Adaptive data compression wind power load test data processing method and system

Abstract

The invention discloses a wind power load test data processing method and system for adaptive data compression, and relates to the related field of data processing, and the method comprises the steps: collecting the load test data of a wind turbine generator through employing a nested sliding window; performing multi-scale feature extraction on the collected data of the nested sliding window, executing multi-scale feature fusion, and establishing a comprehensive change score of each data point; acquiring spatial distribution of acquisition sensors, configuring sensor adjacent coefficients according to the spatial distribution and sensor functions, performing spatial consistency detection of data points, and positioning key frames; and synchronizing the key frame to an adaptive compression channel, constructing a data compression strategy, and executing compression management. The technical problems of insufficient compression efficiency and accuracy caused by insufficient acquisition flexibility and difficulty in accurately capturing dynamic characteristics in existing data processing are solved, and the technical effects of improving the acquisition flexibility, accurately capturing the dynamic change characteristics of the data and further improving the efficiency and accuracy of data compression are achieved.

Description

Technical Field

[0001] The present application relates to data processing related fields, and in particular to a wind power load test data processing method and system with adaptive data compression. Background Art

[0002] In the wind power industry, the processing of wind turbine load test data is crucial to ensuring wind turbine safety, optimizing operation and maintenance strategies. Traditional data acquisition methods often use fixed time intervals or fixed data volumes for sampling, which is not flexible enough when dealing with complex and changing load conditions. In terms of data processing, existing methods usually perform simple statistical analysis or filtering on the collected data, which makes it difficult to fully capture the dynamic characteristics of the data. In terms of data compression, most existing methods use fixed compression algorithms and parameters, and lack the ability to adapt to data characteristics, which may result in the loss of important information during the data compression process, or the compression efficiency is low, and cannot meet the needs of real-time processing.

[0003] In the current relevant technologies, wind power load test data processing has the technical problems of insufficient flexibility in data collection and difficulty in accurately capturing the dynamic characteristics of data, resulting in insufficient efficiency and accuracy of data compression. Summary of the invention

[0004] The present application provides a wind power load test data processing method and system with adaptive data compression, collects data using nested sliding windows, extracts and fuses multi-scale features of the collected data, establishes a comprehensive change score for each data point, performs spatial consistency detection based on sensor adjacency coefficients, locates key frames based on spatial consistency detection results and comprehensive change scores, and constructs an adaptive compression strategy and other technical means, thereby achieving the technical effect of improving the flexibility of data collection, accurately capturing the dynamic change characteristics of data, and thereby improving the efficiency and accuracy of data compression.

[0005] The present application provides a method for processing wind power load test data with adaptive data compression, comprising: collecting load test data of a wind turbine using a nested sliding window, wherein the nested sliding window comprises a basic sliding window and an extended sliding window, and the basic sliding window can slide multiple times within the extended sliding window, and the load test data comprises stress data, vibration data, torque data, and wind speed data; performing multi-scale feature extraction on the collected data of the nested sliding window respectively, and performing multi-scale feature fusion to establish a comprehensive change score for each data point; obtaining the spatial distribution of the collected sensor, configuring the sensor adjacency coefficient according to the spatial distribution and the sensor function, performing spatial consistency detection of the data point based on the sensor adjacency coefficient, and locating the key frame according to the spatial consistency detection result and the comprehensive change score; synchronizing the key frame to an adaptive compression channel, constructing a data compression strategy, and using the data compression strategy to perform compression management of the load test data.

[0006] In a possible implementation, the sensor adjacency coefficient is configured according to the spatial distribution and the sensor function, and the following processing is performed: the proximity distance of the sensor is calculated based on the spatial distribution, and it is determined whether the proximity distance is less than a preset distance threshold; if the proximity distance is less than the preset distance threshold, a first authentication pass result is generated; a function vector is configured according to the sensor function, and the function similarity is calculated by cosine similarity and the function vector, and if the function similarity is higher than the preset function threshold, a second authentication pass result is generated; when the first authentication pass result and the second authentication pass result exist at the same time, the sensor adjacency coefficient is calculated.

[0007] In a possible implementation, the calculation of the sensor proximity coefficient performs the following processing: the sensor proximity coefficient is calculated by the following formula: ;in, Characterizing Sensors and sensors The sensor adjacency coefficient ranges from [0,1], Characterizing Sensors and sensors The adjacent distance, Characterizes the spatial attenuation factor, Characterizing Sensors and sensors functional similarity.

[0008] In a possible implementation, after calculating the sensor adjacency coefficient, the following processing is performed: the differential deviation of the sensor data is calculated by the formula: ;in, Characterizing Sensors and sensors In time The data difference at Characterizing Sensors In time The measured value at Characterizing Sensors In time The measured value at; based on the differential deviation, the sensor proximity coefficient calculates the consistency score: ;in, Representation in time Sensor The consistency score, Characterization and Sensors A set of sensors with sensor adjacency coefficients, is the data difference sensitivity coefficient.

[0009] In a possible implementation, the load test data of the wind turbine generator set is collected by using a nested sliding window, and the following processing is performed: a default window value of a basic sliding window within the nested sliding window is configured, and the basic sliding window is called according to the default window value to perform data pre-collection of the load test data, and a pre-collected data set is established; a data stability analysis of the pre-collected data set is performed, and a reconstruction parameter is generated according to the data stability analysis result, and the reconstruction parameter includes the window value of the basic sliding window and the extended value of the extended sliding window; after reconstructing the nested sliding window according to the window value and the extended value, the load test data of the wind turbine generator set is collected.

[0010] In a possible implementation, the nested sliding window is reconstructed according to the window value and the extension value, and the following processing is performed: data fluctuation analysis of the pre-collected data set is performed, and an extension bias is established according to the data fluctuation analysis result; after reconstructing the basic sliding window with the window value, an extended sliding window is reconstructed based on the extension bias and the extension value with the reconstructed basic sliding window as the center; after collecting the load test data of the wind turbine set using the reconstructed nested sliding window, a window data set corresponding to the reconstructed extended sliding window is obtained; after evenly dividing the window data set, the fluctuation amplitude under the evenly divided window is calculated; and an adaptive sliding step size is configured according to the fluctuation amplitude and the sensitivity coefficient.

[0011] In a possible implementation, the key frames are synchronized to an adaptive compression channel, a data compression strategy is constructed, and the following processing is performed: a quantitative analysis of the key frames is performed; if the quantitative analysis result is higher than a preset quantitative threshold, the key frames are divided into primary key frames and secondary key frames; the primary key frames are compressed with original data retention, the secondary key frames are compressed with a reference low-loss method based on the primary key frames, and the non-key frames are compressed with a loss, so as to construct a data compression strategy.

[0012] The present application also provides a wind power load test data processing system with adaptive data compression, including: a load test data acquisition module, which is used to collect load test data of wind turbines using nested sliding windows, wherein the nested sliding windows include a basic sliding window and an extended sliding window, and the basic sliding window can slide multiple times within the extended sliding window, and the load test data includes stress data, vibration data, torque data, and wind speed data; a multi-scale feature extraction and fusion module, which is used to perform multi-scale feature extraction on the collected data of the nested sliding windows respectively, and perform multi-scale feature fusion to establish a comprehensive change score for each data point; a key frame positioning module, which is used to obtain the spatial distribution of the acquisition sensor, configure the sensor adjacency coefficient according to the spatial distribution and sensor function, perform spatial consistency detection of data points based on the sensor adjacency coefficient, and locate the key frame according to the spatial consistency detection result and the comprehensive change score; a data compression module, which is used to synchronize the key frame to the adaptive compression channel, construct a data compression strategy, and use the data compression strategy to perform compression management of the load test data.

[0013] The wind power load test data processing method and system proposed in this application for adaptive data compression first collects the load test data of the wind turbine using a nested sliding window, wherein the nested sliding window includes a basic sliding window and an extended sliding window, and the basic sliding window can slide multiple times within the extended sliding window, and the load test data includes stress data, vibration data, torque data, and wind speed data, and then multi-scale feature extraction is performed on the collected data of the nested sliding window, and multi-scale feature fusion is performed to establish a comprehensive change score for each data point, and then the spatial distribution of the collected sensor is obtained, and the sensor adjacency coefficient is configured according to the spatial distribution and sensor function, and the spatial consistency detection of the data point is performed based on the sensor adjacency coefficient, and the key frame is located according to the spatial consistency detection result and the comprehensive change score, and finally the key frame is synchronized to the adaptive compression channel, a data compression strategy is constructed, and the compression management of the load test data is performed using the data compression strategy. The technical effect of improving the flexibility of data acquisition, accurately capturing the dynamic change characteristics of data, and thus improving the efficiency and accuracy of data compression is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solution of the embodiment of the present invention, the accompanying drawings of the embodiment of the present invention will be briefly introduced below. A flow chart is used in the present application to illustrate the operations performed by the system according to the embodiment of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or more operations can be removed from these processes.

[0015] Figure 1 A schematic flow chart of a method for processing wind power load test data with adaptive data compression provided in an embodiment of the present application.

[0016] Figure 2 A schematic diagram of the structure of a wind power load test data processing system with adaptive data compression provided in an embodiment of the present application.

[0017] Explanation of the reference numerals: load test data acquisition module 10 , multi-scale feature extraction and fusion module 20 , key frame positioning module 30 , data compression module 40 . DETAILED DESCRIPTION

[0018] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0019] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0020] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict, and the terms "first\second" involved are merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "including" and "having" and any variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by technicians in the technical field of this application. The terms used herein are for the purpose of describing the embodiments of the present application only.

[0021] The embodiment of the present application provides a method for processing wind power load test data with adaptive data compression, such as Figure 1 As shown, the method includes: Step S100, using a nested sliding window to collect load test data of the wind turbine set, the nested sliding window includes a basic sliding window and an extended sliding window, and the basic sliding window can slide multiple times within the extended sliding window, and the load test data includes stress data, vibration data, torque data, and wind speed data.

[0022] Specifically, nested sliding windows are a data processing method that uses windows of different sizes to slide on time series data to capture data changes of different scales. The basic sliding window is a fixed time length (such as 1 second) for data acquisition. It slides on the data stream to continuously capture data fragments and is used to initially capture local changes in the data. The extended sliding window is a longer time length (such as 10 seconds) that covers multiple slides of the basic sliding window and is used to capture a wider range of data change trends. Through the sensor network connected to the wind turbine, load test data such as stress data, vibration data, torque data, and wind speed data are collected in real time. The load test data is various physical quantity data generated by the wind turbine during operation, which is used to evaluate the load state and performance of the unit. The data is stored in time series, and each time point corresponds to a complete set of load test data. The basic sliding window slides on the time series, sliding one data point each time, and recording the data in the window. In the extended sliding window, the basic sliding window can slide multiple times to collect richer data samples.

[0023] In a possible implementation, the load test data of the wind turbine generator set is collected using a nested sliding window, and step S100 further includes step S110, configuring the default window value of the basic sliding window within the nested sliding window, calling the basic sliding window according to the default window value to perform data pre-collection of the load test data, and establishing a pre-collected data set. Specifically, an initial or default window size is set for the basic sliding window. This window size is determined based on empirical values ​​or preliminary data analysis results, and is used to cover a reasonable time range or number of data points so as to effectively capture local features of the data. Subsequently, this configured basic sliding window is used to slide on the load test data stream to collect data. Each time the window slides into place, the data in the window is extracted to form a pre-collected data set. This process is performed continuously until a predetermined amount of pre-collected data or time is reached.

[0024] Step S120, perform data stability analysis of the pre-collected data set, and generate reconstruction parameters according to the data stability analysis results, wherein the reconstruction parameters include the window value of the basic sliding window and the extended value of the extended sliding window. Specifically, the pre-collected data set is analyzed to evaluate the stability and consistency of the data over time. By calculating indicators such as the variance, standard deviation, and autocorrelation coefficient of the data, it is determined whether there is obvious fluctuation or trend in the data. According to the results of the data stability analysis, the window value of the basic sliding window and the extended value of the extended sliding window are adjusted. If the data fluctuates greatly, the window value needs to be increased to include more data points to smooth the fluctuations; if the data is relatively stable, the window value can be appropriately reduced to improve the efficiency of data processing. The extended value of the extended sliding window is used to expand the scope of data acquisition when necessary.

[0025] Step S130, after reconstructing the nested sliding window according to the window value and the extended value, collect the load test data of the wind turbine. Specifically, according to the reconstruction parameters generated in step S120, modify the relevant variables or configurations in the program, adjust the sizes of the basic sliding window and the extended sliding window, and ensure that subsequent data collection meets the new window size requirements. Use the reconstructed nested sliding window to slide on the load test data stream to collect data. At this time, the basic sliding window is used to capture the local features of the data, and the extended sliding window provides the ability to expand the data collection range when needed. This implementation method improves the efficiency and accuracy of data collection and processing by gradually adjusting and optimizing the size of the nested sliding window to adapt to different data characteristics.

[0026] In a possible implementation, the nested sliding window is reconstructed according to the window value and the extension value, and step S130 further includes step S131, performing data fluctuation analysis on the pre-collected data set, and establishing an extension bias according to the data fluctuation analysis result. Specifically, the pre-collected data set is subjected to fluctuation analysis by calculating the standard deviation, coefficient of variation or other statistics of the data. Based on these statistics, trends, periodic changes or outliers in the data are identified. Extension bias refers to determining the bias or extension direction of the extended sliding window relative to the basic sliding window based on the data fluctuation characteristics. For example, if the data fluctuates more on one side, the extended sliding window will expand more toward that side to capture more change information.

[0027] Step S132, after reconstructing the basic sliding window with the window value, reconstruct the extended sliding window based on the expansion bias and the expansion value with the reconstructed basic sliding window as the center. Specifically, first reconstruct the basic sliding window according to the reconstruction parameters generated in step S120 (including the window value and the expansion value of the basic sliding window). Then, with the reconstructed basic sliding window as the center, combined with the expansion bias and the expansion value determined in step S131, reconstruct the extended sliding window, that is, the extended sliding window is dynamically adjusted around the basic sliding window according to the fluctuation characteristics of the data.

[0028] Step S133, after collecting the load test data of the wind turbine using the reconstructed nested sliding window, obtain a window data set corresponding to the reconstructed extended sliding window. Specifically, once the nested sliding window is reconstructed, this window is used to collect the load test data of the wind turbine. During the collection process, the data in each extended sliding window is recorded to form a window data set.

[0029] Step S134, after the window data set is evenly divided, the fluctuation range under the evenly divided window is calculated. Specifically, the window data set is evenly divided into multiple small windows. Then, the fluctuation range of the data in each small window is calculated, including calculating the difference between the maximum and minimum values ​​of the data, the standard deviation or other statistics.

[0030] Step S135, configuring an adaptive sliding step size according to the fluctuation amplitude and sensitivity coefficient. Specifically, based on the fluctuation amplitude calculated in step S134 and the preset sensitivity coefficient, dynamically adjust the sliding step size. The sensitivity coefficient is a preset parameter used to control the degree of influence of the fluctuation amplitude on the adjustment of the sliding step size. If the fluctuation amplitude is large and the sensitivity coefficient is high, reduce the sliding step size to capture data changes more finely. On the contrary, if the fluctuation amplitude is small and the sensitivity coefficient is low, increase the sliding step size to reduce the amount of data. This implementation method can flexibly adapt to the changing characteristics of the data by dynamically adjusting the size and step size of the nested sliding windows. In areas where the data fluctuates greatly, reducing the sliding step size can capture more detailed information; in areas where the data fluctuates less, increasing the sliding step size can reduce unnecessary data redundancy. This can not only reduce the amount of data and reduce storage and transmission costs, but also retain important change information in the data, thereby improving the efficiency and accuracy of data compression.

[0031] Step S200, performing multi-scale feature extraction on the collected data of the nested sliding windows respectively, and performing multi-scale feature fusion to establish a comprehensive change score for each data point.

[0032] Specifically, for the data in each basic sliding window, basic features such as mean, standard deviation, maximum, minimum, etc. are extracted to reflect the overall distribution and discreteness of the data. In the extended sliding window, in addition to extracting basic features, more macro features are also extracted, such as trend features (such as linear regression coefficients) and periodic features (such as Fourier transform results), to reflect the changing trend and periodicity of the data over a longer period of time. The feature extraction results of the basic window and the extended window are combined to form a multi-scale feature set. These features contain both local details of the data and global macro information.

[0033] According to the characteristics of the data and the analysis requirements, key features are selected from the multi-scale feature set for fusion through feature importance evaluation methods (such as information gain, Gini coefficient, etc.), and the selected features are fused using linear combination, nonlinear mapping, etc. For example, the weighted summation method can be used to linearly combine features of different scales, or nonlinear models such as neural networks can be used for feature fusion.

[0034] According to the fused feature vector, the comprehensive change score of each data point is calculated through distance measurement (such as Euclidean distance, Manhattan distance, etc.), similarity measurement (such as cosine similarity, Pearson correlation coefficient, etc.) or custom scoring function, which indicates the degree of change of each data point relative to its surrounding data points. In order to make the comprehensive change score comparable, it is standardized. For example, the score value can be mapped to the range of 0 to 1 so that the score values ​​of different data points have the same scale.

[0035] Step S300, obtaining the spatial distribution of acquisition sensors, configuring sensor adjacency coefficients according to the spatial distribution and sensor functions, performing spatial consistency detection of data points based on the sensor adjacency coefficients, and locating key frames according to the spatial consistency detection results and the comprehensive change score.

[0036] Specifically, the spatial distribution information of sensors on wind turbines, including longitude, latitude and altitude, is obtained through configuration files or databases. Using the spatial distribution data, the physical distance between each pair of sensors is calculated using Euclidean distance, Manhattan distance, etc. According to the functional description of the sensor (such as measurement type, accuracy, range, etc.), the functional similarity between sensors is evaluated by comparing the cosine similarity, Jaccard similarity, etc. of the functional vectors. Combining the adjacent distance and functional similarity, an adjacent coefficient is assigned to each sensor pair through weighted summation, product or other combinations. This coefficient can be a value between 0 and 1, indicating the degree of association between sensors. Using the adjacent coefficient, the consistency between adjacent sensor data points is evaluated by comparing the values, change rates or other statistical features of the data points. Combining the spatial consistency detection results and the comprehensive change score, those data points that are consistent in space and have changed significantly are identified as key frames.

[0037] In a possible implementation, the sensor adjacency coefficient is configured according to the spatial distribution and the sensor function, and step S300 further includes step S310, calculating the proximity distance of the sensor based on the spatial distribution, and judging whether the proximity distance is less than a preset distance threshold. Specifically, the location information of all sensors is collected by GPS or other positioning technologies to form a coordinate data set. For each pair of sensors, the distance between them is calculated using a distance formula (Euclidean distance formula or other distance measurement method). The calculated distance is compared with a preset distance threshold (determined according to the actual situation of the wind farm and the characteristics of the sensor, used to judge whether the two sensors are close enough to be considered adjacent). If the distance is less than the threshold, the two sensors are considered to be physically close.

[0038] Step S320: If the adjacent distance is less than the preset distance threshold, a first authentication pass result is generated. Specifically, the comparison result of each distance calculated in step S310 and the preset distance threshold is checked. For all distances less than the threshold, the corresponding sensor pair is marked as "adjacent" and a first authentication pass result is generated.

[0039] Step S330, configure the function vector according to the sensor function, calculate the function similarity through cosine similarity and the function vector, and if the function similarity is higher than the preset function threshold, generate a second authentication pass result. Specifically, a function vector is generated for each sensor according to its function configuration (obtained from the sensor's technical specification or database, such as measurement type, accuracy, sampling rate, etc.). For each pair of sensors, the cosine similarity formula is used to calculate the similarity between their function vectors. The value range is [-1,1], and the larger the value, the closer the direction. Compare the calculated similarity with the preset function threshold (a pre-set value used to determine whether the functional similarity of the two sensors is high enough, determined according to the data processing requirements of the wind farm and the functional characteristics of the sensor). If the similarity is higher than the threshold, the two sensors are considered to be functionally similar.

[0040] Step S340, when the first authentication pass result and the second authentication pass result exist at the same time, calculate the sensor adjacency coefficient. Specifically, check whether each pair of sensors satisfies both the first authentication pass result (physical adjacency) and the second authentication pass result (functional similarity). For sensor pairs that meet the conditions, use the formula to calculate the adjacency coefficient. The adjacency coefficient can be a value between 0 and 1, indicating the degree of adjacency between sensors. This implementation method combines the physical location and functional configuration of the sensors, and can more accurately evaluate the adjacency between sensors, which helps to more reasonably utilize sensor data in subsequent key frame positioning and data compression strategy construction, ensuring that important information is not lost during the data compression process, while reducing unnecessary data redundancy.

[0041] In a possible implementation, the step S340 of calculating the sensor proximity coefficient further includes a step S341 of calculating the sensor proximity coefficient by the following formula: ; in, Characterizing Sensors and sensors The sensor adjacency coefficient ranges from [0,1], Characterizing Sensors and sensors The adjacent distance, Characterizes the spatial attenuation factor, Characterizing Sensors and sensors functional similarity.

[0042] Specifically, the adjacency distance between sensors is obtained from step S310, and the functional similarity between sensors is obtained from step S330. The spatial attenuation factor is a parameter used to adjust the effect of the adjacency distance on the adjacency coefficient, which determines how the adjacency coefficient decays as the physical distance increases. The value of the spatial attenuation factor is determined based on the actual situation of the wind farm and the data processing requirements. For example, a suitable value can be set through experiments or expert experience. The adjacency coefficient between sensors is calculated using a formula, and the calculated adjacency coefficient is in the range of [0,1], where 1 indicates complete adjacency and 0 indicates no adjacency at all. This implementation provides a method for comprehensively evaluating the degree of adjacency between sensors. By considering the physical and functional characteristics of the sensors, key frames and data redundancy can be more accurately identified, thereby improving the efficiency and quality of data compression.

[0043] In a possible implementation, after calculating the sensor adjacency coefficient, step S300 further includes step S350, calculating the differential deviation of the sensor data by the formula: ; in, Characterizing Sensors and sensors In time The data difference at Characterizing Sensors In time The measured value at Characterizing Sensors In time Specifically, the measurements at the same time point (e.g., each time step) are collected from each sensor. For each pair of sensors ( and ), calculate them at the same time point The data difference is calculated by the absolute difference between the two sensor measurements.

[0044] Step S360, calculating a consistency score based on the differential deviation and the sensor adjacency coefficient: ; in, Representation in time Sensor The consistency score, Characterization and Sensors A set of sensors with sensor adjacency coefficients, is the data difference sensitivity coefficient. Specifically, time Sensor The consistency score is calculated by considering the All sensors with adjacency coefficients The data difference sensitivity coefficient is calculated based on the data difference and adjacency coefficient. The data difference sensitivity coefficient is an adjustment factor used to control the weight of data difference in the consistency score. This implementation provides a method for evaluating data spatial consistency, taking into account both the physical and functional characteristics between sensors (through the adjacency coefficient) and the differences in their measurements at the same time point (through differential deviation). It can accurately identify anomalies and inconsistencies in the data, and by giving priority to retaining data frames with high consistency, it can more effectively compress data while retaining key information.

[0045] Step S400: synchronize the key frame to an adaptive compression channel, construct a data compression strategy, and use the data compression strategy to perform compression management of the load test data.

[0046] Specifically, the located key frame is synchronized to a channel or module specifically used for data compression, and the adaptive compression channel can automatically adjust the compression strategy according to the characteristics of the data. According to the distribution of key frames and data characteristics, a suitable data compression strategy is constructed, such as key frame-based compression, lossy compression, lossless compression, etc. The constructed data compression strategy is used to compress the load test data to reduce data storage space while maintaining the availability and accuracy of the data and improving data processing efficiency. The compressed data is stored and managed to ensure the integrity and availability of the data. The embodiment of the present application uses a nested sliding window to collect data, extracts and fuses multi-scale features of the collected data, establishes a comprehensive change score for each data point, performs spatial consistency detection based on the sensor adjacency coefficient, locates key frames according to the spatial consistency detection results and the comprehensive change score, and constructs an adaptive compression strategy and other technical means, which achieves the technical effect of improving the flexibility of data collection, accurately capturing the dynamic change characteristics of data, and thus improving the efficiency and accuracy of data compression.

[0047] In a possible implementation, the key frame is synchronized to the adaptive compression channel, and a data compression strategy is constructed. Step S400 further includes step S410, performing a quantity analysis of the key frame. Specifically, the system maintains a list or database containing all key frames. The list is traversed, and each key frame is counted to finally obtain the total number of key frames. Assume that in a 10-minute wind power load test data, the system locates 200 key frames.

[0048] Step S420, if the result of the quantity analysis is higher than the preset quantity threshold, the key frame is divided into primary key frames and secondary key frames. Specifically, a threshold for the number of key frames is preset. When the number of key frames exceeds the preset threshold (such as 100 frames), a clustering algorithm (such as K-means), a decision tree or a rule-based method is used to divide the key frames into primary key frames and secondary key frames according to factors such as the comprehensive change score and time interval of the key frames. The primary key frame is the data point with the highest importance or priority among the key frames. The frame with the highest comprehensive change score, the longest time interval or the strongest correlation with other key frames can be selected as the primary key frame. The secondary key frame is a key frame with lower importance or priority than the primary key frame. For example, among the 200 key frames mentioned above, the system identifies 50 primary key frames and 150 secondary key frames through a classification algorithm.

[0049] Step S430, compress the main key frame with the original data retained, perform low-loss compression on the secondary key frame based on the reference of the main key frame, and perform lossy compression on the non-key frame to construct a data compression strategy. Specifically, different compression strategies are used to compress different types of frames according to the classification results of the key frames. The main key frame adopts lossless compression (such as ZIP, GZIP) to ensure the integrity and accuracy of the data; using the main key frame as a reference, the secondary key frame is low-loss compressed using a prediction-based compression algorithm (such as inter-frame prediction in H.264) to reduce the amount of data while maintaining high data quality; non-key frames use lossy compression (such as JPEG, intra-frame compression in H.264) to minimize the amount of data. For example, through the above compression strategy, the system can compress the original 10-minute wind power load test data from 1GB to 200MB while maintaining high data quality and accuracy. This implementation method can more efficiently utilize compression resources and achieve a higher compression ratio by distinguishing between key frames and non-key frames, and further dividing them into primary key frames and secondary key frames. This not only improves compression efficiency but also ensures data quality, thus meeting the needs of the wind power field for data management and storage.

[0050] In the above, refer to Figure 1 The wind power load test data processing method based on adaptive data compression according to an embodiment of the present invention is described in detail. Figure 2 A wind power load test data processing system with adaptive data compression according to an embodiment of the present invention is described.

[0051] The wind power load test data processing system with adaptive data compression according to the embodiment of the present invention is used to solve the technical problems of insufficient flexibility in data collection and difficulty in accurately capturing the dynamic characteristics of data, resulting in insufficient efficiency and accuracy of data compression in the prior art, so as to achieve the technical effect of improving the flexibility of data collection, accurately capturing the dynamic change characteristics of data, and thus improving the efficiency and accuracy of data compression. The wind power load test data processing system with adaptive data compression includes: a load test data acquisition module 10, a multi-scale feature extraction and fusion module 20, a key frame positioning module 30, and a data compression module 40.

[0052] A load test data acquisition module 10 is used to collect load test data of a wind turbine using a nested sliding window, wherein the nested sliding window includes a basic sliding window and an extended sliding window, and the basic sliding window can slide multiple times within the extended sliding window, and the load test data includes stress data, vibration data, torque data, and wind speed data; a multi-scale feature extraction and fusion module 20 is used to perform multi-scale feature extraction on the collected data of the nested sliding window respectively, and perform multi-scale feature fusion to establish a comprehensive change score for each data point; a key frame positioning module 30 is used to obtain the spatial distribution of the collected sensor, configure the sensor adjacency coefficient according to the spatial distribution and sensor function, perform spatial consistency detection of the data point based on the sensor adjacency coefficient, and locate the key frame according to the spatial consistency detection result and the comprehensive change score; a data compression module 40 is used to synchronize the key frame to an adaptive compression channel, construct a data compression strategy, and use the data compression strategy to perform compression management of the load test data.

[0053] The specific configuration of the key frame positioning module 30 will be described in detail below. As described above, the sensor adjacency coefficient is configured according to the spatial distribution and the sensor function, and the key frame positioning module 30 may further include: an adjacency distance calculation unit is used to calculate the adjacency distance of the sensor based on the spatial distribution, and determine whether the adjacency distance is less than the preset distance threshold; an authentication pass unit is used to generate a first authentication pass result if the adjacency distance is less than the preset distance threshold; a functional similarity calculation unit is used to configure a functional vector according to the sensor function, calculate the functional similarity through the cosine similarity and the functional vector, and generate a second authentication pass result if the functional similarity is higher than the preset functional threshold; a sensor adjacency coefficient calculation unit is used to calculate the sensor adjacency coefficient when the first authentication pass result and the second authentication pass result exist at the same time.

[0054] Wherein, the sensor proximity coefficient calculation unit may further include: a sensor proximity coefficient formula construction subunit for constructing a sensor proximity coefficient formula, and calculating the sensor proximity coefficient by the following formula: ,in, Characterizing Sensors and sensors The sensor adjacency coefficient ranges from [0,1], Characterizing Sensors and sensors The adjacent distance, Characterizes the spatial attenuation factor, Characterizing Sensors and sensors functional similarity.

[0055] After calculating the sensor adjacency coefficient, the key frame positioning module 30 may further include: a differential deviation calculation unit for calculating the differential deviation of the sensor data by using the formula: ,in, Characterizing Sensors and sensors In time The data difference at Characterizing Sensors In time The measured value at Characterizing Sensors In time The consistency score calculation unit is used to calculate the consistency score based on the differential deviation and the sensor adjacency coefficient: ,in, Representation in time Sensor The consistency score, Characterization and Sensors A set of sensors with sensor adjacency coefficients, is the data difference sensitivity coefficient.

[0056] The specific configuration of the load test data acquisition module 10 will be described in detail below. As described above, the load test data of the wind turbine is collected using a nested sliding window, and the load test data acquisition module 10 may further include: a data pre-collection unit is used to configure the default window value of the basic sliding window in the nested sliding window, and call the basic sliding window according to the default window value to perform data pre-collection of the load test data, and establish a pre-collected data set; a reconstruction parameter generation unit is used to perform data stability analysis of the pre-collected data set, and generate reconstruction parameters according to the data stability analysis results, and the reconstruction parameters include the window value of the basic sliding window and the extended value of the extended sliding window; the load test data acquisition unit is used to reconstruct the nested sliding window according to the window value and the extended value, and then collect the load test data of the wind turbine.

[0057] Among them, the nested sliding window is reconstructed according to the window value and the extension value, and the load test data acquisition unit may further include: an extension bias establishment subunit is used to perform data fluctuation analysis of the pre-collected data set, and establish an extension bias according to the data fluctuation analysis result; a sliding window expansion subunit is used to reconstruct the extended sliding window based on the extension bias and the extension value, with the reconstructed basic sliding window as the center, after reconstructing the basic sliding window with the window value; a window data set acquisition subunit is used to acquire the window data set corresponding to the reconstructed extended sliding window after using the reconstructed nested sliding window to collect the load test data of the wind turbine group; a fluctuation amplitude calculation subunit is used to evenly divide the window data set and calculate the fluctuation amplitude under the evenly divided window; an adaptive sliding step configuration subunit is used to configure the adaptive sliding step according to the fluctuation amplitude and sensitivity coefficient.

[0058] The specific configuration of the data compression module 40 will be described in detail below. As described above, the key frame is synchronized to the adaptive compression channel, and a data compression strategy is constructed. The data compression module 40 may further include: a quantity analysis unit for performing a quantity analysis of the key frame; a key frame division unit for dividing the key frame into a primary key frame and a secondary key frame if the result of the quantity analysis is higher than a preset quantity threshold; and a data compression strategy construction unit for compressing the primary key frame with original data retention, performing a reference low-loss compression on the secondary key frame based on the primary key frame, and performing a lossy compression on the non-key frame, so as to construct a data compression strategy.

[0059] The wind power load test data processing system with adaptive data compression provided by the embodiment of the present invention can execute the wind power load test data processing method with adaptive data compression provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0060] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0061] The above specific implementation manner does not constitute a limitation to the protection scope of the present application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application should be included in the protection scope of the present application. In some cases, the actions or steps recorded in the present application can be performed in an order different from that in the embodiment and can still achieve the desired results. In addition, the process depicted in the accompanying drawings does not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A method for processing wind power load test data with adaptive data compression, characterized in that: The method comprises: Collecting load test data of the wind turbine generator set using a nested sliding window, wherein the nested sliding window includes a basic sliding window and an extended sliding window, and the basic sliding window can slide multiple times within the extended sliding window, and the load test data includes stress data, vibration data, torque data, and wind speed data; Performing multi-scale feature extraction on the collected data of the nested sliding windows respectively, and performing multi-scale feature fusion to establish a comprehensive change score for each data point; Acquire the spatial distribution of acquisition sensors, configure sensor adjacency coefficients according to the spatial distribution and sensor functions, perform spatial consistency detection of data points based on the sensor adjacency coefficients, and locate key frames according to the spatial consistency detection results and the comprehensive change score; The key frame is synchronized to an adaptive compression channel, a data compression strategy is constructed, and compression management of the load test data is performed using the data compression strategy.

2. The method for processing wind power load test data with adaptive data compression according to claim 1, characterized in that: The configuring of the sensor adjacency coefficient according to the spatial distribution and the sensor function includes: Calculating a proximity distance of a sensor based on the spatial distribution, and determining whether the proximity distance is less than a preset distance threshold; If the adjacent distance is less than the preset distance threshold, generating a first authentication pass result; configuring a function vector according to the sensor function, calculating function similarity by using cosine similarity and the function vector, and generating a second authentication pass result if the function similarity is higher than a preset function threshold; When the first authentication pass result and the second authentication pass result exist at the same time, the sensor adjacency coefficient is calculated.

3. The method for processing wind power load test data with adaptive data compression according to claim 2, characterized in that: The calculating of the sensor adjacency coefficient comprises: The sensor proximity factor is calculated using the following formula: ； in, Characterizing Sensors and sensors The sensor adjacency coefficient ranges from [0,1], Characterizing Sensors and sensors The adjacent distance, Characterizes the spatial attenuation factor, Characterizing Sensors and sensors similarity of functions.

4. The method for processing wind power load test data with adaptive data compression according to claim 3, characterized in that: After calculating the sensor adjacency coefficient, the method further comprises: The differential deviation of the sensor data is calculated by the formula: ； in, Characterizing Sensors and sensors In time The data difference at Characterizing Sensors In time The measured value at Characterizing Sensors In time The measured value at A consistency score is calculated based on the differential deviation and the sensor adjacency coefficient: ； in, Representation in time Sensor The consistency score, Characterization and Sensors A set of sensors with sensor adjacency coefficients, is the data difference sensitivity coefficient.

5. The method for processing wind power load test data with adaptive data compression according to claim 1, characterized in that: The method of collecting load test data of wind turbine generator sets by using nested sliding windows includes: Configuring a default window value of a basic sliding window within the nested sliding window, calling the basic sliding window according to the default window value to perform data pre-collection of load test data, and establishing a pre-collection data set; Performing a data stabilization analysis on the pre-collected data set, and generating a reconstruction parameter according to the data stabilization analysis result, wherein the reconstruction parameter includes a window value of a basic sliding window and an extended value of an extended sliding window; After the nested sliding window is reconstructed according to the window value and the extended value, load test data of the wind turbine generator set is collected.

6. The method for processing wind power load test data with adaptive data compression according to claim 5, characterized in that: The reconstructing the nested sliding window according to the window value and the extension value includes: Perform data fluctuation analysis on pre-collected data sets and establish expansion bias based on the data fluctuation analysis results; After reconstructing the basic sliding window with the window value, reconstructing the extended sliding window based on the extended bias and the extended value with the reconstructed basic sliding window as the center; After collecting the load test data of the wind turbine generator set using the reconstructed nested sliding window, a window data set corresponding to the reconstructed extended sliding window is obtained; After evenly dividing the window data set, the fluctuation range under the evenly divided window is calculated; The adaptive sliding step size is configured according to the fluctuation amplitude and the sensitivity coefficient.

7. The method for processing wind power load test data with adaptive data compression according to claim 1, characterized in that: The step of synchronizing the key frame to the adaptive compression channel and constructing a data compression strategy includes: performing a quantitative analysis of the key frames; If the quantity analysis result is higher than a preset quantity threshold, the key frame is divided into a primary key frame and a secondary key frame; The main key frame is compressed with original data retained, the secondary key frame is compressed with reference low loss based on the main key frame, and the non-key frame is compressed with loss to build a data compression strategy.

8. The wind power load test data processing system with adaptive data compression is characterized by: The system is used to implement the wind power load test data processing method with adaptive data compression according to any one of claims 1 to 7, and the system comprises: A load test data acquisition module, used to acquire load test data of a wind turbine using a nested sliding window, wherein the nested sliding window includes a basic sliding window and an extended sliding window, and the basic sliding window can slide multiple times within the extended sliding window, and the load test data includes stress data, vibration data, torque data, and wind speed data; A multi-scale feature extraction and fusion module, used to extract multi-scale features from the collected data of the nested sliding windows respectively, and perform multi-scale feature fusion to establish a comprehensive change score for each data point; A key frame positioning module is used to obtain the spatial distribution of the acquisition sensor, configure the sensor adjacency coefficient according to the spatial distribution and the sensor function, perform spatial consistency detection of the data points based on the sensor adjacency coefficient, and locate the key frame according to the spatial consistency detection result and the comprehensive change score; The data compression module is used to synchronize the key frames to an adaptive compression channel, construct a data compression strategy, and use the data compression strategy to perform compression management of the load test data.

Images