Flame cutting control method and device for steel billet cutting

Abstract

The invention provides a flame cutting control method and device for steel billet cutting, and relates to the technical field of metallurgy control. The method comprises the following steps: receiving a video image shot by a camera and used for cutting a billet by a flame cutting machine; the video image is recognized based on a first video AI model, and if it is determined that the first recognition result is that the steel billet is cut off, an execution mechanism closing instruction is sent to a PLC control system, so that the PLC control system forwards the mechanism closing instruction to an execution mechanism; and receiving closing state information fed back by the PLC control system after the execution structure executes the mechanism closing instruction, and completing flame cutting control of steel billet cutting this time. The device executes the method. According to the method and the device provided by the embodiment of the invention, manual field operation is not needed, and the invalid combustion time of the flame cutting machine can be shortened, so that the fuel gas is saved, and the service life of the flame cutting machine is prolonged.

Description

Technical Field

[0001] The present invention relates to the technical field of metallurgical control, and particularly relates to a method and device for controlling flame cutting of steel billets. Background Art

[0002] The main idea of continuous casting production is to directly cast molten steel into formed products.

[0003] After the steel billet is drawn out in continuous casting production, the first area it passes through is the secondary cooling channel. In the secondary cooling channel, cooling water is sprayed onto the steel billet, and the steel billet will gradually cool from the outside to the center and enter the straightening machine along the roller path. The function of the straightening machine is to straighten the continuous casting billet to facilitate the next process. Behind the straightening machine is the cutting machine, and a flame cutting machine that moves forward synchronously with the steel billet is used to cut the steel billet.

[0004] The flame cutting machine is a device used to cut the cast billet into a specified length according to requirements and to take the sulfur print of the cross-section of the cast billet. Its electric drive control is composed of three parts: the traveling of the flame cutting vehicle, the movement and lifting of the cutting gun. In the process flow of producing various rod and wire products, due to the cutting site being in a high-temperature and high-brightness environment, after cutting is completed, it usually relies on manual observation, resulting in low accuracy.

[0005] The prior art combines an industrial touch screen with a PLC, which not only expands the functions of the PLC, enables it to have a graphical and interactive working interface, and realizes real-time monitoring of the cutting machine. The continuous casting machine cuts the steel billet, relying on PLC control. After the parameters are set, the flame spraying and cutting time is generally determined by a fixed flow rate and distance.

[0006] The selection of the detection point affects the collected data. Therefore, the PLC system cannot accurately judge whether the steel billet is really cut off. During the process, it is necessary to rely on manual observation to perform a "supplementary cutting" operation when it is found that the billet is not cut off. When the steel billet has not reached the detection point during cutting, if the steel billet has been cut off, but the gas still needs to continue burning, resulting in serious gas waste. At the same time as the ineffective combustion, it will also reduce the actual service life of the flame gun head. In addition, the high-temperature and high-brightness environment at the site poses a safety hazard to manual operations. Summary of the Invention

[0007] Aiming at the problems in the prior art, the embodiments of the present invention provide a method and device for controlling flame cutting of steel billets, which can at least partially solve the problems existing in the prior art.

[0008] On the one hand, the present invention proposes a method for controlling flame cutting of steel billets, and the method for controlling flame cutting of steel billets is applied to a video analysis server; the method for controlling flame cutting of steel billets includes:

[0009] Receive the video image of the billet being cut by the flame cutting machine taken by the camera;

[0010] Based on the first video AI model, identify the video image. If it is determined that the first identification result is that the billet has been cut off, send an actuator closing instruction to the PLC control system, so that the PLC control system forwards the actuator closing instruction to the actuator structure;

[0011] Among them, the first video AI model is pre-trained by LSTM according to the first set of cut pictures samples marked with the corresponding shapes of various billets;

[0012] Receive the closing state information of the actuator structure feedback by the PLC control system after the actuator structure executes the actuator closing instruction, and complete the flame cutting control of the current billet cutting.

[0013] Among them, the actuator closing instruction carries the actuator identification and the actuator closing sequence information; correspondingly, sending the actuator closing instruction to the PLC control system includes:

[0014] Send the actuator closing instruction to the PLC control system, so that the PLC control system forwards the actuator closing instruction to the actuator structure, identify each actuator according to the actuator identification, and control each actuator to close orderly according to the actuator closing sequence information.

[0015] Among them, the flame cutting control method for billet cutting further includes:

[0016] If it is determined that the first identification result is that the billet has been cut off, send the billet cutting statistical record information corresponding to the current billet batch to the application management platform; the application management platform contains the billet batch production plan information.

[0017] Among them, the flame cutting control method for billet cutting further includes:

[0018] Before the step of receiving the video image of the billet being cut by the flame cutting machine taken by the camera, record the cutting control parameters of the flame cutting machine and the billet shape information of the current billet batch;

[0019] After the step of completing the flame cutting control of the current billet cutting, evaluate the quality of the current billet cutting, optimize and adjust the cutting control parameters according to the billet cutting quality evaluation result and the billet shape information, and establish the mapping relationship between the optimized and adjusted cutting control parameters and the billet shape information of the current billet batch.

[0020] Among them, evaluating the quality of the current billet cutting includes:

[0021] Identify the cross-sectional image of the cut billet based on the second video AI model, and determine the cutting seam parameters according to the second recognition result;

[0022] Among them, the second video AI model is pre-trained by a convolutional neural network with an attention mechanism according to a second set of cut image samples marked with various billet shapes respectively;

[0023] Evaluate the cutting quality of the current billet according to the cutting seam parameters.

[0024] Among them, before the step of receiving the video image of the billet cut by the flame cutting machine taken by the camera, the flame cutting control method for billet cutting further includes:

[0025] Obtain the billet shape information of the current billet batch, and determine the optimized cutting control parameters corresponding to the billet shape information according to the mapping relationship, and adjust the flame cutting machine according to the optimized cutting control parameters.

[0026] On the one hand, the present invention proposes a flame cutting control device for billet cutting, and the flame cutting control device for billet cutting is applied to a video analysis server; the flame cutting control device for billet cutting includes:

[0027] A receiving unit for receiving the video image of the billet cut by the flame cutting machine taken by the camera;

[0028] An identification unit for identifying the video image based on the first video AI model. If it is determined that the first recognition result is that the billet has been cut off, an execution mechanism closing instruction is sent to the PLC control system, so that the PLC control system forwards the mechanism closing instruction to the execution structure;

[0029] Among them, the first video AI model is pre-trained by LSTM according to a first set of cut image samples marked with various billet shapes respectively;

[0030] A control unit for receiving the closing state information of the execution structure after executing the mechanism closing instruction fed back by the PLC control system to complete the flame cutting control of the current billet cutting.

[0031] On the other hand, an embodiment of the present invention provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the following method is implemented:

[0032] Receive the video image of the billet cut by the flame cutting machine taken by the camera;

[0033] Based on the first video AI model, identify the video image. If it is determined that the first recognition result is that the billet has been cut, send an actuator closing instruction to the PLC control system, so that the PLC control system forwards the actuator closing instruction to the actuator structure;

[0034] Among them, the first video AI model is pre-trained by LSTM according to the first set of post-cut picture samples marked with the corresponding shapes of various billets;

[0035] Receive the closing state information of the actuator structure after executing the actuator closing instruction feedback by the PLC control system, and complete the flame cutting control of the current billet cutting.

[0036] An embodiment of the present invention provides a computer-readable storage medium, including:

[0037] The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the following method is implemented:

[0038] Receive the video image of the billet cut by the flame cutting machine taken by the camera;

[0039] Based on the first video AI model, identify the video image. If it is determined that the first recognition result is that the billet has been cut, send an actuator closing instruction to the PLC control system, so that the PLC control system forwards the actuator closing instruction to the actuator structure;

[0040] Among them, the first video AI model is pre-trained by LSTM according to the first set of post-cut picture samples marked with the corresponding shapes of various billets;

[0041] Receive the closing state information of the actuator structure after executing the actuator closing instruction feedback by the PLC control system, and complete the flame cutting control of the current billet cutting.

[0042] An embodiment of the present invention further provides a computer program product, the computer program product includes a computer program, and when the computer program is executed by a processor, the following method is implemented:

[0043] Receive the video image of the billet cut by the flame cutting machine taken by the camera;

[0044] Based on the first video AI model, identify the video image. If it is determined that the first recognition result is that the billet has been cut, send an actuator closing instruction to the PLC control system, so that the PLC control system forwards the actuator closing instruction to the actuator structure;

[0045] Among them, the first video AI model is pre-trained by LSTM according to a first set of post-cut picture samples marked with various billet shapes respectively;

[0046] Receive the closing state information of the execution structure after executing the mechanism closing instruction fed back by the PLC control system, and complete the flame cutting control of the current billet cutting.

[0047] The flame cutting control method and device for billet cutting provided by the embodiments of the present invention receive the video image of the billet being cut by the flame cutting machine captured by the camera; identify the video image based on the first video AI model, and if it is determined that the first recognition result is that the billet has been cut off, send an execution mechanism closing instruction to the PLC control system, so that the PLC control system forwards the mechanism closing instruction to the execution structure; among them, the first video AI model is pre-trained by LSTM according to a first set of post-cut picture samples marked with various billet shapes respectively; receive the closing state information of the execution structure after executing the mechanism closing instruction fed back by the PLC control system, and complete the flame cutting control of the current billet cutting, without the need for manual on-site operation, and can also shorten the ineffective combustion time of the flame cutting machine, thereby saving gas usage and increasing the service life of the flame cutting machine. Description of the Drawings

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings. In the drawings:

[0049] Figure 1 is a schematic flowchart of a flame cutting control method for billet cutting provided by an embodiment of the present invention.

[0050] Figure 2 is a schematic structural diagram of a video analysis system architecture provided by an embodiment of the present invention.

[0051] Figure 3 is a schematic flowchart of a flame cutting control method for billet cutting provided by another embodiment of the present invention.

[0052] Figure 4 is a schematic structural diagram of a flame cutting control device for billet cutting provided by an embodiment of the present invention.

[0053] Figure 5 is a schematic diagram of the physical structure of a computer device provided by an embodiment of the present invention. Detailed Embodiments

[0054] To make the objectives, technical solutions and advantages of the embodiments of the present invention more clearly understood, the following further describes in detail the embodiments of the present invention with reference to the accompanying drawings. Herein, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but not to limit the present invention. It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments can be arbitrarily combined with each other.

[0055] Figure 1 is a schematic flowchart of a flame cutting control method for billet cutting provided by an embodiment of the present invention. As Figure 1 shown, the flame cutting control method for billet cutting provided by the embodiment of the present invention is applied to a video analysis server; the flame cutting control method for billet cutting includes:

[0056] Step S1: Receive the video image of a billet being cut by a flame cutting machine captured by a camera.

[0057] Step S2: Identify the video image based on a first video AI model. If it is determined that the first recognition result is that the billet has been cut off, send an actuator closing instruction to the PLC control system, so that the PLC control system forwards the mechanism closing instruction to the actuator;

[0058] Among them, the first video AI model is pre-trained by LSTM according to a first set of cut image samples respectively marked with various billet shapes.

[0059] Step S3: Receive the closing status information of the actuator after the actuator executes the mechanism closing instruction fed back by the PLC control system, and complete the flame cutting control of the current billet cutting.

[0060] In the above step S1, the device receives the video image of a billet being cut by a flame cutting machine captured by a camera. The device can be a computer device that executes this method. In the technical solution of the present application, the acquisition, storage, use, processing, etc. of data all comply with relevant regulations. First, the video analysis system architecture for executing the method of the present invention is described as follows:

[0061] As Figure 2 shown, the video analysis system architecture includes 1 camera, 1 switch, 1 NVR (i.e., network video recorder), 1 video analysis server, 1 PLC controller (i.e., PLC control system) and an application management platform ( Figure 2 not shown), and the camera can be selected as a high-temperature and high-brightness camera.

[0062] NVR: NVR is short for Network Video Recorder, and its main function is the storage and forwarding of video streams. NVR usually works in collaboration with video encoders (DVS) or network cameras (IPC) to complete video recording, storage, and forwarding functions. The computing power of NVR is mainly used for the reception, storage, and management of video data. It is suitable for small-scale monitoring systems, with a limited number of cameras to process, and its main functions are concentrated on video storage and basic management.

[0063] Video analysis server: The video analysis server focuses on the intelligent analysis and processing of video content. It can run artificial intelligence algorithms to perform advanced functions such as behavior analysis and target recognition, and support complex data analysis and system management tasks. The video analysis server has powerful computing power, can process large-scale and high-concurrency video streams, and supports a variety of complex application scenarios.

[0064] One high-temperature and high-brightness camera can be set near the operating range of the flame cutting machine. This high-temperature and high-brightness camera can capture the video image of the flame cutting machine cutting the steel billet, and then this high-temperature and high-brightness camera sends the above video image to the video analysis server.

[0065] The main functions of the switch include signal processing, optimizing network management, and improving data exchange efficiency.

[0066] In the above step S2, the device identifies the video image based on the first video AI model. If it is determined that the first recognition result is that the steel billet has been cut off, an execution mechanism closing instruction is sent to the PLC control system, so that the PLC control system forwards the mechanism closing instruction to the execution structure;

[0067] Among them, the first video AI model is pre-trained by LSTM according to the first set of cut steel billet image samples marked with various corresponding steel billet shapes. The video image can be input into the first video AI model, and the output result of the first video AI model is used as the first recognition result reflecting whether the steel billet has been cut off. Figure 3 As shown, if it is determined that the first recognition result is that the steel billet has not been cut off, the flame cutting machine is maintained to continue executing according to the original program, and the video image is continuously identified through the first video AI model, and subsequent steps are executed until the end of this operation, that is, the flame cutting control of this steel billet cutting is completed. The execution mechanism can include a cutting oxygen valve, a preheating oxygen valve, and a gas valve.

[0068] Various steel billet shapes can include square billets, slab billets, and rectangular billets, etc., which are briefly described as follows:

[0069] Square billet: The cross-section width and height are equal or have little difference, and it is mainly used for rolling section steel and wire rods.

[0070] Slab billet: The ratio of the cross-section width to the height is relatively large, and it is mainly used for rolling plates.

[0071] Rectangular billet: The width and height of the cross-section are unequal and the ratio is not large. It is mainly used for rolling hot-rolled strip steel, deformed steel bars for construction, common wire rods, high-speed wire rods and various small-sized sections.

[0072] Taking the square billet as an example, the first set of cut image samples is a set of image samples composed of a continuous preset number of frame images before and after the cutting moment. The continuous preset number values before and after the cutting moment can be equal or unequal. Taking the equal preset number value of 3 frame images as an example, the set of image samples includes 3 frame images before the cutting moment, 3 frame images after the cutting moment, and the frame image at the cutting moment. These 7 frame images are arranged in sequence according to the time series to form the first set of cut image samples for training the LSTM.

[0073] It can be understood that in order to ensure that the LSTM fully learns the image features before, during and after the cutting action, as well as the running trajectories of the flame cutting machine and the billet to be cut, as many image samples of square billets from different production batches before and after cutting as possible should be selected as the first set of cut image samples.

[0074] The above-mentioned temporally continuous frame images refer to a group of images arranged in chronological order. Such data not only contains the spatial information of each frame image, but also has the temporal front-back correlation. For example, for the movement of an object in a video, there will be a certain coherence in the position and state of the object between the previous frame and the next frame. When processing such data, the model needs to simultaneously understand the spatial features and the change patterns in time.

[0075] The LSTM effectively controls the flow of information by introducing gating mechanisms (input gate, forget gate, output gate) and cell state, and can capture long-term dependencies.

[0076] In the above step S3, the device receives the closing state information of the execution structure after executing the mechanism closing instruction fed back by the PLC control system, and completes the flame cutting control of the current billet cutting. The mechanism closing instruction carries the execution mechanism identifier and the execution mechanism closing sequence information; correspondingly, sending the mechanism closing instruction to the PLC control system includes:

[0077] Sending a mechanism closing instruction to the PLC control system, so that the PLC control system forwards the mechanism closing instruction to the execution structure, identifies each execution mechanism according to the execution mechanism identifier, and controls each execution mechanism to close orderly according to the execution mechanism closing sequence information. Referring to the above description, the cutting oxygen valve can be closed first. After stopping the cutting, the cutting oxygen valve should be closed immediately to terminate the cutting oxygen flow. The closing state information of the cutting oxygen valve is closed after it is confirmed to be closed.

[0078] Then close the acetylene (fuel gas) valve. After closing the cutting oxygen, it is necessary to close the acetylene valve to cut off the fuel gas supply and prevent potential safety hazards caused by residual fuel gas. The closing status information of the fuel gas valve is "closed" after it is confirmed to be closed.

[0079] Finally, close the preheating oxygen valve. After the flame is completely extinguished, close the preheating oxygen valve to ensure that the entire gas circuit system stops working. The closing status information of the preheating oxygen valve is "closed" after it is confirmed to be closed. As Figure 3 shown, the signals of the closing status of the above three valves are fed back to the PLC control system. The PLC control system feeds back the signals of the closing status to the video analysis server, and thus a billet cutting process is completed.

[0080] Closing each actuator in a reasonable order can further ensure the safety of billet production.

[0081] The flame cutting control method for billet cutting further includes:

[0082] If it is determined that the first recognition result is that the billet has been cut off, send the billet cutting statistical record information corresponding to the current billet batch to the application management platform; the application management platform contains billet batch production plan information. As Figure 3 shown, while sending the actuator closing instruction to the PLC control system, the billet cutting statistical record information corresponding to the current billet batch can also be sent to the application management platform. The billet cutting statistical record information can include the current billet batch number, the number of billet cuts in the current billet batch, the length of each cut billet, the cutting completion time, etc.

[0083] The application management platform can summarize and manage the billet cutting statistical record information. By comparing it with the billet batch production plan information regularly or irregularly, it helps relevant personnel accurately and real-time determine the completion progress of billet production, such as the task percentage of completing the overall billet batch production plan.

[0084] The flame cutting control method for billet cutting further includes:

[0085] Before the step of receiving the video image of the billet being cut by the flame cutting machine taken by the camera, record the cutting control parameters of the flame cutting machine and the billet shape information of the current billet batch; the cutting control parameters can include cutting speed, oxygen pressure, preheating flame parameters, etc. The billet shape information can refer to the above description and will not be elaborated here.

[0086] After the step of controlling the flame cutting for completing the current billet cutting, evaluate the quality of the current billet cutting. Optimize and adjust the cutting control parameters according to the billet cutting quality evaluation result and the billet profile information, and establish a mapping relationship between the optimized and adjusted cutting control parameters and the billet profile information of the current billet batch. The evaluation of the quality of the current billet cutting includes:

[0087] Based on the second video AI model, identify the cross-sectional image of the cut billet, and determine the cut seam parameters according to the second identification result. One more camera can be specifically set up to take close-up photos of the cross-section of the cut billet. Since the main purpose of billet cross-section identification is to identify the detailed features of the cut seam on the billet cross-section and the continuity of the image frame time sequence does not need to be considered, the above-mentioned specifically set up one camera can be selected according to the requirement of the highest image clarity to more clearly and comprehensively reflect the cut seam details. The cut seam parameters can include the cut seam width, the surface roughness of the cut seam, the thickness of the slag hanging on the cut, and the cutting stripes, etc., which are described separately as follows:

[0088] Cut seam width: The cut seam width is one of the important indicators for evaluating the cutting quality. The smaller the cut seam width, the better the cutting quality. The cut seam width is related to the diameter of the flame beam spot and the beam mode. For the numerical control flame cutting of metals, the smaller the cut seam width, the better the cutting quality.

[0089] Surface roughness of the cut seam: The surface roughness is also a key indicator for evaluating the cutting quality. There are many parameters for surface roughness. For general numerical control flame cutting machines, the main parameters selected are the ten-point height Rz of the surface micro-irregularities or the arithmetic mean deviation.

[0090] Thickness of the slag hanging on the cut: The thickness of the slag hanging directly reflects the quality of the numerical control cutting machine. Ideal flame cutting should obtain a cut seam without slag hanging.

[0091] Cutting stripes: Cutting stripes will be generated on the surface of the cut seam. The finer the stripes and the shallower the depth, the better the cutting quality.

[0092] The cross-sectional image of the billet can be input into the second video AI model, and the output result of the second video AI model is used as the second identification result.

[0093] Among them, the second video AI model is pre-trained by integrating a convolutional neural network with an attention mechanism according to a second set of cut billet picture samples marked with corresponding billet profiles respectively. Taking square billets as an example, the second set of cut billet picture samples is a set of picture samples composed of images that can more fully reflect the detailed features of the cut seam, so that the convolutional neural network integrated with the attention mechanism can fully and comprehensively learn the cut seam details.

[0094] It can be understood that in order to ensure that the convolutional neural network integrated with the attention mechanism can fully learn the slit detail image features, as many picture samples as possible that can reflect more slit detail images should be selected as the second set of cut pictures samples.

[0095] The attention mechanism dynamically adjusts the feature weights of different regions, enabling the network to automatically focus on the key details in the image and be more sensitive to subtle features (such as textures and edges); the attention mechanism can integrate feature maps at different levels, for example, combining high-resolution details (such as edges) in the shallow layer with semantic information (such as object contours) in the deep layer. This fusion method can not only retain the local fine structure but also utilize the global context to optimize detail discrimination.

[0096] Evaluate the quality of the current billet cutting according to the slit parameters. Referring to the description of the above slit parameters, different weights can be set for various slit parameters respectively, and the quality of the current billet cutting can be evaluated by calculating the weighted scores of different weights.

[0097] Through the second video AI model, the evaluation work of the quality of the current billet cutting can be accurately and efficiently achieved without manual participation.

[0098] Optimize and adjust the cutting control parameters according to the billet cutting quality evaluation result and the billet shape information, as described below:

[0099] Referring to the above discussion, if the weighted score used to reflect the billet cutting quality evaluation result is lower than the preset score (i.e., the billet cutting quality is poor), then the cutting control parameters need to be optimized and adjusted.

[0100] Since the billet shape information is relatively complex, for the sake of brief description, the method of the present invention only takes the billet thickness as the billet shape information for description:

[0101] If the billet thickness is between 30 - 50 mm, then adjust the oxygen pressure so that the adjusted oxygen pressure is between 0.8 - 1.0 MPa.

[0102] If the billet thickness is between 50 - 100 mm, then adjust the oxygen pressure so that the adjusted oxygen pressure is between 0.8 - 1.0 MPa, and also adjust the preset flame parameters so that the adjusted preset flame parameters are a neutral flame (the ratio of acetylene to oxygen is 1:1.15) and the flame core length is 2 - 3 cm.

[0103] If the billet thickness is between 100 - 150 mm, then adjust the preset flame parameters so that the adjusted preset flame parameters are a neutral flame (the ratio of acetylene to oxygen is 1:1.15) and the flame core length is 2 - 3 cm.

[0104] If the thickness of the billet is more than 150 mm, the cutting speed is adjusted. Specifically, a segmented speed reduction strategy is adopted. The speed in the initial segment is reduced by 30%, and the reference value is restored after stabilization. The reference value of the cutting speed can be selected with reference to the existing conventional technology.

[0105] Before the step of receiving the video image of the billet cut by the flame cutting machine captured by the camera, the flame cutting control method for billet cutting further includes:

[0106] Obtain the billet shape information of the current billet batch, and determine the optimized cutting control parameters corresponding to the billet shape information according to the mapping relationship, and adjust the flame cutting machine according to the optimized cutting control parameters. The video image captured by the camera thereafter is the video image of the flame cutting machine cutting the billet according to the optimized cutting control parameters.

[0107] By optimizing the cutting control parameters, it helps to improve the production efficiency and product quality of steel products. Using the flame cutting control method for billet cutting provided by the embodiments of the present invention can save about 5 seconds of gas spraying time during each billet cutting.

[0108] The flame cutting control method for billet cutting provided by the embodiments of the present invention receives the video image of the billet cut by the flame cutting machine captured by the camera; based on the first video AI model, the video image is identified. If it is determined that the first recognition result is that the billet has been cut off, an actuator closing instruction is sent to the PLC control system, so that the PLC control system forwards the mechanism closing instruction to the execution structure; wherein, the first video AI model is pre-trained by LSTM according to the first set of cut pictures samples respectively marked with various billet shapes; receive the closing state information of the execution structure after the execution structure executes the mechanism closing instruction fed back by the PLC control system, and complete the flame cutting control of the current billet cutting. There is no need for manual on-site operation, and it can also shorten the ineffective combustion time of the flame cutting machine, thereby saving gas usage and increasing the service life of the flame cutting machine.

[0109] Further, the actuator closing instruction carries actuator identification and actuator closing sequence information; correspondingly, sending the actuator closing instruction to the PLC control system includes:

[0110] Send an actuator closing instruction to the PLC control system, so that the PLC control system forwards the mechanism closing instruction to the execution structure, identify each execution structure according to the actuator identification, and control each execution structure to close orderly according to the actuator closing sequence information. It can be described with reference to the above embodiments and will not be repeated.

[0111] Further, the flame cutting control method for billet cutting further includes:

[0112] If it is determined that the first recognition result is that the billet has been cut, send the billet cutting statistical record information corresponding to the current billet batch to the application management platform; the application management platform includes billet batch production plan information. The description can refer to the above embodiments and will not be repeated.

[0113] Further, the flame cutting control method for billet cutting further includes:

[0114] Before the step of receiving the video image of the billet being cut by the flame cutting machine captured by the camera, record the cutting control parameters of the flame cutting machine and the billet profile information of the current billet batch. The description can refer to the above embodiments and will not be repeated.

[0115] After the step of completing the flame cutting control of the current billet cutting, evaluate the quality of the current billet cutting. According to the billet cutting quality evaluation result and the billet profile information, optimize and adjust the cutting control parameters, and establish a mapping relationship between the optimized and adjusted cutting control parameters and the billet profile information of the current billet batch. The description can refer to the above embodiments and will not be repeated.

[0116] Further, the evaluation of the quality of the current billet cutting includes:

[0117] Based on the second video AI model, identify the cross-section image of the cut billet, and determine the cutting seam parameters according to the second recognition result. The description can refer to the above embodiments and will not be repeated.

[0118] Among them, the second video AI model is pre-trained by a convolutional neural network with an attention mechanism according to a second set of cut billet picture samples marked with various billet profiles respectively. The description can refer to the above embodiments and will not be repeated.

[0119] Evaluate the quality of the current billet cutting according to the cutting seam parameters. The description can refer to the above embodiments and will not be repeated.

[0120] Further, before the step of receiving the video image of the billet being cut by the flame cutting machine captured by the camera, the flame cutting control method for billet cutting further includes:

[0121] Obtain the billet profile information of the current billet batch, and determine the optimized cutting control parameters corresponding to the billet profile information according to the mapping relationship, and adjust the flame cutting machine according to the optimized cutting control parameters. The description can refer to the above embodiments and will not be repeated.

[0122] Figure 4 is a schematic structural diagram of a flame cutting control device for billet cutting provided by an embodiment of the present invention, as Figure 4As shown in the figure, the flame cutting control device for billet cutting provided by the embodiment of the present invention is applied to a video analysis server; the flame cutting control device for billet cutting includes a receiving unit 401, an identifying unit 402 and a control unit 403, where:

[0123] The receiving unit 401 is configured to receive a video image of a billet being cut by a flame cutting machine captured by a camera; the identifying unit 402 is configured to identify the video image based on a first video AI model. If it is determined that the first identification result is that the billet has been cut off, an actuator closing instruction is sent to the PLC control system, so that the PLC control system forwards the mechanism closing instruction to the actuator structure; wherein, the first video AI model is pre-trained by LSTM according to a first set of post-cut image samples respectively marked with various billet shapes; the control unit 403 is configured to receive the closing state information of the actuator structure after executing the mechanism closing instruction fed back by the PLC control system, and complete the flame cutting control of the current billet cutting.

[0124] Specifically, the receiving unit 401 in the device is configured to receive a video image of a billet being cut by a flame cutting machine captured by a camera; the identifying unit 402 is configured to identify the video image based on a first video AI model. If it is determined that the first identification result is that the billet has been cut off, an actuator closing instruction is sent to the PLC control system, so that the PLC control system forwards the mechanism closing instruction to the actuator structure; wherein, the first video AI model is pre-trained by LSTM according to a first set of post-cut image samples respectively marked with various billet shapes; the control unit 403 is configured to receive the closing state information of the actuator structure after executing the mechanism closing instruction fed back by the PLC control system, and complete the flame cutting control of the current billet cutting.

[0125] The flame cutting control device for billet cutting provided by the embodiment of the present invention receives a video image of a billet being cut by a flame cutting machine captured by a camera; identifies the video image based on a first video AI model. If it is determined that the first identification result is that the billet has been cut off, an actuator closing instruction is sent to the PLC control system, so that the PLC control system forwards the mechanism closing instruction to the actuator structure; wherein, the first video AI model is pre-trained by LSTM according to a first set of post-cut image samples respectively marked with various billet shapes; receives the closing state information of the actuator structure after executing the mechanism closing instruction fed back by the PLC control system, and completes the flame cutting control of the current billet cutting, without the need for manual on-site operation, and can also shorten the ineffective combustion time of the flame cutting machine, thereby saving gas usage and increasing the service life of the flame cutting machine.

[0126] The embodiments of the flame cutting control device for billet cutting provided by the embodiments of the present invention can specifically be used to execute the processing flows of the above-mentioned method embodiments, and their functions will not be elaborated here. For details, reference can be made to the detailed descriptions of the above-mentioned method embodiments.

[0127] Figure 5 It is a schematic physical structure diagram of a computer device provided by an embodiment of the present invention. As Figure 5 shown, the computer device includes: a memory 501, a processor 502, and a computer program stored on the memory 501 and executable on the processor 502. When the processor 502 executes the computer program, the following method is implemented:

[0128] Receive the video image of the billet being cut by the flame cutting machine captured by the camera;

[0129] Based on the first video AI model, identify the video image. If it is determined that the first recognition result is that the billet has been cut off, send an execution mechanism closing instruction to the PLC control system, so that the PLC control system forwards the mechanism closing instruction to the execution structure;

[0130] Among them, the first video AI model is pre-trained by LSTM according to the first set of cut billet picture samples marked with various billet shapes respectively;

[0131] Receive the closing state information of the execution structure after executing the mechanism closing instruction fed back by the PLC control system, and complete the flame cutting control of the current billet cutting.

[0132] This embodiment discloses a computer program product, which includes a computer program. When the computer program is executed by a processor, the following method is implemented:

[0133] Receive the video image of the billet being cut by the flame cutting machine captured by the camera;

[0134] Based on the first video AI model, identify the video image. If it is determined that the first recognition result is that the billet has been cut off, send an execution mechanism closing instruction to the PLC control system, so that the PLC control system forwards the mechanism closing instruction to the execution structure;

[0135] Among them, the first video AI model is pre-trained by LSTM according to the first set of cut billet picture samples marked with various billet shapes respectively;

[0136] Receive the closing state information of the execution structure after executing the mechanism closing instruction fed back by the PLC control system, and complete the flame cutting control of the current billet cutting.

[0137] This embodiment provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the following method is implemented:

[0138] Receive a video image of a steel billet being cut by a flame cutting machine captured by a camera;

[0139] Based on a first video AI model, identify the video image. If it is determined that the first recognition result is that the steel billet has been cut off, send an actuator closing instruction to the PLC control system, so that the PLC control system forwards the mechanism closing instruction to the actuator;

[0140] Wherein, the first video AI model is pre-trained by an LSTM according to a first set of post-cut picture samples marked with corresponding shapes of various steel billets;

[0141] Receive the closing state information of the actuator after executing the mechanism closing instruction feedback by the PLC control system, and complete the flame cutting control of the current steel billet cutting.

[0142] Compared with the technical solutions in the prior art, the flame cutting control method for steel billet cutting provided by the embodiments of the present invention receives a video image of a steel billet being cut by a flame cutting machine captured by a camera; based on a first video AI model, identify the video image. If it is determined that the first recognition result is that the steel billet has been cut off, send an actuator closing instruction to the PLC control system, so that the PLC control system forwards the mechanism closing instruction to the actuator; wherein, the first video AI model is pre-trained by an LSTM according to a first set of post-cut picture samples marked with corresponding shapes of various steel billets; receive the closing state information of the actuator after executing the mechanism closing instruction feedback by the PLC control system, and complete the flame cutting control of the current steel billet cutting, without the need for manual on-site operation, and can also shorten the ineffective combustion time of the flame cutting machine, thereby saving gas usage and increasing the service life of the flame cutting machine.

[0143] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0144] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or means for implementing the functions specified in one block or multiple blocks.

[0145] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that implement the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or means for implementing the functions specified in one block or multiple blocks.

[0146] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or means for implementing the functions specified in one block or multiple blocks.

[0147] In the description of this specification, the description with reference to terms such as "one embodiment", "a specific embodiment", "some embodiments", "for example", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples.

[0148] The above specific embodiments further elaborate on the objectives, technical solutions, and beneficial effects of the present invention. It should be understood that the above are only specific embodiments of the present invention and are not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A flame cutting control method for steel billet cutting, characterized in that: The flame cutting control method for steel billet cutting is applied to a video analysis server; the flame cutting control method for steel billet cutting comprises: receiving a video image captured by a camera of a steel billet being cut by a flame cutting machine; Recognize the video image based on the first video AI model, and if it is determined that the first recognition result is that the steel billet has been cut, send an actuator closing instruction to the PLC control system, so that the PLC control system forwards the actuator closing instruction to the execution structure; The first video AI model is obtained by training the LSTM in advance based on a first cut picture sample set labeled with respective corresponding shapes of various steel billets; The closing state information of the execution structure after executing the mechanism closing instruction is received from the PLC control system to complete the flame cutting control of the current steel billet cutting.

2. The flame cutting control method for steel billet cutting according to claim 1, characterized in that: The actuator closing instruction carries the actuator identification and actuator closing sequence information; accordingly, sending the actuator closing instruction to the PLC control system includes: Send an actuator closing instruction to the PLC control system so that the PLC control system forwards the mechanism closing instruction to the execution structure, identifies each actuator according to the actuator identifier, and controls each actuator to close in an orderly manner according to the actuator closing sequence information.

3. The flame cutting control method for steel billet cutting according to claim 1, characterized in that: The flame cutting control method for steel billet cutting also includes: If it is determined that the first recognition result is that the steel billet has been cut, the steel billet cutting statistical record information corresponding to the current steel billet batch is sent to the application management platform; the application management platform includes the steel billet batch production plan information.

4. The flame cutting control method for steel billet cutting according to claim 1, characterized in that: The flame cutting control method for steel billet cutting also includes: Before the step of receiving the video image of the steel billet being cut by the flame cutting machine taken by the camera, recording the cutting control parameters of the flame cutting machine and the steel billet shape information of the current steel billet batch; After completing the step of flame cutting control for the current billet cutting, the billet cutting quality is evaluated, the cutting control parameters are optimized and adjusted according to the billet cutting quality evaluation result and the billet shape information, and a mapping relationship is established between the optimized and adjusted cutting control parameters and the billet shape information of the current billet batch.

5. The flame cutting control method for steel billet cutting according to claim 4, characterized in that: The evaluation of the billet cutting quality includes: Recognize the cross-sectional image of the cut steel billet based on the second video AI model, and determine the cutting parameters according to the second recognition result; The second video AI model is obtained by training a convolutional neural network with an attention mechanism in advance according to a second cut picture sample set labeled with respective corresponding shapes of various steel billets; The cutting quality of the steel billet is evaluated according to the cutting seam parameters.

6. The flame cutting control method for steel billet cutting according to claim 4, characterized in that: Before the step of receiving the video image of the steel billet being cut by the flame cutting machine taken by the camera, the flame cutting control method for steel billet cutting further includes: The steel billet shape information of the current steel billet batch is obtained, and the optimized cutting control parameters corresponding to the steel billet shape information are determined by the mapping relationship, and the flame cutting machine is adjusted according to the optimized cutting control parameters.

7. A flame cutting control device for billet cutting, characterized in that: The flame cutting control device for steel billet cutting is applied to a video analysis server; the flame cutting control device for steel billet cutting comprises: A receiving unit, used for receiving a video image captured by a camera of the steel billet being cut by the flame cutting machine; an identification unit, configured to identify the video image based on the first video AI model, and if it is determined that the first identification result is that the steel billet has been cut, send an actuator closing instruction to the PLC control system, so that the PLC control system forwards the actuator closing instruction to the execution structure; The first video AI model is obtained by training the LSTM in advance based on a first cut picture sample set labeled with respective corresponding shapes of various steel billets; The control unit is used to receive the closing state information of the execution structure after executing the mechanism closing instruction fed back by the PLC control system, and complete the flame cutting control of this steel billet cutting.

8. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images