Method for estimating the state of the lid of a coke oven and method for generating a state estimation model.

Abstract

To provide a method for estimating a state of coke oven lid parts capable of simultaneously identifying positions of lid parts where different abnormalities are occurring and each abnormality even when the different abnormalities occur simultaneously in a plurality of lid parts of a coke oven.SOLUTION: A method for estimating a state of coke oven lid parts comprises: a monitoring image division step of dividing a region of the coke oven included in monitoring image data into a plurality of divided image data based on positions of a plurality of lid parts in the monitoring image data; and an estimation step of inputting the divided image data in the monitoring image data as input data to a state estimation model, and outputting state information indicating any one of a normal state, a moving machine present state, a gas leakage state, and an ignition state in a state of the lid part included in the divided image data to estimate the state of the lid part.SELECTED DRAWING: Figure 4

Description

Technical Field

[0001] The present invention relates to a method for estimating the state of a furnace lid part of a coke oven and a method for generating a state estimation model, which enable quick grasping of the gas leakage and ignition states in the furnace lid part of the coke oven.

Background Art

[0002] Coke used as a raw material for a blast furnace is produced by heating and carbonizing coal in a carbonization chamber of a coke oven. In the coke oven, after carbonization of the coal (coke cake), it is carried out from the kiln mouth with the furnace lid removed after carbonization of the coal in the carbonization chamber.

[0003] However, when the seal heat insulating material at the kiln mouth is damaged by heat influence and the airtightness of the carbonization chamber is lost, an abnormality occurs in the gas flow in the carbonization chamber, and a flame occurs along with the leakage of gas from the kiln mouth. And when a flame occurs, operations such as extinguishing the flame and stopping the operation of the kiln for repair are generated, and problems such as a decrease in productivity occur. For this reason, research has been conventionally conducted on a technique for quickly grasping an abnormality in equipment in a coke oven.

[0004] Patent Document 1 discloses a method of extracting information on luminance values in pixel units in an image obtained by a monitoring camera, specifying a peripheral light region based on the luminance values, and discriminating the presence or absence of flame generation. Further, Patent Document 2 discloses a method of detecting a gas concentration using a gas sensor whose discoloration degree changes based on a gas concentration or the like, and detecting the gas concentration by a detection current obtained by a light detection sensor that measures the discoloration degree of the gas sensor.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Patent Document 2

[0006] However, while the method disclosed in Patent Document 1 can identify the presence or absence of flame, it is difficult to identify gas leaks, which are precursors to flames, at the same time as identifying flame occurrence. Furthermore, when using the method disclosed in Patent Document 2, it is necessary to install gas sensors near multiple furnace lid sections where gas leaks occur, which leads to cost issues in the equipment. Moreover, in aging coke ovens, gas leaks and flame occurrences often occur simultaneously in multiple locations, and there is a need for technology that can detect these simultaneously.

[0007] The present invention has been made in view of the above circumstances, and its object is to provide a method for estimating the state of a coke oven lid that makes it possible to simultaneously grasp the location of multiple lids where different abnormalities are occurring and the nature of each abnormality, even when different abnormalities are occurring simultaneously in multiple lids of a coke oven. [Means for solving the problem]

[0008] [1] A method for estimating the state of a coke oven lid, comprising: a monitoring image division step of dividing the region of a coke oven included in the monitoring image data into a plurality of divided image data based on the positions of a plurality of furnace lids; and an estimation step of inputting the divided image data in the monitoring image data as input data to a state estimation model, and outputting state information indicating one of the states of the furnace lid included in the divided image data, which is a normal state, a state with a mobile device, a gas leak state, or an ignition state, in order to estimate the state of the furnace lid. [2] A method for generating a state estimation model, wherein a machine learning model is trained using multiple datasets as training data, each dataset consisting of divided image data obtained by dividing the region of a coke oven contained in monitoring image data based on the positions of multiple furnace lids, and state information indicating the state of the furnace lid contained in the divided image data, which is one of the following states: normal state, mobile device present state, gas leak state, or ignition state, and the divided image data is used as training data to generate a state estimation model that takes the divided image data as input and outputs the state information in the divided image data. [Effects of the Invention]

[0009] According to the present invention, even when different abnormalities occur simultaneously in multiple furnace lid sections of a coke oven, it is possible to simultaneously determine the locations of the multiple furnace lid sections where these different abnormalities are occurring, as well as each of the abnormalities themselves. [Brief explanation of the drawing]

[0010] [Figure 1] This is a perspective view showing an example of a coke oven. [Figure 2] This is a schematic diagram showing an example of the system configuration. [Figure 3] This is a schematic diagram showing an example of dividing surveillance image data into multiple segmented image data. [Figure 4] This image shows the results of estimating the state of the furnace lid by applying a state estimation model to monitoring image data. [Modes for carrying out the invention]

[0011] The embodiments of the present invention will be described below with reference to the drawings. The following embodiments are preferred examples of the present invention and are not limiting in any way.

[0012] <Coke oven and lid section> Figure 1 shows the configuration of a coke oven 10 according to this embodiment. Figure 1 is a perspective view showing an example of a coke oven 10. First, the coke oven 10 will be described using Figure 1. The coke oven 10 has a heat storage section 12 composed of multiple heat storage chambers arranged in a row, and multiple carbonization chambers 14 and combustion chambers 16 provided on the heat storage section 12. The carbonization chambers 14 and combustion chambers 16 are arranged alternately adjacent to each other. The coal charging car 18 travels along the longitudinal direction L of the coke oven 10 over the carbonization chambers 14 and combustion chambers 16. Multiple coal charging holes (not shown) are formed in the upper wall of the carbonization chamber 14 along the short direction S of the coke oven 10, and coal, which will be the raw material for coke, is charged into the carbonization chamber 14 from these coal charging holes. Kiln openings 14a are provided on both sides of the carbonization chamber 14, and the kiln openings 14a are covered and closed by removable furnace lids 15. An extruder 20 is positioned on one side of the furnace opening 14a of the carbonization chamber 14, and a guide vehicle, a mobile machine 22, is positioned on the other side of the furnace opening 14a. The extruder 20 and the mobile machine 22 travel along the longitudinal direction L of the furnace.

[0013] In the carbonization chamber 14, coal is carbonized to form coke cake. To carbonize the coal, fuel gas is supplied from each heat storage chamber of the heat storage unit 12 to the combustion chamber 16 and burned, transferring the heat of combustion to the adjacent carbonization chamber 14, thereby heating the inside of the carbonization chamber 14. This raises the temperature of the carbonization chamber 14, causing the coal to carbonize. When the carbonization of the coal is complete, the furnace lid 15 is removed and the extrusion ram of the extruder 20 is inserted into the carbonization chamber 14. By inserting the extrusion ram, the coke cake obtained from the carbonization of the coal is pushed out of the carbonization chamber 14 and received by the mobile machine 22 on the opposite side of the extruder 20. Below the mobile machine 22, a fire extinguishing vehicle 24 is positioned so that it can travel along the longitudinal direction L of the furnace in front of the heat storage unit 12, and the fire extinguishing vehicle 24 receives the coke cake from the mobile machine 22. The fire extinguishing vehicle 24 transports the coke cake to a designated location.

[0014] Here, when coal is charged into the carbonization chamber 14 and gas is generated from within the carbonization chamber, if the sealing insulation material at the kiln opening is deteriorated, flames will erupt from the kiln opening. At this time, gas leakage will first occur from the furnace lid 15. Therefore, in the coke oven 10, it is necessary to instantly grasp the location of the gas leak and ignition that occurs.

[0015] <System Configuration> Next, the configuration of the system 30 according to this embodiment will be described with reference to Figure 2. Figure 2 is a schematic diagram showing an example of the system 30. The system 30 includes an imaging device 31, a transmitter 32, a receiver 33, a processing device 34, and an output device 35.

[0016] The imaging device 31 is positioned to capture all of the furnace lid portions 15 of the coke oven 10 shown in Figure 1 within its field of view. The imaging device 31 continuously images all of the furnace lid portions 15 of the coke oven 10 and generates monitoring image data. The transmitter 32 transmits the monitoring image data generated by the imaging device 31 to the receiver 33. The receiver 33 transfers the received monitoring image data to the processing device 34. The processing device 34 may be a personal computer for monitoring the furnace lid portions 15. The processing device 34 estimates the state of the furnace lid portions 15 of the coke oven 10 based on the monitoring image data. The output device 35 outputs the state of the furnace lid portions 15 estimated by the processing device 34. The output device 35 may be a display or the like with a display unit.

[0017] Here, the monitoring image division step and the estimation step will be described in relation to the method for estimating the state of the furnace lid of a coke oven according to the present invention.

[0018] <Monitoring image division process> The monitoring image segmentation process according to the present invention will be described with reference to FIG. 3. FIG. 3 is a schematic diagram showing an example in which monitoring image data is divided into a plurality of divided image data N. FIG. 3 is based on the monitoring image data generated by imaging the coke oven 10 with the imaging device 31 so as to include all the furnace lid portions 15 within the field of view. The imaging of the coke oven 10 by the imaging device 31 may be performed, for example, with an interval of about 70 m between the coke oven 10 and the imaging device 31. Then, the processing device 34 divides the area T of the coke oven 10 included in the monitoring image data into a plurality of divided image data N as shown by "1" to "8" in the figure based on the positions of the plurality of furnace lid portions 15. Although the image shown in FIG. 3 has a size approximately the same as the area T of the coke oven 10, an image including both the coke oven 10 and the peripheral equipment may be imaged, and the area T of the coke oven 10 included in the captured image may be confirmed and used as the image shown in FIG. 3. Further, the division into the divided image data N based on the positions of the plurality of furnace lid portions 15 may be performed such that only a single furnace lid portion 15 is included in the divided image data N based on the position of a single furnace lid portion 15, or a plurality of furnace lid portions 15 may be included in a single divided image data N based on the positions of the plurality of furnace lid portions 15. Although the plurality of divided image data N shown in FIG. 3 is shown in a form in which the image is divided into a plurality in the horizontal direction, the acquired image may be divided into a plurality in the vertical direction or the image may be divided into a plurality in a matrix form in both the vertical and horizontal directions.

[0019] <Estimation step> The processing device 34 inputs the divided image data N in the monitoring image data as input data into the state estimation model, and outputs state information indicating any one of the states of the furnace lid portion 15 included in the divided image data N, namely, the normal state, the state with a moving mechanism, the gas leakage state, and the ignition state, to estimate the state of the furnace lid portion 15.

[0020] Here, the state of the furnace lid portion 15 in the divided image data N will be described. In the furnace lid portion 15 of the coke oven 10, the state of the furnace lid portion 15 changes according to the internal situation of the closed carbonization chamber 14. In the present embodiment, the state of the furnace lid portion 15 is assumed to be four states: a normal state, a state with a traveling machine, a gas leakage state, and a firing state. The normal state means a state where there is no gas leakage or flame generation from the furnace lid portion 15. The state with a traveling machine means a state where the traveling machine 22 has moved (is positioned) to the furnace lid portion 15 so as to cover the furnace lid portion 15. The gas leakage state means a state where gas is leaking from the furnace lid portion 15. The firing state means a state where a flame is generated at the furnace lid portion 15.

[0021] In this way, in the processing device 34, since the state of the furnace lid portion 15 can be estimated in each region obtained by dividing the monitoring image, it is possible to simultaneously grasp the abnormalities of a plurality of furnace lid portions 15 in which different abnormalities have occurred.

[0022] The processing device 34 estimates the state of the furnace lid portion 15 in each of the divided image data N divided into a plurality in the monitoring image division step, and for each position (for example, "1" to "8" shown in FIG. 3) in the divided monitoring image data, the displays indicating the estimated state information may be superimposed. That is, an image is generated in which the state of the furnace lid portion 15 estimated for each divided image data N is visually displayed at each position in the monitoring image data.

[0023] The processing device 34 transmits the generated image data to the output device 35, and the output device 35 displays the image data. By displaying the image data in this way, as shown in FIG. 4, in the plurality of divided image data N of the coke oven 10, it is possible to simultaneously grasp the furnace lid portions 15 in the "gas leakage state" and the furnace lid portions 15 in the "firing state". And, as shown in FIG. 4, it is also possible to grasp the positions of the furnace lid portions 15 in the "gas leakage state" and the furnace lid portions 15 in the "firing state" respectively.

[0024] When displaying image data on the output device 35, as shown in Figure 4, for furnace lids 15 in a "fire state" requiring immediate attention to the coke oven 10, the word "fire" may be displayed near the furnace lid 15 on the display or the like to proactively notify the manager or other relevant parties. Based on the word "fire" displayed on the output device 35 and the position where the word is displayed, water can be quickly sprayed on the furnace lid 15 in a "fire state," enabling prompt firefighting.

[0025] Furthermore, when performing machine learning as a state estimation model (machine learning model), 16,000 surveillance image data images captured in the past were used as training data. Then, in the estimation process, 1,200 surveillance image data images were used as test data. As a result, for each segmented image data N in the surveillance image data, the estimation rate that was correctly estimated was 88.8% for "normal state (mobile device present)", 83.5% for "gas leak state", and 83.4% for "ignition state", with the average estimation rate for these three states being approximately 85%.

[0026] Regarding the monitoring image data, the estimation process in the processing device 34 classifies the estimated state patterns of the multiple furnace lid sections 15 of the coke oven 10 into the following four patterns. Furthermore, from the perspective of monitoring the occurrence of abnormalities in the furnace lid sections 15 of the coke oven 10, the four estimation patterns are described by including segmented image data N labeled "normal state" and segmented image data N labeled "moving device present state".

[0027] The first is the "normal (mobile device)" estimation pattern. The "normal (mobile device)" estimation pattern is a pattern in which all segmented image data N included in the monitoring image data are in a "normal state (mobile device present)" state.

[0028] The second pattern is the "normal (mobile device) + gas leak" estimation pattern. The "normal (mobile device) + gas leak" estimation pattern is a pattern in which all segmented image data N included in the monitoring image data include a "gas leak state" in addition to the "normal state (mobile device present)". In this case, it also includes states in which multiple segmented image data N included in the monitoring image data are estimated to be in a "gas leak state".

[0029] The third pattern is the "normal (mobile device) + ignition" estimation pattern. The "normal (mobile device) + ignition" estimation pattern is a pattern in which all segmented image data N included in the monitoring image data include the "ignition state" in addition to the "normal state (mobile device present)". In this case, it also includes states in which multiple segmented image data N included in the monitoring image data are estimated to be in the "ignition state".

[0030] The fourth pattern is the "normal (mobile device) + gas leak + ignition" estimation pattern. The "normal (mobile device) + gas leak + ignition" estimation pattern is a pattern in which all segmented image data N included in the monitoring image data include "normal state (mobile device present)", as well as "gas leak state" and "ignition state". In this case, it also includes states in which multiple segmented image data N included in the monitoring image data are estimated to be in a "gas leak state", and states in which multiple segmented image data N are estimated to be in an "ignition state".

[0031] Next, the method for generating a state estimation model according to the present invention will be described.

[0032] The processing device 34 prepares segmented image data N, which is obtained by dividing the region of the coke oven contained in previously captured monitoring image data based on the positions of multiple furnace lids 15. It also prepares state information indicating one of the following states for the furnace lids 15 contained in the prepared segmented image data N: normal state, mobile device present state, gas leak state, or ignition state. Multiple datasets, each consisting of a segmented image data N and state information indicating the state of the furnace lids 15 contained in the segmented image data N, are prepared as training data, and a machine learning model is trained using this training data. A state estimation model is then generated, which takes the segmented image data N as input and outputs the state information in the segmented image data N. Preferably, more than 1000 datasets are prepared as training data.

[0033] The status information may be data accumulated in conjunction with segmented image data N in previously captured monitoring image data. Alternatively, the status information may be information added to newly acquired segmented image data through visual inspection by operators or other personnel.

[0034] In the model training process, the model may be generated using machine learning (deep learning) while adjusting the weight coefficients in the dataset. As an introductory model for the neural network used in machine learning, the InceptioRes NetV2 model may be used from the perspective of improving accuracy in image recognition technology.

[0035] As described above, the coke oven lid state estimation method and state estimation model generation method according to the present invention make it possible to simultaneously grasp the location of each of the coke oven lids 15 where different abnormalities are occurring at the same time, even when different abnormalities are occurring at multiple coke oven lids 15. Furthermore, since the state estimation of the coke oven lids 15 using the state estimation model is performed instantaneously, the states of multiple coke oven lids 15 can be grasped quickly, and consequently, responses to coke oven lids 15 that are leaking or igniting can be taken quickly. Moreover, since the system 30 can be configured using an imaging device 31 that includes multiple coke oven lids 15 in its field of view, the system configuration for monitoring can be simplified. [Explanation of symbols]

[0036] 10 Coke oven 12 Heat storage section 14 Carbonization Chamber 14a Kiln mouth 15 Furnace lid part 16 Combustion chamber 18 Coal loading car 20 Extruders 22 Mobile Units 24 Fire truck 30 Systems 31 Imaging device 32 Transmitters 33 Receiver 34 Processing Unit 35 Output device L Furnace longitudinal direction S Furnace short direction N-segmented image data T Coke oven area

Claims

1. A monitoring image segmentation step in which the region of the coke oven included in the monitoring image data is divided into multiple segmented image data based on the positions of multiple oven lids, An estimation step is to input the divided image data from the monitoring image data into a state estimation model generated by machine learning using multiple datasets, each dataset consisting of divided image data obtained by dividing the region of a coke oven contained in previously captured monitoring image data based on the positions of multiple furnace lids, and state information indicating one of the following states of the furnace lid contained in the divided image data: normal state, with moving device, gas leak state, or ignition state, as training data, and output state information indicating one of the following states of the furnace lid contained in the divided image data, thereby estimating the state of the furnace lid contained in the divided image data. A method for estimating the state of the furnace lid of a coke oven, comprising the characteristics of a coke oven.

2. A machine learning model is trained using multiple datasets as training data, each dataset consisting of segmented image data obtained by dividing the region of the coke oven contained in the monitoring image data based on the positions of multiple furnace lids, and state information indicating one of the following states for the furnace lid contained in the segmented image data: normal state, mobile device present state, gas leak state, or ignition state. A method for generating a state estimation model, which takes the segmented image data as input and generates a state estimation model that outputs the state information in the segmented image data.

Images