A high-temperature bearing test platform and a bearing fault diagnosis and life prediction method

Abstract

The application discloses a high-temperature bearing test platform and a bearing fault diagnosis and life prediction method, and comprises a thrust oil cylinder, a thrust sensor, a heat insulation layer, sliding sealing, a measured mechanism and a frame, rotary sealing, a heat insulation layer, a static pressure bearing, a shaft coupling, a dynamic torque sensor, a torque controller and a high-speed motor. The specific method comprises the following steps: S100, multi-modal data synchronous acquisition; S200, high-temperature drift compensation processing; S300, multi-dimensional feature extraction and fusion; and S400, construction of a physical constraint neural network (PINN) prediction model. The application can simulate deep earth, deep sea and extreme high-temperature environments, has reliable failure protection capability, strong anti-interference capability, accurate fault feature extraction, and high wear amount and life prediction precision.

Description

Technical Field

[0001] This invention relates to the field of mechanical equipment health monitoring technology, and in particular to a high-temperature bearing testing platform and a bearing fault diagnosis and life prediction method. Background Technology

[0002] The thrust bearing in a submersible electric pump protector must withstand the enormous axial forces generated by the pump's operation. As oil extraction progresses towards ultra-deep wells, thrust bearings often face extreme conditions: temperatures of 400°C, heavy loads of 30 tons, and high speeds of 12,000 rpm. Existing testing platforms have limitations in simulating complex operating conditions and in terms of parameter measurement accuracy. In particular, single-sensor monitoring is susceptible to thermal drift interference at high temperatures, making it difficult to accurately capture the physical wear state of the bearing.

[0003] Current technological status indicates that:

[0004] Regarding the maximum temperature resistance: most existing domestic testing devices operate within the range of 220℃ to 250℃ (e.g., the measured bottom temperature of Deep Earth Tak 1 well is about 220℃, and the range of the experimental system of Shanghai Zhongren Company is 250℃); foreign literature and advanced laboratories speculate that it can achieve about 300℃, and a few geothermal wells or ultra-deep well simulation platforms may cover even higher temperatures.

[0005] In terms of maximum load capacity: typical domestic equipment such as the Dongfang Electric test bench has a static load of about 20 tons, and the Gemstone Machinery platform has a dynamic load of about 13 tons; foreign high-precision hydraulic system test benches can reach more than 25 tons.

[0006] Regarding the maximum speed: domestically, it is mostly 1500 r / min (high-voltage variable frequency drive), and some laboratories have achieved short-term high-speed simulation; while similar international equipment can stably reach more than 2000 r / min to adapt to turbine drill bearings and high-speed drilling scenarios.

[0007] In summary, existing thrust bearing testing platforms, both domestically and internationally, still have significant limitations in terms of temperature resistance, load-bearing capacity, and speed performance, making it difficult to cover the operational boundaries of submersible motor thrust bearings under extreme conditions. Especially under high-temperature conditions, the sealing system at the connection between the test chamber and the main shaft is a weak point. Traditional sealing solutions often employ single-lip rubber or mechanical seals, which have the following problems:

[0008] (1) The rubber seals are not heat resistant enough and are prone to heat aging above 220℃;

[0009] (2) Mechanical seals have complex structures and are sensitive to thermal expansion and axial runout;

[0010] (3) Lack of high-temperature elastic compensation design, sealing gap is prone to loss of control;

[0011] (4) Frictional heat generation is concentrated, and the sealing lip is prone to burning or leakage after losing lubrication.

[0012] Apart from rotary seals, hydrostatic bearings achieve contactless load bearing by forming a pressure oil film through external oil supply, and are widely used in heavy-duty high-speed equipment. However, they have significant drawbacks under extreme operating conditions:

[0013] (1) Temperature resistance bottleneck: Traditional hydrostatic bearings cause the spindle and hydraulic oil to heat up at high speeds, and the material strength decreases by more than 40% at high temperatures, making them unsuitable for the high-temperature environment of submersible electric pump testing;

[0014] (2) Load-bearing synergistic failure: High temperature reduces the viscosity of lubricating oil to 1 / 5 of that at room temperature. Traditional straight-hole oil passages are difficult to form a stable oil film. Under a 30T load and a speed of 12000r / min, the oil film is prone to breakage.

[0015] (3) No protection against failure: When the hydrostatic system fails, there is no immediate load-bearing mechanism. The impact load on the shaft directly causes damage to the components. The existing design load limit is mostly ≤30T, which cannot cope with instantaneous overload. Summary of the Invention

[0016] This invention aims to provide a high-temperature bearing testing platform and a bearing fault diagnosis and life prediction method, which can simulate deep-earth, deep-sea and extreme high-temperature environments, has reliable failure protection capabilities, strong anti-interference capabilities, accurate fault feature extraction, and high accuracy in wear and life prediction.

[0017] To achieve the above objectives, the present invention employs the following technical solution:

[0018] This invention discloses a high-temperature bearing testing platform, comprising a thrust cylinder, a thrust sensor, a heat insulation layer, a sliding seal, a tested mechanism and frame, a rotary seal, a heat insulation layer, a hydrostatic bearing, a coupling, a dynamic torque sensor, a torque controller, and a high-speed motor.

[0019] The rotary seal includes a sealing housing, a rotary sleeve fitted inside the sealing housing, a main sealing lip between the rotary sleeve and the head end of the sealing housing, a retaining ring coaxially provided at the tail end of the sealing housing, and an elastic compensation mechanism coaxially fixed in the inner cavity of the tail end of the sealing housing. The elastic compensation mechanism includes a base, the base is fixedly connected to the sealing housing, and the base is connected to a push ring through a wave-shaped elastic compensation component. The push ring pushes the auxiliary sealing lip to tightly adhere to the main sealing lip.

[0020] Preferably, the base is fixed to the sealing housing by positioning screws.

[0021] Preferably, the two ends of the waveform elastic compensation component are fixedly connected to the base and the push ring respectively by high-temperature brazing.

[0022] Preferably, the push ring is slidably connected to the side wall of the base.

[0023] Preferably, the main sealing lip is made of ceramic fiber reinforced graphite material, the auxiliary sealing lip is made of nickel-based alloy coated with polytetrafluoroethylene composite structure, and the waveform elastic compensation component is made of Inconel 718 alloy.

[0024] Preferably, the hydrostatic bearing includes a horizontally arranged hydrostatic bearing body, with end caps fixed at both ends of the hydrostatic bearing body, a hydrostatic bearing spindle sleeved inside the hydrostatic bearing body, and the end caps connected to both ends of the hydrostatic bearing spindle via rolling bearings. The connection between the rolling bearings and the end caps is provided with reverse-threaded sealing threads. A positioning boss is provided at one end of the hydrostatic bearing body, and a thrust boss is provided in the middle of the hydrostatic bearing spindle. The thrust boss is adapted to the positioning boss, and a metal seal is provided between the thrust boss and the positioning boss.

[0025] A bushing is fitted around the outer periphery of the middle part of the hydrostatic bearing spindle. An annular oil cavity is formed between the bushing and the hydrostatic bearing body. An oil inlet is provided at the top of the hydrostatic bearing body, and an oil outlet is provided at the bottom of the hydrostatic bearing body. The annular oil cavity is connected to the oil inlet and the oil outlet.

[0026] Preferably, the rolling bearing is a high-manganese alloy steel thrust bearing.

[0027] Preferably, the inner wall of the hydrostatic bearing body has an annular stepped structure.

[0028] Preferably, the bushing has an annular groove on its outer periphery and branch channels are uniformly provided on the inner periphery of the bushing. The branch channels are connected to the annular groove and are arranged radially. The end of the branch channels is a throttling orifice.

[0029] Preferably, the hydrostatic bearing body is integrally forged from 42CrMo high-temperature alloy.

[0030] Preferably, a cooling jacket is provided on the outside of the hydrostatic bearing body, and a spiral cooling channel is provided between the cooling jacket and the hydrostatic bearing body.

[0031] This invention also discloses a bearing fault diagnosis and life prediction method using the above-mentioned testing platform, comprising the following steps:

[0032] S100, Multimodal Data Synchronous Acquisition: Using a high-temperature testing platform, the potential difference signal, vibration signal, ultrasonic wall thickness signal, and friction torque signal of the thrust bearing are synchronously acquired within the range of 0~400℃.

[0033] S200, High-Temperature Drift Compensation: Nonlinear calibration of ultrasonic velocity and vibration sensor sensitivity is performed using real-time temperature data collected by a K-type thermocouple;

[0034] S300, Multidimensional Feature Extraction and Fusion: Using ultrasonic sensors to monitor the change in the wall thickness of the thrust bearing in real time as a static feature of the bearing's physical wear, and extracting the time-frequency domain features of the vibration signal as a dynamic feature;

[0035] S400. Construct a Physically Constrained Neural Network (PINN) prediction model: Input dynamic features and static wear values ​​into the PINN neural network, embed Archard's wear law into the network's loss function as a hard constraint term, and through collaborative training based on physical mechanisms and data-driven approaches, output the bearing's fault diagnosis conclusion and remaining service life.

[0036]

[0037] in The constrained prediction results must conform to the laws of physical degradation:

[0038]

[0039] in, For wear volume, For real-time load, The sliding distance, The value represents the material hardness.

[0040] Preferably, in step S100, a measuring electrode matrix is ​​arranged in the corresponding area of ​​the bearing under test, a pair of current excitation electrodes are arranged at both ends of the bearing under test, and a vibration sensor is installed at the bottom of the hydrostatic bearing to reduce the thermal load through an isolation structure.

[0041] Preferably, in step S200, a second-order sound velocity compensation model based on the real-time temperature gradient T is established:

[0042]

[0043] in, Reference temperature The speed of sound at that time and This is the pre-calibrated temperature sensitivity coefficient.

[0044] Preferably, in step S300, at a speed of 12000... Under operating conditions, variational mode decomposition is used to denoise the vibration signal and extract time-frequency domain features as dynamic features to characterize the instantaneous operating state of the bearing.

[0045] Preferably, in step S300, a multimodal health index is defined ( This method nonlinearly couples static and dynamic features, suppresses high-frequency random noise through a logarithmic function, and amplifies the degradation trend in the later, severe wear stage through a power function, thereby achieving an accurate characterization of the entire bearing life cycle. The formula is as follows:

[0046]

[0047] in, For multimodal health index, The characteristic value of the signal energy. This is the baseline energy under initial healthy conditions. For the signal energy term weights, Weighting of defect / wear size item. This refers to the actual amount of wear or the depth of the defect. This refers to the allowable wear threshold or failure threshold.

[0048] Preferably, the physical constraint process in step S400 uses the measured physical wear as a dynamic correction factor to correct the bearing degradation trajectory deduced from the vibration signal characteristics in real time.

[0049] Preferably, in step S300, the change in the thrust bearing wall thickness is manifested as a wall thickness reduction with an accuracy of 0.01 mm.

[0050] The beneficial effects of this invention are:

[0051] This invention can solve the problems of difficulty in extracting fault features, inability to monitor wear in real time and quantitatively, and low accuracy of life prediction for thrust bearings under high temperature and strong interference environment of 400℃.

[0052] The rotary seal in this invention can adapt to the 400℃ high-temperature working conditions of the thrust bearing testing system and improves the temperature resistance by more than 60%, the load capacity by 50%, and the rotational speed by more than 5 times. The component adopts a double-lip composite structure and waveform elastic compensation, achieving strong structural integration and an integrated design of heat insulation, sealing, compensation, and cooling. It is suitable for rotating components in deep earth, deep sea, and extreme high temperature environments.

[0053] This invention achieves breakthroughs in high-temperature resistance (42CrMo alloy exhibits stable performance at high temperatures, solving the problem of high-temperature softening in traditional bearings), optimized load-bearing capacity (high-temperature oil film thickness is stabilized at 0.07~0.1mm, achieving synergistic load bearing of 30T / 12000r / min), and reliable failure protection (seamless connection of dual bearings with a response time ≤0.02s, and 42T rated load can withstand instantaneous impact). Attached Figure Description

[0054] Figure 1 This is a schematic diagram of a high-temperature testing platform;

[0055] Figure 2 A cross-sectional view of the rotary seal;

[0056] Figure 3 Exploded view of the front of the rotary seal;

[0057] Figure 4 Exploded view of the rear of the rotary seal;

[0058] Figure 5This is a schematic diagram of a rotary seal;

[0059] Figure 6 This is a front sectional view of a hydrostatic bearing;

[0060] Figure 7 This is an axial sectional view of a hydrostatic bearing;

[0061] Figure 8 This is a schematic diagram of the bearing being tested.

[0062] Figure 9 This is a flowchart of a lifetime prediction algorithm based on multi-source information fusion.

[0063] In the diagram: 1-Thrust cylinder, 2-Thrust sensor, 3-Insulation layer, 4-Sliding seal, 5-Measured mechanism and frame, 6-Rotary seal, 601-Sealing housing, 602-Main sealing lip, 603-Base, 604-Auxiliary sealing lip, 605-Thrust ring, 606-Rotary bushing, 607-Retaining ring, 608-Wave elastic compensation component, 609-Positioning screw, 7-Insulation layer, 8-Hydrostatic bearing, 801-Hydrostatic bearing body, 802-End cap 803-Rolling bearing, 804-Reverse thread seal, 805-Positioning boss, 806-Thrust boss, 807-Metal seal, 808-Hydrostatic bearing spindle, 809-Shaft sleeve, 810-Annular oil cavity, 811-Oil inlet, 812-Oil outlet, 813-Annular groove, 814-Branch flow channel, 9-Coupling, 10-Dynamic torque sensor, 11-Torque controller, 12-High-speed motor, 13-Excitation electrode, 14-Measuring electrode. Detailed Implementation

[0064] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings.

[0065] like Figure 1 As shown, the present invention discloses a high-temperature bearing testing platform, including a thrust cylinder 1, a thrust sensor 2, a heat insulation layer 3, a sliding seal 4, a test mechanism and frame 5, a rotary seal 6, a heat insulation layer 7, a hydrostatic bearing 8, a coupling 9, a dynamic torque sensor 10, a torque controller 11, and a high-speed motor 12.

[0066] In this test platform, the rotary seal 6 includes a sealing housing 601, a rotary bushing 606 fitted inside the sealing housing 601, a main sealing lip 602 between the rotary bushing 606 and the head end of the sealing housing 601, a retaining ring 607 coaxially provided at the tail end of the sealing housing 601, and an elastic compensation mechanism coaxially fixed in the inner cavity of the tail end of the sealing housing 601. The elastic compensation mechanism includes a base 603, which is fixedly connected to the sealing housing 601. The base 603 is connected to a push ring 605 through a wave-shaped elastic compensation component 608. The push ring 605 pushes the auxiliary sealing lip 604 to tightly adhere to the main sealing lip 602.

[0067] The main sealing lip 602, the rotating bushing 606, and the retaining ring 607 are arranged in close proximity on the main shaft from left to right. The left end face of the main sealing lip 602 is the medium end, and the main shaft extends beyond the main sealing lip 602. The distance of the extension depends on the situation. The right end face of the retaining ring 607 is the air end, which is connected to the heat insulation and cooling system. The retaining ring 607 has a positioning device to ensure that the main sealing lip 602, the rotating bushing 606, and the retaining ring 607 are fixed in position on the main shaft.

[0068] The main sealing lip 602 is made of ceramic fiber reinforced graphite material, which has a temperature resistance of over 450℃ and excellent self-lubricating properties, achieving basic sealing.

[0069] The insulation and cooling system is located outside the sealed area, including Figure 1 The middle heat insulation layer 1 and the annular cooling oil channel can simultaneously achieve heat insulation and cooling, preventing heat conduction from affecting the spindle accuracy;

[0070] The lower left end face of the sealing housing 601 is in close contact with the upper end face of the main sealing lip 602, and the left end face inside the sealing housing 601 is aligned with the right end face of the main sealing lip 602. The lower right end face of the sealing housing 601 is aligned with the upper end face of the rotating bushing 606. The space enclosed by the sealing housing 601, the main sealing lip 602 and the rotating bushing 606 includes the elastic compensation device and the sealing cavity.

[0071] The elastic compensation device consists of a base 603, an auxiliary sealing lip 604, a push ring 605, a waveform elastic compensation component 608, and a positioning screw 609. The right end face of the base 603 is in close contact with the inside right end face of the sealing housing 601 and is positioned by the positioning screw 609. The left end face of the base 603 is fixedly connected to the right end of the waveform elastic compensation component 608. The left end of the waveform elastic compensation component 608 is fixedly connected to the right end face of the push ring 605. The upper end face of the push ring 605 is in close contact with the base 603. The left end face of the push ring 605 is in close contact with the right end face of the auxiliary sealing lip 604. The left end face of the auxiliary sealing lip 604 is in close contact with the right end face of the main sealing lip 602.

[0072] The aforementioned sealing cavity is the space between the sealing housing 601 and the elastic compensation device, used to define the sealing space. The inner wall is nitrided. The sealing housing 601 and the main sealing lip 602 form the main seal, and the main sealing lip 602 and the auxiliary sealing lip 604 form the auxiliary seal. A small amount of medium that passes through the main seal will enter the sealing cavity and will then be intercepted by the auxiliary seal.

[0073] The auxiliary sealing lip 604 adopts a nickel-based alloy coated with polytetrafluoroethylene (PTFE) composite structure, which utilizes the high strength of the alloy and the low coefficient of friction of PTFE to compensate for the leakage of small gaps in the main sealing lip 602.

[0074] The waveform elastic compensation component 608 is made of Inconel 718 alloy. It maintains stable elastic deformation under high temperature even when under pressure for a long time, and can continuously provide preload force to the double lips, automatically compensating for thermal expansion and wear.

[0075] In actual use, the sealing housing 601 and the main sealing lip 602 form the main seal, which can isolate most of the medium. After a small amount of medium leaks into the sealing cavity, the main sealing lip 602 and the auxiliary sealing lip 604 form the auxiliary seal, which further isolates the small amount of medium leaked from the main seal. Moreover, due to the continuous compensation effect of the elastic compensation device, it can continuously provide sufficient pre-tightening force under high temperature, high speed and heavy load conditions to ensure sealing performance.

[0076] This rotary seal is suitable for the 400℃ high-temperature conditions of thrust bearing testing systems, with a temperature resistance increase of over 60%, load capacity increase of 50%, and speed increase of over 5 times. The component adopts a double-lip composite structure and waveform elastic compensation, achieving strong structural integration and an integrated design of heat insulation, sealing, compensation, and cooling. It is suitable for rotating components in deep earth, deep sea, and extreme high-temperature environments.

[0077] The hydrostatic bearing 8 includes a horizontally arranged hydrostatic bearing body 801. End caps 802 are fixed at both ends of the hydrostatic bearing body 801. The hydrostatic bearing spindle 808 is sleeved inside the hydrostatic bearing body 801. The two ends of the hydrostatic bearing spindle 808 are connected to the end caps 802 through rolling bearings 803 respectively. The connection between the rolling bearings 803 and the end caps 802 is provided with a reverse-threaded sealing thread 804. One end of the hydrostatic bearing body 801 is provided with a positioning boss 805. The middle part of the hydrostatic bearing spindle 808 is provided with a thrust boss 806. The thrust boss 806 is adapted to the positioning boss 805. A metal seal 807 is provided between the thrust boss 806 and the positioning boss 805.

[0078] A bushing 809 is fitted around the outer periphery of the middle part of the hydrostatic bearing spindle 808. An annular oil cavity 810 is formed between the bushing 809 and the hydrostatic bearing body 801. An oil inlet 811 is provided at the top of the hydrostatic bearing body 801, and an oil outlet 812 is provided at the bottom of the hydrostatic bearing body 801. The annular oil cavity 810 is connected to the oil inlet 811 and the oil outlet 812.

[0079] Rolling bearing 803 is a high manganese alloy steel thrust bearing.

[0080] The hydrostatic bearing body 801 is made of 42CrMo high-temperature alloy, forged and aged, with a tensile strength ≥1200MPa and Vickers hardness ≥450HV. The hydrostatic bearing body 801 has an annular stepped structure, with its inner diameter interference-fitted to the test platform shaft system. An outer diameter has a mounting flange, and an annular oil cavity 810, 20mm wide and 5mm deep, is opened on the end face facing the high-temperature chamber. The oil cavity is connected to an external oil supply system through an 8mm oil inlet 811. A cooling jacket is provided around the outer periphery of the hydrostatic bearing body 801, forming a spiral cooling channel between the cooling jacket and the hydrostatic bearing body 801, suitable for hydraulic oil circulation and heat dissipation.

[0081] The bushing 809 has an annular groove 813 on its outer periphery and branch channels 814 are uniformly provided along the circumferential direction inside the bushing 809. The branch channels 814 are connected to the annular groove 813 and are arranged radially. The end of the branch channels 814 is a throttling orifice.

[0082] The actual use of hydrostatic bearing 8 includes the following steps:

[0083] (I) Assembly process

[0084] The hydrostatic bearing body 801 is fitted onto the test platform shaft system using a heat fitting process (heated to 150°C), and the flange is fixed to the mounting base with M15 bolts;

[0085] The thrust roller bearing is installed on the outside of the main body via a locating sleeve, and the axial clearance is adjusted to 0.04mm;

[0086] Connect the oil inlet 811 (oil supply pressure 1.2MPa), the oil outlet 812, and the temperature sensor (model PT100).

[0087] (II) Work Process

[0088] Normal operating conditions: The hydrostatic bearing hydraulic oil enters the annular oil chamber 810 through the oil inlet 811, and forms a load-bearing oil film through the branch flow channel 814 to support a 30T axial load. The shaft system rotates at 12000r / min.

[0089] Static pressure failure: When the oil supply pressure is lower than 0.3MPa, the rolling bearing 803 immediately contacts the load and triggers the alarm system.

[0090] Overload condition: When the dynamic load rises to 35T, the entire component bears no plastic deformation.

[0091] (III) Parameter Verification

[0092] According to the test bench technical documents, after 1000 hours of continuous operation at 1000 rpm, the bearing temperature rise was ≤18℃, and the oil film stability met the standard. In the hydrostatic failure simulation test, there was no component damage, verifying the effectiveness of the protection.

[0093] The hydrostatic bearing 8 achieves breakthroughs in high-temperature resistance (the 42CrMo alloy exhibits stable performance at high temperatures, solving the problem of high-temperature softening in traditional bearings), optimized load-bearing capacity (the high-temperature oil film thickness is stable at 0.07~0.1mm, achieving synergistic load bearing of 30T / 12000r / min), and reliable failure protection (the seamless connection between the two bearings has a response time ≤0.02s, and the 42T rated load can withstand instantaneous impact).

[0094] like Figure 8 , 9 As shown, the present invention also discloses a bearing fault diagnosis and life prediction method using the above-mentioned testing platform, comprising the following steps:

[0095] S1. Multimodal Data Synchronous Acquisition: Utilizing a high-temperature testing platform, the potential difference signal, vibration signal, ultrasonic wall thickness signal, and friction torque signal of the thrust bearing are simultaneously acquired within the range of 0~400℃. A matrix of measuring electrodes 14 is arranged in the corresponding area of ​​the bearing under test according to certain rules, and a pair of current excitation electrodes 13 are arranged at both ends of the bearing. Any type of corrosion defect can cause a significant change in the measured signal. The vibration sensor is installed at the bottom of the hydrostatic bearing, and the thermal load is reduced through structural isolation. For ultrasonic thickness measurement, the temperature difference data inside and outside the insulation layer is used to compensate for the sound velocity of the ultrasonic wave in the 400℃ medium in real time to maintain measurement accuracy. A heavy-duty hydrostatic bearing 8 supports a load of over 30T, ensuring stable operation of the spindle at a high speed of 12000rpm. This high-rigidity support method minimizes vibration interference from non-measured components, thereby improving the signal-to-noise ratio of the vibration signal (dynamic characteristics).

[0096] S2. High-temperature drift compensation processing: To address signal distortion in a 400℃ environment, the ultrasonic velocity and vibration sensor sensitivity are nonlinearly calibrated using real-time temperature data collected by a K-type thermocouple.

[0097] S3. Multidimensional feature extraction and fusion: Ultrasonic sensors are used to monitor the changes in the wall thickness of the thrust bearing in real time, with a measurement accuracy of 0.01 mm, which serves as a "static feature" of the physical wear of the bearing.

[0098] In this invention, when extracting the time-frequency domain features of the vibration signal as "dynamic features":

[0099] Nonlinear compensation and static feature extraction of high-temperature sound velocity: To address the temperature drift problem in ultrasonic thickness measurement at 400℃, a second-order sound velocity compensation model based on the real-time temperature gradient T is established.

[0100]

[0101] in, Reference temperature The speed of sound at that time and The temperature sensitivity coefficient is pre-calibrated. Using this model to correct the pulse reflection time, the wall thickness reduction with an accuracy of 0.01 mm is obtained as a "static characteristic" of the bearing's physical wear.

[0102] S4, PINN Physical Fusion Modeling: Extracted static and dynamic features are input into a physical information neural network model to establish a bearing fault diagnosis and life prediction model. During model training, Arcard's wear law is embedded in the neural network loss function as a physical constraint term, ensuring that the model's prediction results not only satisfy the data variation patterns but also conform to the physical mechanism of bearing wear evolution. Through the joint constraints of multimodal data features and wear physical laws, the model's ability to identify bearing fault states and degradation trends under high temperature, high speed, and heavy load conditions is improved.

[0103] S5. Dynamic Correction and Output: The actual wall thickness reduction measured by ultrasound, the local wear changes reflected by the potential difference signal, and the friction torque changes are used as dynamic correction factors to correct the output results of the physical information neural network model in real time. When there is a deviation between the model's predicted wear and the measured wear, the prediction results are corrected according to the deviation, making the degradation trajectory output by the model closer to the actual wear process of the bearing. Finally, the bearing fault diagnosis conclusion, wear condition level, and remaining service life prediction results are output, providing a basis for the condition assessment and failure early warning of thrust bearings under high-temperature conditions.

[0104] 2. Dynamic Feature Extraction and Noise Reduction:

[0105] Under high-speed operating conditions of 12000rpm, variational mode decomposition (VMD) is used to denoise the vibration signal and extract time-frequency domain features as "dynamic features" to characterize the instantaneous operating state of the bearing.

[0106] 3. Linear coupling of dynamic and static features

[0107] Define a multimodal health index (HI) that nonlinearly couples the "dynamic" characteristic energy of vibration signals with the "static" physical degradation measured by ultrasound monitoring:

[0108]

[0109] By suppressing high-frequency random noise using a logarithmic function and amplifying the degradation trend in the later, severe wear stage using a power function, a precise characterization of the entire bearing life cycle can be achieved.

[0110] 3. Construct a Physically Constrained Neural Network (PINN) prediction model:

[0111] Dynamic features and static wear values ​​are input into the PINN neural network. Its core innovation lies in embedding Archard's law of wear into the network's loss function as a hard constraint term.

[0112]

[0113] in The constrained prediction results must conform to the laws of physical degradation:

[0114]

[0115] ( For wear volume, For real-time load, The sliding distance, (This refers to the material hardness).

[0116] In practice:

[0117] Hardware implementation: The platform applies a maximum thrust of up to 35T through the thrust cylinder 1, and ensures that the peripheral motors and sensors are not damaged when the core test area is at 400℃ through the heat insulation layer 3 and the heat insulation coupling 9.

[0118] Sensor installation: The vibration sensor is installed at the bottom of the hydrostatic bearing; the ultrasonic thickness sensor is installed on the thrust bearing, and the wall thickness reduction during operation is monitored by real-time pulse reflection method.

[0119] Bearing status is captured through synchronous acquisition of multimodal data: synchronous acquisition of measurement electrode signals, load signals from thrust sensor 2, and friction torque signals from torque sensor 10.

[0120] The extracted feature vectors are input into the PINN neural network, and the Archard wear law is embedded into the model as a physical constraint term in the loss function. Finally, the model synchronously outputs the fault diagnosis conclusion and the remaining useful life (RUL).

[0121] Of course, the present invention may have other various embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art can make various corresponding changes and modifications according to the present invention, but these corresponding changes and modifications should all fall within the protection scope of the appended claims.

Claims

1. A high-temperature bearing testing platform, characterized in that... include: Includes a thrust cylinder (1), a thrust sensor (2), a heat insulation layer (3), a sliding seal (4), a measured mechanism and frame (5), a rotary seal (6), a heat insulation layer (7), a hydrostatic bearing (8), a coupling (9), a dynamic torque sensor (10), a torque controller (11), and a high-speed motor (12). The rotary seal (6) includes a sealing housing (601), a rotary bushing (606) is fitted inside the sealing housing (601), a main sealing lip (602) is provided between the rotary bushing (606) and the head end of the sealing housing (601), a retaining ring (607) is coaxially provided at the tail end of the sealing housing (601), and an elastic compensation mechanism is coaxially fixed in the inner cavity of the tail end of the sealing housing (601). The elastic compensation mechanism includes a base (603), the base (603) is fixedly connected to the sealing housing (601), and the base (603) is connected to a push ring (605) through a wave elastic compensation component (608). The push ring (605) pushes the auxiliary sealing lip (604) to tightly adhere to the main sealing lip (602).

2. The testing platform according to claim 1, characterized in that: The hydrostatic bearing (8) includes a horizontally arranged hydrostatic bearing body (801), with end caps (802) fixed at both ends of the hydrostatic bearing body (801), and a hydrostatic bearing spindle (808) sleeved inside the hydrostatic bearing body (801). The two ends of the hydrostatic bearing spindle (808) are connected to the end caps (802) through rolling bearings (803). The connection between the rolling bearings (803) and the end caps (802) is provided with a reverse-threaded sealing thread (804). One end of the hydrostatic bearing body (801) is provided with a positioning boss (805), and the middle of the hydrostatic bearing spindle (808) is provided with a thrust boss (806). The thrust boss (806) is adapted to the positioning boss (805), and a metal seal (807) is provided between the thrust boss (806) and the positioning boss (805). A bushing (809) is fitted around the outer periphery of the middle part of the hydrostatic bearing spindle (808). An annular oil cavity (810) is formed between the bushing (809) and the hydrostatic bearing body (801). An oil inlet (811) is provided at the top of the hydrostatic bearing body (801), and an oil outlet (812) is provided at the bottom of the hydrostatic bearing body (801). The annular oil cavity (810) is connected to the oil inlet (811) and the oil outlet (812).

3. The testing platform according to claim 2, characterized in that: The bushing (809) has an annular groove (813) on its outer periphery and a branch flow channel (814) is uniformly provided in the bushing (809) along the circumferential direction. The branch flow channel (814) is connected to the annular groove (813) and is arranged radially. The end of the branch flow channel (814) is a throttling hole.

4. The testing platform according to claim 2, characterized in that: A cooling jacket is provided on the outside of the hydrostatic bearing body (801), and a spiral cooling channel is provided between the cooling jacket and the hydrostatic bearing body (801).

5. A method for bearing fault diagnosis and life prediction based on any one of the test platforms described in claims 1-4, characterized in that... Includes the following steps: S100, Multimodal Data Synchronous Acquisition: Using a high-temperature testing platform, the potential difference signal, vibration signal, ultrasonic wall thickness signal, and friction torque signal of the thrust bearing are synchronously acquired within the range of 0~400℃. S200, High-Temperature Drift Compensation: Nonlinear calibration of ultrasonic velocity and vibration sensor sensitivity is performed using real-time temperature data collected by a K-type thermocouple; S300, Multidimensional Feature Extraction and Fusion: Using ultrasonic sensors to monitor the change in the wall thickness of the thrust bearing in real time as a static feature of the bearing's physical wear, and extracting the time-frequency domain features of the vibration signal as a dynamic feature; S400. Construct a Physically Constrained Neural Network (PINN) prediction model: Input dynamic features and static wear values ​​into the PINN neural network, embed Archard's wear law into the network's loss function as a hard constraint term, and through collaborative training based on physical mechanisms and data-driven approaches, output the bearing's fault diagnosis conclusion and remaining service life. ， in The constrained prediction results must conform to the laws of physical degradation: ， in, For wear volume, For real-time load, The sliding distance, The value represents the material hardness.

6. The fault diagnosis and life prediction method according to claim 5, characterized in that... In step S100, a measuring electrode matrix is ​​arranged in the corresponding area of ​​the bearing under test, a pair of current excitation electrodes are arranged at both ends of the bearing under test, and a vibration sensor is installed at the bottom of the hydrostatic bearing to reduce the thermal load through an isolation structure.

7. The fault diagnosis and life prediction method according to claim 5, characterized in that... In step S200, a second-order sound velocity compensation model based on the real-time temperature gradient T is established: ， in, Reference temperature The speed of sound at that time and This is the pre-calibrated temperature sensitivity coefficient.

8. The fault diagnosis and life prediction method according to claim 5, characterized in that... In step S300, at a speed of 12000... Under operating conditions, variational mode decomposition is used to denoise the vibration signal and extract time-frequency domain features as dynamic features to characterize the instantaneous operating state of the bearing. The change in the wall thickness of the thrust bearing is expressed as a wall thickness reduction with an accuracy of 0.01 mm.

9. The fault diagnosis and life prediction method according to claim 5, characterized in that... In step S300, the multimodal health index is defined ( This method nonlinearly couples static and dynamic features, suppresses high-frequency random noise through a logarithmic function, and amplifies the degradation trend in the later, severe wear stage through a power function, thereby achieving an accurate characterization of the entire bearing life cycle. The formula is as follows: ， in, For multimodal health index, The characteristic value of the signal energy. This is the baseline energy under initial healthy conditions. For the signal energy term weights, Weighting of defect / wear size item. This represents the actual amount of wear. This is the failure threshold.

10. The fault diagnosis and life prediction method according to claim 5, characterized in that... The physical constraint process in step S400 uses the measured physical wear as a dynamic correction factor to correct the bearing degradation trajectory deduced from the vibration signal characteristics in real time.

Images