CN112631235A - Iron tower remote monitoring and fault diagnosis system based on improved SOM network - Google Patents
Iron tower remote monitoring and fault diagnosis system based on improved SOM network Download PDFInfo
- Publication number
- CN112631235A CN112631235A CN202011328673.1A CN202011328673A CN112631235A CN 112631235 A CN112631235 A CN 112631235A CN 202011328673 A CN202011328673 A CN 202011328673A CN 112631235 A CN112631235 A CN 112631235A
- Authority
- CN
- China
- Prior art keywords
- iron tower
- signals
- signal
- training
- 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
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 title claims abstract description 134
- 229910052742 iron Inorganic materials 0.000 title claims abstract description 67
- 238000012544 monitoring process Methods 0.000 title claims abstract description 23
- 238000003745 diagnosis Methods 0.000 title claims abstract description 13
- 230000005540 biological transmission Effects 0.000 claims abstract description 31
- 238000003860 storage Methods 0.000 claims abstract description 12
- 238000006073 displacement reaction Methods 0.000 claims abstract description 7
- 238000012549 training Methods 0.000 claims description 30
- 238000013528 artificial neural network Methods 0.000 claims description 29
- 239000013598 vector Substances 0.000 claims description 19
- 238000000034 method Methods 0.000 claims description 17
- 238000012545 processing Methods 0.000 claims description 14
- 238000004422 calculation algorithm Methods 0.000 claims description 13
- 238000012360 testing method Methods 0.000 claims description 13
- 230000006870 function Effects 0.000 claims description 12
- 210000002569 neuron Anatomy 0.000 claims description 12
- 238000000354 decomposition reaction Methods 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 8
- 238000004458 analytical method Methods 0.000 claims description 7
- 210000004027 cell Anatomy 0.000 claims description 6
- 238000013507 mapping Methods 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 6
- 238000003064 k means clustering Methods 0.000 claims description 5
- 239000011248 coating agent Substances 0.000 claims description 4
- 238000000576 coating method Methods 0.000 claims description 4
- 230000002159 abnormal effect Effects 0.000 claims description 3
- 230000002860 competitive effect Effects 0.000 claims description 3
- 238000013500 data storage Methods 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 3
- 238000007689 inspection Methods 0.000 claims description 3
- 238000011056 performance test Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 230000001419 dependent effect Effects 0.000 claims description 2
- 230000004304 visual acuity Effects 0.000 claims description 2
- 238000001514 detection method Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 3
- 238000013459 approach Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 208000025274 Lightning injury Diseases 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000035515 penetration Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
- G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
- G05B23/0259—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterized by the response to fault detection
- G05B23/0262—Confirmation of fault detection, e.g. extra checks to confirm that a failure has indeed occurred
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/20—Pc systems
- G05B2219/24—Pc safety
- G05B2219/24065—Real time diagnostics
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
- Y04S10/52—Outage or fault management, e.g. fault detection or location
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
A remote iron tower monitoring and fault diagnosis system based on an improved SOM network comprises a plurality of iron tower systems, a plurality of remote transport stations and a control room, wherein the iron tower systems are connected with the remote transport stations nearby through short-distance wireless transmission equipment, and the remote transport stations are connected with the control room through the long-distance wireless transmission equipment; the iron tower system comprises a plurality of sensors, a solar cell panel, a storage battery and a tail end collecting device, wherein the sensors comprise stress sensors and displacement sensors and are responsible for collecting stress or displacement signals of each node of the iron tower and sending the signals to the tail end collecting device in a wired transmission mode, and the solar cell panel and the storage battery are responsible for supplying power to the device. The invention improves the performance of the system.
Description
Technical Field
The invention belongs to the field of remote monitoring and fault monitoring, and relates to an iron tower remote monitoring and fault diagnosis system based on an improved SOM network.
Background
Since 1879, China began the construction of transmission lines. The iron tower of the power transmission line is an important part of the power transmission line, the safety of the iron tower of the power transmission line is directly related to the reliability of power supply of a power grid, and the national production construction, the social life order and even the life and property safety are influenced. China is vast, and power transmission lines connecting power supply ends and user ends often need to span various areas with complex meteorological conditions, so that various natural environments and meteorological loads pose great threats to the safety and stability of line iron towers. With the increasing installed capacity and voltage grade of national power grids in recent years, power transmission towers are continuously developed towards high rise, large span and extra-high voltage directions, and higher requirements are provided for the reliability and economy of the towers.
Mutual penetration and deep fusion of the smart grid and the Internet of things are inevitable results of the development of the information communication technology to a certain stage, communication infrastructure resources and electric power infrastructure resources can be effectively integrated, the telecommunication level is improved, and the utilization efficiency of the existing electric power infrastructure is improved. At present, the Internet of things is widely applied in various fields, but the safety research on the electric power iron tower only considers the effects of factors such as earthquake, lightning stroke, ice coating and the like in the design stage, and no corresponding standard and system is established for the safety detection and health evaluation of the built power transmission tower, and no mature detection measures and health evaluation scheme are provided.
Disclosure of Invention
The invention provides a system for remotely monitoring and diagnosing a power tower and a fault based on an improved SOM network, aiming at realizing the intelligent evaluation and diagnosis of the running state and the energy efficiency of the power tower and the online real-time monitoring and intelligent management of fault early warning.
The technical scheme adopted by the invention for solving the technical problems is as follows:
a remote iron tower monitoring and fault diagnosis system based on an improved SOM network comprises a plurality of iron tower systems, a plurality of remote transport stations and a control room, wherein the iron tower systems are connected with the remote transport stations nearby through short-distance wireless transmission equipment, and the remote transport stations are connected with the control room through the long-distance wireless transmission equipment;
the iron tower system comprises a plurality of sensors, a solar cell panel, a storage battery and a tail end collecting device, wherein the sensors comprise stress sensors and displacement sensors and are responsible for collecting stress or displacement signals of each node of the iron tower and sending the signals to the tail end collecting device in a wired transmission mode, and the solar cell panel and the storage battery are responsible for supplying power to the device.
Further, the signals collected in the iron tower are transmitted to a nearby remote transport station to be converted into 5G signals, and then the signals are transmitted to a control room to be processed.
Further, the iron tower system collects signals every 5 minutes, if the collected signals are identified as abnormal signals, the system is in an activated state, and the system in the activated state collects the signals every one second; if the signal returns to normal, the acquisition time of the equipment also returns to 5 minutes, otherwise, the equipment alarms.
The remote transportation station comprises a power supply module, a data acquisition module, a data storage module and a short-distance wireless transmission module, wherein the short-distance wireless transmission module processes electric signals into 5G signals so as to be transmitted, the power supply module is responsible for supplying power to the whole transportation station, and the data acquisition and storage module respectively acquires and stores system signals.
The control room comprises a database, a monitoring center and an expert center, wherein the monitoring center carries out a series of processing of receiving, noise reduction and filtering of systems of all remote transport stations and sends the processed signals to the expert center; and the expert center combines a large amount of data to carry out reasoning and judge the running state of the iron tower, thereby realizing the real-time monitoring of each power transmission iron tower.
The expert center adopts an improved SOM self-organizing feature mapping network, adopts a variable learning rate calculation, and the initial weight of the network is determined by a K-Means clustering algorithm.
Further, the processing procedure of the expert center is as follows:
(1) signal processing
Respectively simulating a plurality of failure modes of the electric power iron tower which can possibly fail under the working conditions of strong wind, ice coating and line breaking, and collecting failure signals when the failure occurs; wherein the fault signal includes the stress of bolt on the key position, the dependent variable of key member, the inclination of iron tower, to measuring the voltage signal that iron tower inclination and output, then need utilize wavelet packet to decompose and carry out time frequency analysis and extract the characteristic value, wavelet packet analysis can be decomposed into a series of wavelet functions that have local characteristic to the signal, all has fine resolving power in low frequency and high frequency range, the time that has adjustable window, the local ability of partial book of frequency, the process is as follows:
a) firstly, carrying out n-layer wavelet packet decomposition on the collected signals, and respectively extracting the nth layer from low frequency to high frequency 2nWavelet packet coefficient at each node, 2nEach node is (i, j) which represents the jth node of the ith layer, where i is n and j is 0,1,2,3 …,2n-1；
b) Reconstructing wavelet packet decomposition coefficient, and extracting signal characteristics of each frequency band range
Let each node wavelet packet coefficient Hi，jThe corresponding reconstructed signal is Si，jAll nodes of the nth layer are analyzed, and the total signal S is represented by the following formula:
c) calculating the total energy of each frequency band signal
Suppose Sn，j(j-0, 1,2,3 …,2n-1) corresponding energy En，j(j is 0,1,2,3 …,2n-1), the energy Sn，jRepresented by the formula:
wherein: h isj,k(j-0, 1,2,3 …,2 n-1; k-1, 2, …, n) represents the reconstructed signal Sn，jThe amplitude of the discrete points of (a);
d) constructing feature vectors
Defining the total energy of the voltage signal as
The relative wavelet packet energy of a certain frequency band is
Then the relative wavelet packet energy eigenvector is
(2) The method comprises the following steps of establishing, training and testing an improved self-organizing feature mapping neural network:
2.1) data preprocessing
Taking the fault characteristic component, the stress of a bolt on a key part and the strain of a key rod piece as input, outputting the input into the judgment of safe operation of the electric power iron tower, and carrying out normalization processing on an input training sample, wherein a normalization equation is as follows:
k＝(x-xmin)/(xmax-xmin)
where k is the normalized value, x is the normalized data, xmin、xmaxRespectively, the minimum value and the maximum value in the normalized data;
2.2) neural network parameter setting
Setting parameters of a neural network, including maximum iteration times epochs, a topological structure, a distance calculation function dist, neighborhood reduction step numbers step and an initial neighborhood IN;
2.3) training neural networks
Inputting a training set to a designated neural network, calculating a center distance to initialize a weight of a first layer, calculating a dot product of the weight W and an input vector X and calculating an Euclidean distance, wherein a value corresponding to a node with the minimum distance is the maximum, finding a winning node through a competitive neuron, updating the weight through a calculated neighborhood and a learning rate, and then repeating the training process until the precision meets a designated requirement or the maximum training frequency is reached, and stopping training;
2.4) training completion
And (3) finishing training by the SOM neural network, inputting the test set to perform performance test, and if the test set meets the precision requirement, using the test set for actual engineering inspection.
According to the system, a simulation model for electric power iron tower safety condition assessment, fault early warning and diagnosis based on big data and artificial intelligence is provided by acquiring, monitoring and analyzing the stress condition of bolts on key parts of an electric tower, and the system has great practical significance and great market value for improving the safety and reliability of an electric power system.
The invention develops an artificial neural network system for on-line monitoring of the running state and the fault of the power iron tower by means of a large amount of actual running data and failure modes of the iron tower, a running state and energy efficiency simulation algorithm of equipment, failure judgment basis and expert knowledge, collects and analyzes the collected signals, and finally realizes the real-time monitoring of the safety of the iron tower.
The invention has the following beneficial effects:
(1) the initial weight of the conventional SOM neural network is difficult to determine, hidden layer neurons are often not fully utilized, and some neurons far away from a learning vector cannot win, so that the neurons become dead nodes. The novel SOM network introduces a K-Means clustering algorithm, calculates the central vector of the whole sample in advance, and superposes a small random number on the basis of the central vector as the initial value of the weight vector, thereby greatly improving the performance of the system.
(2) The algorithm adopts a learning rate changing mode, so that the global searching speed of the system in the early stage is improved, and the local searching capability of the system in the later stage can be ensured.
Drawings
Fig. 1 is a tree structure diagram of wavelet packet 3-layer decomposition.
FIG. 2 is a topological structure diagram of the SOM neural network.
Fig. 3 is an algorithm flow chart.
Fig. 4 is a system flow diagram.
Detailed Description
The invention is further described below with reference to the accompanying drawings.
Referring to fig. 4, a remote iron tower monitoring and fault diagnosis system based on an improved SOM network is composed of a plurality of iron tower systems, a plurality of remote transportation stations and a control room, wherein the plurality of iron tower systems are connected with the nearby remote transportation stations through short-distance wireless transmission equipment, and the remote transportation stations are connected with the control room through long-distance wireless transmission equipment.
The signals collected in the iron tower are transmitted to a nearby remote transport station to be converted into 5G signals, and then the signals are transmitted to a control room to be processed.
Furthermore, the iron tower system consists of a plurality of sensors, a solar cell panel, a storage battery and a tail end acquisition system, wherein the sensors are connected with tail end transmission equipment, the storage battery sensors comprise a stress sensor and a displacement sensor and are used for acquiring parameters of the iron tower in working in real time, and the solar cell panel receives solar energy and generates electric energy which is stored by the storage battery so as to ensure that the system supplies power; and the tail end acquisition system collects signals of a plurality of measuring points of the iron tower and transmits the signals to a nearby remote transportation station.
Further, the control process collects signals every 5 minutes, and if the collected signals are identified as abnormal signals, the system will be in an activated state, and the system in the activated state will collect signals every one second. If the signal returns to normal, the acquisition time of the equipment also returns to 5 minutes, otherwise, the equipment alarms. The staff will carry out the maintenance of stopping operation to the accident iron tower.
Furthermore, the remote transportation station comprises a power module, a data acquisition module, a data storage module and a short-distance wireless transmission module, wherein the power module is responsible for supplying power to the whole transportation station, the data acquisition and storage module is used for acquiring and storing system signals respectively, and the short-distance wireless transmission module is used for processing electric signals into 5G signals so as to facilitate transmission. The remote transportation station is used for receiving nearby electric power iron tower signals, processing the signals, transmitting the processed signals to a control room through 5G signals with high transmission capacity.
Furthermore, the control room also comprises a database, a monitoring center and an expert center, wherein the monitoring center sends a series of processing such as receiving, noise reduction, filtering and the like of the system of each remote transportation station to the expert center. And the expert center combines a large amount of data to carry out reasoning and judge the running state of the iron tower, thereby realizing the real-time monitoring of each power transmission iron tower.
Further, the expert center employs an improved SOM self-organizing feature mapping network. The algorithm adopts a variable learning rate calculation, and the initial weight of the network is determined by a K-Means clustering algorithm; through testing, the algorithm has higher precision and calculation efficiency, and can ensure the stable monitoring of the system.
The artificial neural network fault intelligent diagnosis system of the electric power iron tower comprises: processing signals, constructing a neural network, and diagnosing faults by using the constructed neural network, wherein the details are as follows:
(1) signal processing
Respectively simulating a plurality of failure modes of the electric power iron tower which can possibly fail under working conditions of strong wind, ice coating, line breaking and the like, and collecting failure signals when the failure occurs; the fault signal comprises the stress of a bolt on a key part, the strain of a key rod piece and the inclination angle of the iron tower, and aiming at a voltage signal output by measuring the inclination angle of the iron tower, a wavelet packet analysis is needed to perform time-frequency analysis to extract a characteristic value, the wavelet packet analysis can decompose the signal into a series of wavelet functions with local characteristics, the wavelet functions have good resolution in low-frequency and high-frequency ranges, and the wavelet functions have time-frequency local breaking capacity of an adjustable window, and the structure of the fault signal is shown in figure 1, and the specific process is as follows.
a) Firstly, carrying out n-layer wavelet packet decomposition on the collected signals, and respectively extracting the nth layer from low frequency to high frequency 2nWavelet packet coefficient at each node, 2nEach node is (i, j) which represents the jth node of the ith layer, where i is n and j is 0,1,2,3 …,2n1, e.g. (0,0) node representing the original signal S, and (1,0) node representing the layer 1 low frequency coefficients H of the wavelet packet decomposition1，0The (1,1) node represents the high-frequency coefficient H of the layer 1 of wavelet packet decomposition1，1And so on in turn;
b) reconstructing wavelet packet decomposition coefficient, and extracting signal characteristics of each frequency band range
Let each node wavelet packet coefficient Hi，jThe corresponding reconstructed signal is Si，jAll nodes of the nth layer are analyzed, and the total signal S is represented by the following formula:
c) calculating the total energy of each frequency band signal
Suppose Sn，j(j-0, 1,2,3 …,2n-1) corresponding energy En，j(j is 0,1,2,3 …,2n-1), the energy Sn，jRepresented by the formula:
wherein: h isj,k(j-0, 1,2,3 …,2 n-1; k-1, 2, …, n) represents the reconstructed signal Sn，jThe amplitude of the discrete points of (a);
d) constructing feature vectors
Defining the total energy of the voltage signal as
The relative wavelet packet energy of a certain frequency band is
Then the relative wavelet packet energy eigenvector is
(2) Creation, training and testing of improved self-organizing feature-mapped neural networks
Self-organizing feature mapping neural networks (SOMs), proposed by the netherlands scholars Kohonen in 1981, are also referred to as Kohonen networks. Compared with the huge calculation amount of the RBF and the complexity of solving a linear equation set by using the Gaussian two-multiplication, the SOM network is a guiding-free learning neural network with simple structure and wide application, and the neurons of the SOM network are fully connected. Therefore, the network can learn the distribution characteristics of the input quantity, can also learn the topological structure of the input quantity, and realizes the function of a nonlinear learning algorithm. Currently, SOM networks are most widely used in the clustering problem.
The problem of iron tower fault detection is an unbalanced problem, that is, when a signal is monitored, misjudgment for judging a fault mode as a non-fault mode is more serious than the condition for judging the non-fault mode as the fault mode, and for an iron tower, the fault probability is very small but the influence is very large. Therefore, the modeling performance of the system is better improved, and how to monitor the iron tower in the first time when the iron tower does not work normally is realized.
On the basis of the traditional probabilistic neural network, the density function estimation and the Bayes decision theory are fused, so that the judgment boundary approaches to the Bayes optimal judgment surface.
The performance of the SOM neural network depends strongly on the value of the initial weights. The traditional method adopts random setting, which can fully disperse the weight vectors in a sample space, but in some occasions, the samples are integrally concentrated in some local areas of the space, the weight vectors are distinguished in wide areas of the sample space, the weight vectors close to the whole sample group are adjusted during training, and the vectors far away from the sample group cannot be adjusted. Therefore, in order to improve the performance of the system, in the present invention, a K-Means clustering algorithm (proposed by MacQueen in 1967) is introduced for clustering SOM reference vectors. The algorithm randomly defines k clustering centers, and then, each data point points to a cluster closest to the average value of the data point to initialize the weight, so that part of neurons are prevented from being not strengthened in the operation process;
the mode layer is fully connected with the input layer, no connection exists in the mode layer, and the largest node is found out as a winning node through the dot product of the input vector and the weight in each training. If the input is not normalized, the Euclidean distance is calculated firstly through the following formula, and the value with the minimum distance is selected as the winner node;
wherein d isjThe Euclidean distance is obtained, X is an input vector, W is a weight, and m is the number of the input vectors;
the system updates the weight of all nodes in the neighborhood of the selected node, the initial neighborhood size step can be initially set, and the neighborhood can continuously shrink along with continuous iteration. The winner-point adjustment formula is as follows:
wherein, wij(t) weight of neuron i at time j;
α (t, N) is the learning rate of neuron update, and is a function of the topological distance between the ith neuron and the winning neuron in a neighborhood;
the learning rate of the conventional algorithm is not changed, so that the operation is slow if the learning rate is low in the training process of the neural network; if the learning rate is large, the optimal solution can be easily skipped. The best approach is to initially make the learning rate larger and then decrease it at a faster rate, which is beneficial to quickly capture the approximate structure of the input vector, and the learning rate function is as follows:
wherein, C2Is a constant between 0 and 1; t is the number of iterations, tmIs the maximum iteration number;
the system flow chart is shown in fig. 3, and the steps are as follows:
2.1) data preprocessing
Taking the fault characteristic component, the stress of a bolt on a key part and the strain of a key rod piece as input, outputting the input into the judgment of safe operation of the electric power iron tower, and carrying out normalization processing on an input training sample, wherein a normalization equation is as follows:
k＝(x-xmin)/(xmax-xmin)
where k is the normalized value, x is the normalized data, xmin、xmaxRespectively, the minimum value and the maximum value in the normalized data;
2.2) neural network parameter setting
Setting parameters of a neural network, including maximum iteration times epochs, a topological structure, a distance calculation function dist, neighborhood reduction step numbers step and an initial neighborhood IN;
2.3) training neural networks
Inputting a training set to a designated neural network, calculating a center distance to initialize a weight of a first layer, calculating a dot product of the weight W and an input vector X and calculating an Euclidean distance, wherein a value corresponding to a node with the minimum distance is the maximum, finding a winning node through a competitive neuron, updating the weight through a calculated neighborhood and a learning rate, and then repeating the training process until the precision meets a designated requirement or the maximum training frequency is reached, and stopping training;
2.4) training completion
And (3) finishing training by the SOM neural network, inputting the test set to perform performance test, and if the test set meets the precision requirement, using the test set for actual engineering inspection.
Claims (7)
1. A remote iron tower monitoring and fault diagnosis system based on an improved SOM network is characterized by comprising a plurality of iron tower systems, a plurality of remote transportation stations and a control room, wherein the iron tower systems are connected with the remote transportation stations nearby through short-distance wireless transmission equipment, and the remote transportation stations are connected with the control room through the long-distance wireless transmission equipment;
the iron tower system comprises a plurality of sensors, a solar cell panel, a storage battery and a tail end collecting device, wherein the sensors comprise stress sensors and displacement sensors and are responsible for collecting stress or displacement signals of each node of the iron tower and sending the signals to the tail end collecting device in a wired transmission mode, and the solar cell panel and the storage battery are responsible for supplying power to the device.
2. The system for remotely monitoring and diagnosing the iron tower based on the improved SOM network as claimed in claim 1, wherein the signals collected in the iron tower are transmitted to a nearby remote transportation station, converted into 5G signals and transmitted to a control room for processing.
3. The improved SOM network-based remote monitoring and fault diagnosis system for iron tower according to claim 1 or 2, wherein the iron tower system collects signals once every 5 minutes, if the collected signals are identified as abnormal signals, the system will be in an activated state, and the system in the activated state will collect signals once every second; if the signal returns to normal, the acquisition time of the equipment also returns to 5 minutes, otherwise, the equipment alarms.
4. The system for remotely monitoring and diagnosing the iron tower based on the improved SOM network as claimed in claim 1 or 2, wherein the remote transportation station comprises a power module, a data acquisition module, a data storage module and a short-distance wireless transmission module, the short-distance wireless transmission module processes an electric signal into a 5G signal for transmission, the power module is responsible for supplying power to the whole transportation station, and the data acquisition and storage module respectively acquires and stores the system signal.
5. The improved SOM network-based remote monitoring and fault diagnosis system for iron towers according to claim 1 or 2, wherein the control room comprises a database, a monitoring center and an expert center, and the monitoring center sends a series of processes of receiving, denoising and filtering of the system of each remote transportation station to the expert center; and the expert center combines a large amount of data to carry out reasoning and judge the running state of the iron tower, thereby realizing the real-time monitoring of each power transmission iron tower.
6. The system of claim 5, wherein the expert center uses an improved SOM self-organizing feature mapping network, a learning rate-varying calculation is used, and initial weights of the network are determined by a K-Means clustering algorithm.
7. The system for remote monitoring and fault diagnosis of iron tower based on improved SOM network as claimed in claim 6, wherein the processing procedure of the expert center is:
(1) signal processing
Respectively simulating a plurality of failure modes of the electric power iron tower which can possibly fail under the working conditions of strong wind, ice coating and line breaking, and collecting failure signals when the failure occurs; wherein the fault signal includes the stress of bolt on the key position, the dependent variable of key member, the inclination of iron tower, to measuring the voltage signal that iron tower inclination and output, then need utilize wavelet packet to decompose and carry out time frequency analysis and extract the characteristic value, wavelet packet analysis can be decomposed into a series of wavelet functions that have local characteristic to the signal, all has fine resolving power in low frequency and high frequency range, the time that has adjustable window, the local ability of partial book of frequency, the process is as follows:
a) firstly, carrying out n-layer wavelet packet decomposition on the collected signals, and respectively extracting the nth layer from low frequency to high frequency 2nWavelet packet coefficient at each node, 2nEach node is (i, j) which represents the jth node of the ith layer, where i is n and j is 0,1,2,3 …,2n-1；
b) Reconstructing wavelet packet decomposition coefficient, and extracting signal characteristics of each frequency band range
Let each node wavelet packet coefficient Hi，jThe corresponding reconstructed signal is Si，jAll nodes of the nth layer are analyzed, and the total signal S is represented by the following formula:
c) calculating the total energy of each frequency band signal
Suppose Sn，j(j-0, 1,2,3 …,2n-1) corresponding energy En，j(j is 0,1,2,3 …,2n-1), the energy Sn，jRepresented by the formula:
wherein: h isj,k(j-0, 1,2,3 …,2 n-1; k-1, 2, …, n) represents the reconstructed signal Sn，jThe amplitude of the discrete points of (a);
d) constructing feature vectors
Defining the total energy of the voltage signalIs composed of
The relative wavelet packet energy of a certain frequency band is
Then the relative wavelet packet energy eigenvector is
(2) The method comprises the following steps of establishing, training and testing an improved self-organizing feature mapping neural network:
2.1) data preprocessing
Taking the fault characteristic component, the stress of a bolt on a key part and the strain of a key rod piece as input, outputting the input into the judgment of safe operation of the electric power iron tower, and carrying out normalization processing on an input training sample, wherein a normalization equation is as follows:
k＝(x-xmin)/(xmax-xmin)
where k is the normalized value, x is the normalized data, xmin、xmaxRespectively, the minimum value and the maximum value in the normalized data;
2.2) neural network parameter setting
Setting parameters of a neural network, including maximum iteration times epochs, a topological structure, a distance calculation function dist, neighborhood reduction step numbers step and an initial neighborhood IN;
2.3) training neural networks
Inputting a training set to a designated neural network, calculating a center distance to initialize a weight of a first layer, calculating a dot product of the weight W and an input vector X and calculating an Euclidean distance, wherein a value corresponding to a node with the minimum distance is the maximum, finding a winning node through a competitive neuron, updating the weight through a calculated neighborhood and a learning rate, and then repeating the training process until the precision meets a designated requirement or the maximum training frequency is reached, and stopping training;
2.4) training completion
And (3) finishing training by the SOM neural network, inputting the test set to perform performance test, and if the test set meets the precision requirement, using the test set for actual engineering inspection.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011328673.1A CN112631235A (en) | 2020-11-24 | 2020-11-24 | Iron tower remote monitoring and fault diagnosis system based on improved SOM network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011328673.1A CN112631235A (en) | 2020-11-24 | 2020-11-24 | Iron tower remote monitoring and fault diagnosis system based on improved SOM network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112631235A true CN112631235A (en) | 2021-04-09 |
Family
ID=75303674
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011328673.1A Pending CN112631235A (en) | 2020-11-24 | 2020-11-24 | Iron tower remote monitoring and fault diagnosis system based on improved SOM network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112631235A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2001004691A (en) * | 1999-06-21 | 2001-01-12 | Furukawa Electric Co Ltd:The | Failure zone locating system |
| CN103926526A (en) * | 2014-05-05 | 2014-07-16 | 重庆大学 | Analog circuit fault diagnosis method based on improved RBF neural network |
| CN109491816A (en) * | 2018-10-19 | 2019-03-19 | 中国船舶重工集团公司第七六研究所 | Knowledge based engineering method for diagnosing faults |
| CN109959820A (en) * | 2017-12-24 | 2019-07-02 | 广州奈英科技有限公司 | Monitor the intelligent Detection and its method of steel tower real-time status |
-
2020
- 2020-11-24 CN CN202011328673.1A patent/CN112631235A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2001004691A (en) * | 1999-06-21 | 2001-01-12 | Furukawa Electric Co Ltd:The | Failure zone locating system |
| CN103926526A (en) * | 2014-05-05 | 2014-07-16 | 重庆大学 | Analog circuit fault diagnosis method based on improved RBF neural network |
| CN109959820A (en) * | 2017-12-24 | 2019-07-02 | 广州奈英科技有限公司 | Monitor the intelligent Detection and its method of steel tower real-time status |
| CN109491816A (en) * | 2018-10-19 | 2019-03-19 | 中国船舶重工集团公司第七六研究所 | Knowledge based engineering method for diagnosing faults |
Non-Patent Citations (1)
| Title |
|---|
| 田瑶瑶等: "基于小波包和SOM神经网络的电作动器故障诊断", 《应用科技》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107941537B (en) | A kind of mechanical equipment health state evaluation method | |
| CN109546659B (en) | Power distribution network reactive power optimization method based on random matrix and intelligent scene matching | |
| CN109763944B (en) | Non-contact monitoring system and monitoring method for blade faults of offshore wind turbine | |
| CN103728535B (en) | A kind of extra-high-voltage direct-current transmission line fault location based on wavelet transformation transient state energy spectrum | |
| CN108665112A (en) | Photovoltaic fault detection method based on Modified particle swarm optimization Elman networks | |
| CN111680875B (en) | Unmanned aerial vehicle state risk fuzzy comprehensive evaluation method based on probability baseline model | |
| CN104809722A (en) | Electrical device fault diagnosis method based on infrared thermography | |
| CN102867132A (en) | Aviation direct-current converter online fault combined prediction method based on fractional order wavelet transformation | |
| CN116937579B (en) | Wind power interval prediction considering space-time correlation and interpretable method thereof | |
| CN110866448A (en) | Flutter signal analysis method based on convolutional neural network and short-time Fourier transform | |
| CN112116160B (en) | Important power transmission channel disaster monitoring method based on improved cellular automaton of optimized neural network | |
| CN104318485A (en) | Power transmission line fault identification method based on nerve network and fuzzy logic | |
| CN114169445A (en) | Day-ahead photovoltaic power prediction method, device and system based on CAE and GAN hybrid network | |
| CN113360555A (en) | Fault diagnosis and analysis method and system based on big data of power grid secondary equipment | |
| CN114252103A (en) | Fusion power station operation fault prediction method | |
| CN102087311B (en) | Method for improving measurement accuracy of power mutual inductor | |
| Lu et al. | Event-triggered federated learning for fault diagnosis of offshore wind turbines with decentralized data | |
| Wang et al. | A wind speed prediction method based on improved empirical mode decomposition and support vector machine | |
| CN112631235A (en) | Iron tower remote monitoring and fault diagnosis system based on improved SOM network | |
| CN107069710B (en) | Power system state estimation method considering new energy space-time correlation | |
| CN116187968A (en) | Substation equipment state evaluation method and system, related equipment and medium | |
| Xiaoqian | A sensor fault diagnosis algorithm for uav based on neural network | |
| CN113323823B (en) | AWKELM-based fan blade icing fault detection method and system | |
| Lu et al. | Regional Forecast of Heavy Precipitation and Interpretability Based on TD-VAE | |
| CN118473910B (en) | Electric power Internet of things fault detection method and system based on edge cloud cooperation |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210409 |
|
| RJ01 | Rejection of invention patent application after publication |