A brush seal life prediction method

Abstract

This invention relates to the field of sealing life prediction technology for aero-engines, specifically to a method for predicting the life of brush seals, comprising: establishing a solid model of a forked tube bundle and obtaining the brush filament deformation value and the contact force between the brush filament and the rotor; constructing a fusion dataset; constructing a BP neural network and training it; and predicting the seal life. This invention can accurately predict the wear life of brush seals under different operating conditions.

Description

Technical Field

[0001] This invention relates to the field of sealing life prediction technology for aero-engines, and specifically to a method for predicting the life of brush seals. Background Technology

[0002] As aero engines develop towards higher thrust-to-weight ratios and lower fuel consumption, the operating environment of sealing systems is becoming increasingly demanding. Among various sealing types, brush seals, as a core sealing technology for modern aero engines, gas turbines, and high-end turbomachinery, significantly improve overall efficiency and operational reliability due to their low leakage, high tolerance, and adaptive compensation capabilities. With their flexible brush bristles' adaptive deformation and low leakage capability, they have become the best solution to replace traditional labyrinth seals.

[0003] However, in actual engine operation, the high speed of the main shaft and the high temperature and pressure environment in front of the turbine easily cause the rotor to undergo eccentric motion, thermal expansion, and centrifugal deformation, which in turn interferes with the brush filaments, causing excessive wear on the brush filament tips. This further leads to an increase in the sealing clearance and a shortened seal life. Traditional wear tests are time-consuming and costly; existing brush seal wear tests cannot conduct large-scale, multi-condition coupled tests, and cannot establish an effective database to predict the wear life and sealing performance of brush seals under complex operating conditions.

[0004] Therefore, a method for predicting the lifespan of brush seals is needed to solve the above problems. Summary of the Invention

[0005] To address the problems that existing brush seal simulation methods cannot accurately calculate brush filament wear time, and that existing brush seal wear tests cannot conduct large-scale, multi-condition coupled tests, and cannot establish an effective database to predict the wear life and sealing performance of brush seals under complex working conditions, this invention provides a brush seal life prediction method to solve the existing problems.

[0006] The first aspect of this invention provides a method for predicting the lifespan of a brush seal, employing the following technical solution:

[0007] An integrated three-dimensional compact cross-branch tube bundle solid model was established, including the front baffle, brush bristles, rear baffle, and fluid domain; CFD numerical simulation was performed on the cross-branch tube bundle solid model to obtain the brush bristle deformation value and the contact force between the brush bristles and the rotor;

[0008] The wear length of the brush bristles per unit time is obtained based on the contact force between the brush bristles and the rotor. The geometric model of the worn brush bristles is obtained based on the wear length per unit time. The wear time corresponding to each working condition is obtained by numerical simulation calculation of the geometric model. The first dataset is constructed based on the inlet total pressure, outlet back pressure, rotor speed and wear time corresponding to each working condition under the numerical simulation calculation.

[0009] Numerical simulations were performed iteratively on the worn geometric model under basic operating conditions until all brush filaments were worn down to the interference length. The time taken for all brush filaments to be worn down to the interference length was taken as the wear duration under the basic operating conditions in the numerical simulation iterative calculation. Based on the wear duration corresponding to each operating condition under the numerical simulation calculation, and combined with the equivalent acceleration theory, the acceleration factor was obtained. Based on the acceleration factor and the wear duration corresponding to the basic operating condition under the numerical simulation iterative calculation, the wear duration corresponding to other operating conditions besides the basic operating condition was obtained. A second dataset was constructed based on the wear duration, inlet total pressure, outlet back pressure, and rotor speed corresponding to each operating condition under the numerical simulation iterative calculation.

[0010] The first and second datasets are merged to obtain the merged dataset;

[0011] Construct a BP neural network, and train the neural network on the fused dataset to obtain a trained target neural network;

[0012] The current inlet total pressure, outlet back pressure, and rotor speed are input into the target neural network to predict the wear duration, and the predicted wear duration is used as the seal life.

[0013] A further technical solution of the present invention is that the steps of obtaining the brush filament deformation value and the contact force between the brush filament and the rotor by performing CFD numerical simulation on the solid model of the cross-branch tube bundle are as follows:

[0014] The three-dimensional aerodynamic forces of each brush bristle were obtained by CFD numerical simulation of the solid model of the cross-branch tube bundle.

[0015] The brush filament deformation value and the contact force between the brush filament and the rotor are obtained based on the three-dimensional aerodynamic and mechanical solution model.

[0016] A further technical solution of the present invention is that the step of obtaining the wear length of the brush bristles per unit time based on the contact force between the brush bristles and the rotor is as follows:

[0017] The brush bristle wear volume per unit time is obtained based on the contact force between the bristles and the rotor, combined with the Archard adhesive wear theory.

[0018] The wear length of the brush bristles per unit time is obtained based on the brush bristle wear volume per unit time and the contact area between the brush bristle tip and the rotor surface.

[0019] A further technical solution of the present invention is that the expression for the brush bristle wear volume per unit time is:

[0020]

[0021] In the formula, This indicates the volume of brush bristle wear per unit time. This indicates the contact force between the brush bristles and the rotor; Indicates the hardness of the brush bristles; This indicates the sliding distance between the tip of the brush and the surface of the rotor; This indicates the wear coefficient of the bristle material.

[0022] A further technical solution of the present invention is that the expression of the equivalent acceleration theory is:

[0023]

[0024] In the formula, This indicates the numerical simulation calculation of the first working condition other than the basic working condition. Wear duration corresponding to other operating conditions; This indicates the wear duration corresponding to the basic operating condition under numerical simulation calculation; This represents the entropy increase rate corresponding to the basic operating condition under numerical simulation calculation; This indicates the numerical simulation calculation of the first working condition other than the basic working condition. The entropy increase rate corresponding to other operating conditions.

[0025] A further technical solution of the present invention is to obtain the entropy increase rate generated by wear under various working conditions based on the heat generated by the wear system, the temperature of the brush matrix, and the wear time.

[0026] A further technical solution of the present invention is that the expression for the heat generated by the wear system during the brush bristle wear process is:

[0027]

[0028] In the formula, This indicates the heat generated by the wear system during the brush bristle wear process; Indicates the coefficient of friction; This indicates the contact force between the brush bristles and the rotor; This indicates the rotor linear velocity.

[0029] A further technical solution of the present invention is as follows: the steps of obtaining the acceleration factor based on the wear time corresponding to each working condition under numerical simulation calculation and combined with the equivalent acceleration theory are as follows:

[0030]

[0031] In the formula, Indicates the acceleration factor; This indicates the numerical simulation calculation of the first working condition other than the basic working condition. Wear duration corresponding to other operating conditions; This represents the wear duration corresponding to the basic operating condition under numerical simulation calculation.

[0032] A further technical solution of the present invention is that the steps for constructing a BP neural network are as follows:

[0033] Construct the initial BP neural network;

[0034] The PSO algorithm is used to optimize the BP neural network to obtain the optimized BP neural network.

[0035] The beneficial effects of this invention are:

[0036] This invention obtains the wear duration of brush bristles under basic working conditions through numerical simulation and iterative calculation. Then, based on the principle of equivalent acceleration, a fast calculation method is used to obtain the acceleration factor. Based on the acceleration factor and the wear duration of brush bristles under basic working conditions, the wear duration of brush bristles under other working conditions is obtained through numerical iterative calculation. At this point, a fusion dataset for all working conditions can be established. The fusion dataset includes the wear duration of the entire working condition. That is, a high-precision BP neural network model can be obtained by training a BP neural network using the fusion dataset. The trained neural network model can accurately predict the wear life of brush seals under different working conditions. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 This is a flowchart illustrating a brush seal life prediction method according to the present invention.

[0039] Figure 2 This is a detailed flowchart of a brush seal life prediction method according to the present invention;

[0040] Figure 3 This is a schematic diagram of the structure of the cross-shaped tube bundle solid model in an embodiment of the present invention;

[0041] Figure 4 This is a simulation calculation diagram of the cross-shaped tube bundle solid model in an embodiment of the present invention;

[0042] Figure 5 for Figure 4 A magnified view of the middle brush bristles;

[0043] Figure 6 This is a pressure cloud diagram of the brush seal wear process in an embodiment of the present invention.

[0044] In the diagram: 1. Front baffle; 2. Rear baffle; 3. Brush bristles; 4. Rotor. Detailed Implementation

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] This invention provides an embodiment of a brush seal life prediction method. This embodiment addresses the problems of existing brush seal simulation methods failing to accurately calculate brush filament wear time, and existing brush seal wear tests being unable to conduct large-scale, multi-condition coupled tests, thus failing to establish an effective database for predicting the wear life and sealing performance of brush seals under complex operating conditions. Figure 1 As shown, this embodiment provides a method for predicting the lifespan of a brush seal, the method comprising:

[0047] S1. Establish a solid model of the forked tube bundle and obtain the filament deformation value and the contact force between the filament and the rotor;

[0048] Specifically, such as Figure 3 As shown, an integrated three-dimensional compact staggered tube bundle solid model is established, including front baffle 1, brush bristles 3, rear baffle 2, and fluid domain. The tilt angle of the brush bristles is shown. The angle is 40°; CFD numerical simulation was performed on the solid model of the cross-branch tube bundle to obtain the filament deformation value and the contact force between the filament 3 and the rotor 4.

[0049] For example, in one specific embodiment, the steps for establishing an integrated three-dimensional compact cross-branch tube bundle solid model including a front baffle, brush filaments, a rear baffle, and a fluid domain are as follows: using a mechanical analysis model to solve for the brush filament coordinates of the brush filament model under a certain interference amount, inputting the brush filament coordinates into three-dimensional modeling software, and establishing an integrated three-dimensional compact cross-branch tube bundle solid model including a front baffle, brush filaments, rear baffle, and fluid domain.

[0050] For example, in one specific embodiment, the steps of obtaining the filament deformation value and the contact force between the filament and the rotor by performing CFD numerical simulation on the solid model of the cross-branch tube bundle are as follows: performing CFD numerical simulation on the solid model of the cross-branch tube bundle to obtain the three-dimensional aerodynamic force of each filament; obtaining the filament deformation value and the contact force between the filament and the rotor based on the three-dimensional aerodynamic force and mechanical solution model.

[0051] The CFD numerical simulation steps are as follows: Figure 4As shown, the computational domain of the cross-branch tube bundle solid model was meshed using ICEM software with hexahedral meshes. Then, numerical calculations were performed using the commercial software Fluent. A pressure-based solver was used, with the Simple algorithm as the pressure-velocity coupling algorithm. The fluid medium was an ideal gas, and the standard k-ε turbulence model was employed. The standard k-ε model conforms to the high Reynolds number characteristics of the internal flow of the brush seal, and can predict leakage characteristics and flow details with reasonable accuracy. Furthermore, it has high computational efficiency and strong engineering applicability. Standard wall functions were used at the wall surfaces, and the equation discretization scheme was a second-order upwind scheme. Pressure inlets and outlets were set, and periodic boundaries were used in the circumferential direction to obtain the flow field distribution of the cross-branch tube bundle solid model. A self-compiled UDF program was used to extract the three-dimensional aerodynamic distribution of each brush filament from the flow field distribution. Figure 5 for Figure 4 A magnified view of a section of the medium brush bristles.

[0052] In one specific embodiment, the steps for obtaining the bristle deformation value and the contact force between the bristles and the rotor based on the three-dimensional aerodynamic and mechanical solution model are as follows: The mechanical solution model can comprehensively analyze the various forces acting on the bristles at one time (the three-dimensional aerodynamic force of the airflow on the bristles, the solid contact force and friction between the bristles, the contact normal force and friction between the bristle tip and the shaft, and the contact force and friction between the rear bristles and the rear baffle), achieving a fully coupled analysis of the bristle deformation and obtaining the bristle deformation value; by inputting the three-dimensional aerodynamic force into the mechanical solution model, the contact force between the bristles and the rotor can be obtained.

[0053] S2. Construct a fused dataset;

[0054] Specifically, the wear length of the brush bristles per unit time is obtained based on the contact force between the brush bristles and the rotor. Based on this wear length, a geometric model of the worn brush bristles is obtained. Numerical simulations are performed on this geometric model to obtain the wear duration for each operating condition. A first dataset is constructed based on the inlet total pressure, outlet back pressure, rotor speed, and wear duration for each operating condition under the numerical simulations. Under the basic operating condition, iterative numerical simulations are performed on the worn geometric model until all brush bristles are worn down to the interference length. The time taken to wear all brush bristles down to the interference length is taken as the wear duration for the basic operating condition under the iterative numerical simulations. An acceleration factor is obtained based on the wear duration for each operating condition under the numerical simulations and combined with equivalent acceleration theory. Based on the acceleration factor and the wear duration for the basic operating condition under the iterative numerical simulations, the wear durations for other operating conditions besides the basic operating condition are obtained. A second dataset is constructed based on the wear duration, inlet total pressure, outlet back pressure, and rotor speed for each operating condition under the iterative numerical simulations. The first and second datasets are then merged to obtain a fused dataset.

[0055] S21. Construct the first dataset;

[0056] For example, in one specific embodiment, the step of obtaining the wear length of the brush bristles per unit time based on the contact force between the brush bristles and the rotor is as follows: based on the contact force between the brush bristles and the rotor, and combined with the Arcard adhesive wear theory, the wear volume of the brush bristles per unit time is obtained; based on the wear volume of the brush bristles per unit time and the contact area between the brush bristle tip and the rotor surface, the wear length of the brush bristles per unit time is obtained.

[0057] The wear between the brush bristles of the brush seal and the rotor raceway surface is a typical example of adhesive wear. Therefore, the contact force is substituted into the Archard adhesive wear theory formula to calculate the wear volume, that is, the brush bristle wear volume per unit time. for:

[0058]

[0059] In the formula, This indicates the volume of brush bristle wear per unit time. This indicates the contact force between the brush bristles and the rotor; Indicates the hardness of the brush bristles; This indicates the sliding distance between the tip of the brush and the surface of the rotor; This represents the wear coefficient of the bristle material; in this embodiment, the bristle hardness... The value is 550 MPa; wear coefficient The value is 10 -5 .

[0060] For example, in one specific embodiment, the wear length of the brush bristles per unit time is obtained based on the brush bristle wear volume per unit time and the contact area between the brush bristle tip and the rotor surface. That is, the wear length of the brush bristles per unit time can be obtained by dividing the brush bristle wear volume per unit time by the contact area between the brush bristle tip and the rotor surface.

[0061] For example, in one specific embodiment, a geometric model of the brush bristles after wear is obtained based on the wear length per unit time, and the wear time corresponding to each working condition is obtained by numerical simulation calculation of the geometric model.

[0062] At this point, the time taken for all brush filaments to wear down to the interference length can be used as the wear duration under the basic working condition in the numerical simulation iterative calculation; the inlet total pressure, outlet back pressure, and rotor speed corresponding to each working condition in the numerical simulation calculation are used as input parameters, and the wear duration under each working condition in the numerical simulation calculation is used as output parameters to construct the first dataset.

[0063] S22. Construct the second dataset;

[0064] For example, in one specific embodiment, the step of performing numerical simulation iterative calculations on the worn geometric model under basic operating conditions until all brush bristles are worn down to the interference length, and taking the time taken to wear down all brush bristles to the interference length as the wear duration under the basic operating conditions in the numerical simulation iterative calculation, is as follows: for the wear end time of each brush bristle during the wear process... All of these require judgment: According to the Archard wear theory (Archard wear equation), when the first... Wear volume between the brush bristles and the rotor Volume occupied by all interference quantities When the wear time is equal, the bristles will stop wearing. The wear time at this point is recorded as _____. When the wear volume of each row of brush bristles meets the requirements If the filament bundle is completely worn out, the wear iteration ends. This criterion only requires one scalar comparison to achieve high-precision tracking of the entire wear process. Thus, the wear duration under the basic working condition calculated by numerical simulation can be obtained.

[0065] For example, in one specific embodiment, the step of obtaining the heat generated by the wear system during the brush bristle wear process is as follows:

[0066]

[0067] In the formula, This indicates the heat generated by the wear system during the brush bristle wear process; Indicates the coefficient of friction; This indicates the contact force between the brush bristles and the rotor; This indicates the rotor linear velocity.

[0068] For example, in one specific embodiment, the entropy increase rate generated by wear under various operating conditions is obtained based on the heat generated by the wear system, the temperature of the brush matrix, and the wear duration.

[0069] Based on the equivalent acceleration method, which uses cumulative failure probability as the equivalent quantity for product lifespan and reliability, the acceleration principle is to change the probability distribution function of product failure over time by increasing the operating parameters of the test, thereby accelerating the accumulation rate of the product failure probability. That is, when the product reaches the failure probability accumulated under high operating conditions in a shorter timeframe compared to the failure probability accumulated under normal operating conditions, the two processes are considered equivalent. In this embodiment, by transferring the equivalent acceleration formula to the dissipative wear model, the equivalent quantity of wear life can be transformed from a statistical probability into a deterministic entropy increment, i.e.:

[0070]

[0071] In the formula, This indicates the numerical simulation calculation of the first working condition other than the basic working condition. Wear duration corresponding to other operating conditions; This indicates the wear duration corresponding to the basic operating condition under numerical simulation calculation; This represents the entropy increase rate corresponding to the basic operating condition under numerical simulation calculation; This indicates the numerical simulation calculation of the first working condition other than the basic working condition. The entropy increase rate corresponding to other operating conditions; This indicates the numerical simulation calculation of the first working condition other than the basic working condition. The entropy increment corresponding to each of the other operating conditions; This represents the entropy increment corresponding to the basic operating condition under numerical simulation calculation. In this embodiment, the entropy increase rate of other operating conditions in the wear test is greater than the entropy increase rate of the basic operating condition, and the corresponding wear duration for these other operating conditions is... Wear duration compared to the basic operating condition When the wear test is shortened, the other conditions of the wear test are considered as accelerated conditions of the wear test, that is, the accelerated effect of the wear test is achieved. Therefore, as long as the entropy increase rate is the same as the product of the wear time, that is, the entropy increment is the same, it can be regarded as equivalent acceleration. Since there are errors in the test process, the entropy increment error must be less than 5%.

[0072] Table 1 shows the brush bristle wear time obtained from numerical simulation iterative calculation under the equivalent acceleration theory; Table 2 shows the brush bristle wear time obtained from numerical simulation calculation under the equivalent acceleration theory; and Table 3 shows the acceleration factor error under the two calculation methods. As can be seen from Table 3, when both calculation methods use the acceleration factor from the equivalent acceleration method, the maximum error of the acceleration factor under the same working condition is 4.7%, which does not exceed 5%. Therefore, the wear time under other working conditions can be solved by combining the acceleration factor obtained from numerical simulation calculation under the equivalent acceleration theory with the basic working condition of the numerical simulation iterative calculation, which significantly improves computational efficiency.

[0073] Table 1

[0074]

[0075] Table 2

[0076]

[0077] Table 3

[0078]

[0079] The wear time is directly calculated by evenly distributing the contact force on the brush bristles of the cross-branch tube bundle solid model to each row of brush bristles. The result obtained is significantly shorter than that obtained by iterative calculation. As shown in Table 3, the acceleration factor obtained by numerical simulation under the equivalent acceleration theory is almost identical to that obtained by numerical simulation iterative calculation. That is, the maximum error between the acceleration factor obtained by numerical simulation and numerical simulation iterative calculation does not exceed 5% (the maximum error in Table 3 is 4.7%). Therefore, the wear time of other working conditions under numerical simulation iterative calculation can be solved by the acceleration factor. That is, in this embodiment, if the entropy increment is the same under different working conditions or calculation methods, it is considered as equivalent acceleration. Therefore, the basic working condition and other working conditions are equivalent, and the expression of the acceleration factor is:

[0080]

[0081] In the formula, Indicates the acceleration factor. This indicates the wear duration corresponding to the basic operating condition under numerical simulation calculation; This indicates the numerical simulation calculation of the first working condition other than the basic working condition. Wear duration corresponding to other operating conditions; This represents the wear duration corresponding to the basic operating condition under numerical simulation iterative calculation; This represents the number of iterations in numerical simulation, excluding the basic operating condition. The wear duration corresponding to other operating conditions can be calculated by solving the acceleration factor and the wear duration of the basic operating condition under numerical simulation iterative calculation.

[0082] For example, in this embodiment, the acceleration factor can be obtained based on the wear duration corresponding to each working condition under numerical simulation calculation. The acceleration factor and the wear duration corresponding to the basic working condition under numerical simulation iterative calculation are used, and the wear duration corresponding to other working conditions besides the basic working condition under numerical simulation iterative calculation is obtained by combining the equivalent theory. Thus, the second dataset can be constructed with the wear duration corresponding to each working condition under numerical simulation iterative calculation as the output data and the inlet total pressure, outlet back pressure, and rotor speed as the input parameters.

[0083] Thus, in this embodiment, the first dataset and the second dataset are merged to obtain the merged dataset.

[0084] S3. Construct a BP neural network and train it;

[0085] Specifically, a backpropagation (BP) neural network is constructed, and the neural network is trained based on a fused dataset to obtain a trained target neural network.

[0086] For example, in one specific embodiment, the steps for constructing a BP neural network are: constructing an initial BP neural network; and optimizing the BP neural network using the PSO algorithm to obtain an optimized BP neural network. Specifically, in this embodiment, the root mean square error (RMSE) is used as the core error metric, and the coefficient of determination (R²) is used. 2 This is used to evaluate the performance and quality of BP neural networks.

[0087] S4. Predicting seal life;

[0088] Specifically, the current inlet total pressure, outlet back pressure, and rotor speed are input into the target neural network to predict the wear duration, and the predicted wear duration is used as the seal life.

[0089] The present invention will be described below with reference to specific data:

[0090] Figure 6 The pressure cloud map shows the specific wear condition of the brush bristles at different time points. Through iterative calculation, the wear condition and flow characteristics of the brush bristles at different time points can be observed. At the same time, the aerodynamic force and tip contact force on the brush bristles at that time point can be extracted, so as to update the force and deformation in real time and calculate the wear life of the brush bristles more accurately.

[0091] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for predicting the lifespan of a brush seal, characterized in that, include: An integrated three-dimensional compact cross-branch tube bundle solid model was established, including the front baffle, brush bristles, rear baffle, and fluid domain; CFD numerical simulation was performed on the cross-branch tube bundle solid model to obtain the brush bristle deformation value and the contact force between the brush bristles and the rotor; The wear length of the brush bristles per unit time is obtained based on the contact force between the brush bristles and the rotor. The geometric model of the brush bristles after wear is obtained based on the wear length per unit time. The wear time corresponding to each working condition is obtained by numerical simulation calculation of the geometric model. The first dataset is constructed based on the inlet total pressure, outlet back pressure, rotor speed and wear duration under various operating conditions calculated by numerical simulation. Numerical simulation and iterative calculation were performed on the worn geometric model under the basic working condition until all brush filaments were worn down to the interference length. The time taken to wear down all brush filaments to the interference length was taken as the wear time under the basic working condition in the numerical simulation and iterative calculation. The wear duration under various working conditions is calculated based on numerical simulation, and the acceleration factor is obtained by combining the equivalent acceleration theory. Based on the wear duration corresponding to the basic operating condition under the acceleration factor and numerical simulation iterative calculation, the wear duration corresponding to other operating conditions besides the basic operating condition under the numerical simulation iterative calculation is obtained. A second dataset is constructed based on the wear duration, inlet total pressure, outlet back pressure, and rotor speed corresponding to each operating condition under the numerical simulation iterative calculation; the expression of the equivalent acceleration theory is: In the formula, This indicates the numerical simulation calculation of the first working condition other than the basic working condition. Wear duration corresponding to other operating conditions; This indicates the wear duration corresponding to the basic operating condition under numerical simulation calculation; This represents the entropy increase rate corresponding to the basic operating condition under numerical simulation calculation; This indicates the numerical simulation calculation of the first working condition other than the basic working condition. The steps for obtaining the entropy increase rate corresponding to other working conditions, based on the heat generated by the wear system, the temperature of the brush matrix, and the wear duration, and for obtaining the acceleration factor based on the wear duration corresponding to each working condition through numerical simulation and combined with the equivalent acceleration theory, are as follows: In the formula, Indicates the acceleration factor; This indicates the numerical simulation calculation of the first working condition other than the basic working condition. Wear duration corresponding to other operating conditions; This indicates the wear duration corresponding to the basic operating condition under numerical simulation calculation; The first and second datasets are merged to obtain the merged dataset; Construct a BP neural network, and train the neural network on the fused dataset to obtain a trained target neural network; The current inlet total pressure, outlet back pressure, and rotor speed are input into the target neural network to predict the wear duration, and the predicted wear duration is used as the seal life.

2. The method for predicting the lifespan of a brush seal according to claim 1, characterized in that, The steps for obtaining the brush filament deformation value and the contact force between the brush filament and the rotor through CFD numerical simulation of the cross-branch tube bundle solid model are as follows: The three-dimensional aerodynamic forces of each brush bristle were obtained by CFD numerical simulation of the solid model of the cross-branch tube bundle. The brush filament deformation value and the contact force between the brush filament and the rotor are obtained based on the three-dimensional aerodynamic and mechanical solution model.

3. The method for predicting the lifespan of a brush seal according to claim 1, characterized in that, The steps for obtaining the wear length of the brush bristles per unit time based on the contact force between the bristles and the rotor are as follows: The brush bristle wear volume per unit time is obtained based on the contact force between the bristles and the rotor, combined with the Archard adhesive wear theory. The wear length of the brush bristles per unit time is obtained based on the brush bristle wear volume per unit time and the contact area between the brush bristle tip and the rotor surface.

4. The method for predicting the lifespan of a brush seal according to claim 3, characterized in that, The expression for the bristle wear volume per unit time is: In the formula, This indicates the volume of brush bristle wear per unit time. This indicates the contact force between the brush bristles and the rotor; Indicates the hardness of the brush bristles; This indicates the sliding distance between the tip of the brush and the surface of the rotor; This indicates the wear coefficient of the bristle material.

5. The method for predicting the lifespan of a brush seal according to claim 1, characterized in that, The expression for the heat generated by the wear system during the brush bristle wear process is: In the formula, This indicates the heat generated by the wear system during the brush bristle wear process; Indicates the coefficient of friction; This indicates the contact force between the brush bristles and the rotor; This indicates the rotor linear velocity.

6. The method for predicting the lifespan of a brush seal according to claim 1, characterized in that, The steps to construct a BP neural network are as follows: Construct the initial BP neural network; The PSO algorithm is used to optimize the BP neural network to obtain the optimized BP neural network.

Images