Transmission tower health status assessment method and system based on multi-source information fusion

Abstract

A method and system for evaluating the health status of a transmission tower based on multi-source information fusion. The method first collects data signals required for evaluation, performs feature extraction, obtains time-frequency feature parameters, and divides health status categories. Then, a fault detection model based on an accurate category association algorithm is established. The time-frequency feature parameters are used as input, and the state correlation of different categories is calculated as a diagnosis basis. Then, based on the diagnosis basis, the diagnosis rates of various health states when different data signals are used as the diagnosis basis are calculated, and the rates are used as basic probability distribution values. An evidence table is constructed and data fusion is performed to obtain the final diagnosis result, and the evaluation of the health status of the transmission tower is completed. The present invention is based on multi-source information fusion, and the health status evaluation of the transmission tower is completed through a comprehensive method of multiple influencing dimensions, which effectively solves the defect of low fault diagnosis accuracy of a single information source or a single decision model, so that the present invention has high accuracy and credibility.

Description

Technical Field

[0001] The present invention relates to a tower health status assessment method, and belongs to the technical field of power transmission line operation and maintenance, and in particular to a power transmission tower health status assessment method and system based on multi-source information fusion. Background Art

[0002] Since the lines of power transmission and transformation projects often have a long span, it is often necessary to arrange transmission towers on adverse geological conditions such as foothill slopes, alluvial plains, and deserts. At the same time, affected by severe climate conditions such as crustal movement, rain, snow, and hail, transmission towers are often deformed or tilted due to uneven force and weather factors. In severe cases, they may even collapse, causing transmission line interruptions, which in turn affects the safe and stable operation of the power system. Therefore, it is of great significance to ensure the safe and reliable transmission of electric energy.

[0003] The currently commonly used manual inspection methods often have problems such as large consumption of human and material resources, and it is difficult to ensure the safety of inspectors in extreme weather and harsh environments; and the current sensor-based fault diagnosis technology is mostly based on a single information source signal to identify the health status. This single sensor information processing technology is often affected by complex and changeable external factors in actual operation, and has great uncertainty. It is difficult to meet the requirements of reliability and accuracy of transmission tower fault detection by relying solely on a single information source. Therefore, a means is urgently needed to solve the above defects in the existing technology and meet the needs of transmission tower fault detection. Summary of the invention

[0004] The purpose of the present invention is to overcome the above-mentioned defects and problems existing in the prior art and to provide a method and system for evaluating the health status of a transmission tower based on multi-source information fusion with high accuracy.

[0005] To achieve the above objectives, the technical solution of the present invention is: a method for evaluating the health status of a transmission tower based on multi-source information fusion, comprising:

[0006] S1. Collecting data signals required for transmission tower evaluation; the data signals include tension signals, torque signals, strain signals and angle signals;

[0007] S2. Extract features from the data signal, obtain time-frequency feature parameters of the data signal, and classify the health status of the transmission tower into different categories;

[0008] S3. Establish a fault detection model based on an accurate category association algorithm, use time-frequency feature parameters as input to the fault detection model, calculate the state correlation of different categories, and use the category with the largest state correlation as the basis for diagnosis;

[0009] S4. Based on the diagnosis basis, the diagnosis rates of various health states when different data signals of the transmission tower are obtained as the diagnosis basis are calculated;

[0010] S5. Use the diagnosis rates of various health states as basic probability distribution values, construct an evidence table, and then perform data fusion to obtain the final diagnosis results and complete the evaluation of the health status of the transmission tower.

[0011] The time-frequency characteristic parameters include: mean value, root mean square value, peak value, kurtosis, margin index, pulse index, waveform index, average frequency, center of gravity frequency, and frequency root mean square.

[0012] The step S3 specifically includes:

[0013] S31, confirming the characteristic dimension of the time-frequency characteristic parameter, different categories of health status and K value; the different categories of health status include four health statuses, namely, primary, secondary, tertiary and quaternary; the K value is the number of neighbor samples;

[0014] S32. Based on the nearest neighbor algorithm, calculate the sum of the average accuracy of K values ​​when K values ​​are 3, 5, and 7 under the five-fold cross validation condition, and record the calculation result as P t ;

[0015] S33, exclude the i-th feature dimension, then repeat step S32, and record the calculation result as P i ;

[0016] S34. Calculate the state differentiation capability of different feature dimensions, and the expression is as follows:

[0017] CDA i =1-(P i -P t );

[0018] Among them: CDA j is the state discrimination ability of the i-th feature dimension;

[0019] S35, normalize the i feature dimensions to obtain the corresponding weight W i , which is expressed as follows:

[0020]

[0021] S36. Based on the weights and the ability to distinguish states of different feature dimensions, the Euclidean distance is calculated to optimize the fault detection model. The expression is as follows:

[0022]

[0023] Where: D(X1,X2) is the Euclidean distance based on the ability to distinguish feature states, xli is the position of the target point in the i-th dimension, x 2i is the position of the training sample point in the i-th dimension;

[0024] S37, calculate the proportion of the j-th type of points in the neighboring sample points, and the average Euclidean distance between the target point and the j-th type of neighboring sample points, and the expression is as follows:

[0025]

[0026] Where: M 1j is the proportion of the jth class of points in the neighboring sample points, K is the number of neighboring samples, N j is the number of samples belonging to the jth class state among the K nearest neighbor samples;

[0027]

[0028] Where: M 2j is the average Euclidean distance between the target point and the jth class of neighboring sample points, ∑D(X, X j ) is the sum of the distances from the samples in the jth state to the experimental samples, X is the position of the target point, X j is the position of the jth class of neighboring sample points;

[0029] S38. Calculate the status correlation of different categories, and select the category with the largest status correlation as the diagnosis basis. The expression is as follows:

[0030]

[0031] Among them: CCD j is the state association degree of the jth category.

[0032] In step S4, based on the diagnosis basis, the expression for calculating the diagnosis rate of various health states is as follows:

[0033]

[0034] Where: m ij When the jth signal is used as the basis for judgment, the fault detection model diagnoses the health status of the transmission tower as the ratio of the ith signal, that is, the health status diagnosis rate, M ij N is the number of data classified into the i-th health state by the fault detection model based on the j-th signal, j is the total number of j-th signal data input into the fault detection model.

[0035] The step S5 specifically includes:

[0036] S51, taking the obtained tension signal, torque signal, strain signal, and angle signal as different evidences, and taking the health status diagnosis rate as the basic probability distribution value to construct an evidence table, using the DS evidence reasoning method to perform multi-evidence fusion reasoning, and calculating the fusion probability distribution value; the expression of the fusion probability distribution value is as follows:

[0037]

[0038]

[0039] Where: M(I) is the basic probability distribution value, K is the conflict factor, m1, m2, m3, m4 are different evidences, w, x, y, z are the status categories diagnosed by different evidences, and I is different health status;

[0040] S52. Compare the basic probability distribution values ​​of the health status of different evidences, and select the health status of the transmission tower corresponding to the maximum value as the final diagnosis result.

[0041] A transmission tower health status assessment system based on multi-source information fusion, the system comprising:

[0042] A data signal acquisition module is used to collect data signals required for transmission tower evaluation; the data signals include tension signals, torque signals, strain signals and angle signals;

[0043] The signal feature extraction module is used to extract features from data signals, obtain time-frequency feature parameters of data signals, and classify the health status of transmission towers into different categories;

[0044] The fault detection model building module is used to establish a fault detection model based on the precise category association algorithm, use the time-frequency feature parameters as the input of the fault detection model, calculate the state correlation of different categories, and use the category with the largest state correlation as the basis for diagnosis;

[0045] A health status diagnosis rate calculation module is used to calculate various health status diagnosis rates when different data signals of the transmission tower are obtained as the diagnosis basis based on the diagnosis basis;

[0046] The health status assessment module is used to use various health status diagnosis rates as basic probability distribution values, construct an evidence table, and then perform data fusion to obtain the final diagnosis results and complete the assessment of the health status of the transmission tower.

[0047] The time-frequency characteristic parameters of the acquired data signal include: mean value, root mean square value, peak value, kurtosis, margin index, pulse index, waveform index, average frequency, center of gravity frequency, and frequency root mean square;

[0048] The number of categories includes different health states, namely, primary, secondary, tertiary, and quaternary.

[0049] The fault detection model building module builds a fault detection model according to the following steps;

[0050] S31, confirming the characteristic dimension of the time-frequency characteristic parameter, different categories of health status and K value; the different categories of health status include four health statuses, namely, primary, secondary, tertiary and quaternary; the K value is the number of neighbor samples;

[0051] S32. Based on the nearest neighbor algorithm, calculate the sum of the average accuracy of K values ​​when K values ​​are 3, 5, and 7 under the five-fold cross validation condition, and record the calculation result as P t ;

[0052] S33, exclude the i-th feature dimension, then repeat step S32, and record the calculation result as P i ;

[0053] S34. Calculate the state differentiation capability of different feature dimensions, and the expression is as follows:

[0054] CDA i =1-(P i -P t );

[0055] Among them: CDA i is the state discrimination ability of the i-th feature dimension;

[0056] S35, normalize the i feature dimensions to obtain the corresponding weight W i , which is expressed as follows:

[0057]

[0058] S36. Based on the weights and the ability to distinguish states of different feature dimensions, the Euclidean distance is calculated to optimize the fault detection model. The expression is as follows:

[0059]

[0060] Where: D(X1,X2) is the Euclidean distance based on the ability to distinguish feature states, x 1i is the position of the target point in the i-th dimension, x 2i is the position of the training sample point in the i-th dimension;

[0061] S37, calculate the proportion of the j-th type of points in the neighboring sample points, and the average Euclidean distance between the target point and the j-th type of neighboring sample points, and the expression is as follows:

[0062]

[0063] Where: M 1j is the proportion of the jth class of points in the neighboring sample points, K is the number of neighboring samples, N j is the number of samples belonging to the jth class state among the K nearest neighbor samples;

[0064]

[0065] Where: M 2j is the average Euclidean distance between the target point and the jth class of neighboring sample points, ∑D(X, X j ) is the sum of the distances from the samples in the jth state to the experimental samples, X is the position of the target point, X j is the position of the jth class of neighboring sample points;

[0066] S38. Calculate the status correlation of different categories, and select the category with the largest status correlation as the diagnosis basis. The expression is as follows:

[0067]

[0068] Among them: CCD j is the state association degree of the jth category.

[0069] The health status diagnosis rate calculation module calculates the health status diagnosis rate based on the diagnosis basis according to the following formula;

[0070]

[0071] Where: m ij When the jth signal is used as the basis for judgment, the fault detection model diagnoses the health status of the transmission tower as the ratio of the ith signal, that is, the health status diagnosis rate, M ij N is the number of data classified into the i-th health state by the fault detection model based on the j-th signal, j is the total number of j-th signal data input into the fault detection model.

[0072] The health status assessment module assesses the health status of the transmission tower according to the following steps;

[0073] S51, taking the obtained tension signal, torque signal, strain signal, and angle signal as different evidences, and taking the health status diagnosis rate as the basic probability distribution value to construct an evidence table, using the DS evidence reasoning method to perform multi-evidence fusion reasoning, and calculating the fusion probability distribution value; the expression of the fusion probability distribution value is as follows:

[0074]

[0075]

[0076] Where: M(I) is the basic probability distribution value, K is the conflict factor, m1, m2, m3, m4 are different evidences, w, x, y, z are the status categories diagnosed by different evidences, and I is different health status;

[0077] S52. Compare the basic probability distribution values ​​of the health status of different evidences, and select the health status of the transmission tower corresponding to the maximum value as the final diagnosis result.

[0078] Compared with the prior art, the present invention has the following beneficial effects:

[0079] The present invention discloses a transmission tower health status assessment method and system based on multi-source information fusion. The method first collects data signals required for transmission tower assessment, then performs feature extraction, obtains time-frequency feature parameters, and divides health status categories. Then, a fault detection model based on an accurate category association algorithm is established. The time-frequency feature parameters are used as input to calculate the state association of different categories to obtain a diagnosis basis. Then, based on the diagnosis basis, various health status diagnosis rates are calculated when different data signals of the transmission tower are used as a diagnosis basis. Finally, various health status diagnosis rates are used as basic probability distribution values ​​to construct an evidence table. Then, data fusion is performed to obtain a final diagnosis result, and the health status assessment of the transmission tower is completed. In the application of the present design, based on multi-source information fusion, the influencing factors can be analyzed more deeply and comprehensively, and the health status assessment of the transmission tower is completed through a comprehensive combination of multiple influencing dimensions, which effectively solves the defect of low fault diagnosis accuracy of a single information source or a single decision model, so that the present invention has high accuracy and credibility. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] Figure 1 It is a flow chart of the method steps of the present invention.

[0081] Figure 2 It is a schematic diagram of the system structure of the present invention.

[0082] Figure 3 It is a schematic diagram of the device structure of the present invention.

[0083] In the figure: data signal acquisition module 1, signal feature extraction module 2, fault detection model construction module 3, health status diagnosis rate calculation module 4, health status assessment module 5, processor 6, memory 7, computer program code 71. DETAILED DESCRIPTION

[0084] The present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0085] Embodiment 1:

[0086] See also Figure 1 , a transmission tower health status assessment method based on multi-source information fusion, including:

[0087] S1. Collecting data signals required for transmission tower evaluation; the data signals include tension signals, torque signals, strain signals and angle signals; the data signals can be obtained by installing corresponding sensors on the transmission tower.

[0088] S2. Extract features from the data signal, obtain time-frequency feature parameters of the data signal, and classify the health status of the transmission tower into different categories;

[0089] When processing fault signals, the mean, effective value, skewness, kurtosis, etc. are often used as dimensional indicators. Dimensional indicators are relatively greatly affected by the working conditions of the equipment, and have high requirements on the constancy of the sensor measuring points. In order to monitor the operating status of diagnostic equipment, various dimensionless indicators are also widely used, mainly including: kurtosis index, margin index, waveform index, peak index, pulse index, etc. Dimensionless indicators are less affected by external loads, measuring points and other factors, and are not sensitive to changes in factors such as the frequency and amplitude of the fault signal, so they have relatively strong adaptability. When applied, they also include frequency domain indicators, including center of gravity frequency, root mean square frequency, average frequency, frequency variance, etc.

[0090] Furthermore, the time-frequency characteristic parameters include: mean value, root mean square value, peak value, kurtosis, margin index, pulse index, waveform index, average frequency, center of gravity frequency, and frequency root mean square.

[0091] The different categories include different health states, namely level one, level two, level three, and level four; level one, level two, level three, and level four correspond to good, normal, abnormal, and major abnormal, respectively.

[0092] S3. Establish a fault detection model based on an accurate category association algorithm, use time-frequency feature parameters as input to the fault detection model, calculate the state correlation of different categories, and use the category with the largest state correlation as the basis for diagnosis;

[0093] Further, it specifically includes:

[0094] S31, confirming the characteristic dimension of the time-frequency characteristic parameter, different categories of health status and K value; the different categories of health status include four health statuses, namely, primary, secondary, tertiary and quaternary; the K value is the number of neighbor samples;

[0095] In this embodiment, the characteristic dimensions of the time-frequency characteristic parameters are 10, namely: mean, root mean square value, peak value, kurtosis, margin index, pulse index, waveform index, average frequency, center of gravity frequency, frequency; the different categories of health status are 4, namely: good, normal, abnormal and major abnormal; the K value is set to 7.

[0096] S32. Based on the nearest neighbor algorithm, calculate the sum of the average accuracy of K values ​​when K values ​​are 3, 5, and 7 under the five-fold cross validation condition, and record the calculation result as P t ;

[0097] S33, exclude the i-th feature dimension, then repeat step S32, and record the calculation result as P i ;

[0098] S34. Calculate the state differentiation capability of different feature dimensions, and the expression is as follows:

[0099] CDA i =1-(P i -P t );

[0100] Among them: CDA i is the state distinguishing ability of the i-th feature dimension; j = 1, 2, 3…10, representing 10 different feature dimensions respectively;

[0101] S35, normalize the i feature dimensions to obtain the corresponding weight W i , which is expressed as follows:

[0102]

[0103] S36. Based on the weights and the ability to distinguish states of different feature dimensions, the Euclidean distance is calculated to optimize the fault detection model. The expression is as follows:

[0104]

[0105] Where: D(X1, X2) is the Euclidean distance based on the ability to distinguish feature states, x 1i is the position of the target point in the i-th dimension, x 2i is the position of the training sample point in the i-th dimension;

[0106] S37, calculate the proportion of the j-th type of points in the neighboring sample points, and the average Euclidean distance between the target point and the j-th type of neighboring sample points, and the expression is as follows:

[0107]

[0108] Where: M 1j is the proportion of the jth class of points in the neighboring sample points, K is the number of neighboring samples, N j is the number of samples belonging to the jth state among the K nearest neighbor samples; j = 1, 2, 3, 4, i.e., four different data signals; N is the number of samples belonging to the jth state among the 7 nearest neighbor samples. The larger the value of M1, the more points in the jth class there are.

[0109]

[0110] Where: M 2j is the average Euclidean distance between the target point and the jth class of neighboring sample points, ∑D(X, X j ) is the sum of the distances from the samples in the jth state to the experimental samples, X is the position of the target point, X j is the position of the jth class neighbor sample point; the smaller the value of M2, the smaller the average distance between the sample to be classified and the jth state sample.

[0111] S38. Calculate the status correlation of different categories, and select the category with the largest status correlation as the diagnosis basis. The expression is as follows:

[0112]

[0113] Among them: CCD j is the state association degree of the jth category.

[0114] S4. Based on the diagnosis basis, the diagnosis rates of various health states when different data signals of the transmission tower are obtained as the diagnosis basis are calculated;

[0115] Furthermore, based on the diagnostic basis, the expressions for calculating the diagnostic rates of various health states are as follows:

[0116]

[0117] in: m ij is the ratio of the health status of the transmission tower to the i-th one when the fault detection model uses the j-th signal as the judgment basis, that is, the health status diagnosis rate, M ij N is the number of data classified into the i-th health state by the fault detection model based on the j-th signal, j is the total number of j-th signal data input into the fault detection model.

[0118] In this embodiment, i=1, 2, 3, 4 represent four different health states, namely, level one, level two, level three, and level four; j=1, 2, 3, 4 represent four sensor signals, namely, tension, torque, strain, and angle, respectively.

[0119] S5. Use the diagnosis rates of various health states as basic probability distribution values, construct an evidence table, and then perform data fusion to obtain the final diagnosis results and complete the evaluation of the health status of the transmission tower.

[0120] Further, it specifically includes:

[0121] S51. Use the obtained tension signal, torque signal, strain signal, and angle signal as different evidences, and use the health status diagnosis rate as the basic probability distribution value to construct an evidence table.

[0122] In this embodiment, F1, E2, E3, and E4 are set as four different evidences of tension, torque, strain, and angle, A, B, C, and D represent four health states of primary, secondary, tertiary, and quaternary, and the diagnosis rate is used as the basic probability (BPA) distribution value to construct an evidence table; the evidence table is as follows:

[0123]

[0124] The DS evidence reasoning method is used to perform multi-evidence fusion reasoning and calculate the fusion probability distribution value; the expression of the fusion probability distribution value is as follows:

[0125]

[0126]

[0127] Where: M(I) is the basic probability distribution value, K is the conflict factor, m1, m2, m3, m4 are different evidences, w, x, y, z are the status categories diagnosed by different evidences, i = A, B, C, D, that is, different health status;

[0128] Based on the above expression, the fused BPA value is calculated as follows:

[0129]

[0130]

[0131]

[0132]

[0133] in:

[0134]

[0135] After data fusion, the final data table is obtained:

[0136]

[0137] S52. Compare the basic probability distribution values ​​of M(A), M(B), M(C), and M(D), select the health status of the transmission tower corresponding to the maximum value as the final diagnosis result, complete the health status assessment of the transmission tower, and adopt corresponding operation and maintenance strategies according to the four levels of the operation status results.

[0138] Embodiment 2:

[0139] See also Figure 2 , a transmission tower health status assessment system based on multi-source information fusion, the system comprising:

[0140] The data signal acquisition module 1 is used to collect data signals required for transmission tower evaluation; the data signals include tension signals, torque signals, strain signals and angle signals;

[0141] The signal feature extraction module 2 is used to extract features from the data signal, obtain the time-frequency feature parameters of the data signal, and classify the health status of the transmission tower into different categories;

[0142] Furthermore, the signal feature extraction module 2 obtains time-frequency feature parameters of the data signal including mean value, root mean square value, peak value, kurtosis, margin index, pulse index, waveform index, average frequency, center of gravity frequency, and frequency root mean square.

[0143] Fault detection model building module 3 is used to establish a fault detection model based on an accurate category association algorithm, use time-frequency feature parameters as input to the fault detection model, calculate the state correlation of different categories, and use the category with the largest state correlation as the basis for diagnosis;

[0144] Further, the fault detection model building module 3 obtains the diagnosis basis according to the following steps;

[0145] S31, confirming the characteristic dimension of the time-frequency characteristic parameter, different categories of health status and K value; the different categories of health status include four health statuses, namely, primary, secondary, tertiary and quaternary; the K value is the number of neighbor samples;

[0146] S32. Based on the nearest neighbor algorithm, calculate the sum of the average accuracy of K values ​​when K values ​​are 3, 5, and 7 under the five-fold cross validation condition, and record the calculation result as P t ;

[0147] S33, exclude the i-th feature dimension, then repeat step S32, and record the calculation result as P i ;

[0148] S34. Calculate the state differentiation capability of different feature dimensions, and the expression is as follows:

[0149] CDA i =1-(P i -P t );

[0150] Among them: CDA j is the state discrimination ability of the i-th feature dimension;

[0151] S35, normalize the i feature dimensions to obtain the corresponding weight W i , which is expressed as follows:

[0152]

[0153] S36. Based on the weights and the ability to distinguish states of different feature dimensions, the Euclidean distance is calculated to optimize the fault detection model. The expression is as follows:

[0154]

[0155] Where: D(X1, X2) is the Euclidean distance based on the ability to distinguish feature states, x 1i is the position of the target point in the i-th dimension, x 2i is the position of the training sample point in the i-th dimension;

[0156] S37, calculate the proportion of the j-th type of points in the neighboring sample points, and the average Euclidean distance between the target point and the j-th type of neighboring sample points, and the expression is as follows:

[0157]

[0158] Where: M 1j is the proportion of the jth class of points in the neighboring sample points, K is the number of neighboring samples, N j is the number of samples belonging to the jth class state among the K nearest neighbor samples;

[0159]

[0160] Where: M 2j is the average Euclidean distance between the target point and the jth class of neighboring sample points, ∑D(X, X j ) is the sum of the distances from the samples in the jth state to the experimental samples, X is the position of the target point, X j is the position of the jth class of neighboring sample points;

[0161] S38. Calculate the status correlation of different categories, and select the category with the largest status correlation as the diagnosis basis. The expression is as follows:

[0162]

[0163] Among them: CCD j is the state association degree of the jth category.

[0164] The health status diagnosis rate calculation module 4 is used to calculate the various health status diagnosis rates when different data signals of the transmission tower are obtained as the diagnosis basis based on the diagnosis basis;

[0165] Furthermore, the health status diagnosis rate calculation module 4 calculates the health status diagnosis rate according to the following formula based on the diagnosis basis;

[0166]

[0167] Where: m ij When the jth signal is used as the basis for judgment, the fault detection model diagnoses the health status of the transmission tower as the ratio of the ith signal, that is, the health status diagnosis rate, M ij N is the number of data classified into the i-th health state by the fault detection model based on the j-th signal, j is the total number of j-th signal data input into the fault detection model.

[0168] The health status assessment module 5 is used to use various health status diagnosis rates as basic probability distribution values, construct an evidence table, and then perform data fusion to obtain the final diagnosis result and complete the assessment of the health status of the transmission tower.

[0169] Further, the health status assessment module 5 assesses the health status of the transmission tower according to the following steps;

[0170] S51, taking the obtained tension signal, torque signal, strain signal, and angle signal as different evidences, and taking the health status diagnosis rate as the basic probability distribution value to construct an evidence table, using the DS evidence reasoning method to perform multi-evidence fusion reasoning, and calculating the fusion probability distribution value; the expression of the fusion probability distribution value is as follows:

[0171]

[0172]

[0173] Where: M(I) is the basic probability distribution value, K is the conflict factor, m1, m2, m3, m4 are different evidences, w, x, y, z are the status categories diagnosed by different evidences, and I is different health status;

[0174] S52. Compare the basic probability distribution values ​​of the health status of different evidences, and select the health status of the transmission tower corresponding to the maximum value as the final diagnosis result.

[0175] Embodiment 3:

[0176] See also Figure 3 , a transmission tower health status assessment device based on multi-source information fusion, the device comprising a processor 6 and a memory 7;

[0177] The memory 7 is used to store computer program code 71 and transmit the computer program code 71 to the processor 6;

[0178] The processor 6 is used to execute the transmission tower health status assessment method based on multi-source information fusion described in Example 1 according to the instructions in the computer program code 71.

[0179] This embodiment also includes a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed on a computer, the method for evaluating the health status of a transmission tower based on multi-source information fusion described in Example 1 is implemented.

[0180] Generally speaking, the computer instructions for implementing the method of the present invention may be carried in any combination of one or more computer-readable storage media. Non-transitory computer-readable storage media may include any computer-readable media, except for the signal itself that is temporarily propagating.

[0181] The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) 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 (EKROM 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 thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, device, or device.

[0182] Computer program code for performing the operation of the present invention can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, SMalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages, in particular, Python suitable for neural network computing and platform frameworks based on TensorFlow, PyTorch, etc. can be used. The program code can be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer or to an external computer (for example, using an Internet service provider to connect via the Internet) through any type of network, including a local area network (LAN) or a wide area network (WAN).

[0183] The above-mentioned device and non-temporary computer-readable storage medium can refer to the specific description of a transmission tower health status assessment method based on multi-source information fusion and its beneficial effects, which will not be repeated here.

[0184] Although the embodiments of the present invention have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A method for evaluating the health status of a transmission tower based on multi-source information fusion, characterized in that: include: S1. Collecting data signals required for transmission tower evaluation; the data signals include tension signals, torque signals, strain signals and angle signals; S2. Extract features from the data signal, obtain time-frequency feature parameters of the data signal, and classify the health status of the transmission tower into different categories; S3. Establish a fault detection model based on an accurate category association algorithm, use time-frequency feature parameters as input to the fault detection model, calculate the state correlation of different categories, and use the category with the largest state correlation as the basis for diagnosis; S4. Based on the diagnosis basis, the diagnosis rates of various health states when different data signals of the transmission tower are obtained as the diagnosis basis are calculated; S5. Use the diagnosis rates of various health states as basic probability distribution values, construct an evidence table, and then perform data fusion to obtain the final diagnosis results and complete the evaluation of the health status of the transmission tower; The step S3 specifically includes: S31, confirming the characteristic dimension of the time-frequency characteristic parameter, different categories of health status and K value; the different categories of health status include four health statuses, namely, level one, level two, level three, and level four; the K value is the number of neighbor samples; S32. Based on the nearest neighbor algorithm, calculate the sum of the average accuracy of K values ​​when K values ​​are 3, 5, and 7 under the five-fold cross validation condition, and record the calculation result as P t ; S33, exclude the i-th feature dimension, then repeat step S32, and record the calculation result as P i ; S34. Calculate the state differentiation capability of different feature dimensions, and the expression is as follows: CDA i =1-(P i -P t ); Among them: CDA i is the state discrimination ability of the i-th feature dimension; S35, normalize the i feature dimensions to obtain the corresponding weight W i , which is expressed as follows: S36. Based on the weights and the ability to distinguish states of different feature dimensions, the Euclidean distance is calculated to optimize the fault detection model. The expression is as follows: Where: D(X1, X2) is the Euclidean distance based on the ability to distinguish feature states, x 1i is the position of the target point in the i-th dimension, x 2i is the position of the training sample point in the i-th dimension; S37, calculate the proportion of the j-th type of points in the neighboring sample points, and the average Euclidean distance between the target point and the j-th type of neighboring sample points, and the expression is as follows: Where: M 1j is the proportion of the jth class of points in the neighboring sample points, K is the number of neighboring samples, N j is the number of samples belonging to the jth class state among the K nearest neighbor samples; Where: M 2j is the average Euclidean distance between the target point and the jth class of neighboring sample points, ΣD(X,X j ) is the sum of the distances from the samples in the jth state to the experimental samples, X is the position of the target point, X j is the position of the jth class of neighboring sample points; S38. Calculate the status correlation of different categories, and select the category with the largest status correlation as the diagnosis basis. The expression is as follows: Among them: CCD j is the state association degree of the jth category.

2. The method for evaluating the health status of a transmission tower based on multi-source information fusion according to claim 1 is characterized in that: The time-frequency characteristic parameters include: mean value, root mean square value, peak value, kurtosis, margin index, pulse index, waveform index, average frequency, center of gravity frequency, and frequency root mean square.

3. The method for evaluating the health status of a transmission tower based on multi-source information fusion according to claim 1 is characterized in that: In step S4, based on the diagnosis basis, the expression for calculating the diagnosis rate of various health states is as follows: Where: m ij M is the ratio of the health status of the transmission tower to the i-th type when the fault detection model uses the j-th data signal as the judgment basis, that is, the health status diagnosis rate, ij is the number of data signals classified into the i-th health state by the fault detection model based on the j-th data signal, N j is the total number of j-th data signals input into the fault detection model.

4. The method for evaluating the health status of a transmission tower based on multi-source information fusion according to claim 1 is characterized in that: The step S5 specifically includes: S51, taking the obtained tension signal, torque signal, strain signal, and angle signal as different evidences, and taking the health status diagnosis rate as the basic probability distribution value to construct an evidence table, using the DS evidence reasoning method to perform multi-evidence fusion reasoning, and calculating the fusion probability distribution value; the expression of the fusion probability distribution value is as follows: Where: M(I) is the basic probability distribution value, K is the conflict factor, m1, m2, m3, m4 are different evidences, w, x, y, z are the status categories diagnosed by different evidences, and I is different health status; S52. Compare the basic probability distribution values ​​of the health status of different evidences, and select the health status of the transmission tower corresponding to the maximum value as the final diagnosis result.

5. A transmission tower health status assessment system based on multi-source information fusion, characterized in that: The system comprises: A data signal acquisition module (1) is used to acquire data signals required for transmission tower evaluation; the data signals include tension signals, torque signals, strain signals and angle signals; A signal feature extraction module (2) is used to extract features from data signals, obtain time-frequency feature parameters of the data signals, and classify the health status of transmission towers into different categories; A fault detection model building module (3) is used to establish a fault detection model based on an accurate category association algorithm, using time-frequency feature parameters as input to the fault detection model, calculating the state correlation of different categories, and using the category with the largest state correlation as the basis for diagnosis; A health status diagnosis rate calculation module (4) is used to calculate various health status diagnosis rates when different data signals of the transmission tower are obtained as the diagnosis basis based on the diagnosis basis; The health status assessment module (5) is used to use the diagnosis rates of various health statuses as basic probability distribution values, construct an evidence table, and then perform data fusion to obtain the final diagnosis result, thereby completing the assessment of the health status of the transmission tower; The fault detection model building module (3) obtains the diagnosis basis according to the following steps; S31, confirming the characteristic dimension of the time-frequency characteristic parameter, different categories of health status and K value; the different categories of health status include four health statuses, namely, level one, level two, level three, and level four; the K value is the number of neighbor samples; S32. Based on the nearest neighbor algorithm, calculate the sum of the average accuracy of the K value in the data set when the K value is 3, 5, and 7 under the five-fold cross validation condition, and record the calculation result as P t ; S33, exclude the i-th feature dimension, then repeat step S32, and record the calculation result as P i ; S34. Calculate the state differentiation capability of different feature dimensions, and the expression is as follows: CDA i =1-(P i -P t ); Among them: CDA i is the state discrimination ability of the i-th feature dimension; S35, normalize the i feature dimensions to obtain the corresponding weight W i , which is expressed as follows: S36. Based on the weights and the ability to distinguish states of different feature dimensions, the Euclidean distance is calculated to optimize the fault detection model. The expression is as follows: Where: D(X1, X2) is the Euclidean distance based on the ability to distinguish feature states, x 1i is the position of the target point in the i-th dimension, x 2i is the position of the training sample point in the i-th dimension; S37, calculate the proportion of the j-th type of points in the neighboring sample points, and the average Euclidean distance between the target point and the j-th type of neighboring sample points, and the expression is as follows: Where: M 1j is the proportion of the jth class of points in the neighboring sample points, K is the number of neighboring samples, N j is the number of samples belonging to the jth class state among the K nearest neighbor samples; Where: M 2j is the average Euclidean distance between the target point and the jth class of neighboring sample points, ΣD(X,X j ) is the sum of the distances from the samples in the jth state to the experimental samples, X is the position of the target point, X j is the position of the jth class of neighboring sample points; S38, calculating the status correlation of different categories, and selecting the category with the largest status correlation as the diagnosis basis, the expression is as follows: Among them: CCD j is the state association degree of the jth category.

6. A transmission tower health status assessment system based on multi-source information fusion according to claim 5, characterized in that: The signal feature extraction module (2) obtains the time-frequency feature parameters of the data signal including: mean value, root mean square value, peak value, kurtosis, margin index, pulse index, waveform index, average frequency, center of gravity frequency, and frequency root mean square.

7. A transmission tower health status assessment system based on multi-source information fusion according to claim 6, characterized in that: The health status diagnosis rate calculation module (4) calculates the health status diagnosis rate based on the diagnosis basis according to the following formula; Where: m ij M is the ratio of the health status of the transmission tower to the i-th type when the fault detection model uses the j-th data signal as the judgment basis, that is, the health status diagnosis rate, ij is the number of data signals classified into the i-th health state by the fault detection model based on the j-th data signal, N j is the total number of j-th data signals input into the fault detection model.

8. The transmission tower health status assessment system based on multi-source information fusion according to claim 7 is characterized in that: The health status assessment module (5) assesses the health status of the transmission tower according to the following steps; S51, taking the obtained tension signal, torque signal, strain signal, and angle signal as different evidences, and taking the health status diagnosis rate as the basic probability distribution value to construct an evidence table, using the DS evidence reasoning method to perform multi-evidence fusion reasoning, and calculating the fusion probability distribution value; the expression of the fusion probability distribution value is as follows: Where: M(I) is the basic probability distribution value, K is the conflict factor, m1, m2, m3, m4 are different evidences, w, x, y, z are the status categories diagnosed by different evidences, and I is different health status; S52. Compare the basic probability distribution values ​​of the health status of different evidences, and select the health status of the transmission tower corresponding to the maximum value as the final diagnosis result.

Images