A six-degree-of-freedom load detection system and method for wind turbine blades

Abstract

This invention relates to the field of defect detection technology, and discloses a six-degree-of-freedom load detection system and method for wind turbine blades. The system includes a multi-channel conditioning module, a high-precision synchronous data conversion module, a real-time load decoupling module, a time-domain-frequency-domain fusion processing module, an intelligent condition monitoring and early warning module, and a comprehensive data visualization and integrated monitoring module. It synchronously acquires and conditions multiple raw strain signals, filters out noise, converts them into analog signals, and then performs synchronous analog-to-digital conversion to obtain digital data. It uses phase synchronization compensation and strain amplitude decoupling methods to calculate independent load components in real time. Through fast Fourier transform and order analysis, it extracts the load time history and dynamic response characteristics, analyzes and generates condition warnings in real time, and integrates them into a visualization monitoring platform to achieve real-time monitoring of the entire load state of the wind turbine blades. This invention can improve the accuracy of six-degree-of-freedom load detection.

Description

Technical Field

[0001] This invention relates to the field of defect detection technology, and in particular to a six-degree-of-freedom load detection system and method for wind turbine blades. Background Technology

[0002] Existing technologies for six-degree-of-freedom load testing of wind turbine blades lack systematic optimization in processing multiple raw strain signals. They only perform simple acquisition without standardized processing and targeted high-frequency noise filtering, resulting in a large amount of interference information mixed in the raw signals, making it impossible to form a pure analog signal. At the same time, the analog-to-digital conversion stage does not adopt a combination of time-division multiplexing sampling and programmable gain amplification, nor does it perform digital filtering compensation. This results in insufficient accuracy and poor stability of the converted digital strain data, making it difficult to meet the basic data quality requirements for subsequent load decoupling, thus creating hidden dangers at the data level for six-degree-of-freedom load testing.

[0003] Existing technologies have significant shortcomings in load decoupling and condition monitoring. On the one hand, a decoupling mechanism combining phase synchronization compensation and strain amplitude decoupling has not been established. Processing digital strain data using only a single method cannot accurately separate independent load components in the six degrees of freedom, easily leading to load component confusion and resulting in a large deviation between the detection results and the actual load state of the blade. On the other hand, there is a lack of time-domain-frequency domain fusion load characteristic analysis methods, and no real-time early warning mechanism based on modal parameter identification and adaptive threshold comparison has been built. This means that it is impossible to comprehensively obtain the load time history and frequency response characteristics, or to identify load anomalies in a timely manner. Furthermore, the lack of a multi-dimensional data visualization platform makes it difficult to achieve real-time monitoring of the entire blade load state, resulting in insufficient scientific rigor and practicality of the overall detection process. Summary of the Invention

[0004] This invention provides a six-degree-of-freedom load detection system and method for wind turbine blades to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, this invention provides a six-degree-of-freedom load detection system for wind turbine blades. The system includes a multi-channel conditioning module, a high-precision synchronous data conversion module, a real-time load decoupling module, a time-domain-frequency domain fusion processing module, an intelligent condition monitoring and early warning module, and a comprehensive data visualization and integrated monitoring module, wherein:

[0006] The multi-channel conditioning module is used to acquire multiple raw strain signals and perform noise filtering on the multiple raw strain signals through a channel signal conditioning circuit to generate analog signals of the multiple raw strain signals.

[0007] The high-precision synchronous data conversion module is used to perform synchronous analog-to-digital conversion on the analog signal to obtain digital strain data of the analog signal;

[0008] The real-time load decoupling module is used to decouple the digital strain data in real time based on the phase synchronization compensation and strain amplitude decoupling method, so as to obtain the independent load components in the six degrees of freedom of the digital strain data.

[0009] The time-frequency domain fusion processing module is used to perform time-frequency domain fusion processing on the independent load components based on the fast Fourier transform and order ratio analysis method to obtain the load time history data and frequency response characteristic data of the independent load components.

[0010] The intelligent state monitoring and early warning module is used to perform real-time state signal analysis on the frequency response characteristic data based on an adaptive threshold comparison and modal parameter identification method, and obtain a state early warning signal of the frequency response characteristic data.

[0011] The integrated data visualization and monitoring module is used to integrate the load time history data, the frequency response characteristic data and the status warning signal into the visualization monitoring platform to monitor the load status of the wind turbine blades in real time.

[0012] In a preferred embodiment, when the multi-channel conditioning module acquires multiple raw strain signals and performs noise filtering on the multiple raw strain signals through a channel signal conditioning circuit to generate analog signals of the multiple raw strain signals, it is specifically used for:

[0013] The multiple raw strain signals are standardized to obtain the standard strain signals of the multiple raw strain signals.

[0014] The standard strain signal is subjected to high-frequency noise filtering to obtain the effective frequency band signal of the standard strain signal;

[0015] The effective frequency band signal is subjected to signal simulation analysis to obtain the analog signal of the effective frequency band signal.

[0016] In a preferred embodiment, when the high-precision synchronous data conversion module performs synchronous analog-to-digital conversion on the analog signal to obtain digital strain data of the analog signal, it is specifically used for:

[0017] The analog signal is time-division multiplexed and sampled to obtain multiple discrete signals of the analog signal;

[0018] The multi-channel discrete signals are amplified by programmable gain control to obtain the multi-channel amplified signals of the multi-channel discrete signals;

[0019] The multi-channel amplified signals are quantized and sequence-encoded to obtain a preliminary digital signal sequence of the multi-channel amplified signals;

[0020] The preliminary digital signal sequence is digitally filtered and compensated to obtain digital strain data of the preliminary digital signal sequence.

[0021] In a preferred embodiment, when the real-time load decoupling module performs real-time decoupling of the digital strain data based on a phase synchronization compensation and strain amplitude decoupling method to obtain the independent load components in the six degrees of freedom directions of the digital strain data, it is specifically used for:

[0022] The digital strain data is reconstructed in the spatial domain to obtain the spatial distribution strain data of the digital strain data.

[0023] Phase synchronization compensation is performed on the spatially distributed strain data to obtain aligned strain data of the spatially distributed strain data.

[0024] Based on the coupling relationship between strain and load, the amplitude decoupling processing is performed on the aligned strain data to separate the independent load components.

[0025] Based on the projection relationship in the six degrees of freedom, the independent load components are linearly synthesized to obtain the independent load components in the six degrees of freedom in the digital strain data.

[0026] In a preferred embodiment, when the real-time load decoupling module performs linear synthesis of the independent load components based on the projection relationship in the six degrees of freedom directions to obtain the independent load components in the six degrees of freedom directions in the digital strain data, it is specifically used for:

[0027] Establish the projection mapping relationship between the six degrees of freedom directions and the mutually independent load components;

[0028] A projection coefficient matrix is ​​generated based on the projection mapping relationship. The projection coefficient matrix contains the contribution weights of the load components in different degrees of freedom directions.

[0029] The independent load components are linearly weighted and synthesized with the projection coefficient matrix to obtain a preliminary synthesized vector, wherein the calculation formula for the preliminary synthesized vector is as follows:

[0030] In the formula, For the initial synthesis vector, The projection coefficient matrix is... These are mutually independent load component vectors;

[0031] The preliminary synthesized vector is normalized to obtain the independent load components in the six degrees of freedom of the digital strain data.

[0032] In a preferred embodiment, when the time-domain-frequency domain fusion processing module performs time-domain-frequency domain fusion processing on the independent load components based on the Fast Fourier Transform and order ratio analysis method to obtain the load time history data and frequency response characteristic data of the independent load components, it is specifically used for:

[0033] The independent load components are subjected to spectral conversion processing to obtain the spectral data of the independent load components;

[0034] Multidimensional characteristic frequency analysis is performed on the spectrum data to obtain the characteristic frequency components of the spectrum data;

[0035] Based on the characteristic frequency components and the original time-domain signal, the independent load components are subjected to joint time-frequency domain reconstruction analysis to obtain the load time history data and frequency response characteristic data of the independent load components.

[0036] In a preferred embodiment, when the time-domain-frequency domain fusion processing module performs time-frequency domain joint reconstruction analysis on the independent load components based on the characteristic frequency components and the original time-domain signal to obtain the load time history data and frequency response characteristic data of the independent load components, it is specifically used for:

[0037] The frequency domain feature vectors of the characteristic frequency components are extracted by principal component extraction in the frequency domain to obtain the frequency domain feature vectors of the characteristic frequency components.

[0038] The original time-domain signal is sampled and aligned to obtain the time-domain vector of the original time-domain signal;

[0039] The frequency domain feature vector and the time domain vector are combined to generate a time-frequency domain vector set;

[0040] The time-frequency domain vector set is dynamically weighted to obtain the weighting coefficients of the time-frequency domain vector set;

[0041] Based on the synchronous compression transformation rule, the frequency domain feature vector and the time domain vector are linearly fused to obtain the transformation result of the independent load component. The calculation formula of the transformation result is as follows:

[0042] In the formula, The transformation result is... For the time-frequency domain vector set, the first... The weighting coefficients of each component, The first of the frequency domain feature vectors One portion, The first time domain vector One portion, For regularization parameters, For L2 norm operations, For element-wise multiplication, The frequency domain feature vector, For the time domain vector, The number of feature dimensions;

[0043] The transformation result is then reconstructed by inverse transformation to obtain the load time history data and frequency response characteristic data of the independent load components.

[0044] In a preferred embodiment, when the intelligent state monitoring and early warning module performs real-time state signal analysis on the frequency response characteristic data based on an adaptive threshold comparison and modal parameter identification method to obtain a state early warning signal for the frequency response characteristic data, it is specifically used for:

[0045] Modal parameter identification is performed on the load time history data and frequency response characteristic data to obtain the modal parameters of the load time history data and frequency response characteristic data;

[0046] The modal parameters are dynamically and adaptively adjusted to obtain the threshold range of the modal parameters;

[0047] Based on the threshold range, the modal parameters are analyzed and compared in real time to obtain the abnormal state characteristics of the modal parameters;

[0048] Based on the severity level of the abnormal state characteristics, a state warning signal for the abnormal state characteristics is obtained.

[0049] In a preferred embodiment, when the integrated data visualization and monitoring module integrates the load time history data, the frequency response characteristic data, and the status warning signal into the visualization monitoring platform to monitor the load status of the wind turbine blades in real time, it is specifically used for:

[0050] The load time history data is reconstructed in the time domain to obtain the time domain waveform primitives of the load time history data;

[0051] The frequency response characteristic data is subjected to frequency domain spectrum construction to obtain the frequency domain spectrum of the frequency response characteristic data;

[0052] The state warning signal is state encoded to obtain the state encoding matrix of the state warning signal;

[0053] The time-domain waveform primitives, the frequency-domain spectrum, and the state coding matrix are used to construct a multi-dimensional spatiotemporal framework to generate a comprehensive display framework for the wind turbine blade.

[0054] Based on the visualization monitoring platform, the integrated display framework is monitored in real time to obtain the load status of the wind turbine blades.

[0055] To address the aforementioned problems, the present invention also provides a method for detecting six-degree-of-freedom loads on wind turbine blades, the method comprising:

[0056] S1. Acquire multiple raw strain signals, and perform noise filtering on the multiple raw strain signals through a channel signal conditioning circuit to generate analog signals of the multiple raw strain signals;

[0057] S2. Perform synchronous analog-to-digital conversion on the analog signal to obtain digital strain data of the analog signal;

[0058] S3. Based on the phase synchronization compensation and strain amplitude decoupling method, the digital strain data is decoupled in real time to obtain the independent load components in the six degrees of freedom of the digital strain data;

[0059] S4. Based on the Fast Fourier Transform and order ratio analysis method, the independent load components are subjected to time-domain-frequency domain fusion processing to obtain the load time history data and frequency response characteristic data of the independent load components;

[0060] S5. Based on the adaptive threshold comparison and modal parameter identification method, perform real-time state signal analysis on the frequency response characteristic data to obtain the state warning signal of the frequency response characteristic data;

[0061] S6. Integrate the load time history data, the frequency response characteristic data, and the status warning signal into a visual monitoring platform to monitor the load status of the wind turbine blades in real time.

[0062] Compared with the prior art, the present invention has the following beneficial effects:

[0063] 1. This invention standardizes and filters high-frequency noise from multiple raw strain signals using a multi-channel conditioning module. Combined with time-division multiplexing sampling, programmable gain amplification, and digital filtering compensation from a high-precision synchronous data conversion module, it effectively improves signal quality and data accuracy, providing clean digital strain data for subsequent load decoupling. Simultaneously, the real-time load decoupling module employs phase synchronization compensation and strain amplitude decoupling methods. Through spatial domain reconstruction, alignment processing, and linear synthesis, it accurately separates the independent load components in the six degrees of freedom, significantly improving the accuracy and completeness of load detection.

[0064] 2. This invention achieves joint time-frequency domain reconstruction of independent load components by utilizing the Fast Fourier Transform and order ratio analysis of the time-domain-frequency domain fusion processing module, enabling comprehensive extraction of load time history data and frequency response characteristic data. The intelligent status monitoring and early warning module generates status early warning signals in real time through modal parameter identification and adaptive threshold comparison, ensuring timely detection of load anomalies. Furthermore, the comprehensive data visualization and integrated monitoring module integrates multi-dimensional data into a visualization platform, achieving real-time monitoring of all states. This not only improves the intelligence level of load detection but also provides intuitive and reliable decision-making basis for the safe operation and maintenance of wind turbine blades. Attached Figure Description

[0065] Figure 1 This is a system architecture diagram of a six-degree-of-freedom load detection system for wind turbine blades provided in an embodiment of the present invention;

[0066] Figure 2 This is a flowchart illustrating a six-degree-of-freedom load detection method for wind turbine blades, provided as an embodiment of the present invention.

[0067] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0068] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments belong to some, but not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0069] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “the” and “the” as used in the embodiments of this invention are also intended to include the plural forms, and “multiple” generally includes at least two unless the context clearly indicates otherwise.

[0070] Depending on the context, the word "if" or "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0071] Furthermore, the timing of the steps in the following method embodiments is merely an example and not a strict limitation.

[0072] In practice, the server-side equipment deployed in a six-degree-of-freedom load detection system for a wind turbine blade may consist of one or more devices. This system can be implemented as a business instance, a virtual machine, or hardware devices. For example, it can be implemented as a business instance deployed on one or more devices in a cloud node. Simply put, it can be understood as software deployed on a cloud node to provide a six-degree-of-freedom load detection system for wind turbine blades to various user terminals. Alternatively, it can be implemented as a virtual machine deployed on one or more devices in a cloud node, with application software installed to manage various user terminals. Or, it can also be implemented as a server composed of numerous identical or different types of hardware devices, with one or more hardware devices configured to provide a six-degree-of-freedom load detection system for wind turbine blades to various user terminals.

[0073] In terms of implementation, a six-degree-of-freedom load detection system for wind turbine blades and a user terminal are mutually compatible. That is, if the six-degree-of-freedom load detection system for wind turbine blades is implemented as an application installed on a cloud service platform, then the user terminal is implemented as a client that establishes a communication connection with the application; or if the six-degree-of-freedom load detection system for wind turbine blades is implemented as a website, then the user terminal is implemented as a webpage; or if the six-degree-of-freedom load detection system for wind turbine blades is implemented as a cloud service platform, then the user terminal is implemented as a mini-program in an instant messaging application.

[0074] like Figure 1 The figure shown is a system architecture diagram of a six-degree-of-freedom load detection system for wind turbine blades provided in an embodiment of the present invention.

[0075] The six-degree-of-freedom load detection system 100 for wind turbine blades described in this invention can be installed on a cloud server. In terms of implementation, it can function as one or more service devices, or as an application installed on the cloud (e.g., a mobile service operator's server, server cluster, etc.), or it can be developed into a website. Depending on the functions implemented, the six-degree-of-freedom load detection system 100 for wind turbine blades may include a multi-channel conditioning module 101, a high-precision synchronous data conversion module 102, a real-time load decoupling module 103, a time-domain-frequency-domain fusion processing module 104, an intelligent status monitoring and early warning module 105, and a comprehensive data visualization and integrated monitoring module 106. The modules described in this invention can also be referred to as units, which are a series of computer program segments that can be executed by an electronic device's processor and perform a fixed function, stored in the electronic device's memory.

[0076] In this embodiment of the invention, in a six-degree-of-freedom load detection system for wind turbine blades, each of the above-mentioned modules can be implemented independently and called upon other modules. This calling can be understood as a module connecting to multiple modules of another type and providing corresponding services to those connected modules. The six-degree-of-freedom load detection system for wind turbine blades provided by this embodiment of the invention allows for adjustment of the applicable scope of the system architecture without modifying the program code. This is achieved by adding modules and directly calling them, enabling cluster-based horizontal expansion and flexibly expanding the six-degree-of-freedom load detection system for wind turbine blades. In practical applications, the above modules can be set in the same device or different devices, or they can be set in a virtual device, such as a service instance in a cloud server.

[0077] The following describes, with reference to specific embodiments, each component and its specific workflow of a six-degree-of-freedom load detection system for wind turbine blades:

[0078] The multi-channel conditioning module 101 is used to acquire multiple raw strain signals and perform noise filtering on the multiple raw strain signals through a channel signal conditioning circuit to generate analog signals of the multiple raw strain signals.

[0079] In this embodiment of the invention, when the multi-channel conditioning module acquires multiple raw strain signals and performs noise filtering on the multiple raw strain signals through a channel signal conditioning circuit to generate analog signals of the multiple raw strain signals, it is specifically used for:

[0080] The multiple raw strain signals are standardized to obtain the standard strain signals of the multiple raw strain signals.

[0081] The standard strain signal is subjected to high-frequency noise filtering to obtain the effective frequency band signal of the standard strain signal;

[0082] The effective frequency band signal is subjected to signal simulation analysis to obtain the analog signal of the effective frequency band signal.

[0083] Specifically, when standardizing multiple raw strain signals, the numerical range of each raw strain signal is first obtained, the highest and lowest values ​​of each signal are determined, and a unified target numerical range is set. Each value in each raw strain signal is converted to the corresponding position in the target numerical range according to its proportion in its own numerical range. During the conversion process, it is ensured that the trend and relative relationship of the signal change do not change, so that the raw strain signals of different channels can be presented on the same numerical scale. After such processing, the standard strain signal of multiple raw strain signals is obtained.

[0084] Furthermore, when filtering high-frequency noise from the standard strain signal, an RC low-pass filter circuit consisting of resistors and capacitors is used. The standard strain signal is input into this circuit. The capacitor in the circuit presents a small impedance to the high-frequency signal, so that most of the high-frequency noise components are released to ground through the capacitor. The resistor has a small attenuation to the low-frequency signal, allowing the useful signal in the standard strain signal with a frequency lower than the cutoff frequency of the filter circuit to pass through with almost no attenuation. In this way, the high-frequency noise contained in the standard strain signal is removed, and the useful signal components are retained to obtain the effective frequency band signal of the standard strain signal.

[0085] Furthermore, when performing signal analog analysis on the effective frequency band signal, an analog signal conditioning circuit composed of operational amplifiers is used to process the effective frequency band signal. After receiving the effective frequency band signal, the operational amplifier adjusts the output voltage according to the amplitude change of the signal, so that the amplitude change of the output voltage is consistent with the amplitude change of the effective frequency band signal. At the same time, the capacitors and resistors in the circuit work together to ensure that the waveform of the output signal is the same as the waveform shape of the effective frequency band signal. Throughout the process, the signal always exists in the form of a continuously changing electrical signal, and finally the analog signal of the effective frequency band signal is obtained.

[0086] In summary, standardizing the processing of standard strain signals eliminates the differences in range between multiple signals, avoids deviations in subsequent processing, and provides a unified signal basis for high-frequency noise filtering.

[0087] In summary, high-frequency noise filtering removes signals in the effective frequency band, eliminates high-frequency interference, preserves the true strain signal of the blade, and improves the accuracy of subsequent analysis.

[0088] In summary, the analog signal obtained from the signal simulation analysis enables the effective signal to be adapted to the subsequent analog-to-digital conversion module, providing a reliable source for generating accurate digital strain data.

[0089] The high-precision synchronous data conversion module 102 is used to perform synchronous analog-to-digital conversion on the analog signal to obtain digital strain data of the analog signal.

[0090] In this embodiment of the invention, when the high-precision synchronous data conversion module performs synchronous analog-to-digital conversion on the analog signal to obtain digital strain data of the analog signal, it is specifically used for:

[0091] The analog signal is time-division multiplexed and sampled to obtain multiple discrete signals of the analog signal;

[0092] The multi-channel discrete signals are amplified by programmable gain control to obtain the multi-channel amplified signals of the multi-channel discrete signals;

[0093] The multi-channel amplified signals are quantized and sequence-encoded to obtain a preliminary digital signal sequence of the multi-channel amplified signals;

[0094] The preliminary digital signal sequence is digitally filtered and compensated to obtain digital strain data of the preliminary digital signal sequence.

[0095] Specifically, when performing time-division multiplexing sampling on analog signals, a high-speed electronic switch is used to sequentially turn on the analog signals of different channels in a fixed time sequence. Each time a channel is turned on, the analog signal of that channel is sampled once to obtain a discrete signal value. After the sampling of one channel is completed, the switch is immediately switched to the next channel to perform the same sampling operation. This process is repeated until the analog signals of all channels have been sampled once. By using this time-sequential sampling method, the continuous analog signal is converted into discrete signal values ​​of multiple channels at different time points, resulting in a multi-channel discrete analog signal.

[0096] Furthermore, when performing programmable gain amplification on multiple discrete signals, the multiple discrete signals are input into the programmable gain amplifier respectively. According to the amplitude characteristics of each discrete signal and the requirements of subsequent processing, the control circuit sends corresponding control signals to the programmable gain amplifier to adjust the gain of the amplifier. This ensures that the discrete signals with weaker signals are amplified to an appropriate amplitude, while the discrete signals with stronger signals are not over-amplified to avoid distortion. After such targeted amplification processing, a multi-channel amplified signal of multiple discrete signals is obtained.

[0097] Furthermore, when performing quantization sequence encoding on the multi-channel amplified signals, the amplitude of each amplified signal is compared with multiple preset discrete levels to determine the discrete level corresponding to the amplitude. Then, each discrete level is represented by a specific set of binary numbers. According to the sampling order of the multi-channel amplified signals, the binary numbers corresponding to each signal are arranged sequentially to form a continuous digital sequence. Through this method of converting analog amplitudes into binary numbers and arranging them in order, a preliminary digital signal sequence of the multi-channel amplified signals is obtained.

[0098] Furthermore, when performing digital filtering compensation on the preliminary digital signal sequence, a digital filter is used to process the preliminary digital signal sequence. The filter will identify abnormal values ​​in the sequence caused by noise interference and correct them according to the normal values ​​before and after the abnormal values. At the same time, for the amplitude attenuation or phase shift that may occur during the signal conversion process, it will be adjusted according to the preset compensation rules to restore the amplitude and phase of the signal to their proper state. After such filtering and compensation processing, the digital strain data of the preliminary digital signal sequence is obtained.

[0099] In summary, time-division multiplexing samples multiple discrete signals, acquiring multiple channels of analog signals in turn according to a fixed time sequence. It eliminates the need to configure a separate sampling unit for each signal, simplifies the hardware structure while ensuring synchronous acquisition, solves the problems of asynchronous acquisition of multiple channels of signals and high hardware costs, and provides a discrete signal basis for subsequent unified processing.

[0100] In summary, programmable gain amplification amplifies multiple signals and adjusts the amplification factor according to the amplitude characteristics of each discrete signal to avoid weak signals being overwhelmed and strong signals being distorted. This solves the problem of low processing accuracy caused by differences in signal amplitude and ensures that all signals are within the appropriate amplitude range for subsequent encoding.

[0101] In summary, quantization sequence encoding yields a preliminary digital signal sequence, converting analog amplitudes into binary numbers and arranging them in order. This achieves the core conversion from analog to digital signals, solving the problem that analog signals cannot be directly processed by digital modules, and providing a digital signal carrier for digital filtering compensation.

[0102] In summary, digital filtering compensation yields digital strain data, corrects abnormal values, compensates for amplitude and phase deviations, removes interference during analog-to-digital conversion, and solves the problems of insufficient accuracy and poor stability of the initial digital signal sequence, providing high-quality digital data support for real-time load decoupling.

[0103] The real-time load decoupling module 103 is used to decouple the digital strain data in real time based on the phase synchronization compensation and strain amplitude decoupling method to obtain the independent load components in the six degrees of freedom of the digital strain data.

[0104] In this embodiment of the invention, when the real-time load decoupling module performs real-time decoupling of the digital strain data based on the phase synchronization compensation and strain amplitude decoupling method to obtain the independent load components in the six degrees of freedom of the digital strain data, it is specifically used for:

[0105] The digital strain data is reconstructed in the spatial domain to obtain the spatial distribution strain data of the digital strain data.

[0106] Phase synchronization compensation is performed on the spatially distributed strain data to obtain aligned strain data of the spatially distributed strain data.

[0107] Based on the coupling relationship between strain and load, the amplitude decoupling processing is performed on the aligned strain data to separate the independent load components.

[0108] Based on the projection relationship in the six degrees of freedom, the independent load components are linearly synthesized to obtain the independent load components in the six degrees of freedom in the digital strain data.

[0109] When the real-time load decoupling module performs linear synthesis of the independent load components based on the projection relationship in six degrees of freedom to obtain the independent load components in the six degrees of freedom in the digital strain data, it is specifically used for:

[0110] Establish the projection mapping relationship between the six degrees of freedom directions and the mutually independent load components;

[0111] A projection coefficient matrix is ​​generated based on the projection mapping relationship. The projection coefficient matrix contains the contribution weights of the load components in different degrees of freedom directions.

[0112] The independent load components are linearly weighted and synthesized with the projection coefficient matrix to obtain a preliminary synthesized vector, wherein the calculation formula for the preliminary synthesized vector is as follows:

[0113] In the formula, For the initial synthesis vector, The projection coefficient matrix is... These are mutually independent load component vectors;

[0114] The preliminary synthesized vector is normalized to obtain the independent load components in the six degrees of freedom of the digital strain data.

[0115] Specifically, when reconstructing the spatial domain of digital strain data, the actual spatial location of the acquisition point corresponding to the digital strain data on the structure under test is first determined. The digital strain data of each acquisition point is then matched one-to-one with its spatial coordinates. Then, according to the spatial distribution pattern of the acquisition points, the scattered digital strain data is integrated into a virtual spatial model that matches the shape of the structure under test, so that each spatial location has corresponding strain data. By combining data with spatial location and integrating it into the model, the spatial distribution strain data of digital strain data is obtained.

[0116] Furthermore, when performing phase synchronization compensation on spatially distributed strain data, the time difference of the spatially distributed strain data collected at each acquisition point is first determined, and the earliest acquisition time among all acquisition points is identified as the reference time. Then, based on the time difference of each acquisition point relative to the reference time, the spatially distributed strain data of that acquisition point is adjusted by time offset to ensure that the strain data of all acquisition points are consistent in the time dimension, ensuring that the strain data of different spatial locations correspond to the same time node. After such time synchronization adjustment, the aligned strain data of spatially distributed strain data is obtained.

[0117] Furthermore, based on the coupling relationship between strain and load, when performing amplitude decoupling processing on aligned strain data, the known correspondence between strain and load is first clarified, that is, different types of load will produce strain with specific characteristics. Based on this correlation, the different load characteristics contained in the aligned strain data are analyzed, and strain data with the same load characteristics are grouped into one category. Then, by separating strain data of different categories, the mutual interference between various types of strain data is removed, thereby separating the independent load components.

[0118] Furthermore, based on the projection relationship of the six degrees of freedom, when linearly synthesizing mutually independent load components, the load projection rules corresponding to each of the six degrees of freedom are first clarified, that is, the contribution ratio of each independent load component in different degrees of freedom. Then, according to the projection rules, each mutually independent load component is decomposed into the six degrees of freedom. The decomposition results of all load components in the same degree of freedom are then superimposed to obtain the total load in each degree of freedom. Finally, the total load in the six degrees of freedom is integrated to obtain the independent load components in the six degrees of freedom in the digital strain data.

[0119] Specifically, when establishing the projection mapping relationship between the six degrees of freedom directions and the independent load components, first clarify the six degrees of freedom directions, namely the translation directions along the x-axis, y-axis, and z-axis, and the rotation directions around the x-axis, y-axis, and z-axis. Then determine the correlation of each independent load component's influence in these six directions. Through experimental measurement, determine the degree of influence of each load component on each degree of freedom direction, and determine which load components will act on which degrees of freedom directions and the correlation of their actions. In this way, the projection mapping relationship between the six degrees of freedom directions and the independent load components is established.

[0120] Furthermore, when generating the projection coefficient matrix based on the projection mapping relationship, according to the established projection mapping relationship, the influence degree of each independent load component in the six degrees of freedom is represented by a specific value. This value represents the contribution weight of the load component in the corresponding degree of freedom direction. According to the order of the load components and the order of the six degrees of freedom directions, these contribution weights are arranged in order to form a matrix, where the rows of the matrix correspond to the independent load components, the columns correspond to the six degrees of freedom directions, and each element in the matrix is ​​the contribution weight of the corresponding load component in the corresponding degree of freedom direction, thereby generating the projection coefficient matrix.

[0121] Furthermore, when performing linear weighted synthesis of independent load components with the projection coefficient matrix, the value of each independent load component is multiplied by the contribution weights in the row corresponding to that load component in the projection coefficient matrix to obtain the weighted value of that load component in each degree of freedom direction. Then, the weighted values ​​of all independent load components in the same degree of freedom direction are added together to obtain the synthesized value in each degree of freedom direction. The synthesized values ​​of these six degrees of freedom directions are arranged in order to form a vector, resulting in the preliminary synthesized vector.

[0122] Furthermore, when normalizing the preliminary composite vector, the overall range of the composite values ​​in the six degrees of freedom directions in the preliminary composite vector is first determined, the maximum and minimum values ​​are found, the difference between the maximum and minimum values ​​is calculated, the minimum value is subtracted from the composite value in each degree of freedom direction, and then divided by the difference between the maximum and minimum values, so that each composite value is converted to the range between 0 and 1. After such processing, the independent load components in the six degrees of freedom directions in the digital strain data are obtained.

[0123] Specifically, the initial composite vector in the formula is obtained by linearly weighting the independent load components with the projection coefficient matrix. Specifically, the value of each independent load component is multiplied by the contribution weight of the corresponding row in the projection coefficient matrix to obtain the weighted value of each load component in each degree of freedom direction. Then, the weighted values ​​of all load components in the same degree of freedom direction are added together. Finally, the composite values ​​in the six degrees of freedom directions are arranged in order to form a vector.

[0124] Furthermore, the projection coefficient matrix in the formula is generated based on the projection mapping relationship between the six degrees of freedom and the independent load components. First, the influence degree (i.e., contribution weight) of each independent load component in the six degrees of freedom is determined according to the projection mapping relationship. Then, these contribution weights are arranged in order according to the order of load components and the order of the six degrees of freedom to form a matrix. The rows of the matrix correspond to the independent load components, the columns correspond to the six degrees of freedom, and the elements are the contribution weights of the corresponding load components in the corresponding degrees of freedom.

[0125] Furthermore, the source of the load component vector of the mutually independent load components in the formula is the mutually independent load components obtained after amplitude decoupling processing of the aligned strain data based on the coupling relationship between strain and load, and then the vector formed by arranging these mutually independent load components in a preset order.

[0126] Furthermore, the significance of the formula lies in the fact that by operating the projection coefficient matrix with the load component vectors of the independent load components, a linear weighted synthesis of the independent load components in the six degrees of freedom is achieved. This transforms the dispersed and independent load components into a preliminary synthesis vector that can reflect the load situation in the six degrees of freedom, providing a basis for obtaining the independent load components in the six degrees of freedom, and clearly demonstrating the comprehensive effect of the independent load components in different degrees of freedom.

[0127] Furthermore, the formula shows that when the value of a certain load component in the load component vector of load components with mutually independent parameters increases, and the contribution weight of a certain degree of freedom direction in the corresponding row of that load component in the projection coefficient matrix is ​​fixed, the synthesized value of that degree of freedom direction in the initial synthesized vector will increase accordingly; when the parameters The contribution weight of a certain degree of freedom direction increases, and the parameter When the values ​​of the corresponding load components are fixed, The resultant value in that degree of freedom direction will also increase; conversely, if The value of the medium load component decreased or The corresponding contribution weight decreases. The resultant value in the corresponding degree of freedom direction will decrease accordingly, and the overall result will exhibit parametric characteristics. and numerical changes and The trend of positive correlation between the changes in the corresponding dimension values.

[0128] In summary, by associating discrete digital strain data with the spatial location of the blade acquisition points, the spatial distribution characteristics of strain can be restored, solving the problem of data disconnection from the actual structural location and providing a complete spatial data foundation for subsequent phase synchronization compensation.

[0129] In summary, eliminating the time difference between data from different acquisition points ensures that strain data at each spatial location are synchronized in time, avoiding load decoupling errors caused by time deviations, and providing accurate data with consistent time for amplitude decoupling processing.

[0130] In summary, amplitude decoupling separates independent load components, splits mixed loads based on the coupling relationship between strain and load, solves the problem of load component confusion, and provides a single, pure load basis for the synthesis of loads in six degrees of freedom directions.

[0131] In summary, by combining the projection relationship of degrees of freedom to integrate independent load components, the loads in each degree of freedom direction can be accurately obtained, solving the problem that existing technologies cannot fully and accurately separate six-degree-of-freedom loads, and improving the completeness and accuracy of blade load detection.

[0132] In summary, establishing projection mapping relationships, clarifying the correlation between load components and degree of freedom directions, and resolving corresponding ambiguity issues provide a basis for generating the projection coefficient matrix.

[0133] In summary, generating a weighted projection coefficient matrix quantifies the influence of load components, avoids computational bias, and provides a foundation for linear weighted synthesis.

[0134] In summary, linear weighted synthesis yields a preliminary synthesis vector, integrates the contributions of load components, solves the problem that components cannot directly correspond to degrees of freedom, and builds a transitional bridge.

[0135] In summary, normalization yields six-degree-of-freedom independent load components, unifies the numerical range, corrects deviations, improves detection accuracy, and meets monitoring data requirements.

[0136] The time-frequency domain fusion processing module 104 is used to perform time-frequency domain fusion processing on the independent load components based on the fast Fourier transform and order ratio analysis method to obtain the load time history data and frequency response characteristic data of the independent load components.

[0137] In this embodiment of the invention, when the time-domain-frequency domain fusion processing module performs time-domain-frequency domain fusion processing on the independent load components based on the Fast Fourier Transform and order ratio analysis method to obtain the load time history data and frequency response characteristic data of the independent load components, it is specifically used for:

[0138] The independent load components are subjected to spectral conversion processing to obtain the spectral data of the independent load components;

[0139] Multidimensional characteristic frequency analysis is performed on the spectrum data to obtain the characteristic frequency components of the spectrum data;

[0140] Based on the characteristic frequency components and the original time-domain signal, the independent load components are subjected to joint time-frequency domain reconstruction analysis to obtain the load time history data and frequency response characteristic data of the independent load components.

[0141] When the time-domain-frequency domain fusion processing module performs time-frequency domain joint reconstruction analysis on the independent load components based on the characteristic frequency components and the original time-domain signal to obtain the load time history data and frequency response characteristic data of the independent load components, it is specifically used for:

[0142] The frequency domain feature vectors of the characteristic frequency components are extracted by principal component extraction in the frequency domain to obtain the frequency domain feature vectors of the characteristic frequency components.

[0143] The original time-domain signal is sampled and aligned to obtain the time-domain vector of the original time-domain signal;

[0144] The frequency domain feature vector and the time domain vector are combined to generate a time-frequency domain vector set;

[0145] The time-frequency domain vector set is dynamically weighted to obtain the weighting coefficients of the time-frequency domain vector set;

[0146] Based on the synchronous compression transformation rule, the frequency domain feature vector and the time domain vector are linearly fused to obtain the transformation result of the independent load component. The calculation formula of the transformation result is as follows:

[0147] In the formula, The transformation result is... For the time-frequency domain vector set, the first... The weighting coefficients of each component, The first of the frequency domain feature vectors One portion, The first time domain vector One portion, For regularization parameters, For L2 norm operations, For element-wise multiplication, The frequency domain feature vector, For the time domain vector, The number of feature dimensions;

[0148] The transformation result is then reconstructed by inverse transformation to obtain the load time history data and frequency response characteristic data of the independent load components.

[0149] Specifically, when performing spectrum conversion processing on independent load components, the Fast Fourier Transform (FFT) method is used to convert the independent load components from time-varying time-domain signals to frequency-varying frequency-domain signals. Specifically, the values ​​of the independent load components at different time points are divided according to a certain time interval to obtain a series of discrete time-domain data points. Then, through a specific signal processing procedure, these time-domain data points are converted into corresponding frequency components and the amplitude of each frequency component, forming a spectrum representation with frequency as the horizontal axis and amplitude as the vertical axis, thus obtaining the spectrum data of the independent load components.

[0150] Furthermore, when performing multidimensional characteristic frequency analysis on the spectrum data, the order analysis method is combined to first determine the reference rotational speed information related to the operation of the object under test, calculate the frequency values ​​corresponding to different orders based on the reference rotational speed, and then analyze the amplitude and distribution of each frequency component in the spectrum data to identify those frequency points whose amplitude is significantly higher than the surrounding frequency components and corresponds to the order frequency values. At the same time, attention is paid to the frequency components that continuously appear or change significantly under different operating conditions. These representative frequency points and frequency components are extracted to obtain the characteristic frequency components of the spectrum data.

[0151] Furthermore, when performing time-frequency domain joint reconstruction analysis on independent load components based on characteristic frequency components and the original time-domain signal, the parts of the original time-domain signal of the independent load components related to the characteristic frequency components are first retained, and noise or interference components that do not contain characteristic frequency information are removed. Then, based on the amplitude and phase information of the characteristic frequency components, a signal that can reflect the changes of these characteristic frequencies is reconstructed in the time domain. At the same time, combined with the distribution characteristics of characteristic frequency components in the spectrum data, the response of independent load components at different frequencies is analyzed. Finally, load time history data of independent load components that contain load values ​​that change with time, and frequency response characteristic data of independent load components that contain load response characteristics at different frequencies are obtained.

[0152] Specifically, when extracting principal components of frequency domain eigenvectors from characteristic frequency components, the amplitude and distribution characteristics of each frequency point in the characteristic frequency components are first analyzed to identify the main components that best reflect the overall characteristics of these characteristic frequencies. These main components can reflect the core variation law of the characteristic frequency components. Then, these main components are arranged in order of their importance to form a set of vectors that can summarize the key information of the characteristic frequency components, thus obtaining the frequency domain eigenvectors of the characteristic frequency components.

[0153] Furthermore, when performing time-domain vector resampling and alignment on the original time-domain signal, the sampling time interval of the original time-domain signal is first determined. According to the time reference requirements of subsequent processing, the original time-domain signal is resampled to make the time distribution of the sampling points more uniform and aligned with the preset time nodes. Then, the signal values ​​of each time point obtained after resampling are arranged in chronological order to form a vector that can reflect the signal's change over time, thus obtaining the time-domain vector of the original time-domain signal.

[0154] Furthermore, when combining the frequency domain feature vector and the time domain vector, the frequency domain feature vector and the time domain vector are concatenated in a preset order so that the combined vector set contains both information reflecting frequency characteristics and information reflecting time changes. This ensures that the information of both types of vectors is completely preserved in the vector set without interfering with each other. Through this concatenation method, a time-frequency domain vector set is generated.

[0155] Furthermore, when dynamically weighting the time-frequency domain vector set, we first analyze the importance of the frequency domain feature vectors and time domain vectors in describing the characteristics of independent load components. Based on their importance, we assign corresponding weight coefficients to the frequency domain feature vectors and time domain vectors respectively. Vectors with higher importance are assigned larger weight coefficients, and vectors with lower importance are assigned smaller weight coefficients. These weight coefficients can dynamically reflect the contribution of different vectors in the overall analysis, thus obtaining the weighting coefficients of the time-frequency domain vector set.

[0156] Furthermore, based on the synchronous compression transformation rule, when performing linear hybrid fusion processing on the frequency domain feature vector and the time domain vector, the frequency domain feature vector and the time domain vector are linearly combined according to their respective weighting coefficients, so that the combined result retains both the key information about frequency in the frequency domain feature vector and the key information about time in the time domain vector. At the same time, information redundancy is reduced by compression transformation, highlighting the main features and obtaining the transformation result of the independent load components.

[0157] Furthermore, when performing inverse transformation reconstruction on the transformation result, the inverse transformation method corresponding to the synchronous compression transformation is adopted to restore the transformation result to the original signal form in the time domain and frequency domain. In the time domain, the load value sequence that varies with time is reconstructed to form the load time history data of independent load components. In the frequency domain, the load response characteristics at different frequencies are reconstructed to form the frequency response characteristic data of independent load components. Through such inverse transformation reconstruction, the load time history data and frequency response characteristic data of independent load components are obtained simultaneously.

[0158] Specifically, the source of the parameter transformation result in the formula is the result obtained by linearly mixing and fusing the frequency domain feature vector and the time domain vector based on the synchronous compression transformation rule. Specifically, it is obtained by operating the weighting coefficients of each component of the time-frequency domain vector set with the corresponding frequency domain feature vector component and time domain vector component, and then multiplying the regularization parameter with the relevant operation result.

[0159] Furthermore, in the formula, the time-frequency domain vector set... The weighting coefficients for the first component are derived from the dynamic weighting of the time-frequency domain vector set. These coefficients are based on the relative importance of the frequency-domain eigenvectors and time-domain vectors in describing the characteristics of independent load components. The weighting coefficients assigned to each component are determined by their importance; a higher importance corresponds to a larger coefficient, and a lower importance corresponds to a smaller coefficient.

[0160] Furthermore, the first eigenvector in the frequency domain of the formula... The source of the first component is the first element obtained when extracting the principal components of the frequency domain feature vector from the characteristic frequency components, arranged in order of importance. The main component reflects the core variation law of a certain aspect of the characteristic frequency component.

[0161] Furthermore, the first time-domain vector in the formula The source of the first component is the time-ordered first component in the time-domain vector obtained when the original time-domain signal is resampled and aligned using time-domain vector resampling. The signal value at each time point reflects the change of the original time-domain signal at the corresponding time point.

[0162] Furthermore, the regularization parameter in the formula is a fixed value set to reduce overfitting and ensure the stability of the transformation results during the fusion process. Its value is determined according to the actual processing requirements and data characteristics.

[0163] Furthermore, the source of the frequency domain feature vector in the formula is the vector obtained after extracting the principal components of the frequency domain feature vector from the feature frequency components. It is composed of multiple principal components that can summarize the key information of the feature frequency components, arranged in order of importance.

[0164] Furthermore, the source of the time-domain vector in the formula is the vector obtained after resampling and aligning the original time-domain signal. It is composed of the signal values ​​at each time point after resampling arranged in chronological order, reflecting the change law of the signal over time.

[0165] Furthermore, the source of the feature dimension in the formula is the number of components contained in the frequency domain feature vector and the time domain vector. This number is determined by the complexity of the feature frequency components and the sampling accuracy of the original time domain signal, and is the total number of dimensions describing the feature.

[0166] Furthermore, the significance of the formula lies in the fact that by operating and summing the weighting coefficients of each component of the time-frequency domain vector set with the corresponding frequency and time domain components, and then combining the regularization parameters with the relevant operation results, a linear hybrid fusion of the frequency domain feature vector and the time domain vector is achieved. This not only preserves the key information in the frequency and time domains, but also highlights important features and ensures the stability of the results through weighting and regularization. Finally, a transformation result that can comprehensively reflect the time-frequency domain characteristics of the independent load components is obtained, providing a foundation for subsequent inverse transformation reconstruction.

[0167] Furthermore, the formula trend is as follows: when Increase and corresponding When the value is fixed, It will increase accordingly; when or Increase and corresponding When fixed, It will also increase accordingly; when When the value increases and other parameters remain constant, due to its relationship with... Multiplying the results of operations will cause The contribution of this part increased, and the overall It will also increase; conversely, if or Decrease, assuming other parameters remain constant. The corresponding decrease will occur, and the overall performance will reflect the relationship between various parameters and... The positive correlation between them, at the same time The operation will cause this part to... The rate of increase in the impact is gradually slowing down to avoid imbalances caused by excessively large values.

[0168] In summary, the spectrum data obtained by spectrum conversion is transformed into frequency domain signals by fast Fourier transform, which clearly presents the load distribution at different frequencies. This solves the problem that time domain signals cannot intuitively reflect the load characteristics in the frequency dimension, and provides frequency domain data support for subsequent feature analysis.

[0169] In summary, multidimensional characteristic frequency analysis yields characteristic frequency components, and combined with order analysis, key frequency points related to blade operation are screened out, invalid interference frequencies are eliminated, solving the problem of difficulty in identifying effective information in spectrum data, and providing accurate characteristic basis for joint reconstruction.

[0170] In summary, the joint reconstruction of the time and frequency domains yields load time history and frequency response characteristic data, integrating frequency domain features with time domain signal patterns. This approach preserves the dynamic changes of the load over time while clarifying the response characteristics at different frequencies, thus solving the problem of incompleteness in single time and frequency domain analysis and providing multi-dimensional data for subsequent condition monitoring.

[0171] In summary, the principal component extraction of frequency domain feature vectors yields frequency domain feature vectors, extracts the core information of the feature frequency components, eliminates redundant frequency data, solves the problem of low efficiency in subsequent processing caused by messy frequency domain information, and provides concise core frequency domain data for vector combination.

[0172] In summary, time-domain vector resampling and alignment results in a time-domain vector that ensures a uniform time distribution of the original time-domain signal sampling points and alignment with a preset reference. This eliminates the impact of time deviation on signal analysis, provides time-synchronized time-domain data for vector combination, and guarantees the synergy of time-frequency data.

[0173] In summary, vector combination generates a time-frequency domain vector set, integrating frequency domain feature vectors and time domain vectors to achieve preliminary fusion of time-frequency dimension information, solving the problem that single time-frequency data cannot fully reflect load characteristics, and providing a complete time-frequency data foundation for dynamic weighting.

[0174] In summary, the dynamic weighting method assigns weights based on the importance of the time-frequency vector in describing the load characteristics, highlighting the role of key data, avoiding interference from irrelevant data, providing accurate weighting basis for linear hybrid fusion, and improving the reliability of the fusion results.

[0175] In summary, the linear hybrid fusion transformation result, combined with the synchronous compression transformation rules and calculation formulas, achieves deep fusion of time-frequency data. At the same time, the regularization parameter reduces overfitting, ensures the stability of the transformation result, and solves the problems of insufficient fusion of time-frequency data and easy distortion of results.

[0176] In summary, the inverse transformation reconstructs the load time history and frequency response characteristics data, restoring the transformation result to the original time-frequency signal form. This preserves the dynamic changes of the load over time and clarifies the response characteristics at different frequencies, providing comprehensive and accurate multi-dimensional load data for intelligent condition monitoring and early warning, and assisting in the accurate identification of blade load anomalies.

[0177] The intelligent state monitoring and early warning module 105 is used to perform real-time state signal analysis on the frequency response characteristic data based on the adaptive threshold comparison and modal parameter identification method, and obtain the state early warning signal of the frequency response characteristic data.

[0178] In this embodiment of the invention, when the intelligent state monitoring and early warning module performs real-time state signal analysis on the frequency response characteristic data based on an adaptive threshold comparison and modal parameter identification method to obtain a state early warning signal for the frequency response characteristic data, it is specifically used for:

[0179] Modal parameter identification is performed on the load time history data and frequency response characteristic data to obtain the modal parameters of the load time history data and frequency response characteristic data;

[0180] The modal parameters are dynamically and adaptively adjusted to obtain the threshold range of the modal parameters;

[0181] Based on the threshold range, the modal parameters are analyzed and compared in real time to obtain the abnormal state characteristics of the modal parameters;

[0182] Based on the severity level of the abnormal state characteristics, a state warning signal for the abnormal state characteristics is obtained.

[0183] Specifically, when performing modal parameter identification on load time history data and frequency response characteristic data, the load amplitude and trend changing with time in the load time history data are first extracted, and the response amplitude and phase characteristics corresponding to different frequencies in the frequency response characteristic data are extracted. Then, by combining the correlation between the two, parameters that can reflect the vibration modes of the structure are identified. These parameters include the natural frequency of the structure, the shape of the vibration mode, and the damping characteristics. In this way, the modal parameters of the load time history data and frequency response characteristic data are obtained.

[0184] Furthermore, when dynamically adjusting the modal parameters, historical data of the modal parameters under normal operating conditions are first collected, and the distribution range and variation patterns of these data are analyzed to determine the initial threshold range. Then, based on changes in the structural operating environment, the accumulation of operating time, and experience in handling historical anomalies, the initial threshold range is adjusted periodically. When the structural operating environment changes significantly, the threshold range is promptly narrowed or expanded to adapt to the new operating state. Through this dynamic adjustment, the threshold range of the modal parameters is obtained.

[0185] Furthermore, based on the threshold range, when performing real-time analysis and comparison of modal parameters, the modal parameters collected in real time are compared one by one with the determined threshold range to check whether each modal parameter is within the threshold range. If a modal parameter exceeds the threshold range, the value of the parameter, the extent of the exceedance, and the time of occurrence are recorded. At the same time, the correlation between this parameter and other modal parameters is analyzed. Through this real-time comparison and analysis, the abnormal state characteristics of the modal parameters are obtained.

[0186] Furthermore, when obtaining a status warning signal for an abnormal state feature based on the severity level of the abnormal state feature, a severity level classification standard for the abnormal state feature is first preset, such as minor abnormality, moderate abnormality, and severe abnormality. Among them, minor abnormality refers to modal parameters slightly exceeding the threshold range and having little impact on structural operation; moderate abnormality refers to modal parameters significantly exceeding the threshold range and potentially affecting some functions of the structure; and severe abnormality refers to modal parameters significantly exceeding the threshold range and seriously threatening structural safety.

[0187] Furthermore, based on the abnormal state characteristics obtained from real-time analysis, the severity level of the abnormal state is determined by comparing it with the classification criteria. Then, a signal containing the abnormal location, abnormal type, and handling suggestions is generated according to the corresponding level to obtain a status warning signal for the abnormal state characteristics.

[0188] In summary, modal parameter identification identifies modal parameters and extracts key parameters reflecting the vibration modes of the blade structure from load time history data and frequency response characteristic data. This solves the problem that it is difficult to obtain the essential characteristics of structural vibration by relying on a single data source, and provides a core basis for subsequent state analysis that can characterize the load state of the blade.

[0189] In summary, the threshold range of dynamic threshold adaptive adjustment, combined with historical data of modal parameters during normal blade operation and changes in the operating environment, dynamically optimizes the threshold, avoiding misjudgments or omissions caused by fixed thresholds failing to adapt to environmental fluctuations, and providing a judgment standard that fits the actual operating state for real-time comparison of modal parameters.

[0190] In summary, real-time analysis and comparison of abnormal state characteristics, comparison of real-time modal parameters with dynamic threshold ranges, and accurate capture of parameter changes exceeding the normal range solve the problem of difficulty in real-time identification of load anomalies by manual monitoring, providing clear evidence of anomalies for the generation of early warning signals.

[0191] In summary, the status warning signal is classified according to the severity level. The level is divided according to the impact of the abnormal characteristics and corresponding warnings are generated. This avoids the waste of resources caused by excessive warnings for minor anomalies and prevents the safety risks caused by untimely warnings for serious anomalies. It provides accurate decision support for handling blade load anomalies and ensures the safe operation of blades.

[0192] The integrated data visualization and monitoring module 106 is used to integrate the load time history data, the frequency response characteristic data and the status warning signal into the visualization monitoring platform to monitor the load status of the wind turbine blades in real time.

[0193] In this embodiment of the invention, when the integrated data visualization and monitoring module integrates the load time history data, the frequency response characteristic data, and the status warning signal into the visualization monitoring platform to monitor the load status of the wind turbine blades in real time, it is specifically used for:

[0194] The load time history data is reconstructed in the time domain to obtain the time domain waveform primitives of the load time history data;

[0195] The frequency response characteristic data is subjected to frequency domain spectrum construction to obtain the frequency domain spectrum of the frequency response characteristic data;

[0196] The state warning signal is state encoded to obtain the state encoding matrix of the state warning signal;

[0197] The time-domain waveform primitives, the frequency-domain spectrum, and the state coding matrix are used to construct a multi-dimensional spatiotemporal framework to generate a comprehensive display framework for the wind turbine blade.

[0198] Based on the visualization monitoring platform, the integrated display framework is monitored in real time to obtain the load status of the wind turbine blades.

[0199] Specifically, when reconstructing the time-domain waveform of the load time history data, the load value corresponding to each time node in the load time history data is first extracted. The time node is used as the horizontal axis coordinate and the corresponding load value is used as the vertical axis coordinate. The coordinate points corresponding to each time node and load value are marked one by one in the preset two-dimensional coordinate system. Then, all coordinate points are connected sequentially with smooth line segments in time order to form a curve that can intuitively reflect the trend of load change over time, thus obtaining the time-domain waveform primitive of the load time history data.

[0200] Furthermore, when constructing the frequency domain spectrum of the frequency response characteristic data, the response amplitude corresponding to each frequency value in the frequency response characteristic data is extracted, and the frequency value is used as the horizontal axis coordinate and the corresponding response amplitude is used as the vertical axis coordinate. The coordinate points corresponding to each frequency and response amplitude are marked in the two-dimensional coordinate system.

[0201] Furthermore, based on the distribution of coordinate points, a bar chart or curve is drawn, where the height of the bar chart or the vertical axis of the curve directly corresponds to the magnitude of the response amplitude. Through this graphical presentation, the response characteristics of the load at different frequencies are clearly displayed, and the frequency domain spectrum of the frequency response characteristic data is obtained.

[0202] Furthermore, when encoding the status warning signals, the different types of status warning signals are first determined, such as minor abnormality warning, moderate abnormality warning, severe abnormality warning and normal status indication. A unique encoding value is assigned to each type, where the normal status corresponds to a specific basic code, and different levels of abnormality warnings correspond to different distinguishing codes. Then, the real-time acquired status warning signals are matched with the corresponding encoding values ​​according to their types. These encoding values ​​are arranged into a matrix in chronological order or signal generation order. Each row or column of the matrix corresponds to the warning code of a time point or a monitoring location, thus obtaining the status encoding matrix of the status warning signals.

[0203] Furthermore, when constructing a multi-dimensional spatiotemporal framework for time-domain waveform primitives, frequency-domain spectrograms, and state coding matrices, three independent and related display areas are first divided in the display interface of the visualization monitoring platform, corresponding to the display of time-domain, frequency-domain, and state warning information, respectively. The time-domain waveform primitives are placed in the time-domain display area, and the frequency-domain spectrograms are placed in the frequency-domain display area.

[0204] Furthermore, the encoded values ​​in the status coding matrix are converted into color identifiers or icons and superimposed on the corresponding time-domain waveform primitives or frequency-domain spectrum, so that the abnormal warning location is directly associated with time and frequency information. At the same time, a time synchronization control is set to ensure that the graphic data of the three areas are consistent in the time dimension, forming an overall display structure that can simultaneously present time and frequency characteristics and warning status, and generating a comprehensive display framework for wind turbine blades.

[0205] Furthermore, based on the visualization monitoring platform, when monitoring the real-time status of the integrated display framework, the integrated display framework is loaded onto the display interface of the visualization monitoring platform. The platform receives and updates the data of time-domain waveform elements, frequency-domain spectrum, and status coding matrix in real time. When the time-domain waveform elements show curve changes that exceed the normal fluctuation range, the frequency-domain spectrum shows abnormal peaks, or the status coding matrix displays abnormal codes, the platform will automatically highlight the corresponding abnormal areas. At the same time, it will display the current time, frequency, and warning level information in real time. By observing the dynamic changes of the integrated display framework on the platform interface, the staff can intuitively grasp the load changes and abnormal warning status of the wind turbine blades at different times and frequencies, and obtain the load status of the wind turbine blades.

[0206] In summary, the time-domain waveform reconstruction yields time-domain waveform primitives, transforming load time history data into time-load curves, intuitively presenting time-domain fluctuations, and providing a foundation for understanding real-time dynamics.

[0207] In summary, the frequency domain spectrum constructed from the frequency domain map graphically displays the load response at different frequencies, solving the problem of difficult interpretation of frequency domain data and helping to identify abnormal frequency responses.

[0208] In summary, the state coding matrix obtained by state coding standardizes the format of early warning signals, solves the problem of messy early warning information, and provides unified information for multi-dimensional data fusion.

[0209] In summary, the multi-dimensional framework constructs a comprehensive display framework, integrates time and frequency data with early warnings, establishes correlations, and forms an integrated framework that comprehensively displays the load status.

[0210] In summary, the visualization platform monitors the blade load status in real time, dynamically updates data, highlights anomalies, solves the problem of inefficiency in manual analysis, achieves full-state monitoring, and provides a basis for decision-making.

[0211] Reference Figure 2 The diagram shown is a flowchart illustrating a method for detecting six-degree-of-freedom loads on a wind turbine blade according to an embodiment of the present invention. In this embodiment, the method for detecting six-degree-of-freedom loads on a wind turbine blade includes:

[0212] S1. Acquire multiple raw strain signals, and perform noise filtering on the multiple raw strain signals through a channel signal conditioning circuit to generate analog signals of the multiple raw strain signals;

[0213] S2. Perform synchronous analog-to-digital conversion on the analog signal to obtain digital strain data of the analog signal;

[0214] S3. Based on the phase synchronization compensation and strain amplitude decoupling method, the digital strain data is decoupled in real time to obtain the independent load components in the six degrees of freedom of the digital strain data;

[0215] S4. Based on the Fast Fourier Transform and order ratio analysis method, the independent load components are subjected to time-domain-frequency domain fusion processing to obtain the load time history data and frequency response characteristic data of the independent load components;

[0216] S5. Based on the adaptive threshold comparison and modal parameter identification method, perform real-time state signal analysis on the frequency response characteristic data to obtain the state warning signal of the frequency response characteristic data;

[0217] S6. Integrate the load time history data, the frequency response characteristic data, and the status warning signal into a visual monitoring platform to monitor the load status of the wind turbine blades in real time.

[0218] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0219] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0220] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A six-degree-of-freedom load detection system for wind turbine blades, characterized in that, The system includes a multi-channel conditioning module, a high-precision synchronous data conversion module, a real-time load decoupling module, a time-domain-frequency domain fusion processing module, an intelligent status monitoring and early warning module, and a comprehensive data visualization and integrated monitoring module, wherein: The multi-channel conditioning module is used to acquire multiple raw strain signals and perform noise filtering on the multiple raw strain signals through a channel signal conditioning circuit to generate analog signals of the multiple raw strain signals. The high-precision synchronous data conversion module is used to perform synchronous analog-to-digital conversion on the analog signal to obtain digital strain data of the analog signal; The real-time load decoupling module is used to decouple the digital strain data in real time based on the phase synchronization compensation and strain amplitude decoupling method, to obtain the independent load components in the six degrees of freedom of the digital strain data, including: The digital strain data is reconstructed in the spatial domain to obtain the spatial distribution strain data of the digital strain data. Phase synchronization compensation is performed on the spatially distributed strain data to obtain aligned strain data of the spatially distributed strain data. Based on the coupling relationship between strain and load, the amplitude decoupling processing is performed on the aligned strain data to separate the independent load components. Based on the projection relationship in six degrees of freedom, the mutually independent load components are linearly synthesized to obtain the independent load components in the six degrees of freedom in the digital strain data, including: Establish the projection mapping relationship between the six degrees of freedom directions and the mutually independent load components; A projection coefficient matrix is ​​generated based on the projection mapping relationship. The projection coefficient matrix contains the contribution weights of the load components in different degrees of freedom directions. The independent load components are linearly weighted and synthesized with the projection coefficient matrix to obtain a preliminary synthesized vector, wherein the calculation formula for the preliminary synthesized vector is as follows: ； In the formula, For the initial synthesis vector, The projection coefficient matrix is... This is the load component vector of the mutually independent load components; The preliminary synthesized vector is normalized to obtain the independent load components in the six degrees of freedom of the digital strain data; The time-domain-frequency domain fusion processing module is used to perform time-domain-frequency domain fusion processing on the independent load components based on the Fast Fourier Transform and order ratio analysis methods, to obtain the load time history data and frequency response characteristic data of the independent load components, including: The independent load components are subjected to spectral conversion processing to obtain the spectral data of the independent load components; Multidimensional characteristic frequency analysis is performed on the spectrum data to obtain the characteristic frequency components of the spectrum data; Based on the characteristic frequency components and the original time-domain signal, the independent load components are subjected to joint time-frequency domain reconstruction analysis to obtain the load time history data and frequency response characteristic data of the independent load components, including: The frequency domain feature vectors of the characteristic frequency components are extracted by principal component extraction in the frequency domain to obtain the frequency domain feature vectors of the characteristic frequency components. The original time-domain signal is sampled and aligned to obtain the time-domain vector of the original time-domain signal; The frequency domain feature vector and the time domain vector are combined to generate a time-frequency domain vector set; The time-frequency domain vector set is dynamically weighted to obtain the weighting coefficients of the time-frequency domain vector set; Based on the synchronous compression transformation rule, the frequency domain feature vector and the time domain vector are linearly fused to obtain the transformation result of the independent load component. The calculation formula of the transformation result is as follows: ； In the formula, The transformation result is... For the time-frequency domain vector set, the first... The weighting coefficients of each component, The first of the frequency domain feature vectors One portion, The first time domain vector One portion, For regularization parameters, For L2 norm operations, For element-wise multiplication, The frequency domain feature vector, For the time domain vector, The number of feature dimensions; The transformation result is reconstructed by inverse transformation to obtain the load time history data and frequency response characteristic data of the independent load components; The intelligent state monitoring and early warning module is used to perform real-time state signal analysis on the frequency response characteristic data based on an adaptive threshold comparison and modal parameter identification method, and obtain a state early warning signal of the frequency response characteristic data. The integrated data visualization and monitoring module is used to integrate the load time history data, the frequency response characteristic data and the status warning signal into the visualization monitoring platform to monitor the load status of the wind turbine blades in real time.

2. The six-degree-of-freedom load detection system for wind turbine blades as described in claim 1, characterized in that, When the multi-channel conditioning module acquires multiple raw strain signals and performs noise filtering on these signals through a channel signal conditioning circuit to generate analog signals from the raw strain signals, it is specifically used for: The multiple raw strain signals are standardized to obtain the standard strain signals of the multiple raw strain signals. The standard strain signal is subjected to high-frequency noise filtering to obtain the effective frequency band signal of the standard strain signal; The effective frequency band signal is subjected to signal simulation analysis to obtain the analog signal of the effective frequency band signal.

3. The six-degree-of-freedom load detection system for wind turbine blades as described in claim 1, characterized in that, When the high-precision synchronous data conversion module performs synchronous analog-to-digital conversion on the analog signal to obtain the digital strain data of the analog signal, it is specifically used for: The analog signal is time-division multiplexed and sampled to obtain multiple discrete signals of the analog signal; The multi-channel discrete signals are amplified by programmable gain control to obtain multi-channel amplified signals of the multi-channel discrete signals; The multi-channel amplified signals are quantized and sequence-encoded to obtain a preliminary digital signal sequence of the multi-channel amplified signals; The preliminary digital signal sequence is digitally filtered and compensated to obtain digital strain data of the preliminary digital signal sequence.

4. The six-degree-of-freedom load detection system for wind turbine blades as described in claim 1, characterized in that, When the intelligent state monitoring and early warning module performs real-time state signal analysis on the frequency response characteristic data based on an adaptive threshold comparison and modal parameter identification method to obtain a state early warning signal for the frequency response characteristic data, it is specifically used for: Modal parameter identification is performed on the load time history data and frequency response characteristic data to obtain the modal parameters of the load time history data and frequency response characteristic data; The modal parameters are dynamically and adaptively adjusted to obtain the threshold range of the modal parameters; Based on the threshold range, the modal parameters are analyzed and compared in real time to obtain the abnormal state characteristics of the modal parameters; Based on the severity level of the abnormal state characteristics, a state warning signal for the abnormal state characteristics is obtained.

5. The six-degree-of-freedom load detection system for wind turbine blades as described in claim 1, characterized in that, The integrated data visualization and monitoring module, when integrating the load time history data, the frequency response characteristic data, and the status warning signal into the visualization monitoring platform to monitor the load status of the wind turbine blades in real time, is specifically used for: The load time history data is reconstructed in the time domain to obtain the time domain waveform primitives of the load time history data; The frequency response characteristic data is subjected to frequency domain spectrum construction to obtain the frequency domain spectrum of the frequency response characteristic data; The state warning signal is state encoded to obtain the state encoding matrix of the state warning signal; The time-domain waveform primitives, the frequency-domain spectrum, and the state coding matrix are used to construct a multi-dimensional spatiotemporal framework to generate a comprehensive display framework for the wind turbine blade. Based on the visualization monitoring platform, the integrated display framework is monitored in real time to obtain the load status of the wind turbine blades.

6. A method for detecting six-degree-of-freedom loads on a wind turbine blade, used to implement the six-degree-of-freedom load detection system for a wind turbine blade as described in claim 1, the method comprising: S1. Acquire multiple raw strain signals, and perform noise filtering on the multiple raw strain signals through a channel signal conditioning circuit to generate analog signals of the multiple raw strain signals; S2. Perform synchronous analog-to-digital conversion on the analog signal to obtain digital strain data of the analog signal; S3. Based on the phase synchronization compensation and strain amplitude decoupling method, the digital strain data is decoupled in real time to obtain the independent load components in the six degrees of freedom of the digital strain data; S4. Based on the Fast Fourier Transform and order ratio analysis method, the independent load components are subjected to time-domain-frequency domain fusion processing to obtain the load time history data and frequency response characteristic data of the independent load components; S5. Based on the adaptive threshold comparison and modal parameter identification method, perform real-time state signal analysis on the frequency response characteristic data to obtain the state warning signal of the frequency response characteristic data; S6. Integrate the load time history data, the frequency response characteristic data, and the status warning signal into a visual monitoring platform to monitor the load status of the wind turbine blades in real time.

Images