CN115049988A - Edge calculation method and device for power distribution network monitoring and prejudging - Google Patents

Edge calculation method and device for power distribution network monitoring and prejudging Download PDF

Info

Publication number
CN115049988A
CN115049988A CN202210983658.3A CN202210983658A CN115049988A CN 115049988 A CN115049988 A CN 115049988A CN 202210983658 A CN202210983658 A CN 202210983658A CN 115049988 A CN115049988 A CN 115049988A
Authority
CN
China
Prior art keywords
monitoring
dif
module
abnormal
power distribution
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202210983658.3A
Other languages
Chinese (zh)
Inventor
符杰
习伟
蔡田田
陈波
邓清唐
杨英杰
朱明增
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.)
Southern Power Grid Digital Grid Research Institute Co Ltd
Original Assignee
Southern Power Grid Digital Grid Research Institute 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 Southern Power Grid Digital Grid Research Institute Co Ltd filed Critical Southern Power Grid Digital Grid Research Institute Co Ltd
Priority to CN202210983658.3A priority Critical patent/CN115049988A/en
Publication of CN115049988A publication Critical patent/CN115049988A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
    • G01D21/00—Measuring or testing not otherwise provided for
    • G01D21/02—Measuring two or more variables by means not covered by a single other subclass
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • 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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/40—Scenes; Scene-specific elements in video content
    • G—PHYSICS
    • G08—SIGNALLING
    • G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
    • G08B17/00—Fire alarms; Alarms responsive to explosion
    • G08B17/10—Actuation by presence of smoke or gases, e.g. automatic alarm devices for analysing flowing fluid materials by the use of optical means
    • G—PHYSICS
    • G08—SIGNALLING
    • G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
    • G08B17/00—Fire alarms; Alarms responsive to explosion
    • G08B17/12—Actuation by presence of radiation or particles, e.g. of infrared radiation or of ions
    • G08B17/125—Actuation by presence of radiation or particles, e.g. of infrared radiation or of ions by using a video camera to detect fire or smoke
    • G—PHYSICS
    • G08—SIGNALLING
    • G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
    • G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
    • G08B21/18—Status alarms
    • G08B21/182—Level alarms, e.g. alarms responsive to variables exceeding a threshold
    • G—PHYSICS
    • G08—SIGNALLING
    • G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
    • G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
    • G08B21/18—Status alarms
    • G08B21/20—Status alarms responsive to moisture
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J13/00—Circuit arrangements for providing remote indication of network conditions, e.g. an instantaneous record of the open or closed condition of each circuitbreaker in the network; Circuit arrangements for providing remote control of switching means in a power distribution network, e.g. switching in and out of current consumers by using a pulse code signal carried by the network
    • H02J13/00002—Circuit arrangements for providing remote indication of network conditions, e.g. an instantaneous record of the open or closed condition of each circuitbreaker in the network; Circuit arrangements for providing remote control of switching means in a power distribution network, e.g. switching in and out of current consumers by using a pulse code signal carried by the network characterised by monitoring
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J13/00—Circuit arrangements for providing remote indication of network conditions, e.g. an instantaneous record of the open or closed condition of each circuitbreaker in the network; Circuit arrangements for providing remote control of switching means in a power distribution network, e.g. switching in and out of current consumers by using a pulse code signal carried by the network
    • H02J13/00032—Systems characterised by the controlled or operated power network elements or equipment, the power network elements or equipment not otherwise provided for
    • H02J13/00036—Systems characterised by the controlled or operated power network elements or equipment, the power network elements or equipment not otherwise provided for the elements or equipment being or involving switches, relays or circuit breakers
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J13/00—Circuit arrangements for providing remote indication of network conditions, e.g. an instantaneous record of the open or closed condition of each circuitbreaker in the network; Circuit arrangements for providing remote control of switching means in a power distribution network, e.g. switching in and out of current consumers by using a pulse code signal carried by the network
    • H02J13/00032—Systems characterised by the controlled or operated power network elements or equipment, the power network elements or equipment not otherwise provided for
    • H02J13/00036—Systems characterised by the controlled or operated power network elements or equipment, the power network elements or equipment not otherwise provided for the elements or equipment being or involving switches, relays or circuit breakers
    • H02J13/0004—Systems characterised by the controlled or operated power network elements or equipment, the power network elements or equipment not otherwise provided for the elements or equipment being or involving switches, relays or circuit breakers involved in a protection system
    • 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)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Emergency Management (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Power Engineering (AREA)
  • Evolutionary Computation (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Artificial Intelligence (AREA)
  • Computing Systems (AREA)
  • Databases & Information Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Chemical & Material Sciences (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)
  • Remote Monitoring And Control Of Power-Distribution Networks (AREA)

Abstract

The invention relates to an edge calculation method and device for monitoring and prejudging a power distribution network, belonging to the field of safety management of power distribution rooms, wherein the method comprises the following steps: determining a trigger condition; monitoring an environment variable, and judging whether a triggering condition is met; if not, monitoring again; if yes, sending an alarm signal to obtain a first numerical value; collecting and marking abnormal environment monitoring data sets to obtain a training set, and training a neural network by adopting the training set; determining an abnormal environment key frame; inputting the abnormal environment key frame into a trained neural network to obtain the confidence coefficient of an abnormal target; judging whether the confidence coefficient is greater than a preset threshold value, if so, transmitting the signal into an edge computing device to obtain a second numerical value; the edge computing device obtains a decision according to the first numerical value and the second data. The scheme of the invention can intuitively know the abnormal condition of the on-site power distribution room, and avoid the hidden trouble caused by signal sending by mistake or no signal sending when the signal exceeds the threshold value.

Description

Edge calculation method and device for power distribution network monitoring and prejudging
Technical Field
The invention relates to the field of power distribution room safety management, in particular to an edge calculation method and device for power distribution network monitoring and prejudging.
Background
In a distribution network system of a modern power system, a distribution room serves as a power supply terminal. Widely appears in cities and villages. Most of the power distribution rooms are unattended, so that the environmental monitoring of the power distribution rooms is particularly important. At present, the power distribution room is only provided with a water immersion sensor aiming at the risk of water immersion. To address the risk of fire, the power distribution room is equipped with smoke sensors and sprays. The general camera is equipped with only at the gate to the outsider entering power distribution room. The host system collects the environmental information through the devices, and performs real-time acquisition, processing and reporting.
According to the scheme, only the alarm signal can be acquired in the power distribution room. Background personnel often can not directly understand the abnormal conditions of the on-site power distribution room. The environment monitoring function is only embodied at the level of the alarm signal. If the water rises too high or the fire expands, equipment is easy to damage, and serious results can be caused. And if the sensor fails, signals are sent by mistake or signals are not sent when the threshold value is exceeded, great hidden danger can be caused, and meanwhile, no small burden is caused on power distribution operation and maintenance work.
Disclosure of Invention
The invention aims to provide an edge calculation method and device for monitoring and prejudging a power distribution network, which can visually know the abnormal condition of a field power distribution room and avoid hidden dangers caused by mistaken signal sending or no signal sending when a threshold value is exceeded.
In order to achieve the purpose, the invention provides the following scheme:
an edge calculation method for monitoring and prejudging a power distribution network, the edge calculation method comprising:
determining a trigger condition;
monitoring an environment variable to obtain a monitoring result;
judging whether the monitoring result meets a triggering condition;
if not, returning to the step of 'starting monitoring the environmental variable and obtaining a monitoring result' to continue monitoring;
if yes, sending an alarm signal to obtain a first numerical value;
collecting abnormal environment monitoring data set X = { X = { (X) 1 ,x 2 ,…,x n In which x 1 ,x 2 ,…,x n Monitoring data for abnormal environments;
monitoring the abnormal environment for a data set X = { X = { (X) 1 ,x 2 ,…,x n Marking to obtain a label data set Y = { Y = } 1 ,y 2 ,…,y n In which y 1 ,y 2 ,…,y n Is label data;
monitoring a dataset X = { X) based on the abnormal environment 1 ,x 2 ,…,x n And the label data set Y = { Y = } 1 ,y 2 ,…,y n Determining a training set D = { X, Y };
training the neural network by adopting the training set to obtain a trained neural network;
acquiring a video image in a power distribution room;
filtering redundant frames in the video image to obtain abnormal environment key frames;
inputting the abnormal environment key frame into the trained neural network to obtain an abnormal target confidence coefficient;
judging whether the confidence coefficient is larger than a preset threshold value or not;
if the difference is less than or equal to the preset threshold value, returning to the step of filtering redundant frames in the video image to obtain abnormal environment key frames;
if the signal is greater than the first value, the signal is transmitted to an edge computing device to obtain a second numerical value, and the next step is executed;
the edge computing device makes a decision based on the first numerical value and the second numerical value.
Optionally, the environment variables include: water level, smoke, and door status.
Optionally, the trained neural network is Qint8, and the Qint8 includes:
the device comprises a first convolution unit, a second convolution unit, a backbone network, a pyramid network, a detection head and a coder-decoder; the first convolution unit, the second convolution unit, the backbone network, the pyramid network, the detection head and the codec are connected in sequence.
Optionally, the filtering out the redundant frames in the video image to obtain the abnormal environment key frame specifically includes the following steps:
acquiring four continuous frames of images
Figure 766168DEST_PATH_IMAGE001
；
The four frames of images
Figure 629213DEST_PATH_IMAGE001
Zooming and converting the image into a gray image;
performing Gaussian filtering on the gray level image to obtain a picture
Figure 292275DEST_PATH_IMAGE002
；
To pair
Figure 969332DEST_PATH_IMAGE003
And (3) performing difference to obtain a gray characteristic diagram:
Figure 172780DEST_PATH_IMAGE004
whereindif 1 Anddif 2 the gray characteristic images are obtained after subtraction of gray pictures;
to the abovedif 1 Anddif 2 automatic filling;
will be provided withdif 1 Anddif 2 adjusted to a one-dimensional vector, adif 1 Anddif 2 calculating cosine correlation coefficientcorr；
The camera acquires the next frame of picturep 4 ，
Figure 624752DEST_PATH_IMAGE005
；
Judging the cosine correlation coefficientcorrWhether it is greater than threshold value Thr corr If, ifcorr>Thr corr Then outputting the key framep key =p 4 Otherwise, returning to the step
Figure 288077DEST_PATH_IMAGE006
Scaled and converted to grayscale pictures ".
Optionally, for the abovedif 1 Anddif 2 the following formula is specifically adopted for automatic filling:
Figure 363349DEST_PATH_IMAGE007
wherein the content of the first and second substances,difis a binary image gray-scale value,dif i,j is the characteristic value of the coordinate point of the pixel of the gray-scale picture,Thr s is composed ofdif 1 Anddif 2 the threshold value of (2).
Optionally, the step of the edge computing device obtaining a decision according to the first numerical value and the second numerical value specifically includes:
determining decisions from formulasy：yFirst value + G2 second value of G1, wherein G1 and G2 are weights, and G1>G2；
When the decision is madeyWhen the current time is within the first set range, the edge computing device trips the incoming line switch to inform the main station;
when the decision is madeyAnd when the current time is within the second set range, triggering an alarm signal.
Based on the above method in the present invention, the present invention further provides an edge computing device for monitoring and pre-judging a distribution network, the device comprising:
the trigger condition determining module is used for determining trigger conditions;
the environment variable monitoring module is used for monitoring the environment variable to obtain a monitoring result;
the first judgment module is used for judging whether the monitoring result meets a trigger condition;
the first circulation module is used for returning to the environment variable monitoring module to continue monitoring when the environment variable monitoring module does not meet the requirement;
the first numerical value determining module is used for sending out an alarm signal when the first numerical value is met to obtain a first numerical value;
an abnormal data collection module for collecting an abnormal environment monitoring data set X = { X = { (X) } 1 ,x 2 ,…,x n In which x 1 ,x 2 ,…,x n Monitoring data for abnormal environments;
a data marking module for monitoring the abnormal environment for the data set X = { X = { (X) 1 ,x 2 ,…,x n Marking to obtain a label data set Y = { Y = } 1 ,y 2 ,…,y n In which y 1 ,y 2 ,…,y n Is label data;
a training set determination module to monitor a data set X = { X) based on the abnormal environment 1 ,x 2 ,…,x n And the label data set Y = { Y = } 1 ,y 2 ,…,y n Determining a training set D={X,Y}；
The training module is used for training the neural network by adopting the training set to obtain the trained neural network;
the image acquisition module is used for acquiring video images in the power distribution room;
the preprocessing module is used for filtering redundant frames in the video image to obtain abnormal environment key frames;
the confidence coefficient determining module is used for inputting the abnormal environment key frame into the trained neural network to obtain the confidence coefficient of the abnormal target;
the second judgment module is used for judging whether the confidence coefficient is larger than a preset threshold value or not;
the second circulation module is used for returning to the preprocessing module when the number of the second circulation module is smaller than or equal to the number of the first circulation module;
the second numerical value determining module is used for transmitting the signal to the edge calculating device to obtain a second numerical value when the signal is greater than the first numerical value, and executing the next step;
and the decision module is used for obtaining a decision according to the first numerical value and the second numerical value through an edge calculation device.
Optionally, the trained neural network is Qint8, and the Qint8 includes:
the device comprises a first convolution unit, a second convolution unit, a backbone network, a pyramid network, a detection head and a coder-decoder; the first convolution unit, the second convolution unit, the backbone network, the pyramid network, the detection head and the codec are connected in sequence.
Optionally, the preprocessing module specifically includes the following units:
a first image acquisition unit for acquiring four continuous frames of images
Figure 340795DEST_PATH_IMAGE008
；
A grayscale image conversion unit for converting the four-frame image
Figure 614650DEST_PATH_IMAGE008
Zooming and converting the image into a gray picture;
a Gaussian filtering unit for performing Gaussian filtering on the gray level image to obtain a picture
Figure 36492DEST_PATH_IMAGE002
；
A difference making unit for making a difference
Figure 569367DEST_PATH_IMAGE009
And (3) performing difference to obtain a gray characteristic diagram:
Figure 6296DEST_PATH_IMAGE004
whereindif 1 Anddif 2 the gray characteristic images are obtained after subtraction of gray pictures;
an automatic filling unit for aligning thedif 1 Anddif 2 automatic filling;
a cosine correlation coefficient calculation unit for calculating a cosine correlation coefficientdif 1 Anddif 2 adjusted to a one-dimensional vector, pairdif 1 Anddif 2 calculating cosine correlation coefficientcorr；
A second image acquisition unit for acquiring the next frame of picture by the camerap 4 ，
Figure 603499DEST_PATH_IMAGE005
；
A judging unit for judging the cosine correlation coefficientcorrWhether it is greater than threshold value Thr corr If, ifcorr>Thr corr Then outputting the key framep key =p 4 Otherwise, return to the "gradation map conversion unit".
Optionally, the automatic filling unit specifically adopts the following formula:
Figure 608627DEST_PATH_IMAGE007
wherein the content of the first and second substances,difis a binary image gray-scale value,dif i,j is the characteristic value of the coordinate point of the pixel of the gray-scale picture,Thr s is composed ofdif 1 Anddif 2 the threshold value of (2).
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
according to the method and the device, the edge computing device is added to realize the localization processing, and meanwhile, the method and the device have the function of local decision making, have low requirements on a network, do not need to increase the computing pressure of a background, and can realize the real-time processing on the environment;
by monitoring the water level, the smoke and the door state, namely adding an intelligent camera, the water immersion and the fire degree can be monitored at multiple angles, the intelligent camera is clear at a glance on an image, and the situation that only an alarm signal is known and the influence of the internal environment is unknown in the prior art is broken; a water sensor of a control loop is added, and a control signal is transmitted to an edge computing device, so that a line incoming switch can be quickly cut off, and the expansion of accidents is prevented; the edge calculating device can count the number of the smoke sensors, increase judgment basis, and quickly cut off the inlet wire switch according to the basis to eliminate hidden danger in one step in advance.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
FIG. 1 is a flow chart of an edge calculation method for monitoring and pre-judging a distribution network according to the present invention;
FIG. 2 is a schematic diagram of a yolo-tiny network structure according to the present invention;
FIG. 3 is a schematic diagram of the Qint8 network structure according to the present invention;
FIG. 4 is a flow chart of the environmental waterproof work of the present invention;
FIG. 5 is a simplified diagram of the appearance of a water sensor of the present invention;
FIG. 6 is a schematic diagram of an apparatus for monitoring and predicting environmental water immersion according to the present invention;
FIG. 7 is a flow chart of the work flow of the present invention for preventing fire in the environment;
FIG. 8 is a schematic diagram of an apparatus for monitoring and predicting ambient light according to the present invention;
FIG. 9 is a flowchart illustrating the environment intrusion prevention operation of the present invention;
FIG. 10 is a schematic diagram of an environmental intrusion monitoring and predicting method according to the present invention;
fig. 11 is a schematic structural diagram of an edge computing device for monitoring and predicting a distribution network according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention aims to provide an edge calculation method and device for monitoring and prejudging a power distribution network, which can visually know the abnormal condition of a field power distribution room and avoid hidden dangers caused by mistaken signal sending or no signal sending when a threshold value is exceeded.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
The invention monitors various environmental variables influencing normal operation in the power distribution room in real time through various sensors. And selecting corresponding sensors according to different environment variables. The output of the sensor is transmitted to the edge computing device by an RS-485 line. The edge computing device makes a decision on the incoming sensor output at the proposed algorithm. And after the decision is finished, transmitting the result to the master station. The intelligent, automatic and efficient power distribution room is further improved without background intervention, local acquisition, local monitoring and local decision.
Fig. 1 is a flowchart of an edge calculation method for monitoring and predicting a distribution network according to the present invention, as shown in fig. 1, the edge calculation method includes:
step 1: a trigger condition is determined.
Different degrees of decision making are required for the same environmental variable, where multiple sensors may be provided.
Step 2: and (5) monitoring the environment variable to obtain a monitoring result.
The output quantity of the sensor is saved in the environment monitoring APP of the edge computing device, and the working indicator lamp of the sensor works normally.
And step 3: judging whether the monitoring result meets a triggering condition;
if not, returning to the step 2 to continue monitoring;
if yes, step 8 is executed to send out an alarm signal, and a first numerical value is obtained, that is, when the trigger condition occurs, different trigger conditions obtain different values in the environment monitoring APP of the edge computing device, and the different values are recorded as the first numerical value.
And 4, step 4: collecting abnormal environment monitoring data set X = { X = { (X) 1 ,x 2 ,…,x n In which x 1 ,x 2 ,…,x n For abnormal environment monitoring data, for the abnormal environment monitoring data set X = { X = 1 ,x 2 ,…,x n Marking to obtain a label data set Y = { Y = } 1 ,y 2 ,…,y n In which y 1 ,y 2 ,…,y n Is label data;
monitoring a data set X = { X) based on the abnormal environment 1 ,x 2 ,…,x n And the label data set Y = { Y = } 1 ,y 2 ,…,y n Determining a training set D = { X, Y };
and uploading the data set D = { X, Y } and the yolo-tiny network Q to a cloud platform, training the data set, and quantizing the data set into a model Qint 8. Referring to fig. 2 and 3, fig. 2 is a schematic diagram of a yolo-tiny network structure according to the present invention; FIG. 3 is a schematic diagram of the network structure of Qint8 of the present invention, and the training process is as follows:
and processing the sample data set by sequentially passing through the backbone network, the characteristic pyramid network, the detection head and the coder-decoder, and performing non-maximum suppression processing on the decoder output to obtain a detection result. And after the label data is coded, loss function processing is carried out by combining the output of the detection head, and then the result is transmitted to the detection head through gradient back propagation. Wherein, the process of processing the sample data set by the backbone network comprises the following steps: the convolution unit (208, 208, 32), the convolution unit (104, 104, 64), the residual unit (52, 52, 128), the residual unit (26, 26, 256), the residual unit (13, 13, 512) and the convolution unit (13, 13, 512) are sequentially arranged. The outputs of the residual units (26, 26, 256) are fed through the stem layer 1 to the channel stitching (26 x 384) module of the feature pyramid network. The output of the convolution unit (13, 13, 512) is sent to a convolution (13 × 512) module of a characteristic pyramid network through the trunk layer 2, then sent to a channel splicing (26 × 384) module after being processed by the convolution (13 × 128) + upsampling (26 × 128) module, and sent to the detection head 1 after being spliced with the output data of the residual unit (26, 26, 256) sent from the trunk layer 1. The feature pyramid convolution (13 × 512) module outputs data to the detection head 2 for processing. The processing result of the detection head 1 is sent to a decoder for processing. The outputs of the detector heads 1 and 2 are fed to the loss function module for processing.
And 5: and acquiring a video image in the power distribution room, and filtering redundant frames in the video image to obtain an abnormal environment key frame.
And calling video acquisition equipment to acquire video images, and transmitting the video images into monitoring APP corresponding to the edge computing device through the gigabit network port. And written to the data center via the mqtt message bus. Filtering the redundant frames in the abnormal environment to obtain the key frames in the abnormal environmentp key The method comprises the following specific steps:
step 5.1: acquiring continuous four-frame images from camera
Figure 861754DEST_PATH_IMAGE001
。
Step 5.2: picture taking
Figure 729738DEST_PATH_IMAGE001
Scaling to proper size, converting into gray picture, Gaussian filtering, and rollingThe formula of the product is:
Figure 712607DEST_PATH_IMAGE010
wherein (A), (B), (C), (D), (C), (B), (C)x c ,y c ) The current center point coordinates of the convolution kernel,
Figure 91898DEST_PATH_IMAGE011
the variance corresponding to the convolution kernel.
Step 5.3: obtaining the picture after filtering
Figure 956954DEST_PATH_IMAGE012
To, for
Figure 79893DEST_PATH_IMAGE003
Making a difference:
Figure 386110DEST_PATH_IMAGE004
wherein the content of the first and second substances,dif 1 anddif 2 and obtaining a gray characteristic image after subtracting the gray image.
Step 5.4: to pairdif 1 Anddif 2 automatic filling, the filling mode is as follows:
Figure 733040DEST_PATH_IMAGE007
wherein the content of the first and second substances,Thr s is chromatic aberration, i.e.dif 1 Anddif 2 the threshold value of (a) is set,difis a binary image gray-scale value,dif i,j the characteristic value of the pixel coordinate point of the gray picture is obtained.
Step 5.5: will be provided withdif 1 Anddif 2 adjusted to a one-dimensional vector, pairdif 1 Anddif 2 calculating cosine correlation coefficientcorr:
Figure 429600DEST_PATH_IMAGE013
Wherein the content of the first and second substances,
Figure 412688DEST_PATH_IMAGE014
is composed of
Figure 370148DEST_PATH_IMAGE015
N is
Figure 294504DEST_PATH_IMAGE015
Total number of components.
Step 5.6: the camera acquires the next frame of picturep 4 ：
Figure 806257DEST_PATH_IMAGE005
。
Step 5.7: judging the cosine correlation coefficientcorrWhether it is greater than threshold value Thr corr If, ifcorr>Thr corr Then outputting the key framep key =p 4 Otherwise, return to step 5.2.
Step 6: and inputting the abnormal environment key frame into the trained neural network to obtain the confidence coefficient of the abnormal target.
I.e. the abnormal environment key frame p key The input detection network Qint8 detects the confidence C of the abnormal object through the yolo-tiny network.
And 7: and judging whether the confidence coefficient is larger than a preset threshold value.
If C>
Figure 536578DEST_PATH_IMAGE016
The signal is immediately transmitted to the edge computing device, a value is set in the corresponding monitoring APP, which is recorded as a second value, step 8 is executed, otherwise step 5 is executed.
And 8: for the same physical variable, there are two different types of detection equipment, sensors and intelligent cameras. Therefore, at the time of decision making, the weight occupied by the sensor is set to G1, and the weight occupied by the smart camera is set to G2. The decision is made according to the formula Y = G1 + G2 + second value.
Example 1
When the environment variable is the water level, for the identification and processing of water logging, the on-site power distribution room mainly comprises two water logging sensors, an intelligent camera and an edge computing device, and is specifically shown in fig. 6.
Referring to fig. 4, the working process of the environmental waterproof system of the present invention includes the following steps:
for a water sensor, fig. 5 is a simplified diagram of the water sensor.
Step 1: set height h 1 And h 2 ，h 2 Greater than h 1 . Placing the probe of the water sensor 1 at h 1 To (3). The probe of the water sensor 2 is placed at h 2 And (5) treating the rice. For the water sensor 1, it is set when the water level crosses h 1 And triggering an alarm signal of the edge computing device so as to remind background personnel to pay attention to the water level information. For the water sensor 2, it is set when the water level crosses h 2 And triggering a control signal of the edge computing device to trip off a line inlet switch of the power distribution room, and eliminating hidden danger in advance.
Step 2: the water sensor begins to monitor the water level, and for the edge calculation device, the output quantity of the water sensor is only one switching value, and the simplest yes or no judgment is carried out on the water level. Water sensor 1 and water sensor 2 are gone into edge computing device and are joined in marriage electrical room environmental monitoring APP through RS485 line. And written to the data center via the mqtt message bus. Under normal working conditions, when the water level does not cross the line, the POWER' lamps of the water sensor 1 and the water sensor 2 are kept normally on.
And step 3: when the water sensor 1 detects that the water level has crossed the water level warning line h 1 This signal is immediately passed to the edge computing device. Setting a value in a power distribution room environment monitoring APPa=20, while the 'ALARM' lamp of the water sensor 1 remains constantly on. When the water sensor 1 detects that the water level has crossed the water level warning line h 1 Step 9 is executed, otherwise step 2 is executed.
And 4, step 4: when the water sensor 2 detects that the water level has crossed the water level warning line h 2 Immediately transmitting the signal to the edge computing device to set a value in the power distribution room environment monitoring APPb=1, while the 'ALARM' lamp of the water sensor 2 remains constantly on. When the water sensor 2 detects that the water level has crossed the water level warning line h 2 If so, executing step 9, otherwise, executing step 2.
For camera
And 5: collecting water immersion data set X = { X = { (X) 1 ,x 2 ,…,x n And marking the label data set to obtain a corresponding label data set Y = { Y = } 1 ,y 2 ,…,y n Get data set D ═ X, Y). And uploading the data set D (X, Y) and the yolo-tiny network Q to a cloud platform, training and quantizing the data set D (X, Y) into a model Qint 8.
Step 6: and calling video acquisition equipment to acquire video images, and transmitting the video images into an edge computing device water immersion monitoring APP through a kilomega network port. And written to the data center via the mqtt message bus. Filtering out the water logging redundant frame to obtain a water logging key framep key 。
And 7: soaking the key frame in waterp key The input detection network Qint8 detects the confidence C of the abnormal object through the yolo-tiny network.
And 8: if C>
Figure 286228DEST_PATH_IMAGE016
Immediately transmitting the signal to the edge computing device, setting a value in the flooding monitoring APPdIf C is =1>
Figure 709381DEST_PATH_IMAGE016
Step 9 is executed, otherwise step 5 is executed.
Computing device for edge
And step 9: there are two different types of sensing devices for the same physical variable. The water sensor is used as a main decision-making device, and the intelligent camera is used as an auxiliary device. Therefore, the weight occupied by the water sensor must be greater than that of the intelligent camera in decision making. According to the formulaz=2*（a+b）+dAnd (4) decision making:
if it iszIf the current time is more than or equal to 4, the edge computing device trips the incoming line switch to inform the main station;
if 0<z<And 3, triggering an alarm signal, and reporting the picture to inform the master station by the edge computing device.
Example 2
When the environment variable is smoke, for the identification and treatment of fire, the field power distribution room is provided with a plurality of smoke sensors and sprays according to the floor area of the power distribution room. Meanwhile, the system also comprises an intelligent camera and an edge computing device, and is particularly shown in fig. 8.
Referring to fig. 7, the work flow of preventing fire in the environment of the present invention mainly includes the following steps:
for smoke sensor
Step 1: smoke of the on-site power distribution room is determined. The total number of devices is max and is stored in the power distribution room environment APP. When a part of sensors are triggered, the trigger edge computing control device sprays and releases the fire extinguishing dry powder. When all smoke sensors are triggered, the edge computing control device is triggered to trip off a line inlet switch of the power distribution room, and hidden dangers are eliminated in advance.
Step 2: the smoke sensor is started, and for the edge computing device, the output quantity of the smoke sensor is only one switching value, and the simplest yes or no judgment is carried out on the occurrence of smoke. Smoke sensor passes into edge computing device and joins in marriage electrical room environmental monitoring APP through RS485 line. And written to the data center via the mqtt message bus. Under the normal working state, the sensor indicator lamp is in a light-off state in the normal state, and flickers once every 30 seconds.
And step 3: when a smoke sensor detects that the smoke concentration reaches a trigger alarm value, the signal is immediately transmitted to an edge calculating device. Join in marriage smoke transducer quantity that electrical room environmental monitoring APP recorded an alarm, give m1 with the value of quantity, the smoke transducer's of the while warning sensor pilot lamp is often bright. And (4) when the smoke concentration is monitored to reach the trigger alarm value, executing the step 8, otherwise, executing the step 2.
For camera
And 4, step 4: collecting a fire data set X = { X = { X = } 1 ,x 2 ,…,x n And marking the label data set to obtain a corresponding label data set Y = { Y = } 1 ,y 2 ,…,y n Get data set D ═ X, Y. And uploading the data set D (X, Y) and the yolo-tiny network Q to a cloud platform, training and quantizing the data set D (X, Y) into a model Qint 8.
And 5: and calling video acquisition equipment to acquire video images, and transmitting the video images into an edge computing device (APP) for fire monitoring through a kilomega network port. And written to the data center via the mqtt message bus. Filtering the fire redundant frame to obtain the fire key framep key 。
Step 6: keyframe of firep key The input detection network Qint8 detects the confidence C of the abnormal object through the yolo-tiny network.
And 7: if C>
Figure 442851DEST_PATH_IMAGE016
Immediately, this signal is transmitted to the edge computing device, where a value m2=1 is set for the flare monitoring APP, if C>
Figure 830058DEST_PATH_IMAGE016
Step 9 is executed, otherwise step 5 is executed.
Computing device for edge
And 8: there are two different types of sensing devices for the same physical variable. The smoke sensor is used as main decision-making equipment, and the intelligent camera is used as auxiliary equipment. Therefore, the smoke sensor must be weighted more heavily than the smart camera in the decision making. According to the formulalDecision (= 2 × m1+ m 2):
if it islIf the value is more than or equal to 2 x max, the edge computing device trips the incoming line switch to inform the master station;
if 0<l<2 max, triggering alarm signal to spray and release the extinguishing dry powder, and reporting picture to inform the main station by the edge computing device.
Example 3
When the environment variable is in a door state, for the identification and processing of invasion, the field power distribution room is provided with corresponding door state sensors and warning devices according to the number of doors of the power distribution room. The system also comprises an intelligent camera and an edge computing device, and is particularly shown in figure 10.
Referring to fig. 9, the environment intrusion prevention workflow of the present invention includes the following steps:
for door state sensor
Step 1: the door state sensor starts to work, and for the edge computing device, the output quantity of the door state sensor is only one opening and closing quantity, and the simplest yes or no judgment is carried out on the opening and closing of the door. The door state sensor is transmitted into the edge computing device power distribution room environment monitoring APP through the RS485 line. And written to the data center via the mqtt message bus.
Step 2: when the door state sensor detects that the door is open, it immediately transmits this signal to the edge computing device. In the power distribution room environment monitoring APP, a value u1=1 is set. And (5) if the door is monitored to be opened, executing the step 7, otherwise, executing the step 2.
For camera
And step 3: collecting an internal face dataset X = { X = { X } 1 ,x 2 ,…,x n And marking the label data set to obtain a corresponding label data set Y = { Y = } 1 ,y 2 ,…,y n Get data set D ═ X, Y. And uploading the data set D (X, Y) and the yolo-tiny network Q to a cloud platform, training and quantizing the data set D (X, Y) into a model Qint 8.
And 4, step 4: and calling video acquisition equipment to acquire video images, and transmitting the video images into an edge computing device (APP) for fire monitoring through a kilomega network port. And written to the data center via the mqtt message bus. Filtering the fire redundant frame to obtain the fire key framep key 。
And 5: keyframe of firep key And inputting the abnormal target confidence coefficient C into a detection network Qint8, and detecting the abnormal target confidence coefficient C through a yolo-tiny network.
Step 6: if C<
Figure 637477DEST_PATH_IMAGE016
Immediately, this signal is transmitted to the edge computing device, where a value u2=1 is set in the intrusion monitoring APP whenC<
Figure 231532DEST_PATH_IMAGE016
Step 7 is executed, otherwise step 4 is executed.
Computing device for edge
And 7: there are two different types of detection devices for the same physical variable. The intelligent camera is used as main decision-making equipment, and the door state sensor is used as auxiliary equipment. Therefore, the smart camera must be weighted more heavily than the door status sensor at the time of decision making. The decision is made according to the formula v = u1+2 u 2:
if v is larger than or equal to 2, the edge computing device controls the alarm to give out warning sound, and reports pictures to inform the master station;
if 0< v <2, the edge computing device reports the picture to inform the master station.
Fig. 11 is a schematic structural diagram of an edge computing device for monitoring and predicting a distribution network according to the present invention, and as shown in fig. 11, the system includes:
a trigger condition determining module 101, configured to determine a trigger condition.
And the environment variable monitoring module 102 is configured to start monitoring the environment variable to obtain a monitoring result.
The first judging module 103 is configured to judge whether the monitoring result meets a trigger condition.
And a first loop module 104, configured to, when the environmental variable is not satisfied, return to the "environmental variable monitoring module" to continue monitoring.
And a first value determining module 105, configured to send an alarm signal when the first value is satisfied, so as to obtain a first value.
An abnormal data collection module 106, configured to collect an abnormal environment monitoring data set X = { X = { (X) } 1 ,x 2 ,…,x n }。
A data marking module 107, configured to monitor the abnormal environment for the data set X = { X = { (X) } 1 ,x 2 ,…,x n Marking to obtain a label data set Y = { Y = } 1 ,y 2 ,…,y n }。
A training set determination module 108 for determining a set of abnormal environmental monitoring data X = { X = 1 ,x 2 ,…,x n And the label data set Y = { Y = } 1 ,y 2 ,…,y n Determine training set D = { X, Y }.
And the training module 109 is used for training the neural network by adopting the training set to obtain the trained neural network.
And the image acquisition module 110 is used for acquiring video images in the power distribution room.
And the preprocessing module 111 is configured to filter redundant frames in the video image to obtain an abnormal environment key frame.
A confidence determining module 112, configured to input the abnormal environment key frame into the trained neural network, so as to obtain an abnormal target confidence.
The second judging module 113 is configured to judge whether the confidence is greater than a preset threshold.
A second loop module 114 for returning to the "preprocessing module" when less than or equal to.
And a second value determining module 115, configured to, if the second value is greater than the first value, transmit the signal to the edge calculating device to obtain a second value, and execute the next step.
A decision module 116, configured to obtain a decision according to the first numerical value and the second numerical value by an edge computing device.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.

Claims (10)

1. An edge calculation method for monitoring and prejudging a power distribution network is characterized by comprising the following steps:
determining a trigger condition;
monitoring an environment variable to obtain a monitoring result;
judging whether the monitoring result meets a triggering condition;
if not, returning to the step of 'starting monitoring the environmental variable and obtaining a monitoring result' to continue monitoring;
if yes, sending an alarm signal to obtain a first numerical value;
collecting abnormal environment monitoring data set X = { X = { (X) 1 ,x 2 ,…,x n In which x 1 ,x 2 ,…,x n Monitoring data for abnormal environments;
monitoring the abnormal environment for a data set X = { X = { (X) 1 ,x 2 ,…,x n Marking to obtain a label data set Y = { Y = } 1 ,y 2 ,…,y n In which y 1 ,y 2 ,…,y n Is label data;
monitoring a dataset X = { X) based on the abnormal environment 1 ,x 2 ,…,x n And the label data set Y = { Y = } 1 ,y 2 ,…,y n Determining a training set D = { X, Y };
training the neural network by adopting the training set to obtain a trained neural network;
acquiring a video image in a power distribution room;
filtering redundant frames in the video image to obtain abnormal environment key frames;
inputting the abnormal environment key frame into the trained neural network to obtain an abnormal target confidence coefficient;
judging whether the confidence coefficient is larger than a preset threshold value or not;
if the difference is less than or equal to the preset threshold value, returning to the step of filtering redundant frames in the video image to obtain abnormal environment key frames;
if the signal is greater than the first value, the signal is transmitted to an edge computing device to obtain a second numerical value, and the next step is executed;
the edge computing device makes a decision based on the first numerical value and the second numerical value.
2. The method of claim 1, wherein the environment variables comprise: water level, smoke, and door status.
3. The edge-computing method for power distribution network monitoring and prognostics according to claim 1, wherein the trained neural network is Qint8, the Qint8 comprising:
the device comprises a first convolution unit, a second convolution unit, a backbone network, a pyramid network, a detection head and a coder-decoder; the first convolution unit, the second convolution unit, the backbone network, the pyramid network, the detection head and the codec are connected in sequence.
4. The edge calculation method for power distribution network monitoring and prejudging as claimed in claim 1, wherein the filtering out the redundant frames in the video image to obtain the abnormal environment key frame specifically comprises the following steps:
acquiring four continuous frames of images
Figure 52227DEST_PATH_IMAGE001
；
The four frames of images are processed
Figure 297264DEST_PATH_IMAGE001
Zooming and converting the image into a gray image;
performing Gaussian filtering on the gray level image to obtain a picture
Figure 873739DEST_PATH_IMAGE002
；
To pair
Figure 128002DEST_PATH_IMAGE003
And (3) performing difference to obtain a gray characteristic diagram:
Figure 726997DEST_PATH_IMAGE004
whereindif 1 Anddif 2 the gray characteristic images are obtained after subtraction of gray pictures;
to the abovedif 1 Anddif 2 automatic filling;
will be provided withdif 1 Anddif 2 adjusted to a one-dimensional vector, pairdif 1 Anddif 2 calculating cosine correlation coefficientcorr；
The camera acquires the next frame of picturep 4 ，
Figure 611776DEST_PATH_IMAGE005
；
Judging the cosine correlation coefficientcorrWhether it is greater than threshold value Thr corr If, ifcorr>Thr corr Then output the key framep key =p 4 Otherwise, returning to the step
Figure 206706DEST_PATH_IMAGE006
Scaled and converted to grayscale pictures ".
5. The method of claim 4, wherein the edge calculation is performed on the power distribution networkdif 1 Anddif 2 the automatic filling specifically adopts the following formula:
Figure 467923DEST_PATH_IMAGE007
wherein the content of the first and second substances,difis a binary image gray-scale value,dif i,j is the characteristic value of the coordinate point of the pixel of the gray-scale picture,Thr s is composed ofdif 1 Anddif 2 the threshold value of (2).
6. The method of claim 4, wherein the step of the edge computing device making a decision based on the first and second values comprises:
determining decisions from formulasy：yFirst value + G2 second value of G1, wherein G1 and G2 are weights, and G1>G2；
When the decision is madeyWhen the current time is within the first set range, the edge computing device trips the incoming line switch to inform the main station;
when the decision is madeyAnd when the current time is within the second set range, triggering an alarm signal.
7. An edge computing device for monitoring and prognosticating a power distribution network, the device comprising:
the trigger condition determining module is used for determining trigger conditions;
the environment variable monitoring module is used for monitoring the environment variable to obtain a monitoring result;
the first judgment module is used for judging whether the monitoring result meets a trigger condition;
the first circulation module is used for returning to the environment variable monitoring module to continue monitoring when the environment variable monitoring module does not meet the requirement;
the first numerical value determining module is used for sending out an alarm signal when the first numerical value is met to obtain a first numerical value;
an abnormal data collection module for collecting an abnormal environment monitoring data set X = { X = { (X) } 1 ,x 2 ,…,x n In which x 1 ,x 2 ,…,x n Monitoring data for abnormal environments;
a data marking module for monitoring the abnormal environment for the data set X = { X = { (X) 1 ,x 2 ,…,x n Marking to obtain a label data set Y = { Y = 1 ,y 2 ,…,y n In which y 1 ,y 2 ,…,y n Is label data;
a training set determination module to monitor a data set X = { X) based on the abnormal environment 1 ,x 2 ,…,x n And the label data set Y = { Y = } 1 ,y 2 ,…,y n Determining a training set D = { X, Y };
the training module is used for training the neural network by adopting the training set to obtain the trained neural network;
the image acquisition module is used for acquiring video images in the power distribution room;
the preprocessing module is used for filtering redundant frames in the video image to obtain abnormal environment key frames;
the confidence coefficient determining module is used for inputting the abnormal environment key frame into the trained neural network to obtain the confidence coefficient of the abnormal target;
the second judgment module is used for judging whether the confidence coefficient is larger than a preset threshold value or not;
the second circulation module is used for returning to the preprocessing module when the number of the second circulation module is smaller than or equal to the number of the first circulation module;
the second numerical value determining module is used for transmitting the signal to the edge calculating device to obtain a second numerical value when the signal is greater than the first numerical value, and executing the next step;
and the decision module is used for obtaining a decision according to the first numerical value and the second numerical value through an edge calculation device.
8. The edge computing device for monitoring and prognostics of a power distribution network of claim 7, wherein the trained neural network is Qint8, the Qint8 comprising:
the device comprises a first convolution unit, a second convolution unit, a backbone network, a pyramid network, a detection head and a coder-decoder; the first convolution unit, the second convolution unit, the backbone network, the pyramid network, the detection head and the codec are connected in sequence.
9. The edge computing device for monitoring and pre-judging a power distribution network according to claim 7, wherein the preprocessing module specifically comprises the following units:
a first image acquisition unit for acquiring four continuous frames of images
Figure 497321DEST_PATH_IMAGE008
；
A grayscale image conversion unit for converting the four-frame image
Figure 349739DEST_PATH_IMAGE008
Zooming and converting the image into a gray image;
a Gaussian filtering unit for performing Gaussian filtering on the gray level image to obtain a picture
Figure 369648DEST_PATH_IMAGE009
；
A difference making unit for making
Figure 467179DEST_PATH_IMAGE010
And (3) performing difference to obtain a gray characteristic diagram:
Figure 521723DEST_PATH_IMAGE004
whereindif 1 Anddif 2 the gray characteristic images are obtained after subtraction of gray pictures;
an automatic filling unit for aligning thedif 1 Anddif 2 automatic filling;
a cosine correlation coefficient calculation unit for calculating a cosine correlation coefficientdif 1 Anddif 2 adjusted to a one-dimensional vector, pairdif 1 Anddif 2 calculating cosine correlation coefficientcorr；
A second image acquisition unit for acquiring the next frame of picture by the camerap 4 ，
Figure 810622DEST_PATH_IMAGE005
；
A judging unit for judging the cosine correlation coefficientcorrWhether it is greater than threshold value Thr corr If, ifcorr>Thr corr Then outputting the key framep key =p 4 Otherwise, return to the "gradation map conversion unit".
10. The apparatus according to claim 9, wherein the automatic filling unit specifically uses the following formula:
Figure 412767DEST_PATH_IMAGE007
wherein the content of the first and second substances,difis a binary image gray-scale value,dif i,j is the characteristic value of the coordinate point of the pixel of the gray-scale picture,Thr s is composed ofdif 1 Anddif 2 the threshold value of (2).
CN202210983658.3A 2022-08-17 2022-08-17 Edge calculation method and device for power distribution network monitoring and prejudging Pending CN115049988A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210983658.3A CN115049988A (en) 2022-08-17 2022-08-17 Edge calculation method and device for power distribution network monitoring and prejudging

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210983658.3A CN115049988A (en) 2022-08-17 2022-08-17 Edge calculation method and device for power distribution network monitoring and prejudging

Publications (1)

Publication Number Publication Date
CN115049988A true CN115049988A (en) 2022-09-13

Family

ID=83168216

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210983658.3A Pending CN115049988A (en) 2022-08-17 2022-08-17 Edge calculation method and device for power distribution network monitoring and prejudging

Country Status (1)

Country Link
CN (1) CN115049988A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116071901A (en) * 2023-03-27 2023-05-05 山东金宇信息科技集团有限公司 Multi-element monitoring and early warning method, equipment and medium for machine room environment

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN204791404U (en) * 2015-07-24 2015-11-18 成都安吉斯智能设备有限公司 Long -range thing of electric power networking management system
CN113435105A (en) * 2021-05-25 2021-09-24 杭州申弘智能科技有限公司 Fire early warning system and method based on smoke detection
CN113657207A (en) * 2021-07-29 2021-11-16 威胜信息技术股份有限公司 Intelligent monitoring method and system for fire and light of cloud-edge cooperative power distribution station
CN114320469A (en) * 2021-12-27 2022-04-12 中国矿业大学 Cloud-edge intelligence-based underground hazard source detection method
CN114677584A (en) * 2021-12-29 2022-06-28 广东电网有限责任公司中山供电局 Water immersion identification method and system for power distribution station of double-attention power system
US20220223020A1 (en) * 2020-03-31 2022-07-14 FS, Inc Electric fire prediction monitoring system and the method including the same

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN204791404U (en) * 2015-07-24 2015-11-18 成都安吉斯智能设备有限公司 Long -range thing of electric power networking management system
US20220223020A1 (en) * 2020-03-31 2022-07-14 FS, Inc Electric fire prediction monitoring system and the method including the same
CN113435105A (en) * 2021-05-25 2021-09-24 杭州申弘智能科技有限公司 Fire early warning system and method based on smoke detection
CN113657207A (en) * 2021-07-29 2021-11-16 威胜信息技术股份有限公司 Intelligent monitoring method and system for fire and light of cloud-edge cooperative power distribution station
CN114320469A (en) * 2021-12-27 2022-04-12 中国矿业大学 Cloud-edge intelligence-based underground hazard source detection method
CN114677584A (en) * 2021-12-29 2022-06-28 广东电网有限责任公司中山供电局 Water immersion identification method and system for power distribution station of double-attention power system

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
郭壮志等: "潜在等式约束的配电网遗传算法故障定位", 《现代电力》 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116071901A (en) * 2023-03-27 2023-05-05 山东金宇信息科技集团有限公司 Multi-element monitoring and early warning method, equipment and medium for machine room environment
CN116071901B (en) * 2023-03-27 2023-06-20 山东金宇信息科技集团有限公司 Multi-element monitoring and early warning method, equipment and medium for machine room environment

Similar Documents

Publication Publication Date Title
CN103839373B (en) A kind of unexpected abnormality event Intelligent Recognition alarm device and warning system
JP6867153B2 (en) Abnormality monitoring system
JP6942029B2 (en) Fire monitoring system
KR102195706B1 (en) Method and Apparatus for Detecting Intruder
CN107679471B (en) Indoor personnel air post detection method based on video monitoring platform
CN107911653A (en) The module of intelligent video monitoring in institute, system, method and storage medium
CN108389359B (en) Deep learning-based urban fire alarm method
CN112165600B (en) Drowning identification method and device, camera and computer system
JP2019079445A (en) Fire monitoring system
JP2018088105A (en) Monitoring system
CN113053053B (en) Interval demonstration multi-sensor fusion disaster cellular alarm linkage system based on particle swarm optimization
JP2018072881A (en) Fire disaster monitoring system
CN109867186B (en) Elevator trapping detection method and system based on intelligent video analysis technology
CN112489371A (en) Swimming pool drowning prevention early warning system based on computer vision
CN114724330A (en) Implementation method of self-adaptive mode switching multi-channel video fire real-time alarm system
KR102580434B1 (en) Dangerous situation detection device and dangerous situation detection method
CN115049988A (en) Edge calculation method and device for power distribution network monitoring and prejudging
CN114155595A (en) Behavior detection monitoring method, intelligent camera and intelligent monitoring system
CN113438443A (en) Emergency place monitoring and early warning system with image feature recognition function
CN109637066A (en) A kind of building intelligent the monitoring system of fire protection and method
KR102233679B1 (en) Apparatus and method for detecting invader and fire for energy storage system
CN111908288A (en) TensorFlow-based elevator safety system and method
CN116246445A (en) Knowledge-graph-based warehouse safety multi-source Internet-of-things data early warning method
Jakovčević et al. Notice of Violation of IEEE Publication Principles: Review of wildfire smoke detection techniques based on visible spectrum video analysis
CN115841730A (en) Video monitoring system and abnormal event detection method

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication

Application publication date: 20220913

RJ01 Rejection of invention patent application after publication