Vibration detection unit, monitoring system, monitoring model construction method and monitoring simulation system

Abstract

The invention discloses a vibration monitoring simulation system, which comprises a vibration exciter, a rotary workbench fixed on the vibration exciter and a plurality of vibration detection units, and the vibration detection units are arranged on the rotary workbench and comprise an acrylic bottom plate, an acrylic base plate, a plurality of vibration sensors and a plurality of vibration sensors, the first copper thin film layer is attached to the surface of the acrylic bottom plate; the PET film layer is attached to the upper surface of the first copper film layer; the PTFE thin film layer is attached to the upper surface of the PET thin film layer; the second copper thin film layer is attached to the upper surface of the PTFE thin film layer; the acrylic top plate block is arranged on the upper surface of the second copper film layer; and the at least three springs are connected with the acrylic bottom plate and the acrylic top plate block in a manner of mutually equally dividing included angles, one end of each spring is connected with the acrylic bottom plate, and the other end of each spring is connected with the acrylic top plate block.

Description

Technical Field

[0001] The present invention belongs to the technical field of mechanical testing, and in particular relates to a vibration detection unit, a monitoring system, a monitoring model construction method, and a monitoring simulation system. Background Art

[0002] In the industrial and engineering fields, many aging aerospace, civil and mechanical systems are still in use. These systems may have accumulated structural damage, such as fatigue, cracks, structural deformation, and loose connections caused by vibration. These damages may cause equipment failures, such as accidents caused by fatigue cracks in aircraft wings, collapse of bridges due to wind vibration resonance, and cracking of ship structures. Therefore, an accurate vibration monitoring system is crucial. Through the monitoring system, mechanical status information can be collected and analyzed in a timely manner, and appropriate maintenance measures can be taken only when potential problems are discovered.

[0003] Vibration monitoring typically relies on sensors to detect changes in vibration. Existing sensors are mostly based on piezoelectric, piezoresistive or capacitive principles, but these methods often have some drawbacks, such as requiring an external power supply, high cost and susceptibility to environmental conditions. In addition, existing vibration monitoring systems often perform poorly when capturing low-amplitude vibrations or a wider range of mechanical frequencies, and cannot fully reflect the complexity of structural vibrations.

[0004] Public Content

[0005] One of the embodiments of the present disclosure is a vibration energy monitoring system based on triboelectric effect, which uses one or more vibration detection units as vibration sensors to obtain vibration signals of a monitored object and performs vibration monitoring on the monitored object or structure through a vibration monitoring model. Multiple detection units can be arranged in a grid shape and are arranged at points of interest for vibration monitoring of the monitored object.

[0006] The vibration detection unit is based on the triboelectric effect and is used to detect vibration energy. The unit includes:

[0007] Acrylic base plate;

[0008] A first copper film layer, the first copper film layer is attached to the surface of the acrylic base plate;

[0009] A PET film layer, the PET film layer being attached to the upper surface of the first copper film layer;

[0010] A PTFE film layer, wherein the PTFE film layer is attached to the upper surface of the PET film layer;

[0011] A second copper film layer, the second copper film layer being attached to the upper surface of the PTFE film layer;

[0012] An acrylic top plate, which is disposed on the upper surface of the second copper film layer;

[0013] At least three springs are provided to connect the acrylic bottom plate and the acrylic top plate in a form of equally divided angles between each other, one end of each spring is connected to the acrylic bottom plate, and the other end is connected to the acrylic top plate.

[0014] The present disclosure proposes a vibration energy monitoring system based on the triboelectric effect, which is self-powered, highly adaptable and has efficient energy recovery. The system can automatically generate electric charge when the structure vibrates, and is suitable for remote and inaccessible environments, reducing maintenance costs and improving the convenience of monitoring. Compared with existing monitoring methods, the system based on the triboelectric effect has significant advantages in capturing multi-directional and multi-frequency vibration information, and can provide more accurate data support, especially suitable for the analysis of complex vibration fields and real-time monitoring of structural damage. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The above and other objects, features and advantages of the exemplary embodiments of the present disclosure will become readily understood by reading the detailed description below with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present disclosure are shown in an exemplary and non-limiting manner, in which:

[0016] Figure 1 Flow chart of a vibration energy monitoring method based on triboelectric effect according to one of the embodiments of the present disclosure.

[0017] Figure 2 Schematic diagram of energy distribution algorithm and simulation comparison according to one embodiment of the present disclosure.

[0018] Figure 3 Schematic diagram of a three-dimensional structure based on the triboelectric effect according to one of the embodiments of the present disclosure.

[0019] Figure 4 Schematic diagram of sensor frequency identification according to one embodiment of the present disclosure.

[0020] Figure 5 A schematic diagram of a vibration sensor network according to one embodiment of the present disclosure.

[0021] 1——Acrylic bottom plate, 2——first copper film layer, 3——PET film layer, 4——PTFE film layer, 5——second copper film layer, 6——Acrylic top plate, 7——spring. DETAILED DESCRIPTION

[0022] The purpose of the present disclosure is to provide an all-round, multi-angle vibration monitoring method and system based on the triboelectric effect, so as to solve the problems in the prior art that vibration energy recovery is not comprehensive and multi-dimensional vibration cannot be accurately monitored. The principle of using the triboelectric effect for vibration energy monitoring is based on the coupling effect of friction electrification and electrostatic induction. When two materials with different triboelectric series come into contact and separate during the vibration process, charge transfer will occur, thereby forming charges on the surface of the material. Under the action of electrostatic induction, these charges will form a potential difference between the electrodes of the material, thereby generating current. This vibration energy monitoring method based on the triboelectric effect can monitor the distribution and intensity of vibration energy by analyzing the changes in current and potential difference. The method can also optimize its performance through simulation and experimental research. For example, by adjusting parameters such as the triboelectric series, contact area, and vibration frequency of the material, the energy conversion efficiency and monitoring accuracy can be improved.

[0023] According to one or more embodiments, Figure 1 As shown, a vibration energy monitoring method based on triboelectric effect is used for damage location and structural reinforcement in industrial and engineering structures. The method comprises the following steps: 101. First, a three-dimensional structure of triboelectric effect is constructed. The structure consists of an upper and lower friction layer, the upper layer is PTFE and PET film, and the lower layer is covered with a copper film as a back electrode. The acrylic mass block is supported by three springs symmetrically suspended at an angle of 120° to realize a three-dimensional structure based on triboelectric effect.

[0024] 102, using a signal generator to generate signals of different frequencies to simulate vibration environments at different frequencies. This step is to simulate the energy conversion process under actual vibration conditions.

[0025] 103. Set a rotating workbench, which is fixed on the exciter to collect vibration energy data from multiple angles and simulate vibration scenes from different angles. The voltage signal output by the sensor is converted into frequency domain data to obtain the main frequency information. By comparing the input frequency and the main frequency information, it is ensured that the frequency can be accurately identified. Here, the normal direction of the rotating workbench forms an angle with the vibration direction output by the exciter, and the angle refers to this angle. This means that the vibration output direction of the exciter does not need to be the same as the normal direction of the vibration monitoring unit.

[0026] 104. In order to analyze the influence of angle and frequency on energy output, the angle range of energy data acquisition was set to 0° to 90°, the frequency was set to 2Hz to 11Hz, and the step size was 0.2Hz. Through three repeated experiments, a multidimensional array of 1500 sets of data was formed.

[0027] 105, the CatBoost regression algorithm is used to construct a relationship model between angle, frequency and energy recovery efficiency. The angle and frequency are used as independent variables, and the energy recovery efficiency is used as the dependent variable for model training. Here, since the vibration input direction of the exciter (used to simulate the actual single-point vibration input) does not need to be consistent with the normal direction of the detection unit, even if the power output of the exciter remains constant, the explicit power of the vibration detection unit will vary with the angle difference. The ratio of this explicit power to the actual input power is the energy recovery efficiency. In order to ensure that the vibration energy detection accuracy of the vibration detection unit at a single point is not affected by directionality, its explicit power needs to be corrected.

[0028] 106. The model is validated using the test set and the performance index of the model is calculated using the 9-fold cross validation method. Through this method, the accurate energy conversion efficiency can be predicted under vibration scenarios of different frequencies and angles.

[0029] 107, from single-point precision energy harvesting to grid-type multi-point energy harvesting. A 3x3 sensor network is arranged on a 1mm thick aluminum plate to obtain sensor data.

[0030] 108. Develop a multi-source hierarchical differential evolution algorithm to generate energy cloud maps. Based on the collected vibration energy data, local optimization is performed to estimate the energy distribution of a single vibration source. In the hierarchical multi-source optimization stage, the interaction between multiple vibration sources is considered, the position and intensity of the vibration source are adjusted, and finally the energy cloud map is generated through global optimization.

[0031] The monitoring method of the disclosed embodiment is used to monitor and optimize the vibration energy conversion process based on the triboelectric effect, and improve the efficiency and accuracy of vibration energy conversion through precise experimental design, data collection, model construction and optimization.

[0032] According to one or more embodiments, a vibration energy monitoring method based on triboelectric effect comprises the following steps:

[0033] S1, attach a layer of polyethylene terephthalate (PET) film to the acrylic base plate, and cover it with a copper film as the back electrode. During production, use high-precision bonding equipment to ensure that the copper film and the PET film are flat and firmly attached. A layer of polytetrafluoroethylene (PTFE) film is pasted on the bottom of the acrylic top layer to form a friction pair with the copper electrode. On the top of the acrylic plate, three springs are symmetrically suspended at an angle of 120°. When installing, ensure that the springs are vertically connected to the acrylic mass block so that they can vibrate freely in the vertical and planar directions, thereby forming a three-dimensional structure based on the triboelectric effect, such as Figure 3 The three-dimensional structure here is a three-dimensional friction nanogenerator structure that can simultaneously respond to vibrations in the x, y, and z directions.

[0034] Figure 3 The English terms in include:

[0035] PTFE – Polytetrafluoroethylene, a plastic material with high chemical stability and a low coefficient of friction. PET – Polyethylene terephthalate, a synthetic polymer commonly used to make plastic bottles and other containers.

[0036] Copper - Copper is a metal that conducts electricity well and is often used to make electrodes.

[0037] Acrylic is a transparent plastic material with good weather resistance and processing performance. Spring is a mechanical component that can store elastic potential energy and is used to provide vibration.

[0038] Figure 3 A three-dimensional structure based on triboelectric effect for vibration energy monitoring is described. This structure consists of the following parts:

[0039] 1. Acrylic base plate - serves as the foundation of the entire structure, providing a sturdy platform to support other components.

[0040] 2. PET film - attached to the acrylic base plate as a layer of material for the triboelectric effect. PET film has good insulation properties and can form an effective triboelectric pairing with copper film.

[0041] 3. Copper film - Covered on the PET film as the back electrode. The conductive properties of the copper film make it effective in collecting charges generated by the triboelectric effect.

[0042] 4. PTFE film - pasted on the bottom of the acrylic top layer, forming a friction pair with the copper electrode. The low friction coefficient and high chemical stability of PTFE film make it an ideal triboelectric material.

[0043] 5. Spring and acrylic mass block - symmetrically suspended on the top of the acrylic plate at an angle of 120°. The spring provides the elastic force of vibration, while the acrylic mass block contacts and separates from the PTFE film during vibration, generating triboelectric charges.

[0044] The three-dimensional structural unit based on the triboelectric effect includes: an acrylic base plate; a first copper film layer, which is attached to the surface of the acrylic base plate; a PET film layer, which is attached to the upper surface of the first copper film layer; a PTFE film layer, which is attached to the upper surface of the PET film layer; a second copper film layer, which is attached to the upper surface of the PTFE film layer; an acrylic top plate, which is arranged on the upper surface of the second copper film layer; and at least three springs, which connect the acrylic base plate and the acrylic top plate in the form of equally dividing the angle between each other, and one end of each spring is connected to the acrylic base plate, and the other end is connected to the acrylic top plate.

[0045] During the vibration process, friction charges are generated due to the contact and separation between the PTFE film and the copper electrode. These charges can be collected by the copper electrode and converted into electrical signals, thereby realizing the monitoring of vibration energy. The vibration energy monitoring method based on the triboelectric effect can be used for structural health monitoring, vibration energy harvesting and self-powered sensor systems.

[0046] S2, uses a signal generator to generate signals of different frequencies and transmits them to the exciter through a power amplifier to simulate the vibration environment at different frequencies;

[0047] S3, collects the voltage signal output by the sensor, and converts the time domain data into frequency domain data through Fourier transform, and obtains the main frequency information. Compare the input frequency and the main frequency information to ensure that the sensor can accurately identify the frequency (taking the input frequency of 5.4Hz and 9.3Hz as an example), such as Figure 4 As shown,

[0048] Figure 4 English terms in:

[0049] Main frequency refers to the frequency component with the most concentrated energy in the signal.

[0050] Frequency (Hz) - Frequency (Hertz) indicates the number of times a periodic event is repeated per second.

[0051] Figure 4The frequency domain data obtained after Fourier transformation of the voltage signal output by the sensor is shown. The left figure shows the frequency domain data when the input frequency is 5.4Hz. The most significant peak in the figure appears at 5.4Hz, which indicates that the sensor successfully identifies the main frequency of the input. Other smaller peaks may represent noise or other frequency components in the signal. The right figure shows the frequency domain data when the input frequency is 9.3Hz. Similarly, the most significant peak appears at 9.3Hz, which again proves that the sensor can accurately identify the main frequency of the input. Through accurate frequency identification by the sensor, the operating status of the mechanical system is monitored in real time, and abnormal vibration is detected in time. S4, a rotating table is designed and fixed on the exciter to simulate vibration scenes from different angles, ensuring that each vibration data collection covers multiple directions to form comprehensive multi-angle vibration energy distribution data; the exciter generates vibrations of different frequencies and amplitudes, thereby exciting the vibration response of the rotating table or other structures; in the study of vibration energy monitoring, the exciter is used in conjunction with the sensor and data acquisition system to achieve accurate measurement and analysis of vibration energy.

[0052] S5, in order to analyze the influence of angle and frequency on energy output, the angle range of vibration energy data acquisition is set to the angle array of Angle = [0°, 10°, ..., 90°]. The frequency range is set to the frequency array of Frequency = [2, 2.2, ..., 11], with a step size of 0.2Hz. In the experiment, three repeated measurements were performed to ensure the reliability and consistency of the data, and finally a multidimensional data set containing 1500 sets of data was obtained; here, there are other ways to select the angle and frequency within the target range, including uniformly selecting no less than 10 points each, so it can be set as needed.

[0053] S6, angle and frequency are used as independent variables, and energy recovery efficiency is used as the dependent variable to form the initial data set. The preprocessed data set is randomly divided into a training set and a test set in a ratio of 7:3, ensuring that the training set is used to build the model and the test set is used to verify the generalization ability of the model. The division of the data set must ensure randomness and representativeness to avoid model bias during training.

[0054] S7, before building the model, set the key parameters of the CatBoost regression algorithm (other machine learning methods with regression functions are also acceptable). Specifically, it includes: setting the learning rate to control the convergence speed and accuracy of the model; setting the maximum depth of the decision tree to control the complexity of the model; defining the number of training iterations to ensure that the model fully learns the data features; setting the regularization parameters to avoid overfitting of the model. These parameters need to be tuned through experiments to find the best combination to ensure the accuracy and generalization ability of the model.

[0055] S8, after setting the parameters, the CatBoost regression model is trained using the angle and frequency in the training set as input features and the energy recovery efficiency as the target variable. The model continuously optimizes the decision tree through an iterative process, gradually improving the accuracy of the model in predicting energy recovery efficiency. CatBoost uses a gradient boosting algorithm to update the model weights by minimizing the prediction error in each iteration.

[0056] S9, use the test set to verify the trained model and calculate the performance indicators of the model (such as R 2 , MAE, RMSE, etc.) to evaluate the prediction ability of the model. During the initial verification process, if the model performance is poor, it is necessary to readjust the parameters or adjust the feature encoding method, and use cross-validation methods (such as 9-fold cross-validation) to reduce the risk of overfitting. This method can predict accurate energy conversion efficiency in vibration scenarios of different frequencies and angles;

[0057] S10, from single-point precision energy collection to grid-type multi-point energy collection, the present disclosure arranges a 3x3 sensor network on a 1mm thick aluminum plate (different sensor quantities and layouts can be used according to the actual required collection accuracy and structural area) to obtain sensor data, such as Figure 5 As shown. The sensor records the energy distribution data generated by different vibration sources on the plate;

[0058] S11, according to the vibration field energy attenuation formula, set the position and intensity parameters of the vibration source, where the position of the vibration source is S = (X0, Y0, α), and the sensor is located at (X i ,Y i ), the energy decay formula is as follows:

[0059]

[0060] Among them, α is the vibration source intensity, c is the adjustment factor, and this formula is used to predict the energy attenuation at the sensor; S12, divide the grid area in the vibration field area, assuming that there is only one vibration source in each grid area. Use the differential evolution algorithm for local optimization, and the optimization target is the square sum of the errors between the actual measurement value and the predicted value. The objective function of local optimization is the actual sensor energy measurement value E means and the predicted value E pred The sum of squared errors is:

[0061]

[0062] Where N is the number of sensors. After local optimization is completed, the optimal vibration source position and intensity in each grid are obtained;

[0063] S13, enter the hierarchical multi-source optimization stage, considering the interaction of multiple vibration sources. Introduce the interaction term:

[0064]

[0065] where σ is the spatial scale of the interaction, (X i ,Y i ) and (X j ,Y j ) are the coordinates of the two vibration sources. This interaction term is used to correct the energy coupling between the vibration sources and improve the overall fitting accuracy;

[0066] S14, based on the results of hierarchical optimization, enter the global optimization stage, input the initial parameters of all vibration sources into the differential evolution algorithm, and further minimize the energy prediction error sum of squares through global optimization:

[0067]

[0068] Where M is the total number of vibration sources. The result of global optimization determines the location and intensity of the final vibration source;

[0069] S15, through the above steps, an energy cloud map of the vibration field is generated to show the actual position and intensity distribution of the vibration source. The reliability can be verified by comparing the actual data with the simulated energy cloud map (taking the vibration sources as (50.0, 160.0) and (165.0, 75.0) as an example). Figure 2 shown. Figure 2 The English terms in are as follows:

[0070] Energy refers to the energy distribution of the vibration field in a specific area.

[0071] X(mm)——X coordinate (millimeter), indicating the horizontal position on a two-dimensional plane.

[0072] Y (mm)——Y coordinate (millimeter), indicating the vertical position on the two-dimensional plane.

[0073] Figure 2 In the figure, the left picture shows the actual energy distribution on a two-dimensional plane. The depth of the color in the figure represents the energy level, and the darker the color, the higher the energy. The positions of two vibration sources are marked in the figure, which are (40.7, 165.0) and (176.0, 69.3). These data are obtained through actual measurements. The right picture shows the energy cloud map obtained through simulation. Similarly, the depth of the color represents the energy level. The positions of two vibration sources are marked in the figure, which are (50.0, 160.0) and (165.0, 75.0). These are the vibration source positions predicted by simulation.

[0074] By comparing the energy distribution obtained from actual measurements (left figure) and the energy distribution predicted by simulation (right figure), the accuracy and reliability of the simulation model were verified.

[0075] According to the above-mentioned embodiments of the present disclosure, the innovative energy recovery and vibration monitoring method based on the triboelectric effect proposed in the present disclosure mainly solves the shortcomings of existing sensor technology in terms of vibration monitoring accuracy and applicability. The existing technology not only relies on an external power supply, but also has limitations in its energy recovery efficiency and vibration monitoring capabilities in complex vibration scenarios. It is difficult to fully capture tiny vibrations and multi-frequency vibration information, resulting in low monitoring accuracy and response speed. The present method realizes multi-angle and multi-frequency precise vibration monitoring through a three-dimensional structure, and accurately locates the energy concentration area, so that the system has higher sensitivity and adaptability. The present disclosure significantly improves the accuracy of vibration energy recovery and vibration monitoring, and can be widely used in aerospace, large equipment, bridge monitoring and other fields, especially suitable for vibration monitoring and damage assessment of remote or inaccessible complex structures.

[0076] In summary, the vibration energy monitoring method based on the triboelectric effect disclosed in the present invention comprises the following steps:

[0077] First, a three-dimensional structure based on a 120° symmetrical suspension spring is constructed. This structure realizes all-round three-dimensional energy recovery by combining two working modes: vertical contact separation and in-plane sliding. The simultaneous existence of these two working modes meets the requirement of a single device responding to vibrations in the x, y, and z directions.

[0078] In order to more accurately predict the capture effect of vibration energy, the present disclosure adopts the CatBoost algorithm to construct a connection model between angle, frequency and energy recovery efficiency.

[0079] Finally, the present invention develops a multi-source hierarchical differential evolution algorithm to generate energy cloud maps and accurately locate energy concentration areas. The algorithm can effectively integrate the energy information of multiple vibration sources and generate energy cloud maps that reflect the internal vibration energy distribution of the structure, thereby providing reliable data support for vibration damage analysis of the structure.

[0080] Under this technical process, the present invention can realize all-round three-dimensional vibration energy recovery, and accurately predict and locate vibration energy at multiple angles and frequencies, thereby improving the accuracy of structural damage analysis and the practicality of the energy recovery system, and has broad industrial and engineering application prospects.

[0081] Therefore, the core technical features of the present disclosure include:

[0082] 1. The three-dimensional structure based on the triboelectric effect realizes the combination of vertical and in-plane sliding modes, effectively captures multi-angle and multi-frequency vibration energy, and realizes the precise collection of single-point energy;

[0083] 2. Use machine learning methods to train the nonlinear relationship model between angle, frequency and energy recovery efficiency to accurately predict the energy capture effect, which is suitable for complex industrial and engineering environments;

[0084] 3. By defining the vibration energy attenuation function and combining the hierarchical local optimization with the global correction strategy, an energy cloud map reflecting the internal energy distribution of the structure is generated, the energy concentration area is accurately located, and reliable structural damage data support is provided.

[0085] 4. The execution process and implementation logic of the whole vibration monitoring and energy recovery method. The energy recovery here means that when vibration occurs at the detection point, the three-dimensional friction nanogenerator structure converts the vibration energy into electrical energy for output, thereby realizing the energy recovery process.

[0086] Accordingly, the beneficial effects of the present disclosure include:

[0087] 1. A three-dimensional structure based on the triboelectric effect is realized, which effectively combines the capture of vertical and horizontal vibration energy, breaks through the directional limitations of traditional sensor technology, and significantly improves the comprehensiveness of energy recovery.

[0088] This is because in the three-dimensional structure of the present disclosure, the capacitance of the parallel plate capacitor formed by PTFE and PET will change regardless of vertical or horizontal vibration, while the charge carried by the two friction materials due to friction does not change, so the charge induced on the copper electrode will continue to change with its vertical and horizontal vibrations, resulting in the formation of current in the external circuit. This current output is detected and analyzed as a signal for vibration energy recovery.

[0089] 2. The energy recovery efficiency can be predicted through angle and frequency information, and the vibration energy in complex environments can be predicted with high precision, which improves the intelligence level and accuracy of the vibration monitoring system.

[0090] 3. By placing sensors, a cloud map reflecting the energy distribution of the structural vibration field can be generated, providing reliable data support for the accurate positioning of structural vibration damage and enhancing the practical value of the system.

[0091] The above is only a specific implementation of the present disclosure, but the protection scope of the present disclosure is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present disclosure, and these modifications or replacements should be included in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be based on the protection scope of the claims.

Claims

1. A vibration detection unit, which is based on the triboelectric effect and is used for detecting vibration energy, characterized in that: This unit includes: Acrylic base plate; A first copper film layer, the first copper film layer is attached to the surface of the acrylic base plate; A PET film layer, the PET film layer being attached to the upper surface of the first copper film layer; A PTFE film layer, wherein the PTFE film layer is attached to the upper surface of the PET film layer; A second copper film layer, the second copper film layer being attached to the upper surface of the PTFE film layer; An acrylic top plate, which is disposed on the upper surface of the second copper film layer; At least three springs are provided to connect the acrylic bottom plate and the acrylic top plate in a form of equally divided angles between each other, one end of each spring is connected to the acrylic bottom plate, and the other end is connected to the acrylic top plate.

2. A vibration monitoring simulation system, characterized in that: The system comprises a vibration exciter, a rotating table fixed on the vibration exciter, and one or more vibration detection units according to claim 1, wherein the vibration detection unit is arranged on the rotating table.

3. A vibration monitoring model construction method, characterized in that: Based on the vibration monitoring simulation system as claimed in claim 2, a vibration data training sample set and a test sample set are collected.

4. The vibration monitoring model construction method according to claim 3, characterized in that: The model adopts CatBoost regression algorithm to take angle and frequency as independent variables and energy recovery efficiency as dependent variable to construct a relationship model between angle, frequency and energy recovery efficiency.

5. The vibration monitoring model construction method according to claim 4, characterized in that: The parameter settings of the monitoring model include: Learning rate, which controls the convergence speed and accuracy of the model; The maximum depth of the decision tree controls the model complexity; The number of training iterations to ensure that the model fully learns the data features; Regularization parameters to avoid overfitting of the model.

6. The vibration monitoring model construction method according to claim 4, characterized in that: The vibration data obtained by the vibration detection unit has an angle range of 0° to 90°, a frequency of 2 Hz to 11 Hz, and a step size of 0.2 Hz.

7. The vibration monitoring model construction method according to claim 6, characterized in that: Based on the collected vibration data, an energy cloud diagram is generated for structural vibration damage analysis.

8. The vibration monitoring model construction method according to claim 4, characterized in that: The performance of the monitoring model was calculated using a cross-validation method.

9. A vibration monitoring system, characterized in that: The vibration detection unit as claimed in claim 1 is used as a vibration sensor to obtain a vibration signal of a monitored object, and vibration monitoring of the monitored object is performed using the monitoring model as claimed in claim 3.

10. The system according to claim 9, characterized in that The system comprises a plurality of vibration detection units as claimed in claim 1 arranged in a grid shape, and the vibration detection units are arranged at points of interest for vibration monitoring of the monitored object.

Images