A fault detection method and a training method and device of a fault classification model

Abstract

The disclosure provides a fault detection method and a training method and device of a fault classification model, relates to the technical field of artificial intelligence, and in particular to the technical field of industrial big data. The specific implementation scheme is: acquiring electric signals of different parts of a to-be-detected long-distance power transmission line in multiple time periods; for each part, calculating performance indexes of the electric signals of the part in different time periods to obtain corresponding performance indexes of the part; constructing index features of the performance indexes corresponding to each part respectively; and determining whether faults occur in each part of the to-be-detected long-distance power transmission line based on the index features of each part. The embodiment of the disclosure can realize fault detection of a long-distance power transmission line.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of artificial intelligence, in particular to the technical field of industrial big data, and more particularly to a fault detection method and a training method and device of a fault classification model. BACKGROUND

[0002] A power transmission line is one of the important infrastructures for long-distance power transmission, which bears the function of power stability, and it is very important to ensure the stability of the power transmission line and quickly locate the fault of the power transmission line. SUMMARY

[0003] The present disclosure provides a fault detection method and a training method and device of a fault classification model.

[0004] According to an aspect of the present disclosure, a fault detection method is provided, comprising:

[0005] Obtaining electrical signals of different parts of a long-distance power transmission line to be detected at different time periods;

[0006] For each part, calculating performance indicators of the electrical signals of the part at different time periods to obtain performance indicators corresponding to the part;

[0007] Respectively constructing indicator features of the performance indicators corresponding to each part;

[0008] Based on the indicator features of each part, determining whether a fault occurs in each part of the long-distance power transmission line to be detected.

[0009] According to another aspect of the present disclosure, a training method of a fault classification model is provided, comprising:

[0010] Obtaining sample electrical signals of different parts of a sample long-distance power transmission line at different time periods, the sample electrical signals having classification labels, the classification labels being used to represent that the sample electrical signals belong to a normal category or a fault category;

[0011] For each part, calculating sample performance indicators of the sample electrical signals of the part at different time periods to obtain sample performance indicators corresponding to the part;

[0012] Respectively constructing sample indicator features of the sample performance indicators corresponding to each part;

[0013] Inputting the sample indicator features of each part into a fault classification model for fault classification to obtain predicted classification labels of each part of the sample long-distance power transmission line at different time periods;

[0014] According to the predicted classification labels of each part of the sample long-distance power transmission line at different time periods and the classification labels of the sample electric signals of each part at different time periods, a current loss is calculated, and a training parameter of the fault classification model is adjusted according to the current loss until a preset ending condition is met, so as to obtain a trained fault classification model.

[0015] According to another aspect of the present disclosure, a fault detection device is provided, comprising:

[0016] A signal acquisition module is configured to acquire electric signals of different parts of a long-distance power transmission line to be detected at multiple time periods.

[0017] An index calculation module is configured to calculate, for each part, a performance index of the electric signals of the part at different time periods, to obtain a corresponding performance index of the part.

[0018] A feature construction module is configured to construct an index feature of the corresponding performance index of each part.

[0019] A fault detection module is configured to determine whether each part of the long-distance power transmission line to be detected has a fault based on the index feature of each part.

[0020] According to another aspect of the present disclosure, a training device of a fault classification model is provided, comprising:

[0021] A sample signal acquisition module is configured to acquire sample electric signals of different parts of a sample long-distance power transmission line at multiple time periods, wherein the sample electric signals have classification labels, and the classification labels are used to represent that the sample electric signals belong to a normal category or a fault category.

[0022] A sample index calculation module is configured to calculate, for each part, a sample performance index of the sample electric signals of the part at different time periods, to obtain a corresponding sample performance index of the part.

[0023] A sample feature construction module is configured to construct a sample index feature of the corresponding sample performance index of each part.

[0024] A sample fault classification module is configured to input the sample index feature of each part into a fault classification model for fault classification, to obtain predicted classification labels of each part of the sample long-distance power transmission line at different time periods.

[0025] A classification model training module is configured to calculate a current loss according to the predicted classification labels of each part of the sample long-distance power transmission line at different time periods and the classification labels of each part of the sample electric signals at different time periods, and to adjust a training parameter of the fault classification model according to the current loss until a preset ending condition is met, so as to obtain a trained fault classification model.

[0026] According to another aspect of the present disclosure, an electronic device is provided, comprising:

[0027] at least one processor; and

[0028] a memory connected with the at least one processor in communication; wherein,

[0029] the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the fault detection method or the training method of the fault classification model according to any one of the present disclosure.

[0030] According to another aspect of the present disclosure, a non-transitory computer readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable the computer to perform the fault detection method or the training method of the fault classification model according to any one of the present disclosure.

[0031] According to another aspect of the present disclosure, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the fault detection method or the training method of the fault classification model according to any one of the present disclosure.

[0032] The embodiments of the present disclosure achieve fault detection of long-distance power transmission lines.

[0033] It should be understood that the content described in this part is not intended to identify key or important features of the embodiments of the present disclosure, nor to limit the scope of the present disclosure. Other features of the present disclosure will become apparent through the following description. BRIEF DESCRIPTION OF DRAWINGS

[0034] The accompanying drawings are used to better understand the present scheme, and do not limit the present disclosure. Among them:

[0035] Figure 1 is a schematic diagram of a fault detection method according to the present disclosure;

[0036] Figure 2 is a schematic diagram of a training method of a fault classification model according to the present disclosure;

[0037] Figure 3 is a schematic diagram of a fault detection device according to the present disclosure;

[0038] Figure 4 is a schematic diagram of a training device of a fault classification model according to the present disclosure;

[0039] Figure 5is a block diagram of an electronic device for implementing a fault detection method or a training method of a fault classification model according to an embodiment of the disclosure. DETAILED DESCRIPTION

[0040] Exemplary embodiments of the present disclosure are described herein below with reference to the accompanying drawings, in which various details of the present disclosure are set forth to facilitate an understanding, and should be considered in a descriptive sense. A person skilled in the art will readily recognize that various alternative embodiments of the present disclosure can be made and implemented without departing from the scope and spirit of the present disclosure. Similarly, it will be appreciated that, for the sake of clarity and conciseness, the description below will omit certain well-known features and descriptions.

[0041] A power transmission line is one of important infrastructures for long-distance power transmission, and it is very important to ensure the stability of the power transmission line and quickly locate the fault of the power transmission line. In actual application, the length of the power transmission line is too long, which is easy to cause low fault locating efficiency, affect the stability of the power transmission system, and further affect production activities and resident life. In the related art, a certain number of sensor devices are installed on the power transmission line, data collected by each sensor is transmitted back to a data control center, and then the collected data is observed by an experienced power engineer to determine whether a fault occurs on the corresponding power transmission line or whether there is a possibility of a fault. However, the manual observation and determination of the long-time power transmission line signal are easy to cause low fault identification efficiency and low accuracy of the power transmission line, and if the fault cannot be identified in time, a huge property loss will be caused.

[0042] In order to realize fault detection of a long-distance power transmission line, an embodiment of the present disclosure provides a fault detection method, which comprises: acquiring electric signals of different parts of a long-distance power transmission line to be detected at different time periods; for each part, calculating performance indicators of the electric signals of the part at different time periods to obtain corresponding performance indicators of the part; constructing indicator features of the performance indicators corresponding to each part; and determining whether faults occur in each part of the long-distance power transmission line to be detected based on the indicator features of each part. In the embodiment of the present disclosure, the electric signals of different parts of the long-distance power transmission line to be detected at different time periods are acquired, the performance indicators of the electric signals of each part at different time periods are calculated, the performance indicators of the electric signals of each part at different time periods are used to describe the signal conditions of different parts of the long-distance power transmission line to be detected at different time periods, direct observation of the electric signals is avoided, the indicator features of the performance indicators corresponding to each part are further constructed, and whether faults occur in each part of the long-distance power transmission line to be detected is determined according to the indicator features of each part, which avoids the problem that manual observation of the electric signals is easy to cause low fault identification efficiency and low accuracy of the power transmission line, can timely find the power transmission line with faults based on the indicator features, so as to give an early warning and avoid causing greater property loss.

[0043] The fault detection method provided by the present disclosure is described in detail below through specific embodiments.

[0044] The fault detection method provided by the embodiments of the present disclosure can be applied to electronic devices such as terminal devices, server devices, and the like. The fault detection method provided by the embodiments of the present disclosure can be applied to application scenarios such as long-distance power transmission line fault detection.

[0045] Referring to Figure 1 , Figure 1 A flowchart of a fault detection method provided by the embodiments of the present disclosure includes the following steps:

[0046] S101, acquiring an electric signal of a long-distance power transmission line at different positions of the long-distance power transmission line in different time periods.

[0047] In one example, corresponding power poles are arranged at different positions of the long-distance power transmission line, and sensors can be arranged on the power poles to collect corresponding electric signals, which can be current signals, voltage signals, and the like.

[0048] For example, when detecting the long-distance power transmission line, current fluctuation data, voltage fluctuation data, and the like collected by sensors at different power poles of the long-distance power transmission line in different time periods can be acquired to achieve detection of the long-distance power transmission line. The different time periods can be time periods divided in units of minutes, or time periods divided in units of hours, and the like.

[0049] S102, for each position, calculating a performance index of the electric signal of the position in different time periods to obtain a corresponding performance index of the position.

[0050] For each position, statistical analysis can be performed on the electric signal data of the position in different time periods, the distribution of the electric signal data of the position in different time periods can be observed, the performance index of the electric signal of the position in different time periods can be calculated, and a corresponding performance index of the position can be obtained. Specifically, the corresponding performance index of the position can be a combination of the performance indexes of the electric signals of the position in different time periods in time sequence.

[0051] In one possible implementation, the performance index can include at least one of a mean value, a variance, a number of local spikes, and a phase jitter amplitude.

[0052] For each part, the electrical signal data of the part at different time periods can be statistically analyzed, and the maximum value, minimum value, mean value, variance and other performance indicators of the electrical signal of the part at each time period can be calculated. The local difference of the electrical signal of the part at different time periods can also be compared to obtain the local spike number and phase jitter amplitude of the electrical signal of the part at different time periods. By comparing the local difference of the electrical signal at different time periods, whether the fault exists in the to-be-detected long-distance power transmission line can be detected from a certain time.

[0053] For example, the electrical signal of a part of the to-be-detected long-distance power transmission line at a time period can be represented as [1, 2, 1, 2, 1, 2, 1, …]. Because the length of the electrical signal is too long and the time index of the fault can occur at any time point of the electrical signal, it is not meaningful to directly observe the electrical signal. Therefore, the corresponding mean value, variance, local spike number and phase jitter amplitude can be obtained by statistically analyzing the electrical signal, which is represented as [1.5, 1, 2, 20]. Then, the performance indicator data [1.5, 1, 2, 20] can be observed instead of observing the electrical signal of the part of the to-be-detected long-distance power transmission line at a time period. The time period can be a data sampling interval, for example, in seconds, 10,000 seconds, 30 minutes or one hour, etc.

[0054] In the embodiments of the present disclosure, for each part, the performance indicators of the electrical signal of the part at different time periods can be calculated to obtain the corresponding performance indicators of the part. The performance indicators can include at least one of the mean value, variance, local spike number and phase jitter amplitude. In this way, the performance indicators of the electrical signal of each part at different time periods can be used instead of the electrical signal of each part at different time periods, direct observation of the electrical signal can be avoided, and the fault of the to-be-detected long-distance power transmission line can be located more quickly.

[0055] S103, index features of the performance indicators corresponding to each part are constructed respectively.

[0056] After obtaining the performance indicators corresponding to each part, in one embodiment, the performance indicators of each part can be sequentially combined to obtain index features. For example, the performance indicators include at least one of the mean value, variance, local spike number and phase jitter amplitude. For each part, the performance indicators at different time periods can be sequentially sorted to form an index feature vector. For example, the performance indicators include the mean value, variance, local spike number and phase jitter amplitude, and the index feature vector at one time period can be represented as [mean value, variance, local spike number, phase jitter amplitude] and the like.

[0057] S104, whether the faults occur in each part of the to-be-detected long-distance power transmission line is determined based on the index features of each part.

[0058] For the index features of each part, the index features can be analyzed, such as feature comparison, to determine whether each part of the to-be-detected long-distance power transmission line has a fault, or a fault detection model and the index features of each part are used to determine whether each part of the to-be-detected long-distance power transmission line has a fault, and the like. The fault detection model can be a pre-trained model that can detect whether each part of the to-be-detected long-distance power transmission line has a fault based on the index features of each part.

[0059] In the embodiments of the present disclosure, the electrical signals of the to-be-detected long-distance power transmission line at different parts and different time periods are obtained, the performance indicators of the electrical signals of each part at different time periods are calculated, the performance indicators of the electrical signals of each part at different time periods are used to describe the signal conditions of the to-be-detected long-distance power transmission line at different parts and different time periods, direct observation of the electrical signals is avoided, and the index features of the performance indicators corresponding to each part are further constructed respectively. Then, whether each part of the to-be-detected long-distance power transmission line has a fault is determined according to the index features of each part, which avoids the problem that manual observation of the electrical signals is prone to cause low efficiency and low accuracy of power transmission line fault identification, and enables timely discovery of a fault power transmission line based on the index features, so as to provide early warning and avoid causing greater property losses.

[0060] In a possible implementation, the step S104 of determining whether each part of the to-be-detected long-distance power transmission line has a fault based on the index features of each part can include:

[0061] The index features of each part are input into a pre-trained deep learning model to obtain a fault detection result of each part of the to-be-detected long-distance power transmission line at a specified time period.

[0062] The pre-trained deep learning model is trained according to sample index features of each part of a sample long-distance power transmission line and classification labels of sample electrical signals. The classification labels are used to represent that the sample electrical signals belong to a normal category or a fault category.

[0063] The index features of each part of the to-be-detected long-distance power transmission line are input into the pre-trained deep learning model as input for fault detection to obtain a fault detection result of each part of the to-be-detected long-distance power transmission line at a specified time period.

[0064] The specified time period can be a current time period, or a preset time period after the current time period. For example, if the time period division unit is 1 minute, the preset time period after the current time period is 1 minute after the current time period. If the time period division unit is 1 hour, the preset time period after the current time period is 1 hour after the current time period.

[0065] In a case where the specified period is a current period, detection is performed on whether a fault occurs in each part of the long-distance power transmission line in the current period; in a case where the specified period is a preset period after the current period, prediction is performed on whether a fault occurs in each part of the long-distance power transmission line in the preset period after the current period.

[0066] In the embodiments of the present disclosure, the index features of each part of the long-distance power transmission line to be detected are input into the pre-trained deep learning model for fault detection, to obtain the fault detection result of each part of the long-distance power transmission line to be detected in the specified period, thereby avoiding the problem that manual observation of the electric signal is prone to cause low efficiency and low accuracy of power transmission line fault identification, and enabling timely discovery of a fault power transmission line based on the index features, so as to provide early warning and avoid causing greater property loss.

[0067] In a possible implementation, the deep learning model can be a lightgbm tree classification model, and the index features can include a plurality of performance index parameters. Correspondingly, the input of the index features of each part into the pre-trained deep learning model to obtain the fault detection result of each part of the long-distance power transmission line to be detected in the specified period includes:

[0068] The index features of each part are input into the pre-trained lightgbm tree classification model to obtain the fault detection result of each performance index parameter of each part in the specified period; and the fault detection result of each part of the long-distance power transmission line to be detected in the specified period is obtained according to the fault detection result of each performance index parameter of each part in the specified period.

[0069] Light Gradient Boosting Machine (lightgbm) is a framework for implementing the GBDT (Gradient Boosting Decision Tree) algorithm, supports efficient parallel training, and has faster training speed, lower memory consumption, better accuracy, supports distribution and can quickly process massive data. The lightgbm tree classification model is a classification model established based on the lightgbm tree model framework.

[0070] In lightgbm, features can be processed in parallel, each working node finds the best split point {feature, threshold}, the working node uses point-to-point communication to find the global best split point, and each working node splits the node according to the global best split point.

[0071] In the embodiments of the present disclosure, the index feature can include multiple performance index parameters. After obtaining the index features of each part, the index features of each part can be input into the pre-trained lightgbm tree classification model to obtain the fault detection result of each performance index parameter in the specified period. Further, the fault detection result of each part of the long-distance power transmission line to be detected in the specified period is obtained.

[0072] For example, the specified period is the current period, and the index feature of one part includes four performance index parameters, i.e., mean, variance, local spike number, and phase jitter amplitude, which are represented as [30, 1000, 1, 345]. The first feature (performance index parameter is mean) corresponds to a threshold of 20, the second feature (performance index parameter is variance) corresponds to a threshold of 900, the third feature (performance index parameter is local spike number) corresponds to a threshold of 3, and the fourth feature (performance index parameter is phase jitter amplitude) corresponds to a threshold of 100. The index feature [30, 1000, 1, 345] of the part is input into the pre-trained lightgbm tree classification model to obtain the probability of the first performance index parameter of the part in the current period being faulty, which is 100% (the mean 30 is greater than the threshold 20), the probability of the second performance index parameter in the current period being faulty is 100% (the variance 1000 is greater than the threshold 900), the probability of the third performance index parameter in the current period being faulty is 0% (the local spike number 1 is not greater than the threshold 3), and the probability of the fourth performance index parameter in the current period being faulty is 100% (the phase jitter amplitude 345 is greater than the threshold 100). According to the fault detection result of each performance index parameter of the part in the current period, the fault detection result of the part of the long-distance power transmission line to be detected in the current period is obtained, which is 75% (the mean of the fault probability of each performance index parameter in the current period), or the fault detection result of the part of the long-distance power transmission line to be detected in the current period can also be the weighted sum or maximum value of the fault probability of each performance index parameter in the current period. Of course, the present disclosure is only used as an example for illustration, and it does not constitute a specific limitation on the embodiments of the present disclosure.

[0073] In the embodiments of the present disclosure, the index features of each part of the long-distance power transmission line to be detected are input into the pre-trained lightgbm tree classification model for fault detection to obtain the fault detection result of each part of the long-distance power transmission line to be detected in the specified period, which avoids the problem that manual observation of electrical signals is prone to cause low efficiency and low accuracy of power transmission line fault identification, and can timely find the faulty power transmission line based on the index feature to provide early warning and avoid causing greater property loss.

[0074] Based on the above fault detection method, refer to Figure 2 , Figure 2A flowchart of a training method of a fault classification model provided by an embodiment of the present disclosure includes the following steps:

[0075] S201, sample electric signals of different parts of the sample long-distance power transmission line at different time periods are obtained.

[0076] The sample electric signals have classification labels, and the classification labels are used to represent that the sample electric signals belong to a normal category or a fault category.

[0077] The normal category is a non-fault category, and the implementation process of step S201 can refer to the implementation process of step S101, which will not be described herein again.

[0078] In a possible implementation, the sample electric signals can include positive sample electric signals and negative sample electric signals of different parts of the sample long-distance power transmission line at different time periods, the positive sample electric signals are electric signals of the normal category, and the negative sample electric signals are electric signals of the fault category.

[0079] In the embodiment of the present disclosure, the electric signals of the normal category are used as the positive sample electric signals, and the electric signals of the fault category are used as the negative sample electric signals, the positive sample electric signals and the negative sample electric signals of different parts of the sample long-distance power transmission line at different time periods are obtained respectively, and further, the positive sample electric signals and the negative sample electric signals are used to train the fault classification model, so that the trained fault classification model is more accurate.

[0080] S202, for each part, a sample performance index of the sample electric signals of the part at different time periods is calculated to obtain a sample performance index corresponding to the part.

[0081] In a possible implementation, the sample performance index can include at least one of a mean value, a variance, a number of local spikes, and a phase jitter amplitude.

[0082] S203, sample index features of the sample performance indexes corresponding to each part are constructed respectively.

[0083] The implementation processes of steps S202-S203 can refer to the implementation processes of steps S102-S103, which will not be described herein again.

[0084] S204, the sample index features of each part are input into the fault classification model for fault classification to obtain predicted classification labels of each part of the sample long-distance power transmission line at different time periods.

[0085] S205, according to the predicted classification labels of each part of the sample long-distance power transmission line at different time periods and the classification labels of each part of the sample electric signal at different time periods, a current loss is calculated, and a training parameter of the fault classification model is adjusted according to the current loss until a preset ending condition is met, and a trained fault classification model is obtained.

[0086] The classification labels of each part of the sample electric signal at different time periods are the classification labels (normal category or fault category) possessed by the sample electric signal. The preset ending condition can be a preset iteration number, or a tree depth, or a loss reaching a preset loss threshold, etc.

[0087] In the embodiments of the present disclosure, the positive sample electric signals and the negative sample electric signals of the sample long-distance power transmission line at different parts and at different time periods are obtained, the sample performance indicators of the sample electric signals at different time periods of each part are calculated, the sample performance indicators of the sample electric signals at different time periods of each part are used to describe the signal conditions of different parts of the sample long-distance power transmission line at different time periods, direct observation of the sample electric signal is avoided, and further, the sample index features of the sample performance indicators corresponding to each part are respectively constructed, the sample index features and the classification labels of each part of the sample electric signal at different time periods are used to train the fault classification model, so that the trained fault classification model can detect whether each part of the to-be-detected long-distance power transmission line has a fault, the problem that manual observation of the electric signal is prone to cause low efficiency and low accuracy of power transmission line fault identification is avoided, and the power transmission line with a fault can be found in time so as to give an early warning and avoid causing greater property loss.

[0088] In a possible implementation, the above fault classification model can be a lightgbm tree classification model, and the above sample index features can include a plurality of sample performance indicator parameters. Correspondingly, the step S204 of inputting the sample index features of each part into the fault classification model for fault classification to obtain the predicted classification labels of each part of the sample long-distance power transmission line at different time periods can include:

[0089] inputting the sample index features of each part into the lightgbm tree classification model for fault classification to obtain the fault detection results of each sample performance indicator parameter of each part at different time periods;

[0090] obtaining the predicted classification labels of each part of the sample long-distance power transmission line at different time periods according to the fault detection results of each sample performance indicator parameter of each part at different time periods.

[0091] The embodiments of the present disclosure provide a fault detection device, referring to Figure 3 , the device comprises:

[0092] The signal acquisition module 301 is configured to acquire electrical signals of different parts of the long-distance power transmission line at different time periods.

[0093] The index calculation module 302 is configured to calculate, for each part, a performance index of the electrical signals of the part at different time periods, to obtain a performance index corresponding to the part.

[0094] The feature construction module 303 is configured to construct an index feature of the performance index corresponding to each part.

[0095] The fault detection module 304 is configured to determine whether each part of the long-distance power transmission line under test has a fault based on the index feature of each part.

[0096] In the embodiments of the present disclosure, the electrical signals of different parts of the long-distance power transmission line under test at different time periods are acquired, the performance index of the electrical signals of each part at different time periods is calculated, the performance index of the electrical signals of each part at different time periods is used to describe the signal conditions of different parts of the long-distance power transmission line under test at different time periods, direct observation of the electrical signals is avoided, the index feature of the performance index corresponding to each part is further constructed, and whether each part of the long-distance power transmission line under test has a fault is determined according to the index feature of each part, thereby avoiding the problem that manual observation of the electrical signals is prone to cause low efficiency and low accuracy of power transmission line fault identification, and the fault power transmission line can be discovered in time based on the index feature, so as to give an early warning and avoid causing greater property loss.

[0097] In a possible implementation, the fault detection module 304 is specifically configured to:

[0098] input the index feature of each part into a pre-trained deep learning model to obtain a fault detection result of each part of the long-distance power transmission line under test at a specified time period, wherein the pre-trained deep learning model is trained according to sample index features of each part of a sample long-distance power transmission line and classification labels of sample electrical signals, and the classification label is used to represent that the sample electrical signal belongs to a normal category or a fault category.

[0099] In a possible implementation, the deep learning model is a lightgbm tree classification model, and the index feature includes a plurality of performance index parameters; and the fault detection module 304 is specifically configured to:

[0100] input the index feature of each part into the pre-trained lightgbm tree classification model to obtain a fault detection result of each performance index parameter of each part at a specified time period, and obtain a fault detection result of each part of the long-distance power transmission line under test at the specified time period according to the fault detection result of each performance index parameter of each part at the specified time period.

[0101] In a possible implementation, the performance indicators include at least one of a mean value, a variance, a number of local spikes, and a phase jitter amplitude.

[0102] The embodiments of the present disclosure further provide a device for training a fault classification model, referring to Figure 4 The device comprises:

[0103] The sample signal acquisition module 401 is configured to acquire sample electric signals of the sample long-distance power transmission line at different parts and in different time periods, and the sample electric signals have classification labels, which are used to represent that the sample electric signals belong to a normal category or a fault category.

[0104] The sample indicator calculation module 402 is configured to calculate, for each part, sample performance indicators of the sample electric signals of the part in different time periods, to obtain sample performance indicators corresponding to the part.

[0105] The sample feature construction module 403 is configured to respectively construct sample indicator features of the sample performance indicators corresponding to each part.

[0106] The sample fault classification module 404 is configured to input the sample indicator features of each part into the fault classification model for fault classification, to obtain predicted classification labels of each part of the sample long-distance power transmission line in different time periods.

[0107] The classification model training module 405 is configured to calculate a current loss according to the predicted classification labels of each part of the sample long-distance power transmission line in different time periods and the classification labels of each part of the sample electric signals in different time periods, and adjust training parameters of the fault classification model according to the current loss until a preset end condition is met, to obtain a trained fault classification model.

[0108] In the embodiments of the present disclosure, positive sample electric signals and negative sample electric signals of the sample long-distance power transmission line at different parts and in different time periods are acquired, sample performance indicators of the sample electric signals of each part in different time periods are calculated, the sample performance indicators of the sample electric signals of each part in different time periods are used to describe signal conditions of the sample long-distance power transmission line at different parts and in different time periods, direct observation of the sample electric signals is avoided, sample indicator features of the sample performance indicators corresponding to each part are further constructed, and the fault classification model is trained by using the sample indicator features and the classification labels of each part of the sample electric signals in different time periods, so that the trained fault classification model can detect whether each part of a to-be-detected long-distance power transmission line has a fault, the problem that manual observation of electric signals is prone to cause low power transmission line fault recognition efficiency and low accuracy is avoided, and a power transmission line with a fault can be found in time, so as to give an early warning and avoid causing greater property loss.

[0109] In a possible implementation, the sample electric signals include positive sample electric signals and negative sample electric signals of different parts of the sample long-distance power transmission line at different time periods, the positive sample electric signals are electric signals of a normal category, and the negative sample electric signals are electric signals of a fault category.

[0110] In a possible implementation, the fault classification model is a lightgbm tree classification model, and the sample index features include a plurality of sample performance index parameters.

[0111] The sample index features of each part are input into the lightgbm tree classification model for fault classification, to obtain fault detection results of each sample performance index parameter of each part at different time periods.

[0112] According to the fault detection results of each sample performance index parameter of each part at different time periods, predicted classification labels of each part of the sample long-distance power transmission line at different time periods are obtained.

[0113] In a possible implementation, the sample performance index includes at least one of a mean value, a variance, a number of local spikes, and a phase jitter amplitude.

[0114] In the technical solution of the present disclosure, the collection, storage, use, processing, transmission, provision, and disclosure of user personal information comply with relevant laws and regulations and do not violate public order and good customs. It should be noted that the head model in the present embodiment is not a head model of a specific user and cannot reflect the personal information of a specific user. It should be noted that the two-dimensional face image in the present embodiment comes from a public data set.

[0115] According to the embodiments of the present disclosure, the present disclosure further provides an electronic device, a readable storage medium, and a computer program product.

[0116] The electronic device includes:

[0117] at least one processor; and

[0118] a memory in communication connection with the at least one processor; wherein

[0119] The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method of any one of the present disclosure.

[0120] A non-transient computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable the computer to execute the method of any one of the present disclosure.

[0121] A computer program product comprising a computer program which, when executed by a processor, implements the method of any of the present disclosure.

[0122] Figure 5 A schematic block diagram of an example electronic device 500 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptops, desktops, tablets, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular telephones, smartphones, wearable devices, and other similar computing devices. The components shown here, their connections and relationships, and their functions, are meant to be examples only, and are not meant to limit implementations of the present disclosure described and / or claimed in this document.

[0123] As shown in Figure 5 The device 500 includes a computing unit 501 that can perform various appropriate actions and processes in accordance with a computer program stored in a read-only memory (ROM) 502 or a computer program loaded into a random access memory (RAM) 503 from a storage unit 508. Various programs and data required for the operation of the device 500 can also be stored in the RAM 503. The computing unit 501, the ROM 502, and the RAM 503 are connected to each other through a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0124] Various components in the device 500 are connected to the I / O interface 505, including an input unit 506, such as a keyboard, a mouse, etc.; an output unit 507, such as various types of displays, speakers, etc.; the storage unit 508, such as a magnetic disk, an optical disk, etc.; and a communication unit 509, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 509 allows the device 500 to exchange information / data with other devices through a computer network, such as the Internet, and / or various telecommunication networks.

[0125] The computing unit 501 can be various general and / or special purpose processing components with processing and computing capabilities. Some examples of the computing unit 501 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing unit 501 performs various methods and processes described above, such as the fault detection method or the training method of the fault classification model. For example, in some embodiments, the fault detection method or the training method of the fault classification model can be implemented as a computer software program tangibly embodied in a machine-readable medium, such as the storage unit 508. In some embodiments, part or all of the computer program can be loaded and / or installed onto the device 500 via the ROM 502 and / or the communication unit 509. When the computer program is loaded onto the RAM 503 and executed by the computing unit 501, one or more steps of the fault detection method or the training method of the fault classification model described above can be performed. Alternatively, in other embodiments, the computing unit 501 can be configured to perform the fault detection method or the training method of the fault classification model by any other appropriate means, such as by means of firmware.

[0126] Various implementations of the systems and techniques described above can be realized in digital electronic circuitry, integrated circuitry, a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), a system on a chip (SOC), a complex programmable logic device (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0127] Program code for carrying out methods of the present disclosure can be written in any combination of one or more programming languages. The program code can be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the program code, when executed by the processor or controller, produces a means for implementing the functions / acts specified in the flowcharts and / or block diagrams. The program code can be executed entirely on a machine, partially on a machine, partially on a machine and partially on a remote machine or entirely on a remote machine or server.

[0128] In the context of the present disclosure, a machine-readable medium can be a tangible medium that contains or stores a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include but is not limited to an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of the machine-readable storage medium will include one or more lines of electrical connections, portable computer disks, hard disk drives, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or Flash memory), optical fibers, portable compact disc read-only memories (CD-ROMs), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0129] To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0130] The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0131] The computer system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0132] It should be understood that the various forms of flow shown above can be re-ordered, added to, or have steps deleted, using the steps described above. For example, the steps described in the present disclosure can be performed in parallel, in series, or in a different order, as long as the desired results of the technology disclosed in the present disclosure can be achieved, which is not limited herein.

[0133] The specific implementation described above does not constitute a limitation on the protection scope of the present disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements, and improvements made within the spirit and principles of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A fault detection method, comprising: obtaining electrical signals of different parts of a long-distance power transmission line at different time periods; for each part, calculating performance indicators of the electrical signals of the part at different time periods to obtain performance indicators corresponding to the part; wherein the performance indicators corresponding to the part are combinations of the performance indicators of the electrical signals of the part at different time periods in time sequence; constructing indicator features of the performance indicators corresponding to each part, including: for each part, sequentially combining the performance indicators of the part to obtain an indicator feature of the part; based on the indicator features of each part, determining whether faults occur in each part of the long-distance power transmission line to be detected.

2. The method of claim 1, wherein, The determination of whether faults occur in each part of the long-distance power transmission line to be detected based on the indicator features of each part comprises: inputting the indicator features of each part into a pre-trained deep learning model to obtain fault detection results of each part of the long-distance power transmission line to be detected at a specified time period; wherein the pre-trained deep learning model is trained according to sample indicator features of each part of a sample long-distance power transmission line and classification labels of sample electrical signals, and the classification labels are used to represent that the sample electrical signals belong to a normal category or a fault category.

3. The method of claim 2, wherein, The deep learning model is a lightgbm tree classification model, and the indicator features include multiple performance indicator parameters. The inputting of the indicator features of each part into the pre-trained deep learning model to obtain the fault detection results of each part of the long-distance power transmission line to be detected at the specified time period comprises: inputting the indicator features of each part into a pre-trained lightgbm tree classification model to obtain fault detection results of each performance indicator parameter of each part at a specified time period; obtaining the fault detection results of each part of the long-distance power transmission line to be detected at the specified time period according to the fault detection results of each performance indicator parameter of each part at the specified time period.

4. The method of claim 1, wherein, The performance indicators include at least one of mean, variance, number of local spikes, and phase jitter amplitude.

5. A training method of a fault classification model, comprising: obtaining sample electrical signals of different parts of a sample long-distance power transmission line at different time periods, wherein the sample electrical signals have classification labels, and the classification labels are used to represent that the sample electrical signals belong to a normal category or a fault category; for each part, calculating sample performance indicators of the sample electrical signals of the part at different time periods to obtain sample performance indicators corresponding to the part; wherein the sample performance indicators corresponding to the part are combinations of the sample performance indicators of the sample electrical signals of the part at different time periods in time sequence; constructing sample indicator features of the sample performance indicators corresponding to each part, including: for each part, sequentially combining the sample performance indicators of the part to obtain a sample indicator feature of the part; inputting the sample indicator features of each part into a fault classification model for fault classification to obtain predicted classification labels of each part of the sample long-distance power transmission line at different time periods; According to the predicted classification labels of each of the parts of the sample long-distance power transmission line at different time periods and the classification labels of each of the parts of the sample electric signal at different time periods, a current loss is calculated, and a training parameter of the fault classification model is adjusted according to the current loss until a preset ending condition is met, so as to obtain the trained fault classification model.

6. The method of claim 5, wherein, The sample electric signal includes positive sample electric signals and negative sample electric signals of different parts of the sample long-distance power transmission line at multiple time periods, the positive sample electric signals are electric signals of a normal category, and the negative sample electric signals are electric signals of a fault category.

7. The method of claim 5, wherein, The fault classification model is a lightgbm tree classification model, and the sample index feature includes multiple sample performance index parameters. The sample index feature of each of the parts is input into the fault classification model to perform fault classification, so as to obtain the predicted classification labels of each of the parts of the sample long-distance power transmission line at different time periods. The sample index feature of each of the parts is input into the lightgbm tree classification model to perform fault classification, so as to obtain the fault detection results of each sample performance index parameter of each of the parts at different time periods. The predicted classification labels of each of the parts of the sample long-distance power transmission line at different time periods are obtained according to the fault detection results of each sample performance index parameter of each of the parts at different time periods.

8. The method of claim 5, wherein, The sample performance index includes at least one of a mean value, a variance, a number of local spikes, and a phase jitter amplitude.

9. A fault detection device, comprising: a signal acquisition module configured to acquire electric signals of different parts of a long-distance power transmission line to be detected at multiple time periods; an index calculation module configured to calculate, for each part, a performance index of the electric signals of the part at different time periods to obtain a performance index corresponding to the part; wherein the performance index corresponding to the part is a combination of the performance indexes of the electric signals of the part at different time periods in time sequence; a feature construction module configured to construct an index feature of the performance index corresponding to each of the parts, respectively; a fault detection module configured to determine whether each of the parts of the long-distance power transmission line to be detected has a fault based on the index features of the parts; the feature construction module is specifically configured to sequentially combine the performance index of each part to obtain the index feature of the part.

10. The apparatus of claim 9, wherein, The fault detection module is specifically configured to: input the index features of the parts into a pre-trained deep learning model to obtain fault detection results of each part of the long-distance power transmission line to be detected at a specified time period; wherein the pre-trained deep learning model is trained according to sample index features of each part of a sample long-distance power transmission line and classification labels of a sample electric signal, and the classification labels are used to represent that the sample electric signal belongs to a normal category or a fault category.

11. The apparatus of claim 10, wherein, The deep learning model is a lightgbm tree classification model, and the index feature includes multiple performance index parameters. The fault detection module is specifically configured to: inputting the index features of each of the parts into a pre-trained lightgbm tree classification model to obtain a fault detection result of each performance index parameter of each of the parts in a specified time period; and obtaining a fault detection result of each part of the to-be-detected long-distance power transmission line in the specified time period according to the fault detection result of each performance index parameter of each of the parts in the specified time period.

12. The apparatus of claim 9, wherein, The performance index includes at least one of a mean value, a variance, a local spike number and a phase jitter amplitude.

13. A device for training a fault classification model, comprising: a sample signal acquisition module configured to acquire sample electric signals of different parts of a sample long-distance power transmission line in multiple time periods, the sample electric signals having classification labels, the classification labels being used to represent that the sample electric signals belong to a normal category or a fault category; a sample index calculation module configured to calculate, for each part, sample performance indexes of the sample electric signals of the part in different time periods to obtain sample performance indexes corresponding to the part; wherein the sample performance indexes corresponding to the part are combinations of the sample performance indexes of the sample electric signals of the part in different time periods in time sequence; a sample feature construction module configured to construct sample index features of the sample performance indexes corresponding to each of the parts respectively; a sample fault classification module configured to input the sample index features of each of the parts into a fault classification model to perform fault classification, and obtain predicted classification labels of each of the parts of the sample long-distance power transmission line in different time periods; a classification model training module configured to calculate a current loss according to the predicted classification labels of each of the parts of the sample long-distance power transmission line in different time periods and classification labels of each of the parts of the sample electric signals in different time periods, and adjust training parameters of the fault classification model according to the current loss until a preset end condition is met to obtain a trained fault classification model; the sample feature construction module is specifically configured to sequentially combine the sample performance indexes of each part to obtain sample index features of the part.

14. The apparatus of claim 13, wherein, The sample electric signals include positive sample electric signals and negative sample electric signals of different parts of a sample long-distance power transmission line in multiple time periods, the positive sample electric signals being electric signals of a normal category, and the negative sample electric signals being electric signals of a fault category.

15. The apparatus of claim 13, wherein, The fault classification model is a lightgbm tree classification model, and the sample index features include multiple sample performance index parameters; the sample fault classification module is specifically configured to: input the sample index features of each of the parts into the lightgbm tree classification model to perform fault classification, and obtain fault detection results of each sample performance index parameter of each of the parts in different time periods; obtain predicted classification labels of each of the parts of the sample long-distance power transmission line in different time periods according to the fault detection results of each sample performance index parameter of each of the parts in different time periods.

16. The apparatus of claim 13, wherein, The sample performance index includes at least one of a mean value, a variance, a local spike number and a phase jitter amplitude.

17. An electronic device, comprising: at least one processor; and a memory connected with the at least one processor in communication; wherein, The memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-8.

18. A non-transitory computer readable storage medium having stored thereon computer instructions, wherein, The computer instructions are for causing the computer to perform the method of any one of claims 1-8.

19. A computer program product comprising a computer program which, when executed by a processor, implements the method of any one of claims 1-8.

Images