Wind generating set cable life measuring device and control method

Abstract

The invention discloses a wind generating set cable life measuring device and a control method, and the device comprises a data collection layer which is used for collecting the external environment data of a wind generating set, the operation state data of the wind generating set, and the test data of different positions of a power cable of the wind generating set; the data processing layer is used for carrying out filtering, noise reduction and feature extraction on the data acquired by the data acquisition layer to obtain vibration feature parameters; and the machine learning layer is used for carrying out mode identification and vibration mode prediction on the vibration characteristics of the power cable through a multi-dimensional machine learning algorithm based on the vibration characteristic parameters extracted by the data processing layer to obtain a residual life measurement result of the power cable. The device for measuring the service life of the cable of the wind generating set realizes accurate evaluation and prediction of the service life of the power cable through a multi-level data processing and machine learning method.

Description

Technical Field

[0001] The invention relates to the field of power cables for wind turbine generator sets, and in particular to a wind turbine generator set cable life measuring device and a control method. Background Art

[0002] The operation and maintenance cost of a wind turbine generator set during its entire life cycle is an important component of the total machine cost. There are many systems and components involved in the maintenance of the unit. Specifically, this patent technology focuses on the power cables of the unit. Since the unit shakes during operation, and the number and weight of the power cables of the unit are large, the power cables often slide down under the load of gravity and inertia in a vibrating environment. After sliding down, the cables are prone to friction with the platform, etc., which leads to wear and tear, thus posing electrical safety hazards.

[0003] There are many ways to deal with this pain point problem. Existing patent technologies can be roughly divided into four categories: the first category uses clamping and fixing restraint devices, such as patents CN218733082U, CN109494658B, CN218633153U, etc.; the second category uses cable protection devices, such as patents CN209875394U, CN208461372U, etc.; the third category changes the wiring and routing methods through process optimization, such as patent CN1878329B, etc.; the fourth category directly changes the design plan, cancels multiple power cables, and adopts a collector ring solution, such as patents CN12583793B, CN218498538U, etc.

[0004] However, the applicant discovered that the prior art has the following deficiencies in the process of implementing the present invention:

[0005] 1. Existing technical means or solutions are all based on the static load-bearing perspective, without considering the dynamic load effect of the cable, while the dynamic effect dominates the cable slippage or wear;

[0006] 2. Existing technical means or solutions have a limited lifespan and need to be replaced in a certain period, such as about 2 years, which invisibly increases the operation and maintenance costs of the entire life cycle;

[0007] 3. Most existing methods use local extrusion and clamping of cables, which is not friendly to local stress of cables;

[0008] 4. Most of the power cables are 185 and 240 type cables, which have large diameters, high hardness, and difficult wiring process implementation. In practice, it is difficult to lay out according to the designed linear layout in an environment with limited space;

[0009] 5. The existing technical solutions do not fundamentally solve the problem of power incentive source and consumption;

[0010] 6. A large number of existing units cannot use the collector ring solution to replace the original multiple power cable outlet solution;

[0011] 7. The essence of cable slippage and wear is still about cable life and safety. There is no cable status monitoring and life control method in the existing technical solutions. Summary of the invention

[0012] In view of this, an embodiment of the present invention provides a wind turbine generator cable life measurement device and control method to solve the above technical problems.

[0013] To achieve the above object, in a first aspect, a wind turbine generator cable life measuring device is provided, wherein the device comprises:

[0014] The data collection layer is used to collect the external environment data of the wind turbine generator set, the operating status data of the wind turbine generator set, and the test data of different positions of the power cables of the wind turbine generator set;

[0015] The data processing layer is used to filter, reduce noise and extract features from the data collected by the data collection layer to obtain vibration characteristic parameters;

[0016] The machine learning layer is used to perform pattern recognition and vibration mode prediction of the vibration characteristics of the power cable through a multi-dimensional machine learning algorithm based on the vibration characteristic parameters extracted by the data processing layer, so as to obtain the remaining life measurement result of the power cable.

[0017] In a second aspect, a method for controlling the life of a cable of a wind turbine generator set is provided, the method comprising:

[0018] Collect external environment data, operating status data and test data of different positions of power cables of wind turbines;

[0019] Filtering, denoising and feature extraction are performed on the collected data to obtain vibration characteristic parameters;

[0020] Based on the vibration characteristic parameters extracted in the data processing stage, a multi-dimensional machine learning algorithm is used to perform pattern recognition and vibration mode prediction of the vibration characteristics of the power cable to obtain a remaining life measurement result of the power cable;

[0021] According to the remaining life measurement result, the life of the power cable is improved by adjusting the local acceleration of the power cable.

[0022] In a third aspect, an electronic device is provided, comprising:

[0023] one or more processors;

[0024] a storage device for storing one or more programs,

[0025] When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in the first aspect.

[0026] According to a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored, and when the program is executed by a processor, the method described in the first aspect is implemented.

[0027] The above technical solution has the following beneficial technical effects:

[0028] The wind turbine cable life measurement device achieves accurate assessment and prediction of the life of power cables through multi-level data processing and machine learning methods. The device first comprehensively collects detailed data on the external environment, operating status and different positions of the wind turbine cables through the data acquisition layer. The data processing layer effectively removes data noise and extracts key vibration characteristic parameters through advanced filtering, noise reduction and feature extraction techniques, improving data quality and analysis accuracy. The machine learning layer uses multi-dimensional algorithms to perform pattern recognition and prediction of cable vibration characteristics, which can accurately assess the remaining service life of the cable, thereby realizing intelligent and precise health status monitoring of wind turbine cables, which is conducive to preventing cable failures in advance, optimizing maintenance strategies, and reducing equipment maintenance costs and safety risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The accompanying drawings are used to better understand the present invention and do not constitute an improper limitation of the present invention.

[0030] Figure 1A is a schematic structural diagram of a wind turbine generator set according to an embodiment of the present invention;

[0031] Figure 1B The embodiment of the present invention Figure 1A A cross-sectional view of the AA section;

[0032] Figure 2 1 is an overall schematic diagram of a wind turbine generator set and a power cable wear mitigation device according to an embodiment of the present invention;

[0033] Figure 3 It is a partial schematic diagram of the assembly positions of the first wear reduction device, the second wear reduction device, and the third wear reduction device in the embodiment of the present invention;

[0034] Figure 4 It is a schematic diagram of the specific structure of the first wear reduction device in an embodiment of the present invention;

[0035] Figure 5 for Figure 4 Middle BB cutaway structure diagram;

[0036] Figure 6is a schematic diagram of the specific structure of the second wear reduction device in an embodiment of the present invention;

[0037] Figure 7 for Figure 6 Middle CC cutaway structure diagram;

[0038] Figure 8 for Figure 6 Middle DD cutaway structure diagram;

[0039] Fig. 9 This is a schematic diagram of the specific structure of the third wear reduction device in an embodiment of the present invention;

[0040] Fig.10 for Fig. 9 EE cutaway diagram of the structure;

[0041] Fig.11 This is the cable status measurement point layout of an embodiment of the present invention;

[0042] Fig.12 is a cable status monitoring, analysis and control diagram of an embodiment of the present invention;

[0043] Fig.13 is a comparison diagram of cable predicted mode and real mode classification according to an embodiment of the present invention;

[0044] Fig.14 is a comparison diagram of cable predicted stress and actual stress in an embodiment of the present invention;

[0045] Fig.15 is a curve diagram showing the relationship between cable temperature, vibration and fatigue damage in an embodiment of the present invention;

[0046] Fig.16 is a valve opening and damping control curve diagram of an embodiment of the present invention;

[0047] Fig.17 1 is a graph of effect verification curve (active control of valve opening) of an embodiment of the present invention;

[0048] Fig.18 is a functional block diagram of a wind turbine cable life measuring device according to an embodiment of the present invention;

[0049] Fig.19 is a flow chart of a method for controlling the life of cables of a wind turbine generator set according to an embodiment of the present invention;

[0050] Fig. 20 It is a schematic diagram of the structure of a computer system according to an embodiment of the present invention.

[0051] Description of Figure Numbers:

[0052] 1. Blade; 2. Hub; 3. Drive shaft; 4. Gearbox; 5. Generator; 6. Nacelle; 7. Power cable; 8. Saddle platform; 9. Tower; 10. Converter; 11. Cable clamp;

[0053] 12. A first wear reduction device; 13. A second wear reduction device; 14. A third wear reduction device;

[0054] 12-1, first shell; 12-2, suspended object; 12-3, first damping fluid; 12-4, first isolation ring; 12-5, first buffer;

[0055] 13-1, liquid column; 13-2, flexible fin; 13-3, second damping liquid; 13-4, second shell; 13-5, second buffer; 13-6, second isolation ring; 13-7, regulating valve;

[0056] 14-1, third shell; 14-2, boundary fin; 14-3, spiral plate; 14-4, third damping fluid;

[0057] 15. Unit status measuring point; 16. First measuring point layout position; 17. Second measuring point layout position; 18. Third measuring point layout position. DETAILED DESCRIPTION

[0058] The following is a description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0059] The embodiments of the present invention solve at least one of the following technical problems: starting from the perspective of energy dissipation, solving the problem of sliding and wear of multiple power cables of wind turbines; solving the uneven force and application problems of power cables of wind turbines; improving the operating reliability of power cables and reducing the operating and maintenance costs throughout the life cycle; and proposing a life measurement device and control method suitable for wind turbine cables.

[0060] Figure 1A is a schematic diagram of a wind turbine generator set proposed in the present invention, Figure 1B The embodiment of the present invention Figure 1AIn the cross-sectional view of the middle AA section, the wind turbine generator set mainly includes blades 1, hub 2, transmission shaft 3, gearbox 4, generator 5, nacelle 6, power cable 7, saddle platform 8, tower 9, converter 10 and other components. The blade 1 absorbs the incoming wind energy and converts the wind energy into the rotational mechanical energy of the hub 2 and transmission shaft 3. After that, the coaxial gearbox 4 is driven to increase the speed. The generator 5 converts the mechanical energy into electrical energy, which is transmitted to the converter 10 through the power cable 7 for rectification and inversion. After the box transformer boost processing, it is connected to the grid, thereby realizing the production and transmission of electrical energy. Considering the current carrying capacity of the power cable 7, the electrical energy output by the generator 5 is shunted and carried by multiple power cables 7. The outgoing line of the generator 5 is directly distributed in the nacelle 6, and the cable is twisted through the saddle platform 8. Then, the power cable 7 is fixed to the tower 9 and transmitted to the bottom of the tower 9. In order to prevent multiple power cables 7 from colliding, interfering, etc. in the vibration environment of the unit operation, and to avoid further aggravation of the slippage and wear of the power cables 7 under the influence of these factors, most existing schemes use cable clamps 11 to restrain the power cables. Usually, multiple cable clamps 11 are used for restraint from the generator 5 outlet to the saddle platform 8.

[0061] In the embodiment of the present invention, a device and method for reducing the wear of the wind turbine cable are proposed in combination with the movement characteristics of the power cable 7. Figure 2 As shown, combined with the modal characteristics at different positions of the power cable 7, the first wear reduction device 12 is distributed on the upper part of the power cable 7, which is located near the outlet position of the generator 5, the second wear reduction device 13 is arranged in the middle straight section of the power cable 7, and the third wear reduction device 14 is arranged near the position of the saddle platform 8.

[0062] Further, Figure 3 A partial schematic diagram of the wear mitigation device is given, such as Figure 3As shown, since the power cable 7 involves a straight section and a curved section near the saddle platform 8, for intuitive expression, the power cable 7 is straightened and segmented to show the overall effect of the installation of the wear reduction device. Furthermore, the first wear reduction device 12 and the second wear reduction device 13 corresponding to the upper and middle parts of the power cable 7 are integrated with the power cable 7, and the entire bundle of the power cable 7 passes through the first wear reduction device 12 and the second wear reduction device 13. On the one hand, this design can integrate the wear reduction device with the cable clamp 11, thereby reducing materials, achieving modularization and improving reliability; on the other hand, since there is liquid inside the first wear reduction device 12 and the second wear reduction device 13, the heat of the power cable 7 can be well transferred, thereby reducing energy loss and improving the efficiency of power transmission. The cables near the saddle platform 8 and the bottom of the power cable 7 are generally treated as a single fixed constraint, so the third wear reduction device 14 is set and fixed with the single power cable. Furthermore, a single or multiple wear reduction devices can be configured at the upper, middle and bottom positions of the power cable 7 according to the local vibration level, such as Figure 4 As shown in FIG. , two modular first wear reduction devices 12 are arranged at the upper position of the power cable 7 .

[0063] like Figures 3 to 5 As shown, the first wear reduction device 12 includes a first shell 12-1, a first isolation ring 12-4, a first buffer 12-5, a first damping fluid 12-3 and a suspension 12-2; during assembly, the first isolation ring 12-4 is arranged in the middle of the first shell 12-1 and is used for integrated positioning with the power cable 7, and a closed buffer space is formed between the first isolation ring 12-4 and the first shell 12-1; the first damping fluid 12-3 and the suspension 12-2 are filled in the buffer space formed by the first isolation ring 12-4 and the first shell 12-1; the first buffer 12-5 is filled between the first isolation ring 12-4 and the power cable 7. In the present embodiment, the first shell 12-1 is essentially rectangular in shape to match the relatively low-frequency 1 to 5 Hz modal vibration. During assembly, the power cables 7 are first bundled into a plurality of small groups. The number of small groups can be determined according to the number of power cables. Preferably, if there are eighteen power cables 7, they are divided into six groups, with three power cables 7 in each group. The power cables 7 are partitioned by grouping, and then the power cables 7 are constrained in the first isolation ring 12-4. The multiple power cables 7 in each group are integrated by a first buffer. In the present embodiment, the first isolation ring 12-4 is a ring structure or a polygonal structure that matches the grouping of the power cables 7.

[0064] The material selected for the first buffer 12-5 can alleviate the contact stress between multiple cables to a certain extent, and at the same time has good thermal conductivity, so that the heat generated by the power cable 7 can be better transferred to the first damping liquid 12-3.

[0065] The first damping liquid 12-3 uses a liquid with high viscosity and low temperature resistance, such as silicone oil or ethylene glycol liquid, and can also be replaced by other liquids with similar properties. In order to increase the effective mass of shaking, solid particles can be added to the first damping liquid 12-3.

[0066] In order to prevent the first damping liquid 12-3 from rolling or breaking due to the large shaking of the power cable 7, resulting in poor suppression effect, a suspended object 12-2 is added to the first damping liquid 12-3. The suspended object 12-2 is a low-density fixed body. The suspended object 12-2 is a triangle, rectangle, sphere or polyhedron. The total surface area of ​​the suspended object 12-2 occupies 1 / 4 to 1 / 2 of the liquid surface of the first damping liquid 12-3, and the preferred proportion is 1 / 3.

[0067] like Figure 3 , Figure 6 , Figure 7 and Figure 8 As shown, the second wear reduction device 13 includes a second shell 13-4, at least two liquid columns 13-1, a plurality of flexible fins 13-2, a second isolation ring 13-6, a second buffer 13-5 and a second damping liquid 13-3; the second damping liquid 13-3 adopts a liquid with higher viscosity and low temperature resistance, such as silicone oil or ethylene glycol liquid, and can also be replaced by other liquids with similar properties. In order to increase the effective mass of shaking, solid particles can be added to the second damping liquid 13-3.

[0068] During processing, the liquid columns 13-1 are all integrally formed on the second shell 13-4, and all the liquid columns 13-1 are combined with the second shell 13-4 into a U shape; the U shape can match the relative intermediate frequency 5 to 10HZ modal vibration. The second isolation ring 13-6 is located in the middle of the second shell 13-4, and the second isolation ring 13-6, the second shell 13-4 and the liquid column 13-1 form a closed space; the second damping liquid 13-3 is filled in the closed space; during assembly, the power cables 7 are first bundled into a plurality of groups, and the number of groups can be determined according to the number of power cables. For example, if there are eighteen power cables 7, they are divided into six groups, each group has three power cables 7, and the power cables 7 are divided into groups, and then the power cables 7 are constrained in the first isolation ring 12-4, and the multiple power cables 7 in each group are integrated through the second buffer 13-5. In this embodiment, the first isolation ring 12-4 is a ring structure or a polygonal structure that matches the grouping of the power cables 7.

[0069] At the same time, several flexible fins 13-2 are arranged on the inner wall surface of the liquid column 13-1 to improve the energy dissipation performance of the second damping liquid 13-3 passing through the liquid column 13-1 during shaking; the power cable 7 is arranged in the second isolation ring 13-6 and integrated and positioned; the second buffer 13-5 is filled between the second isolation ring 13-6 and the power cable 7. And the material selected for the second buffer 13-5 can alleviate the contact stress between multiple cables to a certain extent, and at the same time has good thermal conductivity, so that the heat generated by the power cable 7 can be better transferred to the second damping liquid 13-3. In this embodiment, the flexible fin 13-2 can improve the energy dissipation capacity of the second damping liquid 13-3 passing through the liquid column 13-1 during shaking. It is preferably a material with good flexibility, which can fully contact when the second damping liquid 13-3 passes to generate fluid shear force, and restore the original shape when the second damping liquid 13-3 falls back.

[0070] In this embodiment, the flexible fins 13-2 are arranged on the inner wall surface of the liquid column 13-1 at equal or unequal intervals. The height of the flexible fins 13-2 is 1 / 20 to 1 / 8 of the diameter of the liquid column 13-1; the flexible fins 13-2 are arranged at the lower 1 / 4 to 2 / 3 position of the liquid column 13-1. The second isolation ring 13-6 is a ring structure or a polygonal structure that matches the grouping of the power cable 7.

[0071] like Figure 3 , Fig. 9 and Fig.10 As shown, the third wear reduction device 14 includes a third shell 14-1, a spiral plate 14-3 and a third damping liquid 14-4; the spiral plate 14-3 is located in the middle of the third shell 14-1, and the spiral plate 14-3 and the third shell 14-1 form a damping liquid flow channel; the third damping liquid 14-4 is filled in the damping liquid flow channel. The outer surface of the spiral plate 14-3 is provided with a boundary fin 14-2 for increasing the energy dissipation capacity of the third damping liquid 14-4 when it flows. In this embodiment, the spiral plate 14-3 is attached to the third shell 14-1 to form a certain pitch and a certain number of pitch turns, and further determines the length of the damping liquid flow channel according to the corresponding vibration modal frequency at the position where the power cable 7 is installed, and then determines the filling height of the third damping liquid 14-4. The third damping liquid 14-4 uses a liquid with high viscosity and low temperature resistance, such as silicone oil or ethylene glycol liquid, and can also be replaced by other liquids with similar properties. In order to increase the effective mass of sloshing, solid particles can be added to the third damping liquid 14-4. The boundary fins can increase the energy dissipation capacity of the third damping fluid 14 - 4 when it flows.

[0072] In order to monitor the status of different positions of the power cable 7, Fig.11The layout of cable status measurement points is shown in FIG. In order to obtain the real state of the power cable 7 more realistically, the unit status measurement points 15 are arranged inside the cabin 6, including temperature, strain, acceleration, etc. The layout is synchronously carried out at different positions of the power cable 7, including but not limited to the first measurement point layout position 16, the second measurement point layout position 17, and the third measurement point layout position 18. The corresponding measurement point layout positions are preferably located near the outlet position of the generator 5, the middle straight section position of the power cable 7, and the saddle platform 8 position. The measurement points at different positions of the power cable 7 include at least temperature sensors, acceleration sensors, strain gauges, etc.

[0073] like Fig.18 As shown, this embodiment provides a wind turbine cable life measurement device, the device comprising:

[0074] The data collection layer is used to collect the external environment data of the wind turbine generator set, the operating status data of the wind turbine generator set, and the test data of different positions of the power cables of the wind turbine generator set;

[0075] The data processing layer is used to filter, reduce noise and extract features from the data collected by the data collection layer to obtain vibration characteristic parameters;

[0076] The machine learning layer is used to perform pattern recognition and vibration mode prediction of the vibration characteristics of the power cable through a multi-dimensional machine learning algorithm based on the vibration characteristic parameters extracted by the data processing layer, so as to obtain the remaining life measurement result of the power cable.

[0077] In this embodiment, the data acquisition layer includes a multi-sensor integrated system, which is specifically configured as: ambient temperature sensor, humidity sensor, wind speed and direction sensor, vibration acceleration sensor and strain sensor. Among them, the vibration acceleration sensor is deployed at key parts of the power cable, such as cable joints, support points and turning points, with a sampling frequency of not less than 1kHz, and can synchronously collect vibration data at different positions of the cable. The environmental sensor monitors the temperature, humidity, wind speed and other environmental parameters around the wind turbine in real time, and realizes real-time transmission and synchronous recording of data through the data bus. The sensor data is processed by a high-precision analog-to-digital converter to ensure the accuracy and real-time nature of the data.

[0078] In this embodiment, the data processing layer can use a signal processing algorithm based on wavelet transform to first denoise the collected raw vibration data. The specific steps include: selecting a suitable wavelet basis function (such as Haar or Daubechies wavelet), removing high-frequency noise through multi-scale decomposition, and retaining effective vibration characteristic signals. Subsequently, the key vibration characteristic parameters are extracted using time-frequency analysis methods, including but not limited to: peak acceleration, effective value, kurtosis, skewness, spectral entropy, etc. These characteristic parameters can reflect the vibration characteristics and health status of the cable, and provide accurate input data for subsequent machine learning.

[0079] In this embodiment, the machine learning layer adopts an integrated learning method, and specifically, the random forest algorithm and support vector machine (SVM) can be used to identify the vibration mode of the power cable. First, a training set containing historical cable fault data and normal operation data is constructed, and the input features are optimized through feature selection and dimensionality reduction techniques (such as principal component analysis PCA). During the training process, the random forest algorithm can capture the nonlinear relationship of vibration characteristics, and SVM is used to identify potential damage modes of the cable. After the model training is completed, the current cable vibration characteristic parameters can be input in real time to predict the remaining service life of the cable. The prediction results are presented in percentage form, and an early warning mechanism can be triggered based on the prediction results to guide maintenance decisions. In order to improve the generalization ability of the model, this solution also adopts cross-validation and adaptive learning strategies to regularly update the model parameters.

[0080] In this embodiment, the above three levels realize real-time data transmission and collaborative processing through industrial Ethernet and edge computing devices. The sensors of the data acquisition layer are connected to the edge computing gateway through standard communication protocols (such as Modbus TCP / IP), and the algorithms of the data processing layer and machine learning layer are deployed on the GPU or FPGA of the edge computing device to achieve efficient data processing and model reasoning. The system is also equipped with a data storage and visualization module that can generate cable health status reports and trend analysis charts to provide decision support for predictive maintenance of wind turbines.

[0081] In some embodiments, the data acquisition layer includes: an environmental sensor for collecting external environmental data of the wind turbine generator set, including wind speed, wind direction and ambient temperature; a power monitoring sensor for collecting the operating power of the wind turbine generator set; a cabin monitoring sensor for collecting the cabin acceleration of the wind turbine generator set; a monitoring sensor group for multiple power cables, which are respectively arranged at multiple measuring points at different positions of the power cables, and each of the measuring points is correspondingly provided with a temperature sensor, an acceleration sensor and a strain gauge; the temperature sensor is used to detect the temperature data of the power cable; the acceleration sensor is used to detect the vibration acceleration data of the power cable; the strain gauge is used to detect the force and deformation data of the power cable.

[0082] Specifically, the environmental sensor in this embodiment adopts a composite meteorological monitoring system, including an ultrasonic wind speed and direction sensor and a digital temperature sensor. The ultrasonic wind speed and direction sensor is based on the ultrasonic time-of-flight measurement principle, with a measurement range of wind speed of 0-60m / s and an accuracy of ±0.5m / s; the wind direction measurement range is 0-360° and an accuracy of ±2°. The digital temperature sensor adopts a high-precision platinum resistance temperature detector (RTD) with a measurement range of -50 to 200°C and a resolution of 0.1°C. The sensor is installed on the top of the wind turbine tower to avoid obstruction and turbulence, ensuring the accuracy of environmental data.

[0083] Specifically, the power monitoring sensor in this embodiment is directly connected to the output end of the generator, and a high-precision power parameter transmitter is used. The transmitter can synchronously measure key electrical parameters such as active power, reactive power, voltage and current. The measurement accuracy is ±1%, the sampling frequency is 10Hz, and it can reflect the operating status of the wind turbine in real time. The sensor signal is transmitted to the data acquisition system through the standard industrial communication protocol (Modbus RTU), providing important operating condition data for subsequent cable life evaluation.

[0084] Specifically, the cabin monitoring sensor in this embodiment uses a high-performance micro-electromechanical system (MEMS) three-axis acceleration sensor, which is installed in the center of the cabin. The sensor has a wide range (±10g) and a wide frequency response (0-500Hz), and can capture small vibration and impact signals of the cabin. The sensor uses an integrated signal conditioning circuit, has a digital output interface, and has built-in temperature compensation and self-test functions. By monitoring the cabin acceleration, the overall operating stability and potential fault signs of the wind turbine can be evaluated.

[0085] Specifically, based on the specific layout of the power cable 7, in this embodiment, the sensor group is synchronously deployed at the first measuring point arrangement position 16 (near the outlet position of the generator 5), the second measuring point arrangement position 17 (the middle straight section position of the power cable 7), and the third measuring point arrangement position 18 (near the saddle platform 8). Each measuring point includes: (1) a temperature sensor, which uses a PT100 platinum resistance temperature sensor with a measurement range of -50 to 200°C, an accuracy of ±0.1°C, and a dedicated lead compensation; (2) an acceleration sensor, which uses a piezoelectric triaxial accelerometer with a measurement range of ±50g, a frequency response of 0-3000Hz, and an installation direction consistent with the force direction of the cable; (3) a strain gauge, which uses a metal foil resistance strain gauge with a measurement accuracy of ±0.1%, and is precisely bonded with a special strain gauge adhesive to measure the radial and axial strains of the cable.

[0086] Furthermore, the data processing layer includes: an edge computing device, and the edge computing device includes: a finite impulse response filter, which is used to remove noise components in the sensor data obtained by the data acquisition layer to obtain a denoised signal; a wavelet transform module, which is used to apply wavelet transform to remove noise components in the sensor data obtained by the data acquisition layer; and a fast Fourier transform module, which is used to extract vibration modal frequency features of multiple sensor data using fast Fourier transform to identify the main vibration frequency.

[0087] In this embodiment, the edge computing device is equipped with a multi-level noise suppression module. First, a finite impulse response (FIR) filter is used to remove noise from the raw data collected by the sensor. The FIR filter uses the Kaiser window design method, which has a linear phase characteristic and can effectively suppress high-frequency random noise. The filter parameters are dynamically adjusted according to the spectral characteristics of the actual sensor data. The order range is 32-128 orders, and the passband cutoff frequency is set to 80% of the main frequency components of the sensor signal. By reasonably designing the filter parameters, the noise interference can be reduced while retaining the key characteristics of the signal, and a preliminary denoised signal can be obtained.

[0088] In this embodiment, on the basis of FIR filtering, a wavelet transform module is introduced to perform multi-scale signal denoising. The Daubechies wavelet base (db4) is selected as the basic wavelet, and the number of decomposition layers is set to 5-8 layers. The wavelet transform process includes three steps: multi-scale signal decomposition, threshold denoising and reconstruction. Specifically, the sensor signal is first subjected to multi-scale wavelet decomposition to split the signal into approximate coefficients and detail coefficients; then the detail coefficients are denoised using a soft threshold or hard threshold method, and the threshold selection is based on the statistical characteristics of the signal and the noise variance; finally, the denoised signal is reconstructed by an inverse wavelet transform. Compared with traditional filtering methods, wavelet transform can more accurately retain the local time-frequency characteristics of the signal and effectively suppress noise components of different scales.

[0089] In this embodiment, in order to comprehensively analyze the vibration characteristics of the power cable, the edge computing device is equipped with a Fast Fourier Transform (FFT) module to realize the frequency feature extraction of multi-sensor data. The FFT module processes the acceleration sensor data from different measuring points in parallel, and the transformation window length is dynamically adjusted according to the sampling frequency and signal characteristics, and 512-2048 points are selected. During the spectrum analysis process, the Hanning window or Blackman window is used for spectral leakage suppression, and the frequency resolution is better than 0.1Hz. Through the FFT transformation, the main frequency components of the cable vibration can be accurately identified, including the fundamental frequency, harmonics and modulation frequency. For multi-measurement point data, coherence analysis and phase spectrum calculation can also be performed to reveal the correlation between the vibration characteristics of different measuring points.

[0090] Furthermore, the machine learning layer includes: a pattern recognition and classification module, which is used to identify and classify the vibration modal characteristics of the power cable based on the spectral characteristics of the vibration signals of multiple sensors; a vibration mode prediction module, which is used to construct a prediction model of the vibration modal characteristics of the power cable based on the external environment data and historical vibration data of the wind turbine generator set, and predict the vibration modal frequency of the power cable through a multimodal learning method; a model training and life assessment module, which is used to predict the remaining service life of the power cable based on the vibration modal characteristics and historical operation data of the power cable.

[0091] Furthermore, the pattern recognition and classification module specifically includes: a support vector machine submodule, which is used to classify and identify the vibration modes of the power cable according to the spectral characteristics of the vibration signals of multiple sensors in combination with the support vector machine classification algorithm, and output the vibration mode classification results; a recurrent neural network submodule, which is used to input the vibration mode classification results and the time series vibration data of multiple measuring points of the power cable, use the recurrent neural network model to analyze the vibration correlation of different parts of the power cable, and output the time dimension correlation characteristics of the vibration mode.

[0092] Specifically, in this embodiment, the support vector machine submodule realizes the classification and identification of the vibration modes of the power cable through the following steps. First, vibration signals are collected from multiple sensors installed at different positions of the power cable of the wind turbine generator set, and spectrum analysis is performed on these vibration signals to extract spectrum feature vectors. Then, the support vector machine (Support Vector Machine, SVM) classification algorithm is used to map these spectrum feature vectors to a high-dimensional feature space, and nonlinear classification is performed through the kernel function. Specifically, the radial basis kernel function (Radial Basis Function, RBF) can be selected, and the optimal classification hyperplane is established by adjusting the penalty parameter C and the kernel function parameter γ, and different vibration modes (such as normal vibration, abnormal vibration, warning vibration, etc.) are accurately classified. The classification results will be used as input data for subsequent analysis.

[0093] Specifically, in this embodiment, the recurrent neural network (RNN) submodule uses the vibration mode classification results output by the support vector machine submodule and the time series vibration data of multiple measuring points of the power cable to deeply analyze the vibration correlation of different parts of the cable. The specific implementation process includes: First, the time series vibration data of multiple measuring points are input into the long short-term memory network (LSTM) model in chronological order, which can effectively capture the long-term dependencies in the time series. The network learns and memorizes the time evolution characteristics of the vibration signals of different parts of the cable through the gating mechanism (forget gate, input gate and output gate). Then, through the fully connected layer and the softmax activation function, the time dimension correlation characteristics of the vibration mode of each measuring point are output, indicating the potential correlation mode of vibration in different parts of the power cable, which provides an important basis for subsequent vibration mode prediction and life assessment.

[0094] Furthermore, the vibration mode prediction module specifically includes: a gradient decision tree submodule, which is used to construct a mapping model between the external environmental data of the wind turbine and the vibration characteristics through a gradient decision tree algorithm, and extract the influence weights of environmental factors on the vibration characteristics; a deep network learning submodule, which is used to use a deep neural network, with the environmental influence weights output by the gradient decision tree submodule and the historical vibration data of multiple measuring points of the power cable as input, to predict the vibration characteristics of the power cable.

[0095] Specifically, in this embodiment, the gradient decision tree submodule constructs a complex mapping relationship between the external environmental data of the wind turbine and the vibration characteristics through the gradient boosting decision tree (GradientBoosting Decision Tree, GBDT) algorithm. The specific implementation process includes: first, collecting the external environmental data of the wind turbine, such as wind speed, temperature, humidity, atmospheric pressure, wind direction and other multi-dimensional features. Then, using these environmental data as input features and the vibration characteristics of the power cable as the target variable, the decision tree integration model is iteratively constructed using the GBDT algorithm. By repeatedly calculating the residuals and fitting the residuals, the algorithm can gradually optimize the prediction accuracy of the model. During the model training process, the Gini coefficient or information gain is used as the feature selection criterion to automatically extract and quantify the influence weight of each environmental factor on the vibration characteristics, revealing the nonlinear correlation between environmental factors and vibration characteristics.

[0096] Specifically, in this embodiment, the deep network learning submodule uses the deep neural network (DNN) technology to integrate the environmental impact weights output by the gradient decision tree submodule and the historical vibration data of multiple measuring points of the power cable to accurately predict the vibration characteristics of the power cable. The specific implementation process is as follows: First, the environmental impact weights extracted by the gradient decision tree submodule are used as prior knowledge, together with the historical vibration data of multiple measuring points of the power cable (including amplitude, frequency, phase, etc.) as the input layer of the neural network. The network architecture adopts a multi-layer fully connected structure, including an input layer, multiple hidden layers and an output layer. Each hidden layer uses a ReLU activation function, and the random dropout regularization technology is used to prevent overfitting. During the training process, the back propagation algorithm and the Adam optimizer are used to gradually adjust the network weights so that the network can learn complex vibration characteristic mapping relationships. The output layer predicts the future vibration characteristics of the power cable, such as vibration frequency, amplitude change trend, etc., through a softmax or linear activation function, providing key information for cable status monitoring and life prediction.

[0097] Furthermore, the device also includes a feedback control layer for optimizing the vibration control effect of the power cable by adjusting the opening of the regulating valve of the wind turbine generator set according to the measurement result of the remaining life of the power cable.

[0098] The feedback control layer specifically includes a cable life control module, which is specifically used to perform valve opening control based on the cumulative fatigue damage relationship of the cable; the cumulative fatigue damage relationship of the cable is as follows:

[0099]

[0100] In the formula, A max , A min is the maximum and minimum value of the valve opening range; c max is the maximum allowable value of the damping coefficient; c 0 is the reference temperature T ref The damping coefficient under k T is the temperature sensitivity coefficient; T is the real-time temperature collected by the temperature sensor; k σ is the stress sensitivity coefficient; σ eff is the equivalent stress; σ max is the maximum allowable stress limit; c(T,σ) is the damping coefficient that comprehensively considers temperature and stress.

[0101] The following combination Fig.12 The above technical solution is described in detail with the example of the specific structure in:

[0102] Based on the status of the wind turbine generator set and the arrangement of different types of measuring points at different positions of the power cable 7, the status monitoring, status analysis and status control of the power cable 7 can be carried out, such as Fig.12 The cable condition monitoring, analysis and control diagram is given. The system includes at least four levels, namely data acquisition layer, data processing layer, machine learning layer and feedback control layer.

[0103] Furthermore, the data collection layer includes at least two dimensions of data. The first dimension is the unit external environment data and the unit operation status data, including but not limited to wind speed, wind direction, temperature, unit operating power, cabin acceleration a x 、a y 、a z etc.; the second dimension is the test data of different positions of the power cable 7, including the temperature, strain, acceleration of position 1; the temperature, strain, acceleration of position 2 and the temperature, strain, acceleration of the position.

[0104] Furthermore, edge computing devices are configured to build a data processing layer. Based on the data input from these hardware devices and the data acquisition layer, the high-frequency noise components in the data are removed through FIR (Finite Impulse Response) filters and wavelet transform methods.

[0105] Furthermore, Fast Fourier Transform (FFT) is used to extract vibration modal frequency features and identify the main vibration frequency;

[0106] Furthermore, modal identification is performed through frequency domain decomposition to determine the current vibration modal characteristics.

[0107] Specifically, FIR filters are suitable for processing signals with a known frequency range and are used for noise reduction, signal extraction, or real-time filtering. Wavelet transforms are suitable for processing non-stationary signals (such as instantaneous vibration or impact signals) and can be used in scenarios such as fault diagnosis and signal decomposition. FFT is suitable for analyzing the spectral characteristics of periodic and stable signals and is used for harmonic analysis, frequency extraction, etc.

[0108] Since the vibration of power cable 7 presents a complex vibration mode with multi-modal superposition and strong nonlinear characteristics, the vibration characteristics, fatigue life prediction and evaluation are carried out from the data collected by massive sensors through machine learning methods. The machine learning layer includes at least three dimensions, the first dimension is pattern recognition and classification, the second dimension is vibration mode prediction, and the third dimension is model training and life evaluation.

[0109] Furthermore, the main frequency, amplitude, and phase in the sensor data are identified through the first dimension pattern recognition and classification to determine the cable vibration modal characteristics. The corresponding algorithm at least includes: extracting spectral features by combining FFT, classifying vibration modes by using Support Vector Machine (SVM); and learning the coupling relationship between modes from time series data by using Recurrent Neural Network (RNN) model.

[0110] Furthermore, the future vibration characteristics of the cable and the dynamic changes of the vibration modal frequency are estimated through vibration mode prediction. The corresponding algorithm includes at least a gradient boosting decision tree (GBDT) for learning the relationship between the external environment (such as wind speed, wind direction, temperature, etc.) and the vibration mode. Combined with the wind turbine operation data, deep neural network (DNN) is used to predict the cable vibration characteristics.

[0111] Furthermore, combined with historical data, regression models are used to predict the life of fatigue-damaged cables and identify potential failure points.

[0112] In the previous processing, based on the results of the data processing layer, the real-time mode of the power cable is obtained, such as vibration mode, natural frequency, damping ratio and mode shape; based on the data processing results of the machine learning layer, the signal characteristics or modal parameters are analyzed to perform modal identification, classification or prediction; using statistical models or neural network models, regularities are discovered from complex signals, and the real-time stress and predicted stress of the power cable can be obtained, based on which the fatigue level and fatigue prediction can be obtained. Fig.12 In the feedback control layer, it specifically includes: a damping setting module, which is used to optimize the vibration control effect of the power cable by adjusting the opening of the regulating valve of the wind turbine generator set according to the remaining life measurement result of the power cable obtained through damping setting, that is, by adjusting the opening of the regulating valve 13-7, the vibration control effect is optimized and improved, thereby controlling the local and overall life of the cable.

[0113] And through damping setting, that is, by adjusting the opening of the regulating valve 13-7, the vibration control effect is optimized and improved, thereby controlling the local and overall life of the cable.

[0114] The goal of the machine learning layer is to perform pattern recognition, classification, or prediction based on the extracted signal features or modal parameters. This is a data-driven approach that uses statistical models or neural network models to automatically discover patterns or regularities from complex signals.

[0115] Furthermore, the opening of the regulating valve 13-7 is directly related to the damping system, and the damping coefficient is related to the vibration control effect, and the vibration control effect is directly related to the fatigue life of the cable. In the embodiment of the present invention, a valve opening control algorithm based on cable fatigue damage or a cable life control method based on valve opening is proposed. Furthermore, the relationship between the damping coefficient c(T) and the temperature is as follows:

[0116] c(T)=c 0 ·(1+k T ·(TT ref ));

[0117] In the formula, c 0 is the reference temperature T ref The damping coefficient under k T is the temperature sensitivity coefficient; T is the real-time temperature collected by the temperature sensor;

[0118] Furthermore, the relationship between the damping coefficient c(σ) and the cable stress is as follows:

[0119]

[0120] In the formula, k σ is the stress sensitivity coefficient; σ eff is the equivalent stress; σ max is the maximum allowable stress limit;

[0121] Furthermore, the effects of temperature and stress factors on the damping coefficient are as follows:

[0122]

[0123] Furthermore, the relationship between valve opening and damping coefficient is as follows:

[0124]

[0125] In the formula, A max , A min The maximum and minimum values ​​of the valve opening range are preferably 0.2 to 1; c max is the maximum allowable value of the damping coefficient.

[0126] Furthermore, valve opening control based on cable fatigue damage:

[0127] A(t)=A min +(A max -A min )·(1-D(t)); where D(t) is the accumulated fatigue damage.

[0128] Based on the cable life measurement device and control method proposed in the embodiment of the present invention, actual project application and evaluation are carried out, and the following are given: Figure 13 to Figure 17 The comparison between the cable modal prediction and stress prediction and the actual stress through machine learning confirms the effectiveness of the machine learning algorithm proposed in the embodiment of the present invention ( Figure 13-14 ); Through the collection of temperature and stress sensor data at multiple locations, combined with data analysis and machine learning, the influence curve of temperature and vibration factors on the fatigue damage and life of power cable 7 was found ( Fig.15 ); Combined with the evaluation and control of the cable life, the opening of the regulating valve 13-7 is adjusted, and combined with the algorithm provided in the embodiment of the present invention, the control curve of the opening and damping coefficient of the regulating valve 13-7 is obtained ( Fig.16 ), and also compared the vibration of the power cable 7 when the opening of the regulating valve 13-7 is actively controlled and when no control is used ( Fig.17 ), which proves the correctness and effectiveness of the algorithm proposed in the embodiment of the present invention.

[0129] like Fig.19 As shown, this embodiment also provides a method for controlling the life of cables of a wind turbine generator set, the method comprising:

[0130] S10: Collect external environment data, operation status data of the wind turbine generator set and test data of different positions of the power cable;

[0131] S20: filtering, denoising and feature extraction of the collected data to obtain vibration feature parameters;

[0132] S30: Based on the vibration characteristic parameters extracted in the data processing stage, pattern recognition and vibration mode prediction of the vibration characteristics of the power cable are performed through a multi-dimensional machine learning algorithm to obtain a remaining life measurement result of the power cable;

[0133] S40: According to the remaining life measurement result, the life of the power cable is improved by adjusting the local acceleration of the power cable.

[0134] In this embodiment, the data acquisition process adopts a multi-sensor collaborative working method to comprehensively obtain the operating status information of the power cable of the wind turbine generator set. Specifically, the sensor array deployed at different positions of the power cable includes acceleration sensors, strain sensors, temperature sensors and vibration sensors. The acceleration sensor is used to capture the acceleration changes of the cable and accurately measure the dynamic response of the cable under different working conditions; the strain sensor monitors the stress and strain of the cable and evaluates the mechanical load of the cable; the temperature sensor records the ambient temperature around the cable in real time and analyzes the impact of temperature on the performance of the cable; the vibration sensor captures the vibration characteristics of the cable, including key parameters such as vibration frequency, amplitude and phase. At the same time, the external environment data acquisition system of the wind turbine generator set records meteorological information such as wind speed, wind direction, atmospheric pressure, humidity, etc., and transmits it to the central data processing unit in real time through the data bus to ensure the time synchronization and integrity of the data.

[0135] In this embodiment, the data processing stage adopts multi-layer signal processing technology to perform fine processing on the collected raw data. First, the original signal collected by the sensor is denoised using wavelet transform to effectively suppress high-frequency noise and power frequency interference and improve the signal-to-noise ratio. Secondly, Fourier transform and short-time Fourier transform are applied to perform spectral analysis on the signal to extract key frequency components and energy distribution characteristics. Then, through feature engineering technology, multidimensional feature vectors are constructed, including statistical features (mean, variance, skewness, kurtosis), time-frequency domain features (wavelet coefficients, spectral entropy) and nonlinear features (correlation dimension, Lyapunov exponent). These feature vectors comprehensively describe the vibration characteristics of the power cable and provide rich input information for subsequent machine learning.

[0136] In this embodiment, a multimodal intelligent algorithm is used in the machine learning stage to deeply analyze the vibration characteristics of the power cable. First, the support vector machine (SVM) is used to classify the vibration characteristics and divide the cable vibration state into three levels: normal, sub-healthy and abnormal. Subsequently, the long short-term memory network (LSTM) is introduced to analyze the time series characteristics of cable vibration and capture the evolution law of the vibration mode. Through the integrated learning method, combined with the gradient boosting decision tree (GBDT) and the deep neural network (DNN), a power cable remaining life prediction model is constructed. The model output includes the point estimate and confidence interval of the remaining service life, which quantifies the degree of degradation and replacement risk of the cable. Bayesian probabilistic inference is introduced to dynamically adjust the uncertainty of life prediction based on historical data and real-time monitoring information.

[0137] In this embodiment, an intelligent cable life control strategy is determined based on the remaining life measurement results predicted by machine learning. When the predicted remaining life is lower than the preset threshold, the service life is extended by actively adjusting the local acceleration of the power cable. The specific implementation scheme includes: installing fine-tuning vibration reduction devices at key parts of the cable where stress concentration is prone to occur, and using piezoelectric actuators or magnetostrictive actuators to actively adjust the local vibration characteristics of the cable in real time. Through closed-loop feedback control, the output of the actuator is dynamically adjusted according to real-time vibration data to suppress vibrations of harmful frequencies, balance stress distribution, reduce local stress concentration, and thus delay fatigue damage to the cable. In addition, a cable health status file is established, and when the remaining life is lower than the safety critical value, a maintenance warning is triggered to guide precise preventive maintenance.

[0138] The cable life control method of the wind turbine generator set in the embodiment of the present invention is based on state monitoring and machine learning. It monitors the cable stress through acceleration and temperature indicators to monitor the life, and simultaneously adjusts the local acceleration level through the feedback control layer to improve the life of the power cable.

[0139] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0140] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, any one of the above methods is implemented.

[0141] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0142] The present invention also provides an electronic device, comprising: one or more processors; a storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, the one or more processors implement the method provided by the present invention.

[0143] Reference below Fig. 20 , which shows a schematic diagram of the structure of a computer system 800 of an electronic device suitable for implementing an embodiment of the present invention. Fig. 20 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0144] like Fig. 20 As shown, the computer system 800 includes a central processing unit (CPU) 801, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 802 or a program loaded from a storage part 808 into a random access memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the computer system 800 are also stored. The CPU 801, the ROM 802, and the RAM 803 are connected to each other via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0145] The following components are connected to the I / O interface 805: an input section 806 including a keyboard, a mouse, etc.; an output section 807 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 808 including a hard disk, etc.; and a communication section 809 including a network interface card such as a LAN card, a modem, etc. The communication section 809 performs communication processing via a network such as the Internet. A drive 810 is also connected to the I / O interface 805 as needed. A removable medium 811, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 810 as needed so that a computer program read therefrom is installed into the storage section 808 as needed.

[0146] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may occur depending on design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A wind turbine cable life measuring device, characterized in that: The device comprises: The data collection layer is used to collect the external environment data of the wind turbine generator set, the operating status data of the wind turbine generator set, and the test data of different positions of the power cables of the wind turbine generator set; The data processing layer is used to filter, reduce noise and extract features from the data collected by the data collection layer to obtain vibration characteristic parameters; The machine learning layer is used to perform pattern recognition and vibration mode prediction of the vibration characteristics of the power cable through a multi-dimensional machine learning algorithm based on the vibration characteristic parameters extracted by the data processing layer, so as to obtain the remaining life measurement result of the power cable.

2. The device according to claim 1, characterized in that The data collection layer includes: Environmental sensors are used to collect external environmental data of the wind turbine generator set, including wind speed, wind direction and ambient temperature; Power monitoring sensor, used to collect the operating power of wind turbines; Nacelle monitoring sensor, used to collect the nacelle acceleration of the wind turbine; A monitoring sensor group for multiple power cables is respectively arranged at multiple measuring points at different positions of the power cables, and each measuring point is correspondingly provided with a temperature sensor, an acceleration sensor and a strain gauge; the temperature sensor is used to detect the temperature data of the power cable; the acceleration sensor is used to detect the vibration acceleration data of the power cable; the strain gauge is used to detect the stress deformation data of the power cable.

3. The device according to claim 1, characterized in that The data processing layer includes: An edge computing device, the edge computing device comprising: A finite impulse response filter, used to remove noise components in the sensor data obtained by the data acquisition layer to obtain a denoised signal; A wavelet transform module, used to apply wavelet transform to remove noise components in the sensor data obtained by the data acquisition layer; The fast Fourier transform module is used to extract vibration modal frequency features from multiple sensor data using fast Fourier transform and identify the main vibration frequency.

4. The device according to claim 1, characterized in that The machine learning layer includes: A pattern recognition and classification module is used to identify and classify the vibration modal characteristics of the power cable based on the spectral characteristics of the vibration signals of multiple sensors; A vibration mode prediction module is used to construct a prediction model of the vibration modal characteristics of the power cable and predict the vibration modal frequency of the power cable based on the external environment data and historical vibration data of the wind turbine generator set through a multimodal learning method; The model training and life assessment module is used to predict the remaining service life of the power cable based on the vibration modal characteristics and historical operation data of the power cable.

5. The device according to claim 4, characterized in that The pattern recognition and classification module specifically includes: The support vector machine submodule is used to classify and identify the vibration modes of the power cable according to the frequency spectrum characteristics of the vibration signals of multiple sensors in combination with the support vector machine classification algorithm, and output the vibration mode classification results; The recurrent neural network submodule is used to input the vibration mode classification results and the time-series vibration data of multiple measuring points of the power cable, use the recurrent neural network model to analyze the vibration correlation of different parts of the power cable, and output the time dimension correlation characteristics of the vibration mode.

6. The device according to claim 4, characterized in that The vibration mode prediction module specifically includes: The gradient decision tree submodule is used to construct a mapping model between the external environmental data of the wind turbine generator set and the vibration characteristics through the gradient decision tree algorithm, and to extract the influence weight of environmental factors on the vibration characteristics; The deep network learning submodule is used to use a deep neural network to predict the vibration characteristics of the power cable with the environmental impact weights output by the gradient decision tree submodule and the historical vibration data of multiple measuring points of the power cable as input.

7. The device according to claim 1, characterized in that The device also includes a feedback control layer for optimizing the vibration control effect of the power cable by adjusting the opening of the regulating valve of the wind turbine generator set according to the measurement result of the remaining life of the power cable.

8. The device according to claim 7, characterized in that The feedback control layer specifically includes a cable life control module, which is specifically used to perform valve opening control based on the cumulative fatigue damage relationship of the cable; the cumulative fatigue damage relationship of the cable is as follows: In the formula, A max , A min is the maximum and minimum value of the valve opening range; c max is the maximum allowable value of the damping coefficient; c0 is the reference temperature T ref The damping coefficient under k T is the temperature sensitivity coefficient; T is the real-time temperature collected by the temperature sensor; k σ is the stress sensitivity coefficient; σ eff is the equivalent stress; σ max is the maximum allowable stress limit; c(T,σ) is the damping coefficient that comprehensively considers temperature and stress.

9. A method for controlling the life of cables of a wind turbine generator set, characterized in that: The method comprises: Obtain external environment data, operating status data and test data of different positions of power cables of wind turbines; Filtering, denoising and feature extraction are performed on the collected data to obtain vibration characteristic parameters; Based on the vibration characteristic parameters extracted in the data processing stage, a multi-dimensional machine learning algorithm is used to perform pattern recognition and vibration mode prediction of the vibration characteristics of the power cable to obtain a remaining life measurement result of the power cable; According to the remaining life measurement result, the life of the power cable is improved by adjusting the local acceleration of the power cable.

10. An electronic device, characterized in that: include: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method as claimed in claim 8.

Images