Oil casing micro-crack detection method based on quantum computing and topological data analysis

Abstract

The present invention discloses a method for detecting micro cracks in oil casing based on quantum computing and topological data analysis, and specifically relates to the field of crack detection. The present invention first fixes III-V group quantum dots on the surface of a sensor, cleans the oil casing to enhance the adhesion of the quantum dots, excites the quantum dots with a laser to generate a fluorescence signal, and uses a photodetector to collect relevant data; then, a Kalman filter is applied to denoise the original data to improve the data quality; a quantum principal component analysis method is used to perform topological analysis on the pre-processed data and calculate the crack value, and whether a crack exists is determined according to a preset threshold; if so, an ultrasonic sensor is used to collect surface wave and shear wave velocities, and a crack distance value is calculated; finally, the crack position is confirmed by the crack value and the distance value, and a support vector machine model is used to realize real-time recognition of the crack type, and a crack recognition report is provided to facilitate technical personnel to make decisions.

Description

Technical Field

[0001] The present invention relates to the technical field of crack detection, and more specifically, to a method for detecting micro cracks in oil casing based on quantum computing and topological data analysis. Background Art

[0002] Quantum sensing technology is a technology that uses quantum effects to perform high-precision measurements. In the detection of micro-cracks in oil casing, luminescent materials such as quantum dots can be used as sensing elements to monitor micro-cracks on the surface of oil casing in real time through their unique fluorescence properties. Quantum dots have a narrow fluorescence spectrum, a wide excitation spectrum, good optical stability, and are not prone to photobleaching, which can achieve online monitoring of micro-cracks and fluorescence visualization imaging.

[0003] Topological data analysis is a data analysis method based on the principle of topology. It reveals the intrinsic structure and properties of data by extracting topological features (such as points, lines, surfaces, etc.) from the data. In the detection of micro-cracks in oil casing, topological data analysis can be used to identify the shape, size, location and other characteristics of the cracks, providing important information support for crack detection.

[0004] However, there are still some shortcomings in its actual use. For example, traditional crack detection methods often rely on manual visual inspection or simple physical measurement, which is not only inefficient but also difficult to guarantee accuracy. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a method for detecting micro-cracks in oil casing based on quantum computing and topological data analysis to solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] Step A1: Use quantum sensors to scan the surface of the oil casing and collect raw data containing crack information;

[0008] Step A2: performing denoising and filtering preprocessing operations on the collected raw data to improve data quality;

[0009] Step A3: Using quantum algorithms to perform topological data analysis on the preprocessed data, and calculating the crack value;

[0010] Step A4: using an ultrasonic sensor to collect the surface wave velocity and the shear wave velocity, and calculating the surface wave velocity and the shear wave velocity to obtain a crack distance value;

[0011] Step A5: Confirm the crack position according to the crack value and the crack distance value, and identify the cracks on the surface of the oil casing in combination with the known crack feature library.

[0012] Preferably, in step A1, the quantum sensor selects group III-V quantum dots as quantum total materials, fixes the selected quantum dots on the sensor surface by physical adsorption, and then uses a detergent to thoroughly clean the surface of the oil casing to remove dirt, grease and other impurities that may affect the measurement results, and modifies the surface of the oil casing to enhance the interaction between the quantum dots and the surface and improve the adhesion of the quantum dots; uses a laser to irradiate the quantum dot sensor to stimulate the electrons in the quantum dots to transition to a higher energy level; generates detectable fluorescence signals and other signals; detects the fluorescence signals and other signals emitted by the quantum dots by a highly sensitive photodetector device; the other signals contain raw data about cracks on the surface of the oil casing; the raw data includes luminescence intensity, wavelength value, spectral shift value, and fluorescence frequency.

[0013] Preferably, in step A2, the raw data of the surface cracks of the oil casing collected by the high-precision sensor is subjected to Kalman filtering and denoising, and the steps of Kalman filtering and denoising are specifically as follows:

[0014] Step B1: define the system's state transfer matrix, control input matrix, and process noise covariance matrix;

[0015] Step B2: Based on the state estimate and state transfer matrix of the previous moment, predict the state at the current moment; at the same time, update the covariance matrix of the state estimate to reflect the uncertainty of the prediction;

[0016] Step B3: Obtain the observation value at the current moment from the sensor;

[0017] Step B4: Determine the observation matrix and the observation noise covariance matrix;

[0018] Step B5: Use Kalman gain to combine the observed value with the predicted value to obtain an updated state estimate; wherein the Kalman gain is a weight used to weigh the predicted value and the observed value, which is calculated based on the prediction covariance matrix and the observation noise covariance matrix;

[0019] Step B6: Update the covariance matrix of the state estimate to reflect the updated uncertainty;

[0020] Step B7: For each new observation, repeat the prediction and update steps to continuously update the state estimate and covariance matrix;

[0021] Step B8: The state estimation value after Kalman filtering is the denoised signal.

[0022] Preferably, in step A3, the quantum algorithm selects quantum principal component analysis, and selects the eigenvectors corresponding to the data eigenvalues ​​in the largest number of samples as principal components according to the size of the eigenvalues; these principal components can retain the variation information of the original data to the maximum extent; and taking the largest eigenvalue and its corresponding eigenvector as the principal component is one of the key steps.

[0023] Preferably, in step A4, an ultrasonic sensor is selected to ensure that its performance indicators such as frequency, sensitivity and directivity meet the experimental requirements; according to the experimental requirements, a coupling agent is prepared to improve the transmittance of ultrasonic waves on the surface of the material; the surface of the material to be tested is ensured to be clean and flat to reduce the impact on the propagation of ultrasonic waves; the ultrasonic sensor is fixed on a bracket, and its distance and angle with the surface of the material to be tested are adjusted; an ultrasonic signal is generated by a pulse transmitter, and the reflected ultrasonic signal is received by a receiver; the pulse transmitter is started to transmit an ultrasonic signal to the surface of the material to be tested; after the receiver receives the reflected ultrasonic signal, the waveform is observed through an oscilloscope or other display device; according to the characteristics of the waveform, the propagation of surface waves and shear waves is judged, and relevant data is recorded; the collected data is preprocessed to remove noise and interference signals, and the wave velocity of the surface wave and shear wave is calculated using relevant algorithms.

[0024] Preferably, in step A5, the crack value and the crack distance value determine the location of the crack. In the actual detection process, the extracted crack features are input into the trained support vector machine model to realize real-time recognition of the crack type; provide detailed recognition results, including crack type, possible causes and recommended treatment measures; generate a crack recognition report in combination with visualization tools to facilitate technical personnel to review and make maintenance decisions; feed back new detection data and recognition results to the feature library to realize dynamic updating of the feature library and continuously improve the recognition effect; continuously iterate the model through the addition of new data, and adopt online learning or incremental learning and other technologies to keep the model up to date.

[0025] Technical effects and advantages of the present invention:

[0026] The present invention first uses quantum sensors to collect raw data on surface cracks of oil casing, and improves data quality through Kalman filtering and denoising. Subsequently, quantum principal component analysis is used to perform topological data analysis, extract crack features and calculate crack values. Furthermore, an ultrasonic sensor is used to measure crack distance values ​​and accurately locate crack positions. Finally, combined with a crack feature library, a support vector machine model is used to identify crack types and generate a detailed identification report. The present invention combines the high efficiency of quantum computing and the accuracy of topological data analysis, realizes efficient and accurate detection and identification of micro cracks in oil casing, and provides a strong guarantee for safe production in the process of oil and natural gas extraction. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a schematic diagram of the method flow of the present invention. DETAILED DESCRIPTION

[0028] 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.

[0029] See also Figure 1 As shown, the present invention provides a method for detecting micro cracks in oil casing based on quantum computing and topological data analysis, and the method is:

[0030] Step A1: Use quantum sensors to scan the surface of the oil casing and collect raw data containing crack information;

[0031] In the step A1, the quantum sensor selects group III-V quantum dots as quantum total materials, fixes the selected quantum dots on the sensor surface by physical adsorption, and then uses a detergent to thoroughly clean the surface of the oil casing to remove dirt, grease and other impurities that may affect the measurement results, and modifies the surface of the oil casing to enhance the interaction between the quantum dots and the surface and improve the adhesion of the quantum dots; uses a laser to irradiate the quantum dot sensor to stimulate the electrons in the quantum dots to transition to a higher energy level; generates detectable fluorescence signals and other signals; detects the fluorescence signals and other signals emitted by the quantum dots by a highly sensitive photodetector device; the other signals contain raw data about cracks on the surface of the oil casing; the raw data includes luminescence intensity, wavelength value, spectral shift value, and fluorescence frequency.

[0032] Step A2: performing denoising and filtering preprocessing operations on the collected raw data to improve data quality;

[0033] In step A2, the raw data of the surface cracks of the oil casing collected by the high-precision sensor is subjected to Kalman filtering and denoising. The steps of Kalman filtering and denoising are specifically as follows:

[0034] Step B1: define the system's state transfer matrix, control input matrix, and process noise covariance matrix;

[0035] The specific calculation method of the state equation is:

[0036] ,in, It is expressed as the state equation, It is expressed as the state transfer equation, is represented as process noise;

[0037] Step B2: Based on the state estimate and state transfer matrix of the previous moment, predict the state at the current moment; at the same time, update the covariance matrix of the state estimate to reflect the uncertainty of the prediction;

[0038] Step B3: Obtain the observation value at the current moment from the sensor;

[0039] The calculation method of the observed value is as follows:

[0040] ,in, Expressed as observed value, is the observation noise, Expressed as the process noise covariance matrix;

[0041] Step B4: Determine the observation matrix and the observation noise covariance matrix;

[0042] Step B5: Use Kalman gain to combine the observed value with the predicted value to obtain an updated state estimate; wherein the Kalman gain is a weight used to weigh the predicted value and the observed value, which is calculated based on the prediction covariance matrix and the observation noise covariance matrix;

[0043] The calculation method of the predicted value is as follows:

[0044] ,in, Expressed as predicted value, It is expressed as the state transfer equation, It is represented as the posterior estimate of the state x at time step k−1, that is, it is based on all available information at time k−1;

[0045] The calculation method of prediction covariance is as follows:

[0046] ,in, Denoted as the prediction covariance, It is expressed as the state transfer equation, Expressed as the state estimation covariance matrix, Expressed as The transposed matrix of Expressed as the process noise covariance matrix;

[0047] The calculation method of Kalman gain is as follows:

[0048] ,in, Expressed as the Kalman gain, Denoted as the prediction covariance, Expressed as the transposed matrix of the process noise covariance matrix C, Expressed as the measurement noise covariance matrix;

[0049] Step B6: Update the covariance matrix of the state estimate to reflect the updated uncertainty;

[0050] The specific method for calculating the covariance matrix is:

[0051] ,in, Denoted as the covariance matrix of the updated state estimate, Expressed as predicted value, Expressed as the Kalman gain, Expressed as observed value, Expressed as the process noise covariance matrix;

[0052] Step B7: For each new observation, repeat the prediction and update steps to continuously update the state estimate and covariance matrix;

[0053] Step B8: The state estimation value after Kalman filtering is the denoised signal;

[0054] The calculation method of the updated covariance matrix is ​​as follows:

[0055] ,in, is represented as the updated covariance matrix, p is represented as the identity matrix, Expressed as the Kalman gain, Expressed as the process noise covariance matrix, Denoted as the prediction covariance.

[0056] Step A3: Using quantum algorithms to perform topological data analysis on the preprocessed data, and calculating the crack value;

[0057] In step A3, the quantum algorithm selects the quantum principal component analysis method, and selects the eigenvectors corresponding to the data eigenvalues ​​in the largest number of samples as the principal components according to the size of the eigenvalues; these principal components can retain the variation information of the original data to the maximum extent; the largest eigenvalue and its corresponding eigenvector as the principal component is one of the key steps; the calculation method is specifically as follows:

[0058] The calculation method of the covariance matrix is ​​as follows:

[0059] Let the original data matrix be D, where the dimension of D is , It is represented by the number of samples, and o is represented by the number of features; the data in the sample number includes luminescence intensity, wavelength value, spectral shift value, and fluorescence frequency;

[0060] , where E is the covariance matrix, is the number of samples, D is the original data matrix, Represented as the mean vector of the original data matrix D, Expressed as The transposed matrix of

[0061] The covariance matrix is ​​subjected to eigendecomposition. The calculation method of eigendecomposition is as follows:

[0062] ,in, Expressed as eigendecomposition values, Represented as the data feature value in the i-th sample number, Represented as the corresponding eigenvector;

[0063] Project the original data onto the selected principal component direction to obtain the reduced-dimensional data; the reduced-dimensional data matrix The specific calculation method is:

[0064] ,in, Represented as the data matrix after dimensionality reduction, Represented as a , o represents the number of features, q represents the number of selected principal components, that is, the dimension after dimensionality reduction, and E represents the covariance matrix. Expressed as The transposed matrix of

[0065] After the analysis by quantum principal component analysis, through the above steps, the largest eigenvalue and its corresponding eigenvector are selected as the principal component, thereby achieving data dimensionality reduction and information extraction;

[0066] The crack value is calculated by the data analyzed by the quantum principal component analysis method. The specific method for calculating the crack value is:

[0067] , where G represents the crack value, H represents the luminescence intensity, J represents the fluorescence frequency, L represents the wavelength value, and M represents the spectral shift value. , Expressed as a weight factor;

[0068] The calculated crack value is compared with the preset crack threshold. If the calculated crack value is equal to the preset crack threshold, the no-crack information is output without performing too much operation; if the calculated crack value is not equal to the preset crack threshold, the crack information is output and the process goes to step A4 to determine the crack location based on the crack distance value.

[0069] Step A4: using an ultrasonic sensor to collect the surface wave velocity and the shear wave velocity, and calculating the surface wave velocity and the shear wave velocity to obtain a crack distance value;

[0070] In the step A4, an ultrasonic sensor is selected to ensure that its performance indicators such as frequency, sensitivity and directivity meet the experimental requirements; according to the experimental requirements, a coupling agent is prepared to improve the transmittance of ultrasonic waves on the surface of the material; the surface of the material to be tested is ensured to be clean and flat to reduce the impact on the propagation of ultrasonic waves; the ultrasonic sensor is fixed on the bracket, and its distance and angle with the surface of the material to be tested are adjusted; an ultrasonic signal is generated by a pulse transmitter, and the reflected ultrasonic signal is received by a receiver; the pulse transmitter is started to transmit an ultrasonic signal to the surface of the material to be tested; after the receiver receives the reflected ultrasonic signal, the waveform is observed by an oscilloscope or other display device; according to the characteristics of the waveform, the propagation of surface waves and shear waves is judged, and relevant data is recorded; the collected data is preprocessed to remove noise and interference signals, and the wave velocities of surface waves and shear waves are calculated using relevant algorithms;

[0071] The calculation method of surface wave velocity is as follows:

[0072] , where K represents the surface wave velocity, It is expressed as the Poisson's ratio of the material, and L is expressed as the longitudinal wave velocity;

[0073] The calculation method of longitudinal wave velocity is as follows:

[0074] , where L represents the longitudinal wave velocity, Expressed as Young's elastic modulus of the material, Expressed as the density of the material;

[0075] The specific calculation method of shear wave velocity is:

[0076] , where Q represents the shear wave velocity, Expressed as the shear elastic modulus of the material, Expressed as the density of the material;

[0077] The calculation method of the crack distance value is as follows:

[0078] , where Z represents the crack distance value, t represents the time interval between the first echo and the second echo of the surface defect, K represents the surface wave velocity, L represents the longitudinal wave velocity, and Q represents the shear wave velocity;

[0079] The calculated crack distance value can accurately locate the distance between the ultrasonic sensor and the crack.

[0080] Step A5: confirm the crack position according to the crack value and the crack distance value, and identify the cracks on the surface of the oil casing in combination with the known crack feature library;

[0081] In step A5, the crack value and the crack distance value determine the location of the crack. In the actual detection process, the extracted crack features are input into the trained support vector machine model to realize real-time recognition of the crack type; provide detailed recognition results, including crack type, possible causes and recommended treatment measures; generate a crack recognition report in combination with visualization tools to facilitate technical personnel to review and make maintenance decisions; feed back new detection data and recognition results to the feature library to realize dynamic update of the feature library and continuously improve the recognition effect; continuously iterate the model through the addition of new data, and adopt online learning or incremental learning and other technologies to keep the model up to date.

[0082] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for detecting micro cracks in oil casing based on quantum computing and topological data analysis, characterized in that: include: Step A1: Use quantum sensors to scan the surface of the oil casing and collect raw data containing crack information; Step A2: performing denoising and filtering preprocessing operations on the collected raw data to improve data quality; Step A3: Using quantum algorithms to perform topological data analysis on the preprocessed data and calculate the crack value; In step A3, the quantum algorithm selects the quantum principal component analysis method, and selects the eigenvector corresponding to the data eigenvalue in the largest number of samples as the principal component according to the size of the eigenvalue; The calculation method of the covariance matrix is ​​as follows: Let the original data matrix be D, where the dimension of D is , It is represented by the number of samples, and o is represented by the number of features; the data in the sample number includes luminescence intensity, wavelength value, spectral shift value, and fluorescence frequency; , where E is the covariance matrix, is the number of samples, D is the original data matrix, Represented as the mean vector of the original data matrix D, Expressed as The transposed matrix of The covariance matrix is ​​subjected to eigendecomposition. The calculation method of eigendecomposition is as follows: ,in, Expressed as eigendecomposition values, Represented as the data feature value in the i-th sample number, Represented as the corresponding eigenvector; Project the original data onto the selected principal component direction to obtain the reduced-dimensional data; the reduced-dimensional data matrix The specific calculation method is: ,in, Represented as the data matrix after dimensionality reduction, Represented as a The matrix of , E is represented as the covariance matrix, Expressed as The transposed matrix of After analysis by quantum principal component analysis, the largest eigenvalue and its corresponding eigenvector are selected as the principal component, thereby achieving data dimensionality reduction and information extraction; The crack value is calculated by the data analyzed by the quantum principal component analysis method. The specific calculation method of the crack value is: , where G represents the crack value, H represents the luminescence intensity, J represents the fluorescence frequency, L represents the wavelength value, and M represents the spectral shift value. , Expressed as a weight factor; Step A4: using an ultrasonic sensor to collect the surface wave velocity and the shear wave velocity, and calculating the surface wave velocity and the shear wave velocity to obtain a crack distance value; In the step A4, the ultrasonic sensor is fixed on the bracket, and the distance and angle between the ultrasonic sensor and the surface of the material to be measured are adjusted; an ultrasonic signal is generated by a pulse transmitter, and the reflected ultrasonic signal is received by a receiver; the pulse transmitter is started to transmit the ultrasonic signal to the surface of the material to be measured; after the receiver receives the reflected ultrasonic signal, the waveform is observed by an oscilloscope or other display device; The calculation method of surface wave velocity is as follows: , where K represents the surface wave velocity, It is expressed as the Poisson's ratio of the material, and L is expressed as the longitudinal wave velocity; The calculation method of longitudinal wave velocity is as follows: , where L represents the longitudinal wave velocity, Expressed as Young's elastic modulus of the material, Expressed as the density of the material; The specific calculation method of shear wave velocity is: , where Q represents the shear wave velocity, Expressed as the shear elastic modulus of the material, Expressed as the density of the material; The calculation method of the crack distance value is as follows: , where Z represents the crack distance value, t represents the time interval between the first echo and the second echo of the surface defect, K represents the surface wave velocity, L represents the longitudinal wave velocity, and Q represents the shear wave velocity; Step A5: Confirm the crack position according to the crack value and the crack distance value, and identify the cracks on the surface of the oil casing in combination with the known crack feature library.

2. The oil casing micro-crack detection method based on quantum computing and topological data analysis according to claim 1 is characterized in that: In the step A1, the quantum sensor selects group III-V quantum dots as quantum total materials, fixes the selected quantum dots on the sensor surface by physical adsorption, and then uses a detergent to thoroughly clean the surface of the oil casing to remove dirt, grease and other impurities that may affect the measurement results, and modifies the surface of the oil casing to enhance the interaction between the quantum dots and the surface and improve the adhesion of the quantum dots; uses a laser to irradiate the quantum dot sensor to stimulate the electrons in the quantum dots to transition to a higher energy level; generates detectable fluorescence signals and other signals; detects the fluorescence signals and other signals emitted by the quantum dots by a highly sensitive photodetector device; the other signals contain raw data about cracks on the surface of the oil casing; the raw data includes luminescence intensity, wavelength value, spectral shift value, and fluorescence frequency.

3. The oil casing micro-crack detection method based on quantum computing and topological data analysis according to claim 1 is characterized in that: In step A2, the raw data of the surface cracks of the oil casing collected by the high-precision sensor is subjected to Kalman filtering and denoising. The steps of Kalman filtering and denoising are specifically as follows: Step B1: define the system's state transfer matrix, control input matrix, and process noise covariance matrix; The specific calculation method of the state equation is: ,in, It is expressed as the state equation, It is expressed as the state transfer equation, is represented as process noise; Step B2: Based on the state estimate and state transfer matrix of the previous moment, predict the state at the current moment; at the same time, update the covariance matrix of the state estimate to reflect the uncertainty of the prediction; Step B3: Obtain the observation value at the current moment from the sensor; Step B4: Determine the observation matrix and the observation noise covariance matrix; Step B5: Use Kalman gain to combine the observed value with the predicted value to obtain an updated state estimate; wherein the Kalman gain is a weight used to weigh the predicted value and the observed value, which is calculated based on the prediction covariance matrix and the observation noise covariance matrix; Step B6: Update the covariance matrix of the state estimate to reflect the updated uncertainty; The specific method for calculating the covariance matrix is: ,in, Denoted as the covariance matrix of the updated state estimate, Expressed as the predicted value, Expressed as the Kalman gain, is expressed as an observed value, Expressed as the process noise covariance matrix; Step B7: For each new observation, repeat the prediction and update steps to continuously update the state estimate and covariance matrix; Step B8: The state estimation value after Kalman filtering is the denoised signal.

4. The oil casing micro-crack detection method based on quantum computing and topological data analysis according to claim 3 is characterized in that: In step B3, the method for calculating the observed value is specifically as follows: ,in, is expressed as an observed value, is the observation noise, Expressed as the process noise covariance matrix.

5. The oil casing micro-crack detection method based on quantum computing and topological data analysis according to claim 3 is characterized in that: In step B5, the calculation method of the predicted value is specifically as follows: ,in, Expressed as the predicted value, It is expressed as the state transfer equation, It is represented as the posterior estimate of the state x at time step k−1, that is, it is based on all available information at time k−1; The calculation method of prediction covariance is as follows: ,in, Denoted as the prediction covariance, It is expressed as the state transfer equation, Expressed as the state estimation covariance matrix, Expressed as The transposed matrix of Expressed as the process noise covariance matrix; The calculation method of Kalman gain is as follows: ,in, Expressed as the Kalman gain, Denoted as the prediction covariance, Expressed as the transposed matrix of the process noise covariance matrix C, It is represented as the measurement noise covariance matrix.

6. The oil casing micro-crack detection method based on quantum computing and topological data analysis according to claim 3 is characterized in that: In step B6, the covariance matrix calculation method is specifically as follows: ,in, Denoted as the covariance matrix of the updated state estimate, Expressed as the predicted value, Expressed as the Kalman gain, is expressed as an observed value, Expressed as the process noise covariance matrix; In step B8, the updated covariance matrix is ​​calculated as follows: ,in, is represented as the updated covariance matrix, p is represented as the identity matrix, Expressed as the Kalman gain, Expressed as the process noise covariance matrix, Denoted as the prediction covariance.

7. The oil casing micro-crack detection method based on quantum computing and topological data analysis according to claim 1 is characterized in that: In step A5, the crack value and the crack distance value determine the location of the crack. In the actual detection process, the extracted crack features are input into the trained support vector machine model to achieve real-time recognition of the crack type; and provide detailed recognition results, including the crack type, possible causes and recommended treatment measures; Combined with visualization tools, crack identification reports are generated to facilitate technical personnel to review and make maintenance decisions; new inspection data and identification results are fed back to the feature library to achieve dynamic updating of the feature library.

Images