CN113641486A - Intelligent turnout fault diagnosis method based on edge computing network architecture - Google Patents
Intelligent turnout fault diagnosis method based on edge computing network architecture Download PDFInfo
- Publication number
- CN113641486A CN113641486A CN202110757298.0A CN202110757298A CN113641486A CN 113641486 A CN113641486 A CN 113641486A CN 202110757298 A CN202110757298 A CN 202110757298A CN 113641486 A CN113641486 A CN 113641486A
- Authority
- CN
- China
- Prior art keywords
- data
- cloud layer
- fault
- edge
- fault diagnosis
- 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.)
- Granted
Links
- 238000003745 diagnosis Methods 0.000 title claims abstract description 73
- 238000000034 method Methods 0.000 title claims abstract description 63
- 238000007781 pre-processing Methods 0.000 claims abstract description 21
- 238000013528 artificial neural network Methods 0.000 claims abstract description 18
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 18
- 238000003062 neural network model Methods 0.000 claims description 32
- 238000012549 training Methods 0.000 claims description 23
- 238000012545 processing Methods 0.000 claims description 16
- 238000013500 data storage Methods 0.000 claims description 13
- 230000001965 increasing effect Effects 0.000 claims description 10
- 238000000605 extraction Methods 0.000 claims description 9
- 238000012544 monitoring process Methods 0.000 claims description 9
- 238000005516 engineering process Methods 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 5
- 230000011218 segmentation Effects 0.000 claims description 5
- 230000009471 action Effects 0.000 claims description 4
- 238000006243 chemical reaction Methods 0.000 claims description 4
- 230000005540 biological transmission Effects 0.000 claims description 3
- 230000002194 synthesizing effect Effects 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 claims description 2
- 238000012423 maintenance Methods 0.000 abstract description 10
- 238000004422 calculation algorithm Methods 0.000 abstract description 7
- 238000012821 model calculation Methods 0.000 abstract description 4
- 238000004364 calculation method Methods 0.000 description 9
- 238000010586 diagram Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5061—Partitioning or combining of resources
- G06F9/5072—Grid computing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24133—Distances to prototypes
- G06F18/24143—Distances to neighbourhood prototypes, e.g. restricted Coulomb energy networks [RCEN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2209/00—Indexing scheme relating to G06F9/00
- G06F2209/50—Indexing scheme relating to G06F9/50
- G06F2209/502—Proximity
Abstract
The invention discloses a turnout intelligent fault diagnosis method based on an edge computing network architecture, which comprises an access layer, an edge cloud layer and a center cloud layer. The access layer sensor receives turnout current curve data and sends the turnout current curve data to the edge cloud layer for fault diagnosis; after receiving the current curve data, the edge cloud layer performs data preprocessing, and then performs fault diagnosis by using a CNN fault diagnosis model transmitted by the center cloud layer; and the central cloud layer generates a large number of data samples by SMOTE and GAN methods according to a small amount of historical data, trains a CNN model and transmits the CNN model to the edge cloud layer. The central cloud layer receives the preprocessed data of the edge cloud layer to replace a generated sample to train the CNN model; the method solves the problems of unbalanced data samples, complex model calculation of an artificial intelligent neural network fault diagnosis algorithm, occupation of cloud computing bandwidth, insufficient computing capacity of edge nodes and the like in the prior art, and improves the stability and the efficiency of the prior intelligent operation and maintenance system.
Description
Technical Field
The invention belongs to the technical field of rail transit informatization intelligence, and particularly relates to an intelligent turnout fault diagnosis method based on an edge computing network architecture.
Background
With the arrival of the 5G era, high-density, high-strength and high-speed railway operation is brought, and how to ensure the safety and efficiency of train operation has gradually become a research hotspot. The turnout serves as an important component of a railway signal system, and the main function of the turnout is to change the running direction of a train. The turnout is frequently used and the working environment is severe, so that the turnout is a vulnerable device, and once a fault occurs, the normal operation of the railway is seriously influenced. At present, the stability and the safety of a turnout system in China are mainly monitored in real time by a simple alarm threshold device, namely, parameters such as working current, power and the like at a certain moment exceed a threshold value, which indicates that the turnout is abnormal. However, the method has no timeliness, accuracy and intelligence, and cannot find the fault timely and accurately, so that the fault type and the solution can be obtained intelligently, great inconvenience is brought to maintenance personnel, and huge hidden danger is brought to railway safety operation.
With the continuous progress and development of modern signal processing technology, some turnout fault feature extraction and diagnosis methods are proposed according to historical data and working states of turnout switch machines. The method is characterized by processing from the qualitative aspect and the quantitative aspect, and researching turnout fault diagnosis by using methods such as feature extraction, data quantification, current popular machine learning and neural network. However, two problems in the existing method cannot be solved better, firstly, the sample data is not distributed uniformly, and the data volume of the normal curve acquired by the microcomputer monitoring system is far larger than that of the fault curve, so that the data has serious imbalance. And secondly, aiming at the turnout fault diagnosis of the neural network algorithm, the existing current curve data is used for carrying out feature extraction and neural network training, and the problems that the accuracy of a neural network model is not enough under the condition of small data quantity, the training time is too long under the condition of large data quantity, the model calculation is complex and the like exist.
SMOTE is a technology for synthesizing a few-class oversampling, and the algorithm mainly synthesizes a new few-class sample through a part of the few-class samples, and controls the quantity and the distribution condition of the new generated sample, thereby realizing the balance of a plurality of classes of samples and a few classes of samples. The GAN is a generation confrontation network model, and mutual confrontation and game between two networks are achieved through a generator G and a judger D, so that samples generated by the generator have the capability of being falsified. The SMOTE and GAN method can expand the fault data set to balance the number of fault samples and normal samples.
Cloud computing is an effective way for efficiently processing a large amount of data, and can be used as a neural network turnout fault diagnosis network architecture for processing a large amount of data. However, the cloud server is usually deployed at a fixed position, sometimes far away from the terminal, and as the coverage of the railway is further increased, the bandwidth load of the cloud computing network will have a serious influence on the stability of the whole operation and maintenance system, so that it is difficult for a simple cloud computing network architecture to meet the increasing requirements of the railway operation and maintenance system.
The edge calculation is to dispose a server with calculation and storage capabilities at the edge of the network to perform service processing on the edge terminal. The most important characteristic is that the edge node is close to the edge terminal physically, and compared with the traditional cloud computing mode, the characteristic can effectively reduce delay, save energy efficiency, be beneficial to privacy protection, reduce bandwidth occupation, improve system instantaneity and the like. For the railway operation and maintenance system, edge calculation can complete a large number of service processing processes in a local edge layer without being processed by a cloud, so that the service processing efficiency of the whole operation and maintenance system is greatly improved, and the cloud load is reduced. And the mobile terminal is closer to the train and the like, so that the response is faster, and the occurrence of faults is reduced to ensure the safe operation of the train.
Therefore, based on an edge computing network architecture, combined with data enhancement technologies such as SMOTE and GAN networks and the like, and by adopting an artificial intelligent neural network, a large number of data processing models are trained and configured to a central node, and tasks with small computation amount such as data preprocessing and the like are configured to edge nodes, so that the problems of wide distribution of turnout equipment, unbalanced data, occupation of cloud computing bandwidth, insufficient computing capacity of edge nodes and the like can be effectively solved, and the stability and the efficiency of the existing intelligent operation and maintenance system are improved.
Disclosure of Invention
The invention aims to provide an intelligent turnout fault diagnosis method based on an edge computing network architecture, and solves the problems of unbalanced data, complex model calculation aiming at an artificial intelligent neural network fault diagnosis algorithm, cloud computing bandwidth occupation, insufficient edge node calculation capacity and the like in the prior art.
The technical scheme adopted by the invention is as follows: an intelligent turnout fault diagnosis method based on an edge computing network architecture comprises the following specific operation steps:
step 2, the central cloud layer uses the existing real data to perform data enhancement through SMOTE and GAN methods, is used for training an initial neural network model, and transmits the trained model to the edge cloud layer;
step 3, the access layer collects switch machine starting current curve data through a microcomputer monitoring system and sends the data to the edge cloud layer for data storage, data preprocessing and fault diagnosis;
step 4, after preprocessing the data, the edge cloud layer carries out fault diagnosis on one hand, transmits the data to the central cloud layer for data storage and neural network model training on the other hand, and transmits the data which cannot be diagnosed to the central cloud layer for record processing;
step 5, on the basis of the step 4, when the real data volume in the center cloud layer reaches more than 1 ten thousand, the real data from the edge cloud layer is not received, and only the fault data and the unidentified data determined after the intelligent diagnosis of the edge cloud layer are received; and the central cloud layer carries out neural network training according to the time period T and transmits the obtained neural network model to the edge cloud layer.
The present invention is also characterized in that,
the access layer comprises a point switch, a sensor, a microcomputer monitoring system and the like; the edge cloud layer comprises a server and a database of edge nodes; the central cloud layer comprises a large server and a database; the central cloud layer has extremely strong data computing capacity and data storage capacity compared with the servers of the edge cloud layer.
The edge cloud layer is used for preprocessing data and diagnosing faults based on the neural network model, and the neural network model is calculated in the central cloud layer.
The step 2 is as follows:
the central cloud layer stores more than 400 pieces of historical real turnout action current curve data of the same model, and is classified according to normal data and 6 common fault types, wherein more than 50 pieces of 6 similar fault current curve data are stored, and more than 100 pieces of normal data are stored; the central cloud layer performs data enhancement by using a method of synthesizing a few classes of oversampling technologies SMOTE and generating a countermeasure network GAN through the data; increasing 6 kinds of similar fault current curve data to more than 500 pieces respectively, and increasing normal class data to more than 1000 pieces, wherein more than 4000 pieces of data are obtained;
wherein, the normal data type is recorded as f 0;
the type of the fault state of foreign matters existing between the rails is recorded as f 1;
the type of the fluctuation fault state of the conversion stage is recorded as f 2;
the locking stage fluctuation fault state type is recorded as f 3;
the fluctuation fault state type of the starting stage is recorded as f 4;
the locking phase difficult fault state type is recorded as f 5;
the stator and rotor mixed-line fault state type is recorded as f 6.
When current curve data enhancement is carried out, a GAN method is used for four data types of f0, f1, f4 and f6, and a SMOTE method is used for three data types of f2, f3 and f 5;
data generated by SMOTE and GAN methods are used for CNN neural network training to generate an initial neural network model for turnout fault diagnosis; the model can carry out turnout fault diagnosis through data preprocessed by a given current curve, and diagnosis results are divided into a no-fault state, 6 fault states and an unidentified state; the central cloud layer transmits the trained neural network model to each edge cloud layer node;
if normal data and various fault data in the real data reach more than 1 ten thousand, sample data generated by the SMOTE and GAN methods are eliminated, and the real data from the access stratum are all adopted for training the CNN model.
And step 5, defining a period T according to the user requirements, and defaulting to 1 week.
The step 3 is as follows:
the access layer transmits the current curve data received by the microcomputer monitoring system to the edge cloud layer node nearby for data storage, data preprocessing and fault diagnosis;
the data preprocessing is to perform data lengthening, data segmentation and feature extraction on the received current curve data. Then, carrying out fault diagnosis on the preprocessed data by using a neural network diagnosis model transmitted by the central cloud layer;
the data length is to zero the data length of the current curve to 576;
the data segmentation means that compensated data are divided into three sections according to a sliding window method, wherein the three sections are respectively an unlocking stage, a conversion stage and a locking stage;
feature extraction refers to the use of 11 widely used feature extraction equations to describe the geometric features of the curve, including the import-export difference, the mean, the root mean square value, the variance, the score sum, the kurtosis value, the peak factor, the form factor, the pulse factor and the margin factor.
The step 4 is as follows:
step 4.1, each node of the edge cloud layer transmits the preprocessed data stored in the database to the center cloud layer in real time and stores the preprocessed data in the database; and setting the time period to be T days, and merging the stored data by the central cloud layer to train the neural network model. After the neural network model is trained, transmitting the neural network model to each edge cloud layer node;
step 4.2, after the current curve data from the access stratum is subjected to data preprocessing by the edge cloud layer, on one hand, fault diagnosis is carried out, and on the other hand, the current curve data are transmitted to the center cloud layer;
the fault diagnosis result is divided into a normal state, 6 common fault states and an unidentified state; if the state is normal, no processing is performed; if the fault conditions are 6, alarming to a turnout fault maintainer to process the turnout fault according to the fault type; if the cloud layer is not identified, transmitting the curve data to a central cloud layer for filing;
and the record processing means that newly increased uncertain state types are stored for curve data diagnosed in an unidentified state, and a maintainer is appointed to overhaul to a turnout section in which the unidentified state occurs, so as to determine whether the turnout has a fault, the fault reason and the state type. Finally, storing the analysis result in a server;
and when the same type of unidentified current curve data reaches more than 50, expanding the data set by using SMOTE and GAN methods for generating a CNN model of a new fault class.
The step 5 is as follows:
the data volume of the normal current curve is usually far larger than that of the fault current curve, when the data volume of the normal current curve reaches more than 1 ten thousand, the normal current data is not received, and only the fault current curve data and the unidentified current curve data are received, so that the precision of the neural network model is improved; in order to save resources and improve efficiency, the time period can be lengthened, and the transmission frequency can be reduced.
The invention has the beneficial effects that: the invention discloses a turnout fault diagnosis method based on an edge computing network architecture, which aims at the problem of uneven distribution of turnout curve data samples and provides a method for enhancing data of a few fault samples by using SMOTE and GAN networks. Aiming at the problems of complex model calculation, cloud computing bandwidth occupation, insufficient computing capacity of edge nodes and the like of a turnout fault diagnosis method using an artificial intelligent neural network algorithm, the invention adopts a method based on an edge computing network architecture to layer the fault diagnosis network architecture into three layers, namely an access layer, an edge cloud layer and a center cloud layer. Preprocessing calculation and fault diagnosis of a large amount of data are placed on the edge cloud layer, and CNN model training with complex calculation is only carried out on the center cloud layer, so that network time delay and bandwidth of fault diagnosis are greatly reduced, the problems of wide distribution of turnout equipment, occupation of cloud calculation bandwidth, insufficient computing capacity of edge nodes and the like can be effectively solved, and the stability and the efficiency of the existing intelligent operation and maintenance system are improved.
Drawings
FIG. 1 is a schematic diagram of a three-layer network architecture of a turnout fault diagnosis method based on an edge computing network architecture according to the present invention;
FIG. 2 is a schematic diagram of a three-layer network physical structure of a turnout fault diagnosis method based on an edge computing network architecture according to the present invention;
FIG. 3 is a schematic general flow chart of a turnout fault diagnosis method based on an edge computing network architecture according to the present invention;
FIG. 4 is a flow chart of a turnout fault diagnosis method based on an edge computing network architecture according to the present invention;
fig. 5 is a schematic diagram of the SMOTE method of the turnout fault diagnosis method based on the edge computing network architecture.
Detailed Description
As shown in fig. 1, the turnout fault diagnosis method based on the edge computing network architecture of the present invention is specifically implemented according to the following steps:
And 2, performing data enhancement on the central cloud layer by using a small amount of existing real data through an SMOTE (short-time evolution) and GAN (global area network) method, training an initial neural network model, and transmitting the trained model to the edge cloud layer.
And 3, the access layer acquires current curve data of the point switch through the sensor and sends the current curve data to the edge cloud layer for data storage, data preprocessing and fault diagnosis.
And 4, the edge cloud layer sends the data after data preprocessing to the central cloud layer for data storage and neural network training, and the edge cloud layer sends the data which cannot be diagnosed to the central cloud layer for record processing.
And 5, on the basis of the step 4, when the real data volume in the central cloud layer reaches a certain amount (more than 1 ten thousand are recommended), the real data from the edge cloud layer is not received any more, and only the fault data and the unidentified data determined after the intelligent diagnosis of the edge cloud layer are received. And the central cloud layer carries out neural network training according to a time period T, and transmits the obtained neural network model to the edge cloud layer, wherein the T can be defined by a user and defaulted to 1 week.
As shown in fig. 2, step 1 is specifically as follows:
the method divides the turnout fault diagnosis network architecture into three layers, namely an access layer, an edge cloud layer and a center cloud layer. The access layer comprises a point switch, a sensor, a microcomputer monitoring system and the like; the edge cloud layer comprises a server, a database and the like of the edge node; the central cloud layer comprises a large server and a database; the central cloud layer has extremely strong data computing capacity and data storage capacity compared with the servers of the edge cloud layer.
As shown in fig. 3, step 2 is specifically as follows:
the central cloud layer stores a small amount (more than 400 pieces are recommended) of historical real turnout action current curve data of the same model, and can be classified according to normal data and 6 common fault types, wherein each of the 6 similar fault current curve data is more than 50 pieces, and the number of the normal data is more than 100. The central cloud layer performs data enhancement through these small amounts of data using a synthetic minority class oversampling technique (SMOTE) and a generative countermeasure network (GAN) approach. The 6 kinds of similar fault current curve data are respectively increased to more than 500, and the normal data are increased to more than 1000, and more than 4000 pieces of data are obtained.
The SMOTE algorithm flow is as follows:
(1) for each sample in the minority class, calculating the distance from each sample to other samples by using calculation methods such as Euclidean distance and Mahalanobis distance as standards;
(2) then, a value is set, the size is generally 5, and the nearest neighbor sample can be selected; secondly, setting a value which represents the sampling multiplying power and determining the number of finally generated samples;
(3) randomly selecting a neighbor from among its neighbor samples;
(4) finally, a number from 0 to 1 is randomly set, which represents the position proportion of the newly synthesized sample on the connecting line of the original few class samples and the randomly selected neighboring samples.
The steps of the SMOTE process are schematically illustrated in fig. 5. In fig. 5, circles represent normal majority samples, the middle five-pointed star represents minority samples, the four-pointed star represents neighboring samples of minority samples, and black rectangles represent newly generated minority samples.
The GAN algorithm flow is as follows:
(1) initializing a generator G and a discriminator D;
(2) in each iteration process:
firstly, fixing a generator G and only updating parameters of a discriminator D;
fixing the discriminator D and only updating the parameters of the generator G;
GAN target optimization function:
and the data generated by the SMOTE and GAN methods are used for CNN neural network training to generate an initial neural network model for turnout fault diagnosis. The model can carry out turnout fault diagnosis through data preprocessed by a given current curve, and diagnosis results are divided into a normal state, 6 fault states and an unidentified state. And the central cloud layer transmits the trained neural network model to each edge cloud layer node.
As shown in fig. 3 and 4, step 3 is as follows:
and 3.1, as shown in fig. 3, the access layer transmits the current curve data received by the sensor to the edge cloud layer node nearby, and data storage, data preprocessing and fault diagnosis are performed.
The data preprocessing is to perform data lengthening, data segmentation and characteristic value calculation on the received current curve data. And then, carrying out fault diagnosis on the preprocessed data by using the neural network diagnosis model transmitted by the central cloud layer.
Step 3.2, if the diagnosis result of the CNN model is normal, no treatment is carried out; if the fault occurs, different fault processing is respectively carried out according to different fault types. The following six common fault and maintenance suggestions are provided:
fault state f 1: foreign matter exists between the rails.
And (3) fault treatment: and (5) overhauling the switch tongue and the stock rail to see whether foreign matters exist between the switch tongue and the stock rail.
Fault state f 2: the transition phase fluctuates.
And (3) fault treatment: and (5) overhauling the switch machine and the friction belt.
Fault state f 3: the locking phase fluctuates.
And (3) fault treatment: and (6) overhauling the cable box.
Fault state f 4: the start-up phase fluctuates.
And (3) fault treatment: and (5) overhauling the automatic shutter.
Fault state f 5: the locking phase is difficult.
And (3) fault treatment: and (6) overhauling the turnout.
Fault state f 6: and mixing the stator and the rotor.
And (3) fault treatment: and (6) overhauling the switch machine.
As shown in fig. 3, step 4 is specifically as follows:
and 4.1, setting a certain time period T, transmitting the preprocessed data stored in the database to the central cloud layer by the edge cloud layer, storing the preprocessed data received by the central cloud layer from each edge cloud layer node into the database, and combining the preprocessed data with the stored data to train the neural network model. After the neural network model is trained, the neural network model is still sent to each edge cloud layer node.
And 4.2, when the unidentified current curve data is diagnosed by the edge cloud layer, timely alarming to turnout fault maintainers and recording in a central cloud layer server. When the same type of unidentified current curve data reaches more than 50, the data set can be expanded by an SMOTE method and a GAN method and used for generating a CNN model of a new fault class. The new fault class can be added into the conventional fault class by means of manual identification of fault types, fault maintenance modes and generation of a fault diagnosis model.
The step 5 is as follows:
the data volume of the normal current curve collected by the microcomputer monitoring system in the access layer is often far larger than that of the fault current curve, so that the normal curve data set required by the central cloud layer for training the neural network model is much simpler than the fault curve data set. When the data volume of the normal current curve reaches a certain amount (more than 1 ten thousand are suggested), the normal current curve data from the access layer is not needed any more, so that only the fault current curve data and the unidentified current curve data need to be received at the moment, and the accuracy of the neural network model is improved. In order to save resources and improve efficiency, the time period can be lengthened, and the transmission frequency can be reduced.
And when the time period is up, the CNN model is trained through the stored preprocessed data set, and the training model is sent to each edge node.
An example of the identification method of the present invention is to use part of the historical data as the input data set. Data enhancement is firstly carried out on a central cloud layer by using SMOTE and GAN methods. And then, based on a sample set formed by the generated data samples and the real samples as a sample set of the CNN model, performing data preprocessing and model training. And then transmitting the trained CNN model to each node of the edge cloud layer for fault diagnosis, wherein a fault diagnosis flow chart is shown in FIG. 4. After the edge cloud layer carries out data preprocessing on the turnout current curve data from the access layer, on one hand, fault diagnosis is carried out, on the other hand, data storage is carried out, and the data are transmitted to the center cloud layer for model training in a certain time period. Because the data volume of the normal current is far larger than that of the fault current, and the training effect is poor due to uneven data volume distribution, when the current data of the turnout normal action of the same type of the central cloud layer is collected to more than 1 ten thousand, the normal data from the edge cloud layer is not received any more, and only the fault data is received. If normal data and various fault data in the real data reach more than 1 ten thousand, sample data generated by the SMOTE and GAN methods are eliminated, and the real data from the access stratum are all adopted for training the CNN model.
Claims (7)
1. An intelligent turnout fault diagnosis method based on an edge computing network architecture is characterized by comprising the following specific operation steps:
step 1, dividing a turnout fault diagnosis network architecture into three layers, namely an access layer, an edge cloud layer and a center cloud layer;
step 2, the central cloud layer uses the existing real data to perform data enhancement through SMOTE and GAN methods, is used for training an initial neural network model, and transmits the trained model to the edge cloud layer;
step 3, the access layer collects switch machine starting current curve data through a microcomputer monitoring system and sends the data to the edge cloud layer for data storage, data preprocessing and fault diagnosis;
step 4, after preprocessing the data, the edge cloud layer carries out fault diagnosis on one hand, transmits the data to the central cloud layer for data storage and neural network model training on the other hand, and transmits the data which cannot be diagnosed to the central cloud layer for record processing;
step 5, on the basis of the step 4, when the real data volume in the center cloud layer reaches more than 1 ten thousand, the real data from the edge cloud layer is not received, and only the fault data and the unidentified data determined after the intelligent diagnosis of the edge cloud layer are received; and the central cloud layer carries out neural network training according to the time period T and transmits the obtained neural network model to the edge cloud layer.
2. The intelligent turnout fault diagnosis method based on the edge computing network architecture is characterized in that the access layer comprises a switch machine, a sensor, a microcomputer monitoring system and the like; the edge cloud layer comprises a server and a database of edge nodes; the central cloud layer comprises a large server and a database; the central cloud layer has extremely strong data computing capacity and data storage capacity compared with the servers of the edge cloud layer.
The edge cloud layer is used for preprocessing data and diagnosing faults based on the neural network model, and the neural network model is calculated in the center cloud layer.
3. The intelligent turnout fault diagnosis method based on the edge computing network architecture according to claim 2, wherein the step 2 is specifically as follows:
the central cloud layer stores more than 400 pieces of historical real turnout action current curve data of the same model, and is classified according to normal data and 6 common fault types, wherein more than 50 pieces of 6 similar fault current curve data are stored, and more than 100 pieces of normal data are stored; the central cloud layer performs data enhancement by using a method of synthesizing a few classes of oversampling technologies SMOTE and generating a countermeasure network GAN through the data; increasing 6 kinds of similar fault current curve data to more than 500 pieces respectively, and increasing normal class data to more than 1000 pieces, wherein more than 4000 pieces of data are obtained;
wherein, the normal data type is recorded as f 0;
the type of the fault state of foreign matters existing between the rails is recorded as f 1;
the type of the fluctuation fault state of the conversion stage is recorded as f 2;
the locking stage fluctuation fault state type is recorded as f 3;
the fluctuation fault state type of the starting stage is recorded as f 4;
the locking phase difficult fault state type is recorded as f 5;
the stator and rotor mixed-line fault state type is recorded as f 6.
When current curve data enhancement is carried out, a GAN method is used for four data types of f0, f1, f4 and f6, and a SMOTE method is used for three data types of f2, f3 and f 5;
data generated by SMOTE and GAN methods are used for CNN neural network training to generate an initial neural network model for turnout fault diagnosis; the model can carry out turnout fault diagnosis through data preprocessed by a given current curve, and diagnosis results are divided into a no-fault state, 6 fault states and an unidentified state; the central cloud layer transmits the trained neural network model to each edge cloud layer node;
if normal data and various fault data in the real data reach more than 1 ten thousand, sample data generated by the SMOTE and GAN methods are eliminated, and the real data from the access stratum are all adopted for training the CNN model.
4. The intelligent turnout fault diagnosis method based on the edge computing network architecture according to claim 1, wherein the period T in the step 5 is defined according to user requirements and defaults to 1 week.
5. The intelligent turnout fault diagnosis method based on the edge computing network architecture according to claim 3, wherein the step 3 is as follows:
the access layer transmits the current curve data received by the microcomputer monitoring system to the edge cloud layer node nearby for data storage, data preprocessing and fault diagnosis;
the data preprocessing is to perform data lengthening, data segmentation and feature extraction on the received current curve data. Then, carrying out fault diagnosis on the preprocessed data by using a neural network diagnosis model transmitted by the central cloud layer;
the data length is to zero the data length of the current curve to 576;
the data segmentation means that compensated data are divided into three sections according to a sliding window method, wherein the three sections are respectively an unlocking stage, a conversion stage and a locking stage;
feature extraction refers to the use of 11 widely used feature extraction equations to describe the geometric features of the curve, including the import-export difference, the mean, the root mean square value, the variance, the score sum, the kurtosis value, the peak factor, the form factor, the pulse factor and the margin factor.
6. The intelligent turnout fault diagnosis method based on the edge computing network architecture according to claim 3, wherein the step 4 is specifically as follows:
step 4.1, each node of the edge cloud layer transmits the preprocessed data stored in the database to the center cloud layer in real time and stores the preprocessed data in the database; and setting the time period to be T days, and merging the stored data by the central cloud layer to train the neural network model. After the neural network model is trained, transmitting the neural network model to each edge cloud layer node;
step 4.2, after the current curve data from the access stratum is subjected to data preprocessing by the edge cloud layer, on one hand, fault diagnosis is carried out, and on the other hand, the current curve data are transmitted to the center cloud layer;
the fault diagnosis result is divided into a normal state, 6 common fault states and an unidentified state; if the state is normal, no processing is performed; if the fault conditions are 6, alarming to a turnout fault maintainer to process the turnout fault according to the fault type; if the cloud layer is not identified, transmitting the curve data to a central cloud layer for filing;
and the record processing means that newly increased uncertain state types are stored for curve data diagnosed in an unidentified state, and a maintainer is appointed to overhaul to a turnout section in which the unidentified state occurs, so as to determine whether the turnout has a fault, the fault reason and the state type. Finally, storing the analysis result in a server;
and when the same type of unidentified current curve data reaches more than 50, expanding the data set by using SMOTE and GAN methods for generating a CNN model of a new fault class.
7. The intelligent turnout fault diagnosis method based on the edge computing network architecture according to claim 3, wherein the step 5 is as follows:
the data volume of the normal current curve is usually far larger than that of the fault current curve, when the data volume of the normal current curve reaches more than 1 ten thousand, the normal current data is not received, and only the fault current curve data and the unidentified current curve data are received, so that the precision of the neural network model is improved; in order to save resources and improve efficiency, the time period can be lengthened, and the transmission frequency can be reduced.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110757298.0A CN113641486B (en) | 2021-07-05 | 2021-07-05 | Intelligent turnout fault diagnosis method based on edge computing network architecture |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110757298.0A CN113641486B (en) | 2021-07-05 | 2021-07-05 | Intelligent turnout fault diagnosis method based on edge computing network architecture |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113641486A true CN113641486A (en) | 2021-11-12 |
| CN113641486B CN113641486B (en) | 2024-03-01 |
Family
ID=78416724
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110757298.0A Active CN113641486B (en) | 2021-07-05 | 2021-07-05 | Intelligent turnout fault diagnosis method based on edge computing network architecture |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113641486B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114415641A (en) * | 2022-01-29 | 2022-04-29 | 重庆长安汽车股份有限公司 | Privacy protection method and system for remote fault diagnosis |
| CN117081502A (en) * | 2023-10-16 | 2023-11-17 | 锦浪科技股份有限公司 | Photovoltaic direct current arc monitoring platform and control method thereof |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015158198A1 (en) * | 2014-04-17 | 2015-10-22 | 北京泰乐德信息技术有限公司 | Fault recognition method and system based on neural network self-learning |
| CN106709567A (en) * | 2016-12-14 | 2017-05-24 | 河北省科学院应用数学研究所 | Method for diagnosing switch faults based on deep learning model |
| WO2020244134A1 (en) * | 2019-06-05 | 2020-12-10 | 华南理工大学 | Multi-task feature sharing neural network-based intelligent fault diagnosis method |
| CN112348170A (en) * | 2020-11-10 | 2021-02-09 | 交控科技股份有限公司 | Fault diagnosis method and system for turnout switch machine |
| CN112738092A (en) * | 2020-12-29 | 2021-04-30 | 北京天融信网络安全技术有限公司 | Log data enhancement method, classification detection method and system |
| US20210192354A1 (en) * | 2019-12-18 | 2021-06-24 | Sap Se | Generic workflow for classification of highly imbalanced datasets using deep learning |
-
2021
- 2021-07-05 CN CN202110757298.0A patent/CN113641486B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015158198A1 (en) * | 2014-04-17 | 2015-10-22 | 北京泰乐德信息技术有限公司 | Fault recognition method and system based on neural network self-learning |
| CN106709567A (en) * | 2016-12-14 | 2017-05-24 | 河北省科学院应用数学研究所 | Method for diagnosing switch faults based on deep learning model |
| WO2020244134A1 (en) * | 2019-06-05 | 2020-12-10 | 华南理工大学 | Multi-task feature sharing neural network-based intelligent fault diagnosis method |
| US20210192354A1 (en) * | 2019-12-18 | 2021-06-24 | Sap Se | Generic workflow for classification of highly imbalanced datasets using deep learning |
| CN112348170A (en) * | 2020-11-10 | 2021-02-09 | 交控科技股份有限公司 | Fault diagnosis method and system for turnout switch machine |
| CN112738092A (en) * | 2020-12-29 | 2021-04-30 | 北京天融信网络安全技术有限公司 | Log data enhancement method, classification detection method and system |
Non-Patent Citations (2)
| Title |
|---|
| 刘云鹏;许自强;和家慧;王权;高树国;赵军;: "基于条件式Wasserstein生成对抗网络的电力变压器故障样本增强技术", 电网技术, no. 04, 31 December 2020 (2020-12-31) * |
| 张文龙;胡天亮;王艳洁;魏永利;: "云/边缘协同的轴承故障诊断方法", 计算机集成制造系统, no. 03, 31 December 2020 (2020-12-31) * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114415641A (en) * | 2022-01-29 | 2022-04-29 | 重庆长安汽车股份有限公司 | Privacy protection method and system for remote fault diagnosis |
| CN117081502A (en) * | 2023-10-16 | 2023-11-17 | 锦浪科技股份有限公司 | Photovoltaic direct current arc monitoring platform and control method thereof |
| CN117081502B (en) * | 2023-10-16 | 2024-02-06 | 锦浪科技股份有限公司 | Photovoltaic direct current arc monitoring platform and control method thereof |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113641486B (en) | 2024-03-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107941537B (en) | A kind of mechanical equipment health state evaluation method | |
| Li et al. | A deep learning approach for anomaly detection based on SAE and LSTM in mechanical equipment | |
| CN109102005B (en) | Small sample deep learning method based on shallow model knowledge migration | |
| CN104091070B (en) | Rail transit fault diagnosis method and system based on time series analysis | |
| CN110441629B (en) | Fault diagnosis method and device based on turnout switch machine operating current curve | |
| CN109829236A (en) | A kind of Compressor Fault Diagnosis method based on XGBoost feature extraction | |
| CN111582392B (en) | Multi-working-condition health state online monitoring method for key components of wind turbine generator | |
| CN107909118A (en) | A kind of power distribution network operating mode recording sorting technique based on deep neural network | |
| CN113641486A (en) | Intelligent turnout fault diagnosis method based on edge computing network architecture | |
| CN110388315B (en) | Oil transfer pump fault identification method, device and system based on multi-source information fusion | |
| CN111199257A (en) | Fault diagnosis method and device for high-speed rail driving equipment | |
| CN110362045B (en) | Marine doubly-fed wind turbine generator fault discrimination method considering marine meteorological factors | |
| Yang et al. | Fault detection of wind turbine generator bearing using attention-based neural networks and voting-based strategy | |
| CN111680875A (en) | Unmanned aerial vehicle state risk fuzzy comprehensive evaluation method based on probability baseline model | |
| EP4080304A1 (en) | System and method for cyberattack detection in a wind turbine control system | |
| CN113640027A (en) | Heat exchange unit fault diagnosis method and system based on edge calculation and neural network | |
| Saeki et al. | Visual explanation of neural network based rotation machinery anomaly detection system | |
| Li et al. | Framework and case study of cognitive maintenance in Industry 4.0 | |
| US11306705B2 (en) | System and method for monitoring a device | |
| CN114061743A (en) | Vibration monitoring method, device, equipment and medium for wind generating set | |
| Zhang et al. | A novel fault diagnosis method for wind turbine based on adaptive multivariate time-series convolutional network using SCADA data | |
| CN104778340A (en) | Bearing life prediction method based on enhanced particle filter | |
| CN105426665B (en) | Method is determined based on the DYNAMIC RELIABILITY of status monitoring | |
| Jiang et al. | Research on motor bearing fault diagnosis based on the AdaBoost algorithm and the ensemble learning with Bayesian optimization in the industrial Internet of Things | |
| CN116467592A (en) | Production equipment fault intelligent monitoring method and system based on deep learning |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |