A steel plate tracking method in cooling bed area based on multi-model information matching

Through the multi-model information matching method combining visual tracking and logical tracking in the cold bed area, the problems of real-time tracking of steel plates in the cold bed area are solved, and more accurate and comprehensive steel plate status monitoring is achieved, and production efficiency is improved and intelligence level.

CN114037752BActive Publication Date: 2024-11-12BEIJING SCI&TECH UNIV DESIGN RES YUAN CO
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111243605.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-10-25
Publication Date
2024-11-12
Estimated Expiration
2041-10-25

AI Technical Summary

Technical Problem

Real-time tracking of steel plates in cold bed areas is difficult, the existing logic tracking methods have large errors and cannot detect driving lifting operations, resulting in high uncertainty in the status of the steel plates and making it difficult to achieve comprehensive tracking.

Method used

Using a multi-model information matching method, combined with visual tracking and logical tracking, the visual model analyzes the steel plate status information from the image, and matches it with the speed data of the logical model to identify abnormal states and cold bed steel stuck phenomenon to realize real-time steel plate position track.

Benefits of technology

It improves the accuracy and comprehensiveness of steel plate tracking in cold bed areas, reduces manual verification, and improves production rhythm and corporate profits.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114037752B_ABST
    Figure CN114037752B_ABST
Patent Text Reader

Abstract

The present invention provides a method for tracking steel plates in a cooling bed area based on multi-model information matching, and belongs to the field of metallurgical intelligent manufacturing technology. The method first uses a visual tracking model to analyze and obtain the state information of the steel plates in the cooling bed area at time T0, and uses this information to initialize the state of the logic tracking model; the logic tracking model relies on the state of the steel plates in the cooling bed area at time T‑1 and combines the cooling bed speed to calculate the current state information S1 of the steel plates in the cooling bed area at time T; the visual tracking model uses an instance segmentation method to analyze the steel plates in the cooling bed area at time T from the image to obtain the current state information S2 of the steel plates in the cooling bed area; at time T, S1 and S2 are calculated and matched, the abnormal tracking state is analyzed, and S2 is saved for the next calculation; enter the next moment, repeat the above process, and track the state of the steel plates in the cooling bed area in real time. This method reduces the prediction error of the logic tracking model under long time series, and can judge special situations such as stuck steel and hoisting and unloading by using the matching state information, thereby realizing comprehensive, accurate and intelligent tracking of steel plates in the cooling bed area.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of metallurgical intelligent manufacturing, and in particular to a steel plate tracking method in a cooling bed area based on multi-model information matching. Background Art

[0002] During the production of steel products, the position information of each steel plate needs to be tracked in real time, and the tracking of the cooling bed area is the most difficult. The cooling bed area is large and relatively compact, so it is difficult to install optical inspection equipment in it, and there is no sensor to feedback actual information; and there are cranes and steel jams in the cooling bed area, which increases the uncertainty of the steel plate status; the combined impact of these factors increases the difficulty of tracking the steel plates in this area.

[0003] At present, the most commonly used steel plate tracking method in the cooling bed area in enterprises is logical tracking, that is, the speed information of the cooling bed operation is used to calculate and predict the position of the steel plate. Since it is only a theoretical calculation and lacks real-time sensor detection signal feedback, the steel plate position state error obtained by this tracking method is large. In addition, since the logical tracking method cannot detect the operation of the crane lifting the steel plate, the steel plate information cannot be revoked in time, and it is difficult to fully track the steel plates in the cooling bed area.

[0004] The visual tracking technology can intuitively monitor the steel plate status information in the cooling bed area, which is more accurate than the steel plate status information obtained by logical tracking calculation. Therefore, the present invention designs a method based on the fusion matching of logical tracking and visual tracking to track the steel plate in the cooling bed area. By matching and analyzing the detection data of the two models, not only can the position status information of the steel plate in the cooling bed area be accurately obtained, but also the phenomenon of steel stuck in the cooling bed can be identified, so that the steel plate in the cooling bed area can be tracked more comprehensively. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a steel plate tracking method in a cooling bed area based on multi-model information matching.

[0006] The method comprises the following steps:

[0007] (1) The state information of the steel plate in the cooling bed area at time T0 is obtained by analyzing the visual tracking model, and the state of the logical tracking model is initialized by using this information;

[0008] (2) The logic tracking model relies on the steel plate status in the cooling bed area at time T-1 and combines the cooling bed speed to calculate the steel plate status information S1 in the cooling bed area at the current time T;

[0009] (3) The visual tracking model uses the instance segmentation model to analyze the image and obtain the steel plate status information S2 in the cooling bed area at the current time T;

[0010] (4) At time T, the matching status of the steel plate status information S1 in the cooling bed area and the steel plate status information S2 in the cooling bed area is calculated, the abnormal tracking status is analyzed, and then the steel plate status information S2 in the cooling bed area is saved for the next calculation in step (2);

[0011] (5) Entering the next moment, repeating the process from step (2) to step (4), and tracking the status of the steel plate in the cooling bed area in real time.

[0012] In this method, T0 represents the initial time, T represents any time after the start, and T-1 represents the previous time before T.

[0013] In this method, the steel plates in the cooling bed area are transported to the cooling bed area by the upper cooling bed roller conveyor, and then the steel plates leave the cooling bed area by being directly lifted by a crane or transported away from the lower cooling bed roller conveyor.

[0014] The camera used in the above visual tracking model is an infrared monitoring camera, which is installed 2 meters above the driving track. The number of cameras ensures that the camera shooting range covers the entire cooling bed area.

[0015] The steel plate status information includes the x and y coordinate positions of the center of the steel plate at time T and the length and width of the steel plate at time T.

[0016] The instance segmentation model used by the visual tracking model in step (3) is one of SOLO and YOLACT. The steel plate area output by the segmentation model is mapped to the real coordinate system to obtain the steel plate state information S2 of the cooling bed area at time T. S2 includes in and Represents the x and y coordinate positions of the center of the steel plate identified at time T, and Represents the length and width of the steel plate identified at time T.

[0017] In step (2), the logic tracking model obtains the speed data of the cooling bed, and obtains the state quantity change of each steel plate in the cooling bed area per unit time by integrating the speed with time. The state information S1 of the steel plate in the cooling bed area obtained by the logic tracking model is divided into the following three cases:

[0018] ① When the steel plate enters the cooling bed, the logical tracking model is:

[0019]

[0020] ② When the steel plate runs in the cooling bed area, the logical tracking model is:

[0021]

[0022] ③ When the steel plate leaves the cooling bed, the logical tracking model is:

[0023]

[0024] in, and represents the x and y coordinate positions of the center of the steel plate tracked at time T, l T and w T represents the length and width of the steel plate tracked at time T, v is the speed of the cooling bed, and are the x and y coordinate positions of the center of the steel plate tracked at time T-1, l T-1 and w T-1 is the length and width of the steel plate tracked at time T-1.

[0025] In step (4), the steel plate status information of the cooling bed area obtained by the logic tracking model is The steel plate status information in the cooling bed area obtained by the visual tracking model Perform regional intersection-and-union ratio matching. If the intersection-and-union ratio is greater than 90%, it is a match, otherwise it is not a match.

[0026] The relationship between the above matching situation and the steel plate tracking in the cooling bed area is:

[0027] ①S 1 The number of steel plates is greater than S 2 Number of steel plates, S 1 The steel plate cannot be 2 The unmatched steel plates are the ones that were lifted by the crane, so the information is revoked;

[0028] ②S 1 The number of steel plates is less than S 2 Number of steel plates, S 2 The steel plate cannot be 1 The unmatched steel plates are the ones that have just entered the cooling bed area, and the information is added;

[0029] ③S 1 The number of steel plates is equal to S 2 Number of steel plates, S 1 or S 2 If there is a mismatch in the steel plate, it is judged that the cooling bed is stuck and an alarm is issued;

[0030] ④S 1 The number of steel plates is equal to S 2 Number of steel plates, S 1 and S 2 The Chinese steel plate is fully matched, which is the normal steel plate tracking status and timely information update.

[0031] When the crane is lifting the steel plate directly above the cooling bed, the visual tracking model stops tracking due to obstruction. The status information of the steel plate in the cooling bed area is completely obtained by the logical tracking model, and it is stipulated that when the crane is operating in the cooling bed area, the steel plate is not allowed to move on or off the cooling bed along the roller conveyor.

[0032] The beneficial effects of the above technical solution of the present invention are as follows:

[0033] The above solution solves the problem of real-time tracking of steel plates in the cooling bed area, improves the intelligence level of the cooling bed area, reduces the process of manual information verification, improves the production rhythm of the production line, and brings huge profits to steel companies. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a flow chart of the steel plate tracking method in the cooling bed area based on multi-model information matching of the present invention;

[0035] Figure 2 This is a schematic diagram of a steel plate entering a cooling bed area in an embodiment of the present invention;

[0036] Figure 3 It is a schematic diagram of the process of a steel plate moving in a cooling bed area in an embodiment of the present invention;

[0037] Figure 4 Schematic diagram of a steel plate leaving the cooling bed area in an embodiment of the present invention. DETAILED DESCRIPTION

[0038] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0039] The invention provides a cooling bed area steel plate tracking method based on multi-model information matching.

[0040] like Figure 1 As shown, the method comprises the following steps:

[0041] (1) The state information of the steel plate in the cooling bed area at time T0 is obtained by using the visual tracking model analysis, and this information is used to initialize the state of the logic tracking model, that is, the state information obtained by the visual tracking model is used to initialize the variables (position and width and length) by one-to-one assignment. ;

[0042] (2) The logic tracking model relies on the steel plate status in the cooling bed area at time T-1 and combines the cooling bed speed to calculate the steel plate status information S1 in the cooling bed area at the current time T;

[0043] (3) The visual tracking model uses the instance segmentation model to analyze the image and obtain the steel plate status information S2 in the cooling bed area at the current time T;

[0044] (4) At time T, the matching status of the steel plate status information S1 in the cooling bed area and the steel plate status information S2 in the cooling bed area is calculated, the abnormal tracking status is analyzed, and then the steel plate status information S2 in the cooling bed area is saved for the next calculation in step (2);

[0045] (5) Entering the next moment, repeating the process from step (2) to step (4), and tracking the status of the steel plate in the cooling bed area in real time.

[0046] This will be described below with reference to specific embodiments.

[0047] Taking the heat treatment line of a domestic steel plant as an example, the width of the cooling bed is 25 meters. In the case of short plate production, it can accommodate 8 steel plates at the same time, and in the case of long plate production, it can accommodate 4 steel plates at the same time. The steel plates on the cooling bed are transported to the cooling bed area by the upper cooling bed roller. There are two ways for the steel plates to leave the cooling bed area. One way is to be directly lifted away by a crane, and the other way is to be transported away from the lower cooling bed roller.

[0048] The camera in the visual tracking model uses a Hikvision infrared surveillance camera with a 4mm lens, a 79° horizontal field of view, and a captured image size of 2560*1440. The camera is installed 2 meters above the vehicle track and about 12 meters above the ground. A single camera can capture the entire cooling bed area.

[0049] The instance segmentation model used by the visual tracking model can be selected from SOLO and YOLACT, which can independently segment and obtain the position information, length and width information of each steel plate on the cooling bed. The steel plate status information output by the segmentation model is mapped to the physical coordinate system to obtain the status information of each steel plate in the cooling bed area at time T. The status information includes in and Represents the x and y coordinate positions of the center of the steel plate identified at time T, and Represents the length and width information of the steel plate identified at time T.

[0050] The logic tracking model obtains the motor speed data that controls the movement of the cooling bed, and obtains the state change of each steel plate in the cooling bed area per unit time by integrating the speed over time. The state change process of the logic tracking model is divided into the following three categories:

[0051] like Figure 2 As shown in the figure, when the steel plate enters the cooling bed, the logical tracking model is:

[0052]

[0053] like Figure 3 As shown in the figure, when the steel plate is running in the cooling bed area, the logical tracking model is:

[0054]

[0055] like Figure 4 As shown in the figure, when the steel plate leaves the cooling bed, the logical tracking model is:

[0056]

[0057] in and represents the x and y coordinate positions of the center of the steel plate tracked at time T, l T and w T Represents the length and width information of the steel plate tracked at time T, and v is the movement speed of the cooling bed.

[0058] The state of the steel plate in the cooling bed area obtained by the logic tracking model The state of the steel plate in the cooling bed area obtained by the visual tracking model The two states are matched by the regional intersection-and-union ratio. If the intersection-and-union ratio is greater than 90%, it is matched, otherwise it is not matched.

[0059] The relationship between the matching situation and the steel plate tracking in the cooling bed area is:

[0060] S 1 The number of steel plates is greater than S 2 Number of steel plates, S 1 The steel plate cannot be 2 The unmatched steel plates are the ones that were lifted by the crane, so the information is revoked;

[0061] S 1 The number of steel plates is less than S 2 Number of steel plates, S 2 The steel plate cannot be 1 The unmatched steel plates are the ones that have just entered the cooling bed area, and the information is added;

[0062] S 1 The number of steel plates is equal to S 2 Number of steel plates, S 1 or S 2 If there is a mismatch in the steel plates, it is judged that the cooling bed is stuck and an alarm is issued.

[0063] S 1 The number of steel plates is equal to S 2 Number of steel plates, S 1 and S 2 The Chinese steel plate is fully matched, which is the normal steel plate tracking status and timely information update.

[0064] The crane can feedback the position information in real time. When the crane is lifting the steel plate directly above the cooling bed, the visual tracking model stops tracking due to obstruction. The status information of the steel plate in the cooling bed area is completely obtained by the logical tracking model, and it is stipulated that when the crane is operating in the cooling bed area, the steel plate is not allowed to move on or off the cooling bed along the roller conveyor.

[0065] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A steel plate tracking method in a cooling bed area based on multi-model information matching, characterized in that: The steps include: (1) The state information of the steel plate in the cooling bed area at time T0 is obtained by analyzing the visual tracking model, and the state of the logical tracking model is initialized by using this information; (2) The logic tracking model relies on the steel plate status in the cooling bed area at time T-1 and combines the cooling bed speed to calculate the steel plate status information S1 in the cooling bed area at the current time T; (3) The visual tracking model uses the instance segmentation model to analyze the image and obtain the steel plate status information S2 in the cooling bed area at the current time T; (4) At time T, the matching status of the steel plate status information S1 in the cooling bed area and the steel plate status information S2 in the cooling bed area is calculated, the abnormal tracking status is analyzed, and then the steel plate status information S2 in the cooling bed area is saved for the next calculation in step (2); (5) Entering the next moment, repeating the process from step (2) to step (4), and tracking the status of the steel plate in the cooling bed area in real time; In step (2), the logic tracking model obtains the speed data of the cooling bed, and obtains the state quantity change of each steel plate in the cooling bed area per unit time by integrating the speed with time. The state information S1 of the steel plate in the cooling bed area obtained by the logic tracking model is divided into the following three situations: ① When the steel plate enters the cooling bed, the logical tracking model is: ② When the steel plate runs in the cooling bed area, the logical tracking model is: ③ When the steel plate leaves the cooling bed, the logical tracking model is: in, and represents the x and y coordinate positions of the center of the steel plate tracked at time T, l T and w T represents the length and width of the steel plate tracked at time T, v is the speed of the cooling bed, and are the x and y coordinate positions of the center of the steel plate tracked at time T-1, l T-1 and w T-1 is the length and width of the steel plate tracked at time T-1.

2. The method for tracking steel plates in cooling bed area based on multi-model information matching according to claim 1 is characterized in that: In this method, the steel plates in the cooling bed area are transported to the cooling bed area by the upper cooling bed roller conveyor, and then the steel plates leave the cooling bed area by being directly lifted by a crane or transported away from the lower cooling bed roller conveyor.

3. The method for tracking steel plates in cooling bed area based on multi-model information matching according to claim 1 is characterized in that: The camera used in the visual tracking model is an infrared monitoring camera, which is installed 2 meters above the driving track. The number of cameras ensures that the camera shooting range covers the entire cooling bed area.

4. The method for tracking steel plates in cooling bed area based on multi-model information matching according to claim 1 is characterized in that: The steel plate status information includes the x and y coordinate positions of the center of the steel plate at time T and the length and width of the steel plate at time T.

5. The method for tracking steel plates in cooling bed area based on multi-model information matching according to claim 1 is characterized in that: The instance segmentation model used by the visual tracking model in step (3) is one of SOLO and YOLACT. The steel plate area output by the segmentation model is mapped to the real coordinate system to obtain the steel plate state information S2 of the cooling bed area at time T. S2 includes in and Represents the x and y coordinate positions of the center of the steel plate identified at time T, and Represents the length and width of the steel plate identified at time T.

6. The method for tracking steel plates in cooling bed area based on multi-model information matching according to claim 1 is characterized in that: In the step (4), the steel plate status information of the cooling bed area obtained by the logic tracking model is The steel plate status information in the cooling bed area obtained by the visual tracking model Perform regional intersection-and-union ratio matching. If the intersection-and-union ratio is greater than 90%, it is a match, otherwise it is not a match.

7. The method for tracking steel plates in cooling bed area based on multi-model information matching according to claim 6 is characterized in that: The relationship between the matching situation and the steel plate tracking in the cooling bed area is: ① The number of steel plates in S1 is greater than that in S2. The steel plates in S1 cannot be matched in S2. The unmatched steel plates are the ones that were lifted away by the crane, so the information is revoked; ②The number of steel plates in S1 is less than that in S2. The steel plates in S2 cannot be matched in S1. The unmatched steel plates are the ones that have just entered the cooling bed area, so information is added; ③The number of steel plates in S1 is equal to the number of steel plates in S2, and there is a mismatch in S1 or S2, which is judged as steel jam on the cooling bed and an alarm is issued; ④The number of steel plates in S1 is equal to the number of steel plates in S2. The steel plates in S1 and S2 are completely matched, which is the normal steel plate tracking status, and timely information update is carried out.

8. The method for tracking steel plates in cooling bed area based on multi-model information matching according to claim 2 is characterized in that: When the crane is lifting the steel plate directly above the cooling bed, the visual tracking model stops tracking due to obstruction. The status information of the steel plate in the cooling bed area is completely obtained by the logical tracking model, and it is stipulated that when the crane is operating in the cooling bed area, the steel plate is not allowed to move on or off the cooling bed along the roller conveyor.