CN112733947B - Tailing dam space distribution identification system and method - Google Patents
Tailing dam space distribution identification system and method Download PDFInfo
- Publication number
- CN112733947B CN112733947B CN202110050642.2A CN202110050642A CN112733947B CN 112733947 B CN112733947 B CN 112733947B CN 202110050642 A CN202110050642 A CN 202110050642A CN 112733947 B CN112733947 B CN 112733947B
- Authority
- CN
- China
- Prior art keywords
- tailing dam
- layer
- dam
- tailing
- convolution
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computational Linguistics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a system and a method for identifying spatial distribution of a tailing dam, belonging to the technical field of identification and comprising the following steps: constructing a full-element long time sequence data set of the tailing dam; constructing a tailing dam sample database by using the remote sensing image; constructing a deep learning model; training a deep learning model by using a tailing dam sample database to obtain a tailing dam identification model; and identifying the tailing dam by using the tailing dam identification model to finish the identification of the spatial distribution of the tailing dam. The invention can summarize the established and under-construction tailing dam data into the database, not only is the tailing dam data convenient to use, but also the position can be determined according to the database information when flood disasters occur, disaster information such as disaster early warning, tailing dam damage condition identification and the like can be obtained in real time, the invention has important practical application value, and great convenience is provided for people to use water conservancy information.
Description
Technical Field
The invention belongs to the technical field of identification, and particularly relates to a system and a method for identifying spatial distribution of a tailing dam.
Background
The tailing dam is an important component of a tailing pond, is an industrial structure formed by stacking ore tailings, and has extremely strong destructive power once dam break occurs. Therefore, the harmfulness caused by dam break of the tailing dam is extremely high. Previous researches show that unstable dam break of the tailing dam often occurs in the continuous rainfall period, the continuous rainfall can increase the saturation of the dam body, reduce the matrix suction of the soil body, reduce the shear strength of the soil body and finally cause unstable damage of the tailing dam. In view of the reasons, the distribution of the tailing dam is mastered, and timely supervision is carried out in rainy seasons, so that the method has important significance for guaranteeing the life safety and property safety of people.
The existing scheme mainly analyzes the occurrence reason of the accident of the tailing dam, and researches and detects the safety of the tailing dam and an early warning model. However, the traditional scheme has the defects of single monitoring and early warning mode, lagged data use and ineffective information utilization, is only suitable for the tailing dam of a certain area or a certain region, and cannot realize the detection of a large-area. Therefore, how to monitor the tailing dam with a large area and provide a system for comprehensively managing information to update the information of the tailing dam in each place in time is a technical problem to be solved by technical personnel in the field.
Disclosure of Invention
Aiming at the defects in the prior art, the system and the method for identifying the spatial distribution of the tailings dam solve the problems.
In order to achieve the above purpose, the invention adopts the technical scheme that:
the scheme provides a tailing dam spatial distribution recognition system which comprises a data acquisition module, a multi-source tailing dam sample data module, a deep learning module, a tailing dam updating module, a tailing dam dynamic monitoring module and a cloud platform;
the data acquisition module is used for acquiring constructed and under-construction tailing dam engineering information and constructing a tailing dam full-element long time sequence data set according to the tailing dam engineering information;
the tailing dam sample data module is used for constructing a tailing dam sample database by utilizing a tailing dam multi-source remote sensing image;
the deep learning module is used for constructing a deep learning model by using a deep learning method, training the deep learning model by using a sample database of the tailing dam to obtain a tailing dam identification model, carrying out target detection on the tailing dam by using the tailing dam identification model, and endowing the attribute information of a full-element long time sequence data set of the tailing dam to a detected tailing dam target; the full-element long time sequence data set comprises the name, the geographic position, a project main body, an engineering construction condition, engineering investment and an operation condition of the tailing dam;
the tailing dam dynamic monitoring module is used for crawling dynamic information of the tailing dam periodically according to a detection result to obtain updated dynamic information of the tailing dam, and dynamically monitoring the updated dynamic information of the tailing dam by combining a remote sensing image;
and the cloud platform is used for uploading the tailing dam identification model, the dynamic information of the tailing dam and the updated dynamic information of the tailing dam to the cloud platform to complete the identification of the spatial distribution of the tailing dam.
Based on the system, the invention also provides a tailing dam space distribution identification method, which comprises the following steps:
s1, obtaining constructed and under-construction tailing dam engineering information, and constructing a tailing dam full-element long time sequence data set according to the tailing dam engineering information; the full-element long time sequence data set comprises the name, the geographic position, a project main body, an engineering construction condition, engineering investment and an operation condition of the tailing dam;
s2, constructing a tailing dam sample database by using the multi-source remote sensing image of the tailing dam;
s3, constructing a deep learning model by using a deep learning method;
s4, training a deep learning model by utilizing a tailing dam sample database to obtain a tailing dam identification model;
s5, carrying out target detection on the tailing dam by using the tailing dam identification model, and endowing the attribute information of the full-element long time sequence data set of the tailing dam to the detected tailing dam target;
s6, crawling the dynamic information of the tailing dam regularly according to the detection result to obtain updated dynamic information of the tailing dam, and dynamically monitoring the updated dynamic information of the tailing dam by combining a remote sensing image;
s7, uploading the tailing dam identification model, the dynamic information of the tailing dam and the updated dynamic information of the tailing dam to a cloud platform, and completing the identification of the spatial distribution of the tailing dam.
Further, the step S2 includes the following steps:
s201, sketching all multi-source remote sensing images of the tailing dam, obtaining a binary image according to sketched tailing dam areas and non-tailing dam areas, and storing the binary image;
s202, segmenting the original tailing dam image and the binary image, storing a data set with the size of 224x224, and obtaining a tailing dam sample database.
Still further, the deep learning model in step S3 includes an input layer, a first convolutional layer connected to the input layer, a first pooling layer connected to the first convolutional layer, a second convolutional layer connected to the first pooling layer, a second pooling layer connected to the second convolutional layer, a third convolutional layer connected to the second pooling layer, a third pooling layer connected to the third convolutional layer, a fourth convolutional layer connected to the third pooling layer, a fourth pooling layer connected to the fourth convolutional layer, a fifth convolutional layer connected to the fourth pooling layer, a fifth pooling layer connected to the fifth convolutional layer, a first fully-connected layer connected to the fifth pooling layer, a second fully-connected layer connected to the first fully-connected layer, a third fully-connected layer connected to the second fully-connected layer, an HSV image layer connected to the input layer, a second convolutional layer connected to the input layer, a third convolutional layer connected to the second convolutional layer, a third convolutional layer connected to the fourth convolutional layer, a second convolutional layer connected to the fourth convolutional layer, a second convolutional layer, a third convolutional layer connected to the fourth convolutional layer, a third full-connected to the fourth convolutional layer, a sixth image layer connected to the sixth image layer, a sixth image, A sixth convolutional layer connected with the HSV imaging layer, a sixth pooling layer connected with the sixth convolutional layer, a fourth fully-connected layer connected with the sixth pooling layer, a fifth fully-connected layer connected with the fourth fully-connected layer, a sixth fully-connected layer connected with the fifth fully-connected layer, and an output layer connected with the sixth fully-connected layer; the sixth fully connected layer is connected to the third fully connected layer.
Still further, the convolution kernels of the first convolution layer, the second convolution layer, the third convolution layer, the fourth convolution layer, the fifth convolution layer and the sixth convolution layer are all 3 × 3 in size, the step length is 1 in size, and the effective filling size is 1 in size;
the sizes of convolution kernels of the first pooling layer, the second pooling layer, the third pooling layer, the fourth pooling layer, the fifth pooling layer and the sixth pooling layer are all 2 x 2.
Still further, the number of convolution kernels of each of the first convolution layer, the second convolution layer and the sixth convolution layer is 64, the number of convolution kernels of each of the third convolution layer and the fourth convolution layer is 128, and the number of convolution kernels of the fifth convolution layer is 256.
Still further, the expression of converting the RGB image into the HSV image in the HSV image layer is as follows:
when R is max:
H＝(G-B)/(max-min)*60
if G ═ max:
H＝120+(B+R)/(max-min)*60
if B ═ max:
H＝240+(R-G)/(max-min)*60
if H < 0:
H＝H+360
S＝(max-min)/max
max＝max(R,G,B)
min＝min(R,G,B)
wherein, R represents a red channel in the RGB color space, G represents a green channel in the RGB color space, B represents a blue channel in the RGB color space, H represents a hue in the HSV color space, S represents a saturation in the HSV color space, and V represents a lightness in the HSV color space.
Still further, the expression of the loss function of the output layer is as follows:
wherein, ykRepresenting the loss function of the output layer, akRepresenting the kth input signal in the output layer, aiRepresenting the ith input signal in the output layer, exp(ak) Denotes the sum of the exponents of the input signals in the output layer, and n denotes the number of signals of the output signal.
The invention has the beneficial effects that:
(1) the invention uses the network big data to determine the information of the position, the area, the construction condition and the like of the tailing dam, and then uses the remote sensing data to carry out the actual positioning of the tailing dam data, thereby obtaining the spatial distribution information of the tailing dam. The method provided by the invention can be used for summarizing the established and building tailing dam data into the database, so that the tailing dam data can be conveniently used, the position can be determined according to the database information when an flood disaster occurs, disaster information such as disaster early warning and tailing dam damage condition identification can be obtained in real time, the method has important practical application value, and great convenience is provided for people to use water conservancy information.
(2) The invention provides a new tailing dam sample library, a multi-source sample database is constructed by using data corresponding to various remote sensing images such as satellite remote sensing images and unmanned aerial vehicle images.
(3) The method uses the convolutional neural network to train a large number of tailing dam samples to construct the model, but the method does not use the traditional classical deep learning model, but improves the deep model so as to improve the operation efficiency of the improved model.
(4) In view of the fact that the traditional method cannot carry out comprehensive management of a large area, the construction information of the tailing dam is obtained through network data, and then the target identification of the tailing dam is carried out in a targeted mode through remote sensing data.
Drawings
FIG. 1 is a schematic diagram of the system of the present invention.
FIG. 2 is a flow chart of the method of the present invention.
FIG. 3 is a schematic structural diagram of a deep learning model according to the present invention.
Detailed Description
The following description of the embodiments of the present invention is provided to facilitate the understanding of the present invention by those skilled in the art, but it should be understood that the present invention is not limited to the scope of the embodiments, and it will be apparent to those skilled in the art that various changes may be made without departing from the spirit and scope of the invention as defined and defined in the appended claims, and all matters produced by the invention using the inventive concept are protected.
Example 1
As shown in fig. 1, the invention provides a system for identifying spatial distribution of a tailing dam, which comprises a data acquisition module, a multi-source tailing dam sample data module, a deep learning module, a tailing dam updating module, a tailing dam dynamic monitoring module and a cloud platform, wherein the data acquisition module is used for acquiring data of a tailing dam; the data acquisition module is used for acquiring constructed and under-construction tailing dam engineering information and constructing a tailing dam full-element long time sequence data set according to the tailing dam engineering information; the tailing dam sample data module is used for constructing a tailing dam sample database by utilizing the multi-source remote sensing image of the tailing dam; the deep learning module is used for constructing a deep learning model by using a deep learning method, training the deep learning model by using a sample database of the tailing dam to obtain a tailing dam identification model, carrying out target detection on the tailing dam by using the tailing dam identification model, and endowing the attribute information of the full-element long time sequence data set of the tailing dam to the detected target of the tailing dam; the tailing dam dynamic monitoring module is used for crawling dynamic information of the tailing dam periodically according to a detection result to obtain updated dynamic information of the tailing dam, and dynamically monitoring the updated dynamic information of the tailing dam by combining a remote sensing image; and the cloud platform is used for uploading the tailing dam identification model, the dynamic information of the tailing dam and the updated dynamic information of the tailing dam to the cloud platform to complete the spatial distribution identification of the tailing dam.
In the embodiment, the data of the established and building tailings dam can be summarized into the database, so that the data of the tailings dam can be conveniently used, the position can be determined according to the database information when an flood disaster occurs, disaster information such as disaster early warning and tailing dam damage condition identification can be obtained in real time, the practical application value is important, and great convenience is provided for people to use water conservancy information.
Example 2
As shown in fig. 2, the invention provides a method for identifying spatial distribution of a tailing dam, which comprises the following steps:
s1, obtaining the constructed and under-construction tailing dam engineering information, and constructing a tailing dam full-element long time sequence data set according to the tailing dam engineering information.
In this embodiment, the information on the internet is rich in variety, and data crawlers are regularly performed on microblog data, news reports, historical data, yearbook, discount, government bulletins, flood and drought disaster bulletins, chinese and english document libraries and other data, so that maintenance and construction information of the tailing dam occurring at a certain time in a certain area is acquired according to keywords such as keyword 'dam', 'tailing dam', 'water conservancy facility', 'time', 'place', and support is provided for updating dynamic information of the tailing dam every day. Obtaining maintenance and construction information of a tailing dam, and the method comprises the following steps: when a crawler system crawls pages, the pages are sequenced according to time sequence by using a topic web crawler algorithm based on keywords, and the results are crawled and stored, and at the moment, the latest time point of data release of the crawled pages in the previous round is recorded; comparing the crawling time recorded with the last time in the next round of crawling, if the crawling time is earlier than the last time, the webpage which is crawled is shown to be filtered, and if the crawling time is later than the last time point, the webpage is crawled; when the data is stored, simple processing is carried out on the content captured by the crawler, such as title extraction, content extraction, time extraction and the like, and duplicate removal processing is also carried out during storage, so that more processing resources are not wasted. And updating the constructed full-factor dynamic information of the tailing dam in time according to the tailing dam updated every day on the Internet and the related information thereof to form a dynamic tailing dam monitoring system based on ubiquitous network data.
In this embodiment, according to the information of the constructed and in-progress constructed tailings dam, a full-element and long-term data set of the tailings dam is constructed according to the name, the geographical position, the engineering construction condition, the construction purpose, the project subject, the engineering investment, the operation condition, the major event, and the like.
S2, constructing a tailing dam sample database by utilizing the tailing dam multi-source remote sensing image, wherein the implementation method comprises the following steps:
s201, sketching all multi-source remote sensing images of the tailing dam, obtaining a binary image according to sketched tailing dam areas and non-tailing dam areas, and storing the binary image;
s202, segmenting the original tailing dam image and the binary image, storing a data set with the size of 224x224, and obtaining a tailing dam sample database.
In this embodiment, a deep learning technique is used to perform rapid target positioning of the tailing dam target on data sources such as space remote sensing images, aerial remote sensing images, low-altitude remote sensing images, and the like, and the detected target tailing dam is given full-element long-time tailing dam database information constructed in step 2. The way to construct the sample database is as follows: and (3) sketching a tailing dam mask for all tailing dam images, storing a sketched tailing dam area and a non-tailing dam area as binary images, performing image segmentation on the original image and the constructed binary image, and storing a data set with the size of 224x 224.
And S3, constructing a deep learning model by using a deep learning method.
As shown in fig. 3, the deep learning model includes an input layer, a first convolution layer connected to the input layer, a first pooling layer connected to the first convolution layer, a second convolution layer connected to the first pooling layer, a second pooling layer connected to the second convolution layer, a third convolution layer connected to the second pooling layer, a third pooling layer connected to the third convolution layer, a fourth convolution layer connected to the third pooling layer, a fourth pooling layer connected to the fourth convolution layer, a fifth convolution layer connected to the fourth pooling layer, a fifth pooling layer connected to the fifth convolution layer, a first full-link layer connected to the fifth pooling layer, a second full-link layer connected to the first full-link layer, a third full-link layer connected to the second full-link layer, an HSV image layer connected to the input layer, a first convolution layer connected to the first convolution layer, a second convolution layer connected to the first full-link layer, a second convolution layer connected to the second full-link layer, a second convolution layer connected to the fifth convolution layer, a second convolution layer connected to the input layer, a second convolution layer connected to the second convolution layer, a third convolution layer connected to the fifth convolution layer, a third convolution layer, a second convolution layer, a third convolution layer connected to the input layer, a third convolution layer, a second convolution layer, a third convolution layer, a sixth convolutional layer connected with the HSV imaging layer, a sixth pooling layer connected with the sixth convolutional layer, a fourth fully-connected layer connected with the sixth pooling layer, a fifth fully-connected layer connected with the fourth fully-connected layer, a sixth fully-connected layer connected with the fifth fully-connected layer, and an output layer connected with the sixth fully-connected layer; the sixth fully connected layer is connected to the third fully connected layer. The convolution kernel sizes of the first convolution layer, the second convolution layer, the third convolution layer, the fourth convolution layer and the fifth convolution layer are all 3 multiplied by 3, the step lengths are all 1, and the effective filling sizes are all 1; the sizes of convolution kernels of the first pooling layer, the second pooling layer, the third pooling layer, the fourth pooling layer and the fifth pooling layer are all 2 x2, the number of convolution kernels of the first convolution layer, the second convolution layer and the sixth convolution layer is 64, the number of convolution kernels of the third convolution layer and the fourth convolution layer is 128, and the number of convolution kernels of the fifth convolution layer is 256.
In this embodiment, a deep learning model is built, and as shown in fig. 3, the deep learning model includes a deep learning network of 5 convolutional layers and 3 full-connection layers; for a reference deep learning model, adding 1 convolutional layer and 2 full-connection layers to extract color information in an image, performing deep learning model training through one full-connection layer in combination with a basic neural network model and a neural network module for extracting colors, and using a dropout algorithm to enhance the generalization capability of the network in view of the phenomenon that overfitting can be generated in model training.
In this embodiment, the deep learning convolutional layer process: the size of the initial convolution kernel is 3 × 3 × 3, the size of the stride is 1, the size of the effective padding is 1, and the pooling layer pooling adopts a maximum pooling function max pooling of 2 × 2.
In this embodiment, the reference depth learning model is as follows: in the first convolution layer, convolution processing of 64 convolution kernels is used once, and pooling layer posing processing is carried out once; in the second convolution layer, using convolution processing of 64 convolution kernels once and performing pooling layer posing processing once; in the third convolution layer, a convolution process of 128 convolution kernels is used once, and a pooling layer posing process is performed once; in the fourth convolution layer, a convolution process of 128 convolution kernels is used once, and a pooling layer posing process is performed once; in the five convolution layers, convolution processing of 256 convolution kernels is used once, and pooling layer posing processing is carried out once; three full-link layers, a first full-link layer, a second full-link layer and a third full-link layer, are used for processing.
In this embodiment, the model includes a structure of 1 sixth convolution layer, a fourth full-link layer, and a fifth full-link layer, and 2 full-link layers: performing convolution processing on the sixth convolution layer by using 64 convolution kernels once, and performing Pooling layer processing once; and processing by using two full connection layers of a fourth full connection layer and a fifth full connection layer.
In this embodiment, for the depth model module for extracting color, the pixel point is converted from RGB (R represents red; G represents green; B represents blue) pixel point to HSV (H represents hue; S represents saturation; V represents lightness), and the conversion formula is as follows:
when R is max:
H＝(G-B)/(max-min)*60
if G ═ max:
H＝120+(B+R)/(max-min)*60
if B ═ max:
H＝240+(R-G)/(max-min)*60
if H < 0:
H＝H+360
S＝(max-min)/max
max＝max(R,G,B)
min＝min(R,G,B)
wherein, R represents a red channel in the RGB color space, G represents a green channel in the RGB color space, B represents a blue channel in the RGB color space, H represents a hue in the HSV color space, S represents a saturation in the HSV color space, and V represents a lightness in the HSV color space.
In this embodiment, max means the maximum value, max (R, G, B) means the maximum value of three RGB values determined, and max is (5,4,8), which means 8 is the value of max and is also the maximum value, and Min is the minimum value, as above.
In this embodiment, for an HSV image, a convolution process of 64 convolution kernels is performed once, a pooling layer posing process is performed once, and color information features of the image are finally extracted while two full-connected layers are used.
In this embodiment, the expression of the loss function of the output layer is as follows:
wherein, ykRepresenting the loss function of the output layer, akRepresenting the kth input signal in the output layer, aiDenotes the ith input signal in the output layer, exp (a)k) Denotes the sum of the exponents of the input signals in the output layer, and n denotes the number of signals of the output signal.
S4, training the deep learning model by utilizing the tailing dam sample database to obtain a tailing dam identification model.
In this embodiment, the training process is as follows: the method comprises the steps that a multi-source remote sensing data sample is used as an input sample, a tailing dam range sketched in the sample is used as a training label, after a deep learning model is determined, a certain number of samples and labels corresponding to the samples are taken to be input into the deep learning model each time, the deep learning model can automatically learn error values between the samples and the labels, the error values of the deep learning model are smaller and smaller along with the continuous increase of training times, the model precision of final training is higher, and the training process is finished. The correctly trained model can correctly output the water body range in the image of any image.
S5, carrying out target detection on the tailing dam by using the tailing dam identification model, and endowing the attribute information of the tailing dam full-element long time sequence data set to the detected tailing dam target.
In this embodiment, a trained tailing dam recognition model is used to perform target detection on multi-source data such as space remote sensing images, aerial remote sensing images and low-altitude remote sensing images.
S6, crawling the dynamic information of the tailing dam regularly according to the detection result to obtain updated dynamic information of the tailing dam, and dynamically monitoring the updated dynamic information of the tailing dam by combining a remote sensing image;
s7, uploading the tailing dam identification model, the dynamic information of the tailing dam and the updated dynamic information of the tailing dam to a cloud platform to finish the identification of the spatial distribution of the tailing dam
In the embodiment, dynamic detection and information updating of the tailing dam are realized according to a deep learning tailing dam target detection method according to remote sensing data such as satellite remote sensing images, unmanned aerial vehicle remote sensing images and ground actual observation data acquired every day.
In the embodiment, the deep learning model constructed by the method and the daily updated remote sensing data are uploaded to a built cloud platform, updated tailing dam information is automatically acquired in a network every day through a crawler technology, the acquired remote sensing images every day are uploaded to the cloud platform, and monitoring and updating of the data are automatically completed in the cloud platform.
In the embodiment, a remote sensing monitoring and ubiquitous network monitoring method is integrated, a new way for dynamically monitoring the tailing dam of the yellow river basin is created, a dynamic active monitoring system of the tailing dam is established, hot spots and key targets for construction and management of the tailing dam are actively discovered, tracked and locked, and automatic extraction of information of the deep tailing dam is realized. And updating the spatial distribution identification information of the tailing dam according to the remote sensing image and the network big data.
In the embodiment, the actual working conditions of the tailing dam are monitored by using the remote sensing big data and the multi-source remote sensing data, and the deformation of the dam body, the position of the infiltration line, the pore water pressure, the seepage water quantity, the water quality, the soil pressure and the like can be observed and monitored in time. The observation contents are wide, such as the dam slope has no obvious deformation, collapse pit, swampiness, water seepage, crack, ant hole and rat hole, etc. The tailings dam can be divided into four levels, namely a dangerous reservoir, an dangerous reservoir, a sick reservoir and a normal reservoir according to observation data, different detection modes can be adopted for different levels subsequently, and cost saving is facilitated.
Claims (7)
1. A tailing dam space distribution identification system is characterized by comprising a data acquisition module, a multi-source tailing dam sample data module, a deep learning module, a tailing dam updating module, a tailing dam dynamic monitoring module and a cloud platform;
the data acquisition module is used for acquiring constructed and under-construction tailing dam engineering information and constructing a tailing dam full-element long time sequence data set according to the tailing dam engineering information;
the tailing dam sample data module is used for constructing a tailing dam sample database by utilizing a tailing dam multi-source remote sensing image;
the deep learning module is used for constructing a deep learning model by using a deep learning method, training the deep learning model by using a sample database of the tailing dam to obtain a tailing dam identification model, carrying out target detection on the tailing dam by using the tailing dam identification model, and endowing the attribute information of a full-element long time sequence data set of the tailing dam to a detected tailing dam target; the full-element long time sequence data set comprises the name, the geographic position, a project main body, an engineering construction condition, engineering investment and an operation condition of the tailing dam;
the tailing dam dynamic monitoring module is used for crawling dynamic information of the tailing dam periodically according to a detection result to obtain updated dynamic information of the tailing dam, and dynamically monitoring the updated dynamic information of the tailing dam by combining a remote sensing image;
and the cloud platform is used for uploading the tailing dam identification model, the dynamic information of the tailing dam and the updated dynamic information of the tailing dam to the cloud platform to complete the identification of the spatial distribution of the tailing dam.
2. A tailing dam space distribution identification method is characterized by comprising the following steps:
s1, obtaining constructed and under-construction tailing dam engineering information, and constructing a tailing dam full-element long time sequence data set according to the tailing dam engineering information;
s2, constructing a tailing dam sample database by using the multi-source remote sensing image of the tailing dam;
s3, constructing a deep learning model by using a deep learning method;
s4, training a deep learning model by utilizing a tailing dam sample database to obtain a tailing dam identification model;
s5, carrying out target detection on the tailing dam by using the tailing dam identification model, and endowing the attribute information of the full-element long time sequence data set of the tailing dam to the detected tailing dam target; the full-element long time sequence data set comprises the name, the geographic position, a project main body, an engineering construction condition, engineering investment and an operation condition of the tailing dam;
s6, crawling the dynamic information of the tailing dam regularly according to the detection result to obtain updated dynamic information of the tailing dam, and dynamically monitoring the updated dynamic information of the tailing dam by combining a remote sensing image;
s7, uploading the tailing dam identification model, the dynamic information of the tailing dam and the updated dynamic information of the tailing dam to a cloud platform, and completing the identification of the spatial distribution of the tailing dam.
3. The tailings dam spatial distribution identification method of claim 2, wherein the step S2 comprises the steps of:
s201, sketching all multi-source remote sensing images of the tailing dam, obtaining a binary image according to sketched tailing dam areas and non-tailing dam areas, and storing the binary image;
s202, segmenting the original tailing dam image and the binary image, storing a data set with the size of 224x224, and obtaining a tailing dam sample database.
4. The tailings dam spatial distribution identification method of claim 2, wherein the deep learning model in step S3 comprises an input layer, a first convolutional layer connected to the input layer, a first pooling layer connected to the first convolutional layer, a second convolutional layer connected to the first pooling layer, a second pooling layer connected to the second convolutional layer, a third convolutional layer connected to the second pooling layer, a third pooling layer connected to the third convolutional layer, a fourth convolutional layer connected to the third pooling layer, a fourth pooling layer connected to the fourth convolutional layer, a fifth convolutional layer connected to the fourth pooling layer, a fifth pooling layer connected to the fifth convolutional layer, a first fully-connected layer connected to the fifth pooling layer, a second fully-connected layer connected to the first fully-connected layer, a third fully-connected layer connected to the second fully-connected layer, a third fully-connected layer connected to the third fully-connected layer, and a third fully-connected layer, An HSV image layer connected with the input layer, a sixth convolution layer connected with the HSV image layer, a sixth pooling layer connected with the sixth convolution layer, a fourth full-link layer connected with the sixth pooling layer, a fifth full-link layer connected with the fourth full-link layer, a sixth full-link layer connected with the fifth full-link layer and an output layer connected with the sixth full-link layer; the sixth fully connected layer is connected to the third fully connected layer.
5. The tailings dam spatial distribution identification method of claim 4, wherein the convolution kernel sizes of the first convolution layer, the second convolution layer, the third convolution layer, the fourth convolution layer, the fifth convolution layer and the sixth convolution layer are all 3 x 3, the step lengths are all 1, and the effective filling sizes are all 1;
the sizes of convolution kernels of the first pooling layer, the second pooling layer, the third pooling layer, the fourth pooling layer, the fifth pooling layer and the sixth pooling layer are all 2 x 2.
6. The tailings dam spatial distribution identification method of claim 5, wherein the number of convolution kernels of the first convolution layer, the second convolution layer and the sixth convolution layer is 64, the number of convolution kernels of the third convolution layer and the fourth convolution layer is 128, and the number of convolution kernels of the fifth convolution layer is 256.
7. The tailings dam spatial distribution identification method of claim 6, wherein the loss function of the output layer is expressed as follows:
wherein, ykRepresenting the loss function of the output layer, akRepresenting the kth input signal in the output layer, aiDenotes the ith input signal in the output layer, exp (a)k) Indicating the sum of exponentials of input signals in the output layer, n indicating output signalsThe number of signals.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110050642.2A CN112733947B (en) | 2021-01-14 | 2021-01-14 | Tailing dam space distribution identification system and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110050642.2A CN112733947B (en) | 2021-01-14 | 2021-01-14 | Tailing dam space distribution identification system and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112733947A CN112733947A (en) | 2021-04-30 |
| CN112733947B true CN112733947B (en) | 2021-11-09 |
Family
ID=75593149
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110050642.2A Active CN112733947B (en) | 2021-01-14 | 2021-01-14 | Tailing dam space distribution identification system and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112733947B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020034004A1 (en) * | 2018-08-17 | 2020-02-20 | William Smith | Leak detection and location system |
| CN111738144A (en) * | 2020-06-19 | 2020-10-02 | 中国水利水电科学研究院 | Surface water product generation method and system based on Google Earth Engine cloud platform |
| CN111931696A (en) * | 2020-09-04 | 2020-11-13 | 中国水利水电科学研究院 | Lake and reservoir water area remote sensing automatic extraction method based on space constraint |
-
2021
- 2021-01-14 CN CN202110050642.2A patent/CN112733947B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020034004A1 (en) * | 2018-08-17 | 2020-02-20 | William Smith | Leak detection and location system |
| CN111738144A (en) * | 2020-06-19 | 2020-10-02 | 中国水利水电科学研究院 | Surface water product generation method and system based on Google Earth Engine cloud platform |
| CN111931696A (en) * | 2020-09-04 | 2020-11-13 | 中国水利水电科学研究院 | Lake and reservoir water area remote sensing automatic extraction method based on space constraint |
Non-Patent Citations (1)
| Title |
|---|
| 《基于深度学习的SSD模型尾矿库自动提取》;闫凯 等;《中国科学院大学学报》;20200531;第360-367页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112733947A (en) | 2021-04-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Li et al. | Water body classification from high-resolution optical remote sensing imagery: Achievements and perspectives | |
| Gao et al. | MLNet: Multichannel feature fusion lozenge network for land segmentation | |
| CN108416985B (en) | Geological disaster monitoring and early warning system and method based on image recognition | |
| CN105631479A (en) | Imbalance-learning-based depth convolution network image marking method and apparatus | |
| CN105550709A (en) | Remote sensing image power transmission line corridor forest region extraction method | |
| CN111178283B (en) | Unmanned aerial vehicle image-based method for identifying and positioning ground feature of established route | |
| CN111784082B (en) | GIS mountain torrent prevention early warning system based on big data | |
| CN110717493B (en) | License plate recognition method containing stacked characters based on deep learning | |
| CN114998744B (en) | Agricultural machinery track field dividing method and device based on motion and vision dual-feature fusion | |
| CN112750138A (en) | Method for identifying space distribution of silt dam of yellow river basin | |
| CN114898212A (en) | High-resolution remote sensing image multi-terrain change information extraction method | |
| Cai et al. | A comparative study of deep learning approaches to rooftop detection in aerial images | |
| CN117454116A (en) | Ground carbon emission monitoring method based on multi-source data interaction network | |
| Xiao et al. | Novel convolutions for semantic segmentation of remote sensing images | |
| CN114863099A (en) | Foundation cloud atlas segmentation network based on multi-branch asymmetric convolution module and multi-scale feature fusion | |
| CN112733947B (en) | Tailing dam space distribution identification system and method | |
| CN114241310A (en) | Improved YOLO model-based intelligent identification method for piping dangerous case of dike | |
| CN116912700A (en) | River and lake regulation violation problem change detection method based on feature enhancement and refinement | |
| Wang et al. | CCENet: Cascade class-aware enhanced network for high-resolution aerial imagery semantic segmentation | |
| Zhang et al. | Road extraction from multi-source high-resolution remote sensing image using convolutional neural network | |
| Wang et al. | Optimized faster R-CNN for oil wells detection from high-resolution remote sensing images | |
| Gao et al. | IUNet-IF: identification of construction waste using unmanned aerial vehicle remote sensing and multi-layer deep learning methods | |
| CN114170528A (en) | Strong convection region identification method based on satellite cloud picture | |
| Zheng et al. | Multiscale fusion network for rural newly constructed building detection in unmanned aerial vehicle imagery | |
| Liu et al. | A New High-Precision and Lightweight Detection Model for Illegal Construction Objects Based on Deep Learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |