CN-122014529-A - Fan bearing detection method and device for wind power generation equipment

CN122014529ACN 122014529 ACN122014529 ACN 122014529ACN-122014529-A

Abstract

The application discloses a method and a device for detecting a fan bearing of wind power generation equipment, which are used for realizing mutual coupling of qualitative type identification and quantitative severity assessment in fault diagnosis by fusing multisource data of high-frequency vibration and weak impact and introducing a dynamic differential equation as physical constraint, thereby remarkably improving the physical credibility of a diagnosis result and the accurate representation capability of a complex damage state of the bearing.

Inventors

WU SHIDONG
GUO SHUAINAN
SU YIBO
WANG QIAN
ZHU XIAOYI
HAN JUNFEI

Assignees

中国长江三峡集团有限公司

Dates

Publication Date: 20260512
Application Date: 20260210

Claims (10)

1. The method for detecting the fan bearing of the wind power generation equipment is characterized by comprising the following steps of: collecting a high-frequency vibration signal, a weak impact signal and an instantaneous rotating speed signal of the fan bearing; determining fault type probability distribution reflecting fault types and fault severity assessment results reflecting damage sizes or abrasion depths through the high-frequency vibration signals, the weak impact signals and the instantaneous rotating speed signals; Deriving theoretical characteristic frequencies of an inner ring, an outer ring and rolling bodies of the fan bearing based on bearing structure parameters of the fan bearing, and establishing a dynamic differential equation residual term representing a system evolution rule based on the theoretical characteristic frequencies; generating a target model based on the fault type probability distribution, the fault severity assessment result and the dynamic differential equation residual term; when target data comprising a target frequency vibration signal, a target weak impact signal and a target instantaneous rotating speed signal are obtained, a detection result for a target fan bearing is output by adopting the target model based on the target data.
2. The method according to claim 1, wherein the step of determining a fault type and probability distribution reflecting a fault type from the dither signal, the weak impact signal, and the instantaneous rotation speed signal, and a fault severity evaluation result reflecting a damage size or a wear depth includes: The collected high-frequency vibration signal, weak impact signal and instantaneous rotating speed signal are determined as original signals of the bearing; Preprocessing the original signal of the bearing to obtain signal data subjected to preprocessing operation; resampling the pre-processed signal data, converting the pre-processed signal data from a time domain non-stationary signal to an angular domain stationary signal; extracting a time domain statistical index from the angle domain stationary signal to obtain a time-frequency diagram and a multidimensional feature vector which reflect the running state of the bearing; constructing a multi-mode input tensor containing vibration texture information and energy distribution information by adopting the time-frequency diagram and the multi-dimensional feature vector; extracting the characteristics of the multi-mode input tensor, and generating a high-dimensional shared characteristic vector by using a channel attention mechanism to carry out self-adaptive weighting; And determining fault type probability distribution reflecting fault types and fault severity assessment results reflecting damage sizes or abrasion depths based on the high-dimensional shared feature vector.
3. The method according to claim 2, wherein the step of determining a fault type probability distribution reflecting a fault type based on the high-dimensional shared feature vector, and a fault severity assessment result reflecting a damage size or a wear depth comprises: Performing full-connection mapping and normalization calculation on the high-dimensional shared feature vector by adopting a primary fault type classifier to obtain fault type probability distribution reflecting fault types; And extracting the probability distribution of the fault type as a routing control signal to activate a secondary sub-branch network, and guiding the secondary sub-branch network to perform mining operation on the high-dimensional shared feature vector to obtain a fault severity assessment result reflecting the damage size or the abrasion depth.
4. The method of claim 1, wherein the step of deriving theoretical feature frequencies of the inner race, the outer race, and the rolling elements of the fan bearing based on the bearing structure parameters of the fan bearing, and establishing a residual term of a dynamic differential equation characterizing a system evolution law based on the theoretical feature frequencies comprises: According to the Hertz contact theory and a bearing kinematics equation, deriving theoretical characteristic frequencies of an inner ring, an outer ring and rolling bodies of the fan bearing based on bearing structure parameters of the fan bearing, and establishing a dynamic differential equation residual term representing a system evolution rule by adopting the theoretical characteristic frequencies in combination with a single-degree-of-freedom spring damping quality model.
5. The method of any one of claims 1-4, wherein the step of generating a target model based on the fault type probability distribution, the fault severity assessment result, and the dynamic differential equation residual term comprises: establishing a composite total loss function consisting of hierarchical classification loss, physical consistency loss and logic dependence constraint loss by combining the fault type probability distribution, the fault severity evaluation result and the dynamic differential equation residual error term; And adopting a step-by-step combined training mode and utilizing AdamW optimizers to carry out iterative updating on the whole network parameters based on the composite total loss function, and training to obtain a collaborative diagnosis model with physical constraint and nested logic.
6. The method of claim 5, wherein the step of outputting a detection result for a target fan bearing based on the target data using the target model comprises: And inputting the target data into the collaborative diagnosis model, and generating a diagnosis report containing fault types, severity and physical credibility scores by using a primary classification logic and a secondary evaluation subnet of the model.
7. The method of claim 2, wherein the step of preprocessing the bearing raw signal to obtain preprocessed signal data comprises: And filtering out power frequency interference and high-frequency noise of the original signal of the bearing by adopting band-pass filtering to obtain filtered signal data.
8. The utility model provides a fan bearing detection device of wind power generation equipment which characterized in that includes: The data acquisition module is used for acquiring high-frequency vibration signals, weak impact signals and instantaneous rotating speed signals of the fan bearing; The fault type probability distribution and fault severity assessment result determining module is used for determining fault type probability distribution reflecting fault types and fault severity assessment results reflecting damage sizes or abrasion depths through the high-frequency vibration signals, the weak impact signals and the instantaneous rotating speed signals; The dynamic differential equation residual term deducing module is used for deducing theoretical characteristic frequencies of an inner ring, an outer ring and rolling bodies of the fan bearing based on bearing structure parameters of the fan bearing, and establishing dynamic differential equation residual terms representing a system evolution rule based on the theoretical characteristic frequencies; The target model generation module is used for generating a target model based on the fault type probability distribution, the fault severity evaluation result and the dynamic differential equation residual error term; And the detection result generation module is used for outputting a detection result aiming at the target fan bearing based on the target data by adopting the target model when the target data comprising the target high-frequency vibration signal, the target weak impact signal and the target instantaneous rotating speed signal are acquired.
9. An electronic device comprising a processor, a memory and a program or instruction stored on the memory and executable on the processor, which program or instruction when executed by the processor implements the method of claims 1-7.
10. A readable storage medium, characterized in that it stores thereon a program or instructions, which when executed by a processor, implements the method according to claims 1-7.

Description

Fan bearing detection method and device for wind power generation equipment Technical Field The invention relates to the technical field of wind turbine bearing detection of wind power generation equipment, in particular to a wind turbine bearing detection method of wind power generation equipment, a wind turbine bearing detection device of wind power generation equipment, electronic equipment and a readable storage medium. Background The wind power generation equipment runs for a long time in remote and severe environments and is influenced by random wind speed, alternating load and complex vibration environments, and various faults such as fatigue peeling, abrasion and the like are very easy to occur to a core component of a transmission chain, namely a rolling bearing. The health of the bearings directly determines the unplanned downtime and operating costs of the unit. At present, a deep learning method based on data driving is widely applied to fault diagnosis of a fan bearing. However, in practical engineering applications, the existing detection method still faces the following technical bottlenecks. First, the related art relies on a single vibration signal for analysis, but weak impact characteristics are easily masked in early bearing failure or in complex noise environments, resulting in limited detection sensitivity. In addition, the fluctuation of the rotating speed of the fan makes the time domain feature show non-stationarity, and the fault feature is difficult to stably extract. Secondly, the deep learning model of the related art is mostly of a black box structure, and pattern recognition is carried out only by means of massive historical data. Because the internal logic of the model is disjointed with the actual mechanical dynamics rules (such as Hertz contact theory, system damping characteristics and the like) of the bearing, the generalization capability of the model under small sample or extreme working conditions is weaker, and the diagnosis result lacks support of physical basis. Disclosure of Invention Embodiments of the present invention provide a method, an apparatus, an electronic device, and a readable storage medium for detecting a fan bearing of a wind power generation device, so as to overcome or at least partially solve the above-mentioned problems. In order to solve the technical problems, the application is realized as follows: In a first aspect, an embodiment of the present application provides a method for detecting a fan bearing of a wind power generation device, including: collecting a high-frequency vibration signal, a weak impact signal and an instantaneous rotating speed signal of the fan bearing; determining fault type probability distribution reflecting fault types and fault severity assessment results reflecting damage sizes or abrasion depths through the high-frequency vibration signals, the weak impact signals and the instantaneous rotating speed signals; Deriving theoretical characteristic frequencies of an inner ring, an outer ring and rolling bodies of the fan bearing based on bearing structure parameters of the fan bearing, and establishing a dynamic differential equation residual term representing a system evolution rule based on the theoretical characteristic frequencies; generating a target model based on the fault type probability distribution, the fault severity assessment result and the dynamic differential equation residual term; when target data comprising a target frequency vibration signal, a target weak impact signal and a target instantaneous rotating speed signal are obtained, a detection result for a target fan bearing is output by adopting the target model based on the target data. Optionally, the step of determining the fault type and probability distribution reflecting the fault type and the fault severity assessment result reflecting the damage size or the wear depth from the dither signal, the weak impact signal, and the instantaneous rotation speed signal includes: The collected high-frequency vibration signal, weak impact signal and instantaneous rotating speed signal are determined as original signals of the bearing; Preprocessing the original signal of the bearing to obtain signal data subjected to preprocessing operation; resampling the pre-processed signal data, converting the pre-processed signal data from a time domain non-stationary signal to an angular domain stationary signal; extracting a time domain statistical index from the angle domain stationary signal to obtain a time-frequency diagram and a multidimensional feature vector which reflect the running state of the bearing; constructing a multi-mode input tensor containing vibration texture information and energy distribution information by adopting the time-frequency diagram and the multi-dimensional feature vector; extracting the characteristics of the multi-mode input tensor, and generating a high-dimensional shared characteristic vector by using a channel attention mecha