CN115616329A - Five little casees fault recognition system of transformer substation based on convolutional neural network - Google Patents
Five little casees fault recognition system of transformer substation based on convolutional neural network Download PDFInfo
- Publication number
- CN115616329A CN115616329A CN202211413860.9A CN202211413860A CN115616329A CN 115616329 A CN115616329 A CN 115616329A CN 202211413860 A CN202211413860 A CN 202211413860A CN 115616329 A CN115616329 A CN 115616329A
- Authority
- CN
- China
- Prior art keywords
- neural network
- convolutional neural
- small
- fault
- layer
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
- G01D21/00—Measuring or testing not otherwise provided for
- G01D21/02—Measuring two or more variables by means not covered by a single other subclass
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/10—Image acquisition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Multimedia (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Alarm Systems (AREA)
Abstract
The invention provides a transformer substation five-small box fault recognition system based on a convolutional neural network, which comprises a temperature measurement sensing unit, a camera unit and a microprocessor, wherein the temperature measurement sensing unit is connected with the input end of the microprocessor through an I2C interface, and the camera unit is connected with the input end of the microprocessor through a serial camera control bus; the temperature measurement sensing unit is used for acquiring first acquired images of an air switch and a relay in a five-small box of the transformer substation; the camera shooting unit is used for collecting second collected images of an air switch and a relay in five small boxes of the transformer substation; and the microprocessor is used for carrying out characteristic analysis on the first collected image and the second collected image through a built-in convolutional neural network and identifying the fault type according to an analysis result. The invention can effectively identify the fault types of the running conditions of the in-box relay, the air switch and other devices, has the autonomous learning capability and improves the generalization capability of the fault identification method.
Description
Technical Field
The invention belongs to the field of substation box fault identification, and particularly relates to a substation five-small box fault identification system based on a convolutional neural network.
Background
The five small boxes of the transformer substation are general names of outdoor box equipment of the transformer substation such as various mechanism boxes, terminal boxes, maintenance power boxes and the like, serve as auxiliary equipment of a power system, are widely applied to a power transmission and distribution network system, bear the opening and closing operation of the power supply system and control and protect a working power supply of a main equipment system, and directly influence the power supply quality and safety performance of the power system due to the operation reliability of the five small boxes. Due to the fact that the five small boxes of the transformer substation are various in internal insulation structures, complex in actual operation environment (high temperature, dust, humidity and the like), and possible production quality and process defects, insulation performance may be deteriorated in a long-term operation process, partial discharge is caused under the action of an electric field, further development of the partial discharge aggravates insulation aging, and air break or leakage protection action may be caused in severe cases; for example, the overheating phenomenon caused by the loosening of the contact fixing screw of the air switch and the contactor triggers the protection action, and the safe and stable operation of the power distribution system is influenced. At present, the running state of five small boxes of the transformer substation is mainly monitored by means of regular patrol work of technicians. This requires detailed and careful planning, which is necessary to ensure strict enforcement by the management staff.
Therefore, the prevention and elimination of hidden dangers can be greatly ensured by adopting an effective fault identification technology. The image recognition technology is to process and analyze an image by using a computer so as to achieve the purpose of simulating human beings and automatically recognizing objects and targets. Generally, the image recognition method of the power equipment depends on the color feature and the geometric feature of the target, and the methods are often influenced by factors such as background and brightness, so that the generalization capability of the methods is poor.
Disclosure of Invention
In order to solve the defects and defects of poor generalization capability caused by fault identification of five small boxes of a transformer substation depending on color characteristics and geometric characteristics of a target in the prior art, the invention provides a transformer substation five small box fault identification system based on a convolutional neural network, which comprises a temperature measurement sensing unit, a camera unit and a microprocessor, wherein the temperature measurement sensing unit is connected with the input end of the microprocessor through an I2C interface, and the camera unit is connected with the input end of the microprocessor through a serial camera control bus;
the temperature measurement sensing unit is used for acquiring first acquired images of an air switch and a relay in a five-small box of the transformer substation;
the camera shooting unit is used for collecting second collected images of an air switch and a relay in five small boxes of the transformer substation;
and the microprocessor is used for carrying out characteristic analysis on the first collected image and the second collected image through a built-in convolutional neural network and identifying the fault type according to the analysis result.
Optionally, the first collected image is a real-time contact temperature image of an air switch and a relay in the five-small box.
Optionally, the second collected image is a real-time action state image of an air switch and a relay in a five-small box of the transformer substation.
Optionally, the microprocessor performs real-time positioning and fault type identification on the opening and closing states of the air switch and the relay through a deep learning algorithm of a convolutional neural network.
Optionally, an output result of the convolutional layer of the convolutional neural network is represented as:
wherein k is a sublayer number of the convolutional layer, i is a starting row number of pixel scanning of the first and second acquisition images, j is a starting column number of pixel scanning of the first and second acquisition images,shows the output of the convolutional layer k,respectively representing the weights of the filters in sub-layer k, x ij 、x ij+1 、…、x i+2j+2 Pixel representing input sublayer k, b Fb For a predetermined offset, a represents the activation function.
Optionally, the convolutional neural network completes training and learning of five small box collected images of the transformer substation based on an error back propagation method in advance, and determines the number and size of filters adopted by convolutional layers in the convolutional neural network and the preset offset based on a training and learning result.
Optionally, the pooling layer of the convolutional neural network is represented as:
wherein k is the sublayer number of the pooling layer,the output result of the pooling layer k is shown, i and j respectively show the row number and the column number of the neural unit in the sublayer k,respectively representing the inputs of the pooling layer sub-layer k,representing the activation function of pooling layer sublayer k.
Optionally, the number of the neural units in the output layer of the convolutional neural network is equal to the number of the fault types to be identified.
Optionally, the five-small-box fault identification system of the transformer substation further includes a wireless transmission unit, and is configured to send an analysis result of the microprocessor to the fault monitoring platform.
The technical scheme provided by the invention has the following beneficial effects:
the transformer substation five-small-box fault recognition system provided by the invention adopts a deep learning algorithm of a convolutional neural network to effectively identify the operation conditions and fault types of devices such as a five-small-box relay and an air switch in a transformer substation, realizes data transmission and exchange through a wireless transmission and monitoring room, can effectively recognize the fault types of the operation conditions of the devices such as the box relay and the air switch, has an autonomous learning capability, and improves the generalization capability of a fault recognition method.
Drawings
In order to more clearly illustrate the technical solution of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on the drawings without creative efforts.
Fig. 1 is a schematic structural diagram of a transformer substation five-small box fault identification system based on a convolutional neural network according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The terms "first," "second," "third," "fourth," and the like in the description and in the claims, as well as in the drawings, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are capable of operation in sequences other than those illustrated or described herein.
It should be understood that, in various embodiments of the present invention, the sequence numbers of the processes do not mean the execution sequence, and the execution sequence of the processes should be determined by the functions and the internal logic of the processes, and should not constitute any limitation on the implementation process of the embodiments of the present invention.
It should be understood that in the present application, "comprising" and "having" and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
It should be understood that, in the present invention, "a plurality" means two or more. "and/or" is merely an association describing an associated object, meaning that three relationships may exist, for example, and/or B, may mean: a exists alone, A and B exist simultaneously, and B exists alone. The character "/" generally indicates that the former and latter associated objects are in an "or" relationship. "comprising a, B and C", "comprising a, B, C" means that all three of a, B, C are comprised, "comprising a, B or C" means comprising one of a, B, C, "comprising a, B and/or C" means comprising any 1 or any 2 or 3 of a, B, C.
It should be understood that in the present invention, "B corresponding to a", "a corresponds to B", or "B corresponds to a" means that B is associated with a, and B can be determined from a. Determining B from a does not mean determining B from a alone, but may be determined from a and/or other information. And the matching of A and B means that the similarity of A and B is greater than or equal to a preset threshold value.
As used herein, "if" may be interpreted as "at \8230; \8230when" or "when 8230; \8230when" or "in response to a determination" or "in response to a detection", depending on the context.
The technical means of the present invention will be described in detail with reference to specific examples. These several specific embodiments may be combined with each other below, and details of the same or similar concepts or processes may not be repeated in some embodiments.
Example (b):
as shown in fig. 1, the present embodiment provides a system for identifying a fault of a five-small box in a transformer substation based on a convolutional neural network, which includes a temperature measurement sensing unit, a camera unit and a microprocessor, where the temperature measurement sensing unit is connected to an input end of the microprocessor through an I2C interface, and the camera unit is connected to an input end of the microprocessor through a serial camera control bus;
the temperature measurement sensing unit is used for acquiring first acquired images of an air switch and a relay in a five-small box of the transformer substation;
the camera shooting unit is used for collecting second collected images of an air switch and a relay in a five-small box of the transformer substation;
and the microprocessor is used for carrying out characteristic analysis on the first collected image and the second collected image through a built-in convolutional neural network and identifying the fault type according to the analysis result.
In the embodiment, the deep learning algorithm of the convolutional neural network is adopted to carry out effective mode recognition on the operation conditions and fault types of the relay, the air switch and other devices in the five small boxes of the transformer substation, and the transmission and exchange of data are realized through wireless transmission and a monitoring room. The embodiment adopts the deep learning convolutional neural network, can effectively identify the fault types of the running conditions of the devices such as the relay and the air switch in the box, and has the autonomous learning capability.
The first collected image is a real-time contact temperature image of an air switch and a relay in the five small boxes, and in the embodiment, the real-time electric shock temperature image is collected by adopting an infrared temperature measurement dot matrix sensor module MLX90640 ESF-BAB. The infrared temperature measuring dot matrix sensor module MLX90640ESF-BAB is a 32x24 pixel IR which is subjected TO comprehensive calibration, an array tape digital interface packaged by adopting an industry standard 4-pin TO39, and the MLX90640 comprises 768 FIR pixels. An ambient sensor is integrated to measure ambient temperature sensor of the chip and power supply to measure VDD. All output sensors IR, ta and VDD are stored in internal RAM and communication of the sensors and microprocessor can be achieved through an I2C interface. And thermal imaging of each element in the box body is realized.
The second collected image is a real-time action state image of an air switch and a relay in a five-small box of the transformer substation, and in this embodiment, a CameraChip sensor OV7725 is used for collecting the real-time action state image. The CameraChip sensor OV7725 is a high-performance 1/4 inch single-chip VGA camera and image processor, and occupies a small space. In full function operation, the OV7725 meets all pc multimedia and camera phone market requirements for performance, quality and reliability. The low power consumption OV7725 performs well under low light conditions and can operate over a wide temperature range, from-20 ℃ to +70 ℃. The OV7725 contains an array of 640x 480 pictures capable of running the user in VGA mode at 60 frames per second to fully control image quality, format, and output data transfer. The OV7725 provides full frame, sub-sampled or windowed 8 bit/10 bit images in a variety of formats, controlled through a Serial Camera Control Bus (SCCB) interface. This OV7725 is provided with all required camera processing functions including exposure control, gamma, white balance, color saturation, hue control, and the like. These functions are also programmed via the SCCB interface. The method is used for shooting the running state and the fault mode of each device in the box body in real time.
The microprocessor carries out real-time positioning and fault type identification on the opening and closing states of an air switch and a relay through a deep learning algorithm of a convolutional neural network, in the embodiment, STM32F769 kernel is adopted by the microprocessor, and comprises ARM 32-bit Cortex-M7 CPUDPFPU, ART accumulator and 16KB I/D cache kernel, a 0-wait state is allowed to be executed from an embedded Flash and an external memory, and the instruction is as high as 216MHz, MPU,462DMIPS/2.14DMIPS/MHz (Dhrystone2.1) and DSP. The microprocessor is used for data processing and data exchange with each unit.
In this embodiment, the algorithm for failure mode identification is deep learning of a convolutional neural network, and the data types are: the infrared temperature measurement dot matrix 32 × 24 matrix data and the OV7725 camera 640 × 480 matrix data are used as input data of a convolutional neural network, the size of a filter is 5 × 5 matrix data, the matrix data value is obtained through off-line learning, the learning method is an error back propagation method, and values of filter components, weights and offset data and values of output layer weights and offset data are obtained through learning.
In this embodiment, the substation five-small box fault identification system further includes a wireless transmission unit, which is configured to send an analysis result of the microprocessor to the fault monitoring platform. This embodiment specifically adopts SX1278 as the loRa communication module of system, and SX1278 is a long-range, low-power consumption wireless transceiver that Semtech company promoted, is the thing networking wireless transceiver that a performance is high, possesses special loRa modulation mode, has increased communication distance to a certain extent. The characteristics are as follows: (1) The transmission of data, commands and the like is carried out by using a serial port or a USB (universal serial bus), and AT instruction set control is carried out;
(2) Long distance: a sensitivity of 148dB provides a high penetration transmission over long distances of 3000m;
(3) Anti-interference: a SNR of 20dB provides noise immunity;
(4) High capacity: different spreading factors can provide high channel reuse rate to accommodate more sensor nodes;
(5) Low power consumption: the optimized power saving mode can ensure that the product is used for a long time without replacement;
based on the advantages, the SX1278 has the advantages of low power consumption, large capacity, long transmission distance and strong anti-interference capability. Therefore, the device is widely applied to various industrial, electric, petroleum, logistics, workshop equipment control and data acquisition extremely-long-distance industrial automatic control, data acquisition security and alarm systems, various building and building automation consumer electronics products, wireless intelligent transmitters, flowmeters, wireless sensors and intelligent instruments.
The structure of the convolutional neural network employed in this embodiment is as follows:
1) The convolution layer K is the number of the sub-layer of the convolution layer, i and j are the numbers of the initial scanning row and column, and respectively correspond to the infrared temperature measurement lattice 32x24 matrix data and the camera 640x 480 matrix data, and the relationship is that
Wherein k is a sublayer number of the convolutional layer, i is a starting row number of pixel scanning of the first and second collected images, j is a starting column number of pixel scanning of the first and second collected images,represents the output of the convolutional layer k,respectively representing the weight of the filter in sub-layer k, x ij 、x ij+1 、…、x i+2j+2 Pixel representing input sublayer k, b Fb For a preset offset, a represents the activation function.
2) The pooling layer k is the sublayer number of the pooling layer and adopts a 5x5 structure, i and j are the numbers of the rows and columns of the neural units in the sublayer, and the relational expression adopts the maximum pooling layer structure
Wherein k is the sublayer number of the pooling layer,the output result of the sublayer k of the pooling layer is shown, i and j respectively show the row number and the column number of the neural unit in the sublayer k,respectively representing the inputs of the pooling layer sub-layer k,representing the activation function of the pooling layer sublayer k.
3) The output layer N is the number of the neural unit of the output layer and respectively corresponds to the temperature fault modes of contacts such as an air switch, a contactor and the like and the action states of the air switch, the contactor is closed, the contactor is opened and the like, the relational expression is shown as follows,
wherein z is n O Respectively corresponding to 12 data values of a hollow open three-phase incoming line and outgoing line temperature mode in an infrared dot matrix mode, 6 contact temperature modes of a three-phase incoming line and an outgoing line of a contactor, and data values of an opening and closing mode of the hollow open and the contactor in a camera pattern mode,respectively represent the weights between the neural units of the output layer,respectively representing the inputs to the output layer. and a (z) is an activation function and a sigmoid function is adopted, and the output of the activation function is a fault identification type.
In this embodiment, the number of the neural units in the output layer of the convolutional neural network is equal to the number of the fault types to be identified.
The principle of the deep learning method based on the convolutional neural network adopted by the embodiment is as follows:
1) An input layer: directly substituting data of the infrared temperature measurement dot matrix sensor 32X24 pixels and the camera 30W pixels 640X 480 into a nerve unit of an input layer, wherein the data of the input layer is pixel values, and x is used ij The pixel values at i row and j column positions of the image that is read in. In the neural units of the input layer, the input value and the output value are the same if the output of the neural units of i rows and j columns of the input layer is expressed asThen the following relationship holds:
2) Filter and convolutional layer: several filters are prepared to scan the image, where the number of filters is determined based on the results of off-line learning, and furthermore, since the values of the filters are determined by learning the learning data, they are parameters of the model, and the values are expressed as valuesThe size of the filter in this example is 5 × 5. These filters are then used for convolution processing. Multiplying the 5x5 region of the input layer starting from the upper left corner by the corresponding component of the filter to obtain the following convolution valueSequentially stroking the filter, and calculating to obtain convolution value in the same wayThereby obtaining a convolution value using the filterThe initial row and column numbers of the input layer and the area corresponding to the filter in the filter are shown. The set of values thus obtained forms a feature map. This embodiment adds a number b independent of i, j to these convolution values Fk To obtain
Is considered to beNeural units as weighted inputs, the collection of such neural units forming a convolutionOne sub-layer of the layer. b Fk Is the amount of bias common to the convolutional layers. With an activation function a (z), for a weighted inputOutput of its neural unitCan be expressed as:
3) A pooling layer: the convolutional neural network is provided with a pooling layer for compressing convolutional layer information. At the output of the above-mentioned neural unitIn determinant, the present embodiment compresses 2 × 2 nerve units into 1 nerve unit, and these compressed nerve units become the pooling layer. Feature mapped 2x2 neural units are compressed into 1 neural unit. By performing a pooling operation, the number of neurons in the feature map is reduced to one-fourth of the original. The compression method is the maximum pooling method, and since the output and the input are the same value, the pooling layer does not have the concept of an activation function either, and mathematically, the pooling layer activation function a (x) can be regarded as an identity function a (x) = x, and the pooling layer can be expressed as:
here, k is a sublayer number of the pooling layer, and i and j are integers.
4) An output layer: to identify the type of failure, the present embodiment prepares the number of neural units of the desired failure type in the output layer, which accepts outputs (i.e., full connections) from all the neural units of the previous layer (pooling layer). This allows an integrated investigation of the information of the neural units of the pooling layer. The weighted input to the nth neural unit of the output layer may be expressed as:
here, theFor outputting the nth neural unit of the output layer to the neural unit of the pooling layerThe weight of the weight that is assigned to it,is the bias of the nth neural unit of the output layer. In particular toIs expressed as
The relationships between variables and parameters in the above equation are the outputs of the output layer neural units, which form the output of the overall neural network. The value of the nth neural cell of the output layer isThe activation function is a (z),the index n of the maximum value in (b) is the numerical value of the fault type identification to be determined in the present embodiment.
In this embodiment, i rows =32j columns =24 of an input layer of the convolutional neural network for infrared temperature measurement fault identification, a 5 × 5 structure is adopted for the filter, an output layer is 12, node temperature fault types of input and output of the contact and air switch are respectively corresponded, and learning pictures of the node temperature fault types are 186 pieces. The method comprises the steps that i rows =640j columns =480 of an input layer of a convolutional neural network for camera fault identification, a filter is in a 5x5 structure, output of an output layer is 2, the filter corresponds to the on-off fault type and the off-off fault type of a contact switch and an air switch respectively, a learning picture is 102, and the values of components, weights and offsets of the filter and the values of the weights and the offsets of the output layer are obtained offline through the learning data by adopting an error back propagation method algorithm. Sigmoid function is adopted for all the above activation functions. The filter components, weights and offsets obtained by off-line learning and the values of the output layer weights and offsets serve as initial values of a system algorithm and are applied to a microprocessor system programming and a substation 'five-small-box' fault identification algorithm of a convolutional neural network.
The sequence numbers in the above embodiments are merely for description, and do not represent the sequence of the assembly or the use of the components.
The above description is only exemplary of the present invention and should not be taken as limiting the invention, as any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (9)
1. A transformer substation five-small-box fault recognition system based on a convolutional neural network is characterized by comprising a temperature measurement sensing unit, a camera unit and a microprocessor, wherein the temperature measurement sensing unit is connected with the input end of the microprocessor through an I2C interface, and the camera unit is connected with the input end of the microprocessor through a serial camera control bus;
the temperature measurement sensing unit is used for acquiring first acquired images of an air switch and a relay in five small boxes of the transformer substation;
the camera shooting unit is used for collecting second collected images of an air switch and a relay in a five-small box of the transformer substation;
and the microprocessor is used for carrying out characteristic analysis on the first collected image and the second collected image through a built-in convolutional neural network and identifying the fault type according to the analysis result.
2. The convolutional neural network-based substation five-small-box fault recognition system as claimed in claim 1, wherein the first collected image is a real-time contact temperature image of air switches and relays in a five-small box.
3. The convolutional neural network-based substation five-small-box fault identification system as claimed in claim 1, wherein the second collected image is a real-time action state image of an air switch and a relay in a substation five-small box.
4. The system for identifying the fault of the five small boxes of the transformer substation based on the convolutional neural network as claimed in claim 1, wherein the microprocessor is used for carrying out real-time positioning and fault type identification on the opening and closing states of the air switch and the relay through a deep learning algorithm of the convolutional neural network.
5. The system for identifying the five-small-box fault of the transformer substation based on the convolutional neural network as claimed in claim 1, wherein the output result of the convolutional layer of the convolutional neural network is represented as:
wherein k is a sublayer number of the convolutional layer, i is a starting row number of pixel scanning of the first and second acquisition images, j is a starting column number of pixel scanning of the first and second acquisition images,shows the output of the convolutional layer k,respectively denote the filtering in the sub-layer kWeight of the device, x ij 、x ij+1 、…、x i+2j+2 Pixel representing input sublayer k, b Fb For a predetermined offset, a represents the activation function.
6. The substation five-small-box fault recognition system based on the convolutional neural network as claimed in claim 5, wherein the convolutional neural network completes training learning of collected images of the substation five-small box in advance based on an error back propagation method, and determines the number and size of filters adopted by convolutional layers in the convolutional neural network and the preset offset based on a training learning result.
7. The convolutional neural network-based substation five-small-box fault identification system as claimed in claim 1, wherein the pooling layer of the convolutional neural network is represented as:
wherein k is the sublayer number of the pooling layer,the output result of the pooling layer k is shown, i and j respectively show the row number and the column number of the neural unit in the sublayer k,respectively representing the inputs of the pooling layer sub-layer k,representing the activation function of pooling layer sublayer k.
8. The system for identifying the five-small box fault of the transformer substation based on the convolutional neural network as claimed in claim 1, wherein the number of the neural units of the output layer of the convolutional neural network is equal to the number of the fault types to be identified.
9. The convolutional neural network-based substation five-small-box fault recognition system as claimed in claim 1, further comprising a wireless transmission unit for sending the analysis result of the microprocessor to a fault monitoring platform.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211413860.9A CN115616329A (en) | 2022-11-11 | 2022-11-11 | Five little casees fault recognition system of transformer substation based on convolutional neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211413860.9A CN115616329A (en) | 2022-11-11 | 2022-11-11 | Five little casees fault recognition system of transformer substation based on convolutional neural network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115616329A true CN115616329A (en) | 2023-01-17 |
Family
ID=84879232
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211413860.9A Pending CN115616329A (en) | 2022-11-11 | 2022-11-11 | Five little casees fault recognition system of transformer substation based on convolutional neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115616329A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116247824A (en) * | 2023-03-30 | 2023-06-09 | 国网河南省电力公司安阳供电公司 | Control method and system for power equipment |
-
2022
- 2022-11-11 CN CN202211413860.9A patent/CN115616329A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116247824A (en) * | 2023-03-30 | 2023-06-09 | 国网河南省电力公司安阳供电公司 | Control method and system for power equipment |
CN116247824B (en) * | 2023-03-30 | 2023-11-17 | 国网河南省电力公司安阳供电公司 | Control method and system for power equipment |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN115527203B (en) | Cereal drying remote control method and system based on Internet of things | |
CN113436184B (en) | Power equipment image defect discriminating method and system based on improved twin network | |
CN115616329A (en) | Five little casees fault recognition system of transformer substation based on convolutional neural network | |
CN111797725A (en) | Equipment state detection method, device and system | |
CN101833673A (en) | Electric power switchgear switch state image recognition system | |
CN112839200B (en) | Power plant violation behavior identification method and system based on 5G technology and network server | |
CN116054417B (en) | Monitoring system and method for transformer substation | |
CN113177614A (en) | Image recognition system and method for power supply switch cabinet of urban rail transit | |
CN110944156A (en) | Terminal equipment for monitoring power distribution room | |
CN109740527B (en) | Image processing method in video frame | |
CN109141536A (en) | Seismographic station long-distance monitoring method, system and electronic equipment | |
CN113132689A (en) | Low-voltage centralized reading, operation and maintenance simulation device based on AI deep learning algorithm research | |
CN111314707B (en) | Data mapping identification method, device and equipment and readable storage medium | |
CN117347796A (en) | Intelligent gateway-based switching equipment partial discharge diagnosis system and method | |
CN116522274A (en) | Multi-mode monitoring method and device for intelligent power distribution room | |
CN113536607B (en) | Transformer substation signal transmission system evaluation method and system | |
CN113284103B (en) | Substation equipment defect online detection method based on space transformation fast R-CNN model | |
CN115931141A (en) | Temperature identification method of infrared temperature measurement map based on improved ANN algorithm | |
CN114118129A (en) | Method for detecting urban lighting facilities | |
CN115147591A (en) | Transformer equipment infrared image voltage heating type defect diagnosis method and system | |
CN114387391A (en) | Safety monitoring method and device for transformer substation equipment, computer equipment and medium | |
CN114708247A (en) | Cigarette case packaging defect identification method and device based on deep learning | |
CN109030982A (en) | A kind of client cables connector automatic monitoring system | |
CN113326728A (en) | Network anomaly detection method based on D-S evidence theory | |
CN205793036U (en) | A kind of power equipment video intelligent logging table analysis system |
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 |