Pressure steel pipe defect mileage positioning method, device and equipment and medium

Abstract

The invention discloses a pressure steel pipe defect mileage positioning method, device and equipment and a medium, and relates to the technical field of hydropower engineering. The method comprises the following steps: performing weighted fusion on a magnetic flux variation, an accumulated displacement, a temperature correction and a root-mean-square value of a vibration acceleration to obtain multi-source detection data; performing multi-scale wavelet decomposition transformation on the multi-source detection data to obtain a multi-scale wavelet coefficient; performing morphological gradient operation on the multi-scale wavelet coefficient to obtain a multi-scale morphological gradient; fusing the multi-scale wavelet coefficient and the multi-scale morphological gradient to obtain feature enhancement data; and performing defect discrimination through the feature enhancement data, and obtaining the defect mileage according to a space-time correlation analysis method. Through the multi-sensor fusion technology, the multi-scale wavelet-morphological gradient enhanced feature extraction and the optimized mileage positioning algorithm, the positioning error is effectively reduced, and the precision of hydroelectric pressure steel pipe defect mileage positioning is remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of hydropower engineering, and in particular to a method, device, equipment and medium for locating defect mileage of a penstock. Background Art

[0002] Penstocks are core components for energy transmission in hydropower stations. They are subjected to long-term water pressures up to 12 MPa, as well as sediment erosion and chemical corrosion. This can lead to hidden defects such as fatigue cracks and localized corrosion on the pipe walls. According to relevant statistics, the penstock defect detection rate exceeds 70% in hydropower stations operating for more than 10 years, of which hidden cracks account for 45%. Accurately locating these defects is crucial for ensuring the safe operation of hydropower facilities.

[0003] Traditional detection technology primarily relies on magnetic flux detection combined with fixed mileage marker positioning. However, existing magnetic flux detection systems suffer from insufficient baseline stability, with baseline drift of 6-9 μT occurring for every 10°C temperature change. Furthermore, single magnetic flux sensors are susceptible to electromagnetic interference, and mileage sensors are susceptible to mechanical vibration, severely interfering with the accuracy of defect signal recognition. Furthermore, mileage positioning systems based on fixed markers are significantly affected by temperature deformation. When the ambient temperature fluctuates by ±15°C, a positioning error of ±1.5m can accumulate within a standard 200m pipe section, making it difficult to meet the requirements for high-precision defect location. Summary of the Invention

[0004] The present invention provides an optimal method, system and medium for locating the mileage defects of hydroelectric pressure steel pipes, which solve the problem of insufficient detection accuracy and positioning accuracy of existing defect mileage locating methods.

[0005] The present invention is achieved through the following technical solutions:

[0006] A first aspect of the present invention provides a method for locating defect mileage of a penstock, comprising:

[0007] Obtain the magnetic flux change, cumulative displacement, temperature correction, and vibration acceleration detected along the inside of the penstock within the acquisition time window;

[0008] Performing weighted fusion on the magnetic flux variation, cumulative displacement, temperature correction, and root mean square value of vibration acceleration to obtain multi-source detection data;

[0009] Performing a multi-scale wavelet decomposition transform on the multi-source detection data to obtain multi-scale wavelet coefficients;

[0010] Performing a morphological gradient operation on the multi-scale wavelet coefficients to obtain a multi-scale morphological gradient;

[0011] fusing the multi-scale wavelet coefficients and the multi-scale morphological gradient to obtain feature enhanced data;

[0012] Defect identification is performed using the feature-enhanced data, and the defect mileage is located using a spatiotemporal correlation analysis method.

[0013] In view of the fact that magnetic flux detection is affected by temperature and electromagnetic interference, the present invention adopts multi-sensor data fusion including magnetic flux change, cumulative displacement, temperature correction and vibration acceleration to perform defect discrimination and positioning, thereby improving the recognition accuracy of defect signals. When processing defect detection data, a multi-scale wavelet-morphological gradient fusion algorithm is introduced to extract feature enhancement data, improve the signal-to-noise ratio of detection data, effectively reduce positioning error, and significantly improve the accuracy of mileage positioning of hydropower pressure steel pipe defects.

[0014] Furthermore, the multi-source detection data is calculated as:

[0015]

[0016] Among them, L final Represents multi-source detection data; M i represents the change in magnetic flux collected by the i-th magnetic flux sensor within the acquisition time window [k-Δt, k], k represents the end time of the current acquisition time window, Δt represents the length of the current acquisition time window, n represents the number of magnetic flux sensors, α represents the magnetic flux weight coefficient; ΔS represents the cumulative displacement of the laser odometer within the acquisition time window [k-Δt, k], β represents the laser odometer displacement compensation coefficient; T comp represents the temperature correction value output by the temperature and humidity compensation unit, γ represents the temperature correction factor; V avg It represents the RMS value of the vibration acceleration of the three-dimensional vibration monitoring module within the acquisition time window [k-Δt, k], and δ represents the vibration noise suppression factor.

[0017] Furthermore, before performing multi-scale wavelet decomposition transformation on the multi-source detection data, the multi-source detection data is subjected to standardization processing to obtain standardized multi-source detection data;

[0018] Performing a multi-scale wavelet decomposition transform on the standardized multi-source detection data to obtain multi-scale wavelet coefficients;

[0019] The standardized multi-source detection data is calculated as:

[0020]

[0021] Where f(t) represents the standardized multi-source detection data, t represents the acquisition time, t∈[k-Δt,k]; L final Represents multi-source detection data, μ LL is the multi-source detection data of multiple acquisition time windows final The mean of L is the multi-source detection data L final The standard deviation of .

[0022] Furthermore, the feature enhancement data is expressed as:

[0023]

[0024] Among them, F enhanced represents feature enhancement data, W j,k represents the jth layer wavelet coefficient, m represents the number of wavelet decomposition layers, G j (W j,k ) represents the wavelet coefficient W of the jth layer j,k The morphological gradient is obtained by performing the morphological gradient operation.

[0025] Furthermore, the discriminant formula for defect discrimination of the feature enhanced data using a dynamic threshold is:

[0026]

[0027] Among them, DefectFlag = 1 means there is a defect, DefectFlag = 0 means no defect is found; F enhanced represents feature enhancement data, τ(t) is the dynamic threshold,

[0028] τ(t)=μ hist (t)+3σ hist (t)

[0029] Among them, μ hist (t) is the mean of the historical multi-source detection data before time t, σ hist (t) is the standard deviation of the historical multi-source detection data before time t.

[0030] Furthermore, the spatiotemporal correlation analysis method combines the acquisition timestamp and mileage information of the multi-source detection data to determine the defect mileage, which is calculated as:

[0031]

[0032] Among them, L defect Indicates defect mileage, L i To trigger defect judgment, that is, when DefectFlag = 1, the mileage value of the i-th magnetic flux sensor, w i The mileage weight of the i-th magnetic flux sensor, t defect Indicates the time when the defect judgment is triggered, t irepresents the time when the i-th magnetic flux sensor triggers the defect judgment, and ε is a non-negative small constant used to prevent the denominator from being zero.

[0033] Furthermore, before determining the defective mileage, the mileage value L is also included. i Steps to make the correction:

[0034] Using extended Kalman filtering to perform primary correction on the mileage value to obtain an EKF state sequence;

[0035] Clustering the EKF state sequence using fuzzy C-means clustering to obtain cluster centers;

[0036] The cluster center is input into a pre-built BP neural network compensation model to obtain a corrected mileage value.

[0037] A second aspect of the present invention provides a device for locating defect mileage of a penstock, comprising:

[0038] Multi-source data acquisition module, used to obtain the magnetic flux change, cumulative displacement, temperature correction and vibration acceleration detected along the inside of the penstock within the acquisition time window;

[0039] A multi-source data fusion module is used to perform weighted fusion on the magnetic flux change, cumulative displacement, temperature correction and root mean square value of vibration acceleration to obtain multi-source detection data;

[0040] A wavelet transform module is used to perform multi-scale wavelet decomposition transform on the multi-source detection data to obtain multi-scale wavelet coefficients;

[0041] A morphological operation module, configured to perform a morphological gradient operation on the multi-scale wavelet coefficients to obtain a multi-scale morphological gradient;

[0042] A multi-source feature enhancement module is used to fuse the multi-scale wavelet coefficients and the multi-scale morphological gradient to obtain feature enhancement data;

[0043] The defect mileage positioning module is used to identify defects through the feature enhancement data and locate the defect mileage according to the spatiotemporal correlation analysis method.

[0044] The third aspect of the present invention provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method for locating defective mileage of a pressure steel pipe as described in any one of the first aspects of the present invention when executing the computer program.

[0045] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method for locating the defect mileage of a pressure steel pipe as described in any one of the first aspects of the present invention.

[0046] Compared with the prior art, the present invention has the following advantages and beneficial effects: it adopts multi-sensor data fusion including magnetic flux change, cumulative displacement, temperature correction and vibration acceleration to perform defect identification and positioning, and introduces a multi-scale wavelet-morphological gradient fusion algorithm to extract feature enhancement data when processing defect detection data, thereby improving the signal-to-noise ratio of detection data, improving the recognition accuracy of defect signals, effectively reducing positioning errors, and significantly improving the accuracy of mileage positioning of hydropower pressure steel pipe defects; through a three-level error correction mechanism, precise calculations are performed, thereby enhancing the reliability of positioning results, and providing a scientific and accurate basis for the maintenance and management of pressure steel pipes; the data preprocessing and feature extraction method of the present invention reduces the workload of data processing, improves the efficiency of data processing, and can accurately determine the mileage position of defects in a short time. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required for use in the examples. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be considered as limiting the scope. A person of ordinary skill in the art can also derive other relevant drawings based on these drawings without inventive effort. In the drawings:

[0048] Figure 1 This is a flow chart of a method for locating defect mileage of a penstock according to an embodiment of the present invention;

[0049] Figure 2 1 is a schematic diagram of a multi-source sensor arrangement according to an embodiment of the present invention;

[0050] Figure 3 is a defect detection distribution histogram of a penstock according to an embodiment of the present invention;

[0051] Figure 4 The present invention is a schematic diagram of the structure of a device for locating the defect mileage of a penstock according to an embodiment of the present invention. DETAILED DESCRIPTION

[0052] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with examples and drawings. The exemplary embodiments of the present invention and their descriptions are only used to explain the present invention and are not intended to limit the present invention.

[0053] It should be noted that the terms "including" and "having" and any variations thereof in the specification and claims of the present invention and the above-mentioned drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to other steps or units inherent in the device.

[0054] The terms used in various embodiments of the present invention are only used to describe the purpose of specific embodiments and are not intended to limit the various embodiments of the present invention. As used herein, the singular form is intended to also include the plural form, unless the context clearly indicates otherwise. Unless otherwise limited, all terms used here (including technical terms and scientific terms) have the same meaning as those of ordinary skill in the art generally understood by the various embodiments of the present invention. The terms (such as those defined in generally used dictionaries) will be interpreted as having the same meaning as the contextual meaning in the relevant technical field and will not be interpreted as having idealized meaning or too formal meaning, unless clearly defined in various embodiments of the present invention.

[0055] The embodiments of the present invention provide a method, device, equipment and medium for locating defect mileage of a pressure steel pipe, which is suitable for online defect detection and defect mileage positioning of hydropower pressure steel pipes, and is conducive to achieving high-precision and high-efficiency defect mileage positioning.

[0056] See Figure 1 , Figure 1 The figure shows a flow chart of a method for locating defect mileage of a penstock according to the present invention, which includes the following steps.

[0057] S1, obtain the magnetic flux change, cumulative displacement, temperature correction and vibration acceleration detected along the inside of the penstock within the acquisition time window.

[0058] S2, weighted fusion of the magnetic flux change, cumulative displacement, temperature correction and root mean square value of vibration acceleration to obtain multi-source detection data.

[0059] S3, performing multi-scale wavelet decomposition transformation on the multi-source detection data to obtain multi-scale wavelet coefficients.

[0060] S4, performing morphological gradient operation on the multi-scale wavelet coefficients to obtain multi-scale morphological gradients.

[0061] S5, the multi-scale wavelet coefficients and the multi-scale morphological gradient are fused to obtain feature enhanced data.

[0062] S6, perform defect identification through feature enhancement data, and obtain defect mileage based on spatiotemporal correlation analysis method.

[0063] The magnetic flux change is collected by one or more magnetic flux sensors, the cumulative displacement is collected by a laser odometry, the temperature correction is output by the temperature and humidity compensation unit, and the vibration acceleration is collected by the three-dimensional vibration monitoring module. Sensor arrangement: Multiple magnetic flux sensors and laser odometry are arranged on the penstock magnetic flux detector based on parameters such as the length, diameter, and material of the penstock to be inspected. The magnetic flux sensor fits tightly to the surface of the steel pipe and performs detection as the magnetic flux detector moves along the axis of the steel pipe. The magnetic flux sensor's acquisition accuracy should ensure that it can accurately detect even small changes in magnetic flux. The laser odometry sensor is securely mounted in a fixed position on the magnetic flux detector and performs odometry detection as the magnetic flux detector moves along the axis of the steel pipe.

[0064] The process of moving the magnetic flux detector along the axis of the steel pipe to achieve detection is divided into continuous acquisition time windows [k-Δt, k]. The acquisition time window length is Δt, and k represents the end time of the current acquisition time window. Step S1 is to obtain the magnetic flux change, cumulative displacement, temperature correction and vibration acceleration in real time within each acquisition time window. That is, the magnetic flux change includes the magnetic flux change M of one or more magnetic flux sensors within the acquisition time window [k-Δt, k]. i (unit: μT); the cumulative displacement is the cumulative displacement ΔS of the laser odometer in the same acquisition time window [k-Δt, k] (unit: mm); the temperature correction is the temperature correction T output by the temperature and humidity compensation unit in the same acquisition time window [k-Δt, k] comp (Unit: m), satisfying Tcomp=kT·(T current -T ref ), T current is the current collected temperature, T ref =20℃, kT=0.008℃-1; vibration acceleration is the vibration acceleration (unit: m / s) collected by the three-dimensional vibration monitoring module in the same acquisition time window [k-Δt, k] 2 ).

[0065] The cumulative displacement ΔS is calculated by real-time integration of the displacement velocity:

[0066]

[0067] where v(t) is the displacement velocity.

[0068] In step S2, the magnetic flux change, cumulative displacement, temperature correction and vibration acceleration root mean square value are weighted and fused to obtain multi-source detection data, which is expressed as:

[0069]

[0070] Among them, L final Represents multi-source detection data; M i represents the change in magnetic flux collected by the i-th magnetic flux sensor within the acquisition time window [k-Δt, k], k represents the end time of the current acquisition time window, Δt represents the length of the current acquisition time window, n represents the number of magnetic flux sensors, α represents the magnetic flux weight coefficient; ΔS represents the cumulative displacement of the laser odometer within the acquisition time window [k-Δt, k], β represents the laser odometer displacement compensation coefficient; T comp represents the temperature correction value output by the temperature and humidity compensation unit, γ represents the temperature correction factor; V avg It represents the RMS value of the vibration acceleration of the three-dimensional vibration monitoring module within the acquisition time window [k-Δt, k], and δ represents the vibration noise suppression factor.

[0071] Preferably, the weighting coefficient α is 0.58, and after Monte Carlo simulation optimization, the laser mileage displacement compensation coefficient β is 0.25, the temperature correction factor satisfies γ=kT·ΔT, kT=0.008°C-1, and the vibration noise suppression factor δ is 0.05.

[0072] In step S3, the multi-scale wavelet decomposition transform is expressed as:

[0073]

[0074] Among them, W j,k (t) represents the j-th layer coefficient after wavelet decomposition, f(t) represents the multi-source detection data, and k represents the time window.

[0075] In this embodiment, the wavelet decomposition adopts the db6 basis function, and the number of decomposition layers is set to 5, which can be set according to actual needs.

[0076] In one embodiment, before performing wavelet decomposition on the multi-source detection data, the multi-source detection data is normalized to obtain normalized multi-source detection data, which is expressed as:

[0077]

[0078] Where f(t) represents the standardized multi-source detection data, t represents the acquisition time, t∈[k-Δt,k]; L final represents the original multi-source detection data, μ L L is the multi-source detection data of multiple acquisition time windows final The mean of L is the multi-source detection data L final That is, the multi-source detection data input into the wavelet decomposition function adopts the standardized multi-source detection data f(t).

[0079] Specifically, in actual calculation, it is assumed that N sets of multi-source detection data L are acquired in multiple acquisition time windows. final The values are expressed as L final-1 , L fina-2 、……L fina-N . Then the mean and variance are calculated as:

[0080]

[0081] In step S4, morphological gradient operation is further performed on the wavelet decomposition coefficients of each layer to obtain the morphological gradient corresponding to each wavelet coefficient. The morphological gradient operation formula is:

[0082]

[0083] Among them, x is the input of the morphological gradient operation, that is, the wavelet decomposition coefficient W j,k , represents the dilation operation, Represents the erosion operation, the structural element B is a disk-shaped operator with a radius of 3px, and f represents the original function for the morphological operation.

[0084] In step S5, the multi-scale wavelet coefficients and the multi-scale morphological gradient are fused to obtain feature enhancement data, which is expressed as:

[0085]

[0086] Among them, F enhanced represents feature enhancement data, W j,k represents the jth layer wavelet coefficient, m represents the number of wavelet decomposition layers, G j (W j,k ) represents the wavelet coefficient W of the jth layer j,k The morphological gradient is obtained by performing morphological gradient operation. In this embodiment, m is 5, that is, a 5-layer wavelet decomposition is performed.

[0087] This technology improves the signal-to-noise ratio of microcracks (≥0.2mm) by 14dB.

[0088] In one embodiment, a dynamic threshold is used to perform defect discrimination on the feature-enhanced data, and the discriminant formula is as follows:

[0089]

[0090] Among them, DefectFlag = 1 means there is a defect, DefectFlag = 0 means no defect is found; F enhanced represents feature enhancement data, τ(t) is a dynamic threshold, which is generated by historical data statistics.

[0091] τ(t)=μhist (t)+3σ hist (t)

[0092] Among them, μ hist (t) represents the mean of the historical multi-source detection data before time t, σ hist (t) is the standard deviation of the historical multi-source detection data before time t. Compared with the traditional method that mostly uses a fixed threshold method, the present invention improves the detection accuracy through a dynamic threshold.

[0093] The historical multi-source inspection data includes the fusion of magnetic flux change, cumulative displacement, temperature correction, and vibration acceleration data from the current inspection time t during the operation of the penstock. In this implementation, t also represents the collection time, indicating the real-time collection and defect detection of multi-source inspection data and the updating of dynamic thresholds during the operation of the penstock.

[0094] In one embodiment, the dynamic threshold τ(t) is adjusted in real time based on the pipeline's operating status. Based on real-time monitored operating parameters such as pipeline pressure and flow rate, the threshold is modified using a relevant model, taking into account the impact of the pipeline's real-time operating conditions on defect detection.

[0095] Furthermore, when DefectFlag = 1, the timestamp and spatial location information of multi-source data are combined to calculate the defect mileage through the spatiotemporal correlation equation:

[0096]

[0097] Among them, L defect Indicates defect mileage, L i To trigger defect judgment, that is, when DefectFlag = 1, the mileage value of the i-th magnetic flux sensor, w i The mileage weight of the i-th magnetic flux sensor, t defect Indicates the time when the defect judgment is triggered, t i represents the time when the i-th magnetic flux sensor triggers the defect judgment, and ε is a non-negative small constant used to prevent the denominator from being 0. In this embodiment, ε is 10 -6 .

[0098] The spatial positioning accuracy is improved by introducing the submillimeter displacement data of the laser odometry (ΔS resolution is ±0.1mm) into the space-time correlation equation.

[0099] In one embodiment, a weighted fusion is performed on the magnetic flux change of a single sensor, the cumulative displacement, the temperature correction, and the root mean square value of the vibration acceleration to obtain multi-source detection data, and a judgment is made on whether a single sensor triggers a defect judgment. In combination with the situation of adjacent sensors, multi-sensor collaborative defect judgment is performed. For example, at least three adjacent sensors must trigger DefectFlag = 1 at the same time to determine it as a valid defect.

[0100] In this implementation, if the magnetic flux changes of multiple adjacent sensors in a certain area show an abnormal trend before data fusion, even if the magnetic flux change of a certain sensor alone does not account for a prominent proportion in the fused data, after combining the situation of adjacent sensors, it can be comprehensively judged that the feature enhancement data corresponding to the sensors in this area is more likely to meet the defect judgment conditions, and then determine whether the single sensor is triggered.

[0101] In one embodiment, before locating the defect mileage, the mileage value L of the magnetic flux sensor is also included. i Steps to make the correction:

[0102] Primary correction: Use extended Kalman filter to perform primary correction on the mileage value to obtain the EKF state sequence;

[0103] Secondary correction: Fuzzy C-means clustering is used to cluster the EKF state sequence and obtain the cluster center;

[0104] Final correction: Input the cluster center into the pre-built BP neural network compensation model to obtain the corrected mileage data.

[0105] In this embodiment, the mileage value of each magnetic flux sensor can be obtained according to the position parameter information of the magnetic flux sensor, such as the layout position and layout interval, and the accumulated mileage value of the laser sensor.

[0106] Among them, the extended Kalman filter correction model is shown in the following formula:

[0107] Equation of state:

[0108] x k =f(x k-1 ,u k )+w k

[0109] Among them, x k represents the kth state estimate of the EKF state sequence, x k-1 Indicates the state estimate at the previous moment, u k represents the control input of the system, which is the external input information related to positioning, such as the displacement speed measured by the laser odometer, used to assist in updating the system status; w krepresents process noise, which reflects the uncertainty of the system model. In actual pressure pipe inspection, it will be affected by factors such as electromagnetic interference in the environment and measurement errors of the equipment itself, resulting in unpredictable fluctuations in the system state. k It is a quantitative representation of these uncertainties; f(*,*) represents the state transfer function, which is used to describe how the system changes from the current state (represented by the state estimate at the previous moment) and the control input to the state at the next moment. It comprehensively considers the update rules of the mileage position state based on factors such as the sensor movement speed and time interval.

[0110] Observation equation:

[0111] z k =h(x k )+v k

[0112] Among them, z k Indicates the observation value at the kth moment. In the scenario of penstock defect mileage positioning, it can be the measurement data collected by multiple source sensors such as magnetic flux sensors and laser odometry at the kth moment and obtained after certain processing, which is used to observe and verify the system status; k Represents observation noise. Due to the accuracy limitation of the sensor itself and the interference of the external environment, there is a deviation between the actual observation value and the true value. k It is used to represent the noise generated in the observation process, reflecting the uncertainty of the observation data; h(*) represents the observation function, which describes the system state (using x k denoted by) and the observed value (denoted by z k The relationship between the two is determined according to the measurement principles of different sensors and data processing methods, and the system state information is mapped into the observable physical quantity space.

[0113] Covariance update:

[0114] P k| k =(IK k H k )P k| k-1

[0115] Among them, P k|k It represents the state estimation covariance based on all observation data at time k and before, which is used to measure the uncertainty of the current state estimate. The smaller the value, the more accurate the state estimate. I is the unit matrix, which is used to maintain the dimensional consistency of the original covariance matrix in the covariance update calculation to ensure the rationality of the calculation. K k is the Kalman gain, which predicts the covariance (P k|k-1 ), the observation matrix (Hk ) and the observation noise covariance, which is used to balance the weights of the predicted value and the observed value in the state update and determines the degree of correction of the state estimate by the observation data; H k It is the observation matrix, which establishes the linear relationship between the system state space and the observation space, converts the covariance of the state variables into the covariance of the observation variables, and reflects the influence of the system state on the observation value; k|k-1 It represents the predicted covariance of the state at time k based on the observation data at time k-1 at time k, which is the prediction of the uncertainty of the state estimation before the observation data at time k is obtained.

[0116] The input of the extended Kalman filter correction model is feature enhancement data, and the output is the EKF state sequence: {x1,x2,...,x n}.

[0117] The fuzzy C-means clustering model is expressed as:

[0118]

[0119] Where m = 2.5, and the membership update formula is:

[0120]

[0121] The BP neural network compensation model is expressed as:

[0122] y=σ(W3·σ(W2·σ(W1x+b1)+b2))

[0123] Where σ is the ReLU activation function. The model input data is the cluster center v j With historical data D hist The training set contains 2000 sets of historical detection data.

[0124] The feature enhancement data after the three-level correction is expressed as:

[0125] L′ final =BPNN(FCM(EKF(L raw )))

[0126] Among them, L′ final The corrected mileage data is shown in the table. raw represents the original mileage data, EKF(·) represents the extended Kalman filter, FCM(·) represents the fuzzy C-means clustering model, and BPNN(·) represents the BP neural network compensation model.

[0127] Take the defect mileage detection of a large hydropower station as an example. The diameter of the penstock is 8.5m and the pressure is 10Mpa. The multi-source sensor arrangement is shown in Figure 2As shown, a row of eight magnetic flux sensors (0.1μT accuracy, 0.5m spacing) and two laser odometry sensors (±1mm / 10m resolution) are arranged on the penstock magnetic flux detector. A temperature and humidity compensation unit and a three-dimensional vibration monitoring module are also located on the magnetic flux detector. The temperature and humidity compensation unit has an accuracy of ±0.1°C, and the three-dimensional vibration monitoring module has a sampling frequency of 1kHz. The magnetic flux sensors are tightly attached to the steel pipe surface, and detection is achieved by moving the magnetic flux detector along the steel pipe axis. The magnetic flux sensors ensure accurate detection of even small changes in magnetic flux. The laser odometry sensors are securely mounted on the magnetic flux detector, and detection is achieved by moving the magnetic flux detector along the steel pipe axis.

[0128] The data collected by the sensor is transmitted wirelessly to the data acquisition and transmission module. The data acquisition and transmission module uses a high-speed data acquisition chip to collect data in real time at a sampling frequency of 2000 times per second. In order to ensure the reliability of data transmission, redundant transmission technology is adopted, that is, data is transmitted through two different communication lines at the same time. Data encryption technology is used to encrypt the transmitted data to prevent the data from being tampered with or stolen during the transmission process. At the same time, the data transmission protocol is optimized to reduce data transmission delays. After the data processing center receives the collected data, it first pre-processes the data. For the magnetic flux data, the traditional technology and the above-mentioned technology of the present invention (dynamic fusion of multi-source data, defect feature enhancement technology, and three-level error correction mechanism for precise calculation) are used for calculation. The detection results are shown in Table 1, and the defect detection distribution histogram is shown in Table 1. Figure 3 shown.

[0129] Table 1

[0130] index Traditional technology The present invention Positioning error (m) ±1.2 ±0.08 Microcrack detection rate (%) 68 95 Detection efficiency (m / h) 25 150

[0131] According to the results in Table 1 and Figure 3 It can be seen that the present invention significantly improves the positioning accuracy, effectively reduces the positioning error, and significantly improves the accuracy of mileage positioning of hydropower pressure steel pipe defects through multi-sensor fusion technology, advanced data preprocessing and feature extraction methods, and optimized mileage positioning algorithm.

[0132] An embodiment of the present invention further provides a device for locating defective mileage of a penstock, comprising:

[0133] Multi-source data acquisition module, used to obtain the magnetic flux change, cumulative displacement, temperature correction and vibration acceleration detected along the inside of the penstock within the acquisition time window;

[0134] Multi-source data fusion module, used to perform weighted fusion of magnetic flux change, cumulative displacement, temperature correction and vibration acceleration root mean square value to obtain multi-source detection data;

[0135] Wavelet transform module, used to perform multi-scale wavelet decomposition transform on multi-source detection data to obtain multi-scale wavelet coefficients;

[0136] The morphological operation module is used to perform morphological gradient operations on multi-scale wavelet coefficients to obtain multi-scale morphological gradients;

[0137] Multi-source feature enhancement module, used to fuse multi-scale wavelet coefficients and multi-scale morphological gradients to obtain feature enhancement data;

[0138] The defect mileage location module is used to identify defects through feature enhancement data and locate the defect mileage based on the spatiotemporal correlation analysis method.

[0139] Furthermore, a standardization module is included, which is used to perform standardization processing on the multi-source detection data before performing multi-scale wavelet decomposition transformation on the multi-source detection data to obtain standardized multi-source detection data.

[0140] Furthermore, a correction module is included for performing three-level correction on the original mileage data before performing defect identification through feature enhancement data, including:

[0141] The extended Kalman filter is used to perform primary correction on the original mileage data to obtain the EKF state sequence;

[0142] Fuzzy C-means clustering is used to cluster the EKF state sequence and obtain the cluster center;

[0143] The cluster centers are input into the pre-built BP neural network compensation model to obtain the corrected mileage data.

[0144] Furthermore, the system includes a display module that visually displays the calculated defect mileage location results, such as a map showing the specific location of the defect on the penstock and a chart displaying defect characteristics. Furthermore, the system conducts in-depth analysis and evaluation of the location results, formulating appropriate repair and management plans based on the severity and development trends of the defects.

[0145] An embodiment of the present invention further provides an electronic device comprising a processor and a memory, where the processor may be one or more. The memory, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules. The processor executes the software programs, instructions, and modules stored in the memory to perform various functional applications and data processing of the electronic device, thereby implementing the penstock defect mileage location method according to any of the aforementioned embodiments of the present invention.

[0146] The memory may primarily include a program storage area and a data storage area. The program storage area may store an operating system and at least one application required for a function; the data storage area may store data generated based on the use of the terminal, etc. Furthermore, the memory may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state memory device. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories may be connected to the electronic device via a network. Examples of the aforementioned networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0147] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for locating the defect mileage of a pressure steel pipe according to any embodiment of the present invention.

[0148] The computer storage medium of the embodiment of the present invention can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium can be any tangible medium that contains or stores a program, which can be used by an instruction execution system, device or device or used in combination with it.

[0149] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0150] An embodiment of the present invention further provides a computer program product. When the computer program product is run on a computer, the computer is enabled to execute the method for locating the defect mileage of a pressure steel pipe according to any of the above embodiments of the present invention.

[0151] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for locating defect mileage of a penstock, characterized in that: include: Obtain the magnetic flux change, cumulative displacement, temperature correction, and vibration acceleration detected along the inside of the penstock within the acquisition time window; Performing weighted fusion on the magnetic flux variation, cumulative displacement, temperature correction, and root mean square value of vibration acceleration to obtain multi-source detection data; Performing multi-scale wavelet decomposition transformation on the multi-source detection data to obtain multi-scale wavelet coefficients; Performing a morphological gradient operation on the multi-scale wavelet coefficients to obtain a multi-scale morphological gradient; fusing the multi-scale wavelet coefficients and the multi-scale morphological gradient to obtain feature enhanced data; Defect identification is performed using the feature-enhanced data, and the defect mileage is located using a spatiotemporal correlation analysis method.

2. The method for locating defective mileage of a penstock according to claim 1, characterized in that: The multi-source detection data is calculated as: Among them, L final Represents multi-source detection data; M i represents the change in magnetic flux collected by the i-th magnetic flux sensor within the acquisition time window [k-Δt, k], k represents the end time of the current acquisition time window, Δt represents the length of the current acquisition time window, n represents the number of magnetic flux sensors, α represents the magnetic flux weight coefficient; ΔS represents the cumulative displacement of the laser odometer within the acquisition time window [k-Δt, k], β represents the laser odometer displacement compensation coefficient; T comp represents the temperature correction value output by the temperature and humidity compensation unit, γ represents the temperature correction factor; V avg It represents the RMS value of the vibration acceleration of the three-dimensional vibration monitoring module within the acquisition time window [k-Δt, k], and δ represents the vibration noise suppression factor.

3. The method for locating defect mileage of a penstock according to claim 2, characterized in that: Before performing multi-scale wavelet decomposition transformation on the multi-source detection data, performing standardization processing on the multi-source detection data to obtain standardized multi-source detection data; Performing a multi-scale wavelet decomposition transform on the standardized multi-source detection data to obtain multi-scale wavelet coefficients; The standardized multi-source detection data is calculated as: Where f(t) represents the standardized multi-source detection data, t represents the acquisition time, t∈[k-Δt,k]; L final Represents multi-source detection data, μ L L is the multi-source detection data of multiple acquisition time windows final The mean of L is the multi-source detection data L final The standard deviation of .

4. The method for locating defective mileage of a penstock according to claim 1, characterized in that: The feature enhancement data is expressed as: Among them, F enhanced represents feature enhancement data, W j,k represents the jth layer wavelet coefficient, m represents the number of wavelet decomposition layers, G j (W j,k ) represents the wavelet coefficient W of the jth layer j,k The morphological gradient is obtained by performing the morphological gradient operation.

5. The method for locating defect mileage of a penstock according to any one of claims 1 to 4, characterized in that: The discriminant formula for defect discrimination of the feature enhanced data using a dynamic threshold is: Among them, DefectFlag = 1 means there is a defect, DefectFlag = 0 means no defect is found; F enhanced represents feature enhancement data, τ(t) is the dynamic threshold, τ(t)=μ hist (t)+3σ hist (t) Among them, μ hist (t) is the mean of the historical multi-source detection data before time t, σ hist (t) is the standard deviation of the historical multi-source detection data before time t.

6. The method for locating defect mileage of a penstock according to claim 5, characterized in that: The spatiotemporal correlation analysis method combines the acquisition timestamp and mileage information of the multi-source detection data to determine the defect mileage, which is calculated as: Among them, L defect Indicates defect mileage, L i To trigger defect judgment, that is, when DefectFlag = 1, the mileage value of the i-th magnetic flux sensor, w i The mileage weight of the i-th magnetic flux sensor, t defect Indicates the time when the defect judgment is triggered, t i represents the time when the i-th magnetic flux sensor triggers the defect judgment, and ε is a non-negative small constant used to prevent the denominator from being zero.

7. The method for locating defect mileage of a penstock according to claim 6, characterized in that: Before determining the defect mileage, the mileage value L is also included. i Steps to correct the row: Using extended Kalman filtering to perform primary correction on the mileage value to obtain an EKF state sequence; Clustering the EKF state sequence using fuzzy C-means clustering to obtain cluster centers; The cluster center is input into a pre-built BP neural network compensation model to obtain a corrected mileage value.

8. A penstock defect mileage positioning device, characterized in that: include: Multi-source data acquisition module, used to obtain the magnetic flux change, cumulative displacement, temperature correction and vibration acceleration detected along the inside of the penstock within the acquisition time window; A multi-source data fusion module is used to perform weighted fusion on the magnetic flux change, cumulative displacement, temperature correction and root mean square value of vibration acceleration to obtain multi-source detection data; A wavelet transform module is used to perform multi-scale wavelet decomposition transform on the multi-source detection data to obtain multi-scale wavelet coefficients; A morphological operation module, configured to perform a morphological gradient operation on the multi-scale wavelet coefficients to obtain a multi-scale morphological gradient; A multi-source feature enhancement module is used to fuse the multi-scale wavelet coefficients and the multi-scale morphological gradient to obtain feature enhancement data; The defect mileage positioning module is used to identify defects through the feature enhancement data and locate the defect mileage according to the spatiotemporal correlation analysis method.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method for locating defect mileage of a pressure steel pipe according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for locating defect mileage of a pressure steel pipe according to any one of claims 1 to 7 is implemented.

Images