Nondestructive flaw detection method for aero-engine blade of micro unmanned aerial vehicle

Abstract

The invention relates to the technical field of non-destructive testing, and particularly discloses an aero-engine blade non-destructive flaw detection method for a miniature unmanned aerial vehicle, which comprises the following steps: acquiring stress data of a blade in a use process in real time by using an MEMS sensor, analyzing the data, and identifying a stress concentration area; carrying out real-time monitoring on crack data of the stress concentration area, extracting crack propagation fluctuation characteristics by utilizing a wavelet transform algorithm, and calculating a crack propagation fluctuation index; fatigue strength data of the material is monitored, fatigue failure characteristics are extracted, and a fatigue failure index is generated through a support vector machine algorithm; comprehensively analyzing the crack propagation fluctuation index and the fatigue failure index, and judging whether the blade is in a healthy state or not; and finally, based on the blade health state, determining whether to carry out further detection or maintenance.

Description

Technical Field

[0001] The present invention relates to the technical field of nondestructive testing, and in particular to a nondestructive flaw detection method for an aero-engine blade of a micro-UAV. Background Art

[0002] With the rapid development of micro-UAV technology, its application areas have gradually expanded to multiple industries such as agricultural monitoring, environmental exploration, logistics distribution, and military reconnaissance. As an efficient and flexible aerial platform, micro-UAVs are widely used in various complex environments, requiring them to have efficient, stable and safe flight performance. However, during the flight of the UAV, especially in long-term, high-load and complex environments, the blades, as one of the core components of the UAV, are subjected to multiple loads such as high-speed air flow, temperature changes, vibration shock, etc. for a long time, and are prone to stress concentration, crack propagation, fatigue damage and other problems. If these problems are not discovered and effectively solved in time, they may lead to blade failure, and then cause flight accidents or equipment failures.

[0003] Traditional blade inspection methods usually rely on manual regular inspections or post-flight maintenance inspections, which is not only time-consuming and laborious, but may also miss the best time for maintenance due to the long inspection cycle, leading to the accumulation of potential problems and serious consequences. In addition, existing traditional inspection methods, such as visual inspection, manual touch inspection, and non-destructive testing, can detect cracks, corrosion and other damage on blades to some extent, but due to their limitations and the singleness of the inspection methods, it is difficult to conduct comprehensive, real-time and efficient health monitoring of micro-UAV blades. Summary of the invention

[0004] The purpose of the present invention is to provide a non-destructive flaw detection method for aircraft engine blades of micro-UAVs to solve the above-mentioned problems.

[0005] The purpose of the present invention can be achieved through the following technical solutions:

[0006] The nondestructive flaw detection method for aeroengine blades of a micro-UAV comprises the following steps:

[0007] S1: Obtain stress data of the blade during use, analyze the stress data, and determine the stress concentration area;

[0008] S2: Real-time monitoring of crack data in stress concentration areas, processing of crack data, extraction of crack growth fluctuation characteristics, and generation of crack growth fluctuation index based on the crack growth fluctuation characteristics;

[0009] S3: Real-time monitoring of fatigue strength data of materials in stress concentration areas, processing of fatigue strength data, extraction of fatigue failure characteristics, and generation of fatigue failure index based on fatigue failure characteristics;

[0010] S4: Comprehensively analyze the crack growth fluctuation index and fatigue failure index of the stress concentration area to determine whether the blade is healthy, and classify the blade into a healthy blade and an unhealthy blade according to the determination result;

[0011] S5: Based on the healthy blades, the current monitoring state is continued, and based on the unhealthy blades, further detection is performed to determine whether the blades need maintenance.

[0012] As a further solution of the present invention: the step of obtaining stress data on the blade during use, analyzing the stress data, and determining the stress concentration area specifically includes:

[0013] The strain data obtained by the MEMS sensor includes the strain value, combined with the elastic modulus of the material, to calculate the local stress value;

[0014] Determine whether the local stress value is greater than or equal to the preset stress threshold. If so, record it as a stress concentration area.

[0015] As a further solution of the present invention: the real-time monitoring of the crack data in the stress concentration area and the processing of the crack data to extract the crack extension fluctuation characteristics and generate the crack extension fluctuation index according to the crack extension fluctuation characteristics specifically include:

[0016] Real-time collection of blade crack data;

[0017] Among them, the crack data includes: crack extension displacement signal, strain signal, and vibration signal;

[0018] Use wavelet transform algorithm to monitor crack data in real time, including signal preprocessing, wavelet transform decomposition, fluctuation feature extraction, and fluctuation index calculation;

[0019] By extracting high-frequency detail coefficients and calculating their variance, the fluctuation characteristics during crack growth are identified and the crack growth fluctuation index is generated.

[0020] As a further solution of the present invention: the specific process of the wavelet transform algorithm is:

[0021] Select Daubechies wavelet, and the wavelet function is expressed as:

[0022]

[0023] Where ψ(t) represents the basis function of wavelet transform;

[0024] For crack data, multi-level wavelet decomposition is performed;

[0025] Assuming that the crack extension signal is y(t), different approximate coefficients and detail coefficients are obtained after discrete wavelet transform;

[0026] Through discrete wavelet transform, the crack extension signal is represented as multiple wavelet decomposition levels, and the calculation expression is:

[0027]

[0028] In the formula, j represents the number of layers of wavelet transform, j is a positive integer greater than 0, J represents the maximum number of layers of wavelet transform, c j represents the approximate coefficients of the j-th layer wavelet transform, d j represents the detail coefficient of the j-th layer wavelet transform, ψ j (t) represents the wavelet basis function of the jth layer;

[0029] For the high frequency part, detail coefficients are included;

[0030] The high-frequency detail coefficients are extracted from the wavelet decomposition to evaluate the instantaneous fluctuation of crack extension. The variance of the wavelet detail coefficients of each layer is calculated. The calculation expression is:

[0031]

[0032] Where Va represents the value of fluctuation during crack propagation, i represents the number of detail coefficients, N represents the maximum number of detail coefficients, and μ d represents the mean value of detail coefficient;

[0033] Combined with the high-frequency fluctuation characteristics obtained by wavelet transform, the crack extension fluctuation index is calculated, and the calculation expression is:

[0034]

[0035] Where CP represents the crack growth fluctuation index, σ d Represents the standard deviation of the high-frequency detail coefficient.

[0036] As a further solution of the present invention: real-time monitoring of fatigue strength data of materials in stress concentration areas, processing of fatigue strength data, extraction of fatigue failure characteristics, and generation of fatigue failure index according to fatigue failure characteristics specifically include:

[0037] Obtain fatigue strength data of materials in stress concentration areas including stress amplitude, strain amplitude, number of cycles and fatigue damage accumulation value;

[0038] A SVM model is constructed to predict the degree of fatigue failure of the blade and generate a fatigue failure index.

[0039] As a further solution of the present invention: the process of obtaining the fatigue failure index is:

[0040] Collect real-time fatigue strength data of blade materials in stress concentration areas from a sensor network;

[0041] The collected real-time fatigue strength data is preprocessed, including denoising and standardization;

[0042] When analyzing fatigue strength data, the features for fatigue failure prediction are extracted from the fatigue strength data, including stress amplitude features, strain amplitude features, cycle number features, and fatigue damage accumulation features;

[0043] Use historical stress amplitude features, strain amplitude features, cycle number features, and fatigue damage accumulation features as training sets for training;

[0044] Use real-time stress amplitude features, strain amplitude features, cycle number features, and fatigue damage accumulation features as test sets;

[0045] The standardized real-time stress amplitude characteristics, strain amplitude characteristics, cycle number characteristics and fatigue damage accumulation characteristics are used as the input of the SVM model;

[0046] The predicted fatigue failure index is taken as output;

[0047] Optimize the model through the loss function of the SVM model;

[0048] According to the input stress amplitude characteristics, strain amplitude characteristics, cycle number characteristics and fatigue damage accumulation characteristics, the fatigue failure index is calculated through the calculation expression of the SVM model;

[0049] Among them, the calculation expression of the SVM model is:

[0050] f(x)=w T φ(x)+b;

[0051] Where f(x) represents the predicted fatigue failure index, x represents the input feature, and w T is the transpose of the weight vector w, φ(x) is the kernel function, and b is the bias term.

[0052] As a further solution of the present invention: the comprehensive analysis of the crack growth fluctuation index and fatigue failure index of the stress concentration area to calculate the blade health index specifically includes:

[0053] Obtain the crack growth fluctuation index and fatigue failure index in the stress concentration area;

[0054] The crack growth fluctuation index and fatigue failure index are normalized, and the blade health index is calculated.

[0055] As a further solution of the present invention: the step of judging whether the blade is healthy specifically includes:

[0056] Determine whether the leaf health index is greater than or equal to the preset leaf health index threshold. If so, record it as a healthy leaf; if not, record it as an unhealthy leaf.

[0057] As a further solution of the present invention: the further detection based on the unhealthy blades to determine whether the blades need to be repaired specifically includes:

[0058] Based on the unhealthy blade, at this time, the system will trigger an early warning, mark the blade as unhealthy, and start further detection procedures, including increasing the data collection frequency of the sensor and checking for faults in the blade;

[0059] For non-healthy leaves, further inspections include: detailed scanning of the leaves using non-destructive testing techniques;

[0060] Transmit the test results to the system control center;

[0061] Make maintenance decisions based on further test results.

[0062] Beneficial effects of the present invention:

[0063] (1) By real-time monitoring and accurate analysis of the stress, crack growth and fatigue failure data of the micro-UAV aeroengine blades, the safety and reliability of the blades are effectively improved. Traditional blade health monitoring relies on periodic manual inspections or post-flight inspections, which is not only time-consuming and costly, but also often difficult to detect potential damage or failure risks in a timely manner. The present invention integrates MEMS sensors for real-time data acquisition, which can monitor the stress state of the blades during flight at any time, especially the stress concentration area, and detect signs of crack growth and fatigue damage in a timely manner. The crack growth fluctuation characteristics are analyzed by combining the wavelet transform algorithm, and the risk of fatigue failure is predicted by the support vector machine (SVM) algorithm, which can accurately assess the health status of the blades. This real-time monitoring can not only provide early warning of potential faults and prevent small problems from evolving into major faults, but also greatly improve the flight safety of the UAV and avoid flight accidents or equipment failures caused by blade damage. In addition, based on the monitoring status of healthy blades, the system can maintain stable operation, and when unhealthy blades are detected, it can accurately determine whether they need further maintenance or replacement, thereby avoiding unnecessary grounding and maintenance, and ensuring the continuous and efficient operation of the micro-UAV.

[0064] (2) Through efficient data processing and intelligent analysis technology, the maintenance cost and operation cycle of UAV blades are significantly reduced. Traditional blade maintenance mode often relies on regular inspection and manual judgment. The timing and necessity of maintenance are often judged based on experience, which easily leads to premature or late maintenance, wasting a lot of resources and missing the best maintenance time. The present invention adopts sensor-based real-time data acquisition and automated analysis, which can accurately obtain the stress distribution, crack propagation and fatigue failure of the blades during flight, and then generate health index and fatigue failure index through comprehensive analysis to automatically determine whether the blades are in a healthy state. For healthy blades, the system can maintain the existing monitoring frequency without frequent manual intervention. For unhealthy blades, the system can quickly trigger an early warning, accurately locate the potential damage area and decide whether further inspection or maintenance is needed. This accurate prediction and judgment reduces the risk of over-maintenance or neglect of problems that may exist in traditional inspections, reduces maintenance costs, and improves maintenance efficiency. By optimizing maintenance decisions, not only the operating efficiency of the UAV is improved, but also the service life of the blades is extended, and the waste caused by premature replacement of parts is reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] The present invention will be further described below in conjunction with the accompanying drawings.

[0066] Figure 1 It is a flowchart of the specific steps of the non-destructive flaw detection method of the aero-engine blade of the micro-UAV of the present invention;

[0067] Figure 2 It is a flowchart of the steps for obtaining the crack extension fluctuation index in the present invention;

[0068] Figure 3 It is a flowchart of the steps for obtaining the fatigue failure index in the present invention. DETAILED DESCRIPTION

[0069] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0070] See also Figure 1 As shown, the present invention is a non-destructive flaw detection method for aeroengine blades of a micro-UAV, comprising the following steps:

[0071] S1: Obtain stress data of the blade during use, analyze the stress data, and determine the stress concentration area;

[0072] S2: Real-time monitoring of crack data in stress concentration areas, processing of crack data, extraction of crack growth fluctuation characteristics, and generation of crack growth fluctuation index based on the crack growth fluctuation characteristics;

[0073] S3: Real-time monitoring of fatigue strength data of materials in stress concentration areas, processing of fatigue strength data, extraction of fatigue failure characteristics, and generation of fatigue failure index based on fatigue failure characteristics;

[0074] S4: Comprehensively analyze the crack growth fluctuation index and fatigue failure index of the stress concentration area to determine whether the blade is healthy, and classify the blade into a healthy blade and an unhealthy blade according to the determination result;

[0075] S5: Based on the healthy blades, the current monitoring state is continued, and based on the unhealthy blades, further detection is performed to determine whether the blades need maintenance.

[0076] In S1, the stress data of the blade during use is obtained, the stress data is analyzed, and the stress concentration area is determined, including:

[0077] Select appropriate MEMS sensors and calibrate the performance of the sensors to ensure the accuracy of the collected data;

[0078] The MEMS sensors are evenly distributed and installed on the blade surface to collect the stress in each area of ​​the blade;

[0079] During installation, it is necessary to ensure that the sensor is in close contact with the blade surface and that the direction of the sensor is consistent with the force direction of the blade to ensure the accuracy of the data;

[0080] It should be noted that: MEMS sensors are small in size and light in weight, making them suitable for installation in micro-UAVs, and will not burden the overall flight performance;

[0081] After installation, the MEMS sensor transmits the collected data to the ground monitoring system in real time through the wireless transmission module;

[0082] Preprocess the stress data obtained from the MEMS sensor, including:

[0083] Filtering: Remove high-frequency noise or low-frequency interfering signals, usually using a low-pass filter or a band-pass filter;

[0084] Remove outliers: Remove abnormal data caused by sensor failure or environmental interference;

[0085] Time synchronization: If multiple sensors collect data at the same time, it is necessary to ensure the consistency of the data on the time axis to avoid errors caused by sensor asynchrony;

[0086] The preprocessed data will be smoother and more reliable, suitable for subsequent analysis.

[0087] The specific process of acquiring stress data is as follows: using the strain value (expressed as microstrain) obtained by the MEMS sensor and combining it with the elastic modulus of the material, the local stress value is calculated;

[0088] Comparing the local stress value with a preset stress threshold;

[0089] If the local stress value is greater than or equal to a preset stress threshold, the corresponding area is marked as a stress concentration area.

[0090] It should be noted that: in the process of stress data collection, the MEMS sensor converts the recorded strain data into stress data, that is, specific stress values, through the material's stress-strain curve (such as Hooke's law).

[0091] See also Figure 2 As shown, in S2: real-time monitoring of crack data in the stress concentration area is performed, and the crack data is processed to extract crack extension fluctuation characteristics, and a crack extension fluctuation index is generated according to the crack extension fluctuation characteristics, specifically including:

[0092] During the monitoring process, it is first necessary to collect blade crack data in real time through sensors (such as strain gauges, MEMS sensors, etc.);

[0093] Among them, the crack data includes: crack extension displacement signal, strain signal, and vibration signal;

[0094] De-noising the collected crack data. Common de-noising methods include wavelet de-noising and moving average. This process helps to remove high-frequency noise and improve the accuracy of the analysis results.

[0095] The process of obtaining the crack extension fluctuation index is as follows: using the wavelet transform algorithm to monitor the crack data in real time, including signal preprocessing, wavelet transform decomposition, fluctuation feature extraction, and fluctuation index calculation;

[0096] By extracting high-frequency detail coefficients and calculating their variance, the fluctuation characteristics during crack growth are identified, thereby generating a crack growth fluctuation index.

[0097] The crack growth fluctuation index provides an important basis for risk assessment of crack growth and maintenance decision-making.

[0098] Among them, the specific process of wavelet transform algorithm is:

[0099] The crack data is processed using wavelet transform, including:

[0100] Select wavelet basis function (such as Haar wavelet, Daubechies wavelet and Symlet wavelet);

[0101] It should be noted that the result of wavelet transform depends on the selected wavelet basis function. In practical applications, Daubechies wavelet (especially db4 or db6) is selected. Daubechies wavelet has good localization performance in frequency domain and time domain.

[0102] In the present invention, assuming that Daubechies wavelet is selected, the wavelet function is expressed as:

[0103]

[0104] Where ψ(t) represents the basis function of wavelet transform, which is used to decompose and reconstruct the signal;

[0105] The core of wavelet transform is to decompose the signal through filters of different scales;

[0106] Discrete wavelet transform (DWT) decomposes the signal into low-frequency (approximation) and high-frequency (detail) components at multiple scales (frequency bands);

[0107] For crack data, multi-level wavelet decomposition is performed;

[0108] Assuming that the crack extension signal is y(t), the approximate coefficients (low frequency) and detail coefficients (high frequency) of different scales are obtained after discrete wavelet transform;

[0109] Through transformation, the crack extension signal is expressed as multiple wavelet decomposition levels, and the calculation expression is:

[0110]

[0111] In the formula, j represents the number of layers of wavelet transform, j is a positive integer greater than 0, J represents the maximum number of layers of wavelet transform, c j represents the approximate coefficients of the j-th layer wavelet transform, d j represents the detail coefficient of the j-th layer wavelet transform, ψ j (t) represents the wavelet basis function of the jth layer;

[0112] The crack growth fluctuation feature extraction process is as follows:

[0113] Through wavelet transform, we can obtain different frequency band components of crack extension signal;

[0114] For the high-frequency part (such as detail coefficient), because the high-frequency part corresponds to the existence of mutation, fluctuation and instability in the crack propagation process;

[0115] The high-frequency detail coefficients are extracted from the wavelet decomposition to evaluate the instantaneous fluctuation of crack extension. The variance of the wavelet detail coefficients of each layer is calculated to reflect the fluctuation degree of the crack extension signal. The calculation expression is:

[0116]

[0117] Where Va represents the value of fluctuation during crack propagation, i represents the number of detail coefficients, N represents the maximum number of detail coefficients, and μ d represents the mean value of detail coefficient;

[0118] Combined with the high-frequency fluctuation characteristics obtained by wavelet transform, the crack extension fluctuation index is calculated, and the calculation expression is:

[0119]

[0120] Where CP represents the crack growth fluctuation index, σ d Represents the standard deviation of the high-frequency detail coefficient;

[0121] It should be noted that the crack growth fluctuation index reflects the fluctuation intensity during the crack growth process. A higher fluctuation index indicates that the instability during the crack growth process is stronger and there may be a risk of rapid growth.

[0122] See also Figure 3 As shown, in S3, the fatigue strength data of the material in the stress concentration area is monitored in real time, and the fatigue strength data is processed to extract fatigue failure characteristics, and a fatigue failure index is generated according to the fatigue failure characteristics, specifically including:

[0123] Obtain fatigue strength data of materials in stress concentration areas (such as stress amplitude, strain amplitude, number of cycles and fatigue damage accumulation value);

[0124] Among them, the ratio of the number of cycles to the fatigue life is calculated, and the ratios of all loading cycles are summed to obtain the cumulative value of fatigue damage;

[0125] Construct an SVM model to predict the degree of fatigue failure of the blade and generate a fatigue failure index, including:

[0126] It should be noted that the SVM model is a supervised learning algorithm, which is used in the present invention for the problem of blade fatigue state classification. In the fatigue strength data processing, the SVM model is used to classify fatigue failure and non-failure states;

[0127] First, it is necessary to collect real-time fatigue strength data of blade materials in stress concentration areas from the sensor network;

[0128] The collected real-time fatigue strength data is preprocessed, including denoising and standardization, to ensure that the algorithm can effectively process the data;

[0129] When analyzing fatigue strength data, the features for fatigue failure prediction are extracted from the fatigue strength data, including stress amplitude features, strain amplitude features, cycle number features, and fatigue damage accumulation features;

[0130] Use historical stress amplitude features, strain amplitude features, cycle number features, and fatigue damage accumulation features as training sets for training;

[0131] Use real-time stress amplitude features, strain amplitude features, cycle number features, and fatigue damage accumulation features as test sets;

[0132] The standardized real-time stress amplitude characteristics, strain amplitude characteristics, cycle number characteristics and fatigue damage accumulation characteristics are used as the input of the SVM model;

[0133] The predicted fatigue failure index is taken as output;

[0134] The model is optimized through the loss function (such as L2 loss) of the SVM model so that the model can accurately predict the fatigue behavior of the material under a specific load;

[0135] According to the input stress amplitude characteristics, strain amplitude characteristics, cycle number characteristics and fatigue damage accumulation characteristics, the fatigue failure index is calculated through the calculation expression of the SVM model;

[0136] Among them, the calculation expression of the SVM model is:

[0137] f(x)=w T φ(x)+b;

[0138] Where f(x) represents the predicted fatigue failure index, x represents the input feature, and w T is the transpose of the weight vector w, φ(x) is the kernel function, and b is the bias term;

[0139] It should be noted that the fatigue failure index reflects the failure degree of the blade, and the larger the fatigue failure index, the higher the risk of corresponding blade failure.

[0140] In S4, the crack growth fluctuation index and fatigue failure index of the stress concentration area are comprehensively analyzed to determine whether the blade is healthy. According to the determination result, the blade is divided into healthy blade and unhealthy blade, including:

[0141] Obtain the crack growth fluctuation index CP and fatigue failure index f(x) in the stress concentration area;

[0142] The crack growth fluctuation index CP and fatigue failure index f(x) are normalized and the blade health index is calculated;

[0143] The calculation expression of the leaf health index is:

[0144]

[0145] In the formula, S VD represents the leaf health index, c 1 and c 2 is the preset proportional coefficient, and c 1 and c 2 All are greater than 0, e represents the logarithm of the base of a natural number;

[0146] The leaf health index S VD Compare with preset leaf health index threshold;

[0147] If the leaf health index S VD If it is greater than or equal to the preset leaf health index threshold, it means that the corresponding leaf is in good health and is recorded as a healthy leaf;

[0148] If the leaf health index S VD If the leaf health index is less than the preset leaf health index threshold, it means that the corresponding leaf is in poor health and is recorded as an unhealthy leaf.

[0149] In S5, based on the healthy blades, the current monitoring state is continued, and based on the unhealthy blades, further detection is performed to determine whether the blades need maintenance, specifically including:

[0150] Based on healthy blades, the system will continue to maintain the current monitoring status, indicating that the corresponding blades do not show the risk of fatigue damage or failure. The system will periodically record stress, strain and other data and conduct routine inspections to ensure that the blades can continue to operate safely;

[0151] Based on the unhealthy blades, it means that there is a risk of fatigue damage to the blades. At this time, the system will trigger an early warning, mark the blade as "unhealthy", and immediately start further detection procedures, including increasing the data collection frequency of the sensor and checking for blade faults;

[0152] For unhealthy blades, further inspection includes using non-destructive testing techniques (such as ultrasonic, X-ray, eddy current testing, etc.) to scan the blades in detail, especially the stress concentration areas;

[0153] Transmit the test results to the system control center;

[0154] Based on further test results, combined with fatigue failure index and specific damage conditions, the system will make maintenance decisions automatically or through manual intervention;

[0155] If obvious damage such as cracks or corrosion is found, the system will repair or replace the blades to prevent blade failure from causing more serious malfunctions;

[0156] If the damage is minor, the system continues to monitor the blade or perform local repairs until the blade meets safety standards.

[0157] Working principle of the invention: The invention provides a non-destructive flaw detection method for aviation engine blades of micro-UAVs, which aims to monitor the working status of the blades in real time to ensure their reliability and safety during use, especially during long-term flight, when the blades are subjected to constantly changing environmental factors such as stress and temperature, which can easily lead to fatigue damage or crack extension; the method includes five main steps: first, stress data is collected in real time by a MEMS sensor installed on the surface of the blade, and the stress distribution in different areas of the blade surface is calculated using the stress-strain relationship and the elastic modulus of the material, and the stress concentration area is further identified; the stress data is filtered, denoised and synchronously processed to ensure the accuracy and reliability of the data, providing a basis for subsequent analysis; then, for the determined stress concentration area, a sensor is used to monitor the crack extension in real time; the crack data is analyzed by wavelet transform, and the fluctuation characteristics of crack extension are extracted to generate a crack extension fluctuation index, which can reflect the risk of crack extension and provide an important basis for maintenance decision-making; at the same time, Combined with the material fatigue strength data of the blade, the risk of fatigue failure is predicted through the fatigue damage accumulation value and the support vector machine model to generate a fatigue failure index; this index is used to evaluate the fatigue state of the blade, thereby achieving an accurate prediction of the risk of blade failure; based on all data collection and analysis, the crack extension fluctuation index and the fatigue failure index are comprehensively analyzed to calculate the blade health index. If the health index is greater than the set threshold, the blade is considered to be in good condition and the current monitoring state continues; if the health index is lower than the threshold, the blade is marked as unhealthy and further detection procedures are initiated; for unhealthy blades, the system will increase the frequency of data collection and use non-destructive testing technology (such as ultrasound, X-ray, eddy current testing, etc.) to conduct more in-depth inspections of the blades, especially in stress concentration areas, to confirm whether there are cracks, corrosion or other damage; finally, based on the test results and the fatigue failure index, the system makes maintenance decisions. If the damage is serious, it is recommended to replace the blade. If the damage is minor, it will continue to monitor or perform local repairs to ensure that the blade can meet safety standards.

[0158] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0159] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0160] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0161] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0162] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A non-destructive testing method for aircraft engine blades of micro-UAVs, characterized in that: The following steps are involved: S1: Obtain stress data of the blade during use, analyze the stress data, and determine the stress concentration area; S2: Real-time monitoring of crack data in stress concentration areas, processing of crack data, extraction of crack growth fluctuation characteristics, and generation of crack growth fluctuation index based on the crack growth fluctuation characteristics; S3: Real-time monitoring of fatigue strength data of materials in stress concentration areas, processing of fatigue strength data, extraction of fatigue failure characteristics, and generation of fatigue failure index based on fatigue failure characteristics; S4: Comprehensively analyze the crack growth fluctuation index and fatigue failure index of the stress concentration area to determine whether the blade is healthy, and classify the blade into a healthy blade and an unhealthy blade according to the determination result; S5: Based on the healthy blades, the current monitoring state is continued, and based on the unhealthy blades, further detection is performed to determine whether the blades need maintenance.

2. The nondestructive flaw detection method for aircraft engine blades of micro-UAV according to claim 1 is characterized in that: The obtaining of stress data on the blade during use, analyzing the stress data, and determining the stress concentration area specifically includes: The strain data obtained by the MEMS sensor includes the strain value, combined with the elastic modulus of the material, to calculate the local stress value; Determine whether the local stress value is greater than or equal to the preset stress threshold. If so, record it as a stress concentration area.

3. The nondestructive flaw detection method for aircraft engine blades of micro-UAV according to claim 1 is characterized in that: The real-time monitoring of the crack data in the stress concentration area and the processing of the crack data to extract the crack extension fluctuation characteristics and generate the crack extension fluctuation index according to the crack extension fluctuation characteristics specifically include: Real-time collection of blade crack data; Among them, the crack data includes: crack extension displacement signal, strain signal, and vibration signal; Use wavelet transform algorithm to monitor crack data in real time, including signal preprocessing, wavelet transform decomposition, fluctuation feature extraction, and fluctuation index calculation; By extracting high-frequency detail coefficients and calculating their variance, the fluctuation characteristics during crack growth are identified and the crack growth fluctuation index is generated.

4. The nondestructive flaw detection method for aircraft engine blades of micro-UAV according to claim 3 is characterized in that: The specific process of the wavelet transform algorithm is as follows: Select Daubechies wavelet, and the wavelet function is expressed as: Where ψ(t) represents the basis function of wavelet transform; For crack data, multi-level wavelet decomposition is performed; Assuming that the crack extension signal is y(t), different approximate coefficients and detail coefficients are obtained after discrete wavelet transform; Through discrete wavelet transform, the crack extension signal is represented as multiple wavelet decomposition levels, and the calculation expression is: In the formula, j represents the number of layers of wavelet transform, j is a positive integer greater than 0, J represents the maximum number of layers of wavelet transform, c j represents the approximate coefficients of the j-th layer wavelet transform, d j represents the detail coefficient of the j-th layer wavelet transform, ψ j (t) represents the wavelet basis function of the jth layer; For the high frequency part, detail coefficients are included; The high-frequency detail coefficients are extracted from the wavelet decomposition to evaluate the instantaneous fluctuation of crack extension. The variance of the wavelet detail coefficients of each layer is calculated. The calculation expression is: Where Va represents the value of fluctuation during crack propagation, i represents the number of detail coefficients, N represents the maximum number of detail coefficients, and μ d represents the mean value of detail coefficient; Combined with the high-frequency fluctuation characteristics obtained by wavelet transform, the crack extension fluctuation index is calculated, and the calculation expression is: Where CP represents the crack growth fluctuation index, σ d Represents the standard deviation of the high-frequency detail coefficient.

5. The nondestructive flaw detection method for aircraft engine blades of micro-UAV according to claim 1 is characterized in that: The real-time monitoring of the fatigue strength data of the material in the stress concentration area and processing of the fatigue strength data to extract fatigue failure characteristics and generate a fatigue failure index according to the fatigue failure characteristics specifically include: Obtain fatigue strength data of materials in stress concentration areas including stress amplitude, strain amplitude, number of cycles and fatigue damage accumulation value; A SVM model is constructed to predict the degree of fatigue failure of the blade and generate a fatigue failure index.

6. The non-destructive flaw detection method for aircraft engine blades of micro-UAV according to claim 5, characterized in that: The process of obtaining the fatigue failure index is as follows: Collect real-time fatigue strength data of blade materials in stress concentration areas from a sensor network; The collected real-time fatigue strength data is preprocessed, including denoising and standardization; When analyzing fatigue strength data, the features for fatigue failure prediction are extracted from the fatigue strength data, including stress amplitude features, strain amplitude features, cycle number features, and fatigue damage accumulation features; Use historical stress amplitude features, strain amplitude features, cycle number features, and fatigue damage accumulation features as training sets for training; Use real-time stress amplitude features, strain amplitude features, cycle number features, and fatigue damage accumulation features as test sets; The standardized real-time stress amplitude characteristics, strain amplitude characteristics, cycle number characteristics and fatigue damage accumulation characteristics are used as the input of the SVM model; The predicted fatigue failure index is taken as output; Optimize the model through the loss function of the SVM model; According to the input stress amplitude characteristics, strain amplitude characteristics, cycle number characteristics and fatigue damage accumulation characteristics, the fatigue failure index is calculated through the calculation expression of the SVM model; Among them, the calculation expression of the SVM model is: f(x)=w T φ(x)+b; Where f(x) represents the predicted fatigue failure index, x represents the input feature, and w T is the transpose of the weight vector w, φ(x) is the kernel function, and b is the bias term.

7. The nondestructive flaw detection method for aircraft engine blades of micro-UAV according to claim 1, characterized in that: The comprehensive analysis of the crack growth fluctuation index and fatigue failure index of the stress concentration area to calculate the blade health index specifically includes: Obtain the crack growth fluctuation index and fatigue failure index in the stress concentration area; The crack growth fluctuation index and fatigue failure index are normalized, and the blade health index is calculated.

8. The nondestructive flaw detection method for aircraft engine blades of micro-UAV according to claim 1, characterized in that: The step of judging whether the blade is healthy specifically includes: Determine whether the leaf health index is greater than or equal to the preset leaf health index threshold. If so, record it as a healthy leaf; if not, record it as an unhealthy leaf.

9. The nondestructive flaw detection method for aircraft engine blades of micro-UAV according to claim 1, characterized in that: The further detection based on the unhealthy blade to determine whether the blade needs maintenance specifically includes: Based on the unhealthy blade, at this time, the system will trigger an early warning, mark the blade as unhealthy, and start further detection procedures, including increasing the data collection frequency of the sensor and checking for faults in the blade; For non-healthy leaves, further inspections include: detailed scanning of the leaves using non-destructive testing techniques; Transmit the test results to the system control center; Make maintenance decisions based on further test results.

Images