An energy big data governance method and system based on digital twinning

Abstract

The application discloses an energy big data management method and system based on digital twinning, and relates to the field of energy big data management; the method comprises the following steps: collecting vibration data in real time, decomposing the vibration data into multiple frequency scale components, removing background noise signals to obtain multiple denoising frequency scale components, searching for a scale component where a main bearing wear vibration frequency is located, extracting a wear vibration energy feature of the wear vibration frequency from the scale component and calculating a confidence degree of the wear vibration energy feature, performing enhancement processing on the wear vibration energy feature when the confidence degree exceeds a main bearing wear degradation signal threshold, fusing the multiple denoising frequency scale components and the processed wear enhancement feature signal into refined data and transmitting the refined data to a digital twinning platform, and enabling the digital twinning platform to adaptively adjust a main bearing of a wind driven generator. The application can effectively identify and enhance early wear degradation signals of the main bearing of the wind driven generator, avoid misjudgment, provide more accurate refined data for the digital twinning platform, and realize adaptive adjustment on the main bearing.

Description

Technical Field

[0001] This application relates to the field of energy big data governance, specifically to an energy big data governance method and system based on digital twins. Background Technology

[0002] In the operation and management of modern energy infrastructure, digital twin technology is widely used to create virtual copies of physical energy assets to ensure the efficiency and reliability of the system. However, in practical applications, a common challenge is that critical equipment components may undergo slow and uneven physical degradation during long-term operation. The data characteristics generated by this degradation in the early stages are often very weak, and the magnitude of the changes is far below the conventionally set fault warning threshold, making it easy to be misjudged as environmental noise.

[0003] Taking a large wind farm as an example, the main bearing of a wind turbine may experience minute and uneven wear during long-term operation. Due to this subtle, underlying physical change in the frictional resistance of the main bearing, the associated sensor data stream begins to exhibit corresponding, but equally minute, deviations. For example, vibration patterns may show slight spectral shifts, internal temperature readings may record small but persistent increases, and actual power output at a given wind speed may be slightly lower than ideal. While these data variations are real and originate from physical entities, they typically fall within the expected range of normal operating fluctuations and are often difficult to distinguish from the background noise inherent in data acquisition and transmission.

[0004] When operations teams implement aggressive data preprocessing methods in an attempt to "cleanse" data, these subtle signals that should indicate early degradation are inadvertently erased. This causes the digital twin model to fail to accurately perceive the true state of the physical entity, and a subtle but not negligible deviation gradually accumulates between the model and the physical entity. Ultimately, this renders predictive analytics and intelligent decision support functions based on the model ineffective, affecting the overall operational efficiency and reliability of the energy system.

[0005] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0006] This application discloses an energy big data governance method and system based on digital twins, aiming to solve the problem in the prior art that the slow and uneven physical degradation of key equipment components during long-term operation, such as minor wear, produces weak data characteristics that are easily misjudged as environmental noise. This leads to the digital twin model being unable to accurately perceive the real state of the physical entity, thereby affecting the failure of predictive analysis and intelligent decision support functions.

[0007] The technical solution of this application is as follows:

[0008] Firstly, this application discloses a method for energy big data governance based on digital twins, comprising the following steps:

[0009] Real-time acquisition of vibration data of the main bearing of the wind turbine;

[0010] The vibration data is decomposed into multiple frequency scale components. The background noise signal in each frequency scale component is removed according to a preset background noise feature reference table to obtain multiple denoised frequency scale components.

[0011] Find the denoised frequency scale component where the wear vibration frequency of the main bearing is located from multiple denoised frequency scale components, extract the wear vibration energy characteristics of the wear vibration frequency from the found denoised frequency scale component, and calculate the confidence level of the wear vibration energy characteristics.

[0012] The confidence level is compared with the preset threshold for the main bearing wear degradation signal. When the confidence level exceeds the threshold for the main bearing wear degradation signal, the wear vibration energy characteristics are enhanced to obtain the wear enhancement characteristic signal.

[0013] The multiple denoised frequency scale components and the wear enhancement feature signal are fused into refined data, which is then transmitted to the digital twin platform so that the digital twin platform can make adaptive adjustments to the wind turbine main bearing based on the refined data.

[0014] Through this technical solution, this application can effectively identify and enhance the early wear and degradation signals of the main bearing of the wind turbine, avoiding its misjudgment as background noise, thereby providing more accurate refined data for the digital twin platform, realizing adaptive adjustment of the main bearing, and solving the problem of inconsistency between the digital twin model and the physical entity state in the prior art.

[0015] Furthermore, after the step of real-time acquisition of vibration data of the wind turbine main bearing, the following is also included:

[0016] Real-time acquisition of current operating condition data of wind turbines;

[0017] The specific steps for removing background noise signals from each frequency scale component according to a preset background noise feature reference table to obtain multiple denoised frequency scale components include:

[0018] Based on the current operating data, the background noise feature range corresponding to each frequency scale component of the wind turbine under the current operating condition is found in the preset background noise feature reference table.

[0019] Based on the background noise characteristic range, the background noise in each corresponding frequency scale component is removed to obtain multiple denoised frequency scale components.

[0020] Through this technical solution, this application can dynamically adjust the background noise removal strategy based on the real-time operating data of the wind turbine, making the noise removal process more accurate, effectively avoiding misjudgment or over-filtering of wear signals under different operating conditions, and improving the accuracy of data governance.

[0021] Furthermore, after the step of enhancing the wear vibration energy characteristics to obtain the wear enhancement characteristic signal, the process also includes:

[0022] Add enhancement markers to the wear enhancement feature signals;

[0023] The wear enhancement feature signal in the step of fusing the denoised frequency scale component and the wear enhancement feature signal into the refined data is a wear enhancement feature signal with added enhancement label.

[0024] Through this technical solution, this application adds clear identifiers to the enhanced wear signals, enabling the digital twin platform to clearly identify and prioritize these enhanced, high-confidence wear signals, thereby giving them higher weight and attention in subsequent adaptive adjustments, further improving the response speed and accuracy of the digital twin model to early wear.

[0025] Based on the above, this application further proposes that, after the step of real-time acquisition of vibration data of the wind turbine main bearing, the following method is also included:

[0026] Real-time acquisition of main bearing temperature data, output power data, and current operating condition data of wind turbine generators;

[0027] The step of calculating the confidence level of the wear vibration energy characteristics also includes:

[0028] Based on the wear vibration energy characteristics, main bearing temperature data, output power data, and current operating condition data, evaluate the coordinated change trajectory among the main bearing wear vibration energy, main bearing temperature, and output power of the wind turbine under the current operating condition.

[0029] Calculate the divergence between the coordinated change trajectory under the current working condition and the expected correlation trajectory under the corresponding preset working condition, and adjust the confidence level according to the magnitude of the divergence.

[0030] The confidence level in the step of comparing the confidence level with the preset threshold for the main bearing wear degradation signal is the corrected confidence level.

[0031] This technical solution introduces the collaborative change analysis of multi-dimensional data (temperature, power, operating conditions) and calculates the divergence between the data and the expected correlation trajectory to correct the confidence level of wear vibration energy characteristics. This allows for a more comprehensive and accurate assessment of the wear state of the main bearing, effectively avoiding misjudgments that may result from single vibration data and significantly improving the reliability of wear signal identification.

[0032] Preferably, the coordinated change trajectory under the current operating condition includes the main bearing temperature trajectory and the output power trajectory under the current operating condition, and the expected correlation trajectory under the corresponding operating condition includes the expected temperature correlation trajectory and the expected power correlation trajectory under the corresponding operating condition.

[0033] The steps of calculating the divergence between the coordinated change trajectory under the current operating condition and the expected correlation trajectory under the corresponding preset operating condition, and then adjusting the confidence level based on the magnitude of the divergence, specifically include:

[0034] Calculate the first divergence between the main bearing temperature trajectory under the current operating condition and the expected temperature correlation trajectory under the corresponding operating condition, and the second divergence between the output power trajectory under the current operating condition and the expected power correlation trajectory under the corresponding operating condition.

[0035] When the first divergence is lower than the first preset threshold and the second divergence is lower than the second preset threshold, the confidence level is adjusted and increased according to the magnitude of the first divergence and the second divergence.

[0036] If the first divergence is higher than the first preset threshold, and / or the second divergence is higher than the second preset threshold, the confidence level is adjusted and reduced according to the magnitude of the first divergence and / or the second divergence.

[0037] This technical solution calculates the divergence of temperature and power trajectories separately, and then refines the confidence level based on comparisons with preset thresholds, achieving bidirectional dynamic adjustment of the confidence level of wear signals. This correction mechanism can more accurately reflect the indicative role of multi-dimensional data co-changes on wear status, further improving the accuracy and sensitivity of wear signal identification.

[0038] In some preferred embodiments, after the step of comparing the confidence level with a preset main bearing wear degradation signal threshold, the method further includes:

[0039] When the confidence level is less than the main bearing wear degradation signal threshold, the wear vibration energy characteristics are defined as being caused by the structural resonance of the wind turbine, and a structural resonance interference marker is added to the wear vibration energy characteristics.

[0040] The multiple denoised frequency scale components and the wear vibration energy characteristics marked with structural resonance interference are fused into refined data. The refined data is then transmitted to the digital twin platform so that the digital twin platform can make adaptive adjustments to the wind turbine main bearing based on the refined data.

[0041] Through this technical solution, when the confidence level is insufficient to confirm a wear signal, this application can identify and mark it as structural resonance interference, avoiding misjudging non-wear-related vibrations as wear, thereby preventing the digital twin platform from making unnecessary adaptive adjustments based on erroneous information and improving the accuracy and reliability of system decision-making.

[0042] As a technological improvement, the following steps are included before the real-time acquisition of vibration data from the main bearing of the wind turbine:

[0043] During normal operation of the wind turbine, reference vibration data of the main bearing of the wind turbine are collected, and an inertial baseline is initialized for each frequency scale component of the reference vibration data.

[0044] The background noise range in the background noise feature reference table is initialized based on the inert baseline of each frequency scale component.

[0045] Before the step of removing background noise signals from each frequency scale component according to a preset background noise feature reference table to obtain multiple denoised frequency scale components, the following steps are also included:

[0046] The following operations are performed periodically at preset intervals: based on preset rules and the real-time vibration energy of each frequency scale component in the currently collected vibration data of the wind turbine main bearing, the inertial baseline of each frequency scale component is updated.

[0047] This application achieves dynamic adaptive adjustment of the background noise characteristic reference table by introducing an inertial baseline and a periodic update mechanism. This mechanism enables the system to better adapt to long-term changes in the wind turbine operating environment, ensuring the accuracy of background noise removal and thus more effectively identifying weak wear signals.

[0048] Based on this, before updating the inertial baseline of each frequency scale component according to preset rules and the real-time vibration energy of each frequency scale component in the currently collected vibration data of the wind turbine main bearing, the following steps are also included:

[0049] Calculate the deviation of the real-time vibration energy of each frequency scale component from the corresponding inertial baseline in the currently collected vibration data of the wind turbine main bearing, as well as the deviation velocity;

[0050] Each deviation is compared with a preset corresponding short-term fluctuation threshold, and the deviation speed is compared with a preset speed threshold;

[0051] If the deviation of a certain frequency scale component is greater than the short-term fluctuation threshold, the duration is greater than the first preset time, and the deviation speed is greater than the preset speed threshold, the real-time vibration energy of the corresponding frequency scale component is determined to be the main bearing wear and degradation signal; otherwise, the real-time vibration energy of the corresponding frequency scale component is determined to be instantaneous background noise.

[0052] The specific steps for updating the inertial baseline of each frequency scale component based on preset rules and the real-time vibration energy of each frequency scale component in the currently collected vibration data of the wind turbine main bearing include: for each frequency scale component in the currently collected vibration data of the wind turbine main bearing that is determined to be instantaneous background noise, the inertial baseline of the corresponding frequency scale component is updated according to preset rules and real-time vibration energy.

[0053] By introducing multiple judgment criteria such as deviation degree, deviation speed and duration, this application can more accurately distinguish between main bearing wear degradation signals and instantaneous background noise, avoid misjudging short-term fluctuations as wear, thereby improving the accuracy of inert baseline updates and further enhancing the system's ability to identify early wear signals.

[0054] As a further improvement, the step of determining the real-time vibration energy of the corresponding frequency scale component as the main bearing wear degradation signal also includes:

[0055] Calculate the matching degree between the spectrum of the corresponding frequency scale component in this period and the preset main bearing wear characteristic frequency library;

[0056] Compare the match score with the preset match threshold.

[0057] If the matching degree is greater than the preset matching threshold, the step of judging the real-time vibration energy of the corresponding frequency scale component as the main bearing wear degradation signal is executed.

[0058] For each real-time vibration energy in the currently collected vibration data of the wind turbine main bearing, which is judged as the frequency scale component of the instantaneous background noise, the inertial baseline is calculated using the following formula: L_i(t)=alpha*S_i(t)+(1-alpha)*L_i(t-1);

[0059] Where L_i(t) represents the inertial baseline value of frequency scale component i at time t; S_i(t) represents the real-time vibration energy of frequency scale component i at time t; L_i(t-1) represents the inertial baseline value of frequency scale component i at time t-1; and alpha represents the learning rate.

[0060] This technical solution enhances the accuracy of identifying main bearing wear degradation signals by introducing spectral matching judgment, avoiding misidentification of other types of vibration signals as wear. Simultaneously, the exponential smoothing formula is used to update the inertial baseline, enabling the baseline to smoothly adapt to long-term changes in background noise and improving the system's robustness to environmental noise.

[0061] Secondly, this application also discloses an energy big data governance system based on digital twins, including:

[0062] The data acquisition module is used to collect vibration data of the main bearing of the wind turbine in real time.

[0063] The noise reduction module is used to decompose the vibration data into multiple frequency scale components, and remove the background noise signal in each frequency scale component according to the preset background noise feature reference table to obtain multiple noise-reduced frequency scale components.

[0064] The lookup module is used to find the denoised frequency scale component where the wear vibration frequency of the main bearing is located from multiple denoised frequency scale components, extract the wear vibration energy characteristics of the wear vibration frequency from the found denoised frequency scale component, and calculate the confidence level of the wear vibration energy characteristics.

[0065] The comparison module is used to compare the confidence level with the preset threshold of the main bearing wear degradation signal. When the confidence level exceeds the threshold of the main bearing wear degradation signal, the wear vibration energy characteristics are enhanced to obtain the wear enhancement characteristic signal.

[0066] The transmission module is used to fuse the multiple denoised frequency scale components and the wear enhancement feature signal into refined data, and transmit the refined data to the digital twin platform so that the digital twin platform can make adaptive adjustments to the wind turbine main bearing based on the refined data.

[0067] This application provides a system-level solution through a modular design, which enables automated and intelligent management of wear data for wind turbine main bearings. It can efficiently collect, denoise, identify, enhance, and transmit wear signals, providing high-quality refined data for the digital twin platform. This supports precise adaptive adjustments to the main bearings, improving the overall operating efficiency and reliability of the energy system.

[0068] Beneficial effects

[0069] This application discloses a digital twin-based energy big data governance method that collects vibration data of wind turbine main bearings in real time and decomposes it into multiple frequency scale components. Based on this, background noise signals are removed from each frequency scale component according to a preset background noise feature reference table, resulting in multiple denoised frequency scale components. This effectively solves the problem in existing technologies where weak wear signals are easily masked by background noise. Subsequently, wear vibration energy features of the main bearing wear vibration frequency are found and extracted from the denoised frequency scale components, and their confidence levels are calculated, achieving accurate identification of early wear signals. When the confidence level exceeds a preset threshold, the wear vibration energy features are enhanced to obtain wear enhancement feature signals. This step effectively solves the problem in existing technologies where weak wear signals prevent accurate perception by the digital twin model. Finally, the denoised frequency scale components and wear enhancement feature signals are fused into refined data and transmitted to the digital twin platform, enabling the platform to adaptively adjust the wind turbine main bearing based on the refined data. Through the above technical solution, this application can effectively identify and enhance the early wear and degradation signal of the main bearing of the wind turbine, avoid it being misjudged as background noise, thereby providing more accurate refined data for the digital twin platform, realizing adaptive adjustment of the main bearing, solving the problem of inconsistency between the digital twin model and the physical entity state in the prior art, and significantly improving the overall operating efficiency and reliability of the energy system. Attached Figure Description

[0070] Figure 1 This application provides a schematic diagram of a digital twin-based energy big data governance method.

[0071] Figure 2 This application provides a schematic diagram of the structure of an energy big data governance system based on digital twins.

[0072] Figure 2 In the diagram: 1 is the acquisition module, 2 is the noise reduction module, 3 is the search module, 4 is the comparison module, and 5 is the transmission module. Detailed Implementation

[0073] The technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0074] See Figure 1 This application proposes a digital twin-based energy big data governance method, including the following steps:

[0075] S10. Real-time acquisition of vibration data of the main bearing of the wind turbine generator;

[0076] S20. Decompose the vibration data into multiple frequency scale components, and remove the background noise signal in each frequency scale component according to the preset background noise feature reference table to obtain multiple denoised frequency scale components.

[0077] S30. Find the denoised frequency scale component where the wear vibration frequency of the main bearing is located from multiple denoised frequency scale components, extract the wear vibration energy characteristics of the wear vibration frequency from the found denoised frequency scale component, and calculate the confidence level of the wear vibration energy characteristics.

[0078] S40. Compare the confidence level with the preset threshold for the main bearing wear degradation signal. When the confidence level exceeds the threshold for the main bearing wear degradation signal, enhance the wear vibration energy characteristics to obtain the wear enhancement characteristic signal.

[0079] S50. The multiple denoised frequency scale components and the wear enhancement feature signal are fused into refined data, and the refined data is transmitted to the digital twin platform so that the digital twin platform can make adaptive adjustments to the wind turbine main bearing based on the refined data.

[0080] This application refines the vibration data of wind turbine main bearings, particularly through background noise removal, wear vibration energy feature extraction and confidence calculation, and the generation of wear enhancement feature signals. This effectively identifies and enhances early, weak wear degradation signals. Consequently, the digital twin platform receives more accurate and indicative refined data, enabling adaptive adjustments to the wind turbine main bearings. This solves the problem of misjudgment or filtering of weak degradation signals in traditional methods, significantly improving the accuracy of the digital twin model's perception of the physical entity's state, and ultimately enhancing the operational efficiency and reliability of the energy system.

[0081] To better understand the technical solution proposed in this application, some key terms and implementation environments involved will be explained below.

[0082] A "digital twin platform" refers to a virtual digital model that is synchronized in real time with a physical entity (such as the main bearing of a wind turbine). This platform can receive data from the physical entity, perform simulations, analyses, and predictions, and guide or control the physical entity based on the analysis results.

[0083] The main bearing of a wind turbine is a critical component, and its operating condition directly affects the efficiency and lifespan of the generator. The wear and degradation of the main bearing is a gradual process, with early signs being weak and easily masked by noise.

[0084] "Vibration data" refers to vibration signals collected in real time by sensors installed on the main bearing of a wind turbine. This data contains rich information about the operating status of the main bearing, including vibration modes during normal operation and specific vibration characteristics during abnormal wear.

[0085] "Frequency scale components" refer to the signal components that decompose the original vibration data into different frequency ranges through signal processing techniques (such as wavelet decomposition, empirical mode decomposition, etc.). Each frequency scale component represents the energy distribution of the vibration data within a specific frequency range, which helps to identify different types of vibration sources.

[0086] The "Background Noise Characteristic Reference Table" is a pre-built database that stores the background noise characteristics of wind turbine main bearings under different operating conditions. This reference table is used to identify and remove environmental noise and inherent equipment noise during data processing, thereby highlighting true wear signals.

[0087] "Wear vibration frequency" refers to specific vibration frequencies that are directly related to the wear and degradation of main bearings. These frequencies typically appear or intensify when main bearings experience wear and are key indicators for identifying wear.

[0088] "Wear vibration energy characteristics" refer to the intensity of vibration energy exhibited at the wear vibration frequency. Extracting this characteristic helps to quantify the degree of wear.

[0089] "Confidence level" refers to the degree of certainty that the extracted wear vibration energy characteristics are genuine wear signals rather than noise. Calculating confidence level helps assess the reliability of the signal.

[0090] The "main bearing wear degradation signal threshold" is a preset value used to determine whether the wear vibration energy characteristics have reached the wear degradation level that requires attention. When the confidence level exceeds this threshold, it indicates that the wear degradation signal has high reliability.

[0091] "Wear enhancement feature signal" refers to the wear vibration energy characteristics after enhancement processing. The enhancement processing aims to amplify weak wear signals, making them easier for digital twin platforms to identify and utilize.

[0092] "Refined data" refers to high-quality data that has undergone a series of processing steps (including noise reduction, feature extraction, and enhancement). This data more accurately reflects the actual condition of the wind turbine main bearing and can be used for adaptive adjustments to the digital twin platform.

[0093] "Adaptive adjustment" refers to the optimization and adjustment of the operating parameters, maintenance strategies, or prediction models of the wind turbine main bearing by the digital twin platform based on the refined data received, in order to cope with the actual wear and degradation.

[0094] The core of the energy big data governance method based on digital twins proposed in this application lies in the refined processing of vibration data of wind turbine main bearings to accurately identify and enhance early, subtle wear and degradation signals. The main features of this method will be described in detail below.

[0095] First, regarding the step of "real-time acquisition of vibration data from the wind turbine main bearing," appropriate sensors need to be deployed to obtain information about the main bearing's operating status. For example, acceleration sensors or displacement sensors can be installed at key locations on the wind turbine main bearing. These sensors can continuously monitor the vibration of the main bearing.

[0096] Secondly, regarding the step of "decomposing vibration data into multiple frequency scale components and removing background noise signals from each frequency scale component according to a preset background noise feature reference table to obtain multiple denoised frequency scale components," after acquiring the original vibration data, it needs to be processed to separate the useful signals and background noise. One implementation method is to use Fast Fourier Transform (FFT) to convert the time-domain vibration data to the frequency domain, thereby obtaining the energy distribution of different frequency components. Then, the frequency domain signal is divided into multiple frequency scale components according to a preset frequency range. For example, the entire frequency range can be divided into multiple sub-bands such as low frequency, mid frequency, and high frequency. For background noise removal, a background noise feature reference table can be pre-established, recording the spectral characteristics or energy range of background noise corresponding to different frequency scale components under normal wind turbine operation. In actual processing, the real-time spectrum or energy of each frequency scale component is compared with the background noise features in the reference table to identify and filter out the background noise signals, thereby obtaining the denoised frequency scale components. For example, an adaptive filtering algorithm can be used to dynamically adjust the filter parameters according to the background noise feature reference table to remove noise more accurately.

[0097] Secondly, regarding the step of "finding the denoised frequency scale component containing the wear vibration frequency of the main bearing from multiple denoised frequency scale components, extracting the wear vibration energy characteristics of the wear vibration frequency from the found denoised frequency scale components, and calculating the confidence level of the wear vibration energy characteristics," after obtaining the denoised frequency scale components, it is necessary to identify the characteristics related to the main bearing wear. One approach is to pre-determine the specific vibration frequency range that the main bearing may exhibit when wear occurs, i.e., the wear vibration frequency, through experimental or theoretical analysis. Then, in the denoised frequency scale components, find the specific components containing these wear vibration frequencies. For example, if it is known that main bearing wear typically manifests in a certain high-frequency band, then the focus is on analyzing the denoised frequency scale components in that high-frequency band. From the found frequency scale components, extract the vibration energy at the wear vibration frequency as the wear vibration energy characteristic. To evaluate the reliability of this characteristic, its confidence level can be calculated. For example, the confidence level can be determined by comparing the fluctuation range of the current wear vibration energy characteristic with that in historical normal operation data, or by cross-validating multiple sensor data.

[0098] Next, regarding the step of "comparing the confidence level with a preset threshold for the main bearing wear degradation signal, and when the confidence level exceeds the threshold, enhancing the wear vibration energy characteristics to obtain an enhanced wear characteristic signal," after calculating the confidence level of the wear vibration energy characteristics, it needs to be compared with a preset threshold for the main bearing wear degradation signal. This threshold is set based on experience or historical data and is used to determine whether the wear signal reaches a level requiring attention. One implementation is that if the confidence level is higher than the threshold, it indicates that the currently detected wear vibration energy characteristics have high credibility and are likely to represent real wear degradation. At this time, in order to ensure that the digital twin platform can effectively identify these weak but real wear signals, the wear vibration energy characteristics need to be enhanced. For example, methods such as nonlinear amplification, feature weighting, or signal reconstruction can be used to increase the amplitude or intensity of the wear vibration energy characteristics, thereby obtaining an enhanced wear characteristic signal.

[0099] Finally, regarding the step of "fusing the multiple denoised frequency scale components and the wear enhancement feature signal into refined data, and transmitting the refined data to the digital twin platform so that the digital twin platform can adaptively adjust the wind turbine main bearing based on the refined data," after obtaining the wear enhancement feature signal, it needs to be fused with the previously obtained multiple denoised frequency scale components to form a complete, high-quality set of refined data. One implementation method is to package the wear enhancement feature signal as an additional feature dimension, along with the denoised frequency scale components, into a structured data format. For example, this data can be integrated into a multidimensional vector or time-series data stream. Subsequently, this refined data is transmitted to the digital twin platform via a network interface. After receiving the refined data, the digital twin platform will use its internal physical model, simulation algorithm, and machine learning model to evaluate the current state of the wind turbine main bearing. Based on this refined data, the digital twin platform can more accurately identify early wear and degradation trends of the main bearing and make adaptive adjustments to the operating parameters (such as speed and load), maintenance plans, or predictive maintenance strategies of the wind turbine main bearing, thereby extending equipment life, improving operating efficiency, and avoiding potential failures.

[0100] The energy big data governance method based on digital twins proposed in this application aims to address the problem in traditional methods where early, weak wear and degradation signals of wind turbine main bearings are easily masked by background noise and misjudged. This method, through a series of refined data processing steps, ensures that the digital twin platform can receive high-quality, high-confidence wear signals, thereby enabling more accurate adaptive adjustments to the physical entity.

[0101] In summary, this application overcomes the challenge of identifying and utilizing weak wear signals in existing technologies through refined data acquisition, noise reduction, feature extraction, confidence assessment, and signal enhancement. By transmitting this refined data to a digital twin platform, this application enables the digital twin model to more accurately reflect the true state of the physical entity, thereby achieving smarter and more efficient energy big data governance and equipment management.

[0102] The above-mentioned energy big data governance method based on digital twins, after the step of real-time collection of vibration data of wind turbine main bearings, also includes:

[0103] Real-time acquisition of current operating condition data of wind turbines;

[0104] The specific steps for removing background noise signals from each frequency scale component according to a preset background noise feature reference table to obtain multiple denoised frequency scale components include:

[0105] Based on the current operating data, the background noise feature range corresponding to each frequency scale component of the wind turbine under the current operating condition is found in the preset background noise feature reference table.

[0106] Based on the background noise characteristic range, the background noise in each corresponding frequency scale component is removed to obtain multiple denoised frequency scale components.

[0107] Specifically, real-time acquisition of wind turbine operating data refers to obtaining the operating status parameters of the wind turbine at a specific moment or within a specific time period through sensors or SCADA systems. These parameters may include, but are not limited to, wind speed, wind direction, ambient temperature, generator speed, output power, blade pitch angle, and nacelle yaw angle. The purpose is to provide crucial contextual information for subsequent background noise identification.

[0108] The step of removing background noise signals from each frequency scale component based on a preset background noise feature reference table to obtain multiple denoised frequency scale components is further refined. Specifically, the preset background noise feature reference table is used to find the background noise feature range that best matches the current operating conditions based on real-time collected data. This background noise feature reference table can be pre-established by running the wind turbine under different operating conditions for a long time and collecting its vibration data, and then analyzing and modeling this data. For example, the spectral characteristics and energy distribution of background noise may differ significantly under different operating conditions such as low wind speed, high wind speed, no load, and full load. Using the current operating condition data, the system can accurately locate the background noise feature range that matches the current operating state.

[0109] In practical applications, once the background noise characteristic range corresponding to each frequency scale component under the current operating conditions is determined, appropriate signal processing techniques (such as adaptive filtering, wavelet thresholding denoising, spectral subtraction, etc.) can be used to remove the background noise from each corresponding frequency scale component based on this range. This results in cleaner and more accurate denoised frequency scale components.

[0110] This application's solution addresses the issue of insufficient accuracy in background noise removal in traditional methods by introducing real-time acquired current operating condition data. By acquiring current operating condition data in real time, the system can dynamically select or generate the background noise feature range that best matches the current situation from a background noise feature reference table. This dynamic matching mechanism makes the identification and removal of background noise more targeted and accurate, thus avoiding the misidentification of useful wear signals as noise, or vice versa.

[0111] Through the above technical solution, this application can significantly improve the accuracy and adaptability of background noise removal from the vibration data of wind turbine main bearings. Because the denoising process can be dynamically adjusted according to the real-time operating conditions of the wind turbine, the true wear vibration signal can be separated more effectively, reducing the interference of noise on wear feature extraction. Therefore, the extracted wear vibration energy features will be more reliable, and the calculation of their confidence level will be more accurate. Ultimately, the digital twin platform can make more precise and timely adaptive adjustments to the wind turbine main bearing based on this high-quality refined data, thereby effectively extending the service life of the main bearing, reducing maintenance costs, and improving the operating efficiency and safety of the wind turbine.

[0112] This application further proposes a method for energy big data governance based on digital twins, wherein, after the step of enhancing the wear vibration energy characteristics to obtain the wear enhancement characteristic signal, the method further includes:

[0113] Add enhancement markers to the wear enhancement feature signals;

[0114] The wear enhancement feature signal in the step of fusing the denoised frequency scale component and the wear enhancement feature signal into the refined data is a wear enhancement feature signal with added enhancement label.

[0115] Specifically, an enhancement identifier refers to metadata or a flag attached to the wear enhancement feature signal. Its purpose is to explicitly indicate that the signal has been enhanced and that its confidence level has reached or exceeded the main bearing wear degradation signal threshold. This enhancement identifier can be a specific binary bit, a predefined string, a numerical code, or any other mark that can be recognized and parsed by the digital twin platform. For example, a Boolean flag can be set for the wear enhancement feature signal, which is set to "true" or "1" when the signal is enhanced. When the denoised frequency scale component and the wear enhancement feature signal are fused into refined data, the fused wear enhancement feature signal is one that has already been marked with the enhancement identifier. This means that before data fusion, the enhancement identifier has been embedded in the data structure of the wear enhancement feature signal or transmitted as accompanying information. In practical applications, this fusion process can involve packaging the wear enhancement feature signal with the enhancement identifier and other denoised frequency scale components into a unified data packet or data stream for overall reception and processing by the digital twin platform.

[0116] The solution proposed in this application adds an enhancement identifier to the enhanced wear vibration energy characteristic signal, enabling the digital twin platform to clearly identify the signal after it is integrated into refined data and transmitted to the platform. Due to this enhancement identifier, the digital twin platform can distinguish which wear signals have undergone high-confidence assessment and enhancement, thus allowing for priority processing and deeper analysis of these key signals. This explicit labeling mechanism avoids the digital twin platform missing or misjudging key wear signals when processing massive amounts of data, ensuring timely warnings and responses to potential equipment failures.

[0117] This application proposes an improved energy big data governance method based on digital twins, which, after the step of real-time acquisition of vibration data of wind turbine main bearings, further includes:

[0118] Real-time acquisition of main bearing temperature data, output power data, and current operating condition data of wind turbine generators;

[0119] Following the step of calculating the confidence level of the wear vibration energy characteristics, the following further steps are included:

[0120] Based on the wear vibration energy characteristics, main bearing temperature data, output power data, and current operating condition data, evaluate the coordinated change trajectory among the main bearing wear vibration energy, main bearing temperature, and output power of the wind turbine under the current operating condition.

[0121] Calculate the divergence between the coordinated change trajectory under the current working condition and the expected correlation trajectory under the corresponding preset working condition, and adjust the confidence level according to the magnitude of the divergence.

[0122] The confidence level in the step of comparing the confidence level with the preset threshold for the main bearing wear degradation signal is the corrected confidence level.

[0123] Specifically, real-time acquisition of wind turbine main bearing temperature, output power, and current operating condition data refers to continuously acquiring real-time operating temperature of the main bearing, real-time power generation of the wind turbine, and current environmental and operating parameters such as wind speed and direction through various sensors installed on the wind turbine. This data is considered auxiliary information closely related to main bearing wear and degradation, providing a more comprehensive view of the operating status. Evaluating the coordinated change trajectory between the main bearing wear vibration energy, main bearing temperature, and output power under current operating conditions can be understood as analyzing the interrelationships and trends of these multi-source data over time. For example, when the main bearing wears, its vibration energy typically increases, possibly accompanied by an abnormal increase in temperature or fluctuations in output power. This coordinated change trajectory can be modeled and evaluated using multivariate time series analysis, state-space models, or machine learning algorithms. In practical applications, calculating the divergence between the coordinated change trajectory under current operating conditions and the expected correlation trajectory under the corresponding preset operating conditions refers to comparing the currently observed multi-parameter coordinated change pattern with a pre-established baseline pattern under normal operation or known wear patterns. The expected correlation trajectory can be obtained from historical data or simulation models under different operating conditions, representing the normal or typical correlation between parameters under specific operating conditions. Divergence can be quantified using Euclidean distance, Mahalanobis distance, correlation coefficient changes, or anomaly detection indicators based on machine learning. Its purpose is to measure the degree of deviation between the current operating state and the expected state. Furthermore, adjusting the confidence level based on the magnitude of the divergence refers to adjusting the confidence level of the initially calculated wear vibration energy characteristics based on the degree of deviation between the coordinated change trajectory and the expected correlation trajectory.

[0124] This application's solution, by incorporating main bearing temperature data, output power data, and current operating condition data, and evaluating the coordinated change trajectory between these multi-source data and wear vibration energy characteristics, can more comprehensively capture the operating state of the main bearing. Because wear degradation is a complex process involving multiple coupled factors, a single vibration characteristic analysis may not fully reflect its true state. By analyzing the divergence between the coordinated change trajectory and the expected correlation trajectory, vibration anomalies caused by non-wear factors (such as changes in operating conditions, instantaneous impacts, etc.) can be effectively identified, thus avoiding misjudgment as wear signals. Simultaneously, for genuine wear signals, combining abnormal coordinated changes in other parameters can further improve their confidence level, making the identification of wear degradation signals more accurate and robust. This multi-dimensional, coordinated analysis method allows the judgment of the main bearing wear state to move beyond a single feature and instead make comprehensive decisions based on richer contextual information.

[0125] Through the above technical solution, this application can significantly improve the accuracy and reliability of identifying wear and degradation signals of wind turbine main bearings. By comprehensively considering auxiliary data such as main bearing temperature, output power, and current operating conditions, and evaluating their synergistic changes with wear vibration energy characteristics, it is possible to effectively distinguish vibration anomalies caused by actual wear from those caused by other non-wear factors (such as fluctuations in normal operating conditions, environmental interference, etc.). This helps reduce false alarms and missed alarms, enabling the digital twin platform to receive more refined and accurate wear status data, thereby providing a more reliable basis for adaptive adjustments to the main bearing and avoiding unnecessary maintenance or equipment damage due to misjudgment.

[0126] This application further proposes the following steps for calculating the divergence between the cooperative change trajectory under the current operating condition and the expected correlation trajectory under the corresponding preset operating condition, and for correcting the confidence level based on the magnitude of the divergence:

[0127] The coordinated change trajectory under the current operating condition includes the main bearing temperature trajectory and the output power trajectory under the current operating condition. The expected correlation trajectory under the corresponding operating condition includes the expected temperature correlation trajectory and the expected power correlation trajectory under the corresponding operating condition.

[0128] Calculate the first divergence between the main bearing temperature trajectory under the current operating condition and the expected temperature correlation trajectory under the corresponding operating condition, and the second divergence between the output power trajectory under the current operating condition and the expected power correlation trajectory under the corresponding operating condition.

[0129] When the first divergence is lower than the first preset threshold and the second divergence is lower than the second preset threshold, the confidence level is adjusted and increased according to the magnitude of the first divergence and the second divergence.

[0130] If the first divergence is higher than the first preset threshold, and / or the second divergence is higher than the second preset threshold, the confidence level is adjusted and reduced according to the magnitude of the first divergence and / or the second divergence.

[0131] Specifically, the coordinated change trajectory under the current operating condition can be understood as the actual observed sequence of changes in main bearing temperature and output power over time under the current operating condition of the wind turbine. Specifically, the main bearing temperature trajectory under the current operating condition refers to the actual measured sequence of changes in main bearing temperature over time under a specific operating condition; the output power trajectory under the current operating condition refers to the actual measured sequence of changes in wind turbine output power over time under the same specific operating condition. These trajectories reflect the actual behavioral pattern of the wind turbine under its current operating state.

[0132] The expected correlation trajectory under corresponding operating conditions can be understood as a baseline or predicted sequence of changes in main bearing temperature and output power over time under the operating state corresponding to the current operating condition. Specifically, the expected temperature correlation trajectory under corresponding operating conditions refers to the expected trend or range of change in main bearing temperature under the corresponding operating condition; the expected power correlation trajectory under corresponding operating conditions refers to the expected trend or range of change in output power under the corresponding operating condition. These expected correlation trajectories are usually pre-established based on historical data, physical models, or expert experience, and are used as a reference benchmark for evaluating the current system state.

[0133] In practical applications, calculating the first divergence between the main bearing temperature trajectory under the current operating condition and the expected temperature correlation trajectory under the corresponding operating condition, and the second divergence between the output power trajectory under the current operating condition and the expected power correlation trajectory under the corresponding operating condition, aims to quantify the degree of deviation between the actual operating state and the ideal healthy state. The first divergence measures the difference between the actual main bearing temperature trajectory and the expected temperature correlation trajectory, and can be obtained, for example, by calculating indicators such as the mean square error (MSE), dynamic time warping (DTW) distance, or correlation coefficient between the two trajectories. Similarly, the second divergence measures the difference between the actual output power trajectory and the expected power correlation trajectory, and can also be calculated using a similar method.

[0134] Furthermore, when the first divergence is lower than the first preset threshold and the second divergence is lower than the second preset threshold, this indicates that the actual changes in the main bearing temperature and output power are highly consistent with the expected pattern, and that the wind turbine main bearing is experiencing wear. In this case, the confidence level is improved by adjusting the magnitude of the first and second divergences. The purpose is to enhance confidence that the wear vibration energy characteristics are a genuine wear signal; the smaller the divergence, the greater the increase in confidence level.

[0135] Conversely, when the first divergence exceeds a first preset threshold, and / or the second divergence exceeds a second preset threshold, this indicates a significant deviation between the actual change in the main bearing temperature or output power (or both) and the expected pattern. In this case, the confidence level is reduced based on the magnitude of the first and / or second divergence. This is to treat the currently detected wear vibration energy characteristics with caution and avoid misjudgment, as anomalies in other parameters may mean that the vibration signal is not simply caused by wear, but by other faults or changes in operating conditions. For example, the greater the divergence, the greater the reduction in confidence level.

[0136] The proposed solution decomposes the co-variance trajectory into the main bearing temperature trajectory and the output power trajectory under the current operating condition. It then calculates the first and second divergences between these trajectories and the expected temperature and power trajectories under the corresponding operating conditions, enabling a more refined and comprehensive assessment of the wind turbine main bearing's operating status. This refinement allows the system to identify whether the anomaly is due to temperature, power, or both, thus avoiding information loss that might occur with a single comprehensive indicator. When both divergences are below their respective preset thresholds, it indicates that the main bearing's temperature and output power are fluctuating within the corresponding ranges, providing strong evidence for the authenticity of the wear vibration energy characteristics, thus allowing for adjustments to increase confidence. Conversely, when any one or both divergences exceed the preset thresholds, it suggests the possibility of other anomalies, such as sensor malfunction, changes in environmental factors, or non-wear mechanical failures. These factors can complicate the vibration signal, making simple wear vibration energy characteristics insufficient to accurately determine the wear state. By correcting and reducing the confidence level, the solution proposed in this application can effectively avoid misjudging vibration signals caused by other anomalies as wear and degradation signals, thereby improving the accuracy and reliability of wear diagnosis.

[0137] This application further proposes, after comparing the confidence level with a preset main bearing wear degradation signal threshold, the following operations are also included: when the confidence level is less than the main bearing wear degradation signal threshold, the wear vibration energy characteristics are defined as being caused by the structural resonance of the wind turbine, and a structural resonance interference marker is added to the wear vibration energy characteristics; the multiple denoised frequency scale components and the wear vibration energy characteristics with added structural resonance interference markers are fused into refined data, and the refined data is transmitted to the digital twin platform so that the digital twin platform can adaptively adjust the wind turbine main bearing based on the refined data.

[0138] Specifically, when the confidence level of the calculated wear vibration energy characteristic is lower than the preset threshold for the main bearing wear degradation signal, it indicates that the vibration signal is not a strong wear degradation signal. At this point, the system further analyzes the wear vibration energy characteristic and defines it as being caused by the structural resonance of the wind turbine. This definition can be based on a pre-established structural resonance frequency library, achieved by comparing the matching degree between the frequency components of the current vibration signal and known structural resonance frequencies. Once structural resonance is determined, a structural resonance interference marker is added to the wear vibration energy characteristic. This marker is used to clearly indicate the nature of the vibration signal, distinguishing it from the actual wear degradation signal. Subsequently, the multiple denoised frequency scale components and the wear vibration energy characteristic with the added structural resonance interference marker are fused into refined data. The structural resonance interference marker, as part of the refined data, provides the digital twin platform with richer and more accurate contextual information. Finally, this refined data is transmitted to the digital twin platform, enabling it to make more precise and adaptive adjustments to the wind turbine main bearing based on this clearly marked refined data. For example, when structural resonance is detected, the digital twin platform can suggest adjusting the operating parameters of the wind turbine to avoid the resonance frequency, rather than mistakenly triggering wear-related maintenance operations.

[0139] This application's solution introduces a further judgment mechanism for the source of wear vibration energy characteristics when the confidence level is low, and adds a structural resonance interference marker, thereby effectively distinguishing vibration signals caused by actual wear from vibration signals caused by non-wear factors such as structural resonance. It is precisely because of this distinction that the digital twin platform, upon receiving refined data, can make decisions based on more accurate information, avoiding misjudging structural resonance as wear degradation, and thus preventing unnecessary or erroneous adaptive adjustments.

[0140] Prior to the aforementioned step of real-time acquisition of vibration data of the wind turbine main bearing, this application further includes:

[0141] During normal operation of the wind turbine, reference vibration data of the main bearing of the wind turbine are collected, and an inertial baseline is initialized for each frequency scale component of the reference vibration data.

[0142] The background noise range in the background noise feature reference table is initialized based on the inert baseline of each frequency scale component.

[0143] Before the step of removing background noise signals from each frequency scale component according to a preset background noise feature reference table to obtain multiple denoised frequency scale components, the following steps are also included:

[0144] The following operations are performed periodically at preset intervals: based on preset rules and the real-time vibration energy of each frequency scale component in the currently collected vibration data of the wind turbine main bearing, the inertial baseline of each frequency scale component is updated.

[0145] Specifically, when the main bearing of a wind turbine is in normal operation, its vibration data is collected as reference vibration data. This reference vibration data is used to establish an initial background noise model. The reference vibration data is decomposed into multiple frequency scale components, and an inertial baseline is initialized for each frequency scale component. The inertial baseline can be understood as the long-term average vibration energy level of that frequency scale component under normal operating conditions, characterized by a slow response and susceptibility to short-term transient fluctuations.

[0146] The background noise feature reference table is a data structure used to store the background noise feature ranges corresponding to different frequency scale components. After the inertial baseline is initialized, it is used to initialize the background noise range for each corresponding frequency scale component in the background noise feature reference table. Furthermore, to enable the background noise feature reference table to adapt to long-term changes in the wind turbine operating environment, the inertial baseline is updated periodically. Specifically, every preset time period, the system updates the corresponding inertial baseline based on the real-time vibration energy of each frequency scale component in the currently collected vibration data of the wind turbine main bearing, according to preset rules. The preset rules can be a smooth update algorithm, such as an exponentially weighted moving average, to ensure that the inertial baseline can slowly track the long-term trend of background noise while filtering out instantaneous noise or short-term anomalies.

[0147] This application's solution overcomes the limitation of static background noise feature reference tables in traditional methods by introducing an inertial baseline and a periodic update mechanism. First, benchmark vibration data is collected and an inertial baseline is initialized under normal wind turbine operating conditions, ensuring that the initial setting of the background noise feature reference table accurately reflects the background noise level of the wind turbine in a healthy state. Second, by periodically updating the inertial baseline at preset intervals, the background noise feature reference table dynamically adapts to long-term changes in the wind turbine's operating environment, such as changes in ambient temperature and wind speed, or background noise drift caused by slight equipment aging. The slow update characteristic of the inertial baseline ensures its effective differentiation between real background noise changes and transient interference, thus avoiding misjudgment of valid wear signals. Therefore, background noise removal is more accurate, providing a cleaner data foundation for subsequent wear vibration energy feature extraction.

[0148] This application further proposes that, before the step of updating the inertial baseline of each frequency scale component based on the real-time vibration energy of each frequency scale component in the vibration data of the wind turbine main bearing according to preset rules and currently collected data, the following additional steps are included:

[0149] Calculate the deviation of the real-time vibration energy of each frequency scale component from the corresponding inertial baseline in the currently collected vibration data of the wind turbine main bearing, as well as the deviation velocity;

[0150] Each deviation is compared with a preset corresponding short-term fluctuation threshold, and the deviation speed is compared with a preset speed threshold;

[0151] If the deviation of a certain frequency scale component is greater than the short-term fluctuation threshold, the duration is greater than the first preset time, and the deviation speed is greater than the preset speed threshold, the real-time vibration energy of the corresponding frequency scale component is determined to be the main bearing wear and degradation signal; otherwise, the real-time vibration energy of the corresponding frequency scale component is determined to be instantaneous background noise.

[0152] The specific steps for updating the inertial baseline of each frequency scale component based on preset rules and the real-time vibration energy of each frequency scale component in the currently collected vibration data of the wind turbine main bearing include:

[0153] For each real-time vibration energy component in the currently collected vibration data of the wind turbine main bearing that is identified as an instantaneous background noise frequency scale component, the inertial baseline of the corresponding frequency scale component is updated according to preset rules and real-time vibration energy.

[0154] Specifically, deviation refers to the relative change of the current real-time vibration energy relative to its inertial baseline, which can be used to measure the fluctuation amplitude of the real-time vibration energy. Deviation rate refers to the rate of change of the real-time vibration energy relative to the inertial baseline, which can reflect the drastic degree of vibration energy change. Short-term fluctuation threshold and velocity threshold are preset reference values ​​used to distinguish between normal transient background noise and potential wear and degradation signals. The first preset time is used to ensure the stability of the judgment and avoid misjudgment due to brief abnormal fluctuations. When the deviation, duration, and deviation rate of the real-time vibration energy all exceed the preset thresholds, it indicates that the vibration energy is likely not simple background noise, but a degradation signal caused by main bearing wear. Conversely, if these conditions are not met, the vibration energy is considered to be transient background noise. The inertial baseline is updated only for the frequency scale component judged as transient background noise to ensure that the inertial baseline can accurately track the slow changes in background noise without being affected by wear and degradation signals.

[0155] The proposed solution effectively distinguishes between main bearing wear degradation signals and instantaneous background noise by comprehensively assessing the deviation, velocity, and duration of real-time vibration energy. This distinction makes the inertial baseline update process more accurate. The inertial baseline is only updated when real-time vibration energy is identified as instantaneous background noise, thus preventing contamination of the inertial baseline by wear degradation signals. Consequently, the inertial baseline more accurately reflects the background noise level of the wind turbine main bearing under normal operating conditions, providing a more reliable reference for subsequent background noise removal.

[0156] This application further refines the above-mentioned judgment steps and provides a specific calculation method for updating the inert baseline.

[0157] Before determining the real-time vibration energy of the corresponding frequency scale component as the main bearing wear degradation signal, the following steps are also included:

[0158] Calculate the matching degree between the spectrum of the corresponding frequency scale component in this period and the preset main bearing wear characteristic frequency library;

[0159] Compare the match score with the preset match threshold.

[0160] If the matching degree is greater than the preset matching threshold, the step of judging the real-time vibration energy of the corresponding frequency scale component as the main bearing wear degradation signal is executed.

[0161] The steps for updating the inertial baseline of each frequency scale component, which is identified as instantaneous background noise in the vibration data of the wind turbine main bearing, based on preset rules and real-time vibration energy, specifically include:

[0162] For each real-time vibration energy in the currently collected vibration data of the wind turbine main bearing, which is judged as the frequency scale component of the instantaneous background noise, the inertial baseline is calculated using the following formula: L_i(t)=alpha*S_i(t)+(1-alpha)*L_i(t-1);

[0163] Where L_i(t) represents the inertial baseline value of frequency scale component i at time t; S_i(t) represents the real-time vibration energy of frequency scale component i at time t; L_i(t-1) represents the inertial baseline value of frequency scale component i at time t-1; and alpha represents the learning rate, which ranges from 0.0001 to 0.001.

[0164] Specifically, before determining whether the real-time vibration energy of a certain frequency scale component is a signal of main bearing wear degradation, it is first necessary to calculate the matching degree between the spectrum of that frequency scale component in the current period and a preset main bearing wear characteristic frequency library. The main bearing wear characteristic frequency library can be understood as a database storing typical vibration spectrum characteristics corresponding to various known main bearing wear types (e.g., inner ring wear, outer ring wear, rolling element wear, etc.). The matching degree refers to the similarity between the spectrum of the current frequency scale component and the wear characteristic spectrum in the library, which can be calculated using various spectral similarity measurement methods (e.g., correlation coefficient, Euclidean distance, cosine similarity, etc.). This matching degree is then compared with a preset matching threshold. Only when the matching degree exceeds this threshold is the step of determining whether the real-time vibration energy of the corresponding frequency scale component is a signal of main bearing wear degradation performed. This aims to ensure that the identified wear signal is not only abnormal in energy but also highly consistent with the actual wear pattern in terms of spectral characteristics, thereby improving the accuracy of the judgment.

[0165] This application's solution effectively enhances the specificity of wear signal identification by introducing spectral matching degree as a prerequisite for judging main bearing wear degradation signals. When the vibration energy of a certain frequency scale component deviates significantly, lasts for a long time, and deviates rapidly, but its spectral characteristics do not match the known wear patterns well, this signal will not be misidentified as a main bearing wear degradation signal. This avoids the misidentification of strong vibration signals caused by other non-wear factors (such as structural resonance, external impact, etc.), thereby reducing false alarms. Simultaneously, an exponential smoothing formula is used to update the inertial baseline, allowing the background noise baseline to adapt to slow changes in the environment or operating conditions in a controlled and stable manner. The setting of the learning rate alpha ensures that the baseline is not overly sensitive to instantaneous noise fluctuations but rather smoothly reflects the long-term trend of background noise, thus providing a more accurate reference for subsequent denoising processing.

[0166] See Figure 2The specific implementation of this application also discloses an energy big data governance system based on digital twins, including: a data acquisition module 1, a noise reduction module 2, a search module 3, a comparison module 4, and a transmission module 5. The system comprises the following modules: Acquisition module 1, which acquires vibration data of the wind turbine main bearing in real time; Denoising module 2, which decomposes the vibration data into multiple frequency scale components and removes background noise signals from each frequency scale component according to a preset background noise feature reference table, resulting in multiple denoised frequency scale components; Search module 3, which searches for the denoised frequency scale component containing the wear vibration frequency of the main bearing from the multiple denoised frequency scale components, extracts the wear vibration energy features of the wear vibration frequency from the found denoised frequency scale components, and calculates the confidence level of the wear vibration energy features; Comparison module 4, which compares the confidence level with a preset main bearing wear degradation signal threshold, and enhances the wear vibration energy features when the confidence level exceeds the main bearing wear degradation signal threshold, resulting in a wear enhancement feature signal; and Transmission module 5, which fuses the multiple denoised frequency scale components and the wear enhancement feature signal into refined data and transmits the refined data to the digital twin platform, enabling the digital twin platform to adaptively adjust the wind turbine main bearing based on the refined data.

[0167] This system aims to address the problem in traditional energy big data governance where early, weak degradation signals of critical equipment components are easily masked by background noise and misjudged. Through the collaborative work of its modules, the system can refine the vibration data of wind turbine main bearings, effectively identifying and enhancing early, weak wear and degradation signals. As a result, the digital twin platform receives more accurate and indicative refined data, enabling adaptive adjustments to the wind turbine main bearings. This significantly improves the accuracy of the digital twin model's perception of the physical entity's state, thereby enhancing the operational efficiency and reliability of the energy system.

[0168] The key terms and implementation environment involved in the proposed system technical solution have been described in the aforementioned "Energy Big Data Governance Method Based on Digital Twins," and will not be repeated here. The acquisition module can consist of one or more sensors and a data acquisition unit. For example, the sensor could be a piezoelectric accelerometer, mounted on the main bearing housing, used to convert mechanical vibrations into electrical signals.

[0169] The noise reduction module can be a standalone software component or a dedicated hardware processing unit. For example, the module can be implemented as a software program running on a central processing unit (CPU) or digital signal processor (DSP), which includes signal decomposition algorithms (such as Fast Fourier Transform (FFT) or wavelet transform) and noise identification and filtering algorithms.

[0170] The lookup module can be a data analysis engine with a pre-defined library of characteristic frequencies related to main bearing wear. This module uses pattern matching or spectral analysis techniques to identify wear-related frequency components within the denoised frequency scale components.

[0171] The comparison module can be a decision logic unit that receives the confidence level from the lookup module and compares it with a preset threshold for main bearing wear degradation signals. When the enhancement conditions are met, this module triggers a signal enhancement algorithm.

[0172] The transmission module can be a data integration and communication interface. This module is responsible for the structured encapsulation of the denoised frequency scale components and wear-enhanced feature signals, for example, by integrating them into a unified data packet or message stream.

[0173] The energy big data governance system based on digital twins provided in this application has a working principle and execution steps that are basically the same as the energy big data governance method based on digital twins mentioned above, and will not be repeated here.

[0174] In summary, the system of this application overcomes the difficulty in identifying and utilizing weak wear signals in existing technologies through its modular design and refined data processing flow. By transmitting high-quality, high-confidence, and enhanced wear signals to the digital twin platform, this application enables the digital twin model to more accurately reflect the real state of the physical entity, thereby achieving smarter and more efficient energy big data governance and equipment management, significantly improving the overall operating efficiency and reliability of wind turbines.

[0175] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for energy big data governance based on digital twins, characterized in that, Includes the following steps: Real-time acquisition of vibration data of the main bearing of the wind turbine generator; decomposition of the vibration data into multiple frequency scale components; removal of background noise signals from each frequency scale component according to a preset background noise feature reference table to obtain multiple denoised frequency scale components; finding the denoised frequency scale component where the wear vibration frequency of the main bearing is located from the multiple denoised frequency scale components; extracting the vibration energy at the wear vibration frequency from the found denoised frequency scale component as the wear vibration energy feature; and calculating the confidence level of the wear vibration energy feature. The confidence level is compared with a preset threshold for the main bearing wear degradation signal. When the confidence level exceeds the threshold, nonlinear amplification, feature weighting, or signal reconstruction is used to enhance the wear vibration energy characteristics to obtain a wear enhancement feature signal. The multiple denoised frequency scale components and the wear enhancement feature signal are fused into refined data, and the refined data is transmitted to the digital twin platform so that the digital twin platform can adaptively adjust the wind turbine main bearing based on the refined data. Following the step of real-time acquisition of vibration data of the wind turbine main bearing, the process further includes: real-time acquisition of wind turbine main bearing temperature data, output power data, and current operating condition data; following the step of calculating the confidence level of wear vibration energy characteristics, the process further includes: based on the wear vibration energy characteristics, main bearing temperature data, output power data, and current operating condition data, evaluating the coordinated change trajectory between the wind turbine main bearing wear vibration energy, main bearing temperature, and output power under the current operating condition; calculating the divergence between the coordinated change trajectory under the current operating condition and the preset expected correlation trajectory under the corresponding operating condition, and correcting the confidence level based on the magnitude of the divergence; comparing the confidence level with the preset main bearing wear degradation signal threshold, where the confidence level is the corrected confidence level.

2. The energy big data governance method based on digital twins according to claim 1, characterized in that, Following the step of real-time acquisition of vibration data of the wind turbine main bearing, the following steps are also included: real-time acquisition of the current operating condition data of the wind turbine; and the step of removing background noise signals from each frequency scale component according to a preset background noise feature reference table to obtain multiple denoised frequency scale components. Specifically, this includes: finding the background noise feature range corresponding to each frequency scale component under the current operating condition of the wind turbine in the preset background noise feature reference table; and removing background noise from each corresponding frequency scale component according to the background noise feature range to obtain multiple denoised frequency scale components.

3. The energy big data governance method based on digital twins according to claim 1, characterized in that, The process of enhancing the wear vibration energy characteristics to obtain the wear enhanced characteristic signal further includes: adding an enhancement identifier to the wear enhanced characteristic signal; and fusing the denoised frequency scale component and the wear enhanced characteristic signal into refined data. The wear enhanced characteristic signal in the process is the wear enhanced characteristic signal with the added enhancement identifier.

4. The energy big data governance method based on digital twins according to claim 1, characterized in that, The coordinated change trajectory under the current operating condition includes the main bearing temperature trajectory and the output power trajectory under the current operating condition. The expected correlation trajectory under the corresponding operating condition includes the expected temperature correlation trajectory and the expected power correlation trajectory under the corresponding operating condition. The step of calculating the divergence between the coordinated change trajectory under the current operating condition and the preset expected correlation trajectory under the corresponding operating condition, and correcting the confidence level based on the magnitude of the divergence, specifically includes: calculating the first divergence between the main bearing temperature trajectory under the current operating condition and the expected temperature correlation trajectory under the corresponding operating condition, and the second divergence between the output power trajectory under the current operating condition and the expected power correlation trajectory under the corresponding operating condition; when the first divergence is lower than a first preset threshold and the second divergence is lower than a second preset threshold, the confidence level is increased based on the magnitude of the first divergence and the second divergence; when the first divergence is higher than the first preset threshold and / or the second divergence is higher than the second preset threshold, the confidence level is decreased based on the magnitude of the first divergence and / or the second divergence.

5. The energy big data governance method based on digital twins according to claim 4, characterized in that, After comparing the confidence level with the preset main bearing wear degradation signal threshold, the method further includes: if the confidence level is less than the main bearing wear degradation signal threshold, then the wear vibration energy characteristics are defined as being caused by the structural resonance of the wind turbine generator, and a structural resonance interference marker is added to the wear vibration energy characteristics; the multiple denoised frequency scale components and the wear vibration energy characteristics with added structural resonance interference markers are fused into refined data, and the refined data is transmitted to the digital twin platform so that the digital twin platform can adaptively adjust the wind turbine generator main bearing based on the refined data.

6. The energy big data governance method based on digital twins according to claim 1, characterized in that, Before the step of real-time acquisition of vibration data of the wind turbine main bearing, the following steps are included: acquiring reference vibration data of the wind turbine main bearing under normal operating conditions, and initializing an inertial baseline for each frequency scale component of the reference vibration data; initializing the background noise range of each corresponding background noise range in the background noise feature reference table according to the inertial baseline of each frequency scale component; and removing the background noise signal from each frequency scale component according to the preset background noise feature reference table to obtain multiple denoised frequency scale components. Before the step of periodically performing the following operations at preset time intervals, the inertial baseline of each frequency scale component is updated according to preset rules and the real-time vibration energy of each frequency scale component in the currently acquired vibration data of the wind turbine main bearing.

7. The energy big data governance method based on digital twins according to claim 6, characterized in that, Before updating the inertial baseline of each frequency scale component based on preset rules and the real-time vibration energy of each frequency scale component in the currently acquired vibration data of the wind turbine main bearing, the method further includes: calculating the deviation of the real-time vibration energy of each frequency scale component in the currently acquired vibration data of the wind turbine main bearing from the corresponding inertial baseline, and the deviation velocity; comparing each deviation with a preset short-term fluctuation threshold, and comparing the deviation velocity with a preset velocity threshold; when the deviation of a certain frequency scale component is greater than the short-term fluctuation threshold, the duration is greater than a first preset time, and the deviation velocity is greater than the preset velocity threshold, the real-time vibration energy of the corresponding frequency scale component is determined to be a main bearing wear and degradation signal; otherwise, the real-time vibration energy of the corresponding frequency scale component is determined to be instantaneous background noise. The specific steps of updating the inertial baseline of each frequency scale component based on preset rules and the real-time vibration energy of each frequency scale component in the currently acquired vibration data of the wind turbine main bearing include: for each frequency scale component in the currently acquired vibration data of the wind turbine main bearing that is determined to be instantaneous background noise, updating the inertial baseline of the corresponding frequency scale component based on preset rules and real-time vibration energy.

8. The energy big data governance method based on digital twins according to claim 7, characterized in that, Before the step of determining that the real-time vibration energy of the corresponding frequency scale component is the main bearing wear degradation signal, the method further includes: calculating the matching degree between the spectrum of the corresponding frequency scale component in the current period and the preset main bearing wear characteristic frequency library; comparing the matching degree with the preset matching threshold; and if the matching degree is greater than the preset matching threshold, then the step of determining that the real-time vibration energy of the corresponding frequency scale component is the main bearing wear degradation signal is executed.

9. An energy big data governance system based on digital twins, characterized in that, include: The data acquisition module is used to collect vibration data of the main bearing of the wind turbine in real time. After real-time acquisition of vibration data from the wind turbine main bearing, the process includes: real-time acquisition of main bearing temperature data, output power data, and current operating condition data; a denoising module, used to decompose the vibration data into multiple frequency scale components, and remove background noise signals from each frequency scale component according to a preset background noise feature reference table, resulting in multiple denoised frequency scale components; a lookup module, used to find the denoised frequency scale component containing the wear vibration frequency of the main bearing from the multiple denoised frequency scale components, extract the vibration energy at the wear vibration frequency from the found denoised frequency scale component as the wear vibration energy feature, and calculate the confidence level of the wear vibration energy feature; after the step of calculating the confidence level of the wear vibration energy feature, the process further includes: based on the wear vibration energy feature, main bearing temperature data, output power data, and current operating condition data, evaluating the wind turbine main bearing under the current operating condition. The system analyzes the coordinated change trajectory between wear vibration energy, main bearing temperature, and output power; calculates the divergence between the coordinated change trajectory under the current operating condition and the expected correlation trajectory under the corresponding preset operating condition, and corrects the confidence level based on the magnitude of the divergence; compares the confidence level with the preset main bearing wear degradation signal threshold, using the corrected confidence level; a comparison module compares the confidence level with the preset main bearing wear degradation signal threshold, and when the confidence level exceeds the main bearing wear degradation signal threshold, uses nonlinear amplification, feature weighting, or signal reconstruction to enhance the wear vibration energy characteristics, obtaining a wear enhancement feature signal; a transmission module fuses the multiple denoised frequency scale components and the wear enhancement feature signal into refined data, and transmits the refined data to the digital twin platform, enabling the digital twin platform to adaptively adjust the wind turbine main bearing based on the refined data.

Images