CN112187174A - Solar photovoltaic bracket abnormity detection method and system based on artificial intelligence - Google Patents
Solar photovoltaic bracket abnormity detection method and system based on artificial intelligence Download PDFInfo
- Publication number
- CN112187174A CN112187174A CN202011158844.0A CN202011158844A CN112187174A CN 112187174 A CN112187174 A CN 112187174A CN 202011158844 A CN202011158844 A CN 202011158844A CN 112187174 A CN112187174 A CN 112187174A
- Authority
- CN
- China
- Prior art keywords
- branch frame
- current
- photovoltaic
- support
- dimensional coordinate
- 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.)
- Withdrawn
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 39
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 29
- 230000007797 corrosion Effects 0.000 claims abstract description 79
- 238000005260 corrosion Methods 0.000 claims abstract description 79
- 230000002159 abnormal effect Effects 0.000 claims abstract description 35
- 238000000034 method Methods 0.000 claims abstract description 22
- 239000000463 material Substances 0.000 claims abstract description 12
- 238000005259 measurement Methods 0.000 claims abstract description 9
- 238000013210 evaluation model Methods 0.000 claims abstract description 7
- 230000005856 abnormality Effects 0.000 claims description 16
- 238000009434 installation Methods 0.000 claims description 16
- 238000012502 risk assessment Methods 0.000 claims description 14
- 238000011156 evaluation Methods 0.000 claims description 12
- 238000013178 mathematical model Methods 0.000 claims description 8
- 230000001419 dependent effect Effects 0.000 claims description 6
- 238000013528 artificial neural network Methods 0.000 description 11
- 230000008569 process Effects 0.000 description 8
- 230000011218 segmentation Effects 0.000 description 8
- 238000010586 diagram Methods 0.000 description 7
- 238000012549 training Methods 0.000 description 6
- 238000002372 labelling Methods 0.000 description 5
- 238000003062 neural network model Methods 0.000 description 4
- 238000013507 mapping Methods 0.000 description 3
- 241000239290 Araneae Species 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000003628 erosive effect Effects 0.000 description 1
- JEIPFZHSYJVQDO-UHFFFAOYSA-N iron(III) oxide Inorganic materials O=[Fe]O[Fe]=O JEIPFZHSYJVQDO-UHFFFAOYSA-N 0.000 description 1
- 229910052751 metal Inorganic materials 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 150000002739 metals Chemical class 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02S—GENERATION OF ELECTRIC POWER BY CONVERSION OF INFRARED RADIATION, VISIBLE LIGHT OR ULTRAVIOLET LIGHT, e.g. USING PHOTOVOLTAIC [PV] MODULES
- H02S50/00—Monitoring or testing of PV systems, e.g. load balancing or fault identification
-
- 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
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E10/00—Energy generation through renewable energy sources
- Y02E10/50—Photovoltaic [PV] energy
Landscapes
- Image Analysis (AREA)
Abstract
The invention relates to the technical field of artificial intelligence, in particular to a solar photovoltaic bracket abnormity detection method based on artificial intelligence. The method comprises the steps of obtaining a current photovoltaic support image, analyzing each photovoltaic support in the photovoltaic support image, obtaining the current photovoltaic support subsidence through the distance between key points of the photovoltaic supports and the distance between the key points of the photovoltaic supports when the photovoltaic supports are just installed, obtaining the current photovoltaic support inclination metric value through a vector formed by the key points of the photovoltaic supports and a vector formed by the key points of the photovoltaic supports when the photovoltaic supports are just installed, obtaining the photovoltaic support corrosion degree through the area ratio of a corrosion area, the detection time interval and the photovoltaic support material, and further judging the abnormal risk of the current photovoltaic supports through an abnormal risk evaluation model. The method integrates the subsidence degree, the inclination measurement value and the corrosion degree of the photovoltaic support to judge the abnormal risk of the photovoltaic support, and improves the accuracy of judging the abnormal risk of the photovoltaic support.
Description
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a solar photovoltaic bracket abnormity detection method and system based on artificial intelligence.
Background
The solar photovoltaic support is used for fixing and placing the position of the solar photovoltaic electric plate, and the condition of the photovoltaic support of the photovoltaic power station is closely related to the service life and safety of the power station. However, due to the influence of complex natural environment factors, the solar photovoltaic support generally has abnormal conditions such as support foundation settlement, corrosion, damage, inclination and the like along with the lapse of service time, and has a certain influence on the normal operation of the photovoltaic power station. If the foundation of the photovoltaic support is settled, the support is deformed and stress is generated on the solar panel, so that the solar panel is exploded; in addition, the photovoltaic bracket is corroded and damaged commonly, and for example, the root of the bracket, the joint part of the bracket and the solar panel, bolt holes and other parts which are in direct contact with different metals are prone to corrosion. These common photovoltaic support abnormal phenomena then influence the power station inclination lightly, lead to the power station generated energy to descend, and then heavily lead to the support whole to drop, the power station collapses, or counter weight fastness worsens, is blown down easily by wind.
In a comparison document of the application publication No. CN111597904A, which is used for a tunnel cable support inclination recognition method, an included angle between a straight line and a horizontal direction is calculated through an end point coordinate of the detected straight line, and whether the support is inclined or not is judged through the horizontal included angle.
In practice, the inventors found that the above prior art has the following disadvantages:
the factors causing the stent abnormality are considered too singly, and the obtained stent abnormality results have large errors.
Disclosure of Invention
In order to solve the technical problems, the invention aims to provide a solar photovoltaic bracket abnormity detection method and system based on artificial intelligence, and the adopted technical scheme is as follows:
in a first aspect, an embodiment of the present invention provides an artificial intelligence-based method for detecting an abnormality of a solar photovoltaic support, where the method includes: acquiring a current photovoltaic support image as a second image, wherein each photovoltaic support in the second image is a branch support; acquiring a current first three-dimensional coordinate of a connection point of the branch frame and the ground, a current second three-dimensional coordinate of a connection point of the branch frame and the photovoltaic cell panel, and a current distance h between the current first three-dimensional coordinate and the current second three-dimensional coordinate; according to the current distance h and the obtained corresponding initial distance h when the installation is finishedtAcquiring the sinkage degree of the branch frame; obtaining a currentThe inclination amount gamma of the branch frame and the inclination amount gamma which changes when the branch frame is just installed; estimating the corrosion degree upsilon of the branch frame according to a corrosion degree mathematical model which is constructed by taking the corrosion area ratio of the branch frame, the time interval between the current branch frame and the corresponding branch frame when the installation is finished and the corrosion coefficient of the material of the branch frame as independent variables and taking the corrosion degree upsilon of the branch frame as a dependent variable; judging the abnormal risk of the branch frame by combining a preset threshold according to a photovoltaic support abnormal risk evaluation model established by positive correlation of the sinkage degree epsilon of the branch frame, the gradient magnitude gamma of the branch frame, the corrosion degree upsilon of the branch frame and the photovoltaic support abnormal risk evaluation value sigma; the anomaly risk assessment value σ is:
where c is a positive number and is the bias term used to modify the function.
In a second aspect, another embodiment of the invention provides an artificial intelligence-based solar photovoltaic support anomaly detection system, which comprises an image acquisition module, a key point detection module, a support pose judgment module, a support corrosion degree judgment module and an anomaly evaluation module.
The image acquisition module is used for acquiring a current photovoltaic support image which is a second image, and each photovoltaic support in the second image is a branch support.
And the key point detection module is used for acquiring a current first three-dimensional coordinate of a connecting point of the branch frame and the ground, a current second three-dimensional coordinate of a connecting point of the branch frame and the photovoltaic cell panel and a current distance h between the current first three-dimensional coordinate and the current second three-dimensional coordinate.
The support pose judgment module comprises a support sinkage degree judgment unit and a support gradient judgment unit.
A support sinkage degree judging unit used for obtaining the corresponding initial distance h just after the installation according to the current distance htAnd obtaining the branch frame subsidence degree E.
And the support inclination judgment unit is used for acquiring the inclination value gamma which changes when the current branch support and the current branch support are just installed.
And the support corrosion degree judging module is used for estimating the corrosion degree upsilon of the branch frame according to a corrosion degree mathematical model which is constructed by taking the corrosion area occupation ratio of the branch frame, the time interval between the current branch frame and the corresponding branch frame when the installation is finished and the corrosion coefficient of the material of the branch frame as independent variables and taking the corrosion degree upsilon of the branch frame as a dependent variable.
And the anomaly evaluation module is used for judging the anomaly risk of the branch frame by combining a preset threshold according to a photovoltaic support anomaly risk evaluation model established by positive correlation among the sinkage degree epsilon of the branch frame, the gradient magnitude gamma of the branch frame, the corrosion degree upsilon of the branch frame and the photovoltaic support anomaly risk evaluation value sigma. The anomaly risk assessment value σ is:
where c is a positive number and is the bias term used to modify the function.
The invention has at least the following beneficial effects:
when the photovoltaic support is analyzed, the photovoltaic support is not analyzed from a single inclination angle, but analyzed from the sinkage degree, the inclination measurement value and the corrosion degree of the support in multiple aspects, and the sinkage degree, the inclination measurement value and the corrosion degree of the support are combined to analyze the abnormal risk of the photovoltaic support through the photovoltaic support risk assessment model, so that the accuracy of judging the abnormal risk of the photovoltaic support is improved, and convenience is brought to management personnel to process.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions and advantages of the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
Fig. 1 is a flowchart of an artificial intelligence-based solar photovoltaic bracket anomaly detection method according to an embodiment of the present invention;
fig. 2 is a flowchart of detecting a key point in a method for detecting an abnormality of a solar photovoltaic support based on artificial intelligence according to an embodiment of the present invention;
fig. 3 is a schematic diagram of a key point model of a photovoltaic support according to an embodiment of the present invention;
fig. 4 is a flowchart illustrating a method for detecting a sinking degree of a photovoltaic bracket according to an embodiment of the present invention;
fig. 5 is a flow chart of detecting an inclination value of a photovoltaic support in an artificial intelligence-based method for detecting an abnormality of a solar photovoltaic support according to an embodiment of the present invention;
fig. 6 is a flowchart illustrating the detection of the corrosion degree of the photovoltaic support in the artificial intelligence based method for detecting the abnormality of the solar photovoltaic support according to an embodiment of the present invention;
fig. 7 is a flowchart illustrating risk assessment of photovoltaic support abnormality in a method for detecting solar photovoltaic support abnormality based on artificial intelligence according to an embodiment of the present invention;
fig. 8 is a block diagram of an artificial intelligence based solar photovoltaic support anomaly detection system according to an embodiment of the present invention.
Detailed Description
In order to further illustrate the technical means and effects of the present invention adopted to achieve the predetermined objects, the following detailed description, with reference to the accompanying drawings and preferred embodiments, describes specific embodiments, structures, features and effects of an artificial intelligence based solar photovoltaic bracket anomaly detection method and system according to the present invention. In the following description, different "one embodiment" or "another embodiment" refers to not necessarily the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The following describes a specific scheme of the solar photovoltaic bracket abnormality detection method and system based on artificial intelligence in detail with reference to the accompanying drawings.
Referring to fig. 1, a flowchart of an artificial intelligence-based solar photovoltaic bracket anomaly detection method according to an embodiment of the present invention is shown.
Before the embodiment of the invention is implemented, a BIM model of a photovoltaic power station and an information exchange module of the BIM model need to be constructed, wherein the BIM model is a three-dimensional virtual model constructed on the basis of engineering information data of the photovoltaic power station by utilizing a digital technology.
And step S1, acquiring a current photovoltaic support image as a second image, wherein each photovoltaic support in the second image is a branch support. Adopt the photovoltaic power plant that loads the high definition camera to patrol and examine unmanned aerial vehicle and acquire photovoltaic cell's image data, the photovoltaic patrols and examines unmanned aerial vehicle's route and plans according to photovoltaic support's geographical distribution, and the collection of photovoltaic support image should adopt the mode of the fixed point shooting of subregion, the mode of shooting the image is local shooting after whole shooting earlier, the overall structure who acquires photovoltaic support is long-range shooting through the high altitude promptly, then the local image that acquires the support is closely shot from photovoltaic support's rear, utilize wireless transmission system to convey the image data who gathers to the server and handle. In practical implementation, the practitioner can adjust the height and angle of the shot according to the arrangement condition of the photovoltaic supports. Schematic diagram of a photovoltaic panel as shown in fig. 3, the photovoltaic support comprises a first sub-support 10, a second sub-support 20, a photovoltaic panel 30, a ground 40, and a current two-dimensional coordinate K of a connection point of the first sub-support 10 and the ground 401Current two-dimensional coordinate K of the point of connection of the first sub-mount 10 and the photovoltaic panel 302Current two-dimensional coordinate K of the point of connection of the second sub-mount 20 and the photovoltaic panel 303And the current two-dimensional coordinates K of the connection point of the second sub frame 20 and the ground 404Each photovoltaic cell sub-mount comprises two sub-mounts, a first sub-mount 10 and a second sub-mount 20. In the embodiment of the invention, forThe abnormal risk assessment process of a sub-stent 10 is explained.
Step S2, obtaining a current first three-dimensional coordinate of a connection point of the branch frame and the ground, a current second three-dimensional coordinate of a connection point of the branch frame and the photovoltaic cell panel, and a current distance h between the current first three-dimensional coordinate and the current second three-dimensional coordinate.
Specifically, as shown in fig. 2, the trained critical point detection DNN deep neural network is adopted to detect the critical points of the sub-supports of the photovoltaic cell, wherein the critical points comprise the current two-dimensional coordinates K of the connection point of the first sub-support and the ground1Current two-dimensional coordinate K of connection point of first sub-bracket and photovoltaic cell panel2Current two-dimensional coordinate K of connection point of second sub-support and photovoltaic cell panel3And the current two-dimensional coordinate K of the connection point of the first sub-frame and the ground4. In the embodiment of the invention, the HRNet is used as a key point detection model of the photovoltaic support, and the acquired image of the photovoltaic support is input into a key point detection DNN deep neural network, so that the K of the photovoltaic cell is obtained1(x1，y1)、K2(x2，y2)、K3(x3，y3) And K4(x4，y4) Four two-dimensional keypoint coordinates, where K1(x1，y1) And K2(x2，y2) Current two-dimensional coordinates, K, representing key points of the first sub-stent3(x3，y3) And K4(x4，y4) Representing the current two-dimensional coordinates of the keypoints of the second sub-stent.
The specific process of DNN deep neural network model training for photovoltaic scaffold key point detection comprises the following steps:
marking the connection point of the photovoltaic support and the ground and the connection point of the photovoltaic support and the photovoltaic cell panel as key points as real values (ground-truth) of the image by adopting marking tools such as labelme and the like to mark the key points, wherein the obtained key points are of 4 types and are respectively marked as key points K1(x1，y1)、K2(x2，y2)、K3(x3，y3) And K4(x4，y4) And convolving the labeled panel key point scatter diagram with a Gaussian kernel to obtain key point thermodynamic diagrams (heatmaps). Inputting the photovoltaic support image and corresponding note data generated by a labeling tool into a DNN model for end-to-end training, wherein the DNN model for key point detection is an Encoder-Decoder (Encoder-Decoder) structure, and a backbone network of the Encoder is a residual error network (ResNet) and the like for extracting Feature-maps (Feature-maps) of the photovoltaic support image, so that the Encoder outputs the Feature-maps of the photovoltaic support image, and the Decoder up-samples the Feature-maps and updates parameters according to a loss function to finally generate the key point-maps. Key point detection DNN deep neural network Loss function sampled in the training process is heat map Loss function (Heatmaps Loss), and the Loss function is as follows:
wherein, PcijThe score of the photovoltaic support key point in the category C at the image position (i, j) is represented, and the higher the score is, the more likely the score is to be the key point, ycijThe true values (ground-truth) representing the labels are heatmaps generated using a gaussian kernel convolution. N denotes the number of keypoints labeled for the photovoltaic stent, H and W denote the height and width of the image, respectively.
Considering that the two-dimensional coordinates of the photovoltaic support cannot effectively represent the change of the support in space, the TCN network model is adopted to map the two-dimensional key point coordinates of the photovoltaic support to the three-dimensional space to obtain the current three-dimensional coordinates K 'of the key point'1(x1，y1，z1)、K′2(x2，y2，z2)、K′3(x3，y3，z3) And K'4(x4，y4，z4) Wherein, K'1(x1，y1，z1) And K'2(x2，y2，z2) Is the current three-dimensional coordinate, K ', of the first partial stent'3(x3，y3，z3) And K'4(x4，y4，z4) Is the current three-dimensional coordinates of the second sub-stent. From a current first three-dimensional coordinate K 'of a first sub-stent'1(x1，y1，z1) And a current second three-dimensional coordinate K'2(x2，y2，z2) And calculating the current distance h of the first branch bracket as follows:
step S3, according to the current distance h and the obtained corresponding initial distance h when the installation is finishedtAnd obtaining the branch frame subsidence degree E.
Specifically, as shown in fig. 4, before implementation, an initial first three-dimensional coordinate corresponding to a connection point of the current first sub-mount with the ground when installation of the photovoltaic panel is completed is first recorded
And an initial second three-dimensional coordinate of the connection point of the first sub-mount with the photovoltaic panel
And calculating to obtain an initial first three-dimensional coordinate of a connecting point of the first sub-bracket and the ground
Initial second three-dimensional coordinate of connecting point of first branch support of sun and photovoltaic cell panel
Initial distance h oftInitial first three-dimensional coordinates
To the initial second three-dimensional coordinate
Initial vector of
And will initiate a first three-dimensional coordinate
Initial second three-dimensional coordinate
Initial distance htSum vector value
Stored in the server.
Then according to the obtained current first three-dimensional coordinate K'1(x1，y1，z1) And a current second three-dimensional coordinate K'2(x2，y2，z2) The current distance h between and the recorded initial first three-dimensional coordinates
And initial second three-dimensional coordinates
Initial distance h oftThe ratio of the distance h to the branch frame sinkage belongs to a negative correlation relationship, and the mapping model of the actual distance h and the branch frame sinkage belongs to is constructed as follows:
and (4) judging the sinking degree epsilon of the first sub-bracket by the formula (1).
Step S4, obtaining the inclination amount γ changed when the current spider and the spider just mounted are completed.
Specifically, as in FIG. 5, from the currently acquired current first three-dimensional coordinate K 'of the first sub-stent'1And a current second three-dimensional coordinate K'2Obtaining the current second three-dimensional coordinate K'2To the current first three-dimensional coordinate K'1Current vector of
Then from the previously stored initial second three-dimensional coordinate to the initial first three-dimensional coordinate
Initial vector of
And (3) calculating a dimensionless measurement value gamma when the first sub-bracket tilts by using a vector formula:
by the formula (2), a dimensionless measure gamma of the first sub-stent during the inclination is obtained.
And step S5, estimating the corrosion degree upsilon of the branch frame according to a corrosion degree mathematical model which is constructed by taking the corrosion area occupation ratio of the branch frame, the time interval between the current branch frame and the corresponding branch frame when the installation is finished and the corrosion coefficient of the material of the branch frame as independent variables and the corrosion degree upsilon of the branch frame as a dependent variable.
Specifically, as shown in fig. 6, in the embodiment of the present invention, the photovoltaic stent image is semantically segmented by the semantic segmentation deep neural network to obtain a normal photovoltaic stent region and a region where corrosion occurs of the first sub-stent.
For the image of the first sub-support, labeling at pixel level is carried out by using labeling tools such as labelme and the like as an image group-route, and the specific labeling method comprises the following steps: and marking the pixels of the image background as 0, the pixels of the normal photovoltaic support image area as 1, and the pixels of the rusted photovoltaic support image area as 2. And then, inputting the marked first sub-stent image into the trained semantic segmentation deep neural network, and inputting three types of regions, namely a background region, a normal photovoltaic stent region and a rust region of the image.
The semantic segmentation deep neural network training process comprises the following steps:
the method comprises the steps of training a semantic segmentation depth neural network model end to end for a pre-collected photovoltaic support image and a labeled data file generated by a labeling tool, wherein the semantic segmentation depth neural network model is of an Encoder-Decoder structure, the Encoder part can select convolution neural network models such as ResNet and VGG, the Decoder performs up-sampling on a feature image extracted by the Encoder, the class projection of pixels is provided with an original pixel space according to image features, and high-level semantic features of the photovoltaic support image with low resolution are output. The loss function in the training process samples the cross entropy loss function.
Obtaining the number n of normal photovoltaic support pixels of the first sub-support from the currently acquired first sub-support image through a semantic segmentation depth neural networknAnd the number n of pixels of the photovoltaic bracket rusteduThat is, the normal photovoltaic support area n is obtainednAnd the area n of the photovoltaic support where corrosion occursuConsidering that the semantic segmentation depth neural network can only acquire the area of the corrosion area on the surface of the photovoltaic support and cannot acquire the depth of the corrosion area of the photovoltaic support, calculating the corrosion depth of the first sub-support according to the time interval delta t of the time recorded when the current first sub-support is detected and the first sub-support is installed and the corrosion coefficient of the material of the photovoltaic support, and judging the corrosion degree upsilon of the photovoltaic support according to the corrosion area and the corrosion depth:
in the formula (3), the larger upsilon represents that the corrosion degree of the first sub-bracket is higher, p is a corrosion coefficient of the photovoltaic bracket related to the characteristics of the material of the first sub-bracket, is a constant and can be obtained through experimental fitting, in the embodiment of the invention, p is 3.25, and Δ t is a time interval for recording the time after the installation of the first sub-bracket and the time for currently detecting the corrosion of the first sub-bracket.
And step S6, judging the abnormal risk of the branch frame by combining a preset threshold according to a photovoltaic support abnormal risk evaluation model established by positive correlation among the sinkage degree E of the branch frame, the gradient magnitude gamma of the branch frame, the corrosion degree upsilon of the branch frame and the photovoltaic support abnormal risk evaluation value sigma.
Specifically, as shown in fig. 7, in combination with the formula (1), the formula (2) and the formula (3), the subsidence e of the stent, the measure γ of the inclination of the stent, the erosion υ of the stent and the risk assessment value σ of the photovoltaic stent form a positive correlation, and the photovoltaic stent risk assessment model is constructed as follows:
in the formula (4), c is a positive number, is an offset term used for the correction function, and has a value range of (1, 3), and in the embodiment of the present invention, c is 2.7. The larger the value of sigma is, the larger the potential safety hazard of the first sub-support is, and when the value of sigma is larger than 0.4, the abnormal risk of the first sub-support is judged to be too large, and a worker needs to be informed to maintain the photovoltaic support.
And then, projecting the risk assessment value obtained by using the judgment result of the subsidence degree, the inclination degree and the corrosion degree of the first sub-bracket to a constructed BIM (building information modeling) model of the photovoltaic power station by using an information exchange module so that maintenance personnel of the photovoltaic power station can check abnormal conditions of the first sub-bracket, such as subsidence, inclination, corrosion and the like in real time.
In actual operation, the abnormal risk of the first sub-bracket is not only evaluated, but also the abnormal risk of the first sub-bracket and the second sub-bracket of each photovoltaic cell is evaluated,
in conclusion, the three-dimensional coordinate value of the key point is used for judging the sinkage degree and the inclination metric value of the current photovoltaic support relative to the photovoltaic support just installed, the corrosion area, the detection time interval and the support material are used for estimating the corrosion depth, the corrosion degree of the photovoltaic support is estimated, the sinkage degree, the inclination metric value and the corrosion degree of the photovoltaic support are combined by the photovoltaic support risk assessment model to evaluate the abnormal risk of the photovoltaic support, when the abnormal risk assessment value of the photovoltaic cell is larger than the threshold value, the photovoltaic support is judged to have the abnormal risk, maintenance personnel can be informed in time, and due to the existence of the BIM model, the geographical position of the photovoltaic support with the abnormal risk can be better known. The accuracy rate of evaluating the abnormal risk of the photovoltaic bracket is improved.
Based on the same inventive concept as the system/method, another embodiment of the invention also provides an artificial intelligence-based solar photovoltaic bracket abnormality detection system.
Referring to fig. 8, a block diagram of an artificial intelligence based solar photovoltaic bracket anomaly detection system according to another embodiment of the present invention is shown.
The solar photovoltaic support abnormity detection system based on artificial intelligence comprises an image acquisition module 100, a key point detection module 200, a support pose judgment module 300, a support corrosion degree judgment module 400 and an abnormity evaluation module 500.
The image acquisition module is used for acquiring a current photovoltaic support image which is a second image, and each photovoltaic support in the second image is a branch support.
Specifically, adopt the photovoltaic power plant that loads the high definition camera to patrol and examine unmanned aerial vehicle and acquire photovoltaic cell's image data, through patrolling and examining the overall structure that unmanned aerial vehicle high altitude remote shooting acquireed the photovoltaic support, then the local image that acquires the support is shot closely from the rear of photovoltaic support, utilizes wireless transmission system to convey the image data who gathers to the server and handles. Schematic diagram of a photovoltaic panel as shown in fig. 3, the photovoltaic support comprises a first sub-support 10, a second sub-support 20, a photovoltaic panel 30, a ground 40, and a current two-dimensional coordinate K of a connection point of the first sub-support 10 and the ground 401Current two-dimensional coordinate K of the point of connection of the first sub-mount 10 and the photovoltaic panel 302Current two-dimensional coordinate K of the point of connection of the second sub-mount 20 and the photovoltaic panel 303And the current two-dimensional coordinates K of the connection point of the first sub-frame 10 and the floor 404Each photovoltaic cell sub-mount comprises two sub-mounts, a first sub-mount 10 and a second sub-mount 20. In the embodiment of the present invention, the process of evaluating the abnormal risk of the first sub-rack 10 will be explained.
The key point detection module is used for acquiring a current first three-dimensional coordinate of a connecting point of the branch frame and the ground, a current second three-dimensional coordinate of a connecting point of the branch frame and the photovoltaic cell panel and a current distance h between the current first three-dimensional coordinate and the current second three-dimensional coordinate.
Specifically, the trained key point detection DNN deep neural network is adopted to detect key points of the branch frame of the photovoltaic cell, in the embodiment of the invention, the HRNet is used as a key point detection model of the photovoltaic support, and the acquired image of the photovoltaic support is input into the key point detection DNN deep neural network, so that the K of the photovoltaic cell is obtained1(x1，y1)、K2(x2，y2)、K3(x3，y3) And K4(x4，y4) Four two-dimensional keypoint coordinates, where K1(x1，y1) And K2(x2，y2) Current two-dimensional coordinates, K, representing key points of the first sub-stent3(x3，y3) And K4(x4，y4) Representing the current two-dimensional coordinates of the keypoints of the second sub-stent.
Mapping the two-dimensional key point coordinates of the photovoltaic support to a three-dimensional space by adopting a TCN network model to obtain the current three-dimensional coordinates K 'of the key points'1(x1，y1，z1)、K′2(x2，y2，z2)、K′3(x3，y3，z3) And K'4(x4，y4，z4) Wherein, K'1(x1，y1，z1) And K'2(x2，y2，z2) Is the current three-dimensional coordinate, K ', of the first partial stent'3(x3，y3，z3) And K'4(x4，y4，z4) Is the current three-dimensional coordinate of the second partial stent, according to the current first three-dimensional coordinate K 'of the first partial stent'1(x1，y1，z1) And a current second three-dimensional coordinate K'2(x2，y2，z2) Obtaining the first partial cradleThe current distance h is:
the distance h between the two connection points of the current first sub bracket is obtained.
The stent pose determination means includes a stent subsidence determination unit 310 and a stent inclination determination unit 320.
A support sinkage degree judging unit used for obtaining the corresponding initial distance h just after the installation according to the current distance htAnd obtaining the branch frame subsidence degree E.
Specifically, initial first three-dimensional coordinates corresponding to connection points of the current first sub-support and the ground when the photovoltaic cell panel is installed are stored
And an initial second three-dimensional coordinate of the connection point of the first sub-mount with the photovoltaic panel
And calculating to obtain an initial first three-dimensional coordinate of a connecting point of the first sub-bracket and the ground
And an initial second three-dimensional coordinate of the connection point of the first sub-mount with the photovoltaic panel
Initial distance h oftInitial first three-dimensional coordinates
To the initial second three-dimensional coordinate
Initial vector of
Then according to the obtained current first three-dimensional coordinate K'1(x1，y1，z1) And a current second three-dimensional coordinate K'2(x2，y2，z2) The current distance h between and the recorded initial first three-dimensional coordinates
And initial second three-dimensional coordinates
Initial distance h oftThe ratio of the distance h to the branch frame sinkage belongs to a negative correlation relationship, and the mapping model of the actual distance h and the branch frame sinkage belongs to is constructed as follows:
the degree of subsidence ∈ of the first sub-stent can be determined by the formula (1).
And the support inclination judgment unit is used for acquiring the inclination value gamma which changes when the current branch support and the current branch support are just installed.
Specifically, the current three-dimensional coordinate K 'of the currently acquired first sub-stent is determined'1And K'2Obtaining the current second three-dimensional coordinate K'2To the current first three-dimensional coordinate K'1Current vector of
Then the initial second three-dimensional coordinates stored previously are used
To the initial first three-dimensional coordinates
Initial vector of
Calculation using vector formulaAnd (3) obtaining a dimensionless measurement value gamma when the first sub-bracket tilts:
and (4) obtaining a dimensionless measurement value gamma when the photovoltaic bracket inclines by the formula (2).
And the support corrosion degree judging module is used for estimating the corrosion degree upsilon of the branch frame according to a corrosion degree mathematical model which is constructed by taking the corrosion area occupation ratio of the branch frame, the time interval between the current branch frame and the corresponding branch frame when the installation is finished and the corrosion coefficient of the material of the branch frame as independent variables and taking the corrosion degree upsilon of the branch frame as a dependent variable.
Specifically, the area of the corrosion region and the area of the normal region of the first sub-support are obtained through a semantic segmentation depth neural network, then the time interval delta t between the current detection time of the first sub-support and the time when the corresponding first sub-support is just installed is obtained, and the corrosion coefficient p of the material of the corresponding first sub-support is used for calculating the corrosion depth of the first sub-support, so that the corrosion degree of the first sub-support is judged to be:
the higher the upsilon is, the higher the corrosion degree of the first sub-bracket is, p is a corrosion coefficient of the photovoltaic bracket related to the characteristics of the material of the first sub-bracket, and is a constant and can be obtained through experimental fitting, and in the embodiment of the invention, p is 3.25.
And the abnormal evaluation module is used for judging the abnormal risk of the branch frame by combining a preset threshold according to a photovoltaic support abnormal risk evaluation model established by positive correlation among the sinkage degree epsilon of the branch frame, the gradient magnitude gamma of the branch frame, the corrosion degree upsilon of the branch frame and the photovoltaic support abnormal risk evaluation value sigma.
Specifically, the combination of the formula (1), the formula (2) and the formula (3) is formed by the sinking degree epsilon of the branch frame, the measurement gamma of the inclination of the branch frame, the corrosion degree upsilon of the branch frame and the photovoltaic supportThe risk assessment value sigma is in positive correlation, and the photovoltaic support risk assessment model is constructed as follows:
in the formula (4), c is a positive number, is an offset term used for the correction function, and has a value range of (1, 3), and in the embodiment of the present invention, c is 2.7. The larger the value of sigma is, the larger the potential safety hazard of the first sub-support is, and when the value of sigma is larger than 0.4, the first sub-support is judged to have abnormal risk, and a worker needs to be informed to maintain the photovoltaic support.
It should be noted that: the precedence order of the above embodiments of the present invention is only for description, and does not represent the merits of the embodiments. And specific embodiments thereof have been described above. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments.
The present invention is not limited to the above preferred embodiments, and any modifications, equivalent replacements, improvements, etc. within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (8)
1. The solar photovoltaic bracket abnormity detection method based on artificial intelligence is characterized by comprising the following steps:
acquiring a current photovoltaic support image as a second image, wherein each photovoltaic support in the second image is a branch support;
acquiring a current first three-dimensional coordinate of a connection point of the branch frame and the ground, a current second three-dimensional coordinate of a connection point of the branch frame and the photovoltaic cell panel, and a current distance between the current first three-dimensional coordinate and the current second three-dimensional coordinate;
acquiring the sinkage epsilon of the branch frame according to the current distance and the acquired initial distance of the corresponding branch frame just installed;
acquiring the inclination degree value gamma of the current branch frame which changes when the installation of the branch frame is finished;
estimating the corrosion degree upsilon of the branch frame according to a corrosion degree mathematical model which is constructed by taking the corrosion area ratio of the branch frame, the current time interval between the branch frame and the corresponding branch frame when the installation is finished and the corrosion coefficient of the material of the branch frame as independent variables and the corrosion degree upsilon of the branch frame as a dependent variable;
judging the abnormal risk of the branch frame by combining a preset threshold according to a photovoltaic support abnormal risk evaluation model established by positive correlation among the subsidence degree epsilon of the branch frame, the gradient magnitude gamma of the branch frame, the corrosion degree upsilon of the branch frame and the photovoltaic support abnormal risk evaluation value sigma; wherein the abnormal risk assessment value σ is:
where c is a positive number and is the bias term used to modify the function.
2. The method for detecting the abnormality of the solar photovoltaic bracket based on the artificial intelligence as claimed in claim 1, wherein the branch bracket subsidence e is as follows:
3. the method for detecting the abnormality of the artificial intelligence-based solar photovoltaic support according to claim 1, wherein the step of obtaining the inclination magnitude γ of the corresponding sub-support immediately after the sub-support is installed comprises:
acquiring an initial first three-dimensional coordinate corresponding to a connection point with the ground when the sub-frame is just installed and an initial second three-dimensional coordinate corresponding to a connection point with the photovoltaic cell panel;
forming an initial vector according to the initial first three-dimensional coordinate and the initial second three-dimensional coordinate
The current first three-dimensional coordinates and the current second three-dimensional coordinates of the branch frame form a current vector
Obtaining a measurement value gamma when the branch frame tilts by a vector formula:
4. the artificial intelligence-based solar photovoltaic bracket abnormality detection method according to claim 1, wherein the degree of corrosion mathematical model is:
wherein n isuRepresents the area of the corrosion area of the branch frame, nnRepresents the area of the normal region of the sub-mount,
showing the rusted area ratio of the branch frame.
5. The solar photovoltaic bracket abnormity detection system based on artificial intelligence is characterized by comprising an image acquisition module, a key point detection module, a bracket pose judgment module, a bracket corrosion degree judgment module and an abnormity evaluation module;
the image acquisition module is used for acquiring a current photovoltaic support image which is a second image, and each photovoltaic support in the second image is a branch support;
the key point detection module is used for acquiring a current first three-dimensional coordinate of a connecting point of the branch frame and the ground, a current second three-dimensional coordinate of a connecting point of the branch frame and the photovoltaic cell panel and a current distance h between the current first three-dimensional coordinate and the current second three-dimensional coordinate;
the bracket pose judgment module comprises a bracket subsidence judgment unit and a bracket inclination judgment unit;
the support sinkage degree judging unit is used for judging the initial distance h when the installation is finished according to the current distance h and the obtained corresponding initial distance h when the installation is finishedtAcquiring the sinkage degree epsilon of the branch frame;
the bracket inclination judging unit is used for acquiring the inclination value gamma of the current branch bracket which changes when the installation of the branch bracket is finished;
the bracket corrosion degree judging module is used for estimating the corrosion degree upsilon of the branch frame according to a corrosion area ratio of the branch frame, a time interval between the current branch frame and the corresponding branch frame when the installation is finished and a corrosion coefficient of the material of the branch frame which are used as independent variables and a corrosion degree mathematical model which is constructed by using the corrosion degree upsilon of the branch frame as a dependent variable;
the anomaly evaluation module is used for judging the anomaly risk of the branch frame by combining a preset threshold according to a photovoltaic support anomaly risk evaluation model established by positive correlation of the sinkage degree epsilon of the branch frame, the gradient magnitude gamma of the branch frame and the corrosion degree upsilon of the branch frame and the photovoltaic support anomaly risk evaluation value sigma; wherein the abnormal risk assessment value σ is
Where c is a positive number and is the bias term used to modify the function.
6. The artificial intelligence based solar photovoltaic bracket abnormality detection system according to claim 5, wherein the branch bracket subsidence e is:
7. the artificial intelligence based solar photovoltaic support abnormality detection system according to claim 5, wherein the support inclination determination module is further configured to:
obtaining the current first three-dimensional coordinate and the current second three-dimensional coordinate of the branch frame to form a current vector
Storing an initial first three-dimensional coordinate corresponding to a connection point with the ground when the sub-frame is just installed and an initial second three-dimensional coordinate corresponding to a connection point with the photovoltaic cell panel, wherein a vector formed by the initial first three-dimensional coordinate and the initial second three-dimensional coordinate
Obtaining a measurement value gamma when the branch frame tilts by a vector formula:
8. the artificial intelligence based solar photovoltaic support abnormality detection system according to claim 5, wherein: the mathematical model of the corrosion is as follows:
wherein n isuRepresents the area of the corrosion area of the branch frame, nnRepresents the area of the normal region of the sub-mount,
showing the rusted area ratio of the branch frame.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011158844.0A CN112187174A (en) | 2020-10-26 | 2020-10-26 | Solar photovoltaic bracket abnormity detection method and system based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011158844.0A CN112187174A (en) | 2020-10-26 | 2020-10-26 | Solar photovoltaic bracket abnormity detection method and system based on artificial intelligence |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112187174A true CN112187174A (en) | 2021-01-05 |
Family
ID=73922817
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011158844.0A Withdrawn CN112187174A (en) | 2020-10-26 | 2020-10-26 | Solar photovoltaic bracket abnormity detection method and system based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112187174A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113591267A (en) * | 2021-06-17 | 2021-11-02 | 东风汽车集团股份有限公司 | Analysis method and device for suspension strength of gearbox shell |
| CN113838133A (en) * | 2021-09-23 | 2021-12-24 | 上海商汤科技开发有限公司 | State detection method and device, computer equipment and storage medium |
-
2020
- 2020-10-26 CN CN202011158844.0A patent/CN112187174A/en not_active Withdrawn
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113591267A (en) * | 2021-06-17 | 2021-11-02 | 东风汽车集团股份有限公司 | Analysis method and device for suspension strength of gearbox shell |
| CN113591267B (en) * | 2021-06-17 | 2023-12-19 | 东风汽车集团股份有限公司 | Analysis method and device for suspension strength of gearbox shell |
| CN113838133A (en) * | 2021-09-23 | 2021-12-24 | 上海商汤科技开发有限公司 | State detection method and device, computer equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113744270B (en) | Unmanned aerial vehicle visual detection and identification method for crane complex steel structure surface defects | |
| CN108416294B (en) | Fan blade fault intelligent identification method based on deep learning | |
| CN107808133B (en) | Unmanned aerial vehicle line patrol-based oil and gas pipeline safety monitoring method and system and software memory | |
| CN111537515A (en) | Iron tower bolt defect display method and system based on three-dimensional live-action model | |
| CN113554595B (en) | Unmanned aerial vehicle laser radar point cloud tower head deformation detection device and method | |
| CN112187174A (en) | Solar photovoltaic bracket abnormity detection method and system based on artificial intelligence | |
| US20200410710A1 (en) | Method for measuring antenna downtilt based on linear regression fitting | |
| CN110910350B (en) | Nut loosening detection method for wind power tower cylinder | |
| CN113065446B (en) | Deep inspection method for automatically identifying corrosion area of naval vessel | |
| KR102346676B1 (en) | Method for creating damage figure using the deep learning-based damage image classification of facility | |
| CN111914720B (en) | Method and device for identifying insulator burst of power transmission line | |
| CN115240093B (en) | Automatic power transmission channel inspection method based on visible light and laser radar point cloud fusion | |
| CN110910440B (en) | Power transmission line length determination method and system based on power image data | |
| CN113378686A (en) | Two-stage remote sensing target detection method based on target center point estimation | |
| CN106780540B (en) | photovoltaic power generation-oriented ground cloud picture cloud layer tracking and early warning method | |
| CN111079822A (en) | Method for identifying dislocation fault image of middle rubber and upper and lower plates of axle box rubber pad | |
| CN111738206A (en) | Excavator detection method for unmanned aerial vehicle inspection based on CenterNet | |
| CN103679740A (en) | ROI (Region of Interest) extraction method of ground target of unmanned aerial vehicle | |
| CN111507398A (en) | Transformer substation metal instrument corrosion identification method based on target detection | |
| CN117288094B (en) | Real-time positioning system of heading machine based on laser sensor | |
| CN117237597A (en) | Data processing terminal based on Beidou satellite data and AI graph fusion | |
| CN116843738A (en) | Tree dumping risk assessment system and method based on TOF depth camera | |
| CN113947569B (en) | Multi-scale weak damage positioning method for beam structure based on computer vision | |
| CN111457895B (en) | Target size calculation and display method for photoelectric load of unmanned aerial vehicle | |
| CN114332006A (en) | Automatic quantitative assessment method for urban battlement loss |
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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20210105 |
|
| WW01 | Invention patent application withdrawn after publication |