Dynamic response monitoring method and system for large-size fan

Abstract

The invention discloses a dynamic response monitoring method and system for a large-size fan, and belongs to the technical field of remote monitoring of power equipment, and the method comprises the steps: obtaining a component structure image and sensor data of a to-be-detected component; detecting external structure change of the to-be-detected component according to the component structure image to obtain external abnormal features of the to-be-detected component; according to the temperature change data and the material type and the structure size of the to-be-detected component, simulating the internal structure change of the to-be-detected component to obtain the internal abnormal characteristics of the to-be-detected component; and based on the external abnormal features and the internal abnormal features, combining the meteorological change condition and the fan operation data, analyzing fault causes, and obtaining a fault diagnosis result. Therefore, by implementing the method, the device and the system, the problem that a comprehensive and accurate fault diagnosis result is obtained by quickly and accurately detecting fault causes and fault positions outside and inside the fan component under the condition that the fan operation is not suspended in the prior art can be solved.

Description

Technical Field

[0001] The present application belongs to the technical field of remote monitoring of electric power equipment, and specifically relates to a dynamic response monitoring method and system for large-scale wind turbines. Background Art

[0002] As the global energy mix shifts toward cleaner, lower-carbon energy, the proportion of wind power in the power system continues to increase annually. However, wind turbines ("wind turbines") are prone to structural loosening, corrosion, or the presence of foreign matter on their surfaces during extended operation, leading to abnormal wind turbine loads. Traditional inspection methods rely on manual inspections of core and vulnerable parts, but this approach is inefficient and costly. Therefore, remote monitoring of wind turbines is a key approach to ensuring reliable wind power generation and reducing operational costs.

[0003] Currently, determining whether a wind turbine is experiencing an anomaly is often done by capturing structural images of components or collecting operational data. External structural images can only identify external anomalies (such as the presence of foreign matter or cracks on the component surface). Internal anomalies, such as insufficient bearing lubrication or poor gear meshing, require further inspection of the component's internal structure. However, due to safety restrictions, the wind turbine can only be suspended for internal inspection, which results in excessively high inspection costs. Therefore, how to conduct a comprehensive inspection of the external and internal structures of wind turbine components without suspending wind turbine operation to determine the exact cause and location of the fault, provide support for the timely implementation of appropriate maintenance measures, and further improve the operational stability of the power system. Summary of the Invention

[0004] This application proposes a dynamic response monitoring method and system for large-scale wind turbines, which can solve the problem in the prior art of quickly and accurately detecting the cause and location of faults on the outside and inside of wind turbine components without suspending the operation of the wind turbine, and obtaining comprehensive and accurate fault diagnosis results.

[0005] A first aspect of the present application provides a dynamic response monitoring method for a large-scale wind turbine, the method comprising: Acquire component structure images and sensor data of the components to be inspected of the wind turbine; wherein the sensor data includes temperature change data, vibration data, and noise data; Detecting changes in the external structure of the component to be detected based on the component structure image and historical normal structure images to obtain external abnormal features of the component to be detected; According to the temperature change data and the material type and structural dimensions of the component to be detected, the internal structural changes of the component to be detected are simulated by the noise data and the vibration data to obtain internal abnormal characteristics of the component to be detected; Based on the external abnormal characteristics and the internal abnormal characteristics, combined with the meteorological changes in the current period and the wind turbine operation data, the fault cause of the component to be detected is analyzed to obtain a corresponding fault diagnosis result.

[0006] The above scheme detects whether there are any abnormalities in the external structure and internal structure of the components to be inspected in the fan. By comparing the captured component structure image with the historical normal structure image, it is determined whether the external structure of the component to be inspected has changed, and the external abnormality characteristics are obtained. Then, based on the vibration and noise emitted by the component during operation, it is judged whether the internal structure of the component has changed. Based on the fact that the components to be inspected will react differently to temperature changes due to different material types, the degree of internal structure change is quantified, and the internal structure abnormality of the fan components can be judged only by the data collected by the sensor. There is no need to suspend the operation of the fan, which saves a lot of inspection costs and improves inspection efficiency, and obtains accurate internal abnormality characteristics. Finally, the cause of the fault is analyzed based on the impact of meteorological changes and fan operation data on the damage to the components to be inspected, and the fault causes corresponding to the external abnormality characteristics and the internal abnormality characteristics are obtained. Comprehensive and accurate fault diagnosis results are obtained to provide support for subsequent maintenance personnel to implement appropriate maintenance measures.

[0007] In a possible implementation method of the first aspect, external structural changes of the component to be detected are detected based on the component structural image and the historical normal structural image to obtain external abnormal features of the component to be detected, specifically: Acquiring the historical normal structural image of the component to be inspected within a previous period, comparing the historical normal structural image with the component structural image, and cropping the abnormal area in the component structural image based on the comparison result to obtain an image to be identified; Predicting the potential fault type of the image to be identified based on a preset component abnormality type library; Screening the potential fault types by using the image to be identified and the collected weight increase and decrease data of the component to be inspected, and determining the fault type of each component to be inspected; According to the fault type and the image to be identified, an external abnormality feature of the component to be detected is constructed.

[0008] This approach first crops the image to be identified by identifying differences from historical normal structural images. It then matches these images against a library of component anomaly types, predicting the likely type of potential fault for each image. To further determine the type of external fault occurring on the component under inspection, the system uses weight changes to determine whether the component is experiencing inherent aging, structural deformation, or the presence of foreign matter on its surface, enabling comprehensive detection of external anomalies.

[0009] In a possible implementation method of the first aspect, the potential fault types are screened using the image to be identified and the collected weight increase / decrease data of the component to be inspected to determine the fault type of each component to be inspected, specifically: If there is only one potential fault type for the image to be identified, the potential fault type is used as the fault type of the component to be detected; If there is more than one potential fault type in an image to be identified, the weight increase and decrease data of the component to be detected is obtained, and the surface texture and color distribution of the component to be detected are obtained from the image to be identified; the confidence of each potential fault type is calculated based on the weight increase and decrease data, the surface texture and color distribution, and the potential fault type with the highest confidence is selected as the fault type of the component to be detected.

[0010] In this approach, multiple potential fault types are combined with weight change data, surface texture, and color distribution to accurately determine the type of external fault in the component under inspection. For uncertain fault types, the weight change, surface texture, and color distribution can be used to determine whether the component under inspection is experiencing foreign matter, corrosion, or abnormal textures, allowing for accurate visual identification of the fault type for each component under inspection.

[0011] In a possible implementation method of the first aspect, based on the temperature change data and the material type and structural dimensions of the component to be inspected, internal structural changes of the component to be inspected are simulated using the noise data and the vibration data to obtain internal abnormality characteristics of the component to be inspected, specifically: extracting abnormal noise signals and abnormal vibration signals from the noise data and the vibration data, respectively, based on the noise characteristics and vibration characteristics of the component to be inspected under normal working conditions; Align and fuse abnormal noise signals and abnormal vibration signals to obtain abnormal change features; With the goal of achieving the highest similarity with the abnormal change characteristics, the internal structural changes of the component to be detected are simulated through the temperature change data and the material type based on the normal structural dimensions of the component to be detected in the previous period to obtain the internal abnormal characteristics of the component to be detected.

[0012] The above scheme takes into account that the internal structure of the component to be inspected is difficult to capture directly by photographing. Therefore, the noise data and vibration data of the component during operation are collected by sensors, and abnormal data that is different from normal working conditions is extracted from these data. Since vibration and noise are correlated, the acquired abnormal noise signal and abnormal vibration signal are fused to obtain abnormal change characteristics. Then, the operation of the component is simulated based on the normal structural dimensions of the previous period, until the component emits vibrations and noises that are highly similar to the abnormal change characteristics during the simulation. It can be considered that the simulation results at this time are highly similar to the actual internal structure of the component to be inspected, thereby determining which internal structural changes have occurred in the component to be inspected. Accurate internal abnormal characteristics can be safely obtained without suspending the operation of the fan, saving a lot of inspection costs.

[0013] In a possible implementation method of the first aspect, based on the normal structural dimensions of the component to be inspected in a previous period, the internal structural changes of the component to be inspected are simulated using the temperature change data and the material type to obtain internal abnormal characteristics of the component to be inspected, specifically: Simulating the influence of temperature on each material type according to the temperature change data to obtain a first structural characteristic of each component to be inspected; According to the connection structure between the component to be inspected and other components, the degree of damage to the component to be inspected caused by friction generated by the material type during operation of the component to be inspected is simulated to obtain a second structural characteristic of each component to be inspected; By using the first structural characteristic and the second structural characteristic, the internal structural changes of the component to be inspected are continuously simulated on the basis of the normal structural dimensions, and the similarity between the noise signal and the vibration signal generated by the component to be inspected and the abnormal change characteristics under each internal structural change is calculated; When the similarity is the highest, the simulation operation is stopped, and the current simulation result is compared with the normal structure size to construct the internal abnormality feature.

[0014] To accurately determine changes in a component's internal structure, the above scheme considers that different material types can lead to different structural changes and, consequently, different noise and vibration under the same temperature change. Therefore, a first structural characteristic of the component to be tested is constructed. Furthermore, the influence of component connections on noise and vibration, as well as component wear, is considered to construct a second structural characteristic of the component to be tested. Finally, based on these first and second structural characteristics, the internal structural changes of the component to be tested are accurately simulated, resulting in internal anomaly signatures that more closely resemble actual conditions.

[0015] In a possible implementation method of the first aspect, the abnormal noise signal and the abnormal vibration signal are aligned and fused to obtain an abnormal change feature, specifically: Calculate the cross-correlation coefficient and common peak value of abnormal noise signal and abnormal vibration signal; The correlation between abnormal noise signals and abnormal vibration signals in time domain and frequency domain is analyzed by wavelet transform to obtain relevant features; A mapping relationship between abnormal noise signals and abnormal vibration signals is constructed based on the mutual correlation coefficient, the common peak value and the correlation feature, and then data fusion is performed on the abnormal noise signals and the abnormal vibration signals through the mapping relationship to obtain abnormal change features.

[0016] The above solution takes into account the correlation between the noise and vibration emitted by the components during operation. This correlation is captured by finding peak values and aligning them in time, obtaining relevant features to construct a corresponding mapping relationship, providing support for accurately determining the location of the structural changes in the component to be inspected.

[0017] In a possible implementation method of the first aspect, based on the external abnormality characteristics and the internal abnormality characteristics, combined with meteorological changes in the current period and wind turbine operation data, the cause of the failure of the component to be detected is analyzed to obtain a corresponding fault diagnosis result, specifically: Extracting service life and working condition change data of each component to be tested from the wind turbine operation data; According to the external abnormality characteristics and the internal abnormality characteristics, corresponding damage association labels are respectively constructed for the service life, the operating condition change data, and the meteorological change situation; Using the external abnormality features and the internal abnormality features as result variables, causal reasoning is performed on the fault cause of the component to be detected to obtain an influence weight of each damage association label and a causal graph related to the fault cause; The direct and indirect causes of the causal diagram are analyzed by using a path verification method and the influence weights to obtain the fault diagnosis result; wherein the fault diagnosis result includes the fault cause, fault location and fault type of the component to be detected.

[0018] Because the fault causes of external and internal abnormal characteristics are different, and there is a certain coupling relationship between the various fault causes, the above solution constructs corresponding damage association labels for the service life, the operating condition change data, and the meteorological changes, providing support for the subsequent multiple analysis of the fault causes. Then, through reverse reasoning, using the structural abnormality of the component as the result, the damage association label is used to analyze the degree of impact and impact of the component to be tested. By constructing a cause-and-effect diagram to describe the relationship between the fault cause and the component structural abnormality and to characterize the coupling relationship between different fault causes, the direct and indirect causes of the component to be tested are predicted, resulting in accurate and comprehensive fault diagnosis results.

[0019] In a possible implementation method of the first aspect, corresponding damage association labels are constructed for the service life, the operating condition change data, and the meteorological change conditions, respectively, based on the external abnormality characteristics and the internal abnormality characteristics, specifically: Based on historical inspection records, quantify the degree of damage caused by the service life of the component to be inspected, and construct an impact feature of the service life; quantifying the degree of damage influence of the operating condition change data on the component to be inspected based on the sampling frequency and the component load change curve, and constructing the influence characteristics of the operating condition change data; Constructing the impact characteristics of the meteorological change situation based on the long-term characteristics and short-term characteristics of the meteorological change situation; The external abnormality feature and the internal abnormality feature are used as results, and the damage association label is generated for each of the influencing features by searching for the fault cause in the historical fault records of the wind turbine.

[0020] The above scheme uses existing data to quantify the degree of influence of the service life, the operating condition change data and the meteorological change conditions on the damage of the components to be inspected, thereby obtaining corresponding impact characteristics to characterize the damage these factors will cause to the components, providing data support for the subsequent deduction of the cause of the failure.

[0021] In a possible implementation method of the first aspect, using the external abnormality feature and the internal abnormality feature as result variables, causal reasoning is performed on the fault cause of the component to be detected to obtain an influence weight of each damage-related label and a causal graph related to the fault cause, specifically: Based on historical fault records, the external abnormality characteristics and the internal abnormality characteristics are used as result variables, the wind turbine operation data is used as an intermediate response variable, and the damage association label is used as a potential cause variable; By means of a conditional independence test, a causal diagram skeleton is constructed based on the relationship between the potential cause variable and the intermediate response variable and the relationship between the outcome variable and the intermediate response variable; generating an impact weight of each of the damage-related labels according to the causal effect of the potential cause variable on the intermediate response variable; The causal graph skeleton is annotated according to the influence weight, and a causal forest is constructed on the annotated causal graph skeleton with the result variable as an end point to obtain the causal graph.

[0022] The above scheme uses existing historical fault records as causal events, and the external and internal abnormal characteristics as the outcomes of the events. The relationship provided by the causal events is used to deduce the likelihood of the fault cause, resulting in a causal graph skeleton. Because different potential causal variables can cause varying degrees of change in wind turbine operating data, for example, a linear relationship exists between the degree of sudden change in wind speed and the degree of load increase on the wind turbine blades. Based on this causal effect, a weight is assigned to each damage-related label to further analyze whether the fault cause is direct or indirect, resulting in a causal graph that comprehensively displays the various fault causes of the component under inspection and the coupling relationships between them.

[0023] A second aspect of the present application provides a dynamic response monitoring system for a large-scale wind turbine, the system comprising: a data acquisition module, an external feature acquisition module, an internal feature acquisition module, and a fault cause analysis module; The data acquisition module is used to acquire component structure images and sensor data of the components to be inspected of the wind turbine; wherein the sensor data includes temperature change data, vibration data and noise data; The external feature acquisition module is used to detect the external structural changes of the component to be detected based on the component structure image and the historical normal structure image, and obtain the external abnormal features of the component to be detected; The internal feature acquisition module is used to simulate the internal structural changes of the component to be detected based on the temperature change data and the material type and structural dimensions of the component to be detected through the noise data and the vibration data to obtain the internal abnormality features of the component to be detected; The fault cause analysis module is used to analyze the fault cause of the component to be detected based on the external abnormal characteristics and the internal abnormal characteristics, combined with the meteorological changes in the current period and the fan operation data, to obtain the corresponding fault diagnosis result. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solution of the present application, the following is a brief introduction to the drawings required for use in the implementation. Obviously, the drawings described below are only some implementation methods of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0025] Figure 1 This is a schematic diagram of a specific flow chart of a dynamic response monitoring method for a large-scale wind turbine provided in one embodiment of the present application; Figure 2 This is a structural diagram of a dynamic response monitoring system for a large-scale wind turbine provided in one embodiment of the present application. DETAILED DESCRIPTION

[0026] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0027] It should be understood that the step numbers used herein are only for convenience of description and are not intended to limit the order in which the steps are to be executed.

[0028] First embodiment The power generation fan is a key equipment of the wind power generation system, and its structural health directly affects the operating efficiency and safety of the system. In order to improve the efficiency and timeliness of the inspection of fan components, it is now common to collect information on key components such as blades, bearings and gears of the fan through sensors and drones to achieve remote monitoring. However, the monitoring means for the internal structure of the fan are still relatively scarce. It is still difficult to fully analyze the cause and location of the fan failure only through the collected temperature data and vibration data. Further field inspections of the fan are required to obtain accurate fault analysis results. Therefore, the implementation of this application further optimizes the internal structure detection of fan components in response to the above problems. Only through remotely collected multi-source data can the external and internal fault causes and fault locations of the fan components be detected quickly and accurately, and comprehensive and accurate fault diagnosis results can be obtained.

[0029] like Figure 1As shown, in order to solve the problem in the prior art of quickly and accurately detecting the cause and location of faults on the outside and inside of fan components without suspending the operation of the fan, and obtaining comprehensive and accurate fault diagnosis results, the first embodiment of the present application provides a specific flow chart of a dynamic response monitoring method for a large-scale fan. The dynamic response monitoring method for a large-scale fan of this embodiment includes steps S1 to S4, which are detailed as follows: Step S1: Acquire component structure images and sensor data of components to be inspected of a wind turbine.

[0030] Step S1 of the embodiment of the present application mainly involves remotely collecting data related to the structure of the core components and vulnerable parts of the wind turbine through various sensors and drones, so as to obtain a comprehensive internal and external structural condition of the wind turbine, providing sufficient data support for subsequent detection of component abnormalities.

[0031] The external structure of a wind turbine primarily includes components directly exposed to the environment and subject to external influences such as wind loads and climate shocks. These components include blades (for capturing wind energy), towers (for supporting the turbine nacelle and blades and increasing wind-capturing height), and hubs (for connecting the blades to the main shaft). These components can be captured from multiple angles by drones, generating images of their structure. These images can then be used to identify any abnormal changes in the component's structure, surface, or installation position, enabling fault detection within the turbine's external structure.

[0032] Among them, the structural abnormalities of the blades mainly include cracks and fractures caused by long-term use, lightning damage or manufacturing defects, and surface erosion caused by long-term erosion by sand, dust and raindrops; the structural abnormalities of the tower mainly include tower tilting due to strong wind loads or insufficient bolt pre-tightening force, and weld and bolt breakage caused by installation errors or corrosion.

[0033] The fault conditions of these external components of the wind turbine can be observed through images, but drones cannot directly photograph the components inside the wind turbine cabin. Therefore, it is necessary to use sensors to collect multi-source data on the internal structure of the wind turbine to accurately determine what faults have occurred in the internal structure of the wind turbine without manually disassembling the components.

[0034] In the embodiment of the present application, multi-source sensor data including temperature change data, vibration data and noise data are collected by installing temperature sensors, vibration sensors and noise sensors inside the fan, so as to locate the fault position by subsequently detecting whether the temperature of the component is too high and whether the component emits abnormal vibration and noise during operation.

[0035] The internal structure of the wind turbine, which performs energy conversion and control, is concentrated within the nacelle. Its main components include the gearbox (primarily used to control the blade speed), the generator, the control system (used to adjust the blade angle), and the cooling system (including the radiator and refrigeration unit to prevent overheating of the gearbox and generator). These components automatically generate heat during operation. Failure to dissipate heat in a timely manner can lead to overheating of the components, resulting in wind turbine failures such as insulation aging, permanent magnet demagnetization, and mechanical deformation. Furthermore, mechanical failures in the wind turbine can generate abnormal noise during operation, and the vibration frequency of the components can also change. Therefore, collecting vibration and noise data can improve the accuracy of wind turbine fault diagnosis and significantly reduce the losses caused by unplanned downtime due to manual inspections.

[0036] Among them, gearbox failures mainly include pitting and broken teeth caused by poor lubrication or overload; generator failures mainly include generator short circuit caused by insulation aging, reduced power generation efficiency due to high temperature and demagnetization of permanent magnets; control system failures include bearing sticking due to excessive bearing temperature and shortened capacitor life due to long-term use of capacitors.

[0037] Optionally, when the remote monitoring system detects abnormalities in certain electrical parameters of the wind turbine operating data, such as voltage sag / surge, current overload, large power fluctuations, and decreased insulation resistance, it will use drones and sensors to collect data on vulnerable and important components in the wind turbine to obtain component structure images and sensor data.

[0038] Step S2: detecting the external structural changes of the component to be detected based on the component structural image and the historical normal structural image, and obtaining the external abnormal features of the component to be detected.

[0039] Step S2 of the embodiment of the present application mainly detects abnormal conditions on the outside of the fan for components exposed to the external environment by taking real-time structural images of the components to be detected in the fan, further determines what faults have occurred on the outside of the components to be detected, obtains external abnormal characteristics of the components to be detected, and realizes comprehensive detection of external abnormalities of the components.

[0040] First, a historical image of the component under inspection, taken during the previous inspection period, is obtained. This image is taken during the last inspection, showing the component in its normal state. The component image is then compared with a historical image taken from the same angle, specifically looking for cracks, surface textures, foreign matter, movement in the installation position, and component wear, yielding detailed comparison results.

[0041] Based on the comparison results, the abnormal areas in the component structure image that are significantly different from the historical image are marked and cropped, and the cropped abnormal areas are used as the image to be identified to further identify the fault type of the component.

[0042] Based on a pre-defined component anomaly type library, the image to be identified is matched with data in the database, predicting one or more potential fault types for each image. The library, constructed based on historical wind turbine failure records, covers most common fault conditions for wind turbine components and stores numerous images of each type of wind turbine component structural damage, enabling accurate data matching.

[0043] Specifically, each image to be identified will be labeled to indicate which part of the wind turbine component it was taken on. Based on the position of the marked component, the corresponding image is found and matched in the component abnormality type library. When the confidence level of the matching result reaches the set threshold, it means that a fault image that is very similar to the image to be identified has been found, and based on the fault information of the fault image, the potential fault type of the image to be identified is determined.

[0044] For example, for an image to be identified of the leading edge of a blade, there is a small area on the blade whose color is different from the surrounding area. After searching the component abnormality type library, it is determined that there are multiple potential fault types for the image to be identified, including blade icing, coating detachment, or pits.

[0045] If an image to be identified has only one potential fault type, that potential fault type can be directly used as the fault type of the component to be inspected. However, if there are multiple potential fault types, it is necessary to obtain weight increase and decrease data of the component to be inspected and further perform more detailed feature extraction on the image to identify the most likely potential fault type.

[0046] The weight change data can be used to determine if the component has foreign matter or material loss. If the component weight does not change significantly, such as if the component has deformed, visual features related to the surface texture and color distribution of the component will be extracted from the image to further determine if the component has any abnormalities such as cracks, rust, or composite material adhesion failure.

[0047] Combining the weight increase and decrease data, surface texture and color distribution, the confidence level of each potential fault type is calculated, and the potential fault type with the highest confidence level is selected as the fault type of the component to be tested.

[0048] Furthermore, for components such as towers, it is also possible to detect whether the components are vibrating or tilting, and to determine whether the degree of tilt of the components has increased compared to the previous period by using surrounding reference objects in the image, thereby determining whether the components are faulty.

[0049] Step S3, based on the temperature change data and the material type and structural dimensions of the component to be detected, the internal structural changes of the component to be detected are simulated by the noise data and the vibration data to obtain the internal abnormal characteristics of the component to be detected.

[0050] In order to accurately obtain the internal structure of the power generation wind turbine without shutting down the machine, step S3 of the embodiment of the present application sets up multiple sensors to collect multi-source data from aspects such as temperature, vibration and noise to detect the noise characteristics and vibration characteristics of the components to be detected in the current period, and then detect the internal structural changes of the components to be detected, and obtain internal abnormal characteristics that can reflect the internal structural changes of the components to be detected and assist in determining the fault location, so as to realize efficient internal inspection of the wind turbine without implementing unplanned shutdown, thereby saving a lot of inspection costs.

[0051] First, as a control group, the noise and vibration characteristics of the components to be inspected under normal working conditions are obtained. Then, the corresponding abnormal noise signals and abnormal vibration signals are extracted from the noise data and vibration data collected in the current period, and the normal data are compared with the abnormal data to roughly identify whether there are any abnormal conditions inside the components to be inspected in the current period.

[0052] The noise generated by fan components during operation includes mechanical noise, aerodynamic noise, electromagnetic noise, and structural noise. Mechanical noise includes gearbox noise, generator noise, and hydraulic system noise, which are generated by gear meshing, cooling fan rotation, and fluid pulsation when hydraulic pumps and valves are working. They are the main noise sources during fan operation. The main source of aerodynamic noise is the noise generated when air flows through the blades, such as the hissing noise caused by turbulence when air flows over the blade surface. The spectrum of aerodynamic noise is usually low-medium frequency with obvious noise peaks. Electromagnetic noise is caused by unbalanced electromagnetic force causing electromagnetic vibration of the generator and radiation through solid structures. Structural noise is the noise generated by the vibration of structures such as the nacelle and tower under power transmission and external excitation, and is transmitted through the structure.

[0053] Factors influencing fan noise include wind speed, blade design, and operating status. For example, mechanical noise is more pronounced when components are old or poorly maintained. Gears can produce loud metallic clattering noises due to poor meshing, tooth spalling, or broken teeth. In hydraulic systems, air entrainment or leaks can cause bubbling noises in the pipelines. Therefore, analyzing collected noise data can pinpoint the source of the noise, providing support for subsequent fault location. It can also comprehensively determine the cause of component failures by combining vibration, temperature, and other data.

[0054] During wind turbine operation, components generate complex vibration signals due to mechanical motion, airflow disturbances, and other factors. Therefore, these signals can be used to diagnose wind turbine component faults. For example, when a gearbox tooth breaks, the vibration signal generated by the meshing gears can experience a sudden increase in amplitude and broadening of the sidebands. When the generator rotor is eccentric, the amplitude of the vibration signal changes significantly, potentially accompanied by electrical frequency modulation.

[0055] Based on the above description, the noise data and vibration data are compared with the noise characteristics and vibration characteristics of the component to be tested under normal working conditions in terms of signal amplitude, frequency, bandwidth, etc., and signal data that are significantly different from the normal characteristics are extracted as abnormal noise signals and abnormal vibration signals.

[0056] Since the generation of vibration signals and noise signals are both related to mechanical vibration and aerodynamics, there is a certain coupling relationship between the two but they are different. Therefore, by combining abnormal noise signals and abnormal vibration signals for analysis, the accuracy and reliability of fault detection can be improved, and the precise internal abnormal characteristics of the components to be detected can be obtained.

[0057] In this embodiment, the abnormal noise signal is first denoised to remove environmental interference, the trend term of the abnormal vibration signal is eliminated, and the effective frequency band is extracted to complete signal preprocessing. Then, based on the coupling relationship between the two, the cross-correlation coefficient and the common peak value of the preprocessed abnormal noise signal and abnormal vibration signal are calculated.

[0058] For example, when a bearing is damaged, noise impacts and vibration peaks occur simultaneously. At this point, the noise and vibration signals generated by the bearing will have a certain correlation, which can be used to calculate the signal correlation coefficient. When a gear tooth breaks, the vibration and noise signals generated during gear meshing will peak at the same time, and this peak can be used as the shared peak.

[0059] Abnormal noise signals and abnormal vibration signals are decomposed into wavelets of different frequency bands through wavelet transform, and the correlation of signals in time domain and frequency domain is analyzed to obtain relevant features.

[0060] In the time domain, different groups of wavelets are translated or scaled to capture the local characteristics of the signal along the time axis. The energy of the wavelet coefficients within each time window is calculated to reflect the local strength of the signal and collect the local characteristics of the signal along the time axis. Because the impact of component failure will manifest itself in the wavelet coefficients of specific frequency bands in the vibration and noise signals, the signal is decomposed at multiple scales to obtain wavelet coefficients for each frequency band from low to high. From these, the time domain mutation points are identified, allowing for more precise determination of the time domain correlation between abnormal noise and abnormal vibration signals.

[0061] In the frequency domain, time-frequency analysis is performed by calculating the energy proportion of each group of wavelets. The frequency band with concentrated energy in each group of wavelets is found as the dominant frequency band. The correlation between abnormal noise signals and abnormal vibration signals in the frequency domain is determined through the dominant frequency band.

[0062] Then, the correlation in the time domain and the correlation in the frequency domain are combined to construct a correlation feature.

[0063] Optionally, the number of decomposition layers of the wavelet transform is determined according to the highest frequency and sampling rate of the abnormal noise signal and the abnormal vibration signal.

[0064] By analyzing the correlation of signals in the time domain, transient events can be located and the time period of the fault can be accurately pinpointed, providing support for subsequent fault location based on changes in weather and operating conditions. By analyzing the correlation of signals in the frequency domain, periodic shocks can be detected, providing a basis for determining the type of fault.

[0065] The cross-correlation coefficient, the common peak value and the correlation feature are combined to construct a mapping relationship between the abnormal noise signal and the abnormal vibration signal. Then, data fusion of the abnormal noise signal and the abnormal vibration signal is performed through the mapping relationship to obtain abnormal change features.

[0066] Next, aiming for the highest similarity with the abnormal change signature, the internal structural change simulation is performed based on the normal structural dimensions of the component to be inspected in the previous period. Simultaneously, the similarity between the noise and vibration signals emitted by the simulation results and the abnormal change signature is calculated during the internal structural change process. The simulation is terminated when the similarity reaches a maximum value, completing the simulation of the actual change of the component to be inspected. The simulation result corresponding to the maximum value is then compared with the normal structural dimensions, and the corresponding internal abnormal signature is obtained by acquiring the internal structural changes of the component to be inspected.

[0067] Taking into account the different material types and structural shapes of each component to be inspected, and the different degrees of friction generated by the connection structures between the components and other connected components, the characteristics of the component to be inspected are also evaluated based on the temperature and the connection structure, so as to more accurately simulate the internal structural changes of the component to be inspected under the influence of the environment.

[0068] The effects of temperature changes on components made of different materials vary significantly, depending on the physical properties of the material (such as thermal expansion coefficient, thermal conductivity, and temperature resistance limit). The material types of the components to be tested in wind turbines are mainly divided into metals (such as gearbox housings and towers), composite materials (such as blades), polymer materials (such as sealing layers of components and cable insulation layers), ceramic and permanent magnetic materials (such as permanent magnets and ceramic insulators), lubricating materials (such as grease), and electronic materials (such as circuit boards). The embodiment of the present application simulates the degree of influence of temperature on each material type to further obtain the different reactions of each component to be tested to temperature changes, and obtains the first structural characteristics of each component to be tested related to temperature.

[0069] For example, for both gearbox housings, within the same temperature range, aluminum alloy products are more easily deformed than iron products. Therefore, the impact of temperature on the aluminum alloy gearbox housing is greater, and the internal structural deformation of the aluminum alloy gearbox housing is more obvious.

[0070] In addition to temperature, the difference in connection structure or method between the component to be tested and other components will also cause different degrees of damage to the component to be tested. For example, for both bearing connections, the friction generated by sliding bearings is smaller than that of rolling bearings; for both gear meshing, the friction generated by helical gears may be higher than that of straight teeth due to their large contact area. Considering that the friction generated by the same material type also results in different degrees of deformation of each component due to the different connection structures or methods between the components, the embodiment of the present application also simulates the degree of damage to each component to be tested due to the friction generated during operation based on the connection structure or method, and obtains the second structural characteristics of each component to be tested.

[0071] Combining the first structural feature and the second structural characteristic, based on the normal structural dimensions of the component to be detected in the previous period and based on the fan operating conditions in the current period, the internal structural changes of the component to be detected are continuously simulated to obtain the internal abnormal characteristics of the component to be detected.

[0072] Step S4: Based on the external abnormality characteristics and the internal abnormality characteristics, combined with the meteorological changes in the current period and the wind turbine operation data, the cause of the failure of the component to be detected is analyzed to obtain a corresponding fault diagnosis result.

[0073] Because external and internal abnormal characteristics contain a large amount of fault information of the components to be detected, in order to further determine the cause of the fault and assist maintenance personnel in quickly performing repairs and implementing better protective measures, it is also necessary to combine the meteorological changes in the current period and the wind turbine operation data to deduce the cause of the fault and obtain comprehensive and accurate fault diagnosis results.

[0074] The service life and operating condition change data of each component to be tested are extracted from the wind turbine operation data. Then, based on the external and internal abnormal characteristics, corresponding damage association labels are constructed for the above data and meteorological changes, respectively, to associate the abnormal conditions of the components with the corresponding fault causes.

[0075] Specifically, based on historical wind turbine inspection records, we analyze whether there is a correlation between component age and component damage frequency, quantify the relationship between component age and the degree of damage impact, and construct a signature of the impact of age. For example, as operating time increases, blades are more likely to crack due to accumulated mechanical fatigue. When a blade is over 10 years old, its age-related impact signature is more pronounced than when it was first put into operation, with more associated damage labels.

[0076] Based on the sampling frequency and component load variation curve, the degree to which the operating condition variation data affects damage to the component being tested is quantified. This is because the component load variation curve of a wind turbine is closely related to the turbine's operating conditions and can reflect whether the turbine is operating under load within a certain period of time. Large variations in the component load variation curve indicate a high probability of fatigue damage to the component. Extreme conditions in the component load variation curve indicate a high probability of reduced structural safety due to extreme operating conditions. The sampling frequency affects the accuracy of the impact characteristics of the operating condition variation data; the higher the sampling frequency, the more accurate the resulting impact characteristics.

[0077] Then, based on the long-term and short-term characteristics of the weather changes, the impact characteristics of the weather changes are constructed. This is because weather changes can affect the switching of wind turbine operating modes. For example, in high wind speeds, to prevent damage to blades, towers, and other structures due to overload, blade operation may be suspended through measures such as feathering, mechanical braking, and yaw to avoid wind. In sandstorms, derating is required to protect the motor and inverter.

[0078] Furthermore, to improve the accuracy of the impact characteristics of meteorological changes, the embodiments of the present application also divide the impact of meteorological changes into long-term characteristics and short-term characteristics. Short-term characteristics include wind speed turbulence intensity, temperature gradient, and lightning density, while long-term characteristics include the average annual number of freezing rain days and sandstorm frequency. By combining long-term and short-term characteristics, the constructed impact characteristics of meteorological changes can reflect the impact of extreme weather and seasonal changes on the components to be tested.

[0079] In addition, in order to further improve data accuracy, the embodiment of the present application also verifies the meteorological data according to the timestamp and longitude and latitude of the wind turbine to reduce data latency.

[0080] Finally, taking the external and internal abnormal characteristics of the components to be detected as the results, the damage association label of each influencing feature is obtained by searching for the failure events caused by each influencing feature in the existing historical fault records of the wind turbine.

[0081] Then, based on the existing external and internal abnormal features containing the fault conditions of the components to be detected, a causal graph related to each damage association label and the cause of the failure is constructed to facilitate the reverse deduction of the cause of the failure of the components to be detected.

[0082] First, we need to clarify the causal relationship. Based on historical fault records, we use external and internal abnormality characteristics as outcome variables, wind turbine operating data as intermediate response variables, and damage-related labels as potential cause variables. Wind turbine operating data, such as speed, power, current, and voltage, serve as intermediate response variables because they are the result of a fault. The damage-related labels, as factors that may cause the component under test to fail, serve as potential cause variables and provide the basis for subsequent fault cause screening.

[0083] Then, through conditional independence testing, based on the relationships that the potential cause variable causes the change of the intermediate response variable and the intermediate response variable is a representation of the outcome variable, a causal graph skeleton is constructed between the damage association label and the external abnormal features and internal abnormal features.

[0084] For example, a conditional independence test is used to test whether wind speed changes can significantly predict blade vibration changes. A pre-set model is used to fit the effect of wind speed on vibration while controlling for confounding variables such as temperature to construct a directed path between wind speed and vibration. To improve the accuracy of the test results, it is also necessary to eliminate interference from other variables during the test.

[0085] Optionally, the embodiment of the present application adopts the Fisher-Z test method.

[0086] Based on the causal effect of the potential cause variable on the intermediate response variable, an impact weight is generated for each damage-related label. The causal effect describes the degree of influence of the potential cause variable on the intermediate response variable. For example, for every 1 m / s increase in wind speed, the RMS vibration value will increase by 0.2 g. The impact weight indicates the likelihood that the damage-related label is the cause of the component failure.

[0087] The nodes in the causal graph skeleton are then labeled according to the impact weights. A causal forest is constructed using the labeled causal graph skeleton, using the outcome variable as the endpoint, to distinguish between direct and indirect causes. This yields a causal graph of the fault causes associated with both external and internal abnormality characteristics. This causal graph comprehensively displays the various fault causes of the component under inspection and the coupling relationships between them. Distinguishing between direct and indirect causes is achieved based on the impact weights.

[0088] For example, if a generator leaks electricity, the direct cause is the aging of the insulation layer, and the indirect cause is the failure of the insulation layer due to long-term high-temperature operation. Therefore, in the causal diagram, the direct and indirect causes of the generator failure are distinguished by the length of the path, which facilitates the subsequent generation of more comprehensive fault diagnosis results.

[0089] Finally, the direct and indirect causes of the causal diagram are analyzed through the path verification method to obtain the fault cause of the component to be tested. The fault location and fault type are determined through external and internal abnormal characteristics, and accurate and comprehensive fault diagnosis results are obtained, which provide support for the timely implementation of appropriate maintenance measures and can further improve the operational stability of the power system.

[0090] The implementation of the embodiments of the present application has the following beneficial effects: The embodiment of the present application detects whether there are any abnormalities in the external structure and internal structure of the components to be detected in the wind turbine, and compares the captured component structure image with the historical normal structure image to determine whether the external structure of the component to be detected has changed, thereby obtaining external abnormality characteristics. Then, based on the vibration and noise emitted by the component during operation, it is determined whether the component has undergone internal structural changes, and based on the fact that the components to be detected will have different reactions to temperature changes due to different material types, the degree of internal structural changes is quantified, so that the internal structural abnormalities of the wind turbine components can be judged only by the data collected by the sensor, without suspending the operation of the wind turbine, thus saving a lot of inspection costs and improving inspection efficiency, and obtaining accurate internal abnormality characteristics. Finally, the cause of the fault is analyzed based on the impact of meteorological changes and wind turbine operation data on the damage of the components to be detected, and the fault causes corresponding to the external abnormality characteristics and the internal abnormality characteristics are obtained, thereby obtaining comprehensive and accurate fault diagnosis results, and providing support for subsequent maintenance personnel to implement appropriate maintenance measures.

[0091] Furthermore, in order to implement the dynamic response monitoring system of a large-scale wind turbine corresponding to the above method embodiment to achieve the corresponding functions and technical effects, Figure 2 A structural diagram of a dynamic response monitoring system for a large-scale wind turbine is provided. For ease of illustration, only the parts relevant to this embodiment are shown. The dynamic response monitoring system for a large-scale wind turbine provided in this embodiment of the application includes: The data acquisition module 201 is used to acquire component structure images and sensor data of the components to be inspected of the wind turbine; wherein the sensor data includes temperature change data, vibration data and noise data.

[0092] The external feature acquisition module 202 is used to detect the external structural changes of the component to be detected based on the component structure image and the historical normal structure image, and obtain the external abnormal features of the component to be detected.

[0093] In an embodiment of the present application, a historical normal structural image of the component to be inspected within a previous period of time is obtained, the historical normal structural image is compared with the component structural image, and the abnormal area in the component structural image is cropped based on the comparison result to obtain an image to be identified; Predicting the potential fault type of the image to be identified based on a preset component abnormality type library; Screening the potential fault types by using the image to be identified and the collected weight increase and decrease data of the component to be inspected, and determining the fault type of each component to be inspected; According to the fault type and the image to be identified, an external abnormality feature of the component to be detected is constructed.

[0094] The internal feature acquisition module 203 is used to simulate the internal structural changes of the component to be detected through the noise data and the vibration data according to the temperature change data and the material type and structural dimensions of the component to be detected, so as to obtain the internal abnormal features of the component to be detected.

[0095] In the embodiment of the present application, according to the noise characteristics and vibration characteristics of the component to be inspected under normal working conditions, an abnormal noise signal and an abnormal vibration signal are respectively extracted from the noise data and the vibration data; Align and fuse abnormal noise signals and abnormal vibration signals to obtain abnormal change features; With the goal of achieving the highest similarity with the abnormal change characteristics, the internal structural changes of the component to be detected are simulated through the temperature change data and the material type based on the normal structural dimensions of the component to be detected in the previous period to obtain the internal abnormal characteristics of the component to be detected.

[0096] The fault cause analysis module 204 is used to analyze the fault cause of the component to be detected based on the external abnormal characteristics and the internal abnormal characteristics, combined with the meteorological changes in the current period and the wind turbine operation data, to obtain a corresponding fault diagnosis result.

[0097] In the embodiment of the present application, the influence of temperature on each material type is simulated according to the temperature change data to obtain the first structural characteristics of each component to be inspected; According to the connection structure between the component to be inspected and other components, the degree of damage to the component to be inspected caused by friction generated by the material type during operation of the component to be inspected is simulated to obtain a second structural characteristic of each component to be inspected; By using the first structural characteristic and the second structural characteristic, the internal structural changes of the component to be inspected are continuously simulated on the basis of the normal structural dimensions, and the similarity between the noise signal and the vibration signal generated by the component to be inspected and the abnormal change characteristics under each internal structural change is calculated; When the similarity is the highest, the simulation operation is stopped, and the current simulation result is compared with the normal structure size to construct the internal abnormality feature.

[0098] In some embodiments, the data acquisition module 201 is specifically: The external structure of a wind turbine primarily includes components directly exposed to the environment and subject to external influences such as wind loads and climate shocks. These components include blades (for capturing wind energy), towers (for supporting the turbine nacelle and blades and increasing wind-capturing height), and hubs (for connecting the blades to the main shaft). These components can be captured from multiple angles by drones, generating images of their structure. These images can then be used to identify any abnormal changes in the component's structure, surface, or installation position, enabling fault detection within the turbine's external structure.

[0099] Among them, the structural abnormalities of the blades mainly include cracks and fractures caused by long-term use, lightning damage or manufacturing defects, and surface erosion caused by long-term erosion by sand, dust and raindrops; the structural abnormalities of the tower mainly include tower tilting due to strong wind loads or insufficient bolt pre-tightening force, and weld and bolt breakage caused by installation errors or corrosion.

[0100] The fault conditions of these external components of the wind turbine can be observed through images, but drones cannot directly photograph the components inside the wind turbine cabin. Therefore, it is necessary to use sensors to collect multi-source data on the internal structure of the wind turbine to accurately determine what faults have occurred in the internal structure of the wind turbine without manually disassembling the components.

[0101] In the embodiment of the present application, multi-source sensor data including temperature change data, vibration data and noise data are collected by installing temperature sensors, vibration sensors and noise sensors inside the fan, so as to locate the fault position by subsequently detecting whether the temperature of the component is too high and whether the component emits abnormal vibration and noise during operation.

[0102] The internal structure of the wind turbine, which performs energy conversion and control, is concentrated within the nacelle. Its main components include the gearbox (primarily used to control the blade speed), the generator, the control system (used to adjust the blade angle), and the cooling system (including the radiator and refrigeration unit to prevent overheating of the gearbox and generator). These components automatically generate heat during operation. Failure to dissipate heat in a timely manner can lead to overheating of the components, resulting in wind turbine failures such as insulation aging, permanent magnet demagnetization, and mechanical deformation. Furthermore, mechanical failures in the wind turbine can generate abnormal noise during operation, and the vibration frequency of the components can also change. Therefore, collecting vibration and noise data can improve the accuracy of wind turbine fault diagnosis and significantly reduce the losses caused by unplanned downtime due to manual inspections.

[0103] Among them, gearbox failures mainly include pitting and broken teeth caused by poor lubrication or overload; generator failures mainly include generator short circuit caused by insulation aging, reduced power generation efficiency due to high temperature and demagnetization of permanent magnets; control system failures include bearing sticking due to excessive bearing temperature and shortened capacitor life due to long-term use of capacitors.

[0104] Optionally, when the remote monitoring system detects abnormalities in certain electrical parameters of the wind turbine operating data, such as voltage sag / surge, current overload, large power fluctuations, and decreased insulation resistance, it will use drones and sensors to collect data on vulnerable and important components in the wind turbine to obtain component structure images and sensor data.

[0105] In some embodiments, the external feature acquisition module 202 is specifically: First, a historical image of the component under inspection, taken during the previous inspection period, is obtained. This image is taken during the last inspection, showing the component in its normal state. The component image is then compared with a historical image taken from the same angle, specifically looking for cracks, surface textures, foreign matter, movement in the installation position, and component wear, yielding detailed comparison results.

[0106] Based on the comparison results, the abnormal areas in the component structure image that are significantly different from the historical image are marked and cropped, and the cropped abnormal areas are used as the image to be identified to further identify the fault type of the component.

[0107] Based on a pre-defined component anomaly type library, the image to be identified is matched with data in the database, predicting one or more potential fault types for each image. The library, constructed based on historical wind turbine failure records, covers most common fault conditions for wind turbine components and stores numerous images of each type of wind turbine component structural damage, enabling accurate data matching.

[0108] Specifically, each image to be identified will be labeled to indicate which part of the wind turbine component it was taken on. Based on the position of the marked component, the corresponding image is found and matched in the component abnormality type library. When the confidence level of the matching result reaches the set threshold, it means that a fault image that is very similar to the image to be identified has been found, and based on the fault information of the fault image, the potential fault type of the image to be identified is determined.

[0109] For example, for an image to be identified of the leading edge of a blade, there is a small area on the blade whose color is different from the surrounding area. After searching the component abnormality type library, it is determined that there are multiple potential fault types for the image to be identified, including blade icing, coating detachment, or pits.

[0110] If an image to be identified has only one potential fault type, that potential fault type can be directly used as the fault type of the component to be inspected. However, if there are multiple potential fault types, it is necessary to obtain weight increase and decrease data of the component to be inspected and further perform more detailed feature extraction on the image to identify the most likely potential fault type.

[0111] The weight change data can be used to determine if the component has foreign matter or material loss. If the component weight does not change significantly, such as if the component has deformed, visual features related to the surface texture and color distribution of the component will be extracted from the image to further determine if the component has any abnormalities such as cracks, rust, or composite material adhesion failure.

[0112] Combining the weight increase and decrease data, surface texture and color distribution, the confidence level of each potential fault type is calculated, and the potential fault type with the highest confidence level is selected as the fault type of the component to be tested.

[0113] Furthermore, for components such as towers, it is also possible to detect whether the components are vibrating or tilting, and to determine whether the degree of tilt of the components has increased compared to the previous period by using surrounding reference objects in the image, thereby determining whether the components are faulty.

[0114] In some embodiments, the internal feature acquisition module 203 is specifically: First, as a control group, the noise and vibration characteristics of the components to be inspected under normal working conditions are obtained. Then, the corresponding abnormal noise signals and abnormal vibration signals are extracted from the noise data and vibration data collected in the current period, and the normal data are compared with the abnormal data to roughly identify whether there are any abnormal conditions inside the components to be inspected in the current period.

[0115] The noise generated by fan components during operation includes mechanical noise, aerodynamic noise, electromagnetic noise, and structural noise. Mechanical noise includes gearbox noise, generator noise, and hydraulic system noise, which are generated by gear meshing, cooling fan rotation, and fluid pulsation when hydraulic pumps and valves are working. They are the main noise sources during fan operation. The main source of aerodynamic noise is the noise generated when air flows through the blades, such as the hissing noise caused by turbulence when air flows over the blade surface. The spectrum of aerodynamic noise is usually low-medium frequency with obvious noise peaks. Electromagnetic noise is caused by unbalanced electromagnetic force causing electromagnetic vibration of the generator and radiation through solid structures. Structural noise is the noise generated by the vibration of structures such as the nacelle and tower under power transmission and external excitation, and is transmitted through the structure.

[0116] Factors influencing fan noise include wind speed, blade design, and operating status. For example, mechanical noise is more pronounced when components are old or poorly maintained. Gears can produce loud metallic clattering noises due to poor meshing, tooth spalling, or broken teeth. In hydraulic systems, air entrainment or leaks can cause bubbling noises in the pipelines. Therefore, analyzing collected noise data can pinpoint the source of the noise, providing support for subsequent fault location. It can also comprehensively determine the cause of component failures by combining vibration, temperature, and other data.

[0117] During wind turbine operation, components generate complex vibration signals due to mechanical motion, airflow disturbances, and other factors. Therefore, these signals can be used to diagnose wind turbine component faults. For example, when a gearbox tooth breaks, the vibration signal generated by the meshing gears can experience a sudden increase in amplitude and broadening of the sidebands. When the generator rotor is eccentric, the amplitude of the vibration signal changes significantly, potentially accompanied by electrical frequency modulation.

[0118] Based on the above description, the noise data and vibration data are compared with the noise characteristics and vibration characteristics of the component to be tested under normal working conditions in terms of signal amplitude, frequency, bandwidth, etc., and signal data that are significantly different from the normal characteristics are extracted as abnormal noise signals and abnormal vibration signals.

[0119] Since the generation of vibration signals and noise signals are both related to mechanical vibration and aerodynamics, there is a certain coupling relationship between the two but they are different. Therefore, by combining abnormal noise signals and abnormal vibration signals for analysis, the accuracy and reliability of fault detection can be improved, and the precise internal abnormal characteristics of the components to be detected can be obtained.

[0120] In this embodiment, the abnormal noise signal is first denoised to remove environmental interference, the trend term of the abnormal vibration signal is eliminated, and the effective frequency band is extracted to complete signal preprocessing. Then, based on the coupling relationship between the two, the cross-correlation coefficient and the common peak value of the preprocessed abnormal noise signal and abnormal vibration signal are calculated.

[0121] For example, when a bearing is damaged, noise impacts and vibration peaks occur simultaneously. At this point, the noise and vibration signals generated by the bearing will have a certain correlation, which can be used to calculate the signal correlation coefficient. When a gear tooth breaks, the vibration and noise signals generated during gear meshing will peak at the same time, and this peak can be used as the shared peak.

[0122] Abnormal noise signals and abnormal vibration signals are decomposed into wavelets of different frequency bands through wavelet transform, and the correlation of signals in time domain and frequency domain is analyzed to obtain relevant features.

[0123] In the time domain, different groups of wavelets are translated or scaled to capture the local characteristics of the signal along the time axis. The energy of the wavelet coefficients within each time window is calculated to reflect the local strength of the signal and collect the local characteristics of the signal along the time axis. Because the impact of component failure will manifest itself in the wavelet coefficients of specific frequency bands in the vibration and noise signals, the signal is decomposed at multiple scales to obtain wavelet coefficients for each frequency band from low to high. From these, the time domain mutation points are identified, allowing for more precise determination of the time domain correlation between abnormal noise and abnormal vibration signals.

[0124] In the frequency domain, time-frequency analysis is performed by calculating the energy proportion of each group of wavelets. The frequency band with concentrated energy in each group of wavelets is found as the dominant frequency band. The correlation between abnormal noise signals and abnormal vibration signals in the frequency domain is determined through the dominant frequency band.

[0125] Then, the correlation in the time domain and the correlation in the frequency domain are combined to construct a correlation feature.

[0126] Optionally, the number of decomposition layers of the wavelet transform is determined according to the highest frequency and sampling rate of the abnormal noise signal and the abnormal vibration signal.

[0127] By analyzing the correlation of signals in the time domain, transient events can be located and the time period of the fault can be accurately pinpointed, providing support for subsequent fault location based on changes in weather and operating conditions. By analyzing the correlation of signals in the frequency domain, periodic shocks can be detected, providing a basis for determining the type of fault.

[0128] The cross-correlation coefficient, the common peak value and the correlation feature are combined to construct a mapping relationship between the abnormal noise signal and the abnormal vibration signal. Then, data fusion of the abnormal noise signal and the abnormal vibration signal is performed through the mapping relationship to obtain abnormal change features.

[0129] Next, aiming for the highest similarity with the abnormal change signature, the internal structural change simulation is performed based on the normal structural dimensions of the component to be inspected in the previous period. Simultaneously, the similarity between the noise and vibration signals emitted by the simulation results and the abnormal change signature is calculated during the internal structural change process. The simulation is terminated when the similarity reaches a maximum value, completing the simulation of the actual change of the component to be inspected. The simulation result corresponding to the maximum value is then compared with the normal structural dimensions, and the corresponding internal abnormal signature is obtained by acquiring the internal structural changes of the component to be inspected.

[0130] Taking into account the different material types and structural shapes of each component to be inspected, and the different degrees of friction generated by the connection structures between the components and other connected components, the characteristics of the component to be inspected are also evaluated based on the temperature and the connection structure, so as to more accurately simulate the internal structural changes of the component to be inspected under the influence of the environment.

[0131] The effects of temperature changes on components made of different materials vary significantly, depending on the physical properties of the material (such as thermal expansion coefficient, thermal conductivity, and temperature resistance limit). The material types of the components to be tested in wind turbines are mainly divided into metals (such as gearbox housings and towers), composite materials (such as blades), polymer materials (such as sealing layers of components and cable insulation layers), ceramic and permanent magnetic materials (such as permanent magnets and ceramic insulators), lubricating materials (such as grease), and electronic materials (such as circuit boards). The embodiment of the present application simulates the degree of influence of temperature on each material type to further obtain the different reactions of each component to be tested to temperature changes, and obtains the first structural characteristics of each component to be tested related to temperature.

[0132] For example, for both gearbox housings, within the same temperature range, aluminum alloy products are more easily deformed than iron products. Therefore, the impact of temperature on the aluminum alloy gearbox housing is greater, and the internal structural deformation of the aluminum alloy gearbox housing is more obvious.

[0133] In addition to temperature, the difference in connection structure or method between the component to be tested and other components will also cause different degrees of damage to the component to be tested. For example, for both bearing connections, the friction generated by sliding bearings is smaller than that of rolling bearings; for both gear meshing, the friction generated by helical gears may be higher than that of straight teeth due to their large contact area. Considering that the friction generated by the same material type also results in different degrees of deformation of each component due to the different connection structures or methods between the components, the embodiment of the present application also simulates the degree of damage to each component to be tested due to the friction generated during operation based on the connection structure or method, and obtains the second structural characteristics of each component to be tested.

[0134] Combining the first structural feature and the second structural characteristic, based on the normal structural dimensions of the component to be detected in the previous period and based on the fan operating conditions in the current period, the internal structural changes of the component to be detected are continuously simulated to obtain the internal abnormal characteristics of the component to be detected.

[0135] In some embodiments, the fault cause analysis module 204 is specifically: Because external and internal abnormal characteristics contain a large amount of fault information of the components to be detected, in order to further determine the cause of the fault and assist maintenance personnel in quickly performing repairs and implementing better protective measures, it is also necessary to combine the meteorological changes in the current period and the wind turbine operation data to deduce the cause of the fault and obtain comprehensive and accurate fault diagnosis results.

[0136] The service life and operating condition change data of each component to be tested are extracted from the wind turbine operation data. Then, based on the external and internal abnormal characteristics, corresponding damage association labels are constructed for the above data and meteorological changes, respectively, to associate the abnormal conditions of the components with the corresponding fault causes.

[0137] Specifically, based on historical wind turbine inspection records, we analyze whether there is a correlation between component age and component damage frequency, quantify the relationship between component age and the degree of damage impact, and construct a signature of the impact of age. For example, as operating time increases, blades are more likely to crack due to accumulated mechanical fatigue. When a blade is over 10 years old, its age-related impact signature is more pronounced than when it was first put into operation, with more associated damage labels.

[0138] Based on the sampling frequency and component load variation curve, the degree to which the operating condition variation data affects damage to the component being tested is quantified. This is because the component load variation curve of a wind turbine is closely related to the turbine's operating conditions and can reflect whether the turbine is operating under load within a certain period of time. Large variations in the component load variation curve indicate a high probability of fatigue damage to the component. Extreme conditions in the component load variation curve indicate a high probability of reduced structural safety due to extreme operating conditions. The sampling frequency affects the accuracy of the impact characteristics of the operating condition variation data; the higher the sampling frequency, the more accurate the resulting impact characteristics.

[0139] Then, based on the long-term and short-term characteristics of the weather changes, the impact characteristics of the weather changes are constructed. This is because weather changes can affect the switching of wind turbine operating modes. For example, in high wind speeds, to prevent damage to blades, towers, and other structures due to overload, blade operation may be suspended through measures such as feathering, mechanical braking, and yaw to avoid wind. In sandstorms, derating is required to protect the motor and inverter.

[0140] Furthermore, to improve the accuracy of the impact characteristics of meteorological changes, the embodiments of the present application also divide the impact of meteorological changes into long-term characteristics and short-term characteristics. Short-term characteristics include wind speed turbulence intensity, temperature gradient, and lightning density, while long-term characteristics include the average annual number of freezing rain days and sandstorm frequency. By combining long-term and short-term characteristics, the constructed impact characteristics of meteorological changes can reflect the impact of extreme weather and seasonal changes on the components to be tested.

[0141] In addition, in order to further improve data accuracy, the embodiment of the present application also verifies the meteorological data according to the timestamp and longitude and latitude of the wind turbine to reduce data latency.

[0142] Finally, taking the external and internal abnormal characteristics of the components to be detected as the results, the damage association label of each influencing feature is obtained by searching for the failure events caused by each influencing feature in the existing historical fault records of the wind turbine.

[0143] Then, based on the existing external and internal abnormal features containing the fault conditions of the components to be detected, a causal graph related to each damage association label and the cause of the failure is constructed to facilitate the reverse deduction of the cause of the failure of the components to be detected.

[0144] First, we need to clarify the causal relationship. Based on historical fault records, we use external and internal abnormality characteristics as outcome variables, wind turbine operating data as intermediate response variables, and damage-related labels as potential cause variables. Wind turbine operating data, such as speed, power, current, and voltage, serve as intermediate response variables because they are the result of a fault. The damage-related labels, as factors that may cause the component under test to fail, serve as potential cause variables and provide the basis for subsequent fault cause screening.

[0145] Then, through conditional independence testing, based on the relationships that the potential cause variable causes the change of the intermediate response variable and the intermediate response variable is a representation of the outcome variable, a causal graph skeleton is constructed between the damage association label and the external abnormal features and internal abnormal features.

[0146] For example, a conditional independence test is used to test whether wind speed changes can significantly predict blade vibration changes. A pre-set model is used to fit the effect of wind speed on vibration while controlling for confounding variables such as temperature to construct a directed path between wind speed and vibration. To improve the accuracy of the test results, it is also necessary to eliminate interference from other variables during the test.

[0147] Optionally, the embodiment of the present application adopts the Fisher-Z test method.

[0148] Based on the causal effect of the potential cause variable on the intermediate response variable, an impact weight is generated for each damage-related label. The causal effect describes the degree of influence of the potential cause variable on the intermediate response variable. For example, for every 1 m / s increase in wind speed, the RMS vibration value will increase by 0.2 g. The impact weight indicates the likelihood that the damage-related label is the cause of the component failure.

[0149] The nodes in the causal graph skeleton are then labeled according to the impact weights. A causal forest is constructed using the labeled causal graph skeleton, using the outcome variable as the endpoint, to distinguish between direct and indirect causes. This yields a causal graph of the fault causes associated with both external and internal abnormality characteristics. This causal graph comprehensively displays the various fault causes of the component under inspection and the coupling relationships between them. Distinguishing between direct and indirect causes is achieved based on the impact weights.

[0150] For example, if a generator leaks electricity, the direct cause is the aging of the insulation layer, and the indirect cause is the failure of the insulation layer due to long-term high-temperature operation. Therefore, in the causal diagram, the direct and indirect causes of the generator failure are distinguished by the length of the path, which facilitates the subsequent generation of more comprehensive fault diagnosis results.

[0151] Finally, the direct and indirect causes of the causal diagram are analyzed through the path verification method to obtain the fault cause of the component to be tested. The fault location and fault type are determined through external and internal abnormal characteristics, and accurate and comprehensive fault diagnosis results are obtained, which provide support for the timely implementation of appropriate maintenance measures and can further improve the operational stability of the power system.

[0152] The implementation of the embodiments of the present application has the following beneficial effects: The embodiment of the present application detects whether there are any abnormalities in the external structure and internal structure of the components to be detected in the wind turbine, and compares the captured component structure image with the historical normal structure image to determine whether the external structure of the component to be detected has changed, thereby obtaining external abnormality characteristics. Then, based on the vibration and noise emitted by the component during operation, it is determined whether the component has undergone internal structural changes, and based on the fact that the components to be detected will have different reactions to temperature changes due to different material types, the degree of internal structural changes is quantified, so that the internal structural abnormalities of the wind turbine components can be judged only by the data collected by the sensor, without suspending the operation of the wind turbine, thus saving a lot of inspection costs and improving inspection efficiency, and obtaining accurate internal abnormality characteristics. Finally, the cause of the fault is analyzed based on the impact of meteorological changes and wind turbine operation data on the damage of the components to be detected, and the fault causes corresponding to the external abnormality characteristics and the internal abnormality characteristics are obtained, thereby obtaining comprehensive and accurate fault diagnosis results, and providing support for subsequent maintenance personnel to implement appropriate maintenance measures.

[0153] The specific embodiments described above further illustrate the purpose, technical solutions, and beneficial effects of this application. It should be understood that the above description is merely a specific embodiment of this application and is not intended to limit the scope of protection of this application. In particular, it should be noted that for those skilled in the art, any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of this application should be included in the scope of protection of this application.

Claims

1. A dynamic response monitoring method for a large-scale wind turbine, characterized in that: include: Acquire component structure images and sensor data of the components to be inspected of the wind turbine; wherein the sensor data includes temperature change data, vibration data, and noise data; Detecting changes in the external structure of the component to be detected based on the component structure image and historical normal structure images to obtain external abnormal features of the component to be detected; According to the temperature change data and the material type and structural dimensions of the component to be detected, the internal structural changes of the component to be detected are simulated by the noise data and the vibration data to obtain internal abnormal characteristics of the component to be detected; Based on the external abnormal characteristics and the internal abnormal characteristics, combined with the meteorological changes in the current period and the wind turbine operation data, the fault cause of the component to be detected is analyzed to obtain a corresponding fault diagnosis result.

2. The dynamic response monitoring method for a large-scale wind turbine according to claim 1, characterized in that: The external structural changes of the component to be detected are detected based on the component structural image and the historical normal structural image to obtain the external abnormal features of the component to be detected, specifically: Acquiring the historical normal structural image of the component to be inspected within a previous period, comparing the historical normal structural image with the component structural image, and cropping the abnormal area in the component structural image based on the comparison result to obtain an image to be identified; Predicting the potential fault type of the image to be identified based on a preset component abnormality type library; Screening the potential fault types by using the image to be identified and the collected weight increase and decrease data of the component to be inspected, and determining the fault type of each component to be inspected; According to the fault type and the image to be identified, an external abnormality feature of the component to be detected is constructed.

3. The dynamic response monitoring method for a large-scale wind turbine according to claim 2, characterized in that: The potential fault types are screened by using the image to be identified and the collected weight increase and decrease data of the component to be detected to determine the fault type of each component to be detected, specifically: If there is only one potential fault type for the image to be identified, the potential fault type is used as the fault type of the component to be detected; If there is more than one potential fault type in an image to be identified, the weight increase and decrease data of the component to be detected is obtained, and the surface texture and color distribution of the component to be detected are obtained from the image to be identified; the confidence of each potential fault type is calculated based on the weight increase and decrease data, the surface texture and color distribution, and the potential fault type with the highest confidence is selected as the fault type of the component to be detected.

4. The dynamic response monitoring method for a large-scale wind turbine according to claim 1, characterized in that: According to the temperature change data and the material type and structural dimensions of the component to be detected, the internal structural changes of the component to be detected are simulated by the noise data and the vibration data to obtain the internal abnormal characteristics of the component to be detected, specifically: extracting abnormal noise signals and abnormal vibration signals from the noise data and the vibration data, respectively, based on the noise characteristics and vibration characteristics of the component to be inspected under normal working conditions; Align and fuse abnormal noise signals and abnormal vibration signals to obtain abnormal change features; With the goal of achieving the highest similarity with the abnormal change characteristics, the internal structural changes of the component to be detected are simulated through the temperature change data and the material type based on the normal structural dimensions of the component to be detected in the previous period to obtain the internal abnormal characteristics of the component to be detected.

5. The dynamic response monitoring method for a large-scale wind turbine according to claim 4, characterized in that: The internal structural changes of the component to be detected are simulated based on the normal structural dimensions of the component to be detected in the previous period by using the temperature change data and the material type to obtain the internal abnormal characteristics of the component to be detected, specifically: Simulating the influence of temperature on each material type according to the temperature change data to obtain a first structural characteristic of each component to be inspected; According to the connection structure between the component to be inspected and other components, the degree of damage to the component to be inspected caused by friction generated by the material type during operation of the component to be inspected is simulated to obtain a second structural characteristic of each component to be inspected; By using the first structural characteristic and the second structural characteristic, the internal structural changes of the component to be inspected are continuously simulated on the basis of the normal structural dimensions, and the similarity between the noise signal and the vibration signal generated by the component to be inspected and the abnormal change characteristics under each internal structural change is calculated; When the similarity is the highest, the simulation operation is stopped, and the current simulation result is compared with the normal structure size to construct the internal abnormality feature.

6. The dynamic response monitoring method for a large-scale wind turbine according to claim 4, characterized in that: The abnormal noise signal and the abnormal vibration signal are aligned and fused to obtain the abnormal change characteristics, specifically: Calculate the cross-correlation coefficient and common peak value of abnormal noise signal and abnormal vibration signal; The correlation between abnormal noise signals and abnormal vibration signals in time domain and frequency domain is analyzed by wavelet transform to obtain relevant features; A mapping relationship between abnormal noise signals and abnormal vibration signals is constructed based on the mutual correlation coefficient, the common peak value and the correlation feature, and then data fusion is performed on the abnormal noise signals and the abnormal vibration signals through the mapping relationship to obtain abnormal change features.

7. The dynamic response monitoring method for a large-scale wind turbine according to claim 1, characterized in that: Based on the external abnormality characteristics and the internal abnormality characteristics, combined with the meteorological changes in the current period and the wind turbine operation data, the fault cause of the component to be detected is analyzed to obtain the corresponding fault diagnosis result, which is specifically: Extracting service life and working condition change data of each component to be tested from the wind turbine operation data; According to the external abnormality characteristics and the internal abnormality characteristics, corresponding damage association labels are respectively constructed for the service life, the operating condition change data, and the meteorological change situation; Using the external abnormality features and the internal abnormality features as result variables, causal reasoning is performed on the fault cause of the component to be detected to obtain an influence weight of each damage association label and a causal graph related to the fault cause; The direct and indirect causes of the causal diagram are analyzed by using a path verification method and the influence weights to obtain the fault diagnosis result; wherein the fault diagnosis result includes the fault cause, fault location and fault type of the component to be detected.

8. The dynamic response monitoring method for a large-scale wind turbine according to claim 7, characterized in that: The damage association labels corresponding to the service life, the operating condition change data, and the meteorological change conditions are constructed according to the external abnormality characteristics and the internal abnormality characteristics, specifically: Based on historical inspection records, quantify the degree of damage caused by the service life of the component to be inspected, and construct an impact feature of the service life; quantifying the degree of damage influence of the operating condition change data on the component to be inspected based on the sampling frequency and the component load change curve, and constructing the influence characteristics of the operating condition change data; Constructing the impact characteristics of the meteorological change situation based on the long-term characteristics and short-term characteristics of the meteorological change situation; The external abnormality feature and the internal abnormality feature are used as results, and the damage association label is generated for each of the influencing features by searching for the fault cause in the historical fault records of the wind turbine.

9. The dynamic response monitoring method for a large-scale wind turbine according to claim 7, characterized in that: The external abnormality feature and the internal abnormality feature are used as result variables to perform causal reasoning on the fault cause of the component to be detected, and the influence weight of each damage association label and the causal graph related to the fault cause are obtained, specifically: Based on historical fault records, the external abnormality characteristics and the internal abnormality characteristics are used as result variables, the wind turbine operation data is used as an intermediate response variable, and the damage association label is used as a potential cause variable; By means of a conditional independence test, a causal diagram skeleton is constructed based on the relationship between the potential cause variable and the intermediate response variable and the relationship between the outcome variable and the intermediate response variable; generating an impact weight of each of the damage-related labels according to the causal effect of the potential cause variable on the intermediate response variable; The causal graph skeleton is annotated according to the influence weight, and a causal forest is constructed on the annotated causal graph skeleton with the result variable as an end point to obtain the causal graph.

10. A dynamic response monitoring system for a large-scale wind turbine, characterized in that: include: Data acquisition module, external feature acquisition module, internal feature acquisition module and fault cause analysis module; The data acquisition module is used to acquire component structure images and sensor data of the components to be inspected of the wind turbine; wherein the sensor data includes temperature change data, vibration data and noise data; The external feature acquisition module is used to detect the external structural changes of the component to be detected based on the component structure image and the historical normal structure image, and obtain the external abnormal features of the component to be detected; The internal feature acquisition module is used to simulate the internal structural changes of the component to be detected based on the temperature change data and the material type and structural dimensions of the component to be detected through the noise data and the vibration data to obtain the internal abnormality features of the component to be detected; The fault cause analysis module is used to analyze the fault cause of the component to be detected based on the external abnormal characteristics and the internal abnormal characteristics, combined with the meteorological changes in the current period and the fan operation data, to obtain the corresponding fault diagnosis result.

Images