CN115294411B - Power grid power transmission and transformation image data processing method based on neural network - Google Patents

Power grid power transmission and transformation image data processing method based on neural network Download PDF

Info

Publication number
CN115294411B
CN115294411B CN202211221948.0A CN202211221948A CN115294411B CN 115294411 B CN115294411 B CN 115294411B CN 202211221948 A CN202211221948 A CN 202211221948A CN 115294411 B CN115294411 B CN 115294411B
Authority
CN
China
Prior art keywords
power transmission
image
information
transformation
power grid
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.)
Active
Application number
CN202211221948.0A
Other languages
Chinese (zh)
Other versions
CN115294411A (en
Inventor
王光增
方建亮
刘国良
王云烨
周青睐
徐宏
孙钢
姚一杨
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
State Grid Zhejiang Electric Power Co Ltd
Original Assignee
State Grid Zhejiang Electric Power Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by State Grid Zhejiang Electric Power Co Ltd filed Critical State Grid Zhejiang Electric Power Co Ltd
Priority to CN202211221948.0A priority Critical patent/CN115294411B/en
Publication of CN115294411A publication Critical patent/CN115294411A/en
Application granted granted Critical
Publication of CN115294411B publication Critical patent/CN115294411B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • 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
    • G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
    • G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
    • G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
    • G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
    • G06F16/5866—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using information manually generated, e.g. tags, keywords, comments, manually generated location and time information
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
    • G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
    • G06F16/587—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using geographical or spatial information, e.g. location
    • 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
    • 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
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • 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
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Databases & Information Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Multimedia (AREA)
  • General Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Library & Information Science (AREA)
  • Artificial Intelligence (AREA)
  • Health & Medical Sciences (AREA)
  • Software Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Molecular Biology (AREA)
  • Medical Informatics (AREA)
  • Mathematical Physics (AREA)
  • Image Processing (AREA)

Abstract

The invention discloses a neural network-based power transmission and transformation image data processing method for a power grid, which comprises the following steps of: connecting input neurons, hidden neurons and output neurons belonging to the same category one by one; the server receives the power grid power transmission and transformation image sent by the acquisition end in real time, and the input layer of the neural network determines the input neuron corresponding to the power grid power transmission and transformation image according to the type information of the power grid power transmission and transformation image; determining the weight of the corresponding hidden neuron according to time dimension information and/or geographical dimension information corresponding to the power grid power transmission and transformation image substation, and inputting the power grid power transmission and transformation image to the corresponding hidden neuron by the input neuron; the hidden neuron compares the power grid power transmission and transformation image with a preset power transmission and transformation image to obtain an image comparison result; and counting the image comparison results of all the hidden neurons by the output neurons, and generating a final power transmission and transformation monitoring result of the power grid according to the image comparison results and the weights of the hidden neurons.

Description

Power grid power transmission and transformation image data processing method based on neural network
Technical Field
The invention relates to a data processing technology, in particular to a power grid power transmission and transformation image data processing method based on a neural network.
Background
With the acceleration of the pace of economic construction in China, the demand and the daily increase of electric energy are greatly increased, and the scale of equipment in a power grid is increased day by day. Due to the fact that the distance of the power transmission and transformation line of the power grid is very long, the climate change of the environment where each device is located is very violent, the distribution positions are very discrete, and the terrain spanned by the line is very complex, the inspection of the power transmission and transformation line and the device is difficult to implement and low in efficiency.
In recent years, a mode of adopting an unmanned aerial vehicle to replace an inspection worker to inspect the power grid power transmission and transformation line, and an image of the power grid power transmission and transformation line is collected to be used for the inspection worker to inspect, so that the inspection efficiency of the power grid power transmission and transformation line is improved. However, at present, the state of the power grid power transmission and transformation line is mainly recognized by manually observing a large number of inspection images, the mode is low in efficiency and easy to make mistakes, and meanwhile, the corresponding configuration data of different power grid power transmission and transformation lines are different. Therefore, how to combine different configuration data of the power transmission and transformation line of the power grid to carry out targeted and efficient intelligent processing on the power transmission and transformation image of the power grid becomes a technical problem which needs to be solved urgently.
Disclosure of Invention
The invention overcomes the defects of the prior art, provides a power grid power transmission and transformation image data processing method based on a neural network, and can construct a corresponding neural network by combining different configuration data of a power grid power transmission and transformation circuit so as to carry out targeted and efficient intelligent processing on the power grid power transmission and transformation image.
In order to solve the technical problems, the technical scheme of the invention is as follows:
the embodiment of the invention provides a power grid power transmission and transformation image data processing method based on a neural network, which comprises the following steps:
s1, initializing a neural network for image processing, wherein the neural network comprises an input layer, a hidden layer and an output layer which are sequentially connected, and the input layer, the hidden layer and the output layer are respectively provided with corresponding input neurons, hidden neurons and output neurons;
s2, classifying all input neurons, hidden neurons and output neurons according to configuration data, and connecting the input neurons, hidden neurons and output neurons belonging to the same category one by one;
s3, the server receives the power grid power transmission and transformation image sent by the acquisition end in real time, and the input layer of the neural network determines the input neuron corresponding to the power grid power transmission and transformation image according to the type information of the power grid power transmission and transformation image;
s4, determining the weight of the corresponding hidden neuron according to time dimension information and/or geographical dimension information corresponding to the power grid power transmission and transformation image, and inputting the power grid power transmission and transformation image to the corresponding hidden neuron by the input neuron;
s5, the hidden neurons compare the power grid power transmission and transformation image with a preset power transmission and transformation image to obtain an image comparison result, and the image comparison result is output to corresponding output neurons;
and S6, outputting image comparison results of all the hidden neurons of the neuron statistics, and generating a final power grid power transmission and transformation monitoring result according to the image comparison results and the weights of the hidden neurons.
Further, the S2 includes:
extracting the types of the electric transmission and transformation equipment in the configuration data, and dividing all input neurons according to the types of the electric transmission and transformation equipment so that each input neuron corresponds to different types of the electric transmission and transformation equipment;
extracting a fault type corresponding to each power transmission and transformation equipment type in the configuration data, establishing different hidden neurons according to the fault type, and connecting the hidden neurons with corresponding input neurons one by one;
and extracting a fault result corresponding to each power transmission and transformation equipment type in the configuration data, establishing different output neurons according to the fault result, and connecting the output neurons with corresponding hidden neurons.
Further, the S3 includes:
the method comprises the steps that a server side receives a power grid power transmission and transformation image sent by an acquisition side in real time and geographical position information of the corresponding acquisition side, and the geographical position information is compared with a preset position information table to obtain corresponding type information of the power grid power transmission and transformation image;
if the type information is judged to be larger than 1 or the type information is judged to be 0, carrying out image recognition on the power grid power transmission and transformation image to obtain the type information of the power grid power transmission and transformation image;
and if the type information of the power grid power transmission and transformation image cannot be obtained according to the image identification mode, outputting the power grid power transmission and transformation image, and adding the type information to the power grid power transmission and transformation image based on a worker.
Further, the server receives the power grid power transmission and transformation image sent by the acquisition end in real time and the geographical position information of the corresponding acquisition end, compares the geographical position information with a preset position information table, and obtains the corresponding type information of the power grid power transmission and transformation image, including:
determining a preset position interval corresponding to the geographic position information in a preset position information table, wherein the preset position information table is provided with a plurality of preset position intervals and a corresponding relation between each preset position interval and the type information;
if the geographic position information is within 1 preset position interval, taking the type information corresponding to the 1 preset position interval as the type information of the corresponding power transmission and transformation image of the power grid;
if the geographic position information is in the intersection area of the preset position intervals of a plurality of different types of information, the type information is greater than 1;
if the geographic position information is not located in all the preset position intervals, the category information is 0.
Further, if it is determined that the type information of the power grid power transmission and transformation image cannot be obtained according to the image recognition mode, outputting the power grid power transmission and transformation image, and adding the type information to the power grid power transmission and transformation image based on a worker includes:
taking all preset position intervals corresponding to the added type information as first preset position intervals, and acquiring first range intervals of the first preset position intervals, wherein the first range intervals comprise a first longitude sub-range and a first latitude sub-range;
calculating according to the first longitude sub-range and the first latitude sub-range of all the first range intervals to obtain a longitude offset range and a latitude offset range corresponding to the geographic position information;
calculating according to the geographic position information, the longitude offset range and the latitude offset range to obtain a second preset position interval corresponding to the geographic position information;
and correspondingly storing the added type information and the second preset position interval in a preset position information table, and updating the preset position information table.
Further, the calculating according to the geographic location information, the longitude offset range, and the latitude offset range to obtain a second preset location interval corresponding to the geographic location information includes:
longitude information and latitude information in the geographic position information are extracted, a longitude interval maximum value and a longitude interval minimum value are obtained according to the longitude information and a longitude offset range, and a latitude interval maximum value and a latitude interval minimum value are obtained according to the latitude information and the latitude offset range;
combining the longitude interval maximum value, the longitude interval minimum value, the latitude interval maximum value and the latitude interval minimum value to obtain a second preset position interval, calculating the longitude interval maximum value, the longitude interval minimum value, the latitude interval maximum value and the latitude interval minimum value through the following formulas,
Figure 765057DEST_PATH_IMAGE001
wherein the content of the first and second substances,
Figure 151039DEST_PATH_IMAGE002
is the minimum value of the longitude interval of the second preset position interval,
Figure 665197DEST_PATH_IMAGE003
as longitude information in the geographical location information,
Figure 794827DEST_PATH_IMAGE004
is a first
Figure 592887DEST_PATH_IMAGE005
A longitude maximum of a first longitude sub-range of the first range interval,
Figure 149771DEST_PATH_IMAGE006
is a first
Figure 151225DEST_PATH_IMAGE005
A minimum value of longitude of a first sub-range of longitude of the first range interval,
Figure 350125DEST_PATH_IMAGE007
is the upper limit value of the first temporal sub-range,
Figure 753424DEST_PATH_IMAGE008
is a quantitative value for the first measured sub-range,
Figure 481209DEST_PATH_IMAGE009
is the maximum value of the longitude interval of the second preset position interval,
Figure 704380DEST_PATH_IMAGE010
is the minimum value of the latitude interval of the second preset position interval,
Figure 175812DEST_PATH_IMAGE011
is as follows
Figure 682886DEST_PATH_IMAGE012
The maximum value of the first latitude subrange of the first range interval,
Figure 847151DEST_PATH_IMAGE013
is a first
Figure 823197DEST_PATH_IMAGE012
A minimum value of a first latitude subrange of the first range interval,mis the upper limit value of the first latitude sub-range,
Figure 832742DEST_PATH_IMAGE014
is a quantitative value of a first latitude sub-range,
Figure 210633DEST_PATH_IMAGE015
is the maximum value of the latitude interval of the second preset position interval.
Further, the S4 includes:
calling a preset weight corresponding table, wherein the weight corresponding table comprises a plurality of weight corresponding units, each weight corresponding unit corresponds to one hidden neuron, and each weight corresponding unit comprises a corresponding hidden neuron and corresponding weights under different time dimension information and/or geographic dimension information;
and comparing the time dimension information and/or the geographical dimension information corresponding to the power grid power transmission and transformation image station with the weight corresponding units of the corresponding weight corresponding table to obtain the weights of all the hidden neurons.
Further, the S5 includes:
the hidden nerve cell identifies the power transmission and transformation equipment in the power grid power transmission and transformation image to obtain first white light image information and/or first infrared image information, and the power grid power transmission and transformation image is a white light image and/or an infrared image;
comparing the first white light image information with preset white light image information, and/or comparing the first infrared image information with preset infrared image information to obtain an image comparison result;
if the image comparison result is judged to be a normal result, outputting the corresponding image comparison result to an output neuron corresponding to the normal result;
and if the image comparison result is judged to be an abnormal result, outputting the corresponding image comparison result to an output neuron corresponding to the abnormal result.
Further, the hidden neuron identifies the power transmission and transformation device in the power grid power transmission and transformation image to obtain first white light image information and/or first infrared image information, and the power grid power transmission and transformation image is a white light image and/or an infrared image, and the method includes:
if the power grid power transmission and transformation image is a white light image;
selecting all pixel points in the power grid power transmission and transformation image within a first preset pixel value interval as first pixel points;
counting all directly or indirectly connected first pixel points in the power transmission and transformation image of the power grid, and obtaining at least one first region of interest according to all the first pixel points and the region completely surrounded by the first pixel points;
and taking the image information corresponding to the first interested area as first white light image information.
Further, the hidden neuron identifies the power transmission and transformation device in the power grid power transmission and transformation image to obtain first white light image information and/or first infrared image information, and the power grid power transmission and transformation image is a white light image and/or an infrared image, and the method includes:
if the power grid power transmission and transformation image is an infrared image;
selecting all pixel points in a second preset pixel value interval in the power grid power transmission and transformation image as second pixel points;
and counting pixel gray values of all second pixel points in the power transmission and transformation image of the power grid, and taking the pixel gray values of all the second pixel points as first infrared image information.
Further, the comparing the first white light image information with preset white light image information, and/or comparing the first infrared image information with preset infrared image information to obtain an image comparison result, includes:
extracting all edge pixel points of the first interested area to obtain a first equipment outline;
extracting a second device profile corresponding to the preset white light image information, wherein each preset white light image information has a second device profile which is preset correspondingly;
if the first equipment contour is judged to correspond to the second equipment contour, outputting a sub-result with a normal contour, and otherwise, outputting a sub-result with an abnormal contour.
Further, the comparing the first white light image information with preset white light image information, and/or comparing the first infrared image information with preset infrared image information to obtain an image comparison result, includes:
extracting all pixel points of the first region of interest, and if judging that pixel points which do not correspond to a preset pixel value interval exist in all the pixel points of the first region of interest, outputting abnormal sub-results of the object, wherein the pixel points which do not correspond to the preset pixel value interval are surrounded by a plurality of first pixel points;
and if all the pixel points of the first interested area are judged to be the pixel points corresponding to the preset pixel value interval, outputting the normal sub-result of the object.
Further, the comparing the first white light image information with preset white light image information, and/or comparing the first infrared image information with preset infrared image information to obtain an image comparison result, includes:
calculating according to the pixel gray values of all the second pixel points to obtain an average gray pixel value;
counting all pixel points in the preset infrared image information within a second preset pixel value interval as comparison pixel points, and calculating according to pixel gray values of all the comparison pixel points to obtain comparison gray pixel values;
if the difference value between the average gray pixel value and the comparison gray pixel value is less than or equal to the comparison preset value, outputting a temperature normal sub-result;
and if the difference value between the average gray pixel value and the comparison gray pixel value is greater than the comparison preset value, outputting a temperature abnormity sub-result.
Further, the S6 includes:
if the normal output neurons corresponding to the normal results receive the normal sub-results output by all the hidden neurons, the output neurons directly output the normal power grid power transmission and transformation monitoring results;
if the abnormal output neuron corresponding to the abnormal result receives the abnormal sub-result output by the hidden neuron and does not have mutually exclusive abnormal sub-results, the abnormal output neuron outputs all the abnormal sub-results;
if the abnormal output neuron corresponding to the abnormal result receives the abnormal sub-result output by the hidden neuron and has mutually exclusive abnormal sub-results, the abnormal output neuron calls the weight corresponding to the mutually exclusive hidden neuron and determines the abnormal sub-result with the maximum weight in the mutually exclusive abnormal sub-results;
and outputting the abnormal sub-result with the maximum weight and the non-exclusive abnormal sub-result.
The invention has the beneficial effects that:
1. according to the scheme, the input neurons, the hidden neurons and the output neurons of the neural network are classified and configured by configuring the types of the power transmission and transformation equipment in the data and the corresponding fault results, so that different fault images of different power transformation equipment can be correspondingly processed, a relatively accurate power grid power transmission and transformation monitoring result is output, and a worker is assisted to make relatively accurate judgment. According to the scheme, the input neurons corresponding to the power transmission and transformation images are determined according to the type information of the power transmission and transformation images of the power grid, the image data of the corresponding power transformation equipment can be received in a classified mode, and the data can be classified and processed in a targeted, efficient and accurate mode subsequently; according to the scheme, the corresponding hidden neurons are determined according to the fault types corresponding to the types of the electric transmission and transformation equipment, and the hidden neurons are configured to process relevant fault data; the method can determine the weight of the hidden neuron according to the time dimension and the space dimension, and can determine result output with high possibility when multiple abnormal results are mutually exclusive.
2. According to the scheme, the geographical position information corresponding to the power grid power transmission and transformation image station is collected, and the type information of the corresponding power grid power transmission and transformation image is determined by combining the preset position information table, so that the image identification process can be omitted, the data processing amount is reduced, and the determination efficiency of the type information is improved; when the type information is more or the corresponding type information cannot be determined according to the position, the scheme can identify the power grid power transmission and transformation image to obtain the type information of the corresponding power grid power transmission and transformation image. In addition, the longitude deviation range and the latitude deviation range can be obtained by combining the position intervals of the same type of power transformation equipment, the preset position information table can be updated accurately, and the updated preset position information table can be compared comprehensively in the follow-up comparison.
3. The scheme for comparing the white light image information with the infrared image information is laid out. When white light image information is compared, the scheme can determine the region of interest and the corresponding equipment outline, and perform targeted comparison with the preset outline; when infrared image information is compared, the method and the device can perform targeted judgment by combining the pixel gray value. In addition, according to the scheme, after the comparison result is obtained, the abnormal sub-result with the highest possibility and the non-mutually-exclusive abnormal sub-result are determined according to the condition of the comparison result, and finally the abnormal sub-result with the highest weight and the non-mutually-exclusive abnormal sub-result are output to assist workers to make a more accurate fault solving strategy.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the description below are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained according to the drawings without inventive labor.
Fig. 1 is a schematic diagram of a neural network according to the present invention.
Detailed Description
In order that the manner in which the present invention is attained and can be more readily understood, a more particular description of the invention briefly summarized above may be had by reference to the embodiments thereof which are illustrated in the appended drawings.
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 other sequences 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, e.g., a 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, the term "if" may be interpreted as "at \8230; …" or "in response to a determination" or "in response to a detection" depending on the context.
The technical solution of the present invention will be described in detail below with 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.
The embodiment of the invention provides a power grid power transmission and transformation image data processing method based on a neural network, which comprises the following steps of S1-S6:
the method comprises the following steps of S1, initializing a neural network for image processing, wherein the neural network comprises an input layer, a hidden layer and an output layer which are sequentially connected, and the input layer, the hidden layer and the output layer are respectively provided with corresponding input neurons, hidden neurons and output neurons.
Referring to fig. 1, the schematic diagram of a neural network provided in this embodiment includes an input layer, a hidden layer, and an output layer, which are sequentially connected, where the input layer, the hidden layer, and the output layer respectively have corresponding input neurons, hidden neurons, and output neurons.
And S2, classifying all input neurons, hidden neurons and output neurons according to the configuration data, and connecting the input neurons, hidden neurons and output neurons belonging to the same category one by one.
Firstly, the scheme receives configuration data, and then classifies all input neurons, hidden neurons and output neurons by using the configuration data, so that the input neurons, the hidden neurons and the output neurons belonging to the same power transmission and transformation equipment type are connected one by one.
In some embodiments, said S2 comprises S21-S23:
and S21, extracting the types of the electric transmission and transformation equipment in the configuration data, and dividing all input neurons according to the types of the electric transmission and transformation equipment so that each input neuron corresponds to different types of the electric transmission and transformation equipment.
The configuration data of the scheme comprises the types of the power transmission and transformation equipment, and after the configuration data is received, the scheme analyzes the configuration data to obtain the types of the power transmission and transformation equipment, wherein the types of the power transmission and transformation equipment comprise the types of a transformer, a power transmission tower, a power transmission line and the like.
Referring to fig. 1, the scheme divides input neurons, and the diagram includes 2 input neurons corresponding to a transformer and 2 input neurons corresponding to a power transmission tower. It should be noted that, the figure only shows some neurons by way of example, in practical applications, there are a plurality of input neurons, hidden neurons, and output neurons, and the present solution is not shown in the detailed drawings.
And S22, extracting a fault type corresponding to each power transmission and transformation equipment type in the configuration data, establishing different hidden neurons according to the fault type, and connecting the hidden neurons with corresponding input neurons one by one.
It can be understood that the configuration data of the scheme further includes a fault type corresponding to each power transmission and transformation equipment type, and the scheme analyzes the configuration data to obtain the fault type corresponding to each power transmission and transformation equipment type. After the fault types are obtained, different hidden neurons can be established according to the fault types, and the hidden neurons are connected with corresponding input neurons one by one.
For example, for a transformer, the fault types of the transformer include two types, one type is that a suspension object exists on the transformer, and the other type is that the transformer is over-temperature, and after the fault type is obtained, a corresponding hidden neuron is established according to the scheme, the hidden neuron includes two types, one type is a hidden neuron for identifying whether the suspension object exists, the hidden neuron can identify a shot white light image and judge whether the suspension object exists, the other type is a hidden neuron for identifying whether the transformer is over-temperature, and the hidden neuron can identify a shot infrared image and judge whether the transformer is over-temperature.
And S23, extracting a fault result corresponding to each electric transmission and transformation equipment type in the configuration data, establishing different output neurons according to the fault result, and connecting the output neurons with corresponding hidden neurons.
For example, the failure result may be an abnormal result and a normal result, and then two corresponding output neurons may be established, and then the corresponding output neurons and the corresponding hidden neurons may be connected to form a neural network.
It can be understood that the neural network of the present solution may correspond to a plurality of electric transmission and transformation devices, and the input neuron, the hidden neuron and the output neuron corresponding to each electric transmission and transformation device are different.
And S3, the server receives the power grid power transmission and transformation image sent by the acquisition end in real time, and the input layer of the neural network determines the input neuron corresponding to the power grid power transmission and transformation image according to the type information of the power grid power transmission and transformation image.
The power grid power transmission and transformation image that the collection end sent can be received in real time to the server side of this scheme, and the collection end can be for example patrols and examines unmanned aerial vehicle, patrols and examines unmanned aerial vehicle and can gather power grid power transmission and transformation image in real time, then, the input layer of the neural network of this scheme can confirm rather than corresponding input neuron according to the kind information of power grid power transmission and transformation image. For example, after the power transmission and transformation image of the power grid is obtained, the power transmission and transformation image of the power grid is identified, the type information of the power transmission and transformation image of the power grid, such as a transformer, is obtained, and then the input neuron corresponding to the transformer is determined.
In some embodiments, the S3 includes S31-S33:
and S31, the server receives the power grid power transmission and transformation image sent by the acquisition end in real time and the geographical position information of the corresponding acquisition end, and compares the geographical position information with a preset position information table to obtain the corresponding type information of the power grid power transmission and transformation image.
It can be understood that when the acquisition end of the scheme acquires the power grid power transmission and transformation image, corresponding geographic position information can be generated, for example, longitude information and latitude information, the server end of the scheme can receive the power grid power transmission and transformation image sent by the acquisition end and the geographic position information of the corresponding acquisition end in real time, in addition, a preset position information table is further arranged, and then the geographic position information is compared with the preset position information table to obtain the type information of the corresponding power grid power transmission and transformation image.
It should be noted that, the preset position information table in the present embodiment includes a corresponding relationship between the geographic position and the power transmission and transformation equipment.
In some embodiments, S31 (the server receives, in real time, the power transmission and transformation image of the power grid sent by the collection end and the geographic location information of the corresponding collection end, and compares the geographic location information with a preset location information table to obtain the type information of the corresponding power transmission and transformation image of the power grid) includes S311 to S314:
s311, determining a preset position interval corresponding to the geographic position information in a preset position information table, wherein the preset position information table is provided with a plurality of preset position intervals and a corresponding relation between each preset position interval and the type information.
It can be understood that, because the power transmission and transformation line is long, and the span of the power transmission and transformation equipment is large, the preset position information table of the scheme has a plurality of preset position intervals, and each preset position interval has a corresponding relationship with the type information. For example, the preset position interval corresponding to the transformer a is as follows: the longitude is J1-J2 and the latitude is W1-W2.
And S312, if the geographic position information is in 1 preset position interval, taking the type information corresponding to the 1 preset position interval as the type information of the corresponding power transmission and transformation image of the power grid.
Illustratively, if the obtained geographic position information is that the longitude is between J1 and J2, and the latitude is between W1 and W2, and the geographic position information is located in the preset position interval, the type information of the collected power transmission and transformation image of the power grid is the transformer a.
S313, if the geographic position information is in the intersection area of the preset position intervals of the different types of information, the type information is greater than 1.
It can be understood that, in a power transmission and transformation line of a power grid, there may be a case where two power transmission and transformation devices are adjacent to each other, and at this time, the obtained geographic location information may be located in an intersection area of preset location intervals of a plurality of different types of information, that is, the type information is greater than 1, for example, the type information includes both a transformer and a power transmission tower.
S314, if the geographic location information is not located in all the preset location intervals, the category information is 0.
It can be understood that if the geographic location information is not located in all the preset location intervals, it is indicated that there is no corresponding power transmission and transformation equipment, or the preset location information table of the scheme is not updated in time, which results in no preset location interval corresponding to the power transmission and transformation equipment, at this time, the scheme may determine that the category information is 0.
And S32, if the type information is judged to be more than 1 or the type information is judged to be 0, carrying out image recognition on the power grid power transmission and transformation image to obtain the type information of the power grid power transmission and transformation image.
It can be understood that if the category information is determined to be greater than 1, it indicates that there are at least 2 power transmission and transformation point devices, the category information is 0, it indicates that there may be no power transmission and transformation device or power transmission and transformation device, but the preset position information table is not updated in time, and at this time, the scheme performs image recognition on the power grid power transmission and transformation image to obtain the category information of the power grid power transmission and transformation image.
By way of example, according to the scheme, the image recognition is performed on the power grid power transmission and transformation image, the contour of the power transmission and transformation equipment in the power grid power transmission and transformation image can be recognized and obtained by adopting an object contour recognition technology based on Opencv, and the recognized contour is compared with a preset contour of the power transmission and transformation equipment, so that the type information of the power grid power transmission and transformation image is obtained. For example, if the profile is identified as a transformer, the corresponding category information is a transformer, and if the profile is not identified, it can be determined that there is indeed no power transmission and transformation equipment at that location.
And S33, if the type information of the power grid power transmission and transformation image cannot be obtained according to the image identification mode, outputting the power grid power transmission and transformation image, and adding the type information to the power grid power transmission and transformation image based on a worker.
It can be understood that if the type information of the power grid power transmission and transformation image cannot be obtained according to the image recognition mode, the power grid power transmission and transformation image is output according to the scheme, and the worker adds the type information to the power grid power transmission and transformation image.
In some embodiments, S33 (outputting the power grid power transmission and transformation image if it is determined that the category information of the power grid power transmission and transformation image cannot be obtained according to the image recognition method, and adding the category information to the power grid power transmission and transformation image based on a worker) includes S331 to S334:
and S331, taking all the preset position intervals corresponding to the added type information as first preset position intervals, and acquiring a first range interval of the first preset position intervals, wherein the first range interval comprises a first longitude sub-range and a first latitude sub-range.
For example, the type information added by the operator may be a transformer, and in the present solution, all the preset location intervals corresponding to the added type information (transformer) are used as the first preset location interval, for example, there are 5 transformers on the power transmission line, so that the present solution obtains 5 preset location intervals corresponding to the 5 transformers as the first preset location interval, and then obtains a first range interval of the 5 first preset location intervals, where the first range interval includes a first longitude sub-range and a first latitude sub-range.
S332, calculating according to the first longitude sub-range and the first latitude sub-range of all the first range intervals to obtain a longitude offset range and a latitude offset range corresponding to the geographic position information.
After the first longitude sub-range and the first latitude sub-range of the first range interval are obtained, the longitude offset range and the latitude offset range corresponding to the geographic position information are obtained by calculating through the first longitude sub-range and the first latitude sub-range of all the first range intervals.
And S333, calculating according to the geographic position information, the longitude offset range and the latitude offset range to obtain a second preset position interval corresponding to the geographic position information.
After the geographic position information, the longitude deviation range and the latitude deviation range are obtained, the geographic position information, the longitude deviation range and the latitude deviation range are calculated to obtain a second preset position interval corresponding to the geographic position information.
In some embodiments, S333 (the calculation according to the geographic location information, the longitude offset range and the latitude offset range to obtain the second preset location interval corresponding to the geographic location information) includes S3331-S3332:
s3331, extracting longitude information and latitude information in the geographic position information, obtaining a longitude interval maximum value and a longitude interval minimum value according to the longitude information and the longitude offset range, and obtaining a latitude interval maximum value and a latitude interval minimum value according to the latitude information and the latitude offset range.
According to the scheme, longitude information and latitude information in the geographic position information are extracted, then the longitude information and the longitude offset range are combined to obtain the maximum value of a longitude interval and the minimum value of the longitude interval, and finally the latitude information and the latitude offset range are combined to obtain the maximum value of a latitude interval and the minimum value of the latitude interval.
S3332, combining the longitude interval maximum, longitude interval minimum, latitude interval maximum and latitude interval minimum to obtain a second preset position interval, calculating the longitude interval maximum, longitude interval minimum, latitude interval maximum and latitude interval minimum according to the following formulas,
Figure 280221DEST_PATH_IMAGE016
wherein the content of the first and second substances,
Figure 743563DEST_PATH_IMAGE002
is the minimum value of the longitude interval of the second preset position interval,
Figure 556798DEST_PATH_IMAGE003
as longitude information in the geographical location information,
Figure 41394DEST_PATH_IMAGE004
is as follows
Figure 547461DEST_PATH_IMAGE005
A maximum value of longitude of a first sub-range of longitude of the first range interval,
Figure 232521DEST_PATH_IMAGE017
is as follows
Figure 849447DEST_PATH_IMAGE005
A minimum value of longitude of a first sub-range of longitude of the first range interval,
Figure 936351DEST_PATH_IMAGE007
is the upper limit value of the first temporal sub-range,
Figure 613320DEST_PATH_IMAGE008
is a magnitude value of the first measured sub-range,
Figure 785676DEST_PATH_IMAGE009
is the maximum value of the longitude interval of the second preset position interval,
Figure 206293DEST_PATH_IMAGE010
is the minimum value of the latitude interval of the second preset position interval,
Figure 413283DEST_PATH_IMAGE011
is as follows
Figure 510421DEST_PATH_IMAGE012
The maximum value of the first latitude subrange of the first range interval,
Figure 170073DEST_PATH_IMAGE013
is as follows
Figure 128801DEST_PATH_IMAGE012
A minimum value of a first latitude subrange of the first range interval,mis the upper limit value of the first latitude sub-range,
Figure 659140DEST_PATH_IMAGE014
is a quantitative value of a first latitude sub-range,
Figure 677911DEST_PATH_IMAGE015
is the maximum value of the latitude interval of the second preset position interval.
In the above-mentioned formula,
Figure 824859DEST_PATH_IMAGE018
represents the first
Figure 587279DEST_PATH_IMAGE005
The maximum longitude and the second longitude of the first longitude sub-range of the first range interval
Figure 752550DEST_PATH_IMAGE005
A first longitude sub-range of the first range interval having a minimum longitudeThe difference in the values of the two or more,
Figure 676643DEST_PATH_IMAGE019
represents the first
Figure 310887DEST_PATH_IMAGE005
A longitude offset range of the first range interval;
Figure 611418DEST_PATH_IMAGE020
represents the first
Figure 647507DEST_PATH_IMAGE012
The maximum latitude value and the first latitude sub-range of the first range interval
Figure 742502DEST_PATH_IMAGE012
The difference of the latitudinal minimums of the first latitude subrange of the first range interval,
Figure 864042DEST_PATH_IMAGE021
represents the first
Figure 968264DEST_PATH_IMAGE022
A latitude offset range of the first range interval. After the longitude offset range and the latitude offset range are obtained, the scheme combines longitude information and latitude information in the geographic position information to obtain a maximum value of a longitude interval and a minimum value of the longitude interval, and finally combines the latitude information and the latitude offset range to obtain a maximum value of a latitude interval and a minimum value of the latitude interval.
And S334, correspondingly storing the added type information and the second preset position interval in a preset position information table, and updating the preset position information table.
After the second preset position interval is obtained, the added type information and the second preset position interval are correspondingly stored in the preset position information table, and the preset position information table is updated. It can be understood that, according to the scheme, when the position comparison is performed next time, the updated preset position information table can be used for comparison so as to match the corresponding power transformation equipment.
And S4, determining the weight of the corresponding hidden neuron according to the time dimension information and/or the geographical dimension information corresponding to the power grid power transmission and transformation image, and inputting the power grid power transmission and transformation image to the corresponding hidden neuron by the input neuron.
It should be noted that, in some scenarios, the present solution may adopt a plurality of hidden neurons to determine the input data, for example, for the type of the hanging object, the present solution may adopt 2 hidden neurons to determine, one hidden neuron is used to determine whether the hanging object is a sundry such as a plastic bag and clothes, and the other hidden neuron is used to determine whether the hanging object is an ice block, and under different time and space, the weights of the corresponding hidden neurons are different, and relatively accurate determination results are sequentially output.
According to the scheme, the weights of the corresponding hidden neurons are determined by combining time dimension information and/or geographical dimension information.
In some embodiments, the S4 includes S41-S42:
s41, a preset weight corresponding table is called, wherein the weight corresponding table comprises a plurality of weight corresponding units, each weight corresponding unit corresponds to one hidden neuron, each weight corresponding unit comprises a corresponding hidden neuron and corresponding weights under different time dimension information and/or geographic dimension information.
The scheme is provided with a weight corresponding table, the weight corresponding table comprises a plurality of weight corresponding units, each weight corresponding unit corresponds to one hidden neuron, each weight corresponding unit comprises a corresponding hidden neuron, and corresponding weights are set under different time dimension information and/or geographic dimension information.
For example, referring to the following table, to determine whether a pendant is a hidden neuron of a hanging ice cube, a weight correspondence table is provided:
weight of Time dimension information Geography dimension information
0.7 1-12 months Northeast China
0 4-10 months Hainan province
0.6 12-3 months Hainan province
And S42, comparing the time dimension information and/or the geographic dimension information corresponding to the power grid power transmission and transformation image station with the weight corresponding units of the corresponding weight corresponding table to obtain the weights of all the hidden neurons.
According to the scheme, time dimension information and/or geographical dimension information corresponding to the power grid power transmission and transformation image station can be compared with the weight corresponding units of the corresponding weight corresponding table to obtain the weights of all hidden neurons.
And S5, the hidden neurons compare the power grid power transmission and transformation image with a preset power transmission and transformation image to obtain an image comparison result, and the image comparison result is output to the corresponding output neurons.
According to the scheme, the hidden neurons can compare the power grid power transmission and transformation image with a preset power transmission and transformation image to obtain an image comparison result, and then the image comparison result is output to the corresponding output neurons.
The S5 comprises S51-S54:
s51, the hidden nerve cell identifies the power transmission and transformation equipment in the power grid power transmission and transformation image to obtain first white light image information and/or first infrared image information, and the power grid power transmission and transformation image is a white light image and/or an infrared image.
The power grid power transmission and transformation image is a white light image and/or an infrared image. When the images are white light images, the hidden neurons corresponding to the scheme can identify the power transmission and transformation equipment in the power transmission and transformation images of the power grid to obtain first white light image information, and when the images are infrared images, the hidden neurons corresponding to the scheme can identify the power transmission and transformation equipment in the power transmission and transformation images of the power grid to obtain first infrared image information. It can be understood that the white light image is identified, whether a suspended object exists can be judged, the infrared image is identified, and whether the power transformation equipment is over-temperature can be judged.
In some embodiments, S51 (the hidden neuron identifies the power transmission and transformation device in the power grid power transmission and transformation image to obtain first white light image information and/or first infrared image information, and the power grid power transmission and transformation image is a white light image and/or an infrared image) includes S511-S513:
and S511, if the power grid power transmission and transformation image is a white light image, selecting all pixel points in a first preset pixel value interval in the power grid power transmission and transformation image as first pixel points.
It can be understood that, if the power grid power transmission and transformation image is a white light image, all pixel points in the first preset pixel value interval in the power grid power transmission and transformation image are selected as first pixel points according to the scheme.
The first preset pixel value interval may be an interval corresponding to the power transformation device, for example, the surface of the power transmission tower is black, and then the first preset pixel value interval may be a pixel value interval corresponding to black.
S512, counting all directly or indirectly connected first pixel points in the power grid power transmission and transformation image, and obtaining at least one first region of interest according to all the first pixel points and the region completely surrounded by the first pixel points.
According to the scheme, all directly or indirectly connected first pixel points in the power grid power transmission and transformation image are counted to obtain at least one first region of interest. It can be understood that, when there is no impurity on the power transformation equipment, the first region of interest is a power transformation equipment region; when the power transformation equipment is provided with impurities, the first interested area is the power transformation equipment and the impurity area positioned on the power transformation equipment. When the sundries cover the power transformation equipment, the corresponding area is the area completely surrounded by the first pixel points.
And S513, using the image information corresponding to the first region of interest as first white light image information.
According to the scheme, after the first interested area is obtained, the image information corresponding to the first interested area is used as the first white light image information.
In other embodiments, S51 (the hidden neuron identifies the power transmission and transformation equipment in the power transmission and transformation grid image to obtain first white light image information and/or first infrared image information, where the power transmission and transformation grid image is a white light image and/or an infrared image) includes S514 to S515:
and S514, if the power grid power transmission and transformation image is an infrared image, selecting all pixel points in a second preset pixel value interval in the power grid power transmission and transformation image as second pixel points.
When the power grid power transmission and transformation image is the infrared image, all pixel points in the power grid power transmission and transformation image within a second preset pixel value interval can be selected as second pixel points. The second preset pixel value interval is a pixel value interval corresponding to a heating area in the power transformation equipment, and the pixel points corresponding to the heating area in the power transformation equipment can be selected as second pixel points.
And S515, counting the pixel gray values of all second pixel points in the power transmission and transformation image of the power grid, and taking the pixel gray values of all the second pixel points as first infrared image information.
According to the scheme, the pixel gray values of all the second pixels in the power transmission and transformation image of the power grid can be counted, the pixel gray values are the brightness of the corresponding pixels, it can be understood that the higher the temperature is, the higher the corresponding brightness is, and the pixel gray values of all the second pixels can be used as the first infrared image information. It can also be understood that, according to the scheme, whether the temperature of the power transformation equipment is too high can be known through the first infrared image information.
S52, comparing the first white light image information with preset white light image information, and/or comparing the first infrared image information with preset infrared image information to obtain an image comparison result;
after the first white light image information and the first infrared image information are obtained, the scheme compares the first white light image information with preset white light image information and/or compares the first infrared image information with preset infrared image information to obtain an image comparison result. The preset white light image information may be white light image information corresponding to the power transformation equipment without a suspended object state, and the preset infrared image information may be infrared image information corresponding to the power transformation equipment in a normal temperature state.
In some embodiments, S52 (comparing the first white light image information with preset white light image information, and/or comparing the first infrared image information with preset infrared image information to obtain an image comparison result) includes S521-S523:
and S521, extracting all edge pixel points of the first region of interest to obtain a first device outline.
It can be understood that the first region of interest is a white light image, and the scheme extracts all edge pixel points of the first region of interest to obtain the first device outline. For example, the outline of a transformer is obtained, and it should be noted that when a foreign object is hung on the transformer, the corresponding first region of interest also has a region corresponding to the foreign object, and therefore, the outline of the foreign object is also extracted according to the scheme.
And S522, extracting a second device profile corresponding to the preset white light image information, wherein each preset white light image information has a second device profile preset correspondingly.
It can be understood that the preset white light image information is preset, in the scheme, the second device profile corresponding to the preset white light image information is extracted, and each preset white light image information has the second device profile preset corresponding to the preset white light image information. For example, the transformer has a second device profile corresponding thereto.
S523, if the first equipment contour is judged to correspond to the second equipment contour, outputting a sub-result with a normal contour, otherwise, outputting a sub-result with an abnormal contour.
It can be understood that if the first device profile and the second device profile are judged to correspond to each other, it is indicated that no other object exists on the power transformation device, and the phenomenon is normal, and at this time, a sub-result that the profiles are normal can be output; when the first equipment outline and the second equipment outline are judged not to correspond to each other, it is indicated that other objects exist on the power transformation equipment, for example, ice hangs on the power transformation equipment, and an abnormal sub-result of the outline can be output at the moment.
In other embodiments, S52 (the comparing the first white-light image information with preset white-light image information, and/or comparing the first infrared image information with preset infrared image information to obtain an image comparison result) includes S524-S525:
and S524, extracting all pixel points of the first region of interest, and if it is judged that pixel points which do not correspond to the preset pixel value interval exist in all the pixel points of the first region of interest, outputting abnormal sub-results of the object, wherein the pixel points which do not correspond to the preset pixel value interval are surrounded by the plurality of first pixel points.
According to the scheme, all pixel points in the first interested area can be extracted, whether pixel points which do not correspond to the preset pixel value interval exist in all the pixel points of the first interested area is judged, if the pixel points exist, the fact that the covering exists on the power transformation equipment is proved, at the moment, abnormal sub-results of the object can be output, and the pixel points which do not correspond to the covering exist in the preset pixel value interval are surrounded by the first pixel points.
And S525, if all the pixel points of the first region of interest are judged to be pixel points corresponding to the preset pixel value interval, outputting a normal sub-result of the object.
It can be understood that if all the pixel points in the first region of interest are determined to be pixel points corresponding to the preset pixel value interval, it is indicated that no other article exists in the first region of interest, and all the pixel points are pixel points corresponding to the power transformation equipment, and at this moment, the normal sub-result of the object can be output.
In still other embodiments, S52 (comparing the first white light image information with preset white light image information, and/or comparing the first infrared image information with preset infrared image information to obtain an image comparison result) includes S526 to S529:
and S526, calculating according to the pixel gray values of all the second pixel points to obtain an average gray pixel value.
After the pixel gray values of the second pixel points are obtained, the scheme can perform weighted average calculation on the pixel gray values of all the second pixel points to obtain an average gray pixel value.
And S527, counting all pixel points in the preset infrared image information within the second preset pixel value interval as comparison pixel points, and calculating according to pixel gray values of all the comparison pixel points to obtain comparison gray pixel values.
According to the scheme, all pixel points in the preset infrared image information within the second preset pixel value interval are counted as comparison pixel points, and then pixel gray values of all the comparison pixel points are calculated to obtain comparison gray pixel values.
S528, if the difference value between the average gray pixel value and the comparison gray pixel value is smaller than or equal to the comparison preset value, outputting a sub-result with normal temperature.
It can be understood that, if the difference between the average grayscale pixel value and the comparison grayscale pixel value is less than or equal to the comparison preset value, it indicates that the current temperature of the power transformation device is within the normal temperature interval, and at this time, the scheme may output a sub-result of normal temperature.
And S529, if the difference value of the average gray level pixel value and the comparison gray level pixel value is larger than the comparison preset value, outputting a temperature abnormal sub-result.
It can be understood that, if the difference between the average gray level pixel value and the comparison gray level pixel value is greater than the comparison preset value, it indicates that the current temperature of the power transformation equipment is higher than the normal temperature, and at this time, the present scheme may output a temperature anomaly result.
And S53, if the image comparison result is judged to be a normal result, outputting the corresponding image comparison result to the output neuron corresponding to the normal result.
It can be understood that, in the neural network of the present solution, an output neuron corresponding to outputting a normal result and an output neuron corresponding to outputting an abnormal result may be set. And if the image comparison result is a normal result, outputting the comparison result of the corresponding image to an output neuron corresponding to the normal result.
And S54, if the image comparison result is judged to be an abnormal result, outputting the corresponding image comparison result to an output neuron corresponding to the abnormal result.
If the image comparison result is an abnormal result, the scheme outputs the corresponding image comparison result to the output neuron corresponding to the abnormal result.
And S6, outputting image comparison results of all the hidden neurons of the neuron statistics, and generating a final power grid power transmission and transformation monitoring result according to the image comparison results and the weights of the hidden neurons.
The output neurons of the scheme can count the image comparison results of all the hidden neurons, and then the final power grid power transmission and transformation monitoring results which are accurate are generated by using the image comparison results and the weights of the hidden neurons.
In some embodiments, said S6 comprises S61-S64:
and S61, if the normal output neurons corresponding to the normal results receive the normal sub-results output by all the hidden neurons, directly outputting the normal sub-results as normal power transmission and transformation monitoring results of the power grid by the output neurons.
It can be understood that, if the normal output neuron corresponding to the normal result receives the normal sub-results output by all the hidden neurons, it indicates that all the comparison results are normal, for example, the power transformation equipment has no hanging object and is not over-temperature, and at this time, the output neuron of the scheme directly outputs the normal power grid power transmission and transformation monitoring result.
And S62, if the abnormal output neuron corresponding to the abnormal result receives the abnormal sub-result output by the hidden neuron and does not have mutually exclusive abnormal sub-results, the abnormal output neuron outputs all the abnormal sub-results.
It can be understood that, if the abnormal output neuron corresponding to the abnormal result receives the abnormal sub-result output by the hidden neuron and does not have mutually exclusive abnormal sub-results, the present scheme outputs all the abnormal sub-results to the abnormal output neuron.
For example, a plastic bag is hung on the transformer to hide the abnormal sub-result output by the neuron, and an over-temperature transformer is hung on the other abnormal sub-result output by the neuron.
S63, if the abnormal output neuron corresponding to the abnormal result receives the abnormal sub-result output by the hidden neuron and has mutually exclusive abnormal sub-results, the abnormal output neuron calls the weight corresponding to the mutually exclusive hidden neuron and determines the abnormal sub-result with the maximum weight in the mutually exclusive abnormal sub-results.
For example, a plastic bag is hung on the transformer as an abnormal sub-result output by one hidden neuron, and ice blocks are hung on the transformer as an abnormal sub-result output by the other hidden neuron, at this time, mutual exclusion exists between the two abnormal sub-results of the plastic bag hung on the transformer and the ice blocks hung on the transformer, and the abnormal output neuron of the scheme calls the weight corresponding to the mutually exclusive hidden neuron and determines the abnormal sub-result with the largest weight in the mutually exclusive abnormal sub-results. It is understood that the greatest weight indicates the highest likelihood of the corresponding result. For example, in northeast of 12 months, the weight of the ice cubes as the suspended objects is the largest, which indicates that the probability that the ice cubes are the suspended objects is the largest at this time, and then the scheme can output the abnormal sub-result that the ice cubes are the suspended objects. It should be noted that the abnormal sub-result is not necessarily the most accurate result, but is the most probable result in the current scenario, and after the result is output, the result may assist the staff to make a corresponding judgment.
And S64, outputting the abnormal sub-result with the maximum weight and the non-exclusive abnormal sub-result.
After the abnormal sub-result with the maximum weight and the non-exclusive abnormal sub-result are obtained, the abnormal sub-result with the maximum weight and the non-exclusive abnormal sub-result can be output to prompt the staff in time.
Finally, it is to be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.

Claims (10)

1. A power grid power transmission and transformation image data processing method based on a neural network is characterized by comprising the following steps:
s1, initializing a neural network for image processing, wherein the neural network comprises an input layer, a hidden layer and an output layer which are sequentially connected, and the input layer, the hidden layer and the output layer are respectively provided with corresponding input neurons, hidden neurons and output neurons;
s2, classifying all input neurons, hidden neurons and output neurons according to configuration data, and connecting the input neurons, hidden neurons and output neurons belonging to the same category one by one;
s3, the server receives the power grid power transmission and transformation image sent by the acquisition end in real time, and the input layer of the neural network determines the input neurons corresponding to the power grid power transmission and transformation image according to the type information of the power grid power transmission and transformation image;
s4, determining the weight of the corresponding hidden neuron according to time dimension information and/or geographical dimension information corresponding to the power grid power transmission and transformation image station, and inputting the power grid power transmission and transformation image to the corresponding hidden neuron by using an input neuron;
s5, the hidden neuron compares the power grid power transmission and transformation image with a preset power transmission and transformation image to obtain an image comparison result, and the image comparison result is output to a corresponding output neuron;
s6, outputting image comparison results of all the hidden neurons of the neuron statistics, and generating a final power transmission and transformation monitoring result of the power grid according to the image comparison results and the weights of the hidden neurons;
the S2 comprises:
extracting the types of the electric transmission and transformation equipment in the configuration data, and dividing all input neurons according to the types of the electric transmission and transformation equipment so as to enable each input neuron to correspond to different types of the electric transmission and transformation equipment;
extracting a fault type corresponding to each power transmission and transformation equipment type in the configuration data, establishing different hidden neurons according to the fault type, and connecting the hidden neurons with corresponding input neurons one by one;
extracting a fault result corresponding to each power transmission and transformation equipment type in the configuration data, establishing different output neurons according to the fault result, and connecting the output neurons with corresponding hidden neurons;
the S3 comprises the following steps:
the server receives the power grid power transmission and transformation images sent by the acquisition end in real time and the geographical position information of the corresponding acquisition end, and compares the geographical position information with a preset position information table to obtain the corresponding type information of the power grid power transmission and transformation images;
if the type information is judged to be more than 1 or the type information is judged to be 0, carrying out image recognition on the power grid power transmission and transformation image to obtain the type information of the power grid power transmission and transformation image;
if the type information of the power grid power transmission and transformation image cannot be obtained according to the image recognition mode, outputting the power grid power transmission and transformation image, and adding the type information to the power grid power transmission and transformation image based on workers;
the S5 comprises the following steps:
the hidden nerve cell identifies the power transmission and transformation equipment in the power grid power transmission and transformation image to obtain first white light image information and/or first infrared image information, and the power grid power transmission and transformation image is a white light image and/or an infrared image;
comparing the first white light image information with preset white light image information, and/or comparing the first infrared image information with preset infrared image information to obtain an image comparison result;
if the image comparison result is judged to be a normal result, outputting the corresponding image comparison result to an output neuron corresponding to the normal result;
if the image comparison result is judged to be an abnormal result, outputting the corresponding image comparison result to an output neuron corresponding to the abnormal result;
the S6 comprises the following steps:
if the normal output neurons corresponding to the normal results receive the normal sub-results output by all the hidden neurons, the output neurons directly output the normal power grid power transmission and transformation monitoring results;
if the abnormal output neuron corresponding to the abnormal result receives the abnormal sub-result output by the hidden neuron and does not have mutually exclusive abnormal sub-results, the abnormal output neuron outputs all the abnormal sub-results;
if the abnormal output neurons corresponding to the abnormal results receive the abnormal sub-results output by the hidden neurons and have mutually exclusive abnormal sub-results, the abnormal output neurons call weights corresponding to the mutually exclusive hidden neurons, and the abnormal sub-results with the maximum weights in the mutually exclusive abnormal sub-results are determined;
and outputting the abnormal sub-result with the maximum weight and the non-exclusive abnormal sub-result.
2. The neural network-based power grid power transmission and transformation image data processing method according to claim 1,
the server receives the power grid power transmission and transformation images sent by the acquisition end in real time and the geographical position information of the corresponding acquisition end, compares the geographical position information with a preset position information table, and obtains the corresponding type information of the power grid power transmission and transformation images, and the method comprises the following steps:
determining a preset position interval corresponding to the geographic position information in a preset position information table, wherein the preset position information table is provided with a plurality of preset position intervals and a corresponding relation between each preset position interval and the type information;
if the geographic position information is within 1 preset position interval, taking the type information corresponding to the 1 preset position interval as the type information of the corresponding power transmission and transformation image of the power grid;
if the geographic position information is in an intersection area of preset position intervals of a plurality of different types of information, the type information is larger than 1;
if the geographic position information is not located in all the preset position intervals, the category information is 0.
3. The neural network-based power transmission and transformation image data processing method of the power grid according to claim 2,
if the type information of the power grid power transmission and transformation image cannot be obtained according to the image recognition mode, outputting the power grid power transmission and transformation image, and adding the type information to the power grid power transmission and transformation image based on workers, wherein the method comprises the following steps:
taking all preset position intervals corresponding to the added type information as first preset position intervals, and acquiring first range intervals of the first preset position intervals, wherein the first range intervals comprise a first longitude sub-range and a first latitude sub-range;
calculating according to a first longitude sub-range and a first latitude sub-range of all the first range intervals to obtain a longitude offset range and a latitude offset range corresponding to the geographic position information;
calculating according to the geographic position information, the longitude offset range and the latitude offset range to obtain a second preset position interval corresponding to the geographic position information;
and correspondingly storing the added type information and the second preset position interval in a preset position information table, and updating the preset position information table.
4. The neural network-based power grid power transmission and transformation image data processing method according to claim 3,
the calculating according to the geographic position information, the longitude offset range and the latitude offset range to obtain a second preset position interval corresponding to the geographic position information includes:
extracting longitude information and latitude information in the geographic position information, obtaining a longitude interval maximum value and a longitude interval minimum value according to the longitude information and a longitude offset range, and obtaining a latitude interval maximum value and a latitude interval minimum value according to the latitude information and a latitude offset range;
combining the longitude interval maximum value, the longitude interval minimum value, the latitude interval maximum value and the latitude interval minimum value to obtain a second preset position interval, calculating the longitude interval maximum value, the longitude interval minimum value, the latitude interval maximum value and the latitude interval minimum value through the following formulas,
Figure DEST_PATH_IMAGE001
Figure 996794DEST_PATH_IMAGE002
wherein, the first and the second end of the pipe are connected with each other,
Figure DEST_PATH_IMAGE003
is the minimum value of the longitude interval of the second preset position interval,
Figure 831895DEST_PATH_IMAGE004
as the longitude information in the geographical location information,
Figure 139642DEST_PATH_IMAGE005
is a first
Figure 314271DEST_PATH_IMAGE006
A longitude maximum of a first longitude sub-range of the first range interval,
Figure 711754DEST_PATH_IMAGE007
is a first
Figure 971834DEST_PATH_IMAGE006
A longitude minimum of a first longitude sub-range of the first range interval,
Figure 847387DEST_PATH_IMAGE008
is the upper limit value of the first temporal sub-range,
Figure 876522DEST_PATH_IMAGE009
is a magnitude value of the first measured sub-range,
Figure 943442DEST_PATH_IMAGE010
is the maximum value of the longitude interval of the second preset position interval,
Figure 690818DEST_PATH_IMAGE011
is the minimum value of the latitude interval of the second preset position interval,
Figure 370061DEST_PATH_IMAGE012
is as follows
Figure 253704DEST_PATH_IMAGE013
The maximum value of the first latitude sub-range of the first range interval,
Figure 992990DEST_PATH_IMAGE014
is as follows
Figure 227662DEST_PATH_IMAGE013
A minimum latitude value of a first latitude sub-range of the first range interval,
Figure 946481DEST_PATH_IMAGE015
is the upper limit value of the first latitude sub-range,
Figure 215789DEST_PATH_IMAGE016
is a quantitative value of a first latitude sub-range,
Figure 860397DEST_PATH_IMAGE017
is the maximum value of the latitude interval of the second preset position interval.
5. The neural network-based power grid power transmission and transformation image data processing method according to claim 1,
the S4 comprises the following steps:
calling a preset weight corresponding table, wherein the weight corresponding table comprises a plurality of weight corresponding units, each weight corresponding unit corresponds to one hidden neuron, and each weight corresponding unit comprises a corresponding hidden neuron and corresponding weights under different time dimension information and/or geographic dimension information;
and comparing the time dimension information and/or the geographical dimension information corresponding to the power grid power transmission and transformation image station with the weight corresponding units of the corresponding weight corresponding table to obtain the weights of all the hidden neurons.
6. The neural network-based power transmission and transformation image data processing method of the power grid according to claim 1,
the hidden neuron identifies the power transmission and transformation equipment in the power grid power transmission and transformation image to obtain first white light image information and/or first infrared image information, the power grid power transmission and transformation image is a white light image and/or an infrared image, and the hidden neuron comprises:
if the power grid power transmission and transformation image is a white light image;
selecting all pixel points in a first preset pixel value interval in the power grid power transmission and transformation image as first pixel points;
counting all directly or indirectly connected first pixel points in the power transmission and transformation image of the power grid, and obtaining at least one first region of interest according to all the first pixel points and the region completely surrounded by the first pixel points;
and taking the image information corresponding to the first region of interest as first white light image information.
7. The neural network-based power grid power transmission and transformation image data processing method according to claim 6,
the hidden neuron identifies the power transmission and transformation equipment in the power grid power transmission and transformation image to obtain first white light image information and/or first infrared image information, the power grid power transmission and transformation image is a white light image and/or an infrared image, and the hidden neuron comprises:
if the power grid power transmission and transformation image is an infrared image;
selecting all pixel points in a second preset pixel value interval in the power grid power transmission and transformation image as second pixel points;
and (4) counting the pixel gray values of all second pixel points in the power transmission and transformation image of the power grid, and taking the pixel gray values of all the second pixel points as the first infrared image information.
8. The neural network-based power transmission and transformation image data processing method of the power grid according to claim 7,
the comparing the first white light image information with preset white light image information and/or comparing the first infrared image information with preset infrared image information to obtain an image comparison result includes:
extracting all edge pixel points of the first region of interest to obtain a first equipment outline;
extracting a second device outline corresponding to the preset white light image information, wherein each preset white light image information has a second device outline which is preset correspondingly;
if the first equipment contour is judged to correspond to the second equipment contour, outputting a sub-result with a normal contour, and otherwise, outputting a sub-result with an abnormal contour.
9. The neural network-based power transmission and transformation image data processing method of the power grid according to claim 8,
the comparing the first white light image information with preset white light image information and/or comparing the first infrared image information with preset infrared image information to obtain an image comparison result, includes:
extracting all pixel points of the first region of interest, and if judging that pixel points which do not correspond to a preset pixel value interval exist in all the pixel points of the first region of interest, outputting abnormal sub-results of the object, wherein the pixel points which do not correspond to the preset pixel value interval are surrounded by a plurality of first pixel points;
and if all the pixel points of the first interested area are judged to be the pixel points corresponding to the preset pixel value interval, outputting the normal sub-result of the object.
10. The neural network-based power transmission and transformation image data processing method of the power grid according to claim 9,
the comparing the first white light image information with preset white light image information and/or comparing the first infrared image information with preset infrared image information to obtain an image comparison result includes:
calculating according to the pixel gray values of all the second pixel points to obtain an average gray pixel value;
counting all pixel points in the preset infrared image information within a second preset pixel value interval as comparison pixel points, and calculating according to pixel gray values of all the comparison pixel points to obtain comparison gray pixel values;
if the difference value between the average gray pixel value and the comparison gray pixel value is less than or equal to the comparison preset value, outputting a temperature normal sub-result;
and if the difference value between the average gray pixel value and the comparison gray pixel value is greater than the comparison preset value, outputting a temperature abnormity sub-result.
CN202211221948.0A 2022-10-08 2022-10-08 Power grid power transmission and transformation image data processing method based on neural network Active CN115294411B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211221948.0A CN115294411B (en) 2022-10-08 2022-10-08 Power grid power transmission and transformation image data processing method based on neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211221948.0A CN115294411B (en) 2022-10-08 2022-10-08 Power grid power transmission and transformation image data processing method based on neural network

Publications (2)

Publication Number Publication Date
CN115294411A CN115294411A (en) 2022-11-04
CN115294411B true CN115294411B (en) 2022-12-30

Family

ID=83834716

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211221948.0A Active CN115294411B (en) 2022-10-08 2022-10-08 Power grid power transmission and transformation image data processing method based on neural network

Country Status (1)

Country Link
CN (1) CN115294411B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116051497B (en) * 2023-01-04 2023-07-21 杭州启泰信息科技有限公司 Intelligent analysis method for power transmission and transformation images of power grid based on data processing

Citations (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105978144A (en) * 2016-05-17 2016-09-28 南宁正相电气科技有限公司 Infrared monitoring system for automatic cruise transformer station based on APP platform alarm
CN109146062A (en) * 2018-08-14 2019-01-04 视云融聚（广州）科技有限公司 A kind of Space Reconstruction method of video neuron node
CN109784225A (en) * 2018-12-28 2019-05-21 南京大学 A kind of geographical space mode identification method based on reachable tree
EP3576019A1 (en) * 2018-05-29 2019-12-04 Nokia Technologies Oy Artificial neural networks
CN110674931A (en) * 2019-09-29 2020-01-10 河海大学常州校区 Weight importance-based full-connection neural network optimization method and device
CN110929847A (en) * 2019-11-15 2020-03-27 国网浙江省电力有限公司电力科学研究院 Converter transformer fault diagnosis method based on deep convolutional neural network
CN111797986A (en) * 2019-04-09 2020-10-20 Oppo广东移动通信有限公司 Data processing method, data processing device, storage medium and electronic equipment
CN111813130A (en) * 2020-08-19 2020-10-23 江南大学 Autonomous navigation obstacle avoidance system of intelligent patrol robot of power transmission and transformation station
CN112784976A (en) * 2021-01-15 2021-05-11 中山大学 Image recognition system and method based on impulse neural network
CN114091549A (en) * 2021-09-28 2022-02-25 国网江苏省电力有限公司苏州供电分公司 Equipment fault diagnosis method based on deep residual error network
CN114266387A (en) * 2021-12-02 2022-04-01 国网宁夏电力有限公司经济技术研究院 Power transmission and transformation project construction period prediction method, system, equipment and storage medium
CN114881225A (en) * 2022-04-24 2022-08-09 中国电力科学研究院有限公司 Power transmission and transformation inspection model network structure searching method, system and storage medium

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112906769A (en) * 2021-02-04 2021-06-04 国网河南省电力公司电力科学研究院 Power transmission and transformation equipment image defect sample amplification method based on cycleGAN
CN114399496A (en) * 2022-01-17 2022-04-26 浙江大立科技股份有限公司 Method and device for automatically identifying power grid equipment from infrared image

Patent Citations (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105978144A (en) * 2016-05-17 2016-09-28 南宁正相电气科技有限公司 Infrared monitoring system for automatic cruise transformer station based on APP platform alarm
EP3576019A1 (en) * 2018-05-29 2019-12-04 Nokia Technologies Oy Artificial neural networks
CN109146062A (en) * 2018-08-14 2019-01-04 视云融聚（广州）科技有限公司 A kind of Space Reconstruction method of video neuron node
CN109784225A (en) * 2018-12-28 2019-05-21 南京大学 A kind of geographical space mode identification method based on reachable tree
CN111797986A (en) * 2019-04-09 2020-10-20 Oppo广东移动通信有限公司 Data processing method, data processing device, storage medium and electronic equipment
CN110674931A (en) * 2019-09-29 2020-01-10 河海大学常州校区 Weight importance-based full-connection neural network optimization method and device
CN110929847A (en) * 2019-11-15 2020-03-27 国网浙江省电力有限公司电力科学研究院 Converter transformer fault diagnosis method based on deep convolutional neural network
CN111813130A (en) * 2020-08-19 2020-10-23 江南大学 Autonomous navigation obstacle avoidance system of intelligent patrol robot of power transmission and transformation station
CN112784976A (en) * 2021-01-15 2021-05-11 中山大学 Image recognition system and method based on impulse neural network
CN114091549A (en) * 2021-09-28 2022-02-25 国网江苏省电力有限公司苏州供电分公司 Equipment fault diagnosis method based on deep residual error network
CN114266387A (en) * 2021-12-02 2022-04-01 国网宁夏电力有限公司经济技术研究院 Power transmission and transformation project construction period prediction method, system, equipment and storage medium
CN114881225A (en) * 2022-04-24 2022-08-09 中国电力科学研究院有限公司 Power transmission and transformation inspection model network structure searching method, system and storage medium

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
Application of Back Propagation Neuron Network on Data Linkage Transmission of Semiconductor Hydrogen Detection Device;Yi-Chi 等;《2017 IEEE 8th International Conference on Awareness Science and Technology》;20171231;268-272 *
一种基于监督学习的输电线监测中杆塔的检测方法;王孝余 等;《东北电力技术》;20171231;第38卷(第11期);12-14、19 *
基于Haar－CNN模型的自然场景图像分类的研究;张慧娜 等;《四川师范大学学报( 自然科学版)》;20170131;第40卷(第1期);119-126 *

Also Published As

Publication number Publication date
CN115294411A (en) 2022-11-04

Similar Documents

Publication Publication Date Title
CN111259892B (en) Inspection method, inspection device, inspection equipment and inspection medium for state of indicator lamp
CN108898077B (en) Power equipment infrared chart identification method and power equipment infrared identification system
CN109101906A (en) A kind of converting station electric power equipment infrared image exception real-time detection method and device
CN109858367B (en) Visual automatic detection method and system for worker through supporting unsafe behaviors
CN115294411B (en) Power grid power transmission and transformation image data processing method based on neural network
CN103812577A (en) Method for automatically identifying and learning abnormal radio signal type
CN110929592A (en) Extraction method and system for outer boundary of mariculture area
CN110765848A (en) Chemical plant personnel safety guarantee system and early warning method based on artificial intelligence image processing algorithm
CN111428617A (en) Video image-based distribution network violation maintenance behavior identification method and system
CN111209832B (en) Auxiliary obstacle avoidance training method, equipment and medium for substation inspection robot
CN103049739A (en) Tree detection method for use in intelligent monitoring of power transmission line
CN107563356A (en) A kind of unmanned plane inspection pipeline target analysis management method and system
CN106936517A (en) A kind of automatic recognition system and its method of abnormal radio signal
CN112461828A (en) Intelligent pest and disease damage forecasting and early warning system based on convolutional neural network
CN116805204B (en) Intelligent plant monitoring method and system
CN113255691A (en) Method for detecting and identifying harmful bird species target of bird-involved fault of power transmission line
CN117370919B (en) Remote monitoring system for sewage treatment equipment
CN112734637B (en) Thermal infrared image processing method and system for monitoring temperature of lead
CN108876144A (en) A kind of pre- site selecting method of substation based on deep learning algorithm
CN113762115B (en) Distribution network operator behavior detection method based on key point detection
CN116559713A (en) Intelligent monitoring method and device for power supply of communication base station
CN113361968B (en) Power grid infrastructure worker safety risk assessment method based on artificial intelligence and big data
CN115018777A (en) Power grid equipment state evaluation method and device, computer equipment and storage medium
CN111930982A (en) Intelligent labeling method for power grid images
CN110196152A (en) The method for diagnosing faults and system of large-scale landscape lamp group based on machine vision

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