end conditions, the preset first model, the preset second model and the preset third model that respectively reach the corresponding training end conditions are determined as the first prediction model, the second prediction model and the third prediction model respectively.

[0063] Optionally, the training end condition is that the loss functions corresponding to the preset first model, the preset second model, and the preset third model do not exceed a threshold value, and the loss function of the preset first model, the preset second model, or the preset third model is specifically:

[0064]

[0065] Where Loss represents the loss function value, N represents the total amount of data in the seal ring wear degree data, or the unit cumulative load data, or the unit running time data, a represents the sequence number of the data in the seal ring wear degree data, or the unit cumulative load data, or the unit running time data, θ a Indicates the running time of the ath data in the seal ring wear degree data, or the unit cumulative load data, or the unit running time data. It indicates the predicted value outputted after the a-th data is input into the preset first model, the preset second model, or the preset third model.

[0066] S4, obtain the current degree of sealing ring wear, the cumulative load of the unit and the running time of the unit to be tested, and input the degree of sealing ring wear, the cumulative load of the unit and the running time of the unit into the first prediction model, the second prediction model and the third prediction model for processing, respectively, to obtain the first prediction duration, the second prediction duration and the third prediction duration, respectively.

[0067] Among them, each model can be put into use after training is completed. The current degree of sealing ring wear of the unit to be tested, the accumulated load of the unit and the running time of the unit are input into the first prediction model, the second prediction model and the third prediction model for processing respectively, and the first prediction duration, the second prediction duration and the third prediction duration can be obtained respectively.

[0068] S5, multi-source fusion of the first prediction duration, the second prediction duration and the third prediction duration is performed to obtain a fusion result, and the fusion result is determined as a target prediction duration, where the target prediction duration represents the duration that the unit under test can still operate.

[0069] Optionally, the above target prediction duration is specifically:

[0070]

[0071] Wherein, m represents the fusion result, m1 represents the first prediction duration, m2 represents the second prediction duration, m3 represents the third prediction duration, w1 represents the exponential weight of the first prediction duration, w2 represents the exponential weight of the second prediction duration, and w3 represents the exponential weight of the third prediction duration.

[0072] Optionally, the exponential weight of the first predicted duration, the second predicted duration, or the third predicted duration is obtained in the following manner:

[0073] S51, based on the preset initial values ​​of the respective exponential weights, the output values ​​of the preset first model, the preset second model and the preset third model during the training process are preliminarily integrated to obtain a first result.

[0074] S52, using the first result to calculate loss values ​​corresponding to the exponential weights of the first predicted duration, the second predicted duration, and the third predicted duration.

[0075] Optionally, the loss values ​​corresponding to the exponential weights of the first predicted duration, the second predicted duration, and the third predicted duration are specifically:

[0076]

[0077] In the formula, A represents the loss value corresponding to the current exponential weight, m(i) represents the first result, and y i Indicates the runtime corresponding to the first result.

[0078] S53, iteratively update the corresponding exponential weights based on each loss value until the loss value corresponding to each exponential weight reaches the iteration end condition, and use the exponential weights that reach the iteration end condition to perform multi-source fusion on the first prediction duration, the second prediction duration and the third prediction duration to obtain a fusion result.

[0079] Optionally, the corresponding exponential weights are iteratively updated based on each loss value, specifically:

[0080]

[0081] In the formula, w represents the updated exponential weight, wa represents the exponential weight before updating or the preset initial value of the exponential weight, α represents the current learning rate of the preset first model, the preset second model, or the preset third model, represents the current model weight of the preset first model, the preset second model, or the preset third model, and A represents the exponential weight w a The corresponding loss value.

[0082] Embodiment 2:

[0083] The present application embodiment provides a monitoring system for checking and repairing the sealing state of a flat plate, which is applied to a monitoring method for checking and repairing the sealing state of a flat plate in any one of the embodiments 1, such as Figure 2 As shown, including:

[0084] The first module is used to obtain a training data set from the historical operating data of the turbine unit in the same period of time, as well as the operating time from the time point of each data in the training data set during the period to the replacement time of the sealing ring. The training data set includes the sealing ring wear degree data, the unit cumulative load data and the unit operating time data.

[0085] The second module is used to train the preset first model using the sealing ring wear degree data and the corresponding operating time, to train the preset second model using the unit cumulative load data and the corresponding operating time, and to train the preset third model using the unit operating time data and the corresponding operating time.

[0086] The third module is used to determine the preset first model, the preset second model and the preset third model that have reached the corresponding training end conditions as the first prediction model, the second prediction model and the third prediction model respectively until the preset first model, the preset second model and the preset third model respectively reach the corresponding training end conditions.

[0087] The fourth module is used to obtain the current degree of sealing ring wear, the cumulative load of the unit and the running time of the unit to be tested, and input the degree of sealing ring wear, the cumulative load of the unit and the running time of the unit into the first prediction model, the second prediction model and the third prediction model for processing, respectively, to obtain the first prediction duration, the second prediction duration and the third prediction duration, respectively.

[0088] The fifth module is used to perform multi-source fusion of the first prediction duration, the second prediction duration and the third prediction duration to obtain a fusion result, and determine the fusion result as a target prediction duration, which represents the duration that the unit under test can still operate.

[0089] Optionally, the fifth module includes:

[0090] The first submodule is used to preliminarily fuse the output values ​​of the preset first model, the preset second model and the preset third model during the training process based on the preset initial values ​​of each exponential weight and obtain a first result.

[0091] The second submodule is used to calculate the loss values ​​corresponding to the exponential weights of the first predicted duration, the second predicted duration and the third predicted duration by using the first result.

[0092] The third submodule is used to iteratively update the corresponding exponential weights based on each loss value until the loss value corresponding to each exponential weight reaches the iteration end condition, and use the exponential weights that reach the iteration end condition to perform multi-source fusion on the first prediction duration, the second prediction duration and the third prediction duration to obtain a fusion result.

[0093] Embodiment 3:

[0094] The present application embodiment provides an electronic device, such as Figure 3 As shown, it includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, any method in Embodiment 1 is implemented.

[0095] Embodiment 4:

[0096] An embodiment of the present application provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions enable a computer to execute any one of the methods in Example 1.

[0097] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for monitoring the sealing status of a flat plate, characterized in that: The specific steps include: Acquire a training data set from historical operation data of the turbine unit in the same period of time, and the operation time from the time point of each data in the training data set in the period of time to the replacement time of the sealing ring, wherein the training data set includes the wear degree data of the sealing ring, the accumulated load data of the unit and the operation time data of the unit; The preset first model is trained using the seal ring wear degree data and the corresponding operating time, the preset second model is trained using the unit cumulative load data and the corresponding operating time, and the preset third model is trained using the unit running time data and the corresponding operating time; Until the preset first model, the preset second model and the preset third model respectively reach the corresponding training end conditions, the preset first model, the preset second model and the preset third model that respectively reach the corresponding training end conditions are determined as the first prediction model, the second prediction model and the third prediction model respectively; Obtain the current degree of seal ring wear, cumulative load of the unit and running time of the unit for the unit to be tested, and input the degree of seal ring wear, cumulative load of the unit and running time of the unit into the first prediction model, the second prediction model and the third prediction model for processing, respectively, to obtain the first prediction duration, the second prediction duration and the third prediction duration; The first prediction duration, the second prediction duration and the third prediction duration are subjected to multi-source fusion to obtain a fusion result, and the fusion result is determined as a target prediction duration, where the target prediction duration represents the duration during which the unit to be tested can still operate.

2. A method for monitoring the flat plate sealing status inspection according to claim 1, characterized in that: The training end condition is that the loss function corresponding to the preset first model, the preset second model and the preset third model respectively does not exceed the threshold value, and the loss function of the preset first model, the preset second model or the preset third model is specifically: Where Loss represents the loss function value, N represents the total amount of data in the seal ring wear degree data, or the unit cumulative load data, or the unit running time data, a represents the sequence number of the data in the seal ring wear degree data, or the unit cumulative load data, or the unit running time data, θ a Indicates the running time of the ath data in the seal ring wear degree data, or the unit cumulative load data, or the unit running time data. It indicates the predicted value outputted after the a-th data is input into the preset first model, the preset second model, or the preset third model.

3. A method for monitoring the flat plate sealing status inspection according to claim 2, characterized in that: The target prediction duration is specifically: Wherein, m represents the fusion result, m1 represents the first prediction duration, m2 represents the second prediction duration, m3 represents the third prediction duration, w1 represents the exponential weight of the first prediction duration, w2 represents the exponential weight of the second prediction duration, and w3 represents the exponential weight of the third prediction duration.

4. A method for monitoring the flat plate sealing status inspection according to claim 3, characterized in that: The index weight of the first predicted duration, the second predicted duration, or the third predicted duration is obtained in the following manner: Based on the preset initial values ​​of the respective exponential weights, the output values ​​of the preset first model, the preset second model and the preset third model during the training process are preliminarily integrated to obtain a first result; Using the first result, calculate loss values ​​corresponding to the exponential weights of the first predicted duration, the second predicted duration, and the third predicted duration respectively; Based on each loss value, the corresponding exponential weight is iteratively updated until the loss value corresponding to each exponential weight reaches the iteration end condition, and the exponential weight that reaches the iteration end condition is used to perform multi-source fusion on the first prediction duration, the second prediction duration and the third prediction duration to obtain a fusion result.

5. A method for monitoring the flat plate sealing status inspection according to claim 4, characterized in that: The loss values ​​corresponding to the exponential weights of the first prediction duration, the second prediction duration, and the third prediction duration are specifically: In the formula, A represents the loss value corresponding to the current exponential weight, m(i) represents the first result, and y i Indicates the runtime corresponding to the first result.

6. A method for monitoring the flat plate seal status inspection according to claim 5, characterized in that: The corresponding exponential weights are iteratively updated based on each loss value, specifically: In the formula, w represents the updated exponential weight, w a represents the exponential weight before updating or the preset initial value of the exponential weight, α represents the current learning rate of the preset first model, the preset second model, or the preset third model, represents the current model weight of the preset first model, the preset second model, or the preset third model, and A represents the exponential weight w a The corresponding loss value.

7. A monitoring system for flat plate sealing status inspection and maintenance, applied to a monitoring method for flat plate sealing status inspection and maintenance according to any one of claims 1 to 6, characterized in that: include: The first module is used to obtain a training data set from the historical operation data of the turbine unit in the same period of time, and the operation time from the time point of each data in the training data set in the period of time to the replacement time of the sealing ring, and the training data set includes the sealing ring wear degree data, the unit cumulative load data and the unit operation time data; The second module is used to train the preset first model using the sealing ring wear degree data and the corresponding operating time, train the preset second model using the unit cumulative load data and the corresponding operating time, and train the preset third model using the unit running time data and the corresponding operating time; A third module is used for determining the preset first model, the preset second model and the preset third model that respectively reach the corresponding training end conditions as the first prediction model, the second prediction model and the third prediction model respectively until the preset first model, the preset second model and the preset third model respectively reach the corresponding training end conditions; The fourth module is used to obtain the current seal ring wear degree, the cumulative load of the unit and the running time of the unit to be tested, and input the seal ring wear degree, the cumulative load of the unit and the running time of the unit into the first prediction model, the second prediction model and the third prediction model for processing, respectively, to obtain the first prediction time, the second prediction time and the third prediction time; The fifth module is used to perform multi-source fusion of the first prediction duration, the second prediction duration and the third prediction duration to obtain a fusion result, and determine the fusion result as a target prediction duration, wherein the target prediction duration represents the duration that the unit under test can still operate.

8. A monitoring system for flat plate sealing status inspection according to claim 7, characterized in that: The fifth module includes: A first submodule, for preliminarily fusing output values ​​of the preset first model, the preset second model, and the preset third model during the training process based on the preset initial values ​​of the respective exponential weights to obtain a first result; A second submodule is used to calculate the loss values ​​corresponding to the exponential weights of the first predicted duration, the second predicted duration, and the third predicted duration by using the first result; The third submodule is used to iteratively update the corresponding exponential weights based on each loss value until the loss value corresponding to each exponential weight reaches the iteration end condition, and use the exponential weights that reach the iteration end condition to perform multi-source fusion of the first prediction duration, the second prediction duration and the third prediction duration to obtain a fusion result.

9. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method according to any one of claims 1 to 6 is implemented.

10. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores computer instructions, which enable a computer to execute the method of any one of claims 1-6.

Images