Monitoring method, device and equipment of fan variable pitch system and medium

Abstract

The invention relates to the technical field of fan monitoring, in particular to a fan variable pitch system monitoring method, device and equipment and a medium. The method comprises the following steps: acquiring image data of a fan variable pitch system; performing gray processing on the image data based on an adaptive graying algorithm of parameter adjustment; filtering the grayscale image data by adopting median filtering, Gaussian filtering and wavelet filtering to obtain filtered image data; edge detection and feature and feature point recognition are carried out on the filtered image data, and feature point recognition adopts SIFT and SURF fusion algorithm recognition; and based on the identified feature points, calculating the angle of the fan variable pitch system according to a geometrical relationship algorithm, or based on the identified features, calculating the angle of the fan variable pitch system according to a template matching algorithm. According to the method, many defects of a traditional sensor monitoring mode are overcome, the operation condition of the variable-pitch system can be timely and accurately mastered, and the probability that the fan is shut down due to faults of the variable-pitch system is effectively reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of fan monitoring, and particularly to a monitoring method, device, equipment and medium for a fan pitch system. Background Art

[0002] Traditional monitoring methods for fan pitch systems mainly rely on sensor monitoring means. For example, various physical sensors such as angle sensors and displacement sensors are installed on the pitch mechanism. Although this sensor-based monitoring method can obtain relevant parameter information of the pitch system to a certain extent, there are many insurmountable defects. Firstly, the installation process of sensors is extremely complex and requires a certain degree of modification and adaptation of the original structure of the fan, which not only significantly increases the installation cost but also faces many technical challenges and safety risks during the construction process. Secondly, when sensors are in the harsh operating environment of a fan for a long time, they are extremely vulnerable to the erosion and influence of extreme environmental factors such as strong vibration of the fan, complex electromagnetic interference, high temperature, high humidity, and dust. This will inevitably lead to a gradual decline in the measurement accuracy of the sensors, a significant reduction in reliability, and even frequent failures, thus seriously affecting the accuracy, integrity, and continuity of the monitoring data. Thirdly, the monitoring dimension of the sensor monitoring method is relatively single, and only a limited number of specific parameter information can be obtained, making it difficult to comprehensively, deeply, and intuitively reflect the overall operating state of the pitch system and potential various fault hazards, and unable to provide sufficient data support and decision-making basis for the refined operation and maintenance of the fan. Summary of the Invention

[0003] In view of this, the present invention provides a monitoring method, device, equipment and medium for a fan pitch system to solve one of the problems existing in the prior art when using sensors to monitor the fan pitch system.

[0004] In a first aspect, the present invention provides a monitoring method for a fan pitch system, the method comprising: collecting image data of the fan pitch system; performing gray processing on the image data based on an adaptive graying algorithm with parameter adjustment to obtain grayed image data; performing filtering on the grayed image data using median filtering, Gaussian filtering, and wavelet filtering to obtain filtered image data; performing edge detection and feature and feature point recognition on the filtered image data, and the feature point recognition is performed using a fusion algorithm of SIFT and SURF; calculating the angle of the fan pitch system based on the recognized feature points according to a geometric relationship algorithm, or calculating the angle of the fan pitch system based on the recognized features according to a template matching algorithm.

[0005] The monitoring method of the wind turbine pitch system provided by the embodiments of the present invention monitors and calculates the angle by collecting and processing the images of the wind turbine pitch system, overcomes many disadvantages of the traditional sensor monitoring method, can timely and accurately grasp the operating conditions of the pitch system, discover potential faults in advance and take corresponding measures, effectively reduce the probability of the wind turbine stopping due to the pitch system failure, ensure the safe and stable operation of the wind turbine, improve the reliability and power generation efficiency of the wind power generation system, reduce the operation and maintenance costs, and has significant economic and social benefits.

[0006] In an optional implementation manner, before collecting the image data of the wind turbine pitch system, the method further includes: setting multiple groups of observation points in the three-dimensional model of the wind turbine pitch system established in advance; determining the evaluation parameters corresponding to each group of observation points according to the image data observed at each group of observation points, and the image data includes the image data observed when the wind turbine operates under different working conditions; screening out the preliminary candidate observation points from multiple groups of observation points according to the evaluation parameters; arranging the image acquisition device in the wind turbine pitch system according to the preliminary candidate observation points; adjusting the positions of the preliminary candidate observation points based on the comparison result between the image data collected by the image acquisition device and the image data observed at the preliminary candidate observation points; determining the final observation points based on the adjusted preliminary candidate observation points for arranging the image acquisition device, and the image acquisition device is used to collect the image data of the wind turbine pitch system, and the image acquisition device is an industrial-grade CCD or an industrial-grade CMOS.

[0007] In the present invention, an industrial-grade CCD or CMOS camera is used as the image acquisition device, and multiple final observation points are determined through simulation experiments and on-site tests, realizing the all-round and non-blind-spot shooting of the key components of the pitch system and the multi-angle capture of the dynamic changes of the transmission components of the pitch mechanism. Compared with the traditional sensor-based monitoring, it changes the way of obtaining monitoring data and solves the problems of complex installation, susceptibility to environmental influence and single monitoring dimension of traditional sensors.

[0008] In an optional implementation manner, collecting the image data of the wind turbine pitch system includes: collecting the image data of the wind turbine pitch system based on the pre-set image acquisition device and light source; determining the shadow distribution in the image data by using an image processing algorithm; adjusting the light emission intensity and angle of the light source according to the environmental light intensity and shadow distribution; collecting the image data of the wind turbine pitch system based on the pre-set image acquisition device and the adjusted light source.

[0009] In the present invention, the set light source can automatically adjust the light emission intensity, angle and irradiation range according to the real-time light conditions of the wind turbine operating environment, overcomes the influence of unstable light conditions in the wind turbine working environment on image acquisition, avoids shadow interference, improves the image quality, and provides a strong guarantee for subsequent pitch angle recognition.

[0010] In an alternative embodiment, an adaptive grayscale algorithm based on parameter adjustment is used to perform grayscale processing on image data to obtain grayscale image data, including: detecting edge information in the image data based on the Canny edge detection algorithm combined with multi-scale morphological gradient operation; calculating the texture features of the image data; performing grayscale processing on the image data based on the adaptive grayscale algorithm and the adaptive threshold algorithm with dynamically adjusted grayscale conversion parameters to obtain grayscale image data, and the grayscale conversion parameters are dynamically adjusted based on the edge information and texture features.

[0011] In the present invention, an adaptive grayscale algorithm is used to convert a color image into a grayscale image, and the grayscale conversion parameters are automatically adjusted according to the image content and features, while retaining the basic structural information, edge details and texture features of the image and compressing the data volume. Compared with the traditional fixed-parameter grayscale method, the processing speed and real-time performance of the entire monitoring system are improved.

[0012] In an alternative embodiment, edge detection and feature and feature point recognition are performed on the filtered image data, including: detecting edge information in the image data based on the Canny edge detection algorithm combined with multi-scale morphological gradient operation and the local adaptive threshold algorithm, and the threshold in the local adaptive threshold algorithm is adjusted based on the local statistical information, edge intensity and texture features of the image data; respectively using the SIFT algorithm and the SURF algorithm to detect feature points in the image data and merging them; based on the merged feature points, fusing the SIFT feature descriptors and the SURF feature descriptors calculated respectively to obtain a fused feature descriptor; screening the feature points according to the local geometric features and texture features of the wind turbine pitch system, and obtaining the recognized feature points based on the screened feature points and feature descriptors; using a deep convolutional neural network to extract the features of the image data.

[0013] In the present invention, the improved Canny edge detection algorithm is used for edge detection, introducing multi-scale morphological gradient operation to enhance edge contrast and coherence, and combining with an adaptive threshold selection strategy to automatically adjust the threshold according to the local region features of the image, avoiding inaccurate or broken edge detection caused by noise, uneven illumination and blurred component edges in the traditional algorithm, and providing an accurate geometric basis for the pitch angle calculation based on edge geometric features; for feature point recognition, a fused feature point extraction algorithm is proposed, organically combining the advantages of the SIFT and SURF algorithms, comprehensively using their detection capabilities for different scale and type feature points in the feature point detection stage, constructing a hybrid feature descriptor in the feature descriptor generation stage, fusing the scale invariance of the SIFT algorithm and the rapidity and robustness of the SURF algorithm, and also introducing a local feature constraint mechanism to screen and optimize the feature points, solving the problems of inaccurate feature point extraction and insufficient robustness of the feature descriptor in the traditional feature extraction algorithm, and improving the accuracy and reliability of the monitoring system.

[0014] In an alternative embodiment, based on the identified feature points, the angle of the wind turbine pitch system is calculated according to the geometric relationship algorithm, including: training a deep neural network model based on the calibration image and camera parameters to obtain a camera parameter prediction model; matching the identified feature points with the pre-constructed geometric model of the wind turbine pitch system to obtain a matching result; based on the matching result, using the perspective transformation algorithm and the camera parameters predicted by the camera parameter prediction model to convert the two-dimensional feature points into three-dimensional geometric points; calculating the angle of the wind turbine pitch system based on the three-dimensional geometric points and trigonometric functions.

[0015] In the present invention, when calculating the pitch angle based on the geometric relationship, an automatic calibration and optimization method for camera parameters based on deep learning is proposed. By collecting a large number of calibration images and corresponding real camera parameter data to train the deep neural network model, the automatic prediction and real-time optimization adjustment of camera parameters are realized. Combining the perspective transformation algorithm and camera parameters to convert two-dimensional features into three-dimensional geometric points to calculate the pitch angle improves the parameter accuracy and environmental adaptability in the geometric relationship calculation method.

[0016] In an alternative embodiment, based on the identified features, the angle of the wind turbine pitch system is calculated according to the template matching algorithm, including: matching the features with the image features of the template images in the preset template library, where the preset template library includes multiple template images, image features, and corresponding pitch angles, and the preset template library is dynamically updated; determining the angle of the wind turbine pitch system according to the matching result, and the matching result includes a single template image or multiple template images that match successfully. When there are multiple template images, the angle of the wind turbine pitch system is the fusion value of the pitch angles corresponding to the multiple template images.

[0017] In the present invention, when calculating the pitch angle based on template matching, a dynamically updated template image library is constructed, enabling the template image library to adapt to different operating states and environmental conditions, thereby enhancing the robustness of the system. At the same time, a multi-template fusion matching strategy is introduced to solve the problem that traditional template matching methods are difficult to adapt to the image changes of the wind turbine pitch system, providing a new, efficient, and accurate solution for the angle calculation of the wind turbine pitch system.

[0018] In a second aspect, the present invention provides a monitoring device for a wind turbine pitch system. The device includes: an image acquisition module for acquiring image data of the wind turbine pitch system; a grayscale processing module for performing grayscale processing on the image data based on an adaptive grayscale algorithm with parameter adjustment to obtain grayscale image data; a filtering module for filtering the grayscale image data using median filtering, Gaussian filtering, and wavelet filtering to obtain filtered image data; a feature recognition module for performing edge detection and feature and feature point recognition on the filtered image data, and the feature point recognition uses a fusion algorithm of SIFT and SURF; an angle calculation module for calculating the angle of the wind turbine pitch system based on the recognized feature points according to a geometric relationship algorithm, or calculating the angle of the wind turbine pitch system based on the recognized features according to a template matching algorithm.

[0019] In a third aspect, the present invention provides a computer device, including: a memory and a processor, which are communicatively connected to each other. The memory stores computer instructions, and the processor executes the computer instructions to execute the monitoring method of the wind turbine pitch system in the first aspect or any corresponding embodiment thereof.

[0020] In a fourth aspect, the present invention provides a computer-readable storage medium, on which computer instructions are stored. The computer instructions are used to cause a computer to execute the monitoring method of the wind turbine pitch system in the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for use in the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0022] Figure 1 is a flowchart of the monitoring method of the wind turbine pitch system according to an embodiment of the present invention; Figure 2 is a structural block diagram of the monitoring device for the wind turbine pitch system according to an embodiment of the present invention; Figure 3 is a schematic hardware structure diagram of the computer device according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0024] According to an embodiment of the present invention, an embodiment of a monitoring method for a wind turbine pitch system is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0025] In this embodiment, a monitoring method for a wind turbine pitch system is provided, which can be used in electronic devices such as computers, mobile phones, and tablet computers. Figure 1 is a flowchart of the monitoring method for the wind turbine pitch system according to an embodiment of the present invention, as Figure 1 shown, and the process includes the following steps: Step S101, collect image data of the wind turbine pitch system. Specifically, when the wind turbine pitch system is working, it can adjust the blade angle according to the wind speed and other conditions to achieve the best power generation efficiency. Its specific working process can be implemented with reference to the related technology and will not be elaborated here. In this embodiment, the monitoring of the wind turbine pitch system is mainly the monitoring and calculation of this angle, so as to assist the safe and stable operation of the wind turbine pitch system.

[0026] When monitoring the wind turbine pitch system, in this embodiment, an image acquisition device is set in the wind turbine pitch system to collect image data of the wind turbine pitch system for monitoring. Among them, an industrial-grade CCD or industrial-grade CMOS can be used for the image acquisition device. Compared with traditional cameras, such cameras have ultra-high resolution and can capture extremely fine structural features and component details in the pitch system with extremely high clarity. Even the slightest wear marks on the blade surface or the slightest loosening signs at the pitch mechanism connection can be accurately imaged. Its excellent low-light performance can easily adapt to the complex and changeable lighting conditions in the wind turbine operation environment. Whether it is in the dim dawn or evening, or on cloudy or foggy days with insufficient light, or in a completely dark environment at night, it can stably and clearly obtain images. The fast frame rate ensures that in the dynamic operation process of the high-speed rotation of the wind turbine and frequent pitch movements, a series of image sequences can be continuously and accurately captured without missing any key moment, thus providing a rich and complete data source for the real-time status monitoring and analysis of the pitch system.

[0027] Step S102: Perform grayscale processing on the image data based on the adaptive grayscale algorithm with parameter adjustment to obtain the grayscale image data. Specifically, after the image data is collected, the collected image data is a color image. If it is directly processed, it may lead to a large amount of data, high computational complexity, and low processing efficiency. Therefore, the collected image data is first grayscaled to reduce the amount of data.

[0028] Among them, when performing grayscale processing on the image data, this embodiment adopts an adaptive grayscale algorithm. When performing grayscale processing, the grayscale conversion parameters adopted by this algorithm can be adaptively adjusted according to the characteristics of the image data. Thus, while maximizing the retention of image information, a significant reduction in the amount of data can be achieved.

[0029] Step S103: Filter the grayscale image data using median filtering, Gaussian filtering, and wavelet filtering to obtain the filtered image data. Specifically, during the operation of the fan, vibrations, strong electromagnetic interference, and the influence of surrounding environmental factors will introduce various types of noise into the collected image data, seriously affecting the image quality and the accuracy of subsequent analysis results. In this embodiment, a hybrid filtering method is used to filter the noise data in the image data. This hybrid filtering combines the advantages of median filtering, Gaussian filtering, and wavelet filtering and can accurately remove different types and intensities of noise. For the impulse noise caused by fan vibration, median filtering can effectively remove the noise points while maintaining the clarity of the image edges; for the Gaussian noise generated by electromagnetic interference, Gaussian filtering can perform smoothing processing according to the distribution characteristics of the noise; and for the fine texture noise and other complex noise components existing in the image, wavelet filtering can decompose and reconstruct the image at different scales, accurately removing the noise and retaining the important details of the image. Through this hybrid filtering strategy, the clarity and signal-to-noise ratio of the image can be significantly improved, providing reliable image data guarantee for subsequent accurate feature extraction and angle calculation.

[0030] Among them, when filtering the image data, first use median filtering to remove impulse noise, then use Gaussian filtering to smooth the image, and finally use wavelet filtering to remove fine texture noise and other complex noise components. During the filtering process, the filtering parameters can also be dynamically adjusted according to the specific content and characteristics of the image data to determine to retain more detail information while removing the noise. In addition, after filtering, the filtering effect can be fed back in real time, and the filtering strategy can be dynamically adjusted to improve the image quality.

[0031] Step S104: Perform edge detection, feature, and feature point recognition on the filtered image data. The feature point recognition uses a fusion algorithm of SIFT and SURF. Specifically, for the grayscale and filtered image data, first perform edge detection on it to make the edges of the image data clearer, thereby providing an accurate geometric basis for the subsequent calculation of the pitch angle. After performing edge detection on the image data, extract the features and feature points in the image data to provide a data basis for the subsequent angle calculation. Among them, when performing feature point recognition, this embodiment uses a fusion feature point extraction algorithm, that is, uses the SIFT algorithm and the SURF algorithm to fuse and recognize feature points. Thus, in the feature point detection stage, the detection capabilities of the two algorithms for different scales and different types of feature points can be comprehensively utilized to improve the detection quantity and quality of feature points; in the feature descriptor generation stage, by constructing a hybrid feature descriptor, the scale invariance of the SIFT algorithm and the rapidity and robustness of the SURF algorithm are fused. For feature recognition, a neural network model can be used to implement it.

[0032] Step S105: Based on the recognized feature points, calculate the angle of the wind turbine pitch system according to the geometric relationship algorithm, or, based on the recognized features, calculate the angle of the wind turbine pitch system according to the template matching algorithm. In this embodiment, two methods can be used to implement the angle calculation. Among them, when using geometric relationship calculation, use the recognized feature points, map them to the three-dimensional model of the wind turbine pitch system, and then implement the angle calculation based on the geometric relationship between the feature points. When using the template matching algorithm to calculate the angle, match the recognized features with the image features in the template library, and determine the angle based on the matching result.

[0033] The monitoring method of the wind turbine pitch system provided by the embodiment of the present invention monitors and calculates the angle by collecting and processing images of the wind turbine pitch system, overcomes many drawbacks of the traditional sensor monitoring method, and can timely and accurately grasp the operating conditions of the pitch system, discover potential faults in advance and take corresponding measures, effectively reducing the probability of the wind turbine stopping due to pitch system faults, ensuring the safe and stable operation of the wind turbine, improving the reliability and power generation efficiency of the wind power generation system, reducing the operation and maintenance cost, and having significant economic and social benefits.

[0034] Among them, during the image processing process, compared with the traditional fixed-parameter grayscale method, the adaptive grayscale algorithm can automatically adjust the grayscale conversion parameters according to the image content. While retaining the key information, it greatly compresses the data volume, reduces the computational complexity, improves the processing speed and real-time performance, solves the problems of large data volume and low processing efficiency in traditional image analysis methods, and enables the entire monitoring system to respond more quickly to the changes in the pitch system state. When filtering the image, by combining the advantages of median filtering, Gaussian filtering, and wavelet filtering, various noises generated during the operation of the fan are accurately removed, the image clarity and signal-to-noise ratio are improved, reliable image data is provided for subsequent feature extraction and angle calculation, and the problem that traditional single filtering methods cannot effectively remove complex noises and affect the accuracy of subsequent analysis is avoided, ensuring the stability and reliability of the monitoring system.

[0035] In this embodiment, a monitoring method for a fan pitch system is provided, and the method includes the following steps: Step S201, set multiple groups of observation points in the pre-established three-dimensional model of the fan pitch system; determine the evaluation parameters corresponding to each group of observation points according to the image data observed at each group of observation points, where the image data includes the image data observed when the fan operates under different working conditions; screen out the preliminary candidate observation points from multiple groups of observation points according to the evaluation parameters.

[0036] Specifically, in order to enable the collected image data to comprehensively capture each component of the fan pitch system, it is first necessary to determine the arrangement position of the image acquisition device. In this embodiment, the arrangement position is determined by means of simulation tests and on-site tests.

[0037] Among them, during the simulation test, first establish a high-precision three-dimensional model of the fan pitch system; this three-dimensional model can be established using professional computer-aided design software such as CAD software. When establishing the three-dimensional model, according to the actual design parameters of the fan, construct a high-precision three-dimensional model including the blade, hub, pitch mechanism, etc. The shape, size, material properties of each component and the connection relationship between them are reflected in the model, providing accurate basic data for subsequent simulation analysis.

[0038] In addition, in order to evaluate the observation points during the simulation test, it is necessary to first determine the observation target and observation parameters. For the observation target, it is necessary to include the key components and operating state information of the fan pitch system, such as the looseness of the connection part between the blade and the hub, the movement trajectory and angle change of the transmission components of the pitch mechanism, etc. The observation parameters can be parameters used to evaluate the observation effect, such as the coverage rate of the observation area, the clarity of the key component features, the distinguishability under different operating states, etc.

[0039] For multiple sets of observation points, they can be determined in advance. For example, based on the analysis of the wind turbine pitch system, relevant staff can make the determination. Among them, each set of observation points includes multiple observation points distributed at multiple positions of the wind turbine pitch model. Through the multiple observation points in each set of observation points, the complete monitoring of the wind turbine pitch system can be achieved. Thus, during the simulation test, each set of observation points can be arranged in the 3D model in sequence. In this way, the evaluation of each set of observation points can be realized. After arranging a set of observation points in the 3D model, the operation of the wind turbine under various typical working conditions can be simulated, including different wind speeds, wind directions, pitch angles, and lighting conditions, etc. The reflection and refraction of light on the surface of the wind turbine components can be simulated through the ray tracing algorithm, as well as the image data captured by the image acquisition device at different observation point positions.

[0040] For the image data obtained after arranging each set of observation points, the pre-determined observation objectives and observation parameters can be used for evaluation. For example, parameters such as the coverage of the observation area in the image data, the clarity of the key component features, and the recognizability under different operating states can be calculated; then, by using data analysis algorithms and visualization techniques, a set of observation points with better performance in each evaluation parameter can be found as the preliminary candidate observation points. For example, the clustering analysis method can be used to group multiple sets of observation points according to similarity, and then a representative and relatively high comprehensive score of evaluation parameters in each group can be selected as the preliminary candidate observation points.

[0041] Step S202, arrange the image acquisition device in the wind turbine pitch system according to the preliminary candidate observation points; adjust the position of the preliminary candidate observation points based on the comparison result between the image data collected by the image acquisition device and the image data observed at the preliminary candidate observation points; determine the final observation points based on the adjusted preliminary candidate observation points for the arrangement of the image acquisition device, and the image acquisition device is used to collect the image data of the wind turbine pitch system.

[0042] Specifically, after determining the preliminary candidate observation points through the simulation test, further adjustment is still needed by means of on-site testing. Among them, during the on-site testing, the preliminary installation of observation devices such as the image acquisition device and lighting equipment is carried out on the actual wind turbine according to the preliminary candidate observation points. During the installation process, the stability and safety of the equipment should be ensured, and at the same time, preliminary debugging should be carried out to ensure that the equipment can work normally and obtain clear image data.

[0043] During the actual operation of the wind turbine, the image data of the wind turbine pitch system under different working conditions is collected by using the installed image acquisition device. At the same time, combined with other traditional monitoring means (such as sensor monitoring data) and the actual situation records on site, including the operating parameters of the wind turbine, environmental conditions (such as light intensity, temperature, humidity, wind speed, etc.), and the actual space limitations of the equipment installation location, etc., comprehensive data support is provided for subsequent data analysis.

[0044] Compare and analyze the image data collected on-site with the simulation test results in step S101 to evaluate the differences between the actual observation effects and the expected goals. Among them, key attention can be paid to the observation points that performed well in the simulation test but had problems in the actual on-site test. The possible reasons for analysis may include that interference factors in the actual environment were not fully considered in the simulation, the influence of equipment installation accuracy, the influence of small vibrations during the operation of the fan on the observation, etc. Further optimize and adjust the positions of the observation points and equipment parameters according to the analysis results. For example, finely adjust the angles, heights or distances of the observation points, and adjust the luminous intensity and angles of the lighting equipment.

[0045] After multiple rounds of on-site tests, data analysis and optimization adjustments, select the combination of observation points that can stably, clearly and comprehensively observe the key components and operating states of the pitch system during actual operation, and all evaluation parameters reach the optimal or nearly optimal level, and determine them as the final multiple observation points. These final observation points will be used as the formal installation positions of the subsequent long-term monitoring system, providing a reliable hardware foundation for the precise image monitoring of the fan pitch system.

[0046] Thus, through a large number of simulation experiments and on-site tests in this embodiment, multiple optimal observation points are innovatively determined. These installation positions can not only clearly capture the key components of the pitch system, such as the connection part between the blade and the hub, in all directions without dead angles, but also capture the dynamic changes of the transmission components of the pitch mechanism from different angles during operation. Through multi-view image acquisition, more comprehensive component spatial position information can be obtained, laying a solid foundation for subsequent precise three-dimensional reconstruction and angle calculation.

[0047] Step S203, collect the image data of the fan pitch system.

[0048] Specifically, the above step S203 includes: Step S2031, collect the image data of the fan pitch system based on the pre-set image acquisition device and light source.

[0049] Step S2032, use image processing algorithms to determine the shadow distribution in the image data.

[0050] Step S2033, adjust the luminous intensity and angle of the light source according to the ambient light intensity and shadow distribution.

[0051] Step S2034, collect the image data of the fan pitch system based on the pre-set image acquisition device and the adjusted light source.

[0052] Specifically, considering the highly complex working environment of the wind turbine and the extremely unstable lighting conditions, in addition to arranging an image acquisition device at the determined final observation point position in this embodiment, a light source such as an LED light source can also be arranged as the lighting device during image acquisition. During the acquisition process of image data, the luminous intensity and angle of the light source can be adaptively adjusted through a control device. For example, according to the real-time lighting intensity, angle, and shadow distribution of the wind turbine operating environment, the luminous intensity, angle, and irradiation range are automatically adjusted, thereby providing a highly stable, uniform, and shadow-free lighting effect. During the operation of the wind turbine, whether it is the complex shadow area formed due to the blade occlusion under direct sunlight, the low-light environment caused by harsh weather such as strong winds, heavy rains, and dust storms, or the completely dark working scenario at night, through the intelligent adjustment of the light source, it is ensured that the image acquisition area is always in the best lighting state, effectively avoiding any adverse effects of shadows on the image quality. High-quality, shadow-free images greatly improve the accuracy and reliability of subsequent pitch angle recognition, providing a strong guarantee for the high-precision operation of the entire monitoring system.

[0053] Among them, an adaptive adjustment algorithm can be used to intelligently adjust the light source. First, a photosensitive resistor or a photodiode can be used as a light intensity sensor to measure the ambient light intensity in real time. These sensors convert the light signal into an electrical signal, and the analog signal is converted into a digital signal through an analog-to-digital converter (ADC) and sent to the control device for processing. Then, for the image data collected by the image acquisition device, an image processing algorithm is used to detect the shadow area in the image. Specifically, the Canny edge detection algorithm can be used, combined with multi-scale morphological gradient operations, to enhance the edge contrast and coherence and detect the shadow edges in the image; then the image is segmented into multiple regions, and the shadow distribution is determined by calculating the shadow area of each region.

[0054] During the adjustment process of the light source, the control device can use pulse width modulation technology to adjust the luminous intensity of the light source. Specifically, according to the output of the photosensitive sensor, the duty cycle of the PWM signal is dynamically adjusted to control the brightness of the LED. In addition, the control device can also use a small stepping motor or a servo motor to control the angle and irradiation range of the light source. Specifically, according to the shadow distribution, a control signal is sent to the motor to adjust the rotation angle of the motor, realizing the adjustment of the angle and irradiation range of the light source. During the adjustment process, the control device can continuously obtain the ambient light intensity and the shadow distribution of the image data and adjust the light source in real time to ensure that the light source is always in the best state. In addition, a specific adjustment strategy can be pre-determined for adjusting the rotation angle of the motor according to the shadow distribution, and the adjustment can be directly made according to the adjustment strategy during adjustment. This adjustment strategy can also be adjusted according to the shadow distribution obtained in real time to optimize the lighting effect.

[0055] Step S204, perform grayscale processing on the image data based on the adaptive grayscale algorithm with parameter adjustment to obtain the grayscale image data.

[0056] Specifically, the above - mentioned step S204 includes: Step S2041, detect the edge information in the image data based on the Canny edge detection algorithm combined with multi - scale morphological gradient operation. Specifically, use the Canny edge detection algorithm combined with multi - scale morphological gradient operation to enhance the edge contrast and coherence and detect the edge information in the image.

[0057] Step S2042, calculate the texture features of the image data. Specifically, the Local Binary Patterns (LBP) or Gray - level co - occurrence matrix (GLCM) can be used to analyze the texture information of the image.

[0058] Step S2043, perform grayscale processing on the image data based on the adaptive grayscale algorithm with dynamically adjusted grayscale conversion parameters and the adaptive threshold algorithm to obtain the grayscale image data, where the grayscale conversion parameters are dynamically adjusted based on the edge information and texture features. Specifically, when adjusting the grayscale conversion parameters according to the edge information and texture features, it can be adjusted in the following way. For example, for the edge area, the weight of the green channel can be increased to retain more detailed information; for the texture area, the weight of the blue channel can be appropriately adjusted.

[0059] Among them, when performing adaptive grayscale processing using the grayscale conversion parameters, the weighted average method, maximum value method, minimum value method, or average value method combined with the grayscale conversion parameters can be used to calculate the grayscale value. In addition, during the grayscale processing, an adaptive threshold algorithm such as OSTU (Otsu algorithm) can be used to dynamically calculate the threshold of each area of the image data, so as to ensure that the grayscale image retains details while reducing noise and blurring phenomena.

[0060] The adaptive grayscale algorithm adopted by the present invention can automatically adjust the grayscale conversion parameters according to the content and features of the image, and while maximizing the retention of the basic structure information, edge details, and texture features of the image, it realizes a large - scale compression of the data volume. Compared with the traditional fixed - parameter grayscale method, it can better adapt to the differences in the colors, materials, and light reflection characteristics of different components in the wind turbine pitch system image, provides a high - quality and low - data - volume image basis for subsequent edge detection and feature extraction operations, and significantly improves the processing speed and real - time performance of the entire monitoring system.

[0061] Step S205, perform filtering on the grayscale image data using median filtering, Gaussian filtering, and wavelet filtering to obtain the filtered image data; for details, please refer to Figure 1 Step S103 of the embodiment shown, which will not be elaborated here.

[0062] Step S206, perform edge detection and feature and feature point recognition on the filtered image data, and use the SIFT and SURF fusion algorithm for feature point recognition. For details, please refer to Figure 1 Step S104 of the embodiment shown, which will not be elaborated here.

[0063] Step S207, based on the recognized feature points, calculate the angle of the wind turbine pitch system according to the geometric relationship algorithm, or based on the recognized features, calculate the angle of the wind turbine pitch system according to the template matching algorithm. For details, please refer to Figure 1 Step S105 of the embodiment shown, which will not be elaborated here.

[0064] In this embodiment, a monitoring method for a wind turbine pitch system is provided, and the method includes the following steps: Step S301, collect the image data of the wind turbine pitch system; for details, please refer to Figure 1 Step S101 of the embodiment shown, which will not be elaborated here.

[0065] Step S302, perform gray processing on the image data using the adaptive graying algorithm based on parameter adjustment to obtain the grayscale image data; for details, please refer to Figure 1 Step S102 of the embodiment shown, which will not be elaborated here.

[0066] Step S303, perform filtering on the grayscale image data using median filtering, Gaussian filtering, and wavelet filtering to obtain the filtered image data; for details, please refer to Figure 1 Step S103 of the embodiment shown, which will not be elaborated here.

[0067] Step S304, perform edge detection and feature and feature point recognition on the filtered image data, and use the SIFT and SURF fusion algorithm for feature point recognition.

[0068] Specifically, the above Step S304 includes: Step S3041, detect the edge information in the image data based on the Canny edge detection algorithm combined with multi-scale morphological gradient operation and local adaptive threshold algorithm, and the threshold in the local adaptive threshold algorithm is adjusted based on the local statistical information, edge intensity, and texture features of the image data.

[0069] Among them, when the traditional Canny edge detection algorithm processes the image of the wind turbine pitch system, inaccurate edge detection or edge breakage may occur due to problems such as noise, uneven illumination, and blurred component edges in the image. In this embodiment, multi-scale morphological gradient operation is introduced during edge detection, which can perform gradient calculation on the image at different scales, enhancing the contrast and coherence of the edges. Specifically, when using multi-scale morphological gradient operation, combined with structural elements of different scales (such as circular or rectangular structural elements of different sizes), multiple gradient calculations are performed on the image. In this way, it can better adapt to edges of different thicknesses and complexities and enhance the coherence of the edges.

[0070] The multi-scale morphological gradient operation can be implemented using the following process: import cv2 import numpy as np import matplotlib.pyplot as plt def multiscale_morphological_gradient(image, scales=[3, 5, 7]): gradients = [] for scale in scales: kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (scale, scale)) gradient = cv2.morphologyEx(image, cv2.MORPH_GRADIENT, kernel) gradients.append(gradient) return np.max(gradients, axis=0) # Read the image image = cv2.imread('path_to_image.jpg', cv2.IMREAD_GRAYSCALE) # Multi-scale morphological gradient operation gradient_image = multiscale_morphological_gradient(image) # Display the result plt.figure(figsize=(12, 6)) plt.subplot(1, 2, 1) plt.imshow(image, cmap='gray') plt.title('Original Image') plt.axis('off') plt.subplot(1, 2, 2) plt.imshow(gradient_image, cmap='gray') plt.title('Multiscale Morphological Gradient') plt.axis('off') plt.show() In addition, when using the Canny edge detection algorithm combined with multiscale morphological gradient operation for edge detection, a local adaptive threshold algorithm can also be used to adjust the parameters of the Canny algorithm (such as the low threshold and high threshold), further optimizing the edge detection results. This local adaptive threshold algorithm automatically adjusts the threshold according to the characteristics of the local region of the image, avoiding the problems of edge loss or false detection caused by a globally fixed threshold. Thus, through this improved edge detection algorithm, the edge contours of the pitch system components can be detected more accurately and completely. Whether it is the thin edge of the blade or the edge of the complex shape of the pitch mechanism, they can be clearly outlined, providing a highly accurate geometric basis for the subsequent calculation of the pitch angle based on the edge geometric features.

[0071] Among them, the implementation of the Canny edge detection algorithm can refer to the following process: def canny_edge_detection(image, low_threshold=50, high_threshold=150): return cv2.Canny(image, low_threshold, high_threshold) # Apply Canny edge detection edges = canny_edge_detection(gradient_image) # Display the results plt.figure(figsize=(12, 6)) plt.subplot(1, 2, 1) plt.imshow(gradient_image, cmap='gray') plt.title('Gradient Image') plt.axis('off') plt.subplot(1, 2, 2) plt.imshow(edges, cmap='gray') plt.title('Canny Edge Detection') plt.axis('off') plt.show() Specifically, when using the local adaptive threshold algorithm for threshold adjustment, it can be adjusted in combination with the local statistical information of the image (such as mean, variance) and the image content information such as edge intensity and texture features. Among them, when adjusting the threshold according to the image content information, for regions with higher edge intensity, the threshold can be appropriately reduced; for regions with complex texture features, the threshold can be appropriately increased. This can better adapt to images with different lighting conditions and noise levels and improve the edge detection effect.

[0072] Among them, the process of adjusting the threshold using local statistical information can be implemented in the following way: def adaptive_threshold(image, block_size=11, C=2): # Calculate local mean local_mean = cv2.blur(image, (block_size, block_size)) # Calculate local variance local_var = cv2.blur(image^2, (block_size, block_size)) - local_mean^2 # Calculate threshold threshold = local_mean + C×np.sqrt(local_var) # Apply threshold binary_image = np.where(image>threshold, 255, 0).astype(np.uint8) return binary_image # Read image image = cv2.imread('path_to_image.jpg', cv2.IMREAD_GRAYSCALE) # Apply local adaptive thresholding binary_image = adaptive_threshold(image) # Display the result plt.figure(figsize=(12, 6)) plt.subplot(1, 2, 1) plt.imshow(image, cmap='gray') plt.title('Original Image') plt.axis('off') plt.subplot(1, 2, 2) plt.imshow(binary_image, cmap='gray') plt.title('Adaptive Thresholding') plt.axis('off') plt.show() When adjusting the threshold by combining local statistical information of the image (such as mean, variance) and image content information such as edge intensity and texture features, the following process can be referred to for implementation: def content_based_threshold(image, block_size=11, C=2): # Calculate local mean local_mean = cv2.blur(image, (block_size, block_size)) # Calculate local variance local_var = cv2.blur(image ^ 2, (block_size, block_size)) - local_mean ^ 2 # Calculate edge intensity edges = cv2.Canny(image, 50, 150) edge_strength = cv2.blur(edges, (block_size, block_size)) # Calculate texture features texture = cv2.Laplacian(image, cv2.CV_64F) texture_strength = cv2.blur(np.abs(texture), (block_size, block_size)) # Calculate the threshold threshold = local_mean + C × np.sqrt(local_var - edge_strength +texture_strength) # Apply the threshold binary_image = np.where(image>threshold, 255, 0).astype(np.uint8) return binary_image # Apply content - based threshold adjustment binary_image = content_based_threshold(image) # Display the result plt.figure(figsize=(12, 6)) plt.subplot(1, 2, 1) plt.imshow(image, cmap='gray') plt.title('Original Image') plt.axis('off') plt.subplot(1, 2, 2) plt.imshow(binary_image, cmap='gray') plt.title('Content - Based Adaptive Thresholding') plt.axis('off') plt.show() Step S3042: Detect and merge feature points in the image data using the SIFT algorithm and the SURF algorithm respectively; specifically, when performing feature point recognition, use the SIFT algorithm and the SURF algorithm to detect feature points in the image data respectively, then merge the detected feature points and remove duplicate feature points. Among them, the process of detecting feature points in the image data using the SIFT algorithm and the SURF algorithm can be implemented according to the following process: def combined_feature_detection(image): keypoints_sift, descriptors_sift = sift_feature_detection(image) keypoints_surf, descriptors_surf = surf_feature_detection(image) # Merge feature points keypoints = keypoints_sift + keypoints_surf # Remove duplicate feature points keypoints = list(set(keypoints)) return keypoints, (descriptors_sift, descriptors_surf) In some specific cases, the traditional SIFT and SURF algorithms may have problems such as inaccurate feature point extraction or insufficient robustness of feature descriptors. In this embodiment, the advantages of the SIFT and SURF algorithms are organically combined. In the feature point detection stage, the detection capabilities of the two algorithms for different scales and different types of feature points are comprehensively utilized to improve the detection quantity and quality of feature points.

[0073] Step S3043: Based on the merged feature points, fuse the separately calculated SIFT feature descriptors and SURF feature descriptors to obtain the fused feature descriptors. Specifically, for each merged feature point, calculate the SIFT and SURF descriptors respectively, and then fuse the two descriptors. The fusion method can be simple concatenation, weighted average, or other more complex fusion methods. Among them, the descriptors generated by using the SIFT or SURF algorithm are usually vectors of a fixed length. For example, the length of the SIFT descriptor is 128, and the length of the SURF descriptor is 64 or 128. The fused feature descriptors can be determined by concatenating or weighted averaging individual descriptors. The length of the fused feature descriptors generated in this way is the sum of the lengths of the SIFT and SURF descriptors (e.g., 192), or descriptors of the same length are generated by weighted averaging.

[0074] Among them, the generation of the fused feature descriptors can be implemented with reference to the following process: def combined_feature_descriptor(keypoints, descriptors_sift,descriptors_surf): combined_descriptors = [] for kp in keypoints: sift_desc = descriptors_sift[keypoints.index(kp)] surf_desc = descriptors_surf[keypoints.index(kp)] # Concatenate the descriptors combined_desc = np.concatenate((sift_desc, surf_desc)) # Or weighted average # combined_desc = 0.5 × sift_desc + 0.5 ×surf_desc combined_descriptors.append(combined_desc) return np.array(combined_descriptors) In this embodiment, the fused feature descriptors fuse the scale invariance of the SIFT algorithm and the rapidity and robustness of the SURF algorithm, improving the robustness and distinctiveness of the feature descriptors.

[0075] Step S3044, screen the feature points according to the local geometric features and texture features of the wind turbine pitch system, and obtain the recognized feature points based on the screened feature points and feature descriptors; specifically, for the recognized feature points, a local feature constraint mechanism can be introduced, and the feature points are screened and optimized according to the local geometric structure and texture features of the pitch system components to ensure that the extracted feature points have stronger recognition and stability at different pitch angles.

[0076] Among them, the local feature constraint mechanism sets constraint conditions, such as the distribution density of feature points, the distance between feature points and edges, etc., according to the local geometric structure and texture features of the pitch system components, and screens the feature points. For example, when screening according to the edge distance, the distance between the feature point and the edge should be less than a certain threshold to ensure that the feature point is near the edge. When screening according to the texture feature, the texture intensity of the feature point should be less than a certain threshold to ensure that the feature point is in a region with weak texture and reduce the influence of noise. When screening according to the distribution density, the distribution density of the feature points should be moderate, avoiding overly dense or sparse feature points and ensuring the uniform distribution of feature points.

[0077] Among them, the local feature constraint mechanism can be implemented according to the following process: def filter_keypoints(keypoints, image, edge_threshold=10, texture_threshold= 10): filtered_keypoints = [] edges = cv2.Canny(image, 50, 150) texture = cv2.Laplacian(image, cv2.CV_64F) for kp in keypoints: x, y = kp.pt # Check the distance between the feature point and the edge if edges[int(y), int(x)]>edge_threshold: continue # Check the distance between the feature point and the texture if np.abs(texture[int(y), int(x)])>texture_threshold: continue filtered_keypoints.append(kp) return filtered_keypoints Specifically, the above feature point extraction process can be implemented according to the following process: def combined_feature_extraction(image): # Feature point detection keypoints, (descriptors_sift, descriptors_surf) = combined_feature_ detection(image) # Feature descriptor generation combined_descriptors = combined_feature_descriptor(keypoints, descriptors_sift, descriptors_surf) # Local feature constraint filtering filtered_keypoints = filter_keypoints(keypoints, image) return filtered_keypoints, combined_descriptors # Read the image image = cv2.imread('path_to_image.jpg', cv2.IMREAD_GRAYSCALE) # Apply the combined feature point extraction algorithm keypoints, descriptors = combined_feature_extraction(image) # Display the result image_with_keypoints = cv2.drawKeypoints(image, keypoints, None,color=(0, 255, 0)) cv2.imshow('Keypoints', image_with_keypoints) cv2.waitKey(0) cv2.destroyAllWindows() Step S3045, use a deep convolutional neural network to extract the features of the image data. Specifically, use a deep convolutional neural network model to extract the features of the image. These features can be high-level semantic features of the image rather than simple pixel information. For example, a pre-trained CNN model (such as ResNet, VGG, etc.) can be used to extract features. Among them, for the training process of the CNN model, it can learn through a large amount of image data (including images with different illuminations, noises, and deformations), so that the model can automatically learn feature representations that are insensitive to these changes. This learning process enables the model to better recognize and match the image content when facing new images, even if the images have noise, deformation, or illumination changes.

[0078] When using a pre-trained CNN model for feature extraction, through structures such as convolutional layers, pooling layers, and activation functions, gradually extract the deep features of the image. These features include not only low-level information such as the edges and textures of the image, but also high-level information such as the shapes and structures of objects. High-level semantic features are insensitive to the noise and illumination changes of the image, so they can represent the image content more accurately.

[0079] In addition, when using a large amount of image data to train the model, the image data can be processed through data augmentation and regularization techniques to further improve the robustness of the model to noise, deformation, and illumination changes. Among them, the data augmentation process generates more diverse training samples by randomly transforming the image data (such as rotation, scaling, cropping, color adjustment, etc.), enabling the model to learn features in more varying situations. For example, by randomly adjusting the brightness and contrast of the image, the model can better adapt to different illumination conditions. Using regularization techniques (such as Dropout, L2 regularization, etc.) for the image data can prevent the model from overfitting and improve the generalization ability of the model. Regularization techniques help the model maintain good performance when facing unseen images.

[0080] Step S305, based on the identified feature points, calculate the angle of the wind turbine pitch system according to the geometric relationship algorithm, or, based on the identified features, calculate the angle of the wind turbine pitch system according to the template matching algorithm. It should be noted that when calculating the angle, either the geometric relationship algorithm or the template matching algorithm can be selected according to the specific application scenario. In addition, the two algorithms can also be used to calculate the angle respectively, and the angles calculated by the two algorithms can be averaged or weighted averaged to obtain the final angle calculation result.

[0081] Specifically, when using the geometric algorithm to calculate the angle, the above step S305 includes: Step S3051: Train a deep neural network model based on the calibration images and camera parameters to obtain a camera parameter prediction model. When calculating the pitch angle based on the geometric relationship, the internal parameters of the camera (such as focal length, principal point coordinates, etc.) and external parameters (such as installation position, attitude, etc.) need to be accurately known. However, these parameters are often difficult to accurately measure and calibrate in practical applications. Moreover, as the operation time of the wind turbine increases, the camera may undergo small displacements or attitude changes, resulting in inaccurate parameters. In this embodiment, a large number of calibration images and corresponding real camera parameter data are collected to train the deep neural network model. This model can automatically predict the internal and external parameters of the camera based on the features in the image and perform real-time optimization and adjustment.

[0082] Specifically, the calibration image is an image used to calibrate the camera parameters, usually including a calibration board with known geometric shapes and positions (such as a checkerboard or a dot array). In this embodiment, the calibration images for training can be collected in the following way: At different angles and positions of the wind turbine pitch system, use the camera to capture the calibration board images. Ensure that the calibration board covers the entire field of view and capture images under different lighting conditions to obtain diverse calibration images. The real camera parameter data includes the internal parameters of the camera (such as focal length, principal point coordinates, distortion coefficients) and external parameters (such as the installation position and attitude of the camera). In this embodiment, the initial camera parameters for training can be obtained through traditional camera calibration methods (such as Zhang Zhengyou calibration method). Then, use the deep learning model to optimize and calibrate these parameters.

[0083] When training the camera parameter prediction model, use the calibration images and real camera parameter data as the sample set for training the deep learning model. That is, through a large number of calibration images and corresponding real camera parameter data, train the deep neural network model so that the trained camera parameter prediction model can automatically predict the internal and external parameters of the camera based on the features in the image.

[0084] Step S3052: Match the identified feature points with the pre-constructed geometric model of the wind turbine pitch system to obtain a matching result. Specifically, when calculating the angle using a geometric algorithm, it is necessary to first construct a geometric model of the wind turbine pitch system, which includes information such as the geometric shape, installation position, and attitude of the wind turbine blade. When constructing the geometric model, the following geometric parameters are first defined: blade length L, blade width W, distance from the blade root to the center R, pitch axis position (x0, y0), and pitch angle θ. Then, it is assumed that the blade rotates in a two-dimensional plane and the pitch axis is located at the blade root. The geometric shape of the blade can be simplified to a rectangle, and its vertex coordinates are: (x0, y0), (x0 + Lcos(θ), y0 + Lsin(θ)), (x0 + Lcos(θ) - Wsin(θ), y0 + Lsin(θ) + Wcos(θ)), (x0 - Wsin(θ), y0 + Wcos(θ)). Thus, the geometric model of the wind turbine pitch system is obtained. After that, for the identified feature points, the FLANN matcher is used to match them with the points in the geometric model. For example, the feature points on the blade edge in the feature points are matched with the points on the blade edge in the geometric model. Among them, the specific matching process can be implemented according to the following process: def match_keypoints(keypoints1, descriptors1, keypoints2,descriptors2): flann = cv2.FlannBasedMatcher_create() matches = flann.knnMatch(descriptors1, descriptors2, k=2) good_matches = [] for m, n in matches: if m.distance<0.7 × n.distance: good_matches.append(m) return good_matches Step S3053: Based on the matching results, use the perspective transformation algorithm and the camera parameters predicted by the camera parameter prediction model to convert the two-dimensional feature points into three-dimensional geometric points. Specifically, for the two-dimensional feature points to be converted, they can be the feature points that are successfully matched after the identified feature points are matched with the geometric model. When performing the conversion, first use the camera parameter prediction model to predict the internal parameters and external parameters of the camera (i.e., the image acquisition device); then use the internal parameters of the camera to convert the pixel coordinates of the two-dimensional feature points into the normalized coordinates in the camera coordinate system; afterwards, combine the perspective transformation algorithm and the external parameters of the camera to convert the normalized coordinates to obtain the three-dimensional geometric points.

[0085] Among them, the perspective transformation matrix H in the perspective transformation algorithm is calculated in the following way: import cv2 import numpy as np def compute_perspective_transform(src_pts, dst_pts): H, _ = cv2.findHomography(src_pts, dst_pts, cv2.RANSAC, 5.0) return H In addition, the process of conversion using the perspective transformation algorithm, the internal parameter matrix K of the camera, and the external parameter matrix [R∣t] is as follows: def perspective_transform(points, camera_matrix, dist_coeffs, rvec,tvec): points_3d = cv2.undistortPoints(points, camera_matrix, dist_coeffs) points_3d = cv2.perspectiveTransform(points_3d, np.hstack((rvec,tvec))) return points_3d Step S3054: Calculate the angle of the wind turbine pitch system based on the three-dimensional geometric points and trigonometric functions. Specifically, after the three-dimensional geometric points are obtained by conversion, calculate the angle between the two three-dimensional geometric points through their coordinates in the three-dimensional space, so as to obtain the wind turbine pitch angle. Among them, the angle calculation process can be implemented with reference to the following process: def compute_pitch_angle_optimized(p1, p2, camera_matrix, dist_coeffs,rvec, tvec): # 3D reconstruction p1_3d = perspective_transform(p1, camera_matrix, dist_coeffs, rvec,tvec) p2_3d = perspective_transform(p2, camera_matrix, dist_coeffs, rvec,tvec) # Calculate vector v = p2_3d - p1_3d # Calculate pitch angle angle = np.arctan2(v[1], v[0]) * 180 / np.pi return angle In addition, in other embodiments, the angle calculation can also be performed according to the following process: def compute_pitch_angle(p1, p2): # Calculate vector v = p2 - p1 # Calculate pitch angle angle = np.arctan2(v[1], v[0]) * 180 / np.pi return angle Specifically, feature point extraction, matching, perspective transformation, 3D reconstruction, and angle calculation can be implemented according to the following process: import cv2 import numpy as np # Read image image = cv2.imread('path_to_image.jpg', cv2.IMREAD_GRAYSCALE) # Feature point extraction keypoints, descriptors = combined_feature_extraction(image) # Feature point matching template_keypoints, template_descriptors = combined_feature_extraction (template_image) matches = match_keypoints(keypoints, descriptors, template_keypoints,template_descriptors) # Perspective transformation src_pts = np.float32([keypoints[m.queryIdx].pt for m in matches]).reshape(-1, 1, 2) dst_pts = np.float32([template_keypoints[m.trainIdx].pt for m inmatches]).reshape(-1, 1, 2) H = compute_perspective_transform(src_pts, dst_pts) # 3D reconstruction camera_matrix, dist_coeffs, rvec, tvec = load_camera_parameters() p1_3d = perspective_transform(src_pts[0], camera_matrix, dist_coeffs,rvec, tvec) p2_3d = perspective_transform(src_pts[1], camera_matrix, dist_coeffs,rvec, tvec) # Calculate the pitch angle angle = compute_pitch_angle_optimized(p1_3d, p2_3d, camera_matrix,dist_coeffs, rvec, tvec) print(f"Pitch angle: {angle} degrees" Specifically, when calculating the angle using the template matching algorithm, the above step S305 includes: Step S3055: Match the features with the image features of the template images in the preset template library. The preset template library includes multiple template images, image features, and corresponding pitch angles, and the preset template library is dynamically updated.

[0086] Step S3056: Determine the angle of the wind turbine pitch system according to the matching result. The matching result includes a single template image or multiple template images that match successfully. When there are multiple template images, the angle of the wind turbine pitch system is the fusion value of the pitch angles corresponding to the multiple template images.

[0087] Specifically, before using the template matching algorithm, it is necessary to construct a template library first. The template image library contains multiple template images and their corresponding pitch angles. These template images are taken at different pitch angles, and their pitch angles are known. For example: Template image 1, pitch angle is 1 degree; Template image 2, pitch angle is 2 degrees; Template image 3, pitch angle is 3 degrees; Template image 10, pitch angle is 10 degrees.

[0088] After constructing the template library, match the obtained image features with the features in the template image library. Cosine similarity or other similarity measurement methods can be used to calculate the similarity between features. According to the matching result, find the most similar template image and obtain its corresponding pitch angle. For some complex pitch states, a multi-template fusion matching strategy can be adopted to determine the fusion value of the pitch angle. For example, find multiple similar template images, calculate the average value of their corresponding angles, or perform fusion through methods such as weighted average to reduce the error caused by single-template matching, so as to improve the accuracy and reliability of angle calculation.

[0089] Specifically, when using weighted average, the angles of each template can be weighted and averaged according to the similarity. The weight can be a function of the similarity, and the higher the similarity, the greater the weight. This weighted average method can make better use of the information of multiple templates and improve the accuracy of angle calculation.

[0090] In addition, in order to improve the adaptability and robustness of the system, the template image library can be dynamically updated. During the operation of the wind turbine, according to the real-time collected image features and known pitch angles, the template image library is continuously updated, thereby enriching the content of the template library. This dynamic update mechanism enables the template images to learn the image features in more changing situations, further improving the robustness to noise, deformation, and illumination changes. That is, it can adapt to different operating states and environmental conditions. Thus, the accuracy of matching is improved.

[0091] When dynamically updating the template library, if the wind turbine operates under different lighting conditions or in different seasons (such as summer and winter), the template image library will contain the image features under these different conditions, thereby improving the accuracy and robustness of matching. And if a certain template performs poorly in multiple matches, it can be adjusted or replaced to improve the overall performance. In addition, during dynamic update, the content of the template image library can be dynamically adjusted in real time according to the collected image features and pitch angle data. This real-time feedback mechanism ensures that the template image library always contains the latest and most relevant image features, thus improving the accuracy and robustness of matching.

[0092] In the present invention, industrial-grade CCD or CMOS cameras and multi-view observation points are used for image acquisition. These cameras have ultra-high resolution, excellent low-light performance, and fast frame rate, and can accurately capture fine structures and component details, adapting to complex lighting and dynamic operating conditions. Multiple optimal observation points are determined through simulation and on-site testing, and can comprehensively and without dead angles photograph key components and capture the dynamic changes of transmission components, obtaining rich and complete image data, thereby realizing comprehensive, in-depth, and intuitive monitoring of the operating state of the pitch system, and being able to more accurately detect potential fault hazards, such as slight wear of the blade and looseness of the connection part, effectively solving the problem that traditional sensors have a single monitoring dimension and are difficult to comprehensively reflect the overall operating state of the system, and providing sufficient and accurate data support and decision-making basis for the refined operation and maintenance of the wind turbine.

[0093] In the present invention, the light source can be automatically adjusted according to the real-time lighting conditions to ensure that the image acquisition area is always in the best lighting state, avoiding shadow interference, greatly improving the image quality, and then enhancing the accuracy and reliability of subsequent pitch angle recognition, overcoming the influence of unstable lighting in the working environment of the wind turbine on monitoring, ensuring the high-precision operation of the monitoring system, avoiding shadow interference, improving the image quality, and solving the problem that traditional image acquisition is greatly affected by lighting. It provides a strong guarantee for subsequent pitch angle recognition.

[0094] In the present invention, an adaptive grayscale algorithm is adopted: compared with the traditional fixed-parameter grayscale method, it can automatically adjust the grayscale conversion parameters according to the image content, greatly compressing the data volume while retaining key information, reducing the computational complexity, improving the processing speed and real-time performance, solving the problems of large data volume and low processing efficiency in traditional image analysis methods, and enabling the entire monitoring system to respond more quickly to the state changes of the pitch system.

[0095] In the present invention, by combining the advantages of median filtering, Gaussian filtering, and wavelet filtering, various noises generated during the operation of the wind turbine are accurately removed, improving the image clarity and signal-to-noise ratio, providing reliable image data for subsequent feature extraction and angle calculation, avoiding the problem that traditional single filtering methods cannot effectively remove complex noises and affect the accuracy of subsequent analysis, and ensuring the stability and reliability of the monitoring system.

[0096] In the present invention, when performing edge detection using the Canny edge detection algorithm, a multi-scale morphological gradient operation and an adaptive threshold selection strategy are introduced to enhance the edge contrast and coherence, avoid inaccurate or broken edges, provide a highly accurate geometric basis for the pitch angle calculation based on edge geometric features, solve the problem that traditional edge detection algorithms are affected by factors such as noise and uneven illumination when processing images of the wind turbine pitch system, resulting in inaccurate edge detection, and improve the accuracy of the pitch angle calculation.

[0097] In the present invention, by integrating the feature point extraction algorithm, organically combining the advantages of the SIFT and SURF algorithms, optimizing in the feature point detection and descriptor generation stages, and introducing a local feature constraint mechanism, the number and quality of feature points detected are improved, as well as the recognition and stability at different pitch angles, the pitch angle is more accurately determined, the accuracy and reliability of the monitoring system are enhanced, and the problems of inaccurate feature point extraction or insufficient robustness of the feature descriptors existing in traditional feature point extraction algorithms are overcome.

[0098] In the present invention, the camera parameters are automatically predicted and adjusted in real time through a deep learning model, and the pitch angle is calculated by combining an improved perspective transformation algorithm and optimized trigonometric functions, overcoming the problems of inaccurate parameters and poor environmental adaptability in traditional geometric relationship calculation methods, realizing high-precision and adaptive pitch angle calculation, improving the accuracy and stability of the angle calculation, and adapting to the displacement and attitude changes that the camera may undergo during the long-term operation of the wind turbine.

[0099] In the present invention, a dynamically updated template library is constructed, a deep learning image feature matching algorithm is adopted and a multi-template fusion matching strategy is introduced to effectively overcome the influences of image noise, deformation, and illumination changes, more accurately find the matching template to determine the pitch angle, provide an efficient and accurate solution for the angle calculation of the wind turbine pitch system, solve the problem that traditional template matching methods are difficult to adapt to the image changes of the wind turbine pitch system under different operating states and environmental conditions, and improve the accuracy and reliability of the angle calculation.

[0100] In the present invention, through the innovative application of the above series of technical means, non-contact, high-precision, and multi-dimensional monitoring of the wind turbine pitch system is realized, many drawbacks of traditional sensor monitoring methods are overcome, the operating conditions of the pitch system can be timely and accurately grasped, potential faults can be discovered in advance and corresponding measures can be taken, the probability of the wind turbine stopping due to pitch system faults is effectively reduced, the safe and stable operation of the wind turbine is guaranteed, the reliability and power generation efficiency of the wind power generation system are improved, the operation and maintenance costs are reduced, and significant economic and social benefits are achieved.

[0101] In this embodiment, a monitoring device for a wind turbine pitch system is also provided. This device is used to implement the above-mentioned embodiments and preferred implementation manners, and those that have been described will not be repeated here. As used hereinafter, the term "module" can be a combination of software and / or hardware that can achieve a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware is also possible and contemplated.

[0102] This embodiment provides a monitoring device for a wind turbine pitch system, as Figure 2 shown, including: An image acquisition module 21, configured to acquire image data of the wind turbine pitch system; A grayscale processing module 22, configured to perform grayscale processing on the image data based on an adaptive grayscaling algorithm with parameter adjustment to obtain grayscale image data after processing; A filtering module 23, configured to filter the grayscale image data after processing by using median filtering, Gaussian filtering, and wavelet filtering to obtain filtered image data; A feature recognition module 24, configured to perform edge detection and feature and feature point recognition on the filtered image data, and the feature point recognition is performed by using a fusion algorithm of SIFT and SURF; An angle calculation module 25, configured to calculate the angle of the wind turbine pitch system based on the recognized feature points according to a geometric relationship algorithm, or calculate the angle of the wind turbine pitch system based on the recognized features according to a template matching algorithm.

[0103] The further function descriptions of the above-mentioned each module are the same as those in the corresponding above-mentioned embodiments, and will not be repeated here.

[0104] This embodiment of the present invention also provides a computer device having the above-mentioned Figure 2 monitoring device for the wind turbine pitch system.

[0105] Please refer to Figure 3 , Figure 3 which is a schematic structural diagram of a computer device provided by an alternative embodiment of the present invention, as Figure 3As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including a high-speed interface and a low-speed interface. Each component communicates with each other using different buses and can be installed on a common motherboard or in other ways as needed. The processor can process instructions executed within the computer device, including instructions stored in the memory or on the memory to display graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In some alternative embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a set of blade servers, or a multi-processor system). Figure 3 Taking one processor 10 as an example in Figure 3 . The processor 10 can be a central processing unit, a network processor, or a combination thereof. Among them, the processor 10 can further include a hardware chip. The above hardware chip can be an application-specific integrated circuit, a programmable logic device, or a combination thereof. The above programmable logic device can be a complex programmable logic device, a field programmable gate array, a generic array logic, or any combination thereof. Among them, the memory 20 stores instructions executable by at least one processor 10, so that at least one processor 10 executes the method shown in the above embodiments.

[0106] The memory 20 can include a program storage area and a data storage area. Among them, the program storage area can store an operating system and application programs required for at least one function; the data storage area can store data created according to the use of the computer device presented by a kind of landing page of a small program, etc. In addition, the memory 20 can include high-speed random access memory and can also include non-transitory memory, such as at least one disk storage device, a flash memory device, or other non-transitory solid-state storage devices. In some alternative embodiments, the memory 20 can optionally include a memory remotely set relative to the processor 10, and these remote memories can be connected to the computer device through a network. Examples of the above network include but are not limited to the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof. The memory 20 can include volatile memory, such as random access memory; the memory can also include non-volatile memory, such as flash memory, a hard disk, or a solid-state drive; the memory 20 can also include a combination of the above types of memory. The computer device further includes a communication interface 30 for the computer device to communicate with other devices or a communication network.

[0107] Embodiments of the present invention also provide a computer-readable storage medium. The method according to the embodiments of the present invention can be implemented in hardware, firmware, or be implemented as computer code that can be recorded on a storage medium, or be implemented as computer code that is originally stored in a remote storage medium or a non-transitory machine-readable storage medium and downloaded through a network and will be stored in a local storage medium, so that the method described herein can be stored as such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only memory, a random access memory, a flash memory, a hard disk, or a solid-state drive, etc.; further, the storage medium can also include a combination of the above-mentioned types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by the computer, the processor, or the hardware, the method shown in the above embodiments is implemented.

[0108] A part of the present invention can be applied as a computer program product, such as computer program instructions. When executed by a computer, through the operation of the computer, the methods and / or technical solutions according to the present invention can be invoked or provided. Those skilled in the art should be able to understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executes the instructions, or the computer compiles the instructions and then executes the corresponding compiled program, or the computer reads and executes the instructions, or the computer reads and installs the instructions and then executes the corresponding installed program. Herein, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to the computer.

[0109] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A monitoring method for a wind turbine pitch system, characterized in that The method includes: Collecting image data of the wind turbine pitch system; Performing grayscale processing on the image data based on an adaptive grayscale algorithm with parameter adjustment to obtain grayscale image data; Filtering the grayscale image data using median filtering, Gaussian filtering, and wavelet filtering to obtain filtered image data; Performing edge detection and feature and feature point recognition on the filtered image data, and using a fusion algorithm of SIFT and SURF for feature point recognition; Based on the recognized feature points, calculating the angle of the wind turbine pitch system according to the geometric relationship algorithm, or based on the recognized features, calculating the angle of the wind turbine pitch system according to the template matching algorithm.

2. The method according to claim 1, characterized in that, Before collecting the image data of the wind turbine pitch system, the method further includes: Setting multiple groups of observation points in a pre-established three-dimensional model of the wind turbine pitch system; Determining the evaluation parameters corresponding to each group of observation points according to the image data observed at each group of observation points, where the image data includes the image data observed when the wind turbine operates under different working conditions; Screening out preliminary candidate observation points from multiple groups of observation points according to the evaluation parameters; Arranging an image acquisition device in the wind turbine pitch system according to the preliminary candidate observation points; Adjusting the position of the preliminary candidate observation points based on the comparison result between the image data collected by the image acquisition device and the image data observed at the preliminary candidate observation points; Determining the final observation points based on the adjusted preliminary candidate observation points for arranging the image acquisition device, where the image acquisition device is used to collect the image data of the wind turbine pitch system, and the image acquisition device is an industrial-grade CCD or industrial-grade CMOS.

3. The method according to claim 1, wherein Collecting the image data of the wind turbine pitch system includes: Collecting the image data of the wind turbine pitch system based on a pre-set image acquisition device and light source; Using an image processing algorithm to determine the shadow distribution in the image data; Adjusting the emission intensity and angle of the light source according to the ambient light intensity and shadow distribution; Collecting the image data of the wind turbine pitch system based on the pre-set image acquisition device and the adjusted light source.

4. The method according to claim 1, wherein Performing grayscale processing on the image data based on an adaptive grayscale algorithm with parameter adjustment to obtain grayscale image data, including: Detecting the edge information in the image data based on the Canny edge detection algorithm combined with multi-scale morphological gradient operation; Calculating the texture features of the image data; Performing grayscale processing on the image data based on an adaptive grayscale algorithm with dynamically adjusted grayscale conversion parameters and an adaptive threshold algorithm to obtain grayscale image data, where the grayscale conversion parameters are dynamically adjusted based on the edge information and texture features.

5. The method according to claim 1, wherein Performing edge detection and feature and feature point recognition on the filtered image data, including: Detecting the edge information in the image data based on the Canny edge detection algorithm combined with multi-scale morphological gradient operation and a local adaptive threshold algorithm, where the threshold in the local adaptive threshold algorithm is adjusted based on the local statistical information, edge intensity, and texture features of the image data; Detecting the feature points in the image data using the SIFT algorithm and the SURF algorithm respectively and merging them; Based on the merged feature points, the separately calculated SIFT feature descriptors and SURF feature descriptors are fused to obtain the fused feature descriptors; The feature points are screened according to the local geometric features and texture features of the wind turbine pitch system, and the recognized feature points are obtained based on the screened feature points and feature descriptors; A deep convolutional neural network is used to extract the features of the image data.

6. The method according to claim 1, characterized in that Based on the recognized feature points, the angle of the wind turbine pitch system is calculated according to the geometric relationship algorithm, including: The deep neural network model is trained according to the calibration image and camera parameters to obtain the camera parameter prediction model; The recognized feature points are matched with the pre-constructed geometric model of the wind turbine pitch system to obtain the matching result; Based on the matching result, the two-dimensional feature points are converted into three-dimensional geometric points by using the perspective transformation algorithm and the camera parameters predicted by the camera parameter prediction model; The angle of the wind turbine pitch system is calculated based on the three-dimensional geometric points and trigonometric functions.

7. The method according to claim 1, characterized in that, Based on the recognized features, the angle of the wind turbine pitch system is calculated according to the template matching algorithm, including: The features are matched with the image features of the template images in the preset template library. The preset template library includes multiple template images, image features, and corresponding pitch angles, and the preset template library is dynamically updated; The angle of the wind turbine pitch system is determined according to the matching result. The matching result includes a single template image or multiple template images that match successfully. When there are multiple template images, the angle of the wind turbine pitch system is the fusion value of the pitch angles corresponding to the multiple template images.

8. A monitoring device for a wind turbine pitch system, characterized in that, The device includes: An image acquisition module for acquiring image data of the wind turbine pitch system; A grayscale processing module for performing grayscale processing on the image data based on the self-adaptive grayscale algorithm with parameter adjustment to obtain the grayscale image data; A filtering module for filtering the grayscale image data by using median filtering, Gaussian filtering, and wavelet filtering to obtain the filtered image data; A feature recognition module for performing edge detection and feature and feature point recognition on the filtered image data. The feature point recognition is performed by using the SIFT and SURF fusion algorithm; An angle calculation module for calculating the angle of the wind turbine pitch system based on the recognized feature points according to the geometric relationship algorithm, or calculating the angle of the wind turbine pitch system based on the recognized features according to the template matching algorithm.

9. A computer device, characterized in that, Including: A memory and a processor. The memory and the processor are communicatively connected to each other. The memory stores computer instructions, and the processor executes the computer instructions to execute the monitoring method of the wind turbine pitch system according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, Computer instructions are stored on the computer-readable storage medium, and the computer instructions are used to cause the computer to execute the monitoring method of the wind turbine pitch system according to any one of claims 1 to 7.

Images