Guided wave pulse echo detection method for corrosion damage of ground flat steel in different soil environments

Abstract

The invention discloses a guided wave pulse echo detection method for corrosion damage of grounded flat steel under different soil environments, comprising the steps of: S1 obtaining guided wave pulse echo signals of grounded flat steel samples with different corrosion degrees under different soil environments; S2 collecting guided wave pulse echo signals in the grounded flat steel to be tested; S3 extracting the characteristics of the guided wave pulse echo signals of the grounded flat steel to be tested and the pulse guided wave echo signals of each grounded flat steel sample respectively, and obtaining a grounded flat steel corrosion defect echo feature set; S4 fuzzy clustering the obtained grounded flat steel corrosion defect echo feature set through prior experience, and then using a fuzzy C-means algorithm to iteratively update the clustering center and the membership matrix, so that the objective function converges, and the final clustering center and the membership matrix are obtained, and the cluster to which the grounded flat steel to be tested belongs is determined according to the final clustering center and the membership matrix, thereby determining the steel corrosion status of the grounded flat steel to be tested and its surrounding soil environment. It is suitable for power systems.

Description

Technical Field

[0001] The invention relates to the technical field of guided wave pulse echo detection, in particular to a guided wave pulse echo detection method for corrosion damage of ground flat steel under different soil environments. Background Art

[0002] At present, substations are an indispensable part of the power system and also play a key role in power transmission and distribution. Electric energy is transmitted to substations through high-voltage transmission lines, where high voltage is converted into low voltage for users. Among them, grounding flat steel is the key to the long-term stable and safe operation of substations, and is the safety guarantee for electrical equipment and workers in substations.

[0003] However, since the grounding steel bars of the substation are buried deep underground for a long time, their grounding steel network may corrode, threatening the safety of workers and electrical equipment. Since it is not easy to visually inspect the corrosion, the only effective inspection method at this stage is to dig and inspect the grounding steel network on the ground, which is a time-consuming and expensive operation.

[0004] Therefore, it is necessary to maintain the integrity of the grounding flat steel to ensure the safe and stable operation of the substation. However, how to judge whether the current corrosion level of the power station grounding flat steel can still provide sufficient protection for the power station equipment and personnel has become a major problem. Summary of the invention

[0005] The purpose of the present invention is to provide a guided wave pulse echo detection method for corrosion damage of ground flat steel in different soil environments. The method can be implemented without power outage and large-scale excavation of ground flat steel, and can detect corrosion damage of underground ground flat steel.

[0006] The invention discloses a guided wave pulse echo detection method for corrosion damage of ground flat steel under different soil environments, comprising the following steps:

[0007] S1. Obtaining guided wave pulse echo signals of grounded flat steel samples with different corrosion degrees under different soil environments;

[0008] S2, collecting the waveguide pulse echo signal in the grounded flat steel to be tested;

[0009] S3, respectively extracting the characteristics of the guided wave pulse echo signal of the grounded flat steel to be tested and the pulse guided wave echo signal of each grounded flat steel sample to obtain a grounded flat steel corrosion defect echo feature set;

[0010] S4. Fuzzy clustering is performed on the obtained ground flat steel corrosion defect echo feature set based on prior experience, and then the fuzzy C-means algorithm is used to iteratively update the clustering center and the membership matrix to converge the objective function and obtain the final clustering center and the membership matrix. According to the final clustering center and the membership matrix, it is determined which cluster the ground flat steel to be tested belongs to, thereby determining the steel corrosion condition of the ground flat steel to be tested and its surrounding soil environment.

[0011] Preferably, in steps S1 and S2:

[0012] The CRAS sensor is installed on the grounded flat steel, and the guided wave pulse echo signal of the CRAS sensor is collected by the data acquisition equipment.

[0013] Preferably, in step S3:

[0014] For the grounded flat steel guided wave pulse echo signal, it is divided into 10 parts to provide a resolution of 30 cm. The features extracted from each part are the peak value, amplitude, entropy value, attenuation, count and normalized distance from the sensor to the end of each part. The grounded flat steel corrosion defect echo feature set M = {M i ,i=1,2,…,n}, where M i It is the n-dimensional feature vector formed by feature extraction of the ground flat steel corrosion defect echo.

[0015] Preferably, in step S4:

[0016] According to experience, fuzzy clustering is performed on the echo feature set M of the ground flat steel corrosion defect to obtain C clusters C1, C2…C c ,U=(u ij ) represents the membership matrix, which is used to indicate that the sample M i and its subset C j Affiliation; ij The value of is randomly selected from [0,1] and should satisfy:

[0017]

[0018] Preferably, in step S4:

[0019] Setting the objective function of the non-similarity index:

[0020]

[0021] Where J represents the sum of squared distances between instance data and cluster centers; d ij The distance between the i-th data point and the j-th cluster center is a distance metric function, specifically using the Euclidean distance d ij =||M i -Pj ||, where P is the clustering center of each category.

[0022] Preferably, in step S4:

[0023] Make J(U,P) reach the minimum value and construct a new objective function:

[0024]

[0025] where λ j , j = 1, ..., n is the Lagrange multiplier of the n constraint formulas in formula (1), x is used to determine the coefficient of the fuzzy level of the membership matrix U, which is usually also called the smoothing factor. The fuzzy level of U is positively correlated with the value of x, and P is the cluster center of each category, which is calculated by the following formula:

[0026]

[0027] Preferably, each sample data is selected as the initial clustering center P (0) , the objective function value is calculated by formula (3) make That is, when the difference between the two iterations is less than the specified threshold ε, the iteration is terminated. Otherwise, continue to use equations (4) and (5) to update the cluster center and the membership matrix U, iteratively calculate the objective function until the above conditions are met, and output the final membership matrix U and center matrix P

[0028]

[0029] Preferably, the corresponding value of the flat steel data to be tested in the output membership matrix U is used to determine which cluster in the central matrix P it belongs to, that is, which cluster the grounded flat steel to be tested belongs to, thereby determining the steel corrosion condition of the grounded flat steel to be tested and its surrounding soil environment.

[0030] Preferably, the data acquisition device includes a microcontroller having a built-in analog-to-digital converter (ADC) and a digital-to-analog converter (DAC). The CRAS sensor is connected to one side of the grounded flat steel and includes two piezoelectric sheets. The microcontroller is programmed to use its DAC to send an electrical pulse to one of the piezoelectric sheets, and the piezoelectric sheet generates an acoustic pulse in the transducer. At the same time, the microcontroller uses its built-in ADC to record the voltage change in the other piezoelectric sheet at a sampling frequency of 1 MHz and saves it to a memory card. Finally, MATLAB is used to post-process the signal to analyze the recorded signal.

[0031] The beneficial effects of the guided wave pulse echo detection method for corrosion damage of ground flat steel under different soil environments disclosed by the present invention are:

[0032] 1. It can detect the corrosion damage of underground grounding flat steel without power outage and large-scale excavation of grounding flat steel;

[0033] 2. Ability to determine the corrosion damage of ground flat steel in a timely manner with a resolution of 30 cm;

[0034] 3. Be able to determine the corrosion condition of the grounding flat steel and whether it can continue to provide adequate protection for equipment and personnel. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a principle block diagram of a guided wave pulse echo detection method for corrosion damage of ground flat steel under different soil environments of the present invention;

[0036] Figure 2 This is a schematic diagram of the detection method. DETAILED DESCRIPTION

[0037] The present invention will be further described and illustrated below in conjunction with specific embodiments and accompanying drawings:

[0038] See also Figure 1-2 The present invention discloses a guided wave pulse echo detection method for corrosion damage of grounded flat steel under different soil environments. The method comprises the following steps: obtaining guided wave pulse echo signals of grounded flat steel under different corrosion degrees under different soil environments; extracting features of the guided wave pulse echo signals of the grounded flat steel; using a fuzzy C-means algorithm to autonomously discover patterns between the features of the guided wave pulse echo signals, and clustering them into clusters with different corrosion degrees under different soil environments, thereby realizing the detection of corrosion damage of the grounded flat steel and the soil environment surrounding the flat steel.

[0039] The specific steps are as follows:

[0040] Step 1) As a technical solution of the present invention, a data acquisition device or other instrument is used to obtain a guided wave pulse echo signal of a grounded flat steel with different corrosion degrees in different soil environments;

[0041] Step 2) collecting the waveguide pulse echo signal in the ground flat steel to be tested;

[0042] Step 3) extracting the characteristics of the pulse guided wave echo signals of the grounded flat steel to be tested and each grounded flat steel sample respectively;

[0043] Step 4) For the pulse guided wave echo signal characteristics of the ground flat steel and each ground flat steel sample obtained in step 3), fuzzy clustering is performed based on prior experience, and then the fuzzy C-means algorithm is used to cluster them into clusters with different corrosion degrees in each soil environment, thereby realizing the judgment of the corrosion status of the ground flat steel and its surrounding soil environment.

[0044] Specifically in steps 1) and 2):

[0045] The equipment and materials required for the generation and collection of the guided wave pulse echo signal of the grounded flat steel include computers, CRAS sensors, data acquisition equipment, memory cards, acoustic coupling agents, and grounded flat steels. For the soil environment, the experiment uses soils of different densities and humidity to cover the grounded flat steel. The specific structure and working principle of the CRAS sensor can be found in the literature Durham, N.2018. "Modelling and testing of a piezoelectric sensorfor defect detection in steel rods using acoustic guided waves," Masterthesis, University of Manitoba, Winnipeg, MB, Canada.

[0046] Specifically in step 3):

[0047] For each grounded flat steel guided wave pulse echo signal, it is divided into 10 parts to provide a resolution of 30 cm. The features extracted from each part are the peak value, amplitude, entropy value, attenuation (each part relative to the first part), count (threshold set to 0.75) and normalized distance from the sensor to the end of each part.

[0048] The clustering method in step 4) is specific as follows:

[0049] After the pulse echo signals obtained by testing the grounded flat steel and each grounded flat steel sample are subjected to feature extraction in the manner determined in step 3), a grounded flat steel corrosion defect echo feature set M = {M i ,i=1,2,…,n}, where M i is the n-dimensional feature vector formed by the feature extraction of the ground flat steel corrosion defect echo; according to experience, fuzzy clustering is performed on M to obtain C clusters C1, C2…C c ,U=(u ij ) represents the membership matrix, which is used to indicate that the sample M i and its subset C j Affiliation; ij The value of is randomly selected from [0,1] and should satisfy:

[0050]

[0051] Define the objective function of the non-similarity index:

[0052]

[0053] Where J represents the sum of squared distances between instance data and cluster centers; d ij The distance function between the i-th data point and the j-th cluster center is a distance measurement function. Depending on the distribution of the cluster centers, the type of function selected may be different. In general, the Euclidean distance d is used. ij =||M i -P j ||.

[0054] In order to meet the necessary conditions for J(U,P) to reach the minimum value, a new objective function is constructed:

[0055]

[0056] where λ j ,j=1,…,n are the Lagrange multipliers of the n constraints in equation (1). x: coefficient used to determine the fuzzy level of the membership matrix U, usually also called the smoothing factor. The fuzzy level of U is positively correlated with the value of x. P is the cluster center of each category, which is calculated by the following formula:

[0057]

[0058] Select each sample data as the initial clustering center P (0) , the objective function value is calculated by formula (3) If it makes It only needs to meet one of the following conditions: when the difference between the values ​​of two iterations is less than a specified threshold ε (generally ε < 0.01 for accuracy), the iteration is terminated; otherwise, continue to use equations (4) and (5) to update the cluster center and the membership matrix U, iteratively calculate the objective function until the above conditions are met, and output the final membership matrix U and center matrix P.

[0059]

[0060] By obtaining the output membership matrix and center matrix, the clustering of the flat steel data to be tested is achieved, the pulse echo signals of each flat steel to be tested are classified, and the detection of the flat steel is completed. (Since the sum of the membership of each sample to all clusters is 1, if the membership of a sample to a cluster is 1, it means that the sample completely belongs to the cluster. That is, when the membership value is larger, it means that the data point is closer to the cluster center; when the membership value is smaller, it means that the data point is farther away from the cluster center.)

[0061] The data acquisition device has a Teensy 3.6 microcontroller with built-in analog-to-digital (ADC) and digital-to-analog (DAC) converters. The sensor is connected to one side of the grounded flat steel and includes two piezoelectric plates. The microcontroller is programmed to use its DAC to send electrical pulses to one of the piezoelectric plates, which generates an acoustic pulse in the transducer; at the same time, the microcontroller uses its built-in ADC to record the voltage changes in the other piezoelectric plate at a sampling frequency of 1MHz and save it to a memory card. Finally, MATLAB is used to post-process the signal to analyze the recorded signal.

[0062] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

[0063] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention, rather than to limit the scope of protection of the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the essence and scope of the technical solution of the present invention.

Claims

1. A guided wave pulse echo detection method for corrosion damage of ground flat steel under different soil environments, characterized in that: The steps include: S1. Obtaining guided wave pulse echo signals of grounded flat steel samples with different corrosion degrees under different soil environments; S2, collecting the waveguide pulse echo signal in the grounded flat steel to be tested; S3, respectively extracting the characteristics of the guided wave pulse echo signal of the grounded flat steel to be tested and the pulse guided wave echo signal of each grounded flat steel sample to obtain a grounded flat steel corrosion defect echo feature set; S4. Fuzzy clustering is performed on the obtained ground flat steel corrosion defect echo feature set through prior experience, and then the fuzzy C-means algorithm is used to iteratively update the clustering center and the membership matrix to converge the objective function, and the final clustering center and the membership matrix are obtained. According to the final clustering center and the membership matrix, it is determined to which cluster the ground flat steel to be tested belongs, thereby determining the steel corrosion condition of the ground flat steel to be tested and its surrounding soil environment; In steps S1 and S2: The CRAS sensor is installed on the grounded flat steel, and the guided wave pulse echo signal of the CRAS sensor is collected by the data acquisition equipment; In step S3: For the grounded flat steel guided wave pulse echo signal, it is divided into 10 parts to provide a resolution of 30 cm. The features extracted from each part are the peak value, amplitude, entropy value, attenuation, count and normalized distance from the sensor to the end of each part. The grounded flat steel corrosion defect echo feature set is obtained through the extracted features. ,in It is the n-dimensional feature vector formed by the feature extraction of the ground flat steel corrosion defect echo; In step S4: Based on the experience of ground flat steel corrosion defect echo feature set Perform fuzzy clustering and get Cluster , Represents the membership matrix, which is used to indicate the sample and its subset affiliation; The value of is randomly selected from [0,1] and should satisfy: （1）； In step S4: Setting the objective function of the non-similarity index: （2）； in, Represents the sum of squared distances between instance data and cluster centers; Used to determine the membership matrix The coefficient of the blur level, Representative Data points and The cluster center is a distance metric function, specifically using the Euclidean distance ,in, It is the gathering center for various categories.

2. The guided wave pulse echo detection method for corrosion damage of ground flat steel under different soil environments according to claim 1 is characterized in that: According to the corresponding value of the flat steel data to be tested in the output membership matrix U, it is determined which cluster in the cluster center matrix P it belongs to, that is, which cluster the grounded flat steel to be tested belongs to, thereby determining the steel corrosion condition of the grounded flat steel to be tested and its surrounding soil environment.

3. The guided wave pulse echo detection method for corrosion damage of ground flat steel under different soil environments according to claim 2 is characterized in that: The data acquisition device includes a microcontroller with a built-in analog-to-digital converter (ADC) and a digital-to-analog converter (DAC). The CRAS sensor is connected to one side of a grounded flat steel bar and includes two piezoelectric sheets. The microcontroller is programmed to use its DAC to send an electrical pulse to one of the piezoelectric sheets, and the piezoelectric sheet generates an acoustic pulse in the transducer. At the same time, the microcontroller uses its built-in ADC to record the voltage change in the other piezoelectric sheet at a sampling frequency of 1 MHz and saves it to a memory card. Finally, MATLAB is used to post-process the signal to analyze the recorded signal.

Images