Transmission tower bolt looseness detection and positioning method and system based on fractional Fourier transform

Abstract

The invention relates to the technical field of electric power facility maintenance, in particular to a power transmission tower bolt looseness detecting and positioning method and system based on fractional Fourier transform. The method comprises the following steps: an off-line modeling stage: aiming at a preprocessed original vibration signal, obtaining an optimal order for representing bolt looseness in a connection area based on fractional Fourier transform; establishing a plurality of off-line models by using the sample set based on a least square support vector machine of a quantization density target, so as to establish an off-line training model # imgabs1 # of the ith connection region under the optimal order # imgabs0 # based on the support vector set; and an online detection stage: based on the corresponding offline model # imgabs2, obtaining a final detection result, outputting two states of bolt loosening or normal, and if the output result of the # imgabs3 model is loosening, positioning the position of the loosened bolt in the # imgabs4 connection area. According to the invention, the precision and efficiency of bolt looseness detection are improved, the anti-interference capability is enhanced, and a powerful guarantee is provided for safe operation of a power transmission tower.

Description

Technical Field

[0001] The present invention relates to the technical field of power facility maintenance, and in particular to a method and system for detecting and locating loose bolts of a transmission tower based on fractional-order Fourier transform. Background Art

[0002] Transmission towers are an important part of the power transmission system, and their structural stability and safety are directly related to the reliable operation of the power system. Bolt connection is the main connection method in the tower structure. However, after long-term exposure to complex and changeable natural environments, bolts are prone to loosening due to factors such as vibration and corrosion, which in turn affects the overall stability of the tower. Traditional bolt loosening detection methods mostly rely on manual inspections, which are not only inefficient, but also difficult to accurately determine the location and degree of loosening. Therefore, it is particularly important to develop an efficient and accurate bolt loosening detection and positioning method.

[0003] The prior art discloses the following technical solutions: CN118349807A, CN118332506A and CN118961033A all disclose methods for monitoring the loosening of transmission tower bolts, but all of them have certain problems, specifically: In the CN118349807A proposal, an online monitoring method for bolt loosening based on finite element and deep learning is proposed. The innovation of this method lies in the modeling method. The implementation method is to install an online monitoring device on the pole tower, and realize bolt loosening monitoring by regularly collecting vibration signals and uploading them to the control center. If the bolts on the pole tower are loose, the operation and maintenance personnel still need to climb the pole to find the specific location of the loose bolts.

[0004] In the CN118332506A solution, a method and related device for bolt loosening positioning of power transmission pole towers based on PSO-VMD analysis is proposed. In the positioning algorithm, this technology proposes a sensor network for building the entire pole tower by arranging multiple sensors, and the bolt loosening positioning is realized based on the PSO-VMD algorithm. In the specific implementation, this method still requires pole climbing operation and sensor network arrangement.

[0005] In CN118961033A, a method for bolt loose detection of power transmission pole tower based on airborne soundprint detection technology is proposed. This method can avoid manual pole climbing, but the method of bolt loose detection is achieved through aircraft, sound pickup sensors and soundprint analysis technology, which is completely different from the technical means used in this application. Summary of the invention

[0006] Object of the Invention: In response to the problem of inaccurate position of loose bolts in the prior art, the present invention provides a method for detecting and positioning the transmission pole tower bolt based on fractional Fourier transform. The present invention also discloses a system and device for detecting and positioning the transmission pole tower bolt based on fractional Fourier transform. Technical Solution

[0007] First, the offline modeling stage: offline modeling of the bolt connection area of ​​the transmission pole tower of a specific model is established, and the corresponding relationship between the vibration signals under different orders and the type, plane position, and number of loose bolts, and the ratio of loose bolts is defined; Determine the total number of connection areas N of the bottom-up of the tower, place vibration sensors near the four feet of the current connection area, and excite them at a certain distance from the current connection area, thereby collecting the original vibration signals when the ratio of different bolts loosening, and pre-processing the collected original vibration signals; Based on the preprocessed original vibration signals of the current connection area and different bolt looseness ratio values, the optimal order to characterize the bolt looseness of the current connection area is obtained based on fractional Fourier transform; Multiple offline models are established based on the least squares support vector machine of the quantized density target using the sample set of the current connection area, thereby establishing an offline training model of the current connection area under the optimal order under different bolt looseness ratio values ​​based on the support vector set. The sample set is the vibration signal characteristic information under the optimal order under the current connection area and the state estimate under different bolt looseness ratios; The above steps are repeated to obtain the offline training model at the optimal order for all connection areas with different bolt loosening ratio values; Online detection stage: Install a vibration sensor in the bottom connection area, and excite the vibration to collect N vibration signals. Then, based on the offline model obtained in the offline modeling stage, the final detection result is obtained. The output bolts are loose or normal, and if the output bolts are in the first place, the If the output result of the model is loose, locate the position of the loose bolt in the t A connection area.

[0008] Further, including: The offline modeling of the bolt connection area of ​​a specific type of transmission tower includes: The number of loose bolts is the ratio of loose bolts at the same plane position. The loose bolt ratio is defined as: in, Represents different bolt types, is the loosening coefficient of different types of bolts, is the number of loose bolts, m is the number of all bolts in the same plane. Further, including: The preprocessing of the collected vibration signal includes: assuming that the original vibration signal , which is caused by noise and vibration The signal is processed by signal enhancement technology to obtain an enhanced vibration signal. , expressed as: in, is the fractional Fourier transform in order The inverse transformation of It is in the order The spectral amplitude of the fractional Fourier transform is The following will collect The vibration signal is converted into time-frequency matrix , is a vibration signal collected from different connection areas in one excitation. If the total number of connection areas is N, then . Further, including: The method of obtaining the optimal order for characterizing the looseness of the bolts in the current connection area based on the fractional Fourier transform for the original vibration signal preprocessed at the current connection area and different bolt looseness ratio values ​​includes: Step S1: Determine the current bolt loose proportion threshold, the bolt loose proportion threshold is the threshold for estimating the bolt looseness. When it is greater than this threshold, it is determined to be a loose state, otherwise it is determined to be a normal state; calculate the frequency matrix Middle The mutual information value between the frequency point and the bolt loosening ratio and through Select the characteristic frequency set ,in, is the frequency matrix At a certain frequency point in For the The threshold of the vibration signal, The calculation method is: Step S2: Use kernel principal component analysis to reduce the dimension from the feature frequency set Feature extraction is performed in the , and the number of principal components with cumulative contribution greater than 90% is selected as feature information. Step S3: Then obtain the characteristic information corresponding to different feet in the current connection area under the current bolt loosening ratio, according to the objective function Get the optimal order of the connection area Step S4: Adjust the value of the bolt loosening ratio threshold value, and obtain the bolt loosening ratio of the connection area from arrive The corresponding vibration signals at different orders The characteristic information below is recorded as ,in, , M is the bolt loosening ratio from arrive The total number of bolts is used to obtain the optimal order of the current connection area under different bolt loosening ratio thresholds. ;in, . Further, including: The method of using the sample set of the current connection area to establish multiple offline models based on the least squares support vector machine of the quantized density target includes: assuming that the sample set , The optimal order The next Vibration signal characteristic information , Represents the state estimation under different bolt loosening ratios. First, based on the sample set Get the initial training model , and , and define the density radius ; Calculate each sample Density information at and error information , and obtain the quantitative density target information , the sample corresponding to the maximum quantization density target value is recorded as and will Add support vector set ; Update the density information of all samples , and based on the support vector set Get the training model, the performance indicators of the current training model are ; Iterate the above steps and recalculate the quantized density target information until the selected support vector set satisfy , To set the threshold, the final training model at this time is recorded as . Further, including: The final detection result is obtained based on the offline model obtained in the offline modeling stage, specifically including: Vibration sensors are placed at the bottom connection area of ​​the ABCD feet of the tower respectively, and vibration is performed once at a certain distance from the ground at the ABCD feet to collect vibration signals. The optimal order of the vibration signal is extracted based on the fractional Fourier transform. The following characteristic information , and then As input, get the corresponding offline model , according to the offline model The output results of each connection area are obtained. On the other hand, the present invention provides a transmission tower bolt loosening detection and positioning system based on fractional Fourier transform, the system comprising: an offline modeling stage and an online detection stage, the offline modeling stage comprising a signal acquisition module, a preprocessing module, an optimal order calculation module, an iteration module, and a model building module; The signal acquisition module is used to perform offline modeling on the bolt connection area of ​​a specific type of transmission tower, establish the corresponding relationship between vibration signals at different orders and the type, plane position, and number of loose bolts, and define the bolt loosening ratio; The preprocessing module is used to determine the total number N of connection areas of the tower from bottom to top, place vibration sensors near the four feet of the current connection area respectively, and perform vibration excitation at a certain distance from the current connection area, so as to collect original vibration signals at different bolt loosening ratio values, and preprocess the collected original vibration signals; The optimal order calculation module is used to obtain the optimal order characterizing the looseness of the bolts in the current connection area based on the fractional Fourier transform for the original vibration signal preprocessed under the current connection area and different bolt looseness ratio values; The model building module is used to establish multiple offline models based on the least squares support vector machine of the quantitative density target using the sample set of the current connection area, so as to establish an offline training model of the current connection area at the optimal order under different bolt loosening ratio values ​​based on the support vector set, and the sample set is the vibration signal characteristic information at the optimal order under the current connection area and the state estimation under different bolt loosening ratios; The iterative module is used to loop the above modules to obtain an offline training model at the optimal order for all connection areas at different bolt loosening ratio values; Online detection stage: Install a vibration sensor at the bottom connection area, and collect N vibration signals by excitation. Then, based on the offline model obtained in the offline modeling stage, obtain the final detection result. The detection result outputs whether the bolt is loose or in normal state. That is, if the first If the output result of the model is loose, locate the position of the loose bolt in the t A connection area.

[0009] Further, including: In the preprocessing module, the preprocessing of the collected vibration signal includes: In the preprocessing module, the preprocessing of the collected vibration signal includes: assuming that the original vibration signal , which is caused by noise and vibration The signal is processed by signal enhancement technology to obtain an enhanced vibration signal. , expressed as: in, is the fractional Fourier transform in order The inverse transformation of It is in the order The spectral amplitude of the fractional Fourier transform is The following will collect The vibration signal is converted into time-frequency matrix , is a vibration signal collected from different connection areas in one excitation. If the total number of connection areas is N, then .

[0010] Furthermore, this embodiment also includes: In the optimal order calculation module, the original vibration signal preprocessed for the current connection area and different bolt looseness ratio values ​​is used to obtain the optimal order characterizing the bolt looseness in the current connection area based on fractional Fourier transform, including: determining the current bolt looseness ratio threshold, the bolt looseness ratio threshold is the threshold for estimating bolt looseness, when it is greater than the threshold, it is determined to be in a loose state, otherwise, it is determined to be in a normal state; calculating the frequency matrix Middle The mutual information value between the frequency point and the bolt loosening ratio and through Select the characteristic frequency set ,in, is the frequency matrix At a certain frequency point in For the The threshold of the vibration signal, The calculation method is: Using kernel principal component analysis to reduce dimensionality from feature frequency sets Feature extraction is performed in the , and the number of principal components with cumulative contribution greater than 90% is selected as feature information. ; Then, the characteristic information corresponding to different feet in the current connection area under the current bolt loosening ratio is obtained, and according to the objective function Get the optimal order of the connection area ; Adjust the size of the bolt loosening ratio threshold, and obtain the bolt loosening ratio of the connection area from arrive The corresponding vibration signals at different orders The characteristic information below is recorded as ,in, , M is the bolt loosening ratio from arrive The total number of bolts is used to obtain the optimal order of the current connection area under different bolt loosening ratio thresholds. ; in, In a third aspect, the present invention further provides a transmission tower bolt loosening detection and positioning device based on fractional Fourier transform, the device comprising: a vibration sensor and a processor, the vibration sensor is installed on the transmission tower to be tested, and is used to collect vibration signals, the processor is used to perform correlation processing on the collected vibration signals, and obtain an offline model for predicting the connection area position of the loose bolt, so as to judge the specific position of the loose screw according to the offline model, wherein the correlation processing includes the following steps: Step 1: Offline modeling of the bolt connection area of ​​a specific type of transmission tower is performed to establish the corresponding relationship between the vibration signals at different orders and the type, plane position, and number of loose bolts, and define the bolt loosening ratio; Step 2, determining the total number N of connection areas of the tower from bottom to top, placing vibration sensors near the four feet of the current connection area respectively, and exciting at a certain distance from the current connection area, so as to collect original vibration signals at different bolt loosening ratio values, and preprocessing the collected original vibration signals; Step 3: for the original vibration signal after preprocessing in the current connection area and under different bolt loosening ratio values, the optimal order characterizing the loosening of the bolts in the current connection area is obtained based on fractional Fourier transform; Step 4: Using the sample set of the current connection area to establish multiple offline models based on the least squares support vector machine of the quantized density target, thereby establishing an offline training model of the current connection area at the optimal order under different bolt loosening ratio values ​​based on the support vector set, wherein the sample set is the vibration signal characteristic information at the optimal order under the current connection area and the state estimation under different bolt loosening ratios; Step 5: Repeat the above steps to obtain the offline training model at the optimal order for all connection areas at different bolt loosening ratios.

[0011] Beneficial effects: Compared with the prior art, the present invention has the following advantages: The present invention first analyzes the looseness ratio threshold, which sets the bolt looseness to different ratios, and uses fractional Fourier transform to analyze the vibration signal according to different ratios to obtain characteristic information at different orders and obtain the vibration signals at different orders under different bolt looseness ratios in the connection area. The characteristic information under , so as to obtain the optimal order of each connection area; Then, on the basis of the optimal order, the least squares support vector machine classification algorithm based on density quantization is combined to establish an offline training model for each connection area at the optimal order, and finally the loosening of the bolts is detected online, thereby realizing the automatic detection and precise positioning of the loose state of the bolts. Therefore, by the correlation processing of the vibration signal, the loose state of the bolts of the transmission tower can be detected and located without climbing the pole for maintenance, thereby improving the safety and maintenance efficiency of the power facilities. Therefore, the present invention improves the accuracy and efficiency of bolt loosening detection, enhances the anti-interference ability, and provides a strong guarantee for the safe operation of the transmission tower. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 This is a flow chart of a method for detecting and locating loose bolts of a transmission tower based on FRFT according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the numbering of the connection areas of the transmission tower according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the installation position of the vibration sensor according to an embodiment of the present invention; Figure 4 is a flow chart of a feature extraction algorithm according to an embodiment of the present invention; Figure 5 This is a flow chart of a least squares support vector machine model based on a quantized density target according to an embodiment of the present invention; Figure 6 The present invention is a flowchart of the detailed steps of the method for detecting and locating loose bolts on a transmission tower based on FRFT according to an embodiment of the present invention. DETAILED DESCRIPTION

[0013] 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.

[0014] like Figure 1 As shown, the present invention proposes a method for detecting and locating bolt loosening of a transmission tower based on fractional-order Fourier transform. For the same type of transmission towers, it is only necessary to perform fractional-order Fourier analysis on the vibration signals of each connection area in the offline modeling stage to obtain the optimal order of different connection areas in the bolt loosening detection process. In this way, during the online detection process, vibration sensors can be installed on the tower legs, and the loosening state of the bolts of the entire tower can be detected and located based on a handheld detector without manual pole climbing. This is also the first time that fractional-order Fourier transform technology has been applied to the field of bolt loosening detection.

[0015] S1 conducts offline modeling of the bolt connection area of ​​a specific type of transmission tower, establishes the corresponding relationship between the vibration signals at different orders and the type, plane position, and number of loose bolts, and obtains the optimal order that can characterize the looseness of bolts at different plane positions; S2 performs excitation of the same intensity at the same height of the tower base, collects vibration signals at the same position, and extracts frequency domain features at different orders based on fractional Fourier transform.

[0016] S3 realizes automatic detection and location of looseness status of different bolt types based on least squares support vector machine.

[0017] Specifically, the method analyzes the vibration signal through fractional Fourier transform, combined with the least squares support vector machine classification algorithm based on density quantization, to achieve automatic detection and precise positioning of the loose state of the bolt. In order to achieve the above task, the present invention adopts the following technical solutions: like Figure 6 Shown is a detailed flow chart of the method, including: Offline modeling phase: Step S-1 performs offline modeling on the bolt connection area of ​​a specific type of transmission tower, establishes the corresponding relationship between vibration signals at different orders and the type, plane position, and number of loose bolts, and defines the bolt loosening ratio.

[0018] In this embodiment, a specific model of transmission tower is selected to determine its bolt type, plane position and bolt loosening quantity. The bolt loosening quantity is the ratio of bolt loosening at the same plane position, and the bolt loosening ratio is defined as ,in, p Represents different bolt types, is the loosening coefficient of different types of bolts, n is the number of loose bolts, m is the number of all bolts at the same plane position. Step S-2 determines the total number N of connection areas of the tower from bottom to top, places vibration sensors near the four feet of the current connection area, and performs the same excitation at a certain distance from the current connection area, thereby collecting the original vibration signals at different bolt loosening ratios, and preprocessing the collected original vibration signals.

[0019] In this embodiment, if Figure 2 There are 9 connection areas on the tower shown, such as Figure 3 As shown, in the offline training stage, a vibration sensor is placed at each connection area, and vibration is performed at a certain distance from the current connection area, and then the original vibration signal of each connection area is collected, that is, the original collected vibration signal set includes: , and in this embodiment, the bolt loosening ratio is to , this embodiment does not limit the maximum and minimum values, which can generally be 0% to 100%. Here, several threshold data can be selected for collection, such as the loosening ratio threshold of 0%, 20%, 40%, 60%, 80%, 100%. In this embodiment, the method is specifically to collect, first determine that the tower has N connection areas from bottom to top, numbered 1, ..., N. Place a vibration sensor in the first connection area of ​​the A foot of the tower, and excite the A foot 50 cm above the ground. For example, collect the vibration signals of the N connection areas when the bolt loosening ratio is 20%, collect N times in total, that is, collect once for each connection area.

[0020] In this embodiment, it is assumed that the original vibration signal , which is caused by noise and vibration The signal is processed by signal enhancement technology to obtain an enhanced vibration signal. , expressed as: in, is the fractional Fourier transform in order The inverse transformation of It is in the order The spectral amplitude of the fractional Fourier transform is The following will collect The vibration signal is converted into time-frequency matrix , is a vibration signal collected from different connection areas in one excitation. If the total number of connection areas is N, then Step S-3 obtains the optimal order characterizing the looseness of the bolts in the current connection area based on the fractional Fourier transform for the original vibration signal preprocessed at the current connection area and different bolt looseness ratio values.

[0021] In this embodiment, the step further includes: Step S-3-1: Determine the current bolt loosening ratio threshold, which is a threshold for estimating bolt loosening. When it is greater than the threshold, it is determined to be in a loose state, otherwise, it is determined to be in a normal state; calculate the frequency matrix Middle The mutual information value between the frequency point and the bolt loosening ratio and through Select the characteristic frequency set ,in, is the frequency matrix At a certain frequency point in For the The threshold of the vibration signal, The calculation method is: Step S-3-2: Use kernel principal component analysis to reduce the dimension from the feature frequency set Feature extraction is performed in the , and the number of principal components with cumulative contribution greater than 90% is selected as feature information. Step S-3-3: Then obtain the characteristic information corresponding to different feet in the current connection area under the current bolt loosening ratio, according to the objective function Get the optimal order of the connection area ; Step S-3-4: Adjust the size of the bolt loosening ratio threshold, according to the above steps, respectively obtain the bolt loosening ratio of the connection area from respectively obtain the bolt loosening ratio of the connection area from arrive The corresponding vibration signals at different orders The characteristic information below is recorded as ,in, , M is the bolt loosening ratio from arrive The total number of bolts is used to obtain the optimal order of the current connection area under different bolt loosening ratio thresholds. ;in, In this embodiment, the bolt looseness ratio threshold is selected according to actual needs, and its specific data is not limited.

[0022] Step S-4 uses the sample set of the current connection area to establish multiple offline models based on the least squares support vector machine with a quantized density target, thereby establishing an offline training model of the current connection area at the optimal order under different bolt loosening ratios based on the support vector set. The sample set is the vibration signal characteristic information at the optimal order of the current connection area and the state estimation under different bolt loosening ratios.

[0023] In this embodiment, step S-4 specifically includes: S-4-1 Assume that the sample set , The optimal order The next Vibration signal characteristic information , Represents the state estimation under different bolt loosening ratios. First, based on the sample set Get the initial training model , and , and define the density radius ; S-4-2 Calculate each sample Density information at and error information , and obtain the quantitative density target information , the sample corresponding to the maximum quantization density target value is recorded as dand the corresponding samples Add support vector set ; Here, n Represents the value of a random sample, m From 1 to N All samples must be traversed; S-4-3 updates the density information of all samples , and based on the support vector set Get the training model of each connection area. The performance index corresponding to the current training model is ; S-4-4 Iterate steps S-4-2 and S-4-3 to recalculate the quantized density target information until the selected support vector set satisfy , To set the threshold, the final training model at this time is recorded as In the present embodiment, no limitation is imposed on the threshold value. Step S-5: loop the above steps to obtain the offline training model of all connection areas at the optimal order when the bolt loosening ratio is different; Online detection stage: install a vibration sensor in the bottom connection area, that is, install a vibration sensor in the first connection area, and perform excitation to collect vibration signals once or several times, and then obtain the final detection result based on the offline model obtained in the offline modeling stage, and the detection result outputs whether the bolt is loose or in a normal state. In this embodiment, there is no limit on the number of times the vibration signal is collected, but it is more used to improve the reliability of the detection, generally 3 times, and the collected vibration signals are averaged and processed.

[0024] In this embodiment, during the online detection stage, it is possible to consider placing vibration sensors at the first connection area of ​​the ABCD feet of the tower, and conduct an excitation at each of the ABCD feet 50 cm above the ground, thereby collecting vibration signals of the first connection areas of the four feet, a total of 4, and collecting vibration signals of other connection areas at the same time, and then extracting the vibration signal based on FRFT. The following characteristic information Then As input, input to the corresponding offline model The final test result is obtained, and the output is the two states of bolt looseness or normal. If the output result of the offline model is loose, the position of the loose bolt can be located in the iconnection area. In this embodiment, it is not excluded that there are multiple offline models that all output results are in a loose state. On the other hand, the present invention provides a transmission tower bolt loosening detection and positioning system based on fractional Fourier transform, the system comprising: an offline modeling stage and an online detection stage, the offline modeling stage comprising a signal acquisition module, a preprocessing module, an optimal order calculation module, an iteration module, and a model building module; The signal acquisition module is used to perform offline modeling on the bolt connection area of ​​a specific type of transmission tower, establish the corresponding relationship between vibration signals at different orders and the type, plane position, and number of loose bolts, and define the bolt loosening ratio; The preprocessing module is used to determine the total number N of connection areas of the tower from bottom to top, place vibration sensors near the four feet of the current connection area respectively, and perform vibration excitation at a certain distance from the current connection area, so as to collect original vibration signals at different bolt loosening ratio values, and preprocess the collected original vibration signals; The optimal order calculation module is used to obtain the optimal order characterizing the looseness of the bolts in the current connection area based on the fractional Fourier transform for the original vibration signal preprocessed under the current connection area and different bolt looseness ratio values; The model building module is used to establish multiple offline models based on the least squares support vector machine of the quantitative density target using the sample set of the current connection area, so as to establish an offline training model of the current connection area at the optimal order under different bolt loosening ratio values ​​based on the support vector set, and the sample set is the vibration signal characteristic information at the optimal order under the current connection area and the state estimation under different bolt loosening ratios; The iterative module is used to loop the above modules to obtain an offline training model at the optimal order for all connection areas at different bolt loosening ratio values; Online detection stage: Install a vibration sensor at the bottom connection area, and collect N vibration signals by excitation. Then, based on the offline model obtained in the offline modeling stage, obtain the final detection result. The detection result outputs whether the bolt is loose or in normal state. That is, if the first i If the output result of the model is loose, locate the position of the loose bolt in the i A connection area.

[0025] Furthermore, this embodiment also includes: In the preprocessing module, the preprocessing of the collected vibration signal includes: assuming that the original vibration signal , which is caused by noise and vibration The signal is processed by signal enhancement technology to obtain an enhanced vibration signal. , expressed as: in, is the fractional Fourier transform in order The inverse transformation of It is in the order The spectral amplitude of the fractional Fourier transform is The following will collect The vibration signal is converted into time-frequency matrix , is a vibration signal collected from different connection areas in one excitation. If the total number of connection areas is N, then .

[0026] Furthermore, this embodiment also includes: In the optimal order calculation module, the original vibration signal preprocessed for the current connection area and different bolt looseness ratio values ​​is used to obtain the optimal order characterizing the bolt looseness in the current connection area based on fractional Fourier transform, including: determining the current bolt looseness ratio threshold, the bolt looseness ratio threshold is the threshold for estimating bolt looseness, when it is greater than the threshold, it is determined to be in a loose state, otherwise, it is determined to be in a normal state; calculating the frequency matrix Middle The mutual information value between the frequency point and the bolt loosening ratio and through Select the characteristic frequency set ,in, is the frequency matrix At a certain frequency point in For the The threshold of the vibration signal, The calculation method is: Using kernel principal component analysis to reduce dimensionality from feature frequency sets Feature extraction is performed in the , and the number of principal components with cumulative contribution greater than 90% is selected as feature information. ; Then, the characteristic information corresponding to different feet in the current connection area under the current bolt loosening ratio is obtained, and according to the objective function Get the optimal order of the connection area ; Adjust the size of the bolt loosening ratio threshold, and obtain the bolt loosening ratio of the connection area from arrive The corresponding vibration signals at different orders The characteristic information below is recorded as ,in, , M is the bolt loosening ratio from arrive The total number of bolts is used to obtain the optimal order of the current connection area under different bolt loosening ratio thresholds. ;in, In a third aspect, the present invention further provides a transmission tower bolt loosening detection and positioning device based on fractional Fourier transform, the device comprising: a vibration sensor and a processor, the vibration sensor is installed on the transmission tower to be tested, and is used to collect vibration signals, the processor is used to perform correlation processing on the collected vibration signals, and obtain an offline model for predicting the connection area position of the loose bolt, so as to judge the specific position of the loose screw according to the offline model, wherein the correlation processing includes the following steps: Step 1: Offline modeling of the bolt connection area of ​​a specific type of transmission tower is performed to establish the corresponding relationship between the vibration signals at different orders and the type, plane position, and number of loose bolts, and define the bolt loosening ratio; Step 2, determining the total number N of connection areas of the tower from bottom to top, placing vibration sensors near the four feet of the current connection area respectively, and exciting at a certain distance from the current connection area, so as to collect original vibration signals at different bolt loosening ratio values, and preprocessing the collected original vibration signals; Step 3: for the original vibration signal after preprocessing in the current connection area and under different bolt loosening ratio values, the optimal order characterizing the loosening of the bolts in the current connection area is obtained based on fractional Fourier transform; Step 4: Using the sample set of the current connection area to establish multiple offline models based on the least squares support vector machine of the quantized density target, thereby establishing an offline training model of the current connection area at the optimal order under different bolt loosening ratio values ​​based on the support vector set, wherein the sample set is the vibration signal characteristic information at the optimal order under the current connection area and the state estimation under different bolt loosening ratios; Step 5: Repeat the above steps to obtain the offline training model at the optimal order for all connection areas with different bolt loosening ratios.

[0027] Finally, the loose bolts are detected using the offline training model in the device to locate the connection area where the loose bolts are located.

[0028] Other technical features of the transmission tower bolt loosening detection and positioning system based on fractional Fourier transform of the present invention are similar to the corresponding transmission tower bolt loosening detection and positioning method based on fractional Fourier transform, and will not be repeated here.

[0029] In order to illustrate the positioning accuracy of the present invention, this embodiment is combined with actual conditions to carry out experiments, including: The flow chart of the method for detecting and locating loose bolts of transmission towers based on FRFT of the present invention is as follows: Figure 1As shown. Taking a certain type of transmission tower as an example, a specific application of the present invention is given. The transmission tower model is a base drill tower (11BB-ZYT). The bolt diagram and connection area division of the tower are shown in Figure 2 As shown, there are 9 zones in total. Wireless vibration sensors are used to collect vibration signals. The installation method is magnetic. The installation location diagram is shown in Figure 3 In the offline modeling and online detection stages, the vibration signal is processed as shown in the flow Figure 4 As shown in the figure, the online detection and positioning output of bolt loosening is as follows: Figure 5 shown.

[0030] The specific steps include: Offline training phase: The tower has 9 connection areas from bottom to top, each with four legs, ABCD. The loosening threshold of bolts in each connection area is defined as 20%. In the offline modeling stage, the bolts in the first connection area are loosened according to the loosening definition. , at this time A is 20%. Place a vibration sensor at the first connection area of ​​the A foot of the tower, and vibrate the A foot 50 cm above the ground to obtain the vibration signal First, the signal is enhanced and the noise is filtered out to obtain the enhanced vibration signal. Then based on the fractional Fourier transform (FRFT) Next The vibration signal is converted into time-frequency matrix .

[0031] Then calculate Middle The mutual information value between the frequency point and the current bolt loosening ratio and through Select the characteristic frequency set ,in, For the The threshold of the vibration signal is Calculated; Then, the kernel principal component analysis (KPCA) dimensionality reduction technique is used to Feature extraction is performed in the , and the number of principal components with cumulative contribution greater than 90% is selected as feature information. .

[0032] According to the objective function Get the optimal order of the first connection zone .

[0033] The bolt loosening ratio of the first connection zone is further changed from 0% to 50% with an interval of 5%, and the vibration signals under each working condition are collected respectively. The characteristic information under the working condition and the bolt loosening ratio at this time are obtained according to the above method.

[0034] Thus forming a sample set , is the characteristic information of the vibration signal under the optimal order , Indicates a loose state estimate.

[0035] Firstly based on Get the initial training model , and , and define the density radius ; Then, calculate each sample Density information at and error information , and obtain the quantitative density target information , the sample corresponding to the maximum quantization density target value is recorded as d and the corresponding samples Add support vector set ; Secondly, update the density information of all samples , and based on The training model is obtained, and its performance index is ; Recalculate the quantized density target information until the selected support vector set satisfies At this time, we get That is the offline model at the optimal order in the first connection zone.

[0036] According to the above method, obtain the offline model of the remaining connection area In the online detection, vibration sensors are placed in the first connection area of ​​the ABCD feet of the tower, and vibration is excited once at the ABCD feet 50 cm above the ground. Four vibration signals are collected and the vibration signals are extracted based on FRFT. The following characteristic information Then As input, based on the corresponding offline model , get the final test result, output the bolt loosening ratio, if the ratio is greater than 20%, it is considered that the bolt loosening affects the stability of the tower, and the position of the loose bolt can be located in the first i Otherwise, it is considered to be in normal state.

[0037] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0038] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0039] Obviously, those skilled in the art can make various changes and modifications to the embodiments of the present invention without departing from the spirit and scope of the embodiments of the present invention. Thus, if these modifications and variations of the embodiments 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.

Claims

1. A method for detecting and locating loose bolts of transmission towers based on fractional-order Fourier transform, characterized in that: The method includes: Offline modeling stage: Offline modeling of the bolt connection area of ​​a specific type of transmission tower is carried out to establish the corresponding relationship between the vibration signals at different orders and the type, plane position, and number of loose bolts, and define the bolt loosening ratio; Determine the total number N of connection areas of the tower from bottom to top, place vibration sensors near the four feet of the current connection area respectively, and perform vibration excitation at a certain distance from the current connection area, so as to collect original vibration signals at different bolt loosening ratio values, and pre-process the collected original vibration signals; Based on the preprocessed original vibration signals of the current connection area and different bolt looseness ratio values, the optimal order to characterize the bolt looseness of the current connection area is obtained based on fractional Fourier transform; A plurality of offline models are established by using a sample set of the current connection zone based on a least squares support vector machine with a quantized density target, thereby establishing an offline training model of the current connection zone at the optimal order under different bolt loosening ratio values ​​based on the support vector set, wherein the sample set is the vibration signal characteristic information at the optimal order under the current connection zone and the state estimation under different bolt loosening ratios; The above steps are repeated to obtain the offline training model at the optimal order for all connection areas with different bolt loosening ratio values; Online detection stage: Install a vibration sensor at the bottom connection area, and perform excitation to collect N vibration signals. Then, based on the offline model obtained in the offline modeling stage, obtain the final detection result. The detection result outputs whether the bolt is loose or in normal state.

2. The method for detecting and locating loose bolts of transmission towers based on fractional-order Fourier transform according to claim 1, characterized in that: The offline modeling of the bolt connection area of ​​a specific type of transmission tower includes: The number of loose bolts is the ratio of loose bolts at the same plane position. The loose bolt ratio is defined as: in, Represents different bolt types, is the loosening coefficient of different types of bolts, is the number of loose bolts, m is the number of all bolts in the same plane position.

3. The method for detecting and locating loose bolts of transmission towers based on fractional-order Fourier transform according to claim 2, characterized in that: The preprocessing of the collected vibration signal comprises: Assuming the original vibration signal , which is caused by noise and vibration The signal is processed by signal enhancement technology to obtain an enhanced vibration signal. , expressed as: in, is the fractional Fourier transform in order The inverse transformation of It is in the order The spectral amplitude of the fractional Fourier transform is The following will collect The vibration signal is converted into time-frequency matrix , is a vibration signal collected from different connection areas in one excitation. If the total number of connection areas is N, then .

4. The method for detecting and locating loose bolts of transmission towers based on fractional Fourier transform according to claim 3 is characterized in that: The method of obtaining the optimal order for characterizing the looseness of the bolts in the current connection area based on the fractional Fourier transform for the original vibration signal preprocessed at the current connection area and different bolt looseness ratio values ​​includes: Step S1: Determine the current bolt loosening ratio threshold, which is a threshold for estimating bolt loosening. When it is greater than the threshold, it is determined to be in a loose state, otherwise, it is determined to be in a normal state; calculate the frequency matrix Middle The mutual information value between the frequency point and the bolt loosening ratio and through Select the characteristic frequency set ,in, is the frequency matrix At a certain frequency point in For the The threshold of the vibration signal, The calculation method is: Step S2: Use kernel principal component analysis to reduce dimensionality from the feature frequency set Feature extraction is performed in the , and the number of principal components with cumulative contribution greater than 90% is selected as feature information. Step S3: Then obtain the characteristic information corresponding to different feet in the current connection area under the current bolt loosening ratio, according to the objective function Get the optimal order of the connection area Step S4: Adjust the value of the bolt loosening ratio threshold value, and obtain the bolt loosening ratio of the connection area from arrive The corresponding vibration signals at different orders The characteristic information below is recorded as ,in, , M is the bolt loosening ratio from arrive The total number of bolts is used to obtain the optimal order of the current connection area under different bolt loosening ratio thresholds. ; in, .

5. The method for detecting and locating loose bolts of transmission towers based on fractional Fourier transform according to claim 4, characterized in that: The method of establishing multiple offline models using the sample set of the current connection area based on the least squares support vector machine of the quantized density target includes: Assume sample set , The optimal order The next Vibration signal characteristic information , Represents the state estimation under different bolt loosening ratios. First, based on the sample set Get the initial training model , and , and define the density radius ; Calculate each sample Density information at and error information , and obtain the quantitative density target information , the sample corresponding to the maximum quantization density target value is recorded as and will Add support vector set ; Update density information for all samples , and based on the support vector set Get the training model, the performance indicators of the current training model are ; Iterate the above steps and recalculate the quantized density target information until the selected support vector set satisfy , To set the threshold, the final training model at this time is recorded as .

6. The method for detecting and locating loose bolts of transmission towers based on fractional-order Fourier transform according to claim 5, characterized in that: The final detection result is obtained based on the offline model obtained in the offline modeling stage, specifically including: Vibration sensors are placed at the bottom connection area of ​​the ABCD feet of the tower respectively, and vibration is performed once at a certain distance from the ground on the ABCD feet to collect vibration signals, and the optimal order of the vibration signals is extracted based on the fractional Fourier transform. The following characteristic information , and then As input, get the corresponding offline model , according to the offline model Get the output results for each connection area.

7. A transmission tower bolt loosening detection and positioning system based on fractional Fourier transform, characterized in that: The system comprises: an offline modeling stage and an online detection stage, wherein the offline modeling stage comprises a signal acquisition module, a preprocessing module, an optimal order calculation module, an iteration module, and a model building module; The signal acquisition module is used to perform offline modeling on the bolt connection area of ​​a specific type of transmission tower, establish the corresponding relationship between vibration signals at different orders and the type, plane position, and number of loose bolts, and define the bolt loosening ratio; The preprocessing module is used to determine the total number N of connection areas of the tower from bottom to top, place vibration sensors near the four feet of the current connection area respectively, and perform vibration excitation at a certain distance from the current connection area, so as to collect original vibration signals at different bolt loosening ratio values, and preprocess the collected original vibration signals; The optimal order calculation module is used to obtain the optimal order characterizing the looseness of the bolts in the current connection area based on the fractional Fourier transform for the original vibration signal preprocessed under the current connection area and different bolt looseness ratio values; The model building module is used to establish multiple offline models based on the least squares support vector machine of the quantitative density target using the sample set of the current connection area, so as to establish an offline training model of the current connection area at the optimal order under different bolt loosening ratio values ​​based on the support vector set, and the sample set is the vibration signal characteristic information at the optimal order under the current connection area and the state estimation under different bolt loosening ratios; The iterative module is used to loop the above modules to obtain an offline training model at the optimal order for all connection areas at different bolt loosening ratio values; Online detection stage: Install a vibration sensor at the bottom connection area, and perform excitation to collect N vibration signals. Then, based on the offline model obtained in the offline modeling stage, obtain the final detection result. The detection result outputs whether the bolt is loose or in normal state.

8. The transmission tower bolt loosening detection and positioning system based on fractional Fourier transform according to claim 7 is characterized in that: In the preprocessing module, the preprocessing of the collected vibration signal includes: Assuming the original vibration signal , which is caused by noise and vibration The signal is processed by signal enhancement technology to obtain an enhanced vibration signal. , expressed as: in, is the fractional Fourier transform in order The inverse transformation of It is in the order The spectral amplitude of the fractional Fourier transform is The following will collect The vibration signal is converted into time-frequency matrix , is a vibration signal collected from different connection areas in one excitation. If the total number of connection areas is N, then .

9. The transmission tower bolt loosening detection and positioning system based on fractional Fourier transform according to claim 8, characterized in that: In the optimal order calculation module, the original vibration signal preprocessed at the current connection area and different bolt loosening ratio values ​​is used to obtain the optimal order characterizing the loosening of the bolts in the current connection area based on fractional Fourier transform, including: Determine the current bolt loosening ratio threshold, the bolt loosening ratio threshold is the threshold for estimating bolt loosening, when it is greater than the threshold, it is determined to be in a loose state, otherwise, it is determined to be in a normal state; calculate the frequency matrix Middle The mutual information value between the frequency point and the bolt loosening ratio and through Select the characteristic frequency set , in, is the frequency matrix At a certain frequency point in For the The threshold of the vibration signal, The calculation method is: ; Use kernel principal component analysis to reduce dimension from feature frequency set Feature extraction is performed in the , and the number of principal components with cumulative contribution greater than 90% is selected as feature information. ; Then, the characteristic information corresponding to different feet in the current connection area under the current bolt loosening ratio is obtained, and according to the objective function Get the optimal order of the connection area ; Adjust the size of the bolt loosening ratio threshold, and obtain the bolt loosening ratio of the connection area from arrive The corresponding vibration signals at different orders The characteristic information below is recorded as ,in, , M is the bolt loosening ratio from arrive The total number of bolts is used to obtain the optimal order of the current connection area under different bolt loosening ratio thresholds. ; in, .

10. A transmission tower bolt loosening detection and positioning device based on fractional Fourier transform, characterized in that: The device comprises: a vibration sensor and a processor, wherein the vibration sensor is installed on a transmission tower to be tested and is used to collect vibration signals, and the processor is used to perform correlation processing on the collected vibration signals to obtain an offline model for predicting the position of a connection area where a loose bolt is located, thereby determining the specific position of the loose screw according to the offline model, wherein the correlation processing comprises the following steps: Step 1: Offline modeling of the bolt connection area of ​​a specific type of transmission tower is performed to establish the corresponding relationship between the vibration signals at different orders and the type, plane position, and number of loose bolts, and define the bolt loosening ratio; Step 2, determining the total number N of connection areas of the tower from bottom to top, placing vibration sensors near the four feet of the current connection area respectively, and exciting at a certain distance from the current connection area, so as to collect original vibration signals at different bolt loosening ratio values, and preprocessing the collected original vibration signals; Step 3: for the original vibration signal after preprocessing in the current connection area and under different bolt loosening ratio values, the optimal order characterizing the loosening of the bolts in the current connection area is obtained based on fractional Fourier transform; Step 4: Using the sample set of the current connection area to establish multiple offline models based on the least squares support vector machine of the quantized density target, thereby establishing an offline training model of the current connection area at the optimal order under different bolt loosening ratio values ​​based on the support vector set, wherein the sample set is the vibration signal characteristic information at the optimal order under the current connection area and the state estimation under different bolt loosening ratios; Step 5: Repeat the above steps to obtain the offline training model at the optimal order for all connection areas at different bolt loosening ratios.

Images