Wind driven generator main shaft fault detection method, electronic equipment and program product

Abstract

According to the wind driven generator spindle fault detection method, the electronic equipment and the program product, a radar sensor is introduced in spindle fault detection, the surface state of a spindle is directly detected through the micro-Doppler effect, radar can capture the surface roughness change caused by tiny deformation or abrasion of the surface when the spindle rotates, and even if the abrasion loss is extremely small, the surface roughness change is not influenced. And accurate identification can be realized through the phase / frequency change of the reflected wave. And in combination with the eccentric distance characteristics, the balance state abnormity of the main shaft can be further reflected, a more sensitive detection means is provided for early faults, and the fault detection rate is increased. According to the embodiment, multi-mode fusion of radar and the temperature sensor is adopted, the temperature rise rate extracted by the temperature sensor can reflect the abnormity of the friction state of the main shaft and is complementary with the surface roughness and the eccentric distance detected by the radar, and the fault detection rate is also improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wind power generators, in particular to a wind power generator main shaft fault detection method, an electronic device and a program product. BACKGROUND

[0002] The wind power generator main shaft is a core component of the transmission system of the unit, and plays a key role in transmitting the kinetic energy of the wind wheel. It transmits the huge torque captured by the wind wheel blades to the gearbox (in units with a gearbox) or directly to the generator (in direct-drive units), while also needing to withstand the axial thrust, bending moment and other complex dynamic loads from components such as blades and hubs, and transmit these non-torque loads to the support structure of the mainframe. As the hub of the rotating system, the main shaft supports the entire rotor system, and its structure is usually connected to the hub through a flange, installed on the shaft neck with bearings, and connected to the downstream equipment through a shaft coupling, so it must have extremely high strength and fatigue resistance. The main shaft faces various failure risks during long-term operation. Common failure modes include mechanical damage (such as bending, misalignment, axial movement, shaft neck wear), bearing failure (such as roller damage, cage damage, wear due to lubrication failure, corrosion, fracture and aging), and internal cracks caused by long-term exposure to alternating loads (fluctuating loads). If these cracks are not detected in time, they may continue to expand and eventually lead to catastrophic failure.

[0003] The necessity of fault detection for the main shaft is reflected in the following aspects: first, once the main shaft or its bearings fail seriously, the entire nacelle often needs to be hoisted down the tower for repair, with extremely high replacement costs (up to hundreds of thousands of yuan or more) and long repair cycles. Second, unplanned downtime caused by failure can result in significant loss of power generation, directly affecting the economic benefits of the wind farm. Third, most seriously, sudden main shaft failure poses a major safety risk of causing the entire wind turbine to collapse, threatening the safety of personnel and property. In addition, main shaft failure can also cause collateral damage to the bearing installation site of the mainframe, and even to adjacent key components such as the gearbox, further expanding the scope of damage. Traditional periodic maintenance requires downtime and inspection by professionals at high altitudes (over 100 meters), which is not only dangerous and costly, but also difficult to capture early and minor damage changes in the main shaft or bearings, and has obvious lag and limitations.

[0004] The prior art has proposed a scheme for real-time monitoring of the operating state of the main shaft using MEMS vibration sensors, but traditional MEMS vibration sensors have insufficient sensitivity to surface micro-wear of less than 0.5 mm, resulting in a high early failure detection rate. SUMMARY

[0005] The embodiment of the application aims to provide a wind turbine main shaft fault detection method, an electronic device and a program product, so as to solve the problem that the prior art uses a MEMS vibration sensor to monitor the running state of the main shaft in real time, and the early fault omission rate is high.

[0006] The wind turbine main shaft fault detection method provided by the embodiment of the application comprises: The rotating main shaft is detected in real time by using a radar and a temperature sensor to obtain radar information and temperature information; The radar information is preprocessed and feature extracted to obtain surface roughness and eccentricity; The temperature information is preprocessed and feature extracted to obtain a temperature rise rate; The surface roughness, the eccentricity and the temperature rise rate are feature fused to obtain multi-dimensional features; The multi-dimensional features are input into a trained fault detection model to obtain a fault detection result.

[0007] The surface roughness refers to the error of the micro-geometric shape of the processed surface, which is usually described by the height, spacing or shape features of the micro-unevenness. In mechanical engineering, it reflects the degree of fluctuation of the micro-peak and valley of the surface and is an important indicator for evaluating the surface quality of parts. When the radar wave irradiates the surface of the main shaft, the micro-peak and valley structure will change the reflection path of the radar wave, producing a micro-Doppler shift related to the surface roughness (Doppler effect caused by micro-deformation of the surface). By analyzing the spectral features of the radar signal, the surface roughness can be quantitatively extracted, so as to capture early micro-wear.

[0008] The eccentricity refers to the radial distance between the actual rotation center axis of the rotating component (such as the main shaft) and its geometric center axis (ideal axis). Simply put, it reflects the degree of "wobble" of the main shaft during rotation. The radar can indirectly calculate the eccentricity by measuring the phase or frequency change of the reflected wave when the main shaft rotates. For example, when the main shaft radially jumps, the reflection path of the radar wave will be periodically offset, and by analyzing the periodicity and amplitude of the offset, the eccentricity can be quantitatively extracted.

[0009] In the above-mentioned technical solution, a radar sensor is introduced for spindle fault detection. It utilizes the micro-Doppler effect (Doppler frequency shift caused by small target motion) to directly detect the spindle surface condition. The radar can detect subtle surface deformations or surface roughness changes caused by wear during spindle rotation. Even extremely small wear (e.g., micrometer-level) can be accurately identified through phase / frequency changes in the reflected wave. Combined with eccentricity (radial runout during spindle rotation), this can further indicate abnormalities in spindle balance (e.g., eccentricity caused by bearing wear), providing a more sensitive means of detecting early faults and improving the fault detection rate. This embodiment utilizes a multimodal fusion of radar and temperature sensors. The temperature rise rate (rate of temperature change) extracted by the temperature sensor can reflect abnormalities in the spindle friction state (e.g., increased friction due to early wear, significantly increasing the temperature rise rate). This complements the surface roughness and eccentricity detected by the radar, similarly improving the fault detection rate.

[0010] In some optional embodiments, the radar is installed inside the wind turbine housing, the radar is set toward the main shaft, and the direction of the radar is set at a set angle to the center axis of the main shaft, and the set angle is less than 70 degrees and greater than 20 degrees.

[0011] In this technical solution, radar waves strike the spindle surface at an oblique angle within a range of 20°-70°. As the spindle rotates, tiny wear or cracks on the surface alter the radar wave's reflection path, generating a richer Doppler shift signal (the micro-Doppler effect), thereby enhancing detection sensitivity for micron-level surface deformation.

[0012] In some optional embodiments, the method further includes: Use vibration sensors to detect the rotating spindle in real time to obtain vibration information; The vibration information is preprocessed and features are extracted to obtain envelope entropy and harmonic energy ratio.

[0013] Envelope entropy quantifies the periodic impact intensity of the signal by demodulating the envelope of the vibration signal. Early-stage faults (such as pitting corrosion on the inner ring of a bearing) can generate periodic impact vibrations, and envelope entropy effectively captures these low-amplitude, highly periodic signals.

[0014] Harmonic Energy Ratio: By analyzing the energy ratio of the fundamental frequency and its harmonics within a vibration signal, faults such as gear wear and shaft misalignment can be identified. For example, gear wear can significantly increase the energy of specific harmonic components. The harmonic energy ratio quantifies this change, enhancing early detection of gear faults.

[0015] In the technical solution, the vibration features (envelope entropy, harmonic energy ratio) reflect dynamic impact (such as periodic vibration caused by bearing failure) and harmonic component change (such as abnormal harmonic energy distribution caused by gear wear) when the main shaft rotates, and can accurately identify faults of key components such as bearings and gears. The radar features (surface roughness, eccentricity) reflect surface deformation and abnormal balance state (such as shaft neck wear and installation error) of the main shaft. The temperature feature (temperature rise rate) reflects the dynamic process of friction heat generation (such as local overheating caused by poor lubrication). The combination of the three can cover the mechanical, thermal and vibration dimensional features of the main shaft fault, and avoid missed detection caused by the limitations of a single sensor.

[0016] In some optional embodiments, the surface roughness, eccentricity and temperature rise rate are fused into features, including: According to the rotating speed, the weight settings of the radar information, the temperature information and the vibration information are determined; According to the weight settings, the surface roughness, the eccentricity, the temperature rise rate, the envelope entropy and the harmonic energy ratio are fused into features to obtain multi-dimensional features.

[0017] In some optional embodiments, the weight of the radar information is w_uwb = 1 / (1 + k); The weight of the vibration information is w_vib = a1 x k; The weight of the temperature information is w_temp = a2; Wherein, k is a rotating speed adaptive weight; when the rotating speed rpm is greater than a rotating speed setting value, k = a3 x rpm; when the rotating speed rpm is less than or equal to the rotating speed setting value, k = a4; a1, a2, a3 and a4 are fixed coefficient values.

[0018] In the technical solution, at low rotating speed, the weights of the radar information, the vibration information and the temperature information remain unchanged; at high rotating speed, as the rotating speed increases, the weight of the radar information decreases, the weight of the vibration information increases, and the weight of the temperature information remains unchanged. The weight adjustment strategy of the embodiment adapts to the sensitivity difference of the fault features at different rotating speeds, optimizes the sensor data fusion, balances the calculation resources and the detection accuracy, finally reduces the early fault missed detection rate, and improves the system robustness and practicality.

[0019] In some optional embodiments, the surface roughness, the eccentricity and the temperature rise rate are fused into features, including: According to the rotating speed and the current temperature, the weight settings of the radar information, the temperature information and the vibration information are determined; According to the weight settings, the surface roughness, the eccentricity, the temperature rise rate, the envelope entropy and the harmonic energy ratio are fused into features to obtain multi-dimensional features.

[0020] In some optional embodiments, the weight of the radar information is: w_uwb = 1 / (1 + k); When the current temperature is less than or equal to the temperature set value, the weight of the vibration information is: w_vib = a1 x k; when the current temperature is greater than the temperature set value, the weight of the vibration information is: w_vib = a5. The weight of the temperature information is: w_temp = a2. Wherein, k is the speed adaptive weight; when the speed rpm is greater than the speed set value, k = a3 x rpm; when the speed rpm is less than or equal to the speed set value, k = a4; a1, a2, a3, a4 and a5 are fixed coefficient values.

[0021] In the above technical solution, in the low temperature scene, the weight of the vibration information is high, and the weight of the vibration information increases with the increase of the speed. In the high temperature scene, the weight of the vibration information remains at a low weight, and the reason is that: high temperature may be caused by lubrication failure, friction aggravation or thermal expansion, at this time the vibration signal may be distorted due to thermal deformation.

[0022] In some optional embodiments, the temperature sensor and the vibration sensor are installed on the main shaft bearing seat of the wind turbine.

[0023] In some optional embodiments, the fault detection model is a time series sensitive model.

[0024] In the above technical solution, the time series sensitive model (such as LSTM, GRU, Transformer) can remember the historical running state of the main shaft, and capture the gradual evolution of the fault characteristics. For example, early bearing wear will cause the envelope entropy of the vibration signal to gradually increase, and the temperature rise rate to slowly rise. The model can learn these time series patterns and issue a warning before the fault is obvious (such as before the wear reaches a threshold). The main shaft fault may be caused by long-term accumulation of minor abnormalities (such as slow wear caused by poor lubrication). The time series model can effectively capture the dependence relationship across long time windows through the gating mechanism (such as the forget gate and input gate of LSTM) or self-attention mechanism (such as Transformer), avoiding that short-term noise masks long-term trends.

[0025] In some optional embodiments, the output of the fault detection model includes at least one of the following: The probability of surface wear, the probability of bearing pitting, the probability of shaft bending, and the probability of lubrication failure[2].

[0026] Wherein, surface wear is: long-term friction, poor lubrication or load fluctuation causes the gradual loss of surface material, forming micron-level scratches or pits.

[0027] Pitting corrosion of bearing is that local stress concentration, fatigue or lubrication failure leads to surface metal shedding, forming a small concave (pitting pit).

[0028] Shaft bending is that manufacturing defects, installation errors or long-term asymmetric load leads to elastic deformation of shaft body, and eccentricity increases when rotating.

[0029] Lubrication failure is that oil aging, leakage or pollution leads to insufficient oil film thickness, and metal-metal direct contact is caused.

[0030] The electronic device provided by the embodiment of the present application comprises a processor and a memory, the memory stores machine readable instructions executable by the processor, and the machine readable instructions are executed by the processor to perform the method described in any of the above.

[0031] The computer program product provided by the embodiment of the present application comprises a computer program / instruction, which realizes the steps of the method described in any of the above when executed by a processor. BRIEF DESCRIPTION OF DRAWINGS

[0032] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments of the present application. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0033] Figure 1 A step flow chart of a wind turbine main shaft fault detection method provided by the embodiment of the present application; Figure 2 A possible structure schematic diagram of the electronic device provided by the embodiment of the present application. DETAILED DESCRIPTION

[0034] The technical solutions in the embodiments of the present application will be described below in combination with the drawings in the embodiments of the present application.

[0035] Please refer to Figure 1 , Figure 1 A step flow chart of a wind turbine main shaft fault detection method provided by the embodiment of the present application, comprising: Step S1, using radar and temperature sensor to detect the rotating main shaft in real time to obtain radar information and temperature information; Among them, the radar can adopt UWB radar module, for example: NXP SR040 UWB chip, the working frequency is 6.5GHz±1.5GHz, and the detection accuracy is ±0.1mm (surface wear).

[0036] The temperature sensor can adopt a PT1000 platinum resistance, with a measurement range of -40℃ to 150℃ and an accuracy of ±0.5℃. Temperature detection can also use an infrared thermal imaging array to achieve non-contact detection of the temperature information of the main shaft.

[0037] Step S2, pre-processing and feature extraction are performed on the radar information to obtain the surface roughness and eccentricity; Among them, the pre-processing of the radar information includes pulse compression: compressing a wide pulse into a narrow pulse.

[0038] Surface roughness refers to the error of the micro-geometric shape of the processed surface, which is usually described by the height, spacing or shape characteristics of the micro-unevenness. In mechanical engineering, it reflects the degree of micro-peak and valley fluctuation of the surface and is an important indicator for evaluating the surface quality of parts. When radar waves irradiate the surface of the main shaft, the micro-peak and valley structure will change the reflection path of the radar waves, producing a micro-Doppler shift related to the surface roughness (Doppler effect caused by micro-deformation of the surface). By analyzing the spectral characteristics of the radar signal, the surface roughness can be quantitatively extracted, thereby capturing early micro-wear.

[0039] Eccentricity refers to the radial distance between the actual rotation center axis of a rotating component (such as a main shaft) and its geometric center axis (ideal axis). Simply put, it reflects the degree of "wobbling" of the main shaft during rotation. Radar can indirectly calculate the eccentricity by measuring the phase or frequency change of the reflected wave when the main shaft rotates. For example, when the main shaft radially jumps, the reflection path of the radar wave will be periodically offset, and by analyzing the periodicity and amplitude of this offset, the eccentricity can be quantitatively extracted.

[0040] Step S3, pre-processing and feature extraction are performed on the temperature information to obtain the temperature rise rate; Among them, the pre-processing of the temperature information includes moving average: calculating the average value of the values in the continuous window in the data sequence to eliminate short-term fluctuations and highlight long-term trends.

[0041] Temperature rise rate refers to the rate of change of temperature with time during the operation of a device or component, usually expressed in terms of temperature rise per unit time (such as ℃ / min or ℃ / hour). In the fault detection of wind turbine main shafts, the temperature rise rate is one of the key monitoring parameters, and its core role is to identify potential faults (such as lubrication failure, bearing wear or thermal expansion abnormalities) through the dynamic trend of temperature change.

[0042] Step S4, the surface roughness, eccentricity and temperature rise rate are fused to obtain multi-dimensional features; Among them, the surface roughness, eccentricity and temperature rise rate are respectively set with corresponding weights, and the multi-dimensional features are obtained according to the weights.

[0043] Step S5, inputting the multi-dimensional features into the trained fault detection model to obtain a fault detection result.

[0044] The fault detection model can adopt a lightweight edge model TinyFaultNet, which has a small model size, low cost, and low power consumption, and can be applied to an edge device MCU.

[0045] In the main shaft fault detection in the embodiments of the present application, a radar sensor is introduced, and the surface state of the main shaft is directly detected by using micro-Doppler effect (Doppler shift caused by target micro motion). The radar can capture the surface roughness change caused by the micro deformation or wear of the surface of the main shaft during rotation. Even if the wear amount is extremely small (such as microns), it can also be accurately identified through the phase / frequency change of the reflected wave. Combined with the eccentricity (radial runout amount during main shaft rotation) feature, the abnormality of the main shaft balance state (such as eccentricity caused by bearing wear) can be further reflected, providing a more sensitive detection means for early faults and improving the fault detection rate. The multi-modal fusion of radar + temperature sensor is adopted in the embodiments of the present application. The temperature rise rate (temperature change rate) extracted by the temperature sensor can reflect the abnormality of the main shaft friction state (such as early wear leading to increased friction, and significantly increased temperature rise rate). The surface roughness and eccentricity detected by the radar are complementary to the surface roughness and eccentricity, and also improve the fault detection rate.

[0046] Specifically, the surface roughness and eccentricity are extracted from the radar information features. In fact, the following features are first extracted from the radar echo signal, and then the surface roughness and eccentricity are determined according to these features: 1). Vibration characteristics: Vibration amplitude: Monitor the vibration amplitude of the rotating shaft in the radial (X, Y direction) and axial (Z direction) direction. Abnormally increased amplitude usually indicates problems such as imbalance, misalignment, bearing wear, looseness or blade damage.

[0047] Vibration frequency: Accurately measure the fundamental frequency (rotation frequency) and its harmonic, sub-harmonic components of the rotating shaft vibration. Specific faults (such as bearing defects, gear meshing problems, imbalance, misalignment) will produce characteristic frequency components (such as bearing pass frequency, gear meshing frequency and its sidebands). UWB can capture these spectral features.

[0048] Vibration mode: Analyze the time-domain waveform and trajectory (such as shaft center trajectory) of the vibration. Certain vibration modes (such as beat vibration, impact, chaos) are characteristic of certain faults (such as friction, severe looseness, crack development).

[0049] 2). Displacement and deformation: Shaft center position / trajectory: Real-time monitoring of the average position of the rotating shaft center and its trajectory within a rotation (shaft center trajectory). Changes in the trajectory shape (e.g. elliptical, banana-shaped, scattered dots) can directly reflect misalignment, shaft bending, excessive bearing clearance, oil film oscillation, etc.

[0050] Dynamic eccentricity: Measures the instantaneous offset of the rotating shaft relative to the theoretical center during rotation.

[0051] Axial displacement: Monitors the axial displacement of the rotating shaft. Excessive axial displacement may be caused by thrust bearing wear, uneven thermal expansion, coupling problems, or misalignment.

[0052] Bending / flexing: Detects the degree of bending deformation of the rotating shaft during operation, especially during start-stop and load changes. Persistent or excessive bending may be caused by gravity, thermal stress, manufacturing residual stress, or damage (such as cracks).

[0053] 3) Rotational speed and rotational stability: Rotational speed fluctuation: Accurately measures the actual rotational speed of the rotating shaft and its fluctuations. Abnormal rotational speed fluctuations or modulation components in the rotational speed signal may be related to load changes, control system problems, transmission chain failures (such as gear damage), or electrical faults.

[0054] Torsional vibration: Indirectly inferred by analyzing the tangential micro-displacement changes at specific points on the shaft (e.g. at obvious structural features) or specific frequency components (e.g. torsional vibration frequency). Severe torsional vibration can cause significant damage to gears, couplings, and the shaft itself.

[0055] In some optional embodiments, the radar is installed inside the wind turbine housing, the radar is arranged towards the main shaft, and the direction of the radar is arranged at a set angle with the central axis of the main shaft, the set angle being less than 70 degrees and greater than 20 degrees.

[0056] In the embodiments of the present application, the incident angle range of 20°-70° makes the radar wave incident on the surface of the main shaft in a "oblique" manner. At this time, the slight wear or cracks on the surface of the main shaft during rotation will change the reflection path of the radar wave, generating more abundant Doppler shift signals (i.e. micro-Doppler effect), thereby improving the detection sensitivity of micron-level surface deformation.

[0057] In some optional embodiments, further comprising: using a vibration sensor to detect the rotating main shaft in real time to obtain vibration information; preprocessing and feature extraction on the vibration information to obtain envelope entropy and harmonic energy ratio.

[0058] In some optional embodiments, the vibration sensor can be a MEMS three-axis vibration sensor, such as ADI ADXL1002, with a range of ±50g and a sampling rate of 20kHz.

[0059] Envelope Entropy: Quantify the periodic impact intensity of the signal by demodulating the envelope of the vibration signal. Early faults (such as bearing inner race pitting) will generate periodic impact vibrations, and envelope entropy can effectively capture this low-amplitude, high-periodic signal.

[0060] Harmonic Energy Ratio: By analyzing the energy proportion of the fundamental frequency and its harmonics in the vibration signal, gear wear, shaft misalignment, and other faults can be identified. For example, gear wear will cause the energy of specific harmonic components to increase significantly, and the harmonic energy ratio can quantify this change, improving the early identification of gear faults.

[0061] In the embodiments of the present application, the vibration features (envelope entropy, harmonic energy ratio) reflect the dynamic impact (such as periodic vibrations caused by bearing faults) and harmonic component changes (such as abnormal harmonic energy distribution caused by gear wear) when the main shaft rotates, which can accurately identify faults in key components such as bearings and gears. Radar features (surface roughness, eccentricity): reflect surface deformation and balance state abnormalities (such as journal wear, installation errors). Temperature features (temperature rise rate): reflect the dynamic process of frictional heating (such as local overheating caused by poor lubrication). The combination of the three can cover the mechanical, thermal, and vibration dimension features of main shaft faults, avoiding missed detection due to the limitations of a single sensor.

[0062] For offshore wind power scenarios, typhoon-induced vibration information contains both main shaft fault features and typhoon vortex-induced vibration features, which will affect the accuracy of fault judgment. Since the frequency range of typhoon vortex-induced vibration is usually 0.1-5 Hz (low frequency dominant), while the frequency range of main shaft fault vibration is usually 50 Hz-2 kHz (high frequency harmonic), the following methods are used in the embodiments to reduce the interference of typhoon: Perform FFT spectrum analysis by calculating the frequency spectrum of the vibration signal to extract the 0.1-5 Hz low-frequency component.

[0063] Perform energy proportion calculation. If the low-frequency energy proportion is >70%, it is determined to be typhoon interference.

[0064] Use Notch Filter to perform notch filtering in the 0.1-5 Hz frequency band to preserve high-frequency fault features.

[0065] Perform dynamic threshold adjustment, for example: increase the weight of radar information (without typhoon influence) and temperature information, or reduce the weight of vibration information.

[0066] In some optional embodiments, the surface roughness, eccentricity, and temperature rise rate are fused into features, including: According to the rotational speed, determine the weight settings of the radar information, temperature information, and vibration information; According to the weight setting, the surface roughness, eccentricity, temperature rise rate, envelope entropy and harmonic energy ratio are fused to obtain multi-dimensional features.

[0067] In some optional embodiments, the weight of the radar information is: w_uwb = 1 / (1 + k); The weight of the vibration information is: w_vib =a1×k; The weight of the temperature information is: w_temp = a2; Wherein, k is the speed adaptive weight; when the speed rpm is greater than the speed setting value, k=a3×rpm; when the speed rpm is less than or equal to the speed setting value, k=a4; a1, a2, a3 and a4 are fixed coefficient values.

[0068] In one specific embodiment, the speed setting value is 100 revolutions per minute, a1=0.8, a2=0.2, a3=0.02, and a4=0.5.

[0069] In the embodiments of the present application, at low speed, the weights of the radar information, the vibration information and the temperature information remain unchanged; at high speed, as the speed increases, the weight of the radar information decreases, the weight of the vibration information increases, and the weight of the temperature information remains unchanged. The weight adjustment strategy of the present embodiment adapts to the difference in fault feature sensitivity at different speeds, optimizes sensor data fusion, balances calculation resources and detection accuracy, ultimately reduces the early fault omission rate, and improves the system robustness and practicality.

[0070] In some optional embodiments, the surface roughness, eccentricity and temperature rise rate are fused, including: According to the speed and the current temperature, the weight settings of the radar information, the temperature information and the vibration information are determined; According to the weight setting, the surface roughness, eccentricity, temperature rise rate, envelope entropy and harmonic energy ratio are fused to obtain multi-dimensional features.

[0071] In some optional embodiments, the weight of the radar information is: w_uwb = 1 / (1 + k); When the current temperature is less than or equal to the temperature setting value, the weight of the vibration information is: w_vib =a1× k; when the current temperature is greater than the temperature setting value, the weight of the vibration information is: w_vib =a5; The weight of the temperature information is: w_temp = a2; Wherein, k is the speed adaptive weight; when the speed rpm is greater than the speed setting value, k=a3×rpm; when the speed rpm is less than or equal to the speed setting value, k=a4; a1, a2, a3, a4 and a5 are fixed coefficient values.

[0072] In one embodiment, a1=0.8, a2=0.2, a3=0.02, a4=0.5, a5=0.7, the temperature set value is 70 degrees, and the rotational speed set value is 100 revolutions per minute.

[0073] In the embodiments of the present application, in the low-temperature scenario, the weight of the vibration information is high, and the weight of the vibration information increases with the increase of the rotational speed. In the high-temperature scenario, the weight of the vibration information is maintained at a lower weight, and the reason is that high temperature may be caused by lubrication failure, increased friction or thermal expansion, at which time the vibration signal may be distorted due to thermal deformation.

[0074] In some optional embodiments, the temperature sensor and the vibration sensor are installed on the main shaft bearing seat of the wind turbine.

[0075] In some optional embodiments, the fault detection model is a time series sensitive model.

[0076] In the embodiments of the present application, the time series sensitive model (such as LSTM, GRU, and Transformer) can remember the historical running state of the main shaft and capture the gradual evolution of the fault characteristics. For example, early bearing wear will cause the envelope entropy of the vibration signal to gradually increase and the temperature rise rate to slowly rise. The model can learn these time series patterns and issue a warning before the fault is obvious (such as before the wear reaches a threshold value). The main shaft fault may be caused by long-term accumulation of minor abnormalities (such as slow wear caused by poor lubrication). The time series model can effectively capture the dependency relationship across a long time window through a gating mechanism (such as the forget gate and input gate of LSTM) or a self-attention mechanism (such as Transformer), thereby avoiding short-term noise from masking long-term trends.

[0077] In some optional embodiments, the output of the fault detection model includes at least one of the following: the probability of surface wear, the probability of bearing pitting, the probability of shaft bending, and the probability of lubrication failure.

[0078] The surface wear is caused by long-term friction, poor lubrication, or load fluctuation, which gradually causes the surface material to flow away and form micron-level scratches or pits.

[0079] The bearing pitting is caused by local stress concentration, fatigue, or lubrication failure, which causes the surface metal to fall off and form a small depression (pitting pit).

[0080] The shaft bending is caused by manufacturing defects, installation errors, or long-term asymmetric load, which causes the shaft body to elastically deform and the eccentricity to increase when rotating.

[0081] The lubrication failure is caused by aging, leakage, or contamination of the lubricating oil, which causes the oil film to be insufficient in thickness and causes metal-to-metal direct contact.

[0082] In some optional embodiments, the spindle health state can also be comprehensively evaluated according to the probability of surface wear, the probability of bearing pitting, the probability of shaft bending, and the probability of lubrication failure. Subsequently, the future maintenance cycle is determined according to the spindle health state to avoid the situation of shutdown.

[0083] Figure 2 A possible structure of the electronic device provided by the embodiments of the present application is shown. Referring to Figure 2 , the electronic device includes a processor, a memory, and a communication interface, which are interconnected and communicate with each other through a communication bus and / or other forms of connection mechanism (not shown).

[0084] The memory includes one or more (only one is shown in the figure), which can be, but is not limited to, a random access memory (RAM), a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), etc. The processor and other possible components can access the memory to read and / or write data therein.

[0085] The processor includes one or more (only one is shown in the figure), which can be an integrated circuit chip with signal processing capabilities. The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a microcontroller unit (MCU), a network processor (NP), or other conventional processors; it can also be a special-purpose processor, including a neural network processing unit (NPU), a graphics processing unit (GPU), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. Moreover, when there are multiple processors, some of them can be general-purpose processors and others can be special-purpose processors.

[0086] The communication interface includes one or more (only one is shown in the figure) and can be used to communicate directly or indirectly with other devices to exchange data. The communication interface can include an interface for wired and / or wireless communication.

[0087] One or more computer program instructions may be stored in the memory, and the processor may read and execute these computer program instructions to implement the method provided in the embodiment of the present application.

[0088] Understandably, Figure 2 The structure shown is only for illustration, and the electronic device may also include Figure 2 More or fewer components than shown, or with Figure 2 Different structures are shown. Figure 2 The components shown in the figure can be implemented using hardware, software, or a combination thereof. The electronic device can be a physical device, such as a PC, laptop, tablet, mobile phone, server, embedded device, etc., or a virtual device, such as a virtual machine or virtualized container. Furthermore, the electronic device is not limited to a single device and can also be a combination of multiple devices or a cluster consisting of a large number of devices.

[0089] The computer program product provided in the embodiments of the present application comprises computer programs / instructions, which, when executed by a processor, implement the steps of any of the above-described methods.

[0090] In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method can be implemented in other manners. The embodiments described above are merely schematic; for example, the division of the units is only a logical function division; there can be another division manner in actual implementation; for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between the units can be indirect couplings or communication connections through some interfaces, communication interfaces, or a combination of physical or logical interfaces.

[0091] In addition, the units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units; that is, they can be located in one place, or can be distributed on a plurality of network units. In actual implementation, some or all of the units can be selected according to actual needs to achieve the purposes of the embodiments of the present application.

[0092] In addition, each functional module in the various embodiments of the present application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0093] In this document, the terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations.

[0094] The above only describes the embodiments of the present application, and is not used to limit the protection scope of the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for detecting a main shaft fault of a wind turbine generator, characterized in that: include: Use radar and temperature sensors to detect the rotating spindle in real time to obtain radar information and temperature information; Preprocessing and feature extraction are performed on the radar information to obtain surface roughness and eccentricity; Preprocessing and feature extraction are performed on the temperature information to obtain a temperature rise rate; Performing feature fusion on the surface roughness, eccentricity and temperature rise rate to obtain multi-dimensional features; The multidimensional features are input into the trained fault detection model to obtain a fault detection result.

2. The method according to claim 1, wherein The radar is installed inside the wind turbine housing, the radar is arranged toward the main shaft, and the direction of the radar is arranged at a set angle with the central axis of the main shaft, and the set angle is less than 70 degrees and greater than 20 degrees.

3. The method according to claim 1, wherein Also includes: Use vibration sensors to detect the rotating spindle in real time to obtain vibration information; The vibration information is preprocessed and features are extracted to obtain envelope entropy and harmonic energy ratio.

4. The method according to claim 3, wherein The feature fusion of the surface roughness, eccentricity and temperature rise rate includes: Determining weight settings for the radar information, temperature information, and vibration information based on the rotational speed; According to the weight setting, the surface roughness, eccentricity, temperature rise rate, envelope entropy and harmonic energy ratio are subjected to feature fusion to obtain the multi-dimensional feature.

5. The method according to claim 4, wherein The weight of the radar information is: w_uwb = 1 / (1 + k); The weight of the vibration information: w_vib = a1×k; The weight of the temperature information: w_temp = a2; Wherein, k is the speed adaptation weight; when the speed rpm is greater than the speed setting value, k = a3 × rpm; when the speed rpm is less than or equal to the speed setting value, k = a4; a1, a2, a3, and a4 are fixed coefficient values.

6. The method according to claim 3, wherein The feature fusion of the surface roughness, eccentricity and temperature rise rate includes: Determining weight settings for the radar information, temperature information, and vibration information based on the rotation speed and current temperature; According to the weight setting, the surface roughness, eccentricity, temperature rise rate, envelope entropy and harmonic energy ratio are subjected to feature fusion to obtain the multi-dimensional feature.

7. The method according to claim 6, wherein The weight of the radar information is: w_uwb = 1 / (1 + k); When the current temperature is less than or equal to the temperature setting value, the weight of the vibration information is: w_vib = a1 × k; when the current temperature is greater than the temperature setting value, the weight of the vibration information is: w_vib = a5; The weight of the temperature information: w_temp = a2; Wherein, k is the speed adaptation weight; when the speed rpm is greater than the speed setting value, k=a3×rpm; when the speed rpm is less than or equal to the speed setting value, k=a4; a1, a2, a3, a4, and a5 are fixed coefficient values.

8. The method according to claim 2, wherein The temperature sensor and the vibration sensor are installed on the main shaft bearing seat of the wind generator.

9. The method according to claim 1, wherein The fault detection model is a time series sensitive model.

10. The method according to claim 1, wherein The output of the fault detection model includes at least one of the following: Probability of surface wear, probability of bearing pitting, probability of shaft bending, and probability of lubrication failure.

11. An electronic device, characterized in that: include: A processor and a memory, wherein the memory stores machine-readable instructions executable by the processor, and when the machine-readable instructions are executed by the processor, the method according to any one of claims 1 to 10 is performed.

12. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instruction is executed by a processor, the steps of the method according to any one of claims 1 to 10 are implemented.

Images