A sliding bearing fault diagnosis system and method

Abstract

The present invention relates to the technical field of sliding bearing status assessment and health management, and more specifically, to a sliding bearing fault diagnosis system and method. The present invention monitors the inlet and outlet oil temperature, the vibration frequency of the sliding bearing, and the eccentricity of the rotating shaft in different directions within the same cross section; derives an assessment threshold through a series of simulation experiments, and evaluates the operating status of the sliding bearing based on an intelligent algorithm; ultimately, combining real-time monitoring data with simulation data to accurately diagnose the fault location and mode of the sliding bearing. The present invention can perform real-time, online, comprehensive monitoring of the three physical quantities of the sliding bearing: the inlet and outlet oil temperature, vibration, and rotating shaft eccentricity; determine the fault location and mode of the sliding bearing based on the real-time data of the three physical quantities; and, through status classification, enable maintenance personnel to promptly repair or replace sub-healthy or abnormal sliding bearings, greatly improving the maintenance personnel's labor productivity and fault diagnosis capabilities.

Description

Technical Field

[0001] The present invention relates to the technical field of sliding bearing state assessment and health management, and in particular to a sliding bearing fault diagnosis system and method. Background Art

[0002] Sliding bearings are key components in large-scale machinery (such as power machinery, pumps, and fans). Their long service life and excellent vibration resistance have led to their widespread use in industrial machinery and public facilities. However, due to their harsh operating environments, they are also among the most susceptible to damage. According to statistics, 30% of rotating machinery failures are caused by bearing damage. Therefore, timely and effective fault diagnosis of sliding bearings is crucial.

[0003] Common faults in sliding bearings include abnormal vibration and impact caused by foreign matter in the bearing, and oil film oscillation caused by abnormal oil temperature. Common diagnostic methods include monitoring lubricating oil and bearing temperatures, and performing vibration analysis. However, temperature measurement has a hysteresis effect, making it difficult to predict faults. Vibration analysis is the most widely used method, but the energy of the vibration signal is primarily distributed in the low-frequency range below 1500Hz, while the noise signal is also low-frequency, making signal separation difficult.

[0004] The existing technologies for sliding bearing condition assessment and fault diagnosis have the following deficiencies:

[0005] (1) Monitoring lubricating oil temperature and bearing temperature for bearing condition assessment and fault diagnosis has a hysteresis effect and cannot achieve the purpose of predicting faults;

[0006] (2) In the vibration analysis method, the energy of the vibration signal is mainly distributed in the low-frequency area below 1500 Hz, and the noise signal frequency is also low-frequency, making signal separation more difficult. Summary of the Invention

[0007] The technical problem to be solved by the present invention is: in view of the shortcomings of the existing technology, a sliding bearing fault diagnosis system and method are provided, which can accurately diagnose the fault location and fault mode of the sliding bearing, and can realize fault prediction, thereby greatly improving the equipment maintenance efficiency.

[0008] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0009] 1. A sliding bearing fault diagnosis method

[0010] The present invention provides a sliding bearing fault diagnosis method, which mainly includes the following steps:

[0011] S1, monitors the real-time temperature data of the lubricating oil in the sliding bearing, the vibration frequency data of the sliding bearing, and the eccentricity data of the rotating shaft;

[0012] S2, transmits the monitoring data to the data management system;

[0013] S3, processing the monitoring data, conducting simulation experiments to obtain a state assessment threshold, and verifying the rationality of the state assessment threshold;

[0014] S4, through data structured processing, working condition classification, and working condition identification, the operating status of the sliding bearing is evaluated and classified in real time, and early warning information of the operating status of the sliding bearing is generated based on the evaluation results;

[0015] S5, based on the early warning information of the sliding bearing operating status, combined with real-time monitoring data and simulation data, diagnose the fault location and fault mode of the sliding bearing.

[0016] Furthermore, the real-time temperature data includes the lubricating oil temperature at the oil inlet and the lubricating oil temperature at the oil outlet; the monitoring points of the vibration frequency are evenly distributed on the sliding bearing; when measuring the eccentricity data of the rotating shaft, multiple point measurements are performed in different directions of the same cross-section, and each eccentricity monitoring data corresponds to the measurement time.

[0017] Furthermore, the simulation experiment includes a virtual simulation experiment and a faulty part physical experiment; the virtual simulation experiment uses simulation software to simulate and analyze the fault state of the sliding bearing to obtain experimental data; the faulty part physical experiment uses a physical sliding bearing faulty part and collects its relevant data.

[0018] Furthermore, the rationality of the state evaluation threshold is verified by comparing the experimental data obtained from the physical experiment of the faulty component with the experimental data obtained from the virtual simulation experiment. If the deviation between the two is within a preset range, the state evaluation threshold is determined to be a reasonable value.

[0019] Furthermore, the real-time status assessment includes the following steps:

[0020] S41, data structuring processing: performing missing value processing and noise removal on the monitoring data;

[0021] S42, working condition division: dividing the operating conditions of the sliding bearing, and selecting motion state parameters related to the operating state of the sliding bearing to construct a feature vector;

[0022] S43, operating condition identification: Based on the monitoring data and the characteristic vector of the motion state parameters, the real-time operating condition of the sliding bearing is identified, and the probability of occurrence of each operating condition is calculated.

[0023] Furthermore, the missing value processing adopts the hot card filling method; the denoising process includes the following steps:

[0024] 1) Based on the phase space reconstruction theory, the matrix A is reconstructed from the time series signal:

[0025]

[0026] Here, L is selected as half of the signal length. The time series signal is Fourier transformed to determine the number of main frequencies n;

[0027] 2) Perform singular value decomposition on the reconstructed matrix A, using 2n as the effective rank order and setting other singular values ​​to 0, thereby obtaining a new reconstructed matrix B;

[0028] 3) Add the corresponding elements in B and take the average to obtain the denoised signal data.

[0029] Furthermore, the diagnosis of the fault location and fault mode of the sliding bearing is carried out by specifically adopting a mechanism-based fault diagnosis method:

[0030] Based on the historical database of the sliding bearing during normal operation and the simulation data, the functional relationship between the data recording value and the theoretical value during normal operation of the sliding bearing is fitted by the least squares method, and then the real-time monitoring data is compared with the data recording value during normal operation. If the difference is greater than the set threshold, a fault error warning is issued.

[0031] Furthermore, the functional relationship is fitted by the least square method, specifically including:

[0032] Let (x, y) be a pair of observations, and x = [x1, x2, ..., x n ] T ∈R n , y = R satisfies the following theoretical function:

[0033] y=f(x,w)

[0034] Where w=[w1,w2,...,w n ] T is a parameter to be determined;

[0035] In order to find the optimal estimate of the parameter w of the function f(x, w), for a given set of m observations (x i ,y i )(i=1,2,...,m), solve the objective function:

[0036]

[0037] The parameter w that takes the minimum value i(i=1, 2, ..., n).

[0038] 2. A sliding bearing fault diagnosis system

[0039] Based on the same inventive concept, the present invention also provides a sliding bearing fault diagnosis system for implementing the above-mentioned fault diagnosis method, comprising:

[0040] (1) Data monitoring equipment: used to monitor the real-time temperature data of the lubricating oil in the sliding bearing, the vibration frequency data of the sliding bearing, and the eccentricity data of the rotating shaft;

[0041] (2) Data management system: used to receive monitoring data sent by various data monitoring devices, process and evaluate the status, and generate early warning information on the operating status of sliding bearings;

[0042] (3) Fault diagnosis system: Based on the early warning information of the sliding bearing operating status, combined with real-time monitoring data and simulation data, the fault location and fault mode of the sliding bearing are diagnosed.

[0043] Compared with the prior art, the present invention has the following main advantages:

[0044] 1. This invention proposes a sliding bearing fault diagnosis system and method. This system monitors the oil inlet and outlet temperatures, the vibration frequency of the sliding bearing, and the shaft eccentricity in different directions within the same cross section. Furthermore, through a series of simulation experiments, an evaluation threshold is derived. The operating status of the sliding bearing is then evaluated using an intelligent algorithm. Ultimately, by combining real-time monitoring data with simulation data, accurate diagnosis of the sliding bearing fault location and mode is achieved.

[0045] 2. The present invention can perform real-time online comprehensive monitoring of the three physical quantities of the sliding bearing's oil inlet and outlet temperature, vibration, and shaft eccentricity. Based on the real-time data of the three physical quantities, the fault location and fault mode of the sliding bearing are determined. Through status classification, maintenance personnel can promptly repair or replace sliding bearings that are in sub-health or abnormal conditions, greatly improving the maintenance personnel's labor productivity and fault diagnosis capabilities, while also improving factory safety, avoiding major damage, and improving the factory's economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 Flowchart of a sliding bearing fault diagnosis method according to an embodiment of the present invention;

[0047] Figure 2 Schematic diagram of the BP neural network algorithm in an embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for the purpose of explaining the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0049] It should be pointed out that, according to the needs of implementation, the various steps / components described in this application can be split into more steps / components, or two or more steps / components or partial operations of steps / components can be combined into new steps / components to achieve the purpose of the present invention.

[0050] Embodiment 1: This embodiment provides a sliding bearing fault diagnosis method, such as Figures 1-2 As shown, the main steps include:

[0051] S1, monitors the real-time temperature of the lubricating oil in the sliding bearing, the vibration frequency of the sliding bearing, and the eccentricity of the rotating shaft;

[0052] S2, transmits the monitoring data to the data management system;

[0053] S3, based on intelligent algorithms, processes measured data, conducts simulation experiments to derive state thresholds, and verifies their rationality;

[0054] S4, through steps such as data structured processing, operating condition classification, and operating condition identification, achieves real-time evaluation of operating status and issues early warning based on the evaluation results;

[0055] S5, based on the early warning information of the sliding bearing's operating status, combined with real-time monitoring data and simulation data, realizes accurate diagnosis of the sliding bearing's fault location and fault mode.

[0056] Furthermore, the real-time temperature data includes the oil inlet temperature and the oil outlet temperature; the distribution of the vibration frequency measurement points should be uniform and comprehensive; the shaft eccentricity measurement should be performed at multiple points in different directions on the same cross section, and each monitoring data should correspond to the measurement time;

[0057] Furthermore, in the sliding bearing fault diagnosis system and method described in the present invention, the simulation experiment includes a virtual simulation experiment and a physical experiment of a faulty part; the virtual simulation experiment uses simulation software to simulate and obtain experimental data; the physical experiment of the faulty part installs and uses the faulty part to collect relevant data, and the data obtained through the simulation experiment can be applied to the state assessment threshold judgment.

[0058] Furthermore, the status assessment includes the following steps:

[0059] (1) Data structured processing. In order to eliminate the influence of factors such as the collection environment, technology, and equipment as much as possible, the data is processed for missing values ​​and denoising;

[0060] (2) Working condition classification. Considering the real-time changes in the operating state of the sliding bearing, the operating conditions of the sliding bearing are classified, and the parameters that are relevant to the operating state of the sliding bearing and have a greater impact are selected to construct the feature vector;

[0061] (3) Operating condition identification. Refer to the health benchmark and identify the real-time operating condition of the sliding bearing based on the monitoring data, and calculate the probability of the real-time operating condition occurring;

[0062] (4) Status classification. By measuring the distance between the state feature vector of the sliding bearing in the abnormal state and the healthy benchmark, the degradation degree of the operating state is evaluated and graded. Based on this, the operating state of the sliding bearing is matched with healthy, subhealthy, and abnormal.

[0063] Furthermore, for typical faults during the operation of sliding bearings, appropriate diagnostic algorithms are employed based on parameter influence curves, performance correlation curves, and experimental / simulation data to achieve real-time diagnosis of the fault location and mode of the sliding bearing. These fault diagnosis methods include mechanism-based methods, data-driven methods, and fusion-driven methods.

[0064] (1) Mechanism-based diagnostic method: Based on the historical database of the system during normal operation and the simulation data of its mechanism, performance and other models, a priori information is obtained. The functional relationship between the observed value and the theoretical value of the data during normal operation of the system is fitted by the least squares method. The real-time status information measurement data of the system is then compared with the data during normal operation. If the error is greater than the set threshold, a fault error warning is issued;

[0065] (2) Data-driven diagnostic method: pre-processing is performed based on the collected data, and outliers can be eliminated according to the 3δ rule.

[0066] Feature extraction is then performed. Selecting the most sensitive feature values ​​for a specific machine can enhance the targeted nature of monitoring and diagnosis, improving diagnostic accuracy. A series of steps, including feature dimensionality reduction and pattern recognition, are then performed to map the relationship between operating data features and fault categories.

[0067] (3) Fusion-driven diagnosis method: After normalizing the historical monitoring data for different operating conditions and boundary conditions, the weights, thresholds, fault characteristics and probabilities are identified and corrected through BP neural network and other training models to obtain a more accurate mechanism diagnosis model and realize personalized diagnosis of generator system faults.

[0068] Embodiment 2: This embodiment provides a sliding bearing fault diagnosis method, comprising:

[0069] First, measure the inlet and outlet oil temperatures, the vibration frequency of the sliding bearing, and the eccentricity of the rotating shaft.

[0070] Thermocouple temperature sensors can be used to measure the inlet and outlet oil temperatures. In the actual testing process, the thermocouple temperature sensors are fixed on the inlet and outlet of the oil. The data acquisition system collects the temperature signals of the oil inlet and outlet in real time.

[0071] When measuring vibration frequency, the appropriate sensor should be selected based on the bearing's structure. For bearings with exposed bearing seats, absolute vibration velocity or acceleration sensors can be used; for bearings with concealed bearing seats, relative vibration displacement sensors are often used. Measurement directions are primarily radial, vertical, and horizontal. The measurement point should be located where the vibration transmission path is shortest and most sensitive to the vibration source.

[0072] The eccentricity of the rotating shaft can be measured by the displacement sensor to measure the displacement change x, y of the journal in the horizontal and vertical directions relative to the center of the shaft hole, and then the eccentricity formula is used to calculate the displacement change x, y. The eccentricity can be calculated. Based on this principle, the eccentricity in multiple directions can be measured for the same cross section.

[0073] Furthermore, simulation software, such as the workbench in Ansys, can be used in virtual simulation experiments to set parameter properties such as materials, forces, spatial relationships, and meshing to simulate and obtain experimental data. In physical experiments on faulty parts, the described method can be used to measure temperature, pressure, vibration, and eccentricity during operation, and the data obtained in combination with the simulation experiment can be used to determine the evaluation threshold.

[0074] After determining the evaluation threshold, the monitored data can be processed based on the intelligent algorithm to evaluate the operating status of the sliding bearing.

[0075] The status assessment includes the following steps:

[0076] Step 1: Data Structuring

[0077] In order to minimize the impact of factors such as the acquisition environment, technology, and equipment, the data is processed for missing values ​​and denoising to enhance the non-ideal original vibration signal data, resolve issues such as missing data, insufficient samples, and high noise, and provide support for subsequent feature extraction and fault prediction. The missing value processing method is the hot card filling method. In the complete data, the object that is most similar to it is found and the current value is filled with the most similar value. This is done using the KNN algorithm with fixed parameters and only one reference. The denoising method is the singular value decomposition. The noise reduction process is as follows:

[0078] (1) Based on the phase space reconstruction theory, the matrix A is reconstructed from the time series signal:

[0079]

[0080] Here, L is selected as half of the signal length. The time series signal is Fourier transformed to determine the number of main frequencies n.

[0081] (2) Perform singular value decomposition on the reconstructed matrix A, take 2n as the effective rank order, set other singular values ​​to 0, and thus obtain a new reconstructed matrix B.

[0082] (3) Add the corresponding elements in B and take the average to obtain the noise-reduced vibration signal data.

[0083] Step 2: Working Condition Division

[0084] Considering the real-time changes in the operating state of the sliding bearing, the operating conditions of the sliding bearing are divided, and the parameters related to the operating state of the sliding bearing and with the greatest influence are selected to construct the feature vector;

[0085] Step 3: Working Condition Identification

[0086] Refer to the health benchmark, identify the real-time operating conditions of the sliding bearing based on the monitoring data, and calculate the probability of the real-time operating conditions occurring;

[0087] Step 4: Status Classification

[0088] By measuring the distance between the state characteristic vector of the sliding bearing under abnormal conditions and the healthy benchmark, the degree of deterioration of the operating status is evaluated and graded. Based on this, the operating status of the sliding bearing is matched with healthy, sub-healthy, and abnormal.

[0089] Furthermore, based on the early warning information of the sliding bearing's operating status, combined with real-time monitoring data and simulation data, accurate diagnosis of the sliding bearing's fault location and fault mode can be achieved.

[0090] In the process of accurately diagnosing the fault location and fault mode of the sliding bearing, a mechanism-based fault diagnosis method is adopted based on parameter influence curves, performance correlation curves, and experimental / simulation data for typical faults during the operation of the sliding bearing to achieve real-time diagnosis of the fault location and fault mode of the sliding bearing.

[0091] Embodiment 3, based on the same inventive concept, this embodiment further provides a sliding bearing fault diagnosis system for implementing the above-mentioned fault diagnosis method, comprising:

[0092] (1) Data monitoring equipment: used to monitor the real-time temperature data of the lubricating oil in the sliding bearing, the vibration frequency data of the sliding bearing, and the eccentricity data of the rotating shaft;

[0093] (2) Data management system: used to receive monitoring data sent by various data monitoring devices, process and evaluate the status, and generate early warning information on the operating status of sliding bearings;

[0094] (3) Fault diagnosis system: Based on the early warning information of the sliding bearing operating status, combined with real-time monitoring data and simulation data, the fault location and fault mode of the sliding bearing are diagnosed.

[0095] Furthermore, all parts of this application that are not described in detail are the same as the existing technology or are implemented using the existing technology.

[0096] In summary:

[0097] 1. This invention proposes a sliding bearing fault diagnosis system and method. This system monitors the oil inlet and outlet temperatures, the vibration frequency of the sliding bearing, and the shaft eccentricity in different directions within the same cross section. Furthermore, through a series of simulation experiments, an evaluation threshold is derived. The operating status of the sliding bearing is then evaluated using an intelligent algorithm. Ultimately, by combining real-time monitoring data with simulation data, accurate diagnosis of the sliding bearing fault location and mode is achieved.

[0098] 2. The present invention can perform real-time online comprehensive monitoring of the three physical quantities of the sliding bearing's oil inlet and outlet temperature, vibration, and shaft eccentricity. Based on the real-time data of the three physical quantities, the fault location and fault mode of the sliding bearing are determined. Through status classification, maintenance personnel can promptly repair or replace sliding bearings that are in sub-health or abnormal conditions, greatly improving the maintenance personnel's labor productivity and fault diagnosis capabilities, while also improving factory safety, avoiding major damage, and improving the factory's economic benefits.

[0099] It will be easily understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements 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 sliding bearing fault diagnosis method, characterized in that: The steps include: S1, monitors the real-time temperature data of the lubricating oil in the sliding bearing, the vibration frequency data of the sliding bearing, and the eccentricity data of the rotating shaft; S2, transmits the monitoring data to the data management system; S3, processing the monitoring data, conducting simulation experiments to obtain a state assessment threshold, and verifying the rationality of the state assessment threshold; S4, through data structured processing, working condition classification, and working condition identification, the operating status of the sliding bearing is evaluated and classified in real time, and early warning information of the operating status of the sliding bearing is generated based on the evaluation results; S5, based on the early warning information of the sliding bearing operating status, combined with real-time monitoring data and simulation data, diagnose the fault location and fault mode of the sliding bearing.

2. A sliding bearing fault diagnosis method according to claim 1, characterized in that: The real-time temperature data includes the lubricating oil temperature at the oil inlet and the lubricating oil temperature at the oil outlet; the monitoring points of the vibration frequency are evenly distributed on the sliding bearing; when measuring the eccentricity data of the rotating shaft, multiple points are measured in different directions of the same cross section, and each eccentricity monitoring data corresponds to the measurement time.

3. A sliding bearing fault diagnosis method according to claim 1, characterized in that: The simulation experiment includes a virtual simulation experiment and a physical experiment of a faulty part; the virtual simulation experiment uses simulation software to simulate and analyze the fault state of the sliding bearing to obtain experimental data; The faulty part entity experiment uses a physical sliding bearing faulty part and collects relevant data thereof.

4. A sliding bearing fault diagnosis method according to claim 3, characterized in that: The rationality of the verification state assessment threshold is specifically to compare the experimental data obtained from the physical experiment of the faulty component with the experimental data obtained from the virtual simulation experiment. If the deviation value between the two is within a preset range, the state assessment threshold is determined to be a reasonable value.

5. A sliding bearing fault diagnosis method according to claim 1, characterized in that: The real-time status assessment includes the following steps: S41, data structuring processing: performing missing value processing and noise removal on the monitoring data; S42, working condition division: dividing the operating conditions of the sliding bearing, and selecting motion state parameters related to the operating state of the sliding bearing to construct a feature vector; S43, operating condition identification: Based on the monitoring data and the characteristic vector of the motion state parameters, the real-time operating condition of the sliding bearing is identified, and the probability of occurrence of each operating condition is calculated.

6. A sliding bearing fault diagnosis method according to claim 5, characterized in that: The missing value processing adopts the hot card filling method; the denoising process includes the following steps: 1) Based on the phase space reconstruction theory, the matrix A is reconstructed from the time series signal: Among them, L is selected as half of the signal length, and the time series signal is Fourier transformed to determine the number of main frequencies n; 2) Perform singular value decomposition on the reconstructed matrix A, using 2n as the effective rank order and setting other singular values ​​to 0, thereby obtaining a new reconstructed matrix B; 3) Add the corresponding elements in B and take the average to obtain the denoised signal data.

7. A sliding bearing fault diagnosis method according to claim 1, characterized in that: The diagnosis of the fault location and fault mode of the sliding bearing is carried out by specifically adopting a fault diagnosis method based on a mechanism: Based on the historical database of the sliding bearing during normal operation and the simulation data, the functional relationship between the data recording value and the theoretical value during normal operation of the sliding bearing is fitted by the least squares method, and then the real-time monitoring data is compared with the data recording value during normal operation. If the difference is greater than the set threshold, a fault error warning is issued.

8. A sliding bearing fault diagnosis method according to claim 7, characterized in that: Fitting functional relationships through the least squares method includes: Let (x, y) be a pair of observations, and x = [x1, x2, ..., x n ] T ∈R n , y = R satisfies the following theoretical function: y=f(x,w) Where w=[w1,w2,...,w n ] T is a parameter to be determined; In order to find the optimal estimate of the parameter w of the function f(x, w), for a given set of m observations (x i ,y i )(i=1,2,...,m), solve the objective function: The parameter w that takes the minimum value i (i=1, 2, ..., n).

9. A sliding bearing fault diagnosis system for implementing the fault diagnosis method according to any one of claims 1 to 8, characterized in that: include: Data monitoring equipment: used to monitor the real-time temperature data of the lubricating oil in the sliding bearing, the vibration frequency data of the sliding bearing, and the eccentricity data of the rotating shaft; Data management system: used to receive monitoring data sent by various data monitoring devices, process and evaluate the status, and generate early warning information on the operating status of sliding bearings; Fault diagnosis system: Based on the early warning information of the sliding bearing's operating status, combined with real-time monitoring data and simulation data, the fault location and fault mode of the sliding bearing are diagnosed.

Images