CN110597165A - A steel pile monitoring system and a steel pile monitoring method - Google Patents

Abstract

An embodiment of the present invention provides a steel pile monitoring system, which relates to the field of steel pile monitoring, which includes a monitoring subnet, a control subnet, and a central controller; the central controller is connected to the monitoring subnet and the control subnet respectively; the monitoring subnet The network is used to collect red steel images and extract characteristic parameters based on the red steel images. The central controller generates control commands according to the characteristic parameters sent by the monitoring subnet and sends the control commands to the control subnet; wherein, the monitoring The subnet includes multiple visible light monitoring nodes and multiple infrared monitoring nodes; the control subnet is connected to the external equipment in the steel rolling area to control the external equipment to deal with the steel pile accident scene. The invention provides a steel pile monitoring method. Compared with manual monitoring, the present invention can monitor whether there is a steel pile accident for a long time without being on duty, greatly reducing the manpower burden, and will not be affected by human factors .

Description

A steel pile monitoring system and a steel pile monitoring method

technical field

The invention relates to the field of steel pile monitoring, in particular to a steel pile monitoring system and a steel pile detection method.

Background technique

At present, the existing steel pile monitoring mainly relies on manual monitoring. When the pile of steel occurs, the on-site personnel press the emergency stop switch to stop the rolling production line. Or use the signal generated by the collision between the piled steel and external objects when the piled steel occurs, or use a non-contact looper monitor to monitor the rolling process, and then pass the monitoring results to the PLC controller, and then the PLC Handle the incident accordingly.

However, due to the negligence or fatigue of on-site personnel, manual monitoring usually cannot respond in time. However, the current automatic monitoring scheme needs to install the sensor close to the steel rolling area. Due to the high temperature of the steel rolling and the physical collision of the red steel, the sensor is easily damaged, so that the monitoring system itself needs regular maintenance, which increases the maintenance cost. In addition, the existing monitoring scheme needs to be connected to the PLC controller, and the PLC program needs to be changed. For the ordinary steel rolling production line, the installation of the monitoring system requires a long time for debugging and modification, which affects production.

Contents of the invention

In order to solve the problems mentioned above in the background technology, the purpose of the embodiments of the present invention is to provide a steel pile monitoring system and a steel pile monitoring method.

A preferred embodiment of the present invention provides a steel pile monitoring system, including a monitoring subnet, a control subnet and a central controller;

The central controller is connected with the monitoring subnet and the control subnet respectively; the monitoring subnet is used to collect red steel images and extract characteristic parameters based on the red steel images, and the central controller according to the The characteristic parameter sent by the monitoring subnet generates a control command and sends the control command to the control subnet; wherein, the monitoring subnet includes a plurality of visible light monitoring nodes and a plurality of infrared monitoring nodes; the The control sub-network is connected with external equipment, and the external equipment is controlled to handle the piled steel accident site.

Further, the monitoring subnet is connected to the central controller through a communication protocol of TCP/IP.

Further, the control subnet is connected to the central controller through the communication protocol of Ethercat industrial Ethernet.

Further, the central controller is an upper computer, and each visible light monitoring node is equipped with a first lower computer and a high-speed industrial camera.

Further, the central controller is an upper computer, and each infrared monitoring node is provided with a second lower computer and a far-infrared imager.

Further, the first lower computer and the second lower computer are industrial personal computers.

Further, the field devices include alarm lights and flying shears.

Embodiments of the present invention also provide a method for monitoring piled steel, wherein the method uses the above-mentioned piled steel monitoring system to monitor piled steel, and the steps include:

Obtain the red steel image; wherein, the red steel image includes the visible light image collected by the high-speed industrial camera and the infrared image collected by the far-infrared imager;

Image processing is performed on the red steel image to obtain an image to be detected; wherein, the image processing includes performing region-of-interest division and image segmentation processing on the red steel image;

Carrying out steel stacking phenomenon recognition based on the image to be detected, to determine the steel rolling area corresponding to the image to be detected corresponding to the steel stacking accident;

Based on the phenomenon of piled steel, the external equipment is controlled to handle the piled steel accident on-site.

Further, based on the image to be detected, the phenomenon of steel stacking is identified, so as to determine that the image to be detected corresponds to the steel rolling area where the steel stacking accident occurs, including:

Performing area cross-border detection on the image to be detected to obtain the first pile of steel phenomenon recognition results;

Carrying out time-of-arrival detection on the image to be detected to obtain a recognition result of the second pile of steel phenomena;

Based on the recognition result of the first pile of steel phenomenon and the recognition result of the second pile of steel phenomenon, it is determined that the steel rolling area where the steel pile accident occurs corresponds to the image to be detected.

Further, it also includes:

Acquiring the red steel shape feature information of the image to be detected;

Based on the shape feature information of the red steel, the recognition result of the first pile of steel phenomenon and the recognition result of the second pile of steel phenomenon, a machine learning method is used to construct a pile of steel monitoring model to predict the occurrence of a pile of steel accident.

The steel pile monitoring system and steel pile monitoring method provided by the present invention, compared with manual monitoring, the present invention can monitor whether there is a steel pile accident for a long time without being on duty, which greatly reduces the manpower burden, and Will not be affected by human factors. Compared with the existing automatic monitoring scheme, the present invention can place the visual sensor far away from the steel rolling area, and will not be affected by the high temperature and physical collision of the steel rolling, and has a long working time. In addition, the system signal judgment does not go through the PLC, which has little impact on the existing control system, is convenient for construction, and does not affect production.

Description of drawings

In order to illustrate the technical solutions of the embodiments of the present invention more clearly, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention, and thus It should be regarded as a limitation on the scope, and those skilled in the art can also obtain other related drawings based on these drawings without creative work.

Fig. 1 is a topological schematic diagram of the principle structure of a stack steel monitoring system provided by the first embodiment of the present invention;

Fig. 2 is a schematic flow chart of a method for monitoring piled steel provided in the second embodiment of the present invention;

Fig. 3 is a schematic diagram of ROI area division in a steel stack monitoring method provided by the second embodiment of the present invention;

Fig. 4 is a schematic diagram of area cross-border detection in a steel stack monitoring method provided by the second embodiment of the present invention;

Fig. 5 is a schematic diagram of changes in the area of red steel in different exposure windows in a method for monitoring stacked steel provided by the second embodiment of the present invention;

Fig. 6 is a schematic diagram of the arrangement of monitoring points in a steel pile monitoring method provided by the second embodiment of the present invention.

Icons: 1-High-speed industrial camera; 2-First lower computer; 3-Far infrared imager; 4-Second lower computer; 5-Central controller; 6-Control subnet; 7-Alarm lights; 8-Flying Cut.

Detailed ways

The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention. The components of the embodiments of the invention generally described and illustrated in the figures herein may be arranged and designed in a variety of different configurations. Accordingly, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative efforts belong to the protection scope of the present invention.

It should be noted that like numerals and letters denote similar items in the following figures, therefore, once an item is defined in one figure, it does not require further definition and explanation in subsequent figures. Meanwhile, in the description of the present invention, the terms "first", "second", etc. are only used to distinguish descriptions, and cannot be understood as indicating or implying relative importance.

Please refer to FIG. 1 , the first embodiment of the present invention provides a steel stack monitoring system, including a monitoring subnet, a control subnet 6 and a central controller 5;

The central controller 5 is connected with the monitoring subnet and the control subnet 6 respectively; the monitoring subnet is used to collect red steel images and extract feature parameters based on the red steel images, and the central controller 5 Generate a control command according to the characteristic parameters sent by the monitoring subnet and send the control command to the control subnet 6; wherein, the monitoring subnet includes a plurality of visible light monitoring nodes and a plurality of infrared monitoring nodes Node; the control subnet 6 is connected to the external equipment in the steel rolling area, and is used to control the external equipment to deal with the piled steel accident scene.

In this implementation, the monitoring subnet is responsible for red steel data collection, feature extraction and some monitoring and alarm functions in the steel rolling area. There are two types of monitoring nodes, visible light monitoring nodes and infrared monitoring nodes. Both types of monitoring nodes receive the command of the central controller 5, collect the red steel image of the steel rolling area under its control, perform image segmentation and extract the characteristic parameters of the red steel and transmit it to the central controller 5.

Further, in a preferred embodiment of the present invention, the central controller 5 is an upper computer, and each visible light monitoring node is provided with a first lower computer 2 and a high-speed industrial camera 1, The high-speed industrial camera 1 is responsible for capturing images of Honggang at a rate of more than 50 frames per second. Each of the infrared monitoring nodes is provided with a second lower computer 4 and a far-infrared imager 3, and the acquisition rate of the far-infrared imager 3 is 27 frames/s.

Among them, in this embodiment, the high-speed industrial camera 1 adopts an industrial camera with a USB3.0 interface, which can collect images of Honggang at a rate of 90 frames per second at a maximum resolution of 1080*1024, and can capture the dynamics of Honggang. characteristic. However, due to the interference of many uncertain factors in visible light images, a red indicator light may cause misjudgment. Therefore, the present invention also uses the far-infrared imager 3 to image the red steel. The far-infrared imager 3 can image the temperature of objects in the range of 0-1200°C, so it can separate the red steel at 900°C from the background through an algorithm. In this embodiment, two imaging devices with different principles are used to collect data on red steel, which further ensures the accuracy of data collection and subsequent image analysis.

Further, the first lower computer 2 and the second lower computer 4 are industrial computers. Industrial computer is an industrial control computer, which has computer attributes and characteristics, such as computer motherboard, CPU, hard disk, memory, peripherals and interfaces, as well as operating system, control network and protocol, computing power, and friendly man-machine interface. Therefore, the red steel image can be segmented and the red steel characteristic parameters can be extracted through the industrial computer of these two types of nodes in the monitoring subnet, and the extracted red steel characteristic parameters can be transmitted through the transmission protocol between the upper computer and the lower computer. , will be transmitted to the host computer for the identification of steel stacking phenomenon.

In this embodiment, the red steel image collected by the high-speed industrial camera 1 and the far-infrared imager 3 can be transmitted to the first lower computer 2 and the second lower computer 4 through the USB3.0 interface for image segmentation and red steel image Feature parameter extraction, the specific steps include image segmentation after dividing the region of interest (ROI) of the red steel image, and after separating the red steel image from the background, it is ensured that the collected red steel feature parameters are all included in the ROI area On the other hand, through the delineated ROI area, the image transmission data is reduced, the calculation of the industrial computer is reduced, and the efficiency is improved.

Further, the monitoring subnet and the central controller 5 communicate with each other through the transmission protocol TCP/IP, that is, the first lower computer 2 of the monitoring subnet transfers the image-processed red steel via the TCP/IP transmission protocol. The image and red steel feature parameters are sent to the central controller 5 . The central controller 5 receives the red steel image and the characteristic parameters of the red steel sent by the monitoring subnet and conducts a comprehensive analysis again to judge whether the piled steel phenomenon occurs. If the phenomenon of stacking steel occurs, when the alarm is activated, a corresponding control command is generated and sent to the control subnetwork 6, and the control subnetwork 6 controls the external equipment in the steel rolling area to deal with the piled steel accident scene. Wherein, in this embodiment, the central controller 5 adopts a Beckhoff industrial computer to run the Twincat real-time environment to ensure its high reliability.

In addition, in this embodiment, in order to ensure the normal operation of the monitoring subnet and the central controller 5 due to system failure, the central controller 5 is generally placed in the monitoring room with a good operating environment, and adopts a real-time modified Beckhoff industrial computer itself has high reliability and is not easy to crash. And considering that the first lower computer 2 and the second lower computer 4 of the monitoring subnet need to be installed at the monitoring site, the environment is bad, and there is a certain possibility of crash. Therefore, the industrial computer system of the monitoring subnet adopts two methods of software and hardware to ensure reliability. In terms of hardware, a fully enclosed fanless industrial computer with high reliability is selected to ensure that no iron filings will enter the interior of the industrial computer and cause damage to the industrial computer. In terms of software, the software watchdog method is adopted, and each industrial computer of the monitoring subnet regularly sends a network signal to the central controller 5 to feed the dog. When the central controller 5 receives the dog feeding signal overtime, the central controller 5 Send a forced restart to the corresponding industrial computer of the monitoring subnet.

Further, the control subnet 6 and the central controller 5 are in communication connection through the transmission protocol Ethercat industrial Ethernet. Ethercat master-slave network, the central controller 5 can be used as the Ethercat master station, and the control subnet 6 can be used as a slave station. The control subnet 6 can include one or more of bus coupling terminals, Profinet bus conversion modules, and IO input and output modules. kind. The slave station module can be directly connected to the production line control network in the steel rolling area or replace the emergency stop switch through the IO module. There is no need to make major modifications to the production line. The production line is stopped first, and then the external equipment is controlled to handle the on-site steel pile accident, which is faster. . Wherein, in this embodiment, the external equipment includes a warning light 7 and a flying shear 8, the warning light 7 is used to remind workers of the steel pile phenomenon, and the flying shear 8 is used to deal with the steel pile phenomenon.

Further, the high-speed industrial camera 1 and the far-infrared imager 3 are provided with a transparent protective cover. This is due to the harsh environment of the steel rolling site, which may affect the industrial camera and thermal imager, so the far-infrared imager 3 is protected by a customized water-cooled cover, and in order to avoid the impact of iron filings on the industrial camera, it is also necessary to A custom-made protective cover protects the camera.

The steel pile monitoring system provided by the first embodiment of the present invention, compared with manual monitoring, the present invention can monitor whether there is a steel pile accident for a long time without being on duty, which greatly reduces the burden on manpower, and does not will be affected by human factors. Compared with the existing automatic monitoring scheme, the present invention can place the visual sensor far away from the steel rolling area, and will not be affected by the high temperature and physical collision of the steel rolling, and has a long working time. In addition, the system signal judgment does not go through the PLC, which has little impact on the existing control system, is convenient for construction, and does not affect production.

Please refer to Fig. 2, the second embodiment of the present invention provides a kind of stacked steel monitoring method, described method uses above-mentioned stacked steel monitoring system to carry out stacked steel monitoring, and it can be controlled by the industrial control of monitoring sub-network and central controller 5 respectively computer equipment, and at least include the following steps:

S201. Acquire the red steel image; wherein, the red steel image includes a visible light image collected by the high-speed industrial camera 1 and an infrared light image collected by the far-infrared imager 3 .

Among them, in this embodiment, the difficulty of monitoring the pile of steel is to separate the red steel from the background. Therefore, in this embodiment, the monitoring subnet of the pile of steel monitoring system uses two kinds of image acquisition sensors to collect the red steel image. These two sensors are a high-speed industrial camera 1 and a far-infrared imager 3, which are used to collect visible light images and infrared light images of red steel images, respectively. The industrial computer equipment that controls these image acquisition sensors to acquire red steel images is the industrial computer equipment in the monitoring subnet, including the first lower computer 2 and the second lower computer 4 . In this embodiment, two imaging devices with different principles are used to collect images of red steel, which further ensures the accuracy of image collection and subsequent image processing and analysis.

S202. Perform image processing on the red steel image to obtain an image to be detected; wherein the image processing includes performing interest region division and image segmentation processing on the red steel image.

Among them, in the red steel image collected by the image acquisition equipment, most of the image area of the red steel image is an environmental image image that has nothing to do with the pile steel monitoring. The existence of these environmental images will interfere with the segmentation of the red steel from the background, and add an additional unnecessary computational burden to the image processing. Therefore, before image segmentation, it is necessary to delineate the region of interest (ROI region) to reduce the processed data and improve the calculation efficiency.

In a preferred embodiment of the present invention, please refer to Fig. 3, the region of interest division step of described red steel image comprises:

Adopt Gaussian mixture model image segmentation algorithm to segment described red steel image, to mark red steel region;

Converting the red steel image processed by the Gaussian mixture model image segmentation algorithm into a binary image; wherein, the background is 0, and the red steel area is 1;

Hough transform is used to process the red steel image processed by the Gaussian mixture model image segmentation algorithm to fit the straight line that the red steel passes through;

Based on the straight line, a certain area threshold is expanded on the red steel image to obtain the ROI area.

Among them, Gaussian mixture model image segmentation (GMM) is a nonlinear learning model, which learns according to image samples and obtains a nonlinear segmentation model for segmentation. Its characteristics are that the more complete the sample, the better the segmentation effect and the more robust it is. high. However, the learning samples of the algorithm need to be calibrated manually, so it is necessary to input samples before the operation of the pile steel monitoring system. In addition, since the installation positions of the high-speed industrial camera 1 and the far-infrared imager 3 are fixed, the region of interest demarcated through the above steps can be used for a period of time in the subsequent acquired red steel images. By setting a Time threshold, the region of interest of the red steel image acquired during this period of time continues to use the region of interest that has been divided before, reducing operations and improving efficiency. After the time threshold is exceeded, the region of interest is re-divided to avoid errors caused by the possible offset of the lens positions of the high-speed industrial camera 1 and the far-infrared imager 3 due to the influence of gravity.

In this embodiment, after the red steel image defines the region of interest, in order to segment the red steel from the background of the region of interest, this embodiment also uses Gaussian mixture model image segmentation, color space segmentation and temperature threshold segmentation. One or more of the three segmentation algorithms perform image segmentation on the red steel image delimiting the region of interest.

Color space segmentation is a relatively simple segmentation algorithm. This algorithm uses the color characteristics of red steel. There is no similar color in the environment around red steel, so it can separate red steel from the background. The operation process is to convert the image from the RGB color space to the HSI color space, and use the H (hue) layer to perform binary segmentation. The color segmentation speed is fast, and the segmentation effect is good near the red steel, but it will be affected by the ambient light at a position farther away from the red steel, so this algorithm can be used to obtain the red steel segmentation in the visible light image.

The temperature threshold segmentation needs to rely on the far-infrared imager 3 for segmentation. This is because the temperature of red steel exceeds 800°C, which is much higher than the ambient temperature. By setting a certain threshold, the red steel can be separated from the background. This image segmentation method is simple and fast, so it is recognized as the final segmentation standard.

By selecting one or more of the above three segmentation algorithms, and then matching the segmentation results of the visible light image with the segmentation results of the infrared light image, a more accurate red steel segmentation image can be obtained.

On the basis of the foregoing embodiments, in a preferred embodiment of the present invention, the image segmentation step of the red steel image includes:

Segmenting the infrared light image by using a temperature threshold segmentation algorithm to obtain a first segmented image;

Segmenting the visible light image by using a Gaussian mixture model image segmentation algorithm or a color space segmentation algorithm to obtain a second segmented image;

matching the first segmented image and the second segmented image to obtain an image to be detected.

S203. Perform steel stacking phenomenon recognition based on the image to be detected, so as to determine a steel rolling area where the steel stacking accident occurs corresponding to the image to be detected.

Further, based on the image to be detected, the phenomenon of steel stacking is identified, so as to determine that the image to be detected corresponds to the steel rolling area where the steel stacking accident occurs, including:

Performing area cross-border detection on the image to be detected to obtain the first pile of steel phenomenon recognition results;

Carrying out time-of-arrival detection on the image to be detected to obtain a recognition result of the second pile of steel phenomena;

Based on the recognition result of the first pile of steel phenomenon and the recognition result of the second pile of steel phenomenon, the corresponding steel rolling area where the steel pile accident occurs in the image to be detected is determined.

Among them, in this embodiment, two algorithms are used to identify the phenomenon of piled steel on the image to be detected, and then the two kinds of recognition results are finally judged by the central controller 5 whether there is a piled steel accident, and then the corresponding occurrence of piled steel is found. In the steel rolling area of the accident, a control command is generated and sent to the control subnetwork 6 for processing. Among them, the two algorithms are the area cross-border detection algorithm and the arrival time sequence (velocity feature) monitoring algorithm.

The basic idea of the regional cross-border detection algorithm is to determine the normal activity range of the red steel through initial monitoring in advance. As shown in Figure 4, when the stacked steel detection system is in the pre-operation stage, the red steel area of the red steel window collected by the monitoring subnet in real time is A' (not shown in the figure), after a period of time, the accumulated A' is ANDed to obtain the central area A as shown in Figure 4, that is, A=A'∪A'∪A'...., and then Use the structural element D to perform the expansion operation on A to obtain B, that is Thereby setting B is the normal red steel activity area. When the pile steel detection system is transferred from the pre-operation to the operation stage, the area where the real-time collected image A' exceeds B is C, and C=A'-B. The area of C area can be used as an indicator for steel pile monitoring. When C exceeds the threshold T, it can be considered that a pile of steel accident has been sent, so that the first recognition result of the pile of steel phenomenon can be obtained.

Among them, in the region crossing detection algorithm of this embodiment, the threshold T is a key parameter of this algorithm. If T is too large, although it can ensure that the phenomenon of piled steel can be detected, the sensitivity is low and the effect is limited. If T is too small, there is no guarantee that the phenomenon of stacking steel can be detected. Therefore, the steel pile monitoring method also uses the threshold optimization algorithm to self-adapt the threshold selection. In the initial state, first select a relatively rough threshold Ti to ensure that piles of steel can be detected. When a pile of steel is detected, the threshold is updated to a slightly smaller value Ti+1. At this time, both thresholds are detected in parallel, so The cycle repeats until one of the thresholds is missed, and it is considered to be close to the optimal threshold. The system restores the original threshold and enters the normal operation state.

In this embodiment, the red steel image collected in real time is shown in FIG. 6 , and the color block in the center of the black background is the window where the red steel is exposed. The time series (velocity feature) monitoring algorithm can count the red steel area S in the window. When the red steel passes by, the change of the area S of each different exposed window with respect to time t is shown in Figure 5.

In the coordinate system of Fig. 5, the abscissa is time, and the ordinate is the area of exposed red steel, and the three coordinate systems correspond to three different exposure windows. By performing differential operations on this, the time point t _i at which the area S jumps can be obtained, i belongs to {1..n}, and the difference operation is performed on t _i in different windows to obtain the time difference sequence: Ni={t ₂ -t ₁ , t ₃ -t ₂ ，...t _n -t _n-1 }; this time series reflects the dynamic characteristics of red steel. Among them, a single N _i only reflects the observation situation of a certain monitoring node. When all monitoring nodes have extracted N _i , the central controller 5 then integrates the time series characteristics of each monitoring node to obtain the reach time series. The field of view of each monitoring node overlaps (as shown in Figure 6), each monitoring node is sorted by position, and the local time series of each monitoring node is summarized to the central controller 5 to form a long global time series N _g , where, Ng=N ₁ ∪N ₂ ....∪N _i , by judging whether the global time series is normal, the recognition result of the second pile of steel phenomena can be obtained. When there is a difference in one of the time series, the steel rolling area where the pile-up accident occurred can be found according to the red steel position corresponding to the time series, so that the central controller 5 sends the control command to the control subnetwork 6, and the control subnetwork 6 controls external devices for processing.

It should be noted that, in this step, the detection of area crossing detection and arrival time detection can be analyzed by the monitoring subnetwork, and can also be analyzed by the central controller 5, but the recognition results of the two must be comprehensively analyzed by the central controller 5. In order to determine whether the phenomenon of stacking steel occurs, to ensure accuracy. In addition, in Figure 6, each monitoring node is equipped with a high-speed industrial camera 1 and a far-infrared imager 3 to ensure the accuracy of collecting the red steel area.

S204, based on the piled steel phenomenon, control the external equipment to process.

Further, it also includes:

Acquiring the red steel shape feature information of the image to be detected;

Based on the shape feature information of the red steel, the recognition result of the first pile of steel phenomenon and the recognition result of the second pile of steel phenomenon, a machine learning method is used to construct a pile of steel monitoring model to predict the occurrence of a pile of steel accident.

The red steel shape feature information of the image to be detected can be provided by the industrial computer of the monitoring subnet or the industrial computer equipment of the central controller 5 . In a preferred embodiment of the present invention, the industrial computer equipment of the monitoring subnet first performs image processing on the image of the red steel to obtain the change of the angle between the red steel and the horizontal direction, the change of the curvature of the red steel, the change of the center of gravity of the red steel and the exposure of the red steel Change the four morphological features of the area, and send the obtained features to the central controller 5 . These shape feature information and the recognition results of the first pile of steel phenomenon and the recognition result of the second pile of steel phenomenon constitute the learning samples of pile of steel accidents. Using this sample for machine learning can obtain a nonlinear model for steel pile monitoring. By using this model, it is possible to predict in real time whether a steel pile accident will occur according to the shape characteristics, which will greatly advance the early warning time.

In the steel pile monitoring method provided by the second embodiment of the present invention, the system uses two types of image acquisition sensors and three image segmentation algorithms in parallel to monitor the steel pile. Its key technology is the image segmentation and feature extraction of red steel, and the analysis and judgment based on the extracted features. Once a "stack of steel" is found, an alarm will be issued in time, and the corresponding upstream flying shear 8 will be automatically controlled to prevent a large number of red steel from escaping. Cause equipment or other accidents and minimize losses. In addition, the pile steel monitoring system can also automatically save accident images, and use the accident images as sample materials for machine learning, so as to realize the self-evolution of the system and improve the sensitivity and reliability of accident prediction.

It should be noted that the device or module embodiments described above are only illustrative, and the units described as separate components may or may not be physically separated, and the components shown as units may or may not be It is not a physical unit, that is, it can be located in one place, or it can be distributed to multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. In addition, in the drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that they have a communication connection, which can be specifically implemented as one or more communication buses or signal lines. It can be understood and implemented by those skilled in the art without creative effort.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and changes. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims (10)

1. A pile of steel monitoring system, characterized in that, comprises a monitoring subnet, a control subnet and a central controller; The central controller is connected with the monitoring subnet and the control subnet respectively; the monitoring subnet is used to collect red steel images and extract characteristic parameters based on the red steel images, and the central controller according to the The characteristic parameter sent by the monitoring subnet generates a control command and sends the control command to the control subnet; wherein, the monitoring subnet includes a plurality of visible light monitoring nodes and a plurality of infrared monitoring nodes; the The control sub-network is connected with external equipment, and the external equipment is controlled to handle the piled steel accident scene. 2. The pile steel monitoring system according to claim 1, characterized in that, the monitoring subnet and the central controller are connected through a communication protocol of TCP/IP. 3. The pile steel monitoring system according to claim 2, characterized in that, the control subnet and the central controller are communicated through the Ethercat industrial Ethernet through a transmission protocol. 4. The pile steel monitoring system according to claim 3, wherein the central controller is a host computer, and each visible light monitoring node is provided with a first lower computer and a high-speed industrial camera . 5. The stack steel monitoring system according to claim 4, wherein the central controller is a host computer, and each infrared monitoring node is provided with a second slave computer and a far-infrared imager. 6. The stack steel monitoring system according to claim 5, characterized in that, the first lower computer and the second lower computer are industrial computers. 7. The pile steel monitoring system according to claim 6, characterized in that, the field devices include warning lights and flying shears. 8. A pile of steel monitoring method is characterized in that, the method uses the pile of steel monitoring system as claimed in claim 7 to carry out pile of steel monitoring, and its steps include: Obtain the red steel image; wherein, the red steel image includes the visible light image collected by the high-speed industrial camera and the infrared image collected by the far-infrared imager; Image processing is performed on the red steel image to obtain an image to be detected; wherein, the image processing includes performing region-of-interest division and image segmentation processing on the red steel image; Carrying out steel stacking phenomenon recognition based on the image to be detected, to determine the steel rolling area corresponding to the image to be detected corresponding to the steel stacking accident; Based on the phenomenon of piled steel, the external equipment is controlled to handle the piled steel accident on-site. 9. The pile-up steel monitoring method according to claim 8, characterized in that, carrying out pile-up steel phenomenon recognition based on the image to be detected, to determine that the step of the steel-rolling area corresponding to the pile-up steel accident in the image to be detected comprises: Performing area cross-border detection on the image to be detected to obtain the first pile of steel phenomenon recognition results; Carrying out time-of-arrival detection on the image to be detected to obtain a recognition result of the second pile of steel phenomena; Based on the recognition result of the first pile of steel phenomenon and the recognition result of the second pile of steel phenomenon, it is determined that the steel rolling area where the steel pile accident occurs corresponds to the image to be detected. 10. pile steel monitoring method according to claim 9, is characterized in that, also comprises: Acquiring the red steel shape feature information of the image to be detected; Based on the shape feature information of the red steel, the recognition result of the first pile of steel phenomenon and the recognition result of the second pile of steel phenomenon, a machine learning method is used to construct a pile of steel monitoring model to predict the occurrence of a pile of steel accident.