Blade tip timing measurement system and method based on multi-sensor fusion

Abstract

The embodiment of the invention provides a blade tip timing measurement system and method based on multi-sensor fusion, belongs to the technical field of blade tip monitoring, and solves the problems that in an existing method, a compressed sensing model carries out blade tip timing analysis singly based on blade tip acceleration, blade working parameters collected by multiple sensors cannot be fused, and the blade tip timing measurement accuracy is high. And the complex state of the blade is difficult to comprehensively and accurately judge. The method comprises the steps that working parameter signals are filtered, the filtered working parameter signals are fused, a single-point fusion signal set is recognized and analyzed through a blade end monitoring model, and a blade end diagnosis result is output. According to the invention, the multi-sensor group is adopted to collect the working parameters of the blade, fusion processing is carried out to output the fusion signal set, various different types of information can be integrated, and various working parameter information of the blade can be more comprehensively obtained through multi-sensor fusion and generation of the fusion signal set, so that the running state of the blade is more accurately monitored.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of blade tip monitoring, and in particular to a blade tip timing measurement system and method based on multi-sensor fusion. Background Art

[0002] Steam turbines, gas turbines, aircraft engines, and other major equipment are critical dynamic devices. High-speed blades, as core components of these devices, play a crucial role in converting thermal or kinetic energy into mechanical energy during operation. Because these blades are required to operate for long periods in harsh environments such as high temperature, high pressure, and high speed, and are subjected to cyclical loads such as centrifugal force and aerodynamic forces, they are highly susceptible to high-cycle fatigue, resulting in microscopic damage. If not discovered and addressed promptly, these damages can gradually accumulate, eventually leading to serious failures such as blade corner loss, fragment loss, or even outright fracture, threatening the safe operation of the equipment and potentially causing major safety accidents. According to statistical analysis, one of the main causes of damage or failure of high-speed blades is vibration, particularly synchronous and asynchronous vibration.

[0003] Chinese patent CN116878652A discloses a method and system for identifying high-order vibration of blade tip timing based on blade tip acceleration. The method includes: determining the number and angle of circumferential installation of rotor blade casings by optimizing sensor positions based on the need for identifying high-frequency vibration parameters of blades; conducting experimental measurements to obtain the actual arrival time series of rotor blades passing through the sensors; grouping three sensors together, converting the blade tip timing analysis from blade tip vibration displacement to blade tip vibration acceleration using the arrival time and the installation angle of the sensors, and determining the blade vibration resonance area using the classic SDOF algorithm; establishing a compressed sensing model of undersampled blade tip vibration acceleration signals, and determining the positions of non-zero elements of the vibration sparse coefficient vector after solving it using a sparse algorithm. The design matrix of the vibration equation corresponding to the rotor blade high-order vibration frequency and blade high-order vibration acceleration can be determined without prior information, and the amplitude and phase of the blade high-order vibration can be determined using a circumferential Fourier algorithm.

[0004] However, the existing method establishes a compressed sensing model of undersampled blade tip vibration acceleration signals. The compressed sensing model performs blade tip timing analysis based solely on blade tip acceleration and is unable to fuse the blade operating parameters collected by multiple sensors, resulting in incomplete information and making it difficult to make a comprehensive and accurate judgment on the complex state of the blade. Summary of the Invention

[0005] The embodiments of the present invention aim to solve at least one of the technical problems existing in the prior art, and provide a blade tip timing measurement system and method based on multi-sensor fusion.

[0006] In a first aspect, an embodiment of the present invention provides a blade tip timing measurement method based on multi-sensor fusion, the method comprising:

[0007] The multi-sensor group collects the blade operating parameters in real time based on a preset sampling period, performs analog-to-digital conversion on the blade operating parameters, and obtains a digital operating parameter signal;

[0008] loading at least one set of working parameter signals, filtering the working parameter signals, fusing the filtered working parameter signals, and outputting a fused signal set;

[0009] A three-dimensional finite element model of the blade is pre-built based on the basic parameters of the rotor blade. The mapping relationship between the single-point parameters of the blade and the full-field parameters of the blade is established through the three-dimensional finite element model of the blade. Based on the mapping relationship, a single-point fusion signal set at the blade tip is obtained;

[0010] Pre-build a leaf-end monitoring model based on sparse reconstruction and motion compensation mechanism, iteratively train the leaf-end monitoring model using historical signal sets, and output a converged leaf-end monitoring model;

[0011] The leaf end monitoring model is executed with the single-point fusion signal set at the leaf end as input. The leaf end monitoring model identifies and analyzes the single-point fusion signal set and outputs the leaf end diagnosis result, where the leaf end diagnosis result includes the leaf end health and the leaf end fault type.

[0012] In some possible embodiments, the method for the multi-sensor group to collect blade operating parameters in real time based on a preset sampling period includes:

[0013] Determine the sensor weight of the sampling sensor based on the blade working parameter type collected by the multi-sensor group, and determine the sampling importance of the sampling sensor in combination with the sensor weight and parameter type;

[0014] The sampling importance is calculated by the following formula:

[0015]

[0016]

[0017] Among them, G cg Indicates the sampling importance, C,q cg Represent the number of parameter types and sensor weights, H w ,w z ,K d are respectively blade inertia time constant, single mass blade model, damping coefficient, T m ,T e ,w m Respectively represent mechanical torque, electromagnetic torque, and mechanical speed;

[0018] Load the sampling importance of the sampling sensor and determine whether the sampling importance exceeds the preset importance threshold;

[0019] If the preset critical threshold is exceeded, the preset sampling period of the corresponding sampling sensor is shortened by half;

[0020] If the preset important threshold is not exceeded, the corresponding sampling sensor maintains the preset sampling period;

[0021] The blade operating parameters are collected in real time based on the adjusted sampling period, and analog-to-digital conversion is performed on the blade operating parameters to obtain operating parameter signals in a digital form.

[0022] In some possible embodiments, the method for filtering the working parameter signal includes:

[0023] Loading the working parameter signal, constructing a signal interpolation function for the working parameter signal, obtaining a reconstruction formula of the working parameter signal in the frequency domain based on the signal interpolation function, reconstructing the working parameter signal according to the reconstruction formula, and obtaining a signal reconstruction set;

[0024] Obtaining a signal reconstruction set, performing discrete sampling of the signal reconstruction set at equal intervals with a fixed sampling frequency, and performing secondary reconstruction of the discrete sampling points using a cubic strip interpolation polynomial with a non-kink boundary condition to obtain a secondary reconstruction set;

[0025] The secondary reconstruction set is subjected to low-pass filtering based on a low-pass filter, a filtering threshold of the secondary reconstruction set is preset based on an interpolation FFT algorithm of a Hanning window, and frequency estimation is performed on the secondary reconstruction set after low-pass filtering based on the filtering threshold under the condition of the same signal-to-noise ratio, and a frequency estimation set is output;

[0026] Load the frequency estimation set, fuse the filtered working parameter signals, and output the fused signal set.

[0027] In some possible embodiments, the reconstruction formula is expressed as:

[0028]

[0029]

[0030] Among them, c(t) represents the output representation of the signal reconstruction set, B,f z are the bandwidth and center frequency of the working parameter signal respectively, S(t) represents the signal interpolation function, and α represents the sampling angle of the sampling sensor corresponding to the working parameter signal.

[0031] In some possible embodiments, the method of fusing the filtered working parameter signal includes:

[0032] Obtain the filtered frequency estimation set, perform dimensionality reduction processing on the frequency estimation set based on the frequency dimensionality reduction model, perform spatiotemporal synchronization and coordinate system alignment on the frequency estimation set after dimensionality reduction, and obtain a frequency estimation set aligned with heterogeneous data;

[0033] Load the frequency estimation set of heterogeneous data alignment, perform multi-scale decomposition on the frequency estimation set based on wavelet transform, obtain the high-frequency features corresponding to the frequency estimation set, perform feature-level fusion on the high-frequency features, and obtain the feature fusion set;

[0034] A sparse representation model of the feature fusion set is established based on the undersampling method. The blade tip fusion frequency is solved based on the sparse algorithm. The amplitude and phase of the blade tip fusion frequency are determined by combining the circumferential Fourier algorithm. The feature fusion set containing the blade tip fusion frequency, amplitude and phase is obtained.

[0035] In some possible embodiments, establishing a mapping relationship between a blade single-point parameter and a blade full-field parameter using a three-dimensional finite element model of the blade includes:

[0036] Obtain the basic parameters of the rotor blades, identify the blade length, width, height, thickness, chord length, and torsion angle parameters in the basic parameters of the rotor blades, and construct a three-dimensional model of the blade using three-dimensional modeling software;

[0037] Load the pre-built blade 3D model, import the blade 3D model into the finite element analysis software, mesh the blade 3D model, and assign the elastic modulus, Poisson's ratio, and density to the blade 3D model to obtain the blade 3D finite element model;

[0038] Modal analysis is performed on the three-dimensional finite element model of the blade, and the modal parameters of the three-dimensional finite element model of the blade are extracted. The three-dimensional finite element model of the blade is solved according to different working conditions to obtain the distribution state of the full-field displacement, stress, and strain parameters of the blade. The mapping relationship between the single-point parameters of the blade and the full-field parameters of the blade is established using a polynomial fitting algorithm.

[0039] In some possible embodiments, the method for the leaf-end monitoring model to identify and analyze a single-point fusion signal set includes:

[0040] Obtain a single-point fusion signal set, and the input layer of the leaf-end monitoring model encodes the single-point fusion signal to obtain a signal encoding set;

[0041] The LSTM architecture of the blade end monitoring model uses the Newmark algorithm to perform discrete solution on the signal coding set and calculate the transient dynamic balance value of the blade end equivalent stress;

[0042] The LSTM architecture loads the transient dynamic balance value of the blade tip equivalent stress and predicts the real-time dynamic stress of the blade tip based on sparse reconstruction and motion compensation mechanisms;

[0043] Load the real-time dynamic stress of the blade tip and set at least one set of cylindrical constraints at the blade shaft. Keep the time step consistent with the flow field calculation. Analyze the real-time dynamic stress results of the blade tip under rated operating conditions and low flow conditions, and output the blade tip health.

[0044] In some possible embodiments, the method for identifying and analyzing a single-point fusion signal set by the leaf-end monitoring model specifically further includes:

[0045] Determine whether the health of the leaf tip exceeds the preset health threshold;

[0046] If the leaf health is lower than the preset health threshold, the leaf health is used as an index to trigger a leaf graded alarm and output the leaf fault type.

[0047] Integrate the blade end health and blade end fault type into the blade end diagnosis result, and output the blade end diagnosis result.

[0048] In a second aspect, an embodiment of the present invention provides a blade tip timing measurement system based on multi-sensor fusion, which is used to implement the blade tip timing measurement method based on multi-sensor fusion as described above. The blade tip timing measurement system based on multi-sensor fusion includes:

[0049] The signal acquisition module collects the blade operating parameters in real time based on a preset sampling period through a multi-sensor group, performs analog-to-digital conversion on the blade operating parameters, and obtains a digital working parameter signal;

[0050] a signal fusion module, configured to load at least one set of operating parameter signals, filter the operating parameter signals, fuse the filtered operating parameter signals, and output a fused signal set;

[0051] The parameter mapping module pre-builds a three-dimensional finite element model of the blade based on the basic parameters of the rotor blade, establishes a mapping relationship between the single-point parameters of the blade and the full-field parameters of the blade through the three-dimensional finite element model of the blade, and obtains a single-point fusion signal set at the blade tip based on the mapping relationship;

[0052] The leaf-end diagnosis module is used to pre-build a leaf-end monitoring model based on sparse reconstruction and motion compensation mechanism, iteratively train the leaf-end monitoring model using historical signal sets, output a converged leaf-end monitoring model, and execute the leaf-end monitoring model with the single-point fusion signal set at the leaf end as input. The leaf-end monitoring model identifies and analyzes the single-point fusion signal set and outputs the leaf-end diagnosis results.

[0053] In some possible embodiments, the blade end diagnostic module includes:

[0054] A model building unit is used to pre-build a leaf-end monitoring model based on sparse reconstruction and motion compensation mechanism, iteratively train the leaf-end monitoring model using a historical signal set, and output a converged leaf-end monitoring model;

[0055] The leaf end analysis unit is used to execute the leaf end monitoring model with the single-point fusion signal set at the leaf end as input. The leaf end monitoring model identifies and analyzes the single-point fusion signal set and outputs the leaf end diagnosis result.

[0056] The blade tip timing measurement system and method based on multi-sensor fusion in the embodiments of the present invention utilizes a multi-sensor group to collect blade operating parameters, performs fusion processing, and outputs a fused signal set. This system integrates multiple types of information. Through multi-sensor fusion and the generation of fused signal sets, it more comprehensively acquires information on various blade operating parameters, thereby enabling more accurate monitoring of the blade's operating status and reducing the risk of misjudgment or missed judgments due to information loss. Furthermore, through multiple processing methods such as constructing a signal interpolation function, performing frequency domain reconstruction, performing quadratic reconstruction using cubic spline interpolation polynomials with non-kinked boundary conditions, and combining a low-pass filter with a Hanning window interpolation FFT algorithm, the system significantly improves signal processing accuracy and anti-interference capabilities, effectively addressing the shortcomings of existing signal processing technologies.

[0057] Furthermore, in the blade tip timing measurement system and method based on multi-sensor fusion of the embodiments of the present invention, when the multi-sensor group collects blade operating parameters in real time based on a preset sampling period, the sampling importance of the sampling sensor is scientifically and reasonably determined by combining the type of blade operating parameters collected by the multi-sensor group with the sensor weight, and the sensors and parameters that are more critical to blade monitoring can be accurately identified, so that the subsequent sampling strategy can focus more specifically on the collection of important information, thereby improving the efficiency and effectiveness of monitoring. By judging whether the sampling importance exceeds the preset importance threshold, the key sensors and relatively minor sensors can be effectively distinguished. Based on this judgment result, differentiated sampling strategies can be adopted for different sensors, thereby providing a key decision-making basis for optimizing the sampling period, improving monitoring efficiency and data quality.

[0058] Furthermore, the blade tip timing measurement system and method based on multi-sensor fusion in the embodiments of the present invention utilizes a low-pass filter to perform low-pass filtering on the secondary reconstruction set, effectively removing high-frequency noise interference from the signal, retaining the meaningful low-frequency components in the blade operating parameter signal, and improving the signal-to-noise ratio. Simultaneously, the Hanning window-based interpolation FFT algorithm presets the filtering threshold, enabling more accurate frequency estimation of the filtered signal under the same signal-to-noise ratio conditions. The filtered operating parameter signals are then fused and output as a fused signal set. This organically integrates different types of blade operating parameter signals, fully leveraging the advantages and complementarity of each parameter signal, ensuring that the fused signal set contains richer and more comprehensive blade status information, thereby improving information availability and diagnostic accuracy.

[0059] Furthermore, the blade tip timing measurement system and method based on multi-sensor fusion in the embodiments of the present invention establishes a sparse representation model of the feature fusion set based on the undersampling method and utilizes a sparse algorithm to solve the blade tip fusion frequency. This model can represent a complex feature fusion set with fewer non-zero coefficients, highlight key frequency components, remove redundant information, and achieve efficient expression and compression of features. At the same time, the amplitude and phase of the blade tip fusion frequency are determined in conjunction with the circumferential Fourier algorithm, further accurately determining the characteristic parameters of the blade tip fusion frequency, resulting in a complete feature fusion set containing the blade tip fusion frequency, amplitude, and phase, providing accurate and comprehensive feature information for subsequent blade health diagnosis and status assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0061] Figure 1 1 is a schematic diagram of the implementation flow of the blade tip timing measurement method based on multi-sensor fusion provided by an embodiment of the present invention;

[0062] Figure 2 A schematic diagram of the implementation process of a method for real-time acquisition of blade operating parameters by a multi-sensor group based on a preset sampling period is shown;

[0063] Figure 3 A schematic diagram of a method for implementing filtering of a working parameter signal is shown;

[0064] Figure 4 A schematic diagram of a method for implementing a fusion process of filtered working parameter signals is shown;

[0065] Figure 5 A schematic diagram of the process for establishing a mapping relationship between blade single-point parameters and blade full-field parameters using a three-dimensional finite element model of the blade is shown;

[0066] Figure 6 The schematic diagram of the implementation process of the leaf-end monitoring model for the method of identifying and analyzing a single-point fusion signal set is shown;

[0067] Figure 7 The schematic diagram of the structure of the blade tip timing measurement system based on multi-sensor fusion is shown. DETAILED DESCRIPTION

[0068] To enable those skilled in the art to better understand the technical solutions of the present invention, the present invention is further described below in conjunction with the accompanying drawings and specific embodiments. It is apparent that the described embodiments are only a portion of the embodiments of the present invention, rather than all of them. Based on the described embodiments of the present invention, all other embodiments obtained by those skilled in the art without requiring creative effort are within the scope of protection of the present invention.

[0069] Unless otherwise specified, the technical terms or scientific terms used in the embodiments of the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The terms "including" or "comprising" used in the embodiments of the present invention neither limit the shapes, numbers, steps, actions, operations, components, originals and / or their groups mentioned, nor exclude the appearance or addition of one or more other different shapes, numbers, steps, actions, operations, components, originals and / or their groups, or the addition of these. In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number and order of the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

[0070] Unless otherwise specifically stated, the relative arrangements of the components and steps, numerical expressions, and numerical values ​​set forth in these embodiments do not limit the scope of the present invention. At the same time, it should be understood that, for ease of description, the dimensions of the various parts shown in the drawings are not drawn in accordance with actual proportional relationships, and that the techniques, methods, and devices known to those of ordinary skill in the relevant art may not be discussed in detail, but where appropriate, the techniques, methods, and devices shown should be considered part of the authorized specification. In all examples shown and discussed herein, any specific other examples may have different values. It should be noted that similar symbols and letters represent similar items in the following figures, and therefore, once an item is defined in one figure, it does not need to be further discussed in subsequent figures.

[0071] In the description of the embodiments of the present invention, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In the embodiments of the present invention, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine different embodiments or examples described in the embodiments of the present invention and the features of different embodiments or examples, unless they are mutually inconsistent.

[0072] Below, the exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments of the present invention, and it should be understood that the present invention is not limited to the exemplary embodiments described herein.

[0073] The existing method establishes a compressed sensing model of undersampled blade tip vibration acceleration signals. The compressed sensing model performs blade tip timing analysis based solely on blade tip acceleration and is unable to fuse the blade operating parameters collected by multiple sensors, resulting in incomplete information and making it difficult to make a comprehensive and accurate judgment on the complex state of the blade.

[0074] In response to the above problems, an embodiment of the present invention proposes a blade tip timing measurement system and method based on multi-sensor fusion. In short, when the method is implemented, a multi-sensor group first collects blade working parameters in real time based on a preset sampling period, performs analog-to-digital conversion on the blade working parameters, obtains working parameter signals in digital form, filters the working parameter signals, and fuses the filtered working parameter signals to output a fusion signal set. Then, a three-dimensional finite element model of the blade is pre-constructed based on the basic parameters of the rotor blade. A mapping relationship between the single-point parameters of the blade and the full-field parameters of the blade is established through the three-dimensional finite element model of the blade. Based on the mapping relationship, a single-point fusion signal set at the blade tip is obtained. At the same time, a blade tip monitoring model based on sparse reconstruction and motion compensation mechanism is pre-constructed, and the blade tip monitoring model is iteratively trained using a historical signal set to output a converged blade tip monitoring model. Finally, the blade tip monitoring model identifies and analyzes the single-point fusion signal set and outputs a blade tip diagnosis result.

[0075] In an embodiment of the present invention, a multi-sensor group is used to collect blade operating parameters, and fusion processing is performed to output a fusion signal set. This can integrate a variety of different types of information. Through multi-sensor fusion and the generation of fusion signal sets, a variety of blade operating parameter information can be obtained more comprehensively, thereby more accurately monitoring the operating status of the blade and reducing the risk of misjudgment or missed judgment due to information loss. In addition, by constructing a signal interpolation function, performing frequency domain reconstruction, using a cubic spline interpolation polynomial with non-kinked boundary conditions for secondary reconstruction, and combining a low-pass filter with a Hanning window interpolation FFT algorithm and other processing methods, the accuracy and anti-interference ability of signal processing are greatly improved, effectively solving the shortcomings of the existing technology in signal processing.

[0076] The embodiment of the present invention provides a blade tip timing measurement method based on multi-sensor fusion, Figure 1 The following is a schematic diagram of the implementation process of the blade-end timing measurement method based on multi-sensor fusion. The blade-end timing measurement method based on multi-sensor fusion specifically includes:

[0077] In step S10, the multi-sensor group collects blade operating parameters in real time based on a preset sampling period, performs analog-to-digital conversion on the blade operating parameters, and obtains digital operating parameter signals. This multi-sensor group collects blade operating parameters in real time based on the preset sampling period, enabling timely acquisition of various real-time status information during blade operation, ensuring the timeliness of monitoring. Collecting multiple blade operating parameters provides a richer and more comprehensive data base than collecting a single parameter.

[0078] It should be noted that the multi-sensor group includes but is not limited to vibration sensors, strain sensors, temperature sensors, and pressure sensors, and the blade operating parameters include but are not limited to vibration parameters (vibration acceleration, vibration velocity, vibration displacement), strain parameters (axial strain, tangential strain, radial strain), temperature parameters, and pressure parameters.

[0079] Step S20 , loading at least one set of operating parameter signals, filtering the operating parameter signals, fusing the filtered operating parameter signals, and outputting a fused signal set.

[0080] Step S30: pre-build a three-dimensional finite element model of the blade based on the basic parameters of the rotor blade, establish a mapping relationship between the blade single-point parameters and the blade full-field parameters through the three-dimensional finite element model of the blade, and obtain a single-point fusion signal set at the blade tip based on the mapping relationship.

[0081] In an embodiment of the present invention, a three-dimensional finite element model of the blade is pre-constructed based on the basic parameters of the rotor blade, and a mapping relationship between the single-point parameters of the blade and the full-field parameters of the blade is established through the model. This can establish a connection between the single-point working parameters collected locally and the full-field state of the entire blade, realizing a comprehensive mapping from the local to the whole, and making up for the problem of insufficient grasp of the overall state of the blade that may be caused by only focusing on local parameters.

[0082] Step S40 , pre-constructing a leaf-end monitoring model based on sparse reconstruction and motion compensation mechanism, iteratively training the leaf-end monitoring model using a historical signal set, and outputting a converged leaf-end monitoring model.

[0083] In step S50 , the blade end monitoring model is executed with the single-point fusion signal set at the blade end as input. The blade end monitoring model identifies and analyzes the single-point fusion signal set and outputs a blade end diagnosis result, wherein the blade end diagnosis result includes the blade end health and the blade end fault type.

[0084] In an embodiment of the present invention, a multi-sensor group is used to collect blade operating parameters, and fusion processing is performed to output a fusion signal set. This can integrate a variety of different types of information. Through multi-sensor fusion and the generation of fusion signal sets, a variety of blade operating parameter information can be obtained more comprehensively, thereby more accurately monitoring the operating status of the blade and reducing the risk of misjudgment or missed judgment due to missing information.

[0085] The embodiment of the present invention provides a method for real-time acquisition of blade operating parameters by a multi-sensor group based on a preset sampling period. Figure 2 The present invention shows a flow chart of a method for collecting blade operating parameters in real time based on a preset sampling period by a multi-sensor group. The method for collecting blade operating parameters in real time based on a preset sampling period by a multi-sensor group specifically includes:

[0086] In step S101, the sensor weight of the sampling sensor is determined based on the type of blade working parameters collected by the multi-sensor group, and the sampling importance of the sampling sensor is determined in combination with the sensor weight and parameter type; among them, highly sensitive parameters (such as electromagnetic torque mutation and speed fluctuation) can be given higher weights to dynamically increase their sampling frequency, while low-frequency sampling is maintained for non-critical parameters (such as ambient temperature) to reduce data redundancy.

[0087] The sampling importance is calculated by the following formula:

[0088]

[0089] Among them, G cg Indicates the sampling importance, C,q cg Represent the number of parameter types and sensor weights, H w ,w z ,Kd are respectively blade inertia time constant, single mass blade model, damping coefficient, T m ,T e ,w m They represent mechanical torque, electromagnetic torque, and mechanical speed respectively.

[0090] Step S102: Load the sampling importance of the sampling sensor.

[0091] Step S103: determine whether the importance exceeds a preset importance threshold, which may be 0.4-0.6.

[0092] Step S104: if the value exceeds the preset critical threshold, the preset sampling period of the corresponding sampling sensor is shortened by half.

[0093] Step S105: If the preset important threshold is not exceeded, the corresponding sampling sensor maintains the preset sampling period.

[0094] In step S106, the blade working parameters are collected in real time based on the adjusted sampling period, and the blade working parameters are converted into digital form to obtain working parameter signals in digital form. The collected blade working parameters are converted into digital form to obtain working parameter signals in digital form, which facilitates the subsequent use of digital signal processing technology for efficient and accurate analysis and processing, ensures the data quality and availability of the entire monitoring process, and lays a good foundation for subsequent signal processing, fusion and diagnosis steps.

[0095] In an embodiment of the present invention, when the multi-sensor group collects blade operating parameters in real time based on a preset sampling period, the sampling importance of the sampling sensor is scientifically and reasonably determined by combining the type of blade operating parameters collected by the multi-sensor group with the sensor weight, so that the sensors and parameters that are more critical to blade monitoring can be accurately identified, so that the subsequent sampling strategy can focus more specifically on the collection of important information, thereby improving the efficiency and effectiveness of monitoring. By judging whether the sampling importance exceeds the preset importance threshold, the key sensors and relatively minor sensors can be effectively distinguished. Based on this judgment result, differentiated sampling strategies can be adopted for different sensors, thereby providing a key decision-making basis for optimizing the sampling period, improving monitoring efficiency and data quality.

[0096] The embodiment of the present invention provides a method for filtering a working parameter signal. Figure 3 The following is a flow chart showing a method for filtering a working parameter signal. The method for filtering a working parameter signal specifically includes:

[0097] Step S201: Load the working parameter signal, construct a signal interpolation function for the working parameter signal, obtain a reconstruction formula of the working parameter signal in the frequency domain based on the signal interpolation function, reconstruct the working parameter signal according to the reconstruction formula, and obtain a signal reconstruction set. Interpolation and reconstruction of the high-frequency vibration signal can compensate for the frequency band loss problem caused by insufficient sampling rate of the fixed sensor group and improve the frequency domain coverage. By constructing the signal interpolation function and obtaining the reconstruction formula of the working parameter signal in the frequency domain based on it, the original signal can be described and reconstructed more comprehensively and accurately, making the signal more complete in the frequency domain.

[0098] In this embodiment, the reconstruction formula is expressed as:

[0099]

[0100] Among them, c(t) represents the output representation of the signal reconstruction set, B,f z are the bandwidth and center frequency of the working parameter signal respectively, S(t) represents the signal interpolation function, and α represents the sampling angle of the sampling sensor corresponding to the working parameter signal.

[0101] Step S202: Obtain a signal reconstruction set, perform equally spaced discrete sampling on the signal reconstruction set at a fixed sampling frequency, and perform secondary reconstruction on the discrete sampling points using a cubic strip-spline interpolation polynomial with a non-kink boundary condition to obtain a secondary reconstruction set. In this embodiment, performing equally spaced discrete sampling on the signal reconstruction set at a fixed sampling frequency ensures uniformity and regularity of the sampling, facilitating subsequent stable signal analysis. Secondary reconstruction of the discrete sampling points using a cubic strip-spline interpolation polynomial with a non-kink boundary condition effectively avoids abnormal fluctuations or distortion at signal boundaries, making the entire signal smoother and more continuous, further improving signal quality and reliability.

[0102] Step S203, low-pass filtering is performed on the secondary reconstruction set based on a low-pass filter, and a filtering threshold of the secondary reconstruction set is preset based on the interpolation FFT algorithm of the Hanning window. Under the condition of the same signal-to-noise ratio, frequency estimation is performed on the secondary reconstruction set after low-pass filtering based on the filtering threshold, and a frequency estimation set is output.

[0103] It should be noted that low-pass filtering the secondary reconstruction set effectively removes high-frequency noise interference from the signal, retains the meaningful low-frequency components of the blade operating parameter signal, and improves the signal-to-noise ratio. Furthermore, the Hanning window-based interpolation FFT algorithm, with a preset filter threshold, enables more accurate frequency estimation of the filtered signal under the same signal-to-noise ratio.

[0104] Step S204: load the frequency estimation set, fuse the filtered working parameter signals, and output a fused signal set.

[0105] In an embodiment of the present invention, a low-pass filter is used to perform low-pass filtering on the secondary reconstruction set, which can effectively remove high-frequency noise interference in the signal, retain the meaningful low-frequency components in the blade operating parameter signal, and improve the signal-to-noise ratio. At the same time, the interpolation FFT algorithm based on the Hanning window presets the filtering threshold, which can more accurately estimate the frequency of the filtered signal under the same signal-to-noise ratio conditions. The filtered operating parameter signals are fused and output as a fused signal set. This can organically integrate different types of blade operating parameter signals, fully leveraging the advantages and complementarity of each parameter signal, so that the fused signal set contains richer and more comprehensive blade status information, improving the availability of information and the accuracy of diagnosis.

[0106] The embodiment of the present invention provides a method for fusing filtered working parameter signals. Figure 4 A schematic flow chart of a method for fusing filtered working parameter signals is shown. The method for fusing filtered working parameter signals specifically includes:

[0107] Step S301: obtain a filtered frequency estimation set, perform dimensionality reduction processing on the frequency estimation set based on a frequency dimensionality reduction model, perform spatiotemporal synchronization and coordinate system 1 on the frequency estimation set after dimensionality reduction processing, and obtain a frequency estimation set aligned with heterogeneous data.

[0108] In an embodiment of the present invention, the frequency estimation set is reduced in dimension based on the frequency dimensionality reduction model, which can effectively reduce the dimension and scale of the data. While retaining key information, it reduces the complexity and redundancy of the data and reduces the amount of data for subsequent processing. The frequency dimensionality reduction model can be a principal component analysis PCA model to perform spatiotemporal synchronization and coordinate system on the frequency estimation set after dimensionality reduction processing to solve the spatiotemporal misalignment problem caused by differences in sensor installation positions.

[0109] Step S302 loads the frequency estimate set for heterogeneous data alignment, performs multi-scale decomposition on the frequency estimate set based on wavelet transform, obtains the high-frequency features corresponding to the frequency estimate set, and performs feature-level fusion on the high-frequency features to obtain a fusion feature set. Multi-scale decomposition of the frequency estimate set based on wavelet transform decomposes the signal into components at different frequency scales, obtaining low-frequency approximations and high-frequency details, capturing the characteristic information of the signal at different scales. Feature-level fusion of the high-frequency features organically combines the detailed features of different sensors in the high-frequency band, fully leveraging the advantages of each sensor in different frequency ranges to obtain a more comprehensive and rich fusion feature set.

[0110] Step S303: establish a sparse representation model of the feature fusion set based on the undersampling method, solve the blade tip fusion frequency based on the sparse algorithm, and determine the amplitude and phase of the blade tip fusion frequency in combination with the circumferential Fourier algorithm to obtain a feature fusion set containing the blade tip fusion frequency, amplitude and phase.

[0111] In this embodiment of the present invention, a sparse representation model for the feature fusion set is established based on an undersampling method, and a sparse algorithm is used to solve the blade tip fusion frequency. This allows for the representation of complex feature fusion sets with fewer non-zero coefficients, highlighting key frequency components, removing redundant information, and achieving efficient feature expression and compression. Simultaneously, the amplitude and phase of the blade tip fusion frequency are determined using a circumferential Fourier algorithm, further accurately determining the characteristic parameters of the blade tip fusion frequency. This results in a complete feature fusion set containing the blade tip fusion frequency, amplitude, and phase, providing accurate and comprehensive feature information for subsequent blade health diagnosis and status assessment.

[0112] The embodiment of the present invention provides a method for establishing a mapping relationship between blade single-point parameters and blade full-field parameters through a blade three-dimensional finite element model. Figure 5 The figure shows a flow chart of a method for establishing a mapping relationship between a blade single point parameter and a blade full field parameter using a three-dimensional finite element model of the blade. The method for establishing a mapping relationship between a blade single point parameter and a blade full field parameter using a three-dimensional finite element model of the blade specifically includes:

[0113] Step S401, obtain the basic parameters of the rotor blade, identify the blade length, width, height, thickness, chord length, and torsion angle parameters among the basic parameters of the rotor blade, and use 3D modeling software to construct a 3D model of the blade; using 3D modeling software to construct a 3D model of the blade can intuitively display the geometric shape and structural characteristics of the blade, provide a precise geometric basis for subsequent finite element analysis and parameter mapping relationship establishment, and ensure the accuracy and reliability of the entire analysis process.

[0114] Step S402 , load the pre-built blade three-dimensional model, import the blade three-dimensional model into the finite element analysis software, mesh the blade three-dimensional model, and assign the elastic modulus, Poisson's ratio, and density to the blade three-dimensional model to obtain the blade three-dimensional finite element model.

[0115] It should be noted that the three-dimensional finite element model of the blade supports the simulation of the interface characteristics between the composite material and the metal matrix, and can diagnose complex faults such as delamination damage.

[0116] In step S403, a modal analysis is performed on the three-dimensional finite element model of the blade to extract the modal parameters of the three-dimensional finite element model of the blade. The three-dimensional finite element model of the blade is solved according to different working conditions to obtain the distribution state of the full-field displacement, stress, and strain parameters of the blade. A polynomial fitting algorithm is used to establish a mapping relationship between the single-point parameters of the blade and the full-field parameters of the blade.

[0117] In an embodiment of the present invention, modal analysis is performed on the three-dimensional finite element model of the blade to extract the modal parameters of the blade, such as the natural frequency and vibration mode. These modal parameters are crucial for understanding the vibration characteristics and dynamic behavior of the blade. The three-dimensional finite element model of the blade is solved according to different working conditions to obtain the distribution state of the full-field displacement, stress, and strain parameters of the blade, which comprehensively reflects the mechanical response and state information of the blade under various actual working conditions. A polynomial fitting algorithm is used to establish a mapping relationship between the single-point parameters of the blade and the full-field parameters of the blade, which can organically link the local single-point parameters with the overall full-field parameters, realizing information mapping and conversion from local to overall.

[0118] The embodiment of the present invention provides a method for identifying and analyzing a single-point fusion signal set using a leaf-end monitoring model. Figure 6 The following is a flow chart showing a method for identifying and analyzing a single-point fusion signal set by a leaf-end monitoring model. The method for identifying and analyzing a single-point fusion signal set by a leaf-end monitoring model specifically includes:

[0119] In step S501, a single-point fusion signal set is obtained. The input layer of the leaf-end monitoring model encodes the single-point fusion signal to obtain a signal code set. The input layer of the leaf-end monitoring model encodes the single-point fusion signal set, converting complex signal data into a more compact and representative signal code set. This not only reduces the data volume and improves the efficiency of subsequent processing, but also extracts key feature information from the signal.

[0120] In step S502, the LSTM architecture of the blade tip monitoring model uses the Newmark algorithm to perform a discrete solution on the signal coding set and calculate the transient dynamic balance value of the blade tip equivalent stress. This LSTM architecture uses the Newmark algorithm to perform a discrete solution on the signal coding set, accurately calculating the transient dynamic balance value of the blade tip equivalent stress. The Newmark algorithm offers high accuracy and stability when processing dynamic problems. Combined with the LSTM architecture's ability to capture time series information, it can better reflect the temporal variation of blade tip stress.

[0121] In step S503, the LSTM architecture loads the transient dynamic balance values ​​of the blade tip equivalent stresses and predicts the blade tip's real-time dynamic stresses based on sparse reconstruction and motion compensation. After loading the transient dynamic balance values ​​of the blade tip equivalent stresses, the LSTM architecture predicts the blade tip's real-time dynamic stresses based on sparse reconstruction and motion compensation. Sparse reconstruction highlights key features in stress data and reduces data redundancy, while motion compensation corrects for stress variation deviations caused by blade motion, resulting in a more accurate prediction of the blade tip's real-time dynamic stresses during actual operation.

[0122] Step S504: Load the blade tip's real-time dynamic stresses. Set at least one set of cylindrical constraints on the blade shaft, using a time step consistent with the flow field calculation. Analyze the blade tip's real-time dynamic stress results for both rated and low-flow conditions, and output the blade tip's health. Setting at least one set of cylindrical constraints on the blade shaft, consistent with the flow field calculation time step, and analyzing the blade tip's real-time dynamic stress results for both rated and low-flow conditions allows for a comprehensive assessment of the blade tip's stress state and health under different operating conditions.

[0123] Step S505 , determining whether the health of the blade end exceeds a preset health threshold.

[0124] Step S506 : If the health of the leaf end is less than a preset health threshold, a leaf end classification alarm is triggered with the leaf end health as an index, and the leaf end fault type is output.

[0125] In step S507, if the blade health is greater than a preset health threshold (in this embodiment, the health threshold may be 0.8-0.95), the blade health is saved. This allows for the accumulation of a large amount of blade health data, facilitating subsequent analysis and tracing of blade health trends. This historical data allows for a better understanding of the long-term performance changes of the blades, supporting optimized maintenance strategies and predictive maintenance.

[0126] Step S508 : Integrate the blade health and the blade fault type into a blade diagnosis result, and output the blade diagnosis result.

[0127] The embodiment of the present invention provides a blade tip timing measurement system based on multi-sensor fusion, Figure 7 The schematic diagram of the structure of the blade tip timing measurement system based on multi-sensor fusion is shown. The blade tip timing measurement system based on multi-sensor fusion specifically includes:

[0128] The signal acquisition module 100 collects blade operating parameters in real time based on a preset sampling period through a multi-sensor group, performs analog-to-digital conversion on the blade operating parameters, and obtains operating parameter signals in digital form.

[0129] The signal fusion module 200 is used to load at least one set of working parameter signals, filter the working parameter signals, fuse the filtered working parameter signals, and output a fused signal set.

[0130] The parameter mapping module 300 pre-builds a three-dimensional finite element model of the blade based on the basic parameters of the rotor blade, establishes a mapping relationship between the single-point parameters of the blade and the full-field parameters of the blade through the three-dimensional finite element model of the blade, and obtains a single-point fusion signal set at the blade tip based on the mapping relationship.

[0131] The leaf-end diagnosis module 400 is used to pre-build a leaf-end monitoring model based on sparse reconstruction and motion compensation mechanism, iteratively train the leaf-end monitoring model using a historical signal set, output a converged leaf-end monitoring model, and execute the leaf-end monitoring model with a single-point fusion signal set at the leaf end as input. The leaf-end monitoring model identifies and analyzes the single-point fusion signal set and outputs a leaf-end diagnosis result.

[0132] In this embodiment, the blade tip diagnosis module 400 includes:

[0133] The model construction unit 410 is used to pre-construct a leaf-end monitoring model based on sparse reconstruction and motion compensation mechanism, iteratively train the leaf-end monitoring model using a historical signal set, and output a converged leaf-end monitoring model.

[0134] The leaf end analysis unit 420 is used to execute the leaf end monitoring model with the single point fusion signal set at the leaf end as input. The leaf end monitoring model identifies and analyzes the single point fusion signal set and outputs the leaf end diagnosis result.

[0135] In summary, the present invention provides a blade tip timing measurement system and method based on multi-sensor fusion. In an embodiment of the present invention, a multi-sensor group is used to collect blade operating parameters, and fusion processing is performed to output a fusion signal set. It is possible to integrate a variety of different types of information. Through multi-sensor fusion and the generation of fusion signal sets, a variety of blade operating parameter information can be obtained more comprehensively, thereby more accurately monitoring the operating status of the blade and reducing the risk of misjudgment or missed judgment due to missing information.

[0136] In an embodiment of the present invention, a low-pass filter is used to perform low-pass filtering on the secondary reconstruction set, which can effectively remove high-frequency noise interference in the signal, retain the meaningful low-frequency components in the blade operating parameter signal, and improve the signal-to-noise ratio. At the same time, the interpolation FFT algorithm based on the Hanning window presets the filtering threshold, which can more accurately estimate the frequency of the filtered signal under the same signal-to-noise ratio conditions. The filtered operating parameter signals are fused and output as a fused signal set. This can organically integrate different types of blade operating parameter signals, fully leveraging the advantages and complementarity of each parameter signal, so that the fused signal set contains richer and more comprehensive blade status information, improving the availability of information and the accuracy of diagnosis.

[0137] In this embodiment of the present invention, a sparse representation model for the feature fusion set is established based on an undersampling method, and a sparse algorithm is used to solve the blade tip fusion frequency. This allows for the representation of complex feature fusion sets with fewer non-zero coefficients, highlighting key frequency components, removing redundant information, and achieving efficient feature expression and compression. Simultaneously, the amplitude and phase of the blade tip fusion frequency are determined using a circumferential Fourier algorithm, further accurately determining the characteristic parameters of the blade tip fusion frequency. This results in a complete feature fusion set containing the blade tip fusion frequency, amplitude, and phase, providing accurate and comprehensive feature information for subsequent blade health diagnosis and status assessment.

[0138] It will be understood that the above embodiments are merely exemplary embodiments for illustrating the principles of the present invention, and the present invention is not limited thereto. Those skilled in the art will appreciate that various modifications and improvements can be made without departing from the spirit and substance of the present invention, and such modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A blade tip timing measurement method based on multi-sensor fusion, characterized in that: The method comprises: The multi-sensor group collects the blade operating parameters in real time based on a preset sampling period, performs analog-to-digital conversion on the blade operating parameters, and obtains a digital operating parameter signal; loading at least one set of working parameter signals, filtering the working parameter signals, fusing the filtered working parameter signals, and outputting a fused signal set; A three-dimensional finite element model of the blade is pre-built based on the basic parameters of the rotor blade. The mapping relationship between the single-point parameters of the blade and the full-field parameters of the blade is established through the three-dimensional finite element model of the blade. Based on the mapping relationship, a single-point fusion signal set at the blade tip is obtained; Pre-build a leaf-end monitoring model based on sparse reconstruction and motion compensation mechanism, iteratively train the leaf-end monitoring model using historical signal sets, and output a converged leaf-end monitoring model; The leaf end monitoring model is executed with the single-point fusion signal set at the leaf end as input. The leaf end monitoring model identifies and analyzes the single-point fusion signal set and outputs the leaf end diagnosis result, where the leaf end diagnosis result includes the leaf end health and the leaf end fault type.

2. The blade tip timing measurement method based on multi-sensor fusion according to claim 1 is characterized in that: The method for collecting blade operating parameters in real time based on a preset sampling period by a multi-sensor group includes: Determine the sensor weight of the sampling sensor based on the blade working parameter type collected by the multi-sensor group, and determine the sampling importance of the sampling sensor in combination with the sensor weight and parameter type; The sampling importance is calculated by the following formula: Among them, G cg Indicates the sampling importance, C,q cg Represent the number of parameter types and sensor weights, H w ,w z ,K d are respectively blade inertia time constant, single mass blade model, damping coefficient, T m ,T e ,w m Respectively represent mechanical torque, electromagnetic torque, and mechanical speed; Load the sampling importance of the sampling sensor and determine whether the sampling importance exceeds the preset importance threshold; If the preset critical threshold is exceeded, the preset sampling period of the corresponding sampling sensor is shortened by half; If the preset important threshold is not exceeded, the corresponding sampling sensor maintains the preset sampling period; The blade operating parameters are collected in real time based on the adjusted sampling period, and analog-to-digital conversion is performed on the blade operating parameters to obtain operating parameter signals in a digital form.

3. The blade tip timing measurement method based on multi-sensor fusion according to claim 1, characterized in that: The method for filtering the working parameter signal includes: Loading the working parameter signal, constructing a signal interpolation function for the working parameter signal, obtaining a reconstruction formula of the working parameter signal in the frequency domain based on the signal interpolation function, reconstructing the working parameter signal according to the reconstruction formula, and obtaining a signal reconstruction set; Obtaining a signal reconstruction set, performing discrete sampling of the signal reconstruction set at equal intervals with a fixed sampling frequency, and performing secondary reconstruction of the discrete sampling points using a cubic strip interpolation polynomial with a non-kink boundary condition to obtain a secondary reconstruction set; The secondary reconstruction set is subjected to low-pass filtering based on a low-pass filter, a filtering threshold of the secondary reconstruction set is preset based on an interpolation FFT algorithm of a Hanning window, and frequency estimation is performed on the secondary reconstruction set after low-pass filtering based on the filtering threshold under the condition of the same signal-to-noise ratio, and a frequency estimation set is output; Load the frequency estimation set, fuse the filtered working parameter signals, and output the fused signal set.

4. The blade tip timing measurement method based on multi-sensor fusion according to claim 3 is characterized in that: The reconstruction formula is expressed as: Among them, c(t) represents the output representation of the signal reconstruction set, B,f z are the bandwidth and center frequency of the working parameter signal respectively, S(t) represents the signal interpolation function, and α represents the sampling angle of the sampling sensor corresponding to the working parameter signal.

5. The blade tip timing measurement method based on multi-sensor fusion according to claim 3, characterized in that: The method for fusing the filtered working parameter signal comprises: Obtain the filtered frequency estimation set, perform dimensionality reduction processing on the frequency estimation set based on the frequency dimensionality reduction model, perform spatiotemporal synchronization and coordinate system alignment on the frequency estimation set after dimensionality reduction, and obtain a frequency estimation set aligned with heterogeneous data; Load the frequency estimation set of heterogeneous data alignment, perform multi-scale decomposition on the frequency estimation set based on wavelet transform, obtain the high-frequency features corresponding to the frequency estimation set, perform feature-level fusion on the high-frequency features, and obtain the feature fusion set; A sparse representation model of the feature fusion set is established based on the undersampling method. The blade tip fusion frequency is solved based on the sparse algorithm. The amplitude and phase of the blade tip fusion frequency are determined by combining the circumferential Fourier algorithm. The feature fusion set containing the blade tip fusion frequency, amplitude and phase is obtained.

6. The blade tip timing measurement method based on multi-sensor fusion according to any one of claims 1 to 5, characterized in that: The mapping relationship between the single-point parameters of the blade and the full-field parameters of the blade is established by using the three-dimensional finite element model of the blade, including: Obtain the basic parameters of the rotor blades, identify the blade length, width, height, thickness, chord length, and torsion angle parameters in the basic parameters of the rotor blades, and construct a three-dimensional model of the blade using three-dimensional modeling software; Load the pre-built blade 3D model, import the blade 3D model into the finite element analysis software, mesh the blade 3D model, and assign the elastic modulus, Poisson's ratio, and density to the blade 3D model to obtain the blade 3D finite element model; Modal analysis is performed on the three-dimensional finite element model of the blade, and the modal parameters of the three-dimensional finite element model of the blade are extracted. The three-dimensional finite element model of the blade is solved according to different working conditions to obtain the distribution state of the full-field displacement, stress, and strain parameters of the blade. The mapping relationship between the single-point parameters of the blade and the full-field parameters of the blade is established using a polynomial fitting algorithm.

7. The blade tip timing measurement method based on multi-sensor fusion according to claim 6, characterized in that: The method for identifying and analyzing a single-point fusion signal set by the leaf-end monitoring model includes: Obtain a single-point fusion signal set, and the input layer of the leaf-end monitoring model encodes the single-point fusion signal to obtain a signal encoding set; The LSTM architecture of the blade end monitoring model uses the Newmark algorithm to perform discrete solution on the signal coding set and calculate the transient dynamic balance value of the blade end equivalent stress; The LSTM architecture loads the transient dynamic balance value of the blade tip equivalent stress and predicts the real-time dynamic stress of the blade tip based on sparse reconstruction and motion compensation mechanisms; Load the real-time dynamic stress of the blade tip and set at least one set of cylindrical constraints at the blade shaft. Keep the time step consistent with the flow field calculation. Analyze the real-time dynamic stress results of the blade tip under rated operating conditions and low flow conditions, and output the blade tip health.

8. The blade tip timing measurement method based on multi-sensor fusion according to claim 7, characterized in that: The method for identifying and analyzing a single-point fusion signal set by the leaf-end monitoring model specifically includes: Determine whether the health of the leaf tip exceeds the preset health threshold; If the leaf health is lower than the preset health threshold, the leaf health is used as an index to trigger a leaf graded alarm and output the leaf fault type. Integrate the blade end health and blade end fault type into the blade end diagnosis result, and output the blade end diagnosis result.

9. A blade tip timing measurement system based on multi-sensor fusion, used to implement the blade tip timing measurement method based on multi-sensor fusion according to any one of claims 1 to 8, characterized in that: The blade tip timing measurement system based on multi-sensor fusion includes: The signal acquisition module collects the blade operating parameters in real time based on a preset sampling period through a multi-sensor group, performs analog-to-digital conversion on the blade operating parameters, and obtains a digital working parameter signal; a signal fusion module, configured to load at least one set of operating parameter signals, filter the operating parameter signals, fuse the filtered operating parameter signals, and output a fused signal set; The parameter mapping module pre-builds a three-dimensional finite element model of the blade based on the basic parameters of the rotor blade, establishes a mapping relationship between the single-point parameters of the blade and the full-field parameters of the blade through the three-dimensional finite element model of the blade, and obtains a single-point fusion signal set at the blade tip based on the mapping relationship; The leaf-end diagnosis module is used to pre-build a leaf-end monitoring model based on sparse reconstruction and motion compensation mechanism, iteratively train the leaf-end monitoring model using historical signal sets, output a converged leaf-end monitoring model, and execute the leaf-end monitoring model with the single-point fusion signal set at the leaf end as input. The leaf-end monitoring model identifies and analyzes the single-point fusion signal set and outputs the leaf-end diagnosis results.

10. The blade tip timing measurement system based on multi-sensor fusion according to claim 9, characterized in that: The blade tip diagnostic module includes: A model building unit is used to pre-build a leaf-end monitoring model based on sparse reconstruction and motion compensation mechanism, iteratively train the leaf-end monitoring model using a historical signal set, and output a converged leaf-end monitoring model; The leaf end analysis unit is used to execute the leaf end monitoring model with the single-point fusion signal set at the leaf end as input. The leaf end monitoring model identifies and analyzes the single-point fusion signal set and outputs the leaf end diagnosis result.

Images