A building friction pendulum isolation bearing state monitoring and fault diagnosis system

Abstract

The application discloses an intelligent friction pendulum isolation bearing and a state monitoring and fault diagnosis system. First, finite element analysis and calculation are carried out on the friction pendulum isolation bearing, so that the hysteresis curve and stress threshold change characteristics of the bearing under the operating condition are obtained; second, the hardware of the bearing state monitoring and fault diagnosis system is built, and the real-time acquisition and transmission of the stress, acceleration and displacement state of each component unit in the bearing and the overall structure of the bearing are completed; third, the system application sensors are accurately arranged according to the finite element analysis result; and finally, the pre-processing, deep learning and information storage of the multi-element data are realized in combination with the software of the bearing state monitoring and fault diagnosis system. The application can timely and accurately monitor and evaluate the operating state of the bearing and analyze and diagnose the health state of the bearing, so that the fatigue damage and structural damage of the bearing caused by the influence of operating conditions, environment and other factors for a long time are prevented, and the reduction of the operating reliability and service safety of the bearing is avoided.

Description

Technical Field

[0001] This invention relates to the field of building bearing technology, specifically to a system for monitoring the condition and diagnosing faults of building friction pendulum seismic isolation bearings. Background Technology

[0002] Friction pendulum seismic isolation bearings, installed between the superstructure and foundation of a building, are a vibration control technology that reduces earthquake damage to the superstructure by adjusting the equivalent radius of curvature, isolation period, horizontal stiffness, and damping ratio. They are widely used in building seismic isolation. However, during use, bearings are subjected to long-term vertical pressure, lateral shear force, and environmental factors, leading to fatigue damage and structural failure, which directly threatens the overall safety of the building. To ensure the operational stability and safety of the bearings during service, a condition monitoring and fault diagnosis system for friction pendulum seismic isolation bearings has been designed and developed. This system enables real-time monitoring and evaluation of the performance of critical bearings, representing a crucial step towards making engineering structures safer, healthier, and smarter.

[0003] In practical engineering, the long-term use of friction pendulum seismic isolation bearings in buildings faces three main problems: first, the bearings undergo structural deformation and damage to the friction material and the bearing as a whole under long-term medium-to-high loads; second, the bearings are affected by environmental factors such as humidity, dryness, high temperature, and low temperature, leading to a weakening or loss of their seismic isolation function; and third, the accurate monitoring and evaluation of the bearings' sliding capacity, resistance to torsion, resistance to tension, and restoring capacity during seismic isolation. Due to limitations in the engineering environment and manual testing methods, the performance monitoring and fault diagnosis of friction pendulum seismic isolation bearings are often difficult to detect and address in a timely and effective manner. Furthermore, as the main load-bearing components and seismic isolation devices, the occurrence of disasters involving the bearings can cause enormous losses. Therefore, timely and accurate monitoring and evaluation of the bearing's operating status and analysis and diagnosis of its health condition are key technologies for improving the performance of friction pendulum seismic isolation bearings. Summary of the Invention

[0004] The purpose of this invention is to provide a system for monitoring and diagnosing the condition of a building friction pendulum seismic isolation bearing. This system can monitor and evaluate the bearing's operating status and analyze and diagnose its health status in a timely and accurate manner. It can prevent fatigue damage and structural failure caused by long-term exposure to operating conditions and environmental factors, thereby reducing the bearing's operational reliability and service safety. This system ensures the stable operation of the bearing's load-bearing capacity and seismic isolation performance, thus solving the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a system for monitoring and diagnosing the condition of a building friction pendulum seismic isolation bearing, comprising four parts, including S1: finite element analysis and calculation of the performance of the building friction pendulum seismic isolation bearing, S2: hardware of the bearing condition monitoring and fault diagnosis system, S3: arrangement of bearing sensors in the system application, and S4: software of the bearing condition monitoring and fault diagnosis system.

[0006] Preferably, the present invention provides a system for monitoring and diagnosing the condition of a building friction pendulum seismic isolation bearing. Specifically, S1 involves: determining the bearing structural model based on the friction pendulum seismic isolation bearing design scheme proposed for the superstructure of a building in actual engineering practice; calculating the hysteresis curve and stress threshold variation characteristics of the bearing under actual working conditions through finite element analysis; and obtaining the bearing performance technical parameters based on the finite element analysis results. The bearing dynamic friction coefficient is... and post-yield stiffness Calculate the expression:

[0007]

[0008] In the formula: The initial stiffness of the support; To support the vertical load force; This is the yield displacement; The equivalent radius of curvature;

[0009] Support hysteresis energy dissipation Calculate the expression:

[0010]

[0011] In the formula: This represents the elastic deformation energy of the support. Equivalent stiffness; This represents the horizontal displacement of the support.

[0012] The Von Mises stress calculation expression for a support structure is as follows:

[0013]

[0014] In the formula: This is the first principal stress; This is the second principal stress; The third principal stress; through the deep learning network of the S4: support condition monitoring and fault diagnosis system software, the technical parameters of support performance are set to diagnose the failure behavior of the support.

[0015] Preferably, the present invention provides a system for monitoring and diagnosing the condition of a building friction pendulum seismic isolation bearing. Specifically, S2 comprises four main hardware components: S201: The bearing material strain performance acquisition system consists of a strain sensor, a bridge box, a strain meter, a data acquisition unit, and a host computer. This module aims to acquire the dynamic stress between the various components within the bearing, enabling the host computer to monitor and diagnose the accurate identification of the structural mechanical properties of the bearing during long-term service. Based on the stress threshold variation characteristics of the friction pendulum bearing through finite element analysis, the bearing stress is mainly concentrated between the spherical cap and the sliding surfaces of the upper and lower bearing plates. The commonly used friction material for the bearing is polytetrafluoroethylene (PTFE) with an elastic modulus of 280 MPa and a design strength of 30 MPa. Due to the long-term high-pressure load and environmental factors during service, the friction material is prone to structural deformation and damage. Therefore, strain sensors need to be embedded at the stress concentration points where the friction material connects to the upper and lower bearing plates to collect real-time dynamic stress changes within the bearing. The strain performance acquisition system has six strain sensors, each containing: working strain gauge S1, compensating strain gauge S2 and compensating strain gauge S3. The six strain gauges form a full bridge and are connected to the strain tester through a bridge box. The data acquisition unit then transmits the acquired strain signal to the host computer.

[0016] S202: The bearing photoelectric displacement acquisition system consists of a reflective photoelectric sensor, a data acquisition unit, and a host computer. This module aims to collect the vertical distance and tangential sliding distance between the upper and lower bearing plates during the bearing's service life, so that the host computer monitoring and diagnostic system can identify the bearing's load performance, tensile strength, seismic isolation performance, and self-recovery performance. According to the hysteresis curve of the friction pendulum bearing based on finite element analysis, the upper and lower bearing plates should always maintain horizontal movement in the same direction during bearing sliding, and the bearing can automatically return to its original center position without residual displacement under gravity without tangential excitation. The photoelectric sensor of the photoelectric displacement acquisition system is glued to the outside of the limiter on the lower bearing plate of the bearing. Different colored markers are drawn on the parallel lines of the upper bearing plate. When light shines on the markers, it is reflected and transmitted to the sensor, and the collected displacement signal is transmitted to the host computer by the connected data acquisition unit.

[0017] S203: The bearing acceleration acquisition system consists of an acceleration sensor, a data acquisition unit, and a host computer. This module aims to collect acceleration changes during the bearing isolation process so that the host computer, equipped with a monitoring and diagnostic system, can identify the hysteretic energy dissipation effect of the bearing. Based on the hysteresis curve of the friction pendulum bearing through finite element analysis, the area enclosed by the hysteresis curve during bearing sliding represents the hysteretic dissipation energy. If the equivalent stiffness and the bearing's elastic deformation energy remain constant, the bearing's hysteretic energy dissipation effect is stable. The acceleration sensor of the acceleration acquisition system is glued to the outside of the upper bearing plate limiter. When the upper bearing plate reciprocates and slides, the sensor transmits the collected acceleration signal to the host computer via the connected data acquisition unit.

[0018] S204: The host computer for the support condition monitoring and fault diagnosis system uses a Raspberry Pi microcomputer. The device features an SOC chip, USB interface, DSI display, Wi-Fi and Bluetooth modules, and can internally build software systems to process, analyze, and store diverse data. Furthermore, the Raspberry Pi can be equipped with expansion boards to further enrich the monitoring and diagnostic system's functionality. For example, it can install buzzers or indicator lights for system safety alarms, power control modules for long-term independent operation, wireless transmission modules for internet connectivity, and storage expansion modules to increase the system's data storage capacity.

[0019] Preferably, in the building friction pendulum seismic isolation bearing condition monitoring and fault diagnosis system provided by the present invention, S3 specifically refers to: the arrangement method of strain sensors, reflective photoelectric sensors, and acceleration sensors in the system application:

[0020] S301: The strain sensors are arranged in two locations: one embedded in the stress concentration point where the friction material connects to the upper support plate; the other embedded in the stress concentration point where the friction material connects to the lower support plate. Each location contains three sets of strain sensors, each set including: working strain gauge S1, compensating strain gauge S2, and compensating strain gauge S3. The strain gauges should be embedded during the support design and assembly process.

[0021] S302: The reflective photoelectric sensor is arranged outside the limiter of the lower support plate of the support, and different colored markings are drawn on the parallel line of the upper support plate.

[0022] S303: The acceleration sensor is located on the outside of the upper support plate limiter of the support.

[0023] Preferably, the present invention provides a system for monitoring and diagnosing the condition of a building friction pendulum seismic isolation bearing, wherein S4 specifically comprises: the bearing condition monitoring and fault diagnosis system software mounted on a Raspberry Pi host computer consists of three parts: data preprocessing, a deep learning model, and information storage.

[0024] S401: The function of the data preprocessing module is to set the sampling rate of the acquisition system and to restore the acquired data as accurately as possible to the most basic data of the actual service condition of the support. During signal preprocessing, the acquired data should first be calibrated and transformed to restore it to digital signal data with corresponding physical units. Simultaneously, due to factors such as sensor instability and environmental interference around the sensor, the data obtained by the acquisition system may deviate from the true values. Therefore, in the preprocessing stage, methods such as eliminating polynomial trend terms, smoothing, and noise filtering should be used to remove zero-point drift and eliminate noise.

[0025] S402: The deep learning model consists of two parts: feature extraction based on wavelet packet transform and a convolutional fault diagnosis model based on Siamese networks. In the process of multivariate data monitoring and diagnosis, various factors (such as illumination, temperature, and jitter) cause the acquired signals to contain varying degrees of noise. To address the difficulty in identifying faults during condition monitoring and fault diagnosis in noisy environments, a feature extraction method based on wavelet packet transform is adopted. Wavelet envelope basis has stronger time-frequency resolution than ordinary wavelet basis, which is conducive to extracting more accurate time-frequency local information from the original signal as target features, keenly sensing and distinguishing the generation of abnormal information in multivariate data, and ensuring the accuracy and reliability of recognition and classification. In view of the problem of imbalance of fault samples caused by the difficulty in collecting fault samples and the large number of normal samples, a convolutional neural network based on Siamese network is used for processing. By inputting the network through sample pairs, the dependence of the network on data samples during the model calculation can be effectively reduced. Moreover, the similarity measurement of Siamese network can also effectively reduce the distance between similar samples, highlighting the subtle differences between similar samples. Combined with the bearing performance technical parameters and the failure criteria of building friction pendulum seismic isolation bearing, the failure behavior of the bearing during service can be accurately diagnosed.

[0026] S403: The information storage module stores the data preprocessing and deep learning diagnostic results. The data preprocessing results serve as the information source for the deep learning model, while the deep learning diagnostic results are stored in the storage module. Under normal service conditions, the information storage of condition monitoring and fault diagnosis results is limited to short-term unit time. If any monitoring or diagnostic anomalies are detected, the information will be uploaded to the monitoring center via the Raspberry Pi host computer's Wi-Fi function to remind users of subsequent inspections and maintenance.

[0027] Compared with the prior art, the beneficial effects of the present invention are:

[0028] (1) This invention uses finite element analysis to model the friction pendulum seismic isolation bearing of a building, and determines the hysteresis curve and stress threshold variation characteristics of the bearing. This provides accurate numerical basis and criteria for the design of the bearing condition monitoring and fault diagnosis system, and solves the problem of the stability of accuracy and reliability in the system design process.

[0029] (2) The hardware composition of the system of the present invention can realize the real-time acquisition and monitoring of stress, acceleration and displacement states of each component unit inside the support and the overall structure of the support. At the same time, refining the system composition is conducive to comprehensively characterizing the dynamic characteristics of the friction pendulum support and providing real-time and reliable multi-dimensional data for the upper computer software.

[0030] (3) The system software of this invention can realize preprocessing, deep learning and information storage of multivariate data from the acquisition system. Signal preprocessing and time-frequency analysis eliminate background noise in the signal acquisition process and restore the support operating status as realistically as possible to the actual data; deep learning adopts feature extraction based on wavelet packet transform and convolutional fault diagnosis model based on Siamese network, which solves the problem of large number of parameters in multivariate data and high requirements for system hardware. At the same time, the Siamese network is input in the form of sample pairs, which effectively highlights the small differences between similar samples, and can obtain accurate and real-time monitoring and diagnosis results for fault problems of various sizes and types.

[0031] (4) This invention adopts a three-in-one system design approach integrating modeling, hardware, and software to comprehensively characterize and pattern recognize the dynamic characteristics of the friction pendulum support, thereby improving the system's computational accuracy, real-time monitoring, diagnostic reliability, and service stability. The engineering significance of condition monitoring lies in its ability to respond to abnormal information in a timely and accurate manner, while the engineering significance of fault diagnosis lies in its ability to accurately and effectively diagnose the type of fault, thus improving the safety and reliability of the support during long-term service. Attached Figure Description

[0032] Figure 1 This is a flowchart of the present invention;

[0033] Figure 2 This is a flowchart of the support performance analysis of the present invention;

[0034] Figure 3 This is a schematic diagram of the building friction pendulum seismic isolation bearing of the present invention;

[0035] Figure 4 This is a schematic diagram of the hysteresis curve of the support performance technical parameters of the present invention;

[0036] Figure 5 This is a schematic diagram of the hysteresis curve for the support performance analysis of the present invention;

[0037] Figure 6 This is a schematic diagram of stress distribution for the support performance analysis of the present invention;

[0038] Figure 7 This is a functional flowchart of the system hardware composition in this invention;

[0039] Figure 8This is a schematic diagram of the system hardware composition in this invention;

[0040] Figure 9 This is a schematic diagram showing the arrangement of the system's application to the support sensors;

[0041] Figure 10 This is a schematic diagram of the system's application to the support sensor, viewed from below.

[0042] Figure 11 This is a top view of the system's application to the support sensor layout;

[0043] Figure 12 This is a functional flowchart of the system software components in this invention;

[0044] Figure 13 This is a schematic diagram of the system software composition in this invention;

[0045] Figure 14 This is a table of performance failure criteria for building friction pendulum seismic isolation bearings in this invention.

[0046] In the figure: 1. Upper support plate, 2. Friction material A, 3. Friction material B, 4. Spherical cap, 5. Lower support plate, 6. Photoelectric sensor, 7. Accelerometer, 8. Marker point, 9. Strain sensor A, 10. Strain sensor B, 11. Strain sensor C, 12. Strain sensor D, 13. Strain sensor E, 14. Detailed Implementation

[0047] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0048] It should be noted that in the description of this invention, the terms "inner", "outer", "upper", "lower", "both sides", "one end", "the other end", "left", "right", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting this invention.

[0049] This invention provides a technical solution: a building friction pendulum seismic isolation bearing condition monitoring and fault diagnosis system comprising four parts: finite element analysis and calculation of building friction pendulum seismic isolation bearing performance, hardware composition of the bearing condition monitoring and fault diagnosis system, bearing sensor arrangement in system application, and software composition of the bearing condition monitoring and fault diagnosis system.

[0050] The finite element analysis and calculation of the performance of building friction pendulum seismic isolation bearings involves finite element modeling based on the bearing design scheme, calculating the hysteresis curve and stress threshold variation characteristics of the bearing under actual working conditions, combining the analysis results to derive the bearing's performance technical parameters, and diagnosing the bearing's failure behavior using performance technical failure criteria. The hardware components of the bearing condition monitoring and fault diagnosis system enable real-time acquisition and monitoring of stress, displacement, and acceleration states of each internal component and the overall structure of the bearing, providing real-time and reliable multi-dimensional data for the host computer software. The system application describes the arrangement of strain sensors, photoelectric sensors, and accelerometers during installation. The software components of the bearing condition monitoring and fault diagnosis system enable preprocessing, deep learning, and information storage of the acquired multi-dimensional data. Through the building friction pendulum seismic isolation bearing condition monitoring and fault diagnosis system, real-time and accurate monitoring and diagnosis of various sizes and types of working conditions can be performed.

[0051] like Figure 2 , Figure 3 , Figure 4 As shown, the finite element analysis and calculation of the performance of the building friction pendulum seismic isolation bearing is based on the design scheme of the friction pendulum seismic isolation bearing proposed in the superstructure of the building in actual engineering practice. The bearing structure model is constructed, and the hysteresis curve and stress threshold variation characteristics of the bearing under actual working conditions are calculated by finite element analysis. The bearing performance technical parameters are obtained by combining the analysis results.

[0052] Among them, the dynamic friction coefficient of the support and post-yield stiffness Calculate the expression:

[0053]

[0054] In the formula: The initial stiffness of the support; To support the vertical load force; This is the yield displacement; It is the equivalent radius of curvature.

[0055] Support hysteresis energy dissipation Calculate the expression:

[0056]

[0057] In the formula: This represents the elastic deformation energy of the support. Equivalent stiffness; This represents the horizontal displacement of the support.

[0058] The Von Mises stress calculation expression for a support structure is as follows:

[0059]

[0060] In the formula: This is the first principal stress; is the second principal stress; is the third principal stress.

[0061] like Figure 2 , Figure 5 and Figure 6 As shown, Table 1 ( ) is used to determine the performance failure criteria according to the requirements of the "Building Friction Pendulum Seismic Isolation Bearing" standard. Figure 14 Finite element parametric analysis is performed on the support model to obtain the critical thresholds of various technical parameters within the support criteria range and to accurately locate the support monitoring parts. By setting the support performance technical parameter indicators through the deep learning network of the S4: support condition monitoring and fault diagnosis system software component, the failure behavior of the support is diagnosed.

[0062] like Figure 7 and Figure 8 As shown, the hardware components of the support condition monitoring and fault diagnosis system include: a material strain performance acquisition system, a photoelectric displacement acquisition system, an acceleration acquisition system, and a Raspberry Pi microcomputer.

[0063] The bearing material strain performance acquisition system collects dynamic stress between the various components of the bearing by setting up a strain acquisition device, and then... Figure 6 The finite element analysis stress threshold results indicate that the stress concentration point of the support is located between the sliding surfaces of the upper support plate 1, the lower support plate 5, and the spherical cap 4. Since the polytetrafluoroethylene (PTFE) friction material is prone to structural deformation and damage under long-term high-pressure loads and environmental influences, strain sensors A9, B10, C11, D12, E13, and F14 are embedded at the stress concentration points where the friction material connects to the upper and lower support plates 1 and 4. These sensors allow for real-time acquisition of dynamic stress changes within the support, enabling the system to identify the mechanical properties of the support structure.

[0064] The bearing photoelectric displacement acquisition system collects the vertical and tangential displacement changes of the upper bearing plate 1 during the service life of the bearing by setting up a photoelectric displacement acquisition device. Figure 5 According to the hysteresis curve results of finite element analysis, the upper support plate 1 and the lower support plate 5 should always maintain horizontal movement in the same direction when the support slides, and the support can automatically return to its original center position without residual displacement under gravity without tangential excitation. The photoelectric sensor 6 of the photoelectric displacement acquisition system is glued to the outside of the limiter of the lower support plate 5 of the support. Different colored markers 8 are drawn on the parallel line of the upper support plate 1. When light shines on the markers, it is reflected and transmitted to the sensor, which can transmit the acquired displacement signal to the host computer for the system to identify the load performance, tensile strength, seismic isolation performance and self-recovery performance of the support.

[0065] The bearing acceleration acquisition system collects acceleration changes during the service life of the bearing by setting up an acceleration acquisition device, and then... Figure 5 According to the hysteresis curve results of finite element analysis, when the upper support plate 1 undergoes reciprocating sliding, the acceleration sensor 7, which is glued to the outside of the upper support plate limiter, can transmit the collected acceleration signal to the host computer. This is used to compare the deviation between the actual hysteretic energy dissipation effect of the support and the calculated value, and to analyze the factors affecting the seismic isolation performance of the support.

[0066] The host computer for the support condition monitoring and fault diagnosis system uses a Raspberry Pi microcomputer. The device includes an SOC chip, USB interface, DSI display, Wi-Fi and Bluetooth modules, allowing for the construction of internal software systems to process, analyze, and store diverse data. Furthermore, the Raspberry Pi can be equipped with expansion boards to further enrich the monitoring and diagnostic system's functionality. These expansion boards can be installed for various purposes, such as adding buzzers or indicator lights for system safety alarms, power control modules for long-term independent operation, wireless transmission modules for internet connectivity, and storage expansion modules to increase the system's data storage capacity.

[0067] like Figure 9-11 As shown, the arrangement of strain sensors A9, B10, C11, D12, E13, and F14, photoelectric sensor 6, and acceleration sensor 7 in the system application is as follows: Strain sensors A9, B10, and C11 are embedded in the stress concentration points where friction material B3 connects to the lower support plate 5; strain sensors D12, E13, and F14 are embedded in the stress concentration points where friction material A2 connects to the upper support plate 1. Each group contains a working strain gauge S1, a compensating strain gauge S2, and a compensating strain gauge S3. The strain gauges should be embedded during the support design and assembly process. Photoelectric sensor 6 is arranged outside the limiter of the lower support plate 5 of the support, and different colored markings 8 are drawn on the parallel line of the upper support plate 1. Acceleration sensor 7 is arranged outside the limiter of the upper support plate 1 of the support.

[0068] like Figure 12 and Figure 13 As shown, the bearing condition monitoring and fault diagnosis system software consists of three parts: data preprocessing, deep learning model, and information storage.

[0069] The data preprocessing module is responsible for setting the sampling rate of the acquisition system and restoring the acquired data to the most basic data reflecting the actual service condition of the support as accurately as possible. During signal preprocessing, the acquired data should first be calibrated and transformed to restore it to digital signal data with corresponding physical units. Simultaneously, due to factors such as sensor instability and environmental interference around the sensor, the data obtained by the acquisition system may deviate from the true values. Therefore, the preprocessing stage should employ methods such as eliminating polynomial trend terms, smoothing, and noise filtering to remove zero-point drift and eliminate noise.

[0070] The deep learning model consists of two parts: feature extraction based on wavelet packet transform and a convolutional fault diagnosis model based on Siamese networks. In multivariate data monitoring and diagnosis, various factors (such as illumination, temperature, and jitter) cause the acquired signals to contain varying degrees of noise. To address the difficulty of identifying faults in noisy environments, a feature extraction method based on wavelet packet transform is adopted. This method involves decomposing the sampled time-domain random signal sequence into wavelet packets, mapping it to random sequences within subspaces of the time scale domain. The most stable information state contained in the wavelet packet decomposition result represents the feature state corresponding to different types of targets. The entropy value of the optimal subspace and its position parameter in the complete binary tree are used as feature quantities, which can be classified and used as the identification result.

[0071] To address the imbalance of fault samples caused by the difficulty in collecting fault samples and the abundance of normal samples, a convolutional neural network based on Siamese networks is employed. By inputting sample pairs into the network, the dependence of the network on data samples during model computation is effectively reduced. Furthermore, the Siamese network similarity metric effectively reduces the distance between similar samples, highlighting subtle differences among them. The convolutional fault diagnosis model transforms the time-domain image generated after wavelet packet transform into an image recognition problem. Diagnostic analysis is performed by constructing a Siamese network, which includes two identical sub-convolutional networks for feature extraction. Each sub-network consists of alternating convolutional and pooling layers, all using 3x3 convolutional kernels with 32, 64, and 128 channels. Inputting fault sample pairs of the same or different categories into the sub-networks enables real-time monitoring of the support status and the identification and extraction of abnormal fault features.

[0072] The information storage module stores the data preprocessing and deep learning diagnostic results. The data preprocessing results serve as the information source for the deep learning model, while the deep learning diagnostic results are stored in the storage module. Under normal service conditions, the information storage of condition monitoring and fault diagnosis results is limited to short-term unit time. If any monitoring or diagnostic anomalies are detected, the information will be uploaded to the monitoring center via the Raspberry Pi host computer's Wi-Fi function to remind users of subsequent inspections and maintenance.

[0073] This invention provides timely and accurate monitoring and evaluation of bearing operating status and analysis and diagnosis of bearing health status, preventing bearings from suffering fatigue damage and structural failure due to long-term operating conditions and environmental factors, which would reduce the reliability and safety of bearing operation and ensure the stable operation of bearing load performance and seismic isolation performance.

[0074] Any aspects of this invention not described in detail are well-known to those skilled in the art.

[0075] Finally, it should be noted that the above specific embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the embodiments, those skilled in the art should understand that modifications and equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications and substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An intelligent friction pendulum isolation bearing and its condition monitoring and fault diagnosis system, characterized in that, It consists of four parts, including S1: finite element analysis and calculation of the performance of the friction pendulum seismic isolation bearing; S2: hardware of the bearing condition monitoring and fault diagnosis system; S3: arrangement of bearing sensors in system application; and S4: software of the bearing condition monitoring and fault diagnosis system. Specifically, S1 involves: determining the bearing structural model based on the friction pendulum seismic isolation bearing design scheme proposed for the superstructure of a building in actual engineering practice; calculating the hysteresis curve and stress threshold variation characteristics of the bearing under actual working conditions through finite element analysis; and obtaining the bearing performance technical parameters based on the finite element analysis results, including the bearing dynamic friction coefficient. and post-yield stiffness Calculate the expression: (1) In the formula: The initial stiffness of the support; To support the vertical load force; This is the yield displacement; The equivalent radius of curvature, Support hysteresis energy dissipation Calculate the expression: (2) In the formula: This represents the elastic deformation energy of the support. Equivalent stiffness; For the horizontal displacement of the support, The Von Mises stress calculation expression for the support structure is as follows: (3) In the formula: This is the first principal stress; This is the second principal stress; The third principal stress, By setting bearing performance technical parameters in the deep learning network of the S4 bearing condition monitoring and fault diagnosis system software, the failure behavior of the bearing can be diagnosed.

2. The intelligent friction pendulum isolation bearing and condition monitoring and fault diagnosis system according to claim 1, characterized in that, Specifically, S2 refers to the intelligent friction pendulum vibration isolation bearing and its condition monitoring and fault diagnosis system. The hardware mainly consists of four parts: S201: The bearing material strain performance acquisition system consists of strain sensors, a bridge box, a strain tester, a data acquisition unit, and a host computer. The bearing material strain performance acquisition system aims to collect the dynamic stress between the various components inside the bearing so that the host computer monitoring and diagnostic system can accurately identify the structural mechanical performance of the bearing during long-term service. According to the stress threshold change characteristics of the friction pendulum bearing through finite element analysis, the bearing stress is mainly concentrated between the spherical cap and the sliding surface of the upper and lower bearing plates. The commonly used friction material for the bearing is polytetrafluoroethylene with an elastic modulus of 280 MPa and a design strength of 30 MPa. Due to the long-term high-pressure load and environmental factors during the service of the bearing, the friction material is prone to structural deformation and damage. Strain sensors need to be embedded at the stress concentration points where the friction material connects to the upper and lower bearing plates to collect the dynamic stress changes inside the bearing in real time. The strain performance acquisition system has six sets of strain sensors, each set containing: working strain gauge S1, compensation strain gauge S2, and compensation strain gauge S3. The six sets of strain gauges form a full bridge and are connected to the strain tester through the bridge box. The data acquisition unit then transmits the collected strain signals to the host computer. S202: The bearing photoelectric displacement acquisition system consists of a reflective photoelectric sensor, a data acquisition unit, and a host computer. The bearing photoelectric displacement acquisition system aims to collect the vertical distance and tangential sliding distance between the upper and lower bearing plates during the service of the bearing, so that the host computer monitoring and diagnostic system can identify the load performance, tensile strength, vibration isolation performance, and self-recovery performance of the bearing. According to the hysteresis curve of the friction pendulum bearing finite element analysis, the upper and lower bearing plates should always maintain horizontal movement in the same direction when the bearing slides, and the bearing can automatically return to its original center position by gravity without residual displacement without tangential excitation. The photoelectric sensor of the photoelectric displacement acquisition system is glued to the outside of the limiter of the lower bearing plate of the bearing. Different colored markers are drawn on the parallel line of the upper bearing plate. When light shines on the markers, it is reflected and transmitted to the sensor, and the collected displacement signal can be transmitted to the host computer by the connected data acquisition unit. S203: The bearing acceleration acquisition system consists of an acceleration sensor, a data acquisition unit, and a host computer. The bearing acceleration acquisition system aims to collect the acceleration changes during the bearing isolation process so that the host computer can carry a monitoring and diagnostic system to identify the hysteretic energy dissipation effect of the bearing. According to the hysteretic curve of the friction pendulum bearing by finite element analysis, the area enclosed by the hysteretic curve when the bearing slides is the hysteretic dissipation energy. If the equivalent stiffness and the elastic deformation energy of the bearing remain constant, the hysteretic energy dissipation effect of the bearing is stable. The acceleration sensor of the acceleration acquisition system is glued to the outside of the upper bearing plate limiter of the bearing. When the upper bearing plate reciprocates and slides, the sensor will transmit the collected acceleration signal to the host computer through the connected data acquisition unit. S204: The host computer of the bearing condition monitoring and fault diagnosis system adopts a microcomputer and builds a software system inside to realize the processing, analysis and storage of multi-dimensional data. At the same time, an expansion board is installed to further enrich the functions of the monitoring and diagnosis system.

3. The intelligent friction pendulum isolation bearing and condition monitoring and fault diagnosis system according to claim 1, characterized in that, Specifically, S3 refers to the arrangement method of the strain sensor, reflective photoelectric sensor, and accelerometer in the system application: S301: The strain sensors are arranged in two locations. One location is embedded in the stress concentration point where the friction material connects to the upper support plate, and the other location is embedded in the stress concentration point where the friction material connects to the lower support plate. Each location contains three sets of strain sensors. Each set includes: working strain gauge S1, compensating strain gauge S2 and compensating strain gauge S3. The strain gauges should be embedded during the support design and assembly process. S302: The reflective photoelectric sensor is arranged outside the limiter of the lower support plate of the support, and different colored markings are drawn on the parallel line of the upper support plate. S303: The acceleration sensor is located on the outside of the upper support plate limiter of the support.

4. The intelligent friction pendulum isolation bearing and condition monitoring and fault diagnosis system according to claim 1, characterized in that, Specifically, S4 refers to the support status monitoring and fault diagnosis system software mounted on the host computer, which consists of three parts: data preprocessing, deep learning model, and information storage. S401: The function of the data preprocessing module is to set the sampling rate of the acquisition system and to restore the acquired data to the most basic data of the actual service status of the support. In the signal preprocessing process, the acquired data should first be calibrated and transformed to restore it to digital signal data with corresponding physical units. At the same time, due to factors such as unstable sensor performance and environmental interference around the sensor, the data obtained by the acquisition system will deviate from the true value. In the preprocessing stage, any one or more of the following methods are used to remove zero drift and eliminate noise: elimination of polynomial trend terms, smoothing, and noise reduction filtering. S402: The deep learning model consists of two parts: feature extraction based on wavelet packet transform and a convolutional fault diagnosis model based on Siamese network. In the process of multivariate data monitoring and diagnosis, the collected signals contain varying degrees of noise due to interference from various factors. To solve the problem of difficulty in identifying state monitoring and fault diagnosis in noisy environments, a feature extraction method based on wavelet packet transform is adopted. Wavelet packet basis has stronger time-frequency resolution than ordinary wavelet basis, which is conducive to extracting more accurate time-frequency local information in the original signal as target features. It can keenly perceive and distinguish the generation of abnormal information in multivariate data, ensuring the accuracy and reliability of identification and classification. In view of the problem of imbalance of fault samples caused by the difficulty in collecting fault samples and the large number of normal samples, a convolutional neural network based on Siamese network is used for processing. The network is input through sample pairs, which can effectively reduce the network's dependence on data samples during model calculation. Moreover, the similarity measurement method of Siamese network can also effectively reduce the distance between similar samples and highlight the small differences between similar samples. Combined with the bearing performance technical parameters and the performance failure criteria of friction pendulum seismic isolation bearing, it can accurately diagnose the failure behavior of the bearing during service. S403: The function of the information storage module is to store the data preprocessing and deep learning diagnostic results. The data preprocessing results are used as the information source for the deep learning model, and the deep learning diagnostic results are stored in the storage module. Under normal service conditions, the information storage of condition monitoring and fault diagnosis results is retained in short-term unit time. If there is an abnormality in monitoring and diagnosis, the information will be uploaded to the monitoring center through the upper computer Wi-Fi function to remind subsequent inspection and maintenance.

Images