CN111860241A - Power equipment discharge fault identification method based on wavelet packet analysis - Google Patents
Power equipment discharge fault identification method based on wavelet packet analysis Download PDFInfo
- Publication number
- CN111860241A CN111860241A CN202010647020.3A CN202010647020A CN111860241A CN 111860241 A CN111860241 A CN 111860241A CN 202010647020 A CN202010647020 A CN 202010647020A CN 111860241 A CN111860241 A CN 111860241A
- Authority
- CN
- China
- Prior art keywords
- sound
- wavelet packet
- packet analysis
- discharge fault
- node
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
- G06F2218/06—Denoising by applying a scale-space analysis, e.g. using wavelet analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Signal Processing (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Investigating Or Analyzing Materials By The Use Of Ultrasonic Waves (AREA)
Abstract
The invention belongs to the technical field of power equipment fault identification, and discloses a power equipment discharge fault identification method based on wavelet packet analysis. The feature vector extraction is carried out on the sound sample by a wavelet packet analysis method, the characteristic of low high-frequency resolution of a commonly-used Mel feature vector is overcome, the recognition performance is good, the recognition of the discharge fault is completed by combining machine learning, and the fault monitoring and recognition are realized by utilizing the non-contact mode.
Description
Technical Field
The invention relates to the technical field of power equipment fault identification, in particular to a power equipment discharge fault identification method based on wavelet packet analysis.
Background
With the progress of times and the development of science and technology, the power industry has been developed greatly, and the electrical industry also becomes the prop industry of various countries' economy, and provides basic power for various industries. Various power equipment is widely applied, such as transformers, reactors and the like are key carriers for safe and stable operation of a power system, and equipment faults and health conditions directly relate to normal operation of the power system and reliable power supply of users. Often, the failure of the power equipment easily causes great economic loss and even possibly causes safety accidents. For example, in the case of the 'great blackout' accident in 2003, the experts estimate that the economically visible loss caused by the accident is $ 250 to $ 300 billion, at least 8 people die directly, and the indirect economic loss and social impact are even immeasurable. Therefore, the most basic requirement of the power system is that the power system can be stably and safely operated for a long time, but various faults are generated in the operation process of the power equipment, and compared with the serious consequences caused by the faults, the monitoring and identifying system for better developing the power equipment is much more cost-effective, so that the monitoring and identifying of the power equipment faults is developed into an important branch in the discipline of the power system.
However, most common methods for monitoring equipment faults need to directly measure the electric quantity of equipment in operation to identify the faults, which inevitably increases the complexity of a power system, and many detection methods need to be power failure detection, which is time-consuming and labor-consuming and high in cost.
Because the equipment can emit various sounds in the operation process, the information of the operation condition of the equipment can be obtained from the change and the difference of the sounds, and whether the equipment has faults or not and what faults occur can be judged. For example, if the transformer is in a normal operation state, the transformer will generate a slight "buzz", and the sound is regularly heavy, which is generated by the vibration of the iron core, and when the transformer generates an irregular abnormal sound, it represents that the transformer has a fault, and different fault abnormal sounds are different. For example: when the transformer sometimes makes a "pungency" sound like water boiling, the short-circuit current may occur due to turn-to-turn short circuit, so that the transformer oil is abnormally heated, and thus has a sound of boiling; if an irregular, occasional "gouty" sound occurs, it is likely due to core grounding problems, which should be checked for transformer deactivation to avoid a re-escalation of the fault.
However, at present, the fault of the transformer is mainly identified by hearing the sound by an experienced technician. Although the operation difficulty is not great, the method has high requirements on the experience of workers, cannot be popularized and further cannot realize real-time monitoring, which is not in accordance with the development requirements of the society. In addition, because a large amount of abundant working experience is needed for judging whether the fault occurs, the accuracy is greatly reduced.
Disclosure of Invention
The invention provides a method for identifying the discharge fault of power equipment based on wavelet packet analysis, which solves the problems that the existing sound identification of transformer faults only depends on the experience of workers, the accuracy is low, the real-time monitoring cannot be realized, and the like.
The invention can be realized by the following technical scheme:
a method for identifying a discharge fault of power equipment based on wavelet packet analysis comprises the steps of collecting a sound sample of the power equipment to be identified, preprocessing the sound sample, extracting a feature vector of the preprocessed sound sample through the wavelet packet analysis method, and finally identifying the discharge fault of the power equipment to be identified through machine learning.
And further, carrying out wavelet packet decomposition on each frame of sound signals of the preprocessed sound samples by using a wavelet packet analysis method to obtain wavelet coefficient groups corresponding to each node, then calculating energy groups corresponding to each node, and carrying out normalization processing, wherein the energy groups corresponding to each node after the normalization processing are the feature vectors.
Further, a pywavelets toolkit in python software is adopted to carry out n-layer wavelet packet decomposition on each frame of sound signal to obtain a wavelet coefficient group corresponding to each node, and the jth wavelet coefficient of the ith node is recorded as xi,jThe number of nodes is 2nThe number of wavelet coefficients corresponding to each node is equal to the number of sampling points N divided by the number of nodes, and the energy group of the ith node is EiThen, there are:
then, normalization processing is performed by using the following equation, and the node energy group after normalization processing is marked as Ei'，
Then the normalized energy per node group E 'will be'1、E'2… … as characteristic parameters, form a 2nFeature vectors of the dimension.
Furthermore, the collected sound samples are subjected to mute shearing, then the rest sound samples are subjected to normalization processing, then the sound samples subjected to normalization processing are subjected to pre-emphasis processing, and finally the sound samples subjected to pre-emphasis processing are subjected to framing and windowing processing, so that the sound samples are pre-processed.
Further, a limit is set to the average energy E of the whole sound sample signalmeanAnd 40% of the total volume, cutting the silence below the limit from the whole sound sample, performing zero-mean normalization processing on the residual sound sample data, and then performing pre-emphasis processing on the sound sample data after the normalization processing, wherein the transfer formula of the pre-emphasis filter is set as: h (z) ═ 1- α z -1And taking alpha as 0.9, dividing the pre-emphasized sound sample into a plurality of frames of sound signals with equal length by adopting an overlapped sliding frame dividing method, and then performing windowing processing, thereby completing the pre-processing of the sound sample.
Further, sound samples are collected and preprocessed under a normal working condition and a discharge fault working condition respectively, feature vectors of the preprocessed sound samples are extracted by a wavelet packet analysis method respectively, the extracted feature vectors are used as a database, two different working conditions are respectively assigned with 1 and-1 and are used as a training set of a support vector machine to build a training model, and discharge fault recognition of power equipment to be recognized is achieved by machine learning.
The beneficial technical effects of the invention are as follows:
the feature vector extraction is carried out on the sound sample by a wavelet packet analysis method, the characteristic of low high-frequency resolution of a commonly-used Mel feature vector is overcome, the recognition performance is good, the recognition of the discharge fault is completed by combining machine learning, and the fault monitoring and recognition are realized by utilizing the non-contact mode. The discharge fault identification method based on the sound signals does not need to rely on the experience of workers to judge, greatly reduces practical risk and cost, and can overcome the defects that manual inspection cannot be carried out in real time, the experience requirement on detection personnel is high, the influence of subjective factors is caused, the working environment of inspection personnel is severe, the economic cost and the time cost are high, and the like.
Drawings
FIG. 1 is a schematic overall flow diagram of the present invention;
FIG. 2 is a schematic diagram of a wavelet packet decomposition tree according to the present invention;
FIG. 3 is a schematic diagram of the band distribution of wavelet packet decomposition according to the present invention;
FIG. 4 is a spectrogram schematic of a first frame of the present invention;
FIG. 5 is a diagram illustrating the result of the four-layer wavelet packet decomposition of the present invention;
FIG. 6 is a feature vector after normalization according to the present invention.
Detailed Description
The following detailed description of the preferred embodiments will be made with reference to the accompanying drawings.
At present, on-line monitoring systems of power equipment such as transformers, switches and the like are all on-line monitoring devices based on contact sensors, and the on-line monitoring systems are possible risk points, particularly secondary on-line monitoring systems integrated by non-original manufacturers, have the problems of poor stability and poor monitoring effect, and cannot find equipment faults in time; and the non-contact monitoring mode has image, sound and infrared modes, wherein the sound mode has the lowest cost and has the advantages that: (1) the sound is easy to collect, and the sound is in a non-contact type, so that the information collection stage is considered to be absolutely safe, and the cost is lower than that of an image and infrared mode; (2) the large-range continuous real-time monitoring is realized, and the equipment abnormality is judged at first time with high possibility mainly depending on the precision of a sound identification technology; (3) the technology based on sound diagnosis is complementary with the intelligent diagnosis technology of the existing electric equipment, so that the practical risk and cost can be greatly reduced by using the method.
As shown in fig. 1, the invention provides a method for identifying a discharge fault of an electrical device based on wavelet packet analysis, which is characterized in that a sound signal in the operation process of the electrical device is collected by a sensor to obtain a sound sample, and the sound sample is trained by a machine learning method to monitor the operation state of the device in real time and warn the fault condition. Firstly, the sensor for acquiring the sound signal can be arranged on the shell of the machine and does not need to be part of the power system, so that the equipment does not need to be stopped for acquiring the sound signal; secondly, the optical fiber sensor is adopted to collect signals, so that electromagnetic interference can be shielded, and convenience is provided for signal processing in the later period; thirdly, preprocessing the collected sound sample, extracting the feature vector of the preprocessed sound sample by a wavelet packet analysis method, and finally, recognizing the discharge fault of the power equipment to be recognized by machine learning.
The specific process is as follows:
step 1, performing mute clipping on the collected sound sample, wherein the purpose is to take out a mute section contained in natural recording, because the mute section only contains noise and does not contain useful information, the average energy of the mute section is far lower than that of the whole sound sample signal, 40% is taken as a boundary, and the calculation formula is as follows:
Wherein, L represents the number of sample points, and wave _ data represents the sample point data.
Assuming that the length of the acquired sound sample segment is 0.5s, the calculation formula of the average energy Ev of the segment is:
And 3, performing pre-emphasis processing on the data obtained after the normalization processing, wherein the loss of a high-frequency signal is more serious than that of a low-frequency signal when a sound signal is transmitted through an air channel, the distortion of the high-frequency signal is compensated by adopting the pre-emphasis processing, and the transmission formula of a pre-emphasis filter is as follows:
H(z)＝1-αz-1wherein α is 0.9.
And 4, framing the pre-emphasized sound sample, wherein the sound signal needs to be divided into a plurality of small sound segments, namely sound frames, by using some methods to process each sound frame in order to apply the conventional method for processing the stable random signal to the processing of the sound signal due to the short-time stationarity of the sound signal. The invention adopts the overlapping sliding frame dividing method, and the frame dividing method is used for enabling the transition between adjacent sound frames to be smoother and keeping the continuous characteristic of the sound signal. The sliding between adjacent sound frames is called frame shift, and for the discharge fault sound, the length of the sound frame, namely the frame length, is found to be 0.5s, and 1/10 of the frame shift frame length is reasonable. Through frame processing, the audio signal is finally divided into a plurality of frames of sound signals with equal length, so that input units with equal data number are obtained, then windowing is performed, spectrum leakage generated at the position where each frame of signal is truncated is solved, and a Hamming window can be selected for windowing.
Step 5, utilizing a wavelet packet analysis method to carry out n-layer wavelet packet decomposition on each frame of sound signals, and obtaining the jth wavelet coefficient of the ith node as xi,jThe number of the nodes is 2nThe number of wavelet coefficients of each node is equal to the number of sampling points N divided by the number of nodes, and if the energy group of the ith node is recorded as EiThen, there are:
step 6, normalizing the energy group of each node after wavelet packet decomposition, wherein the energy group after normalization is recorded as Ei', the formula is as follows:
thereby obtaining the energy group E of each node after the normalization processing1'、E'2……。
As shown in fig. 3, it can be seen that each node after wavelet packet decomposition represents time domain parameters of different frequency bands, so the node energy obtained through the above process represents energy of different frequency bands. Due to sound under discharge conditionsThe energy of the system is greatly different from the frequency band distribution and the normal working condition: generally speaking, the proportion of the low-frequency part of the sound of the power equipment under the normal working condition is high, the high-frequency part is less, the proportion of the high-frequency part under the discharge fault is far higher than that under the normal working condition, the specific frequency distribution subdivided by the high-frequency part is also different, therefore, the node energy obtained after decomposing the wavelet packet can be used as a characteristic parameter to carry out sound identification, and according to the steps, 2 is extracted nFeature vectors of the dimension.
Finally, the feature vectors extracted from the sound samples of the normal working condition and the discharge fault working condition by the method are used as a database, the values of 1 and-1 are respectively assigned to two different working conditions, and the two different working conditions are used as a training set of a support vector machine to build a training model, so that the discharge fault sound can be identified.
Taking the first frame as an example, firstly, 20000 sample points are taken as one frame at a sampling frequency of 40000Hz, that is, the content of 0.5s, 2000 sample points are taken for frame shift, and finally about 6000 sound frames are obtained, and feature vectors are extracted for the sound frames respectively, and the first frame is taken as an example below, and fig. 4 is a sound spectrogram thereof.
After the preprocessing is finished, a sample is decomposed by adopting a pywavelet packet in python, db5 is used as a basic wavelet function to decompose 5 layers of wavelet packets, the wavelet packet decomposition tree and the frequency band distribution are shown in figures 2 and 3, the specific decomposition result is shown in figure 5, the wavelet coefficient of each node after the wavelet packet decomposition is obtained, only the wavelet packet decomposition result of the first four layers is shown due to the resolution, the node energy is calculated by the wavelet coefficient of each node obtained by the wavelet packet decomposition of the fifth layer, then normalization processing is carried out, the feature vectors of the first frame before and after normalization are shown in figure 6, the node energy is used as the feature vector, and the feature vector has 32 dimensions.
The feature vectors extracted from the sound samples are used as a training set of a support vector machine to build a training model, 20% of data are selected as a test set, and the test success rate is shown in the following table:
wherein, gamma is the parameter of the kernel function, C is the penalty factor, and both are the internal parameters of the support vector machine, and it can be seen that under the appropriate parameters, the success rate of the identification can reach 96%, and the effect is relatively ideal.
Although specific embodiments of the present invention have been described above, it will be appreciated by those skilled in the art that these are merely examples and that many variations or modifications may be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is therefore defined by the appended claims.
Claims (6)
1. A method for identifying a discharge fault of power equipment based on wavelet packet analysis is characterized by comprising the following steps: collecting a sound sample of the power equipment to be identified, preprocessing the sound sample, extracting a feature vector of the preprocessed sound sample by a wavelet packet analysis method, and finally identifying the discharge fault of the power equipment to be identified by machine learning.
2. The electrical equipment discharge fault identification method based on wavelet packet analysis according to claim 1, characterized by: and performing wavelet packet decomposition on each frame of sound signals of the preprocessed sound samples by using a wavelet packet analysis method to obtain wavelet coefficient groups corresponding to each node, then calculating energy groups corresponding to each node, and performing normalization processing, wherein the energy groups corresponding to each node after the normalization processing are the feature vectors.
3. The electrical equipment discharge fault identification method based on wavelet packet analysis according to claim 2, characterized by: adopting a pywavelets toolkit in python software to carry out n-layer wavelet packet decomposition on each frame of sound signal to obtain a wavelet coefficient group corresponding to each node, and recording the jth wavelet coefficient of the ith node as xi,jThe number of nodes is 2nNumber of wavelet coefficients corresponding to each nodeEqual to the number of sampling points N divided by the number of nodes, the energy group of the ith node is EiThen, there are:
then, normalization processing is performed by the following equation, and the node energy group after normalization processing is denoted as E'i，
Then the normalized energy per node group E 'will be'1、E'2… … as characteristic parameters, form a 2nFeature vectors of the dimension.
4. The electrical equipment discharge fault identification method based on wavelet packet analysis according to claim 1, characterized by: the method comprises the steps of carrying out mute cutting on collected sound samples, then carrying out normalization processing on the rest sound samples, then carrying out pre-emphasis processing on the sound samples subjected to the normalization processing, and finally carrying out framing and windowing processing on the sound samples subjected to the pre-emphasis processing, thereby completing the pre-processing on the sound samples.
5. The electrical equipment discharge fault identification method based on wavelet packet analysis according to claim 4, characterized by: setting a limit to the average energy E of the whole sound sample signalmeanAnd 40% of the total volume, cutting the silence below the limit from the whole sound sample, performing zero-mean normalization processing on the residual sound sample data, and then performing pre-emphasis processing on the sound sample data after the normalization processing, wherein the transfer formula of the pre-emphasis filter is set as: h (z) ═ 1- α z-1And finally, dividing the pre-emphasized sound sample into a plurality of frames of sound signals with equal length by adopting an overlapped sliding framing method, and then performing windowing processing, thereby finishing the pre-processing of the sound sample.
6. The electrical equipment discharge fault identification method based on wavelet packet analysis according to claim 1, characterized by: respectively acquiring and preprocessing sound samples under a normal working condition and a discharge fault working condition, respectively extracting feature vectors of the preprocessed sound samples by a wavelet packet analysis method, respectively assigning 1 and-1 to two different working conditions by taking the extracted feature vectors as a database, and building a training model by taking the two different working conditions as a training set of a support vector machine, and realizing discharge fault recognition of power equipment to be recognized by machine learning.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010647020.3A CN111860241A (en) | 2020-07-07 | 2020-07-07 | Power equipment discharge fault identification method based on wavelet packet analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010647020.3A CN111860241A (en) | 2020-07-07 | 2020-07-07 | Power equipment discharge fault identification method based on wavelet packet analysis |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111860241A true CN111860241A (en) | 2020-10-30 |
Family
ID=73153166
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010647020.3A Pending CN111860241A (en) | 2020-07-07 | 2020-07-07 | Power equipment discharge fault identification method based on wavelet packet analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111860241A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112562698A (en) * | 2020-12-02 | 2021-03-26 | 国网山西省电力公司大同供电公司 | Power equipment defect diagnosis method based on fusion of sound source information and thermal imaging characteristics |
| CN117233589A (en) * | 2023-09-14 | 2023-12-15 | 中国南方电网有限责任公司超高压输电公司广州局 | GIS equipment fault diagnosis method and device, computer equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108470570A (en) * | 2018-01-23 | 2018-08-31 | 浙江大学山东工业技术研究院 | Motor abnormal sound detection method |
| CN108844725A (en) * | 2018-04-24 | 2018-11-20 | 天津大学 | A kind of automobile engine bearing wear fault diagnosis method |
| CN109357749A (en) * | 2018-09-04 | 2019-02-19 | 南京理工大学 | A kind of power equipment audio signal analysis method based on DNN algorithm |
| US20200200648A1 (en) * | 2018-02-12 | 2020-06-25 | Dalian University Of Technology | Method for Fault Diagnosis of an Aero-engine Rolling Bearing Based on Random Forest of Power Spectrum Entropy |
-
2020
- 2020-07-07 CN CN202010647020.3A patent/CN111860241A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108470570A (en) * | 2018-01-23 | 2018-08-31 | 浙江大学山东工业技术研究院 | Motor abnormal sound detection method |
| US20200200648A1 (en) * | 2018-02-12 | 2020-06-25 | Dalian University Of Technology | Method for Fault Diagnosis of an Aero-engine Rolling Bearing Based on Random Forest of Power Spectrum Entropy |
| CN108844725A (en) * | 2018-04-24 | 2018-11-20 | 天津大学 | A kind of automobile engine bearing wear fault diagnosis method |
| CN109357749A (en) * | 2018-09-04 | 2019-02-19 | 南京理工大学 | A kind of power equipment audio signal analysis method based on DNN algorithm |
Non-Patent Citations (1)
| Title |
|---|
| 李春雷;董志学;王新杰;: "发电机声音检测与故障诊断研究", 内蒙古工业大学学报(自然科学版), no. 03, pages 2 - 3 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112562698A (en) * | 2020-12-02 | 2021-03-26 | 国网山西省电力公司大同供电公司 | Power equipment defect diagnosis method based on fusion of sound source information and thermal imaging characteristics |
| CN112562698B (en) * | 2020-12-02 | 2022-11-04 | 国网山西省电力公司大同供电公司 | Power equipment defect diagnosis method based on fusion of sound source information and thermal imaging characteristics |
| CN117233589A (en) * | 2023-09-14 | 2023-12-15 | 中国南方电网有限责任公司超高压输电公司广州局 | GIS equipment fault diagnosis method and device, computer equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109357749B (en) | DNN algorithm-based power equipment audio signal analysis method | |
| CN102426835B (en) | Method for identifying local discharge signals of switchboard based on support vector machine model | |
| CN109740523B (en) | Power transformer fault diagnosis method based on acoustic features and neural network | |
| CN111860241A (en) | Power equipment discharge fault identification method based on wavelet packet analysis | |
| CN112329914B (en) | Fault diagnosis method and device for buried transformer substation and electronic equipment | |
| CN112660745A (en) | Intelligent diagnosis method and system for carrier roller fault and readable storage medium | |
| CN112420055A (en) | Substation state identification method and device based on voiceprint characteristics | |
| CN112599134A (en) | Transformer sound event detection method based on voiceprint recognition | |
| CN116778964A (en) | Power transformation equipment fault monitoring system and method based on voiceprint recognition | |
| CN113257249A (en) | Power equipment fault diagnosis method, device and equipment based on voiceprint recognition | |
| CN115687969A (en) | Low-voltage transformer fault diagnosis method based on sound characteristic analysis | |
| CN113345399A (en) | Method for monitoring sound of machine equipment in strong noise environment | |
| CN112052712B (en) | Power equipment state monitoring and fault identification method and system | |
| CN115376526A (en) | Power equipment fault detection method and system based on voiceprint recognition | |
| CN117289067B (en) | Transformer running state on-line monitoring system | |
| CN115910097A (en) | Audible signal identification method and system for latent fault of high-voltage circuit breaker | |
| CN117763460A (en) | Transformer fault detection method and system based on voiceprint analysis | |
| CN114157023B (en) | Distribution transformer early warning information acquisition method | |
| CN111933186A (en) | Method, device and system for fault identification of on-load tap-changer | |
| CN114295195B (en) | Abnormality judgment method and system for optical fiber sensing vibration signals based on feature extraction | |
| CN115062867A (en) | Bearing fault prediction method based on Teager energy spectrum and improved CNN model | |
| CN114121025A (en) | Voiceprint fault intelligent detection method and device for substation equipment | |
| CN112850408A (en) | Elevator emergency stop trapped person fault detection method based on multi-model fusion | |
| CN113343887A (en) | Multi-sensor mixed fault signal blind separation method based on edge calculation and machine learning | |
| CN116416975A (en) | Port equipment operation and maintenance fault sound monitoring method based on collaborative neural network algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |