Intelligent monitoring method, system and equipment for fan bolt and medium

Abstract

The invention discloses an intelligent monitoring method, system, equipment and medium for a fan bolt, and the method comprises the steps: firstly, synchronously collecting a vibration acoustic signal and surface strain distribution data of a target bolt, then extracting acoustic features and strain features based on the vibration acoustic signal and the surface strain distribution data, carrying out the fusion of the acoustic features and the strain features to form a joint feature vector, and carrying out the recognition of the joint feature vector. The method comprises the steps of obtaining a joint feature vector of a bolt, inputting the joint feature vector into a mixed deep learning model, carrying out bolt state recognition to obtain a current health state of the bolt, and finally generating decision information according to the joint feature vector and the current health state of the bolt. By adopting the method, the fault early warning sensitivity and the environmental adaptability of fan bolt monitoring can be improved, and reliable guarantee is provided for safe operation of a wind generating set.

Description

Technical Field

[0001] This invention relates to the field of mechanical structure health monitoring technology, and in particular to a method, system, equipment and medium for intelligent monitoring of fan bolts. Background Technology

[0002] Wind turbine generators operate for extended periods in a complex and variable alternating load environment consisting of wind loads, gravity, and inertial forces. The connecting bolts of key components such as the tower, blades, and nacelle are subjected to enormous dynamic stresses, making them highly susceptible to preload loosening, loosening, and even fatigue fracture, which seriously threatens the structural safety and operational stability of the generator set.

[0003] Monitoring technologies for bolt status mainly include mechanical continuity monitoring, optical measurement methods, direct sensor monitoring, and image recognition methods, but all have significant limitations. Mechanical continuity monitoring primarily uses conductors such as enameled wire connected in series in the bolt connection circuit, relying on bolt loosening to break the conductor and trigger an alarm (e.g., Chinese patent CN222415084U). This method is essentially a binary switch monitoring, unable to quantify the degree of loosening, and requires manual repair after a break, resulting in high maintenance costs and an inability to determine status trends or provide early warnings. Optical measurement methods mainly determine bolt displacement by detecting the continuity or deviation of a light beam through a specific measuring hole (e.g., Chinese patent CN114263571B). This method requires extremely high cleanliness along the monitoring path; in the harsh environment of wind turbine sites with abundant oil and dust, lens contamination can easily lead to false alarms or failures, resulting in insufficient environmental adaptability and long-term reliability. Direct sensor monitoring mainly involves installing angle sensors, displacement sensors, or resistance strain gauges on bolts or flanges (e.g., Chinese patents CN120778353A and CN113090472B). While this method can provide quantitative information, it typically requires configuring an independent sensor and wired signal transmission line for each monitoring point. When implemented on wind turbine structures with numerous and widely distributed bolts, it suffers from problems such as complex wiring, high costs, and poor system scalability. Image recognition methods utilize cameras to capture the markings or lines at the bolt positions and identify displacement through image processing algorithms (e.g., Chinese patent CN212177330U). However, its effectiveness is heavily dependent on lighting conditions, and its performance degrades at night, in shadows, or under obstructed conditions. Furthermore, processing high-definition image data requires significant computing resources and transmission bandwidth, making real-time performance difficult to guarantee.

[0004] In summary, existing bolt condition monitoring systems suffer from problems such as limited monitoring dimensions, weak early fault warning capabilities, and poor environmental adaptability. Summary of the Invention

[0005] The purpose of this invention is to provide a method, system, equipment, and medium for intelligent monitoring of wind turbine bolts, thereby improving the fault early warning sensitivity and environmental adaptability of wind turbine bolt monitoring and providing reliable protection for the safe operation of wind turbine generator sets.

[0006] To achieve the above objectives, the present invention provides an intelligent monitoring method for wind turbine bolts, comprising: Simultaneously acquire vibration acoustic signals and surface strain distribution data of the target bolt; Based on the vibration acoustic signal and the surface strain distribution data, acoustic features and strain features are extracted and fused to form a joint feature vector; The joint feature vector is input into a hybrid deep learning model to identify the bolt state and obtain the current health status of the bolt. Decision information is generated based on the joint feature vector and the current health status of the bolt.

[0007] Optionally, the synchronous acquisition of the vibration acoustic signal and surface strain distribution data of the target bolt includes: Based on the nonlinear acoustic modulation effect, a dual-frequency composite ultrasonic excitation signal containing high-frequency excitation components and low-frequency excitation components is applied to the target bolt to collect the vibration acoustic signal of the target bolt. Simultaneously measure the micro-strain at multiple locations on the surface of the target bolt to generate surface strain distribution data.

[0008] Optionally, the step of extracting acoustic features and strain features based on the vibration acoustic signal and the surface strain distribution data, and fusing them to form a joint feature vector, includes: From the vibration acoustic signal, linear features, modulation index generated by nonlinear acoustic modulation effect, nonlinear coefficient and higher harmonic components are extracted as acoustic features; From the surface strain distribution data, strain gradient, distribution symmetry, strain extrema and their locations are extracted as strain features. The acoustic features and the strain features are fused to obtain a joint feature vector.

[0009] Optionally, the hybrid deep learning model includes a one-dimensional convolutional neural network and a long short-term memory network.

[0010] Optionally, the step of inputting the joint feature vector into a hybrid deep learning model to perform bolt state recognition and obtain the current health status of the bolt includes: When the joint feature vector matches the baseline feature vector acquired by the bolt under standard preload, the bolt condition is identified as normal. When the joint feature vector deviates from the baseline feature vector, the bolt status is identified as a warning state. When the joint feature vector exceeds a preset abnormal threshold, the bolt status is identified as an alarm state.

[0011] Optionally, generating decision information based on the joint feature vector and the current health status of the bolt includes: Based on the joint feature vector, the remaining useful life of the bolt is predicted using a time-series prediction model; The data sampling adjustment frequency is generated based on the current health status of the bolt.

[0012] To achieve the above objectives, the present invention also provides an intelligent monitoring system for wind turbine bolts, comprising: A composite sensor module is used to simultaneously acquire the vibration acoustic signals and surface strain distribution data of the target bolt; The feature extraction module is used to extract acoustic features and strain features based on the vibration acoustic signal and the surface strain distribution data, and fuse them to form a joint feature vector; The deep learning diagnostic module is used to input the joint feature vector into the hybrid deep learning model to identify the bolt status and obtain the current health status of the bolt. The information output module is used to generate decision information based on the joint feature vector and the current health status of the bolt.

[0013] Optionally, the composite sensor module includes a piezoelectric energy harvesting unit, which generates electricity by harvesting the mechanical vibration energy of the target bolt.

[0014] To achieve the above objectives, the present invention also provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the intelligent monitoring method for wind turbine bolts as described above.

[0015] To achieve the above objectives, the present invention also provides a computer-readable storage medium comprising a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform the intelligent monitoring method for wind turbine bolts as described above.

[0016] Compared with existing technologies, this invention provides a method, system, device, and medium for intelligent monitoring of wind turbine bolts. By simultaneously acquiring the vibration acoustic signals and surface strain distribution data of the target bolts, it overcomes the limitations of single-parameter monitoring in existing technologies, achieving multi-dimensional data complementarity to comprehensively reflect the true working condition of the bolts. By extracting features from the two types of data and fusing them to form a joint feature vector, combined with the strong nonlinear mapping and pattern recognition capabilities of a hybrid deep learning model, the sensitivity to identifying faults such as early preload relaxation and slight loosening of bolts can be improved. By generating decision information including state assessment and acquisition frequency and power consumption adjustment, the real-time performance of monitoring can be improved. This enhances the reliability of wind turbine bolt monitoring and provides a reliable guarantee for the safe operation of wind turbine generators. In addition, the application of a piezoelectric energy harvesting unit to generate electricity by collecting the mechanical vibration energy of the target bolts avoids the drawbacks of complex wiring and high costs, and enhances the adaptability to the harsh outdoor environment of wind turbines. Attached Figure Description

[0017] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of an intelligent monitoring method for wind turbine bolts provided in an embodiment of the present invention; Figure 2 This is a structural block diagram of a wind turbine bolt intelligent monitoring system provided in an embodiment of the present invention; Figure 3 This is a structural block diagram of a terminal device provided in an embodiment of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] See Figure 1 , Figure 1 This is a flowchart of a smart monitoring method for wind turbine bolts provided by an embodiment of the present invention. The smart monitoring method for wind turbine bolts includes steps S1 to S4: Step S1: Synchronously acquire the vibration acoustic signal and surface strain distribution data of the target bolt; In an optional embodiment, step S1 includes steps S101 to S102: Step S101: Based on the nonlinear acoustic modulation effect, apply a dual-frequency composite ultrasonic excitation signal containing high-frequency excitation components and low-frequency excitation components to the target bolt to collect the vibration acoustic signal of the target bolt. It should be noted that, through the nonlinear acoustic modulation effect, signal changes can be detected even with minute variations in bolt torque (<5%), whereas traditional methods typically require torque losses exceeding 15-20% for reliable detection. The nonlinear acoustic modulation principle is based on the parametric resonance effect of the bolt connection interface under the action of dual-frequency sound waves. When the bolt preload decreases, the stiffness of the contact interface nonlinearly increases, leading to an increase in the amplitude of the modulation frequency components. By calculating the nonlinear modulation index, the degree of bolt loosening can be quantified. : ; in, To modulate the sideband amplitude, This represents the high-frequency excitation amplitude. The MI value is positively correlated with bolt preload loss.

[0021] For example, a dual-frequency composite ultrasonic excitation containing high frequency (f_high, e.g., 100kHz) and low frequency (f_low, e.g., 5kHz) is applied to the bolt, and its response signal is acquired simultaneously. When the bolt preload changes, the nonlinear acoustic characteristics of the bolt contact interface (such as harmonic generation and modulation frequency components) will change significantly. The amplitude of the sideband signal (f_high±n*f_low) generated by the nonlinear effect of the bolt contact interface is monitored, and the MI value can be calculated.

[0022] Step S102: Simultaneously measure the micro-strain at multiple locations on the surface of the target bolt to generate surface strain distribution data.

[0023] For example, a flexible strain sensor is used, directly attached to the bolt surface, to measure the micro-strain distribution in the axial and circumferential directions of the bolt, thereby generating surface strain distribution data. It is worth noting that, compared to traditional strain gauges, the flexible sensor can adapt to the curved surface structure of the bolt and can acquire richer strain distribution information.

[0024] Step S2: Based on the vibration acoustic signal and the surface strain distribution data, extract acoustic features and strain features, and fuse them to form a joint feature vector; It should be noted that since the raw data collected (acoustic signals, strain distribution) cannot be directly used for diagnosis, it is first necessary to extract key characteristic indicators that can characterize the bolt condition.

[0025] In one optional embodiment, step S2 includes steps S201 to S203: Step S201: Extract linear features, modulation index generated by nonlinear acoustic modulation effect, nonlinear coefficient and higher harmonic components from the vibration acoustic signal as acoustic features; For example, linear features (amplitude, frequency, phase) are extracted from the vibration acoustic signal, and modulation index (MI), nonlinear parameter β, and amplitude of higher harmonic components are extracted from the nonlinear acoustic response. These parameters are extremely sensitive to changes in the micro-contact state of the bolted connection interface (early loosening).

[0026] Step S202: Extract strain gradient, distribution symmetry, strain extrema and their locations from the surface strain distribution data as strain features; For example, the strain gradient, distribution symmetry, maximum / minimum strain values ​​and their locations of the bolt as a whole can be extracted from the data of the distributed strain sensor array. These features can reflect the macroscopic uniformity of the bolt stress and whether there is local stress concentration.

[0027] Step S203: Fuse the acoustic features and the strain features to obtain a joint feature vector.

[0028] Specifically, the aforementioned acoustic and strain features are combined into a unified high-dimensional joint feature vector. This fused feature vector simultaneously contains microscopic and macroscopic, dynamic (acoustic) and static (strain) information about the bolt's state, providing a comprehensive data foundation for subsequent intelligent diagnosis.

[0029] Step S3: Input the joint feature vector into the hybrid deep learning model to identify the bolt status and obtain the current health status of the bolt; In one alternative embodiment, the hybrid deep learning model includes a one-dimensional convolutional neural network and a long short-term memory network.

[0030] In one optional embodiment, the step of inputting the joint feature vector into a hybrid deep learning model to perform bolt state recognition and obtain the current health status of the bolt includes: When the joint feature vector matches the baseline feature vector acquired by the bolt under standard preload, the bolt condition is identified as normal. When the joint feature vector deviates from the baseline feature vector, the bolt status is identified as a warning state. When the joint feature vector exceeds a preset abnormal threshold, the bolt status is identified as an alarm state.

[0031] It should be noted that when the bolt is under standard preload (i.e., healthy state), the vibration acoustic signal and surface strain distribution data of the target bolt are collected simultaneously as a baseline dataset, and the baseline feature vector is calculated accordingly.

[0032] For example, the hybrid deep learning model is a hybrid model employing a one-dimensional convolutional neural network (1D-CNN) and a long short-term memory network (LSTM). This model has been trained on massive amounts of laboratory data and is capable of recognizing multiple patterns ranging from healthy to faulty. Its network structure specifically includes a 1D-CNN branch for processing acoustic signals and extracting local features in the frequency domain; an LSTM branch for processing strain time-series data and capturing long-term dependencies; and a feature fusion layer for concatenating the feature vectors from the two branches.

[0033] Specifically, the obtained high-dimensional joint feature vector data is input into a hybrid model of 1D-CNN and LSTM. This hybrid model learns the complex mapping relationship between this "joint feature vector" and the bolt's health status, thus outputting a probability distribution. For example: normal state probability: 95%; warning state (early loosening) probability: 4%; alarm state (significant loosening) probability: 1%. Then, based on the principle of maximum probability, the bolt's current state is determined to be one of the following three categories: Normal state: The joint feature vector is in high agreement with the baseline state, indicating that the bolt connection is reliable.

[0034] Warning status: A slight deviation in the joint feature vector indicates that the bolt may loosen early (e.g., preload loss <10%), which requires close monitoring.

[0035] Alarm status: The joint feature vector shows a significant anomaly, indicating that the bolts are obviously loose or have cracks, posing a safety hazard.

[0036] Step S4: Generate decision information based on the joint feature vector and the current health status of the bolt.

[0037] In an optional embodiment, step S4 includes steps S401 to S402: Step S401: Based on the joint feature vector, predict the remaining useful life of the bolt using a time-series prediction model; For example, the joint feature vector and bolt health status results of each cycle are continuously stored to form a bolt health status time series. Then, time series prediction models such as long short-term memory networks are used to predict the future development trend of key features (such as modulation index and strain gradient). When the predicted value exceeds the fault threshold, the remaining useful life of the bolt is calculated and output to provide a quantitative basis for maintenance planning.

[0038] Step S402: Generate a data sampling adjustment frequency based on the current health status of the bolt.

[0039] For example, when the bolt's current health status is normal, a "normal" heartbeat packet is sent to the monitoring center, the data is archived, and the existing monitoring frequency is maintained; when the bolt's current health status is in an alert state, an alert notification is immediately sent to maintenance personnel, prompting them to pay attention to the specific bolt. At the same time, the monitoring frequency is increased, and a more refined diagnostic procedure is initiated; when the bolt's current health status is in an alarm state, the highest level alarm information is sent to the main control system and relevant personnel, recommending prompt inspection, and the monitoring frequency is increased to continuous or quasi-continuous mode.

[0040] Furthermore, the diagnostic results and status level can be transmitted back to the sensing modules related to the wind turbine bolts. Based on this feedback, power consumption strategies can be dynamically adjusted (e.g., temporarily relaxing power consumption limits and prioritizing data acquisition during warning / alarm states), achieving a fully closed-loop adaptive optimization from perception to decision-making.

[0041] In summary, the intelligent monitoring method for wind turbine bolts provided by this invention overcomes the limitations of single-parameter monitoring in existing technologies by simultaneously collecting vibration acoustic signals and surface strain distribution data of the target bolts. This enables multi-dimensional data complementarity to comprehensively reflect the actual working conditions of the bolts. By extracting features from the two types of data and fusing them to form a joint feature vector, combined with the strong nonlinear mapping and pattern recognition capabilities of a hybrid deep learning model, the sensitivity to identifying faults such as early preload relaxation and slight loosening of bolts can be improved. By generating decision information that includes state assessment and adjustment of acquisition frequency and power consumption, the real-time performance of monitoring can be improved. This enhances the reliability of wind turbine bolt monitoring and provides a reliable guarantee for the safe operation of wind turbine generators.

[0042] Based on the above method items, the present invention provides corresponding system items embodiments.

[0043] See Figure 2 , Figure 2 This is a structural block diagram of a wind turbine bolt intelligent monitoring system provided in an embodiment of the present invention. The wind turbine bolt intelligent monitoring system includes: The composite sensor module 21 is used to simultaneously acquire the vibration acoustic signal and surface strain distribution data of the target bolt; Feature extraction module 22 is used to extract acoustic features and strain features based on the vibration acoustic signal and the surface strain distribution data, and fuse them to form a joint feature vector; The deep learning diagnostic module 23 is used to input the joint feature vector into the hybrid deep learning model to identify the bolt status and obtain the current health status of the bolt. The information output module 24 is used to generate decision information based on the joint feature vector and the current health status of the bolt.

[0044] In one alternative embodiment, the composite sensor module 21 includes: The nonlinear acoustic excitation and detection unit is used to apply a dual-frequency composite ultrasonic excitation signal containing high-frequency excitation components and low-frequency excitation components to the target bolt based on the nonlinear acoustic modulation effect, so as to acquire the vibration acoustic signal of the target bolt. A distributed flexible strain sensor array is used to simultaneously measure the micro-strain at multiple locations on the surface of the target bolt, generating surface strain distribution data.

[0045] In one optional embodiment, the feature extraction module 22 includes: The acoustic feature extraction module is used to extract linear features, modulation index generated by nonlinear acoustic modulation effect, nonlinear coefficient and higher harmonic components from the vibration acoustic signal as acoustic features; The strain feature extraction module is used to extract strain gradient, distribution symmetry, strain extrema and their locations as strain features from the surface strain distribution data. The feature fusion module is used to fuse the acoustic features and the strain features to obtain a joint feature vector.

[0046] In one optional embodiment, the deep learning diagnostic module 23 is configured to: When the joint feature vector matches the baseline feature vector acquired by the bolt under standard preload, the bolt condition is identified as normal. When the joint feature vector deviates from the baseline feature vector, the bolt status is identified as a warning state. When the joint feature vector exceeds a preset abnormal threshold, the bolt status is identified as an alarm state.

[0047] In one optional embodiment, the information output module 24 includes: The trend prediction and life assessment unit is used to predict the remaining useful life of the bolt based on the joint feature vector and using a time-series prediction model. The feedback adjustment unit is used to generate a data sampling adjustment frequency based on the current health status of the bolt.

[0048] It should be noted that the intelligent monitoring system for wind turbine bolts provided in this embodiment of the invention is used to execute all the process steps of the intelligent monitoring method for wind turbine bolts in the above embodiment. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0049] In one alternative embodiment, the composite sensor module includes a piezoelectric energy harvesting unit that generates electricity by harvesting the mechanical vibration energy of the target bolt.

[0050] Exemplarily, this invention employs a fully self-powered scheme, efficiently converting the mechanical vibration of the bolt into electrical energy to power the composite sensor module, thereby achieving complete energy self-sufficiency and reducing maintenance requirements. First, the piezoelectric transducer of the electrical energy harvesting unit uses high-performance lead zirconate titanate piezoelectric ceramics such as PZT-5H, giving it high electromechanical conversion efficiency to generate greater strain and higher output voltage under minor vibrations, directly converting the periodic mechanical vibration energy of the bolt into alternating current. Then, through a multistable magnetic coupling structure, utilizing the nonlinear repulsion / attraction between magnets, the system possesses two or more stable states (i.e., "multistable"), effectively widening the operating frequency band and ensuring good energy harvesting performance within the typical vibration frequency range of the wind turbine (1-30Hz). Finally, a supercapacitor is used to store the harvested energy and continuously power the sensing nodes when there is no vibration.

[0051] Specifically, the composite sensor module is mounted onto the target bolts (such as tower bolts or blade flange bolts) using a specialized fixture. After installation, the built-in piezoelectric energy harvesting unit begins to collect energy from environmental vibrations and charge the supercapacitor.

[0052] It is worth noting that the self-powered configuration in this embodiment of the invention avoids the drawbacks of complex wiring and high cost. The dual-parameter fusion acquisition and the anti-interference design of the model enhance the adaptability to harsh environments such as oil pollution, dust, and variable lighting in the field.

[0053] In addition, the intelligent monitoring system for wind turbine bolts also includes a communication network for establishing low-power wireless communication connections (such as LoRaWAN or ZigBee) between the composite sensor module and a nearby gateway or between the feature extraction module and the deep learning diagnostic module.

[0054] This invention also provides a terminal device, such as... Figure 3 The diagram shown is a structural block diagram of a preferred embodiment of a terminal device provided by the present invention. The terminal device includes a processor 31, a memory 32, and a computer program stored in the memory 32 and configured to be executed by the processor 31. When the processor 31 executes the computer program, it implements the intelligent monitoring method for wind turbine bolts as described in any of the above embodiments.

[0055] In addition, embodiments of the present invention also provide a computer-readable storage medium, the computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform the intelligent monitoring method for wind turbine bolts as described in any of the above embodiments.

[0056] When the processor 31 executes the computer program, it implements the steps in the above-described embodiment of the intelligent monitoring method for wind turbine bolts, for example... Figure 1 All steps of the intelligent monitoring method for wind turbine bolts shown are illustrated. Alternatively, when the processor 31 executes the computer program, it implements the functions of each module in the above-described intelligent monitoring system embodiment for wind turbine bolts, for example... Figure 2 The functions of each module in the intelligent monitoring system for wind turbine bolts are shown.

[0057] Preferably, the computer program can be divided into one or more modules / units, which are stored in the memory 32 and executed by the processor 31 to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program in the terminal device.

[0058] The processor 31 can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor, or the processor 31 can be any conventional processor. The processor 31 is the control center of the terminal device, connecting various parts of the terminal device through various interfaces and lines.

[0059] The memory 32 mainly includes a program storage area and a data storage area. The program storage area can store the operating system, applications required for at least one function, etc., and the data storage area can store related data, etc. In addition, the memory 32 can be a high-speed random access memory, or a non-volatile memory, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, and a flash card, etc., or the memory 32 can also be other volatile solid-state storage devices.

[0060] It should be noted that the aforementioned terminal devices may include, but are not limited to, processors and memory, as will be understood by those skilled in the art. Figure 3 The structural block diagram shown is merely a structural example of the terminal device described above and does not constitute a limitation on the structure of the terminal device. The terminal device may include more or fewer components than shown, or combine certain components, or use different components.

[0061] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for intelligent monitoring of wind turbine bolts, characterized in that, include: Simultaneously acquire vibration acoustic signals and surface strain distribution data of the target bolt; Based on the vibration acoustic signal and the surface strain distribution data, acoustic features and strain features are extracted and fused to form a joint feature vector; The joint feature vector is input into a hybrid deep learning model to identify the bolt state and obtain the current health status of the bolt. Decision information is generated based on the joint feature vector and the current health status of the bolt.

2. The intelligent monitoring method for wind turbine bolts as described in claim 1, characterized in that, The synchronous acquisition of vibration acoustic signals and surface strain distribution data of the target bolt includes: Based on the nonlinear acoustic modulation effect, a dual-frequency composite ultrasonic excitation signal containing high-frequency excitation components and low-frequency excitation components is applied to the target bolt to collect the vibration acoustic signal of the target bolt. Simultaneously measure the micro-strain at multiple locations on the surface of the target bolt to generate surface strain distribution data.

3. The intelligent monitoring method for wind turbine bolts as described in claim 1, characterized in that, The process of extracting acoustic features and strain features based on the vibration acoustic signal and the surface strain distribution data, and fusing them to form a joint feature vector, includes: From the vibration acoustic signal, linear features, modulation index generated by nonlinear acoustic modulation effect, nonlinear coefficient and higher harmonic components are extracted as acoustic features; From the surface strain distribution data, strain gradient, distribution symmetry, strain extrema and their locations are extracted as strain features. The acoustic features and the strain features are fused to obtain a joint feature vector.

4. The intelligent monitoring method for wind turbine bolts as described in claim 1, characterized in that, The hybrid deep learning model includes a one-dimensional convolutional neural network and a long short-term memory network.

5. The intelligent monitoring method for wind turbine bolts as described in claim 1, characterized in that, The step of inputting the joint feature vector into a hybrid deep learning model to identify the bolt state and obtain the current health state of the bolt includes: When the joint feature vector matches the baseline feature vector acquired by the bolt under standard preload, the bolt condition is identified as normal. When the joint feature vector deviates from the baseline feature vector, the bolt status is identified as a warning state. When the joint feature vector exceeds a preset abnormal threshold, the bolt status is identified as an alarm state.

6. The intelligent monitoring method for wind turbine bolts as described in claim 1, characterized in that, The step of generating decision information based on the joint feature vector and the current health status of the bolt includes: Based on the joint feature vector, the remaining useful life of the bolt is predicted using a time-series prediction model; The data sampling adjustment frequency is generated based on the current health status of the bolt.

7. A smart monitoring system for wind turbine bolts, characterized in that, include: A composite sensor module is used to simultaneously acquire the vibration acoustic signals and surface strain distribution data of the target bolt; The feature extraction module is used to extract acoustic features and strain features based on the vibration acoustic signal and the surface strain distribution data, and fuse them to form a joint feature vector; The deep learning diagnostic module is used to input the joint feature vector into the hybrid deep learning model to identify the bolt status and obtain the current health status of the bolt. The information output module is used to generate decision information based on the joint feature vector and the current health status of the bolt.

8. The intelligent monitoring system for wind turbine bolts as described in claim 7, characterized in that, The composite sensor module includes a piezoelectric energy harvesting unit, which generates electricity by collecting the mechanical vibration energy of the target bolt.

9. A terminal device, characterized in that, The method includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the intelligent monitoring method for wind turbine bolts as described in any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform the intelligent monitoring method for wind turbine bolts as described in any one of claims 1 to 6.

Images