Image monitoring equipment control method and system

Abstract

The embodiment of the invention provides an image monitoring equipment control method and system, is applied to image monitoring equipment control in a refining heating furnace, and belongs to the technical field of equipment monitoring. The method comprises the following steps: acquiring real-time image information of flame of each furnace nozzle, and evaluating the abnormal state probability of the flame of each furnace nozzle based on the image information; positioning and evaluating the furnace nozzles with the abnormal probability greater than a preset threshold value to form an abnormal furnace nozzle set; generating a posture adjustment scheme of the image monitoring equipment array based on the position of each abnormal furnace nozzle in the abnormal furnace nozzle set; and executing the attitude adjustment scheme, and executing attitude adjustment of each image monitoring device in the image monitoring array. The scheme of the invention solves the problem that the accuracy of the existing in-furnace flame state monitoring scheme cannot be guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment monitoring, and in particular to an image monitoring equipment control method and an image monitoring equipment control system. Background Art

[0002] The refining heating furnace is an important equipment in the refining enterprise. Its function is to heat the furnace tube through the flame of the furnace nozzle and heat the fluid in the furnace tube to a predetermined temperature, so that the subsequent refining equipment can perform the corresponding refining process. Therefore, the operation result of the refining equipment will directly depend on the heating effect of the refining heating furnace. The heating effect of the refining heating furnace directly depends on the flame combustion state of the furnace nozzle. If the furnace nozzle or the fuel supply equipment fails, the fluid heating will not be able to be performed as expected, and the refining needs will not be met. In order to ensure the heating effect of the refining heating furnace, it is necessary to monitor the combustion state of the furnace nozzle flame and determine whether the furnace nozzle flame meets expectations. In the existing scheme, a fire viewing hole is opened on the outer wall of the furnace of the refining heating furnace, so that the supervisor can observe the flame combustion state of each furnace nozzle in the furnace through the fire viewing hole, and then judge whether there is any abnormality in the current flame combustion state based on his own experience. This method has great limitations:

[0003] 1) There is no guarantee that the supervisor can continuously observe the flame status through the fire viewing hole. Therefore, when a sudden abnormality occurs in the furnace, there is no guarantee that the supervisor can discover the abnormality in the first time.

[0004] 2) The abnormal judgment of flame status needs to rely on the personal experience of the supervisor, and the accuracy of the monitoring results cannot be guaranteed.

[0005] In response to this problem, some solutions exist that monitor the flame status by setting up image monitoring equipment and then collecting flame image information. However, because the furnace environment is very complex and there are multiple furnace nozzles, the accuracy of existing image monitoring solutions cannot be guaranteed due to too many interference factors. Existing image monitoring solutions all use image monitoring equipment to collect wide-area images of the furnace, and then identify abnormal flames in the images. However, this method is not very targeted and is prone to identifying abnormal flames of other furnace nozzles as the flame of the current furnace nozzle, and it is also easy to identify flame areas with other noise. In view of the problem that the accuracy of existing furnace flame status monitoring solutions cannot be guaranteed, it is necessary to create a new image monitoring equipment control solution. Summary of the invention

[0006] The purpose of the embodiments of the present invention is to provide an image monitoring device control method and system to at least solve the problem that the accuracy of existing furnace flame state monitoring solutions cannot be guaranteed.

[0007] In order to achieve the above-mentioned purpose, the first aspect of the present invention provides an image monitoring equipment control method, which is applied to the image monitoring equipment control in a refining heating furnace, and the method comprises: collecting real-time image information of each furnace nozzle flame, and evaluating the abnormal probability of each furnace nozzle flame state based on the image information; locating furnace nozzles with an abnormal probability greater than a preset threshold to form an abnormal furnace nozzle set; based on the position of each abnormal furnace nozzle in the abnormal furnace nozzle set, generating a posture adjustment plan for the image monitoring equipment array; executing the posture adjustment plan to perform posture adjustment of each image monitoring device in the image monitoring array.

[0008] Optionally, the abnormal probability assessment of the flame state of each furnace nozzle based on the image information includes: preprocessing the image information to obtain a preprocessed image; identifying the flame area of ​​the preprocessed image; performing three-dimensional rubbing based on pre-constructed furnace three-dimensional information and the identified flame area, and identifying the furnace nozzle serial number belonging to each flame area; based on a pre-constructed abnormal recognition model, performing abnormal probability assessment on the flame of each furnace nozzle one by one, and obtaining abnormal probability prediction results corresponding to each furnace nozzle.

[0009] Optionally, the preprocessing of the image information to obtain preprocessed image information includes: denoising the image information to obtain a denoised image; binarizing the denoised image to obtain a binarized image, and using the binarized image as the preprocessed image.

[0010] Optionally, the flame area recognition of the preprocessed image includes: performing grayscale recognition on each pixel in the binarized image, and treating the areas connected by pixels whose grayscale is lower than a preset grayscale threshold as background areas; removing the background area, and connecting the areas in the remaining image based on grayscale similarity to obtain multiple independent calibrated areas, and treating these calibrated areas as flame areas.

[0011] Optionally, three-dimensional rubbing is performed based on pre-constructed furnace three-dimensional information and identified flame areas to identify the furnace nozzle serial numbers belonging to each flame area, including: determining the image monitoring device from which the image information corresponding to the flame area identification result comes, and reading the posture information of the image monitoring device; determining the field of view information of the corresponding image monitoring device based on the posture information; based on the field of view information, three-dimensional rubbing is performed in the pre-constructed furnace three-dimensional information, and image calibration information under the corresponding field of view is projected; based on the image calibration information, the furnace nozzle serial numbers belonging to each flame area are calibrated for the image that determines the flame area identification result.

[0012] Optionally, the method further includes: constructing three-dimensional information of the furnace, including: collecting design information of the refining heating furnace, and reading structural information of the furnace in the design information; importing the structural information into MATLAB to obtain three-dimensional information of the furnace.

[0013] Optionally, the method further includes: constructing an abnormality recognition model, including: collecting historical image information of the furnace nozzle flame, the historical image information including: normal flame image information and labeled abnormal flame image information; using the historical image information as a training sample, performing model training in a preset neural network, and obtaining an abnormality recognition model.

[0014] Optionally, based on the position of each abnormal furnace nozzle in the abnormal furnace nozzle set, a posture adjustment plan for the image monitoring device array is generated, including: identifying the furnace nozzle serial number with an abnormal probability greater than a preset threshold, and locating the position of each abnormal furnace nozzle in the design information of the refining heating furnace based on the determined furnace nozzle serial number; based on the position of each abnormal furnace nozzle, identifying the priority of each image monitoring device relative to each abnormal furnace nozzle in the preferred image monitoring device comparison table; wherein, the higher the priority, the better the field of view of the image information of the abnormal furnace nozzle collected by the corresponding image monitoring device; based on the priority of each image monitoring device relative to each abnormal furnace nozzle, image monitoring devices are allocated to each abnormal furnace nozzle in the image monitoring device array, and each abnormal furnace nozzle is allocated at least two image monitoring devices; each image monitoring device generates a posture adjustment plan for the corresponding abnormal furnace nozzle based on the allocation result.

[0015] Optionally, each image monitoring device generates a posture adjustment plan for the corresponding abnormal furnace nozzle based on the allocation result, including: if an image monitoring device is assigned to multiple abnormal furnace nozzles at the same time, then the supervision operation order of the image monitoring device is determined based on the priority between the image monitoring device and each abnormal furnace nozzle, and the higher the priority, the higher the supervision operation order of the image monitoring device relative to the abnormal furnace nozzle; if an image monitoring device is assigned to multiple abnormal furnace nozzles at the same time, and the image monitoring device has the same priority as at least two of the abnormal furnace nozzles, then the abnormal probabilities of the abnormal furnace nozzles with the same priority are compared, and the greater the abnormal probability, the higher the supervision operation order of the image monitoring device relative to the abnormal furnace nozzle; based on the determined supervision operation order, a field of view alignment order and a focusing plan are generated for each image monitoring device, and the field of view alignment order and the focusing plan are used as the posture adjustment plan.

[0016] Optionally, based on the determined supervision operation sequence, a field of view alignment sequence and a focusing scheme are generated for each image monitoring device, including: based on the determined supervision operation sequence, generating a sequential alignment scheme for each image monitoring device relative to each abnormal furnace nozzle, and generating an alignment timing table based on the sequential alignment scheme; determining the distance information between the abnormal furnace nozzle corresponding to each alignment moment and the assigned image monitoring device, and determining the focusing information at each alignment moment based on the distance information, and adding the focusing information to the alignment timing table in sequence to obtain the corresponding field of view alignment sequence and focusing scheme.

[0017] Optionally, the method also includes: after completing the posture adjustment of each image monitoring device, increasing the sampling frequency based on the original sampling frequency, and completing a preset period of flame image acquisition corresponding to the abnormal furnace nozzle as a supervision image; performing abnormality probability evaluation on the supervision image again, if the abnormality probability evaluation result is still greater than the preset threshold, outputting an alarm message; otherwise, correcting the abnormal furnace nozzle to a normal furnace nozzle.

[0018] The second aspect of the present invention provides an image monitoring equipment control system, which is applied to the control of image monitoring equipment in a refining heating furnace. The system includes: an acquisition unit, which is used to acquire real-time image information of the flame of each furnace nozzle, and evaluate the abnormal probability of the flame state of each furnace nozzle based on the image information; a positioning unit, which is used to locate the furnace nozzles with an abnormal probability greater than a preset threshold to form an abnormal furnace nozzle set; a scheme generation unit, which is used to generate a posture adjustment scheme for the image monitoring equipment array based on the position of each abnormal furnace nozzle in the abnormal furnace nozzle set; and an execution unit, which is used to execute the posture adjustment scheme and perform posture adjustment of each image monitoring device in the image monitoring array.

[0019] On the other hand, the present invention provides a computer-readable storage medium, on which instructions are stored, which, when executed on a computer, enable the computer to execute the above-mentioned image monitoring device control method.

[0020] Through the above technical solution, the solution of the present invention sets an image monitoring device in the furnace to form an image monitoring array, and continuously collects the flame image information of each furnace nozzle. Then, through the corresponding image processing method, the flame state is automatically monitored through the flame image information, and the abnormal probability is output. For the furnace nozzles with an abnormal probability greater than the preset threshold, the posture information of the image monitoring device is adjusted in a targeted manner, and the abnormal furnace nozzles are continuously monitored, thereby realizing differentiated supervision of abnormal furnace nozzles and ensuring monitoring accuracy.

[0021] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the accompanying drawings:

[0023] Figure 1 is a flowchart of the steps of an image monitoring device control method provided by an embodiment of the present invention;

[0024] Figure 2 It is a system structure diagram of an image monitoring equipment control system provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0025] The specific implementation of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the present invention, and is not used to limit the present invention.

[0026] The refining heating furnace is an important equipment in the refining enterprise. Its function is to heat the furnace tube through the flame of the furnace nozzle and heat the fluid in the furnace tube to a predetermined temperature so that the subsequent refining equipment can perform the corresponding refining process. Therefore, the operation result of the refining equipment will directly depend on the heating effect of the refining heating furnace. The heating effect of the refining heating furnace directly depends on the flame combustion state of the furnace nozzle. If the furnace nozzle or fuel supply equipment fails, the fluid heating will not be able to be performed as expected, and the refining needs will not be met.

[0027] In order to ensure the heating effect of the refining heating furnace, it is necessary to monitor the combustion status of the nozzle flame and determine whether the nozzle flame meets expectations. In the existing solution, a fire viewing hole is opened on the outer wall of the furnace of the refining heating furnace, so that the supervisor can observe the combustion status of each nozzle flame in the furnace through the fire viewing hole, and then judge whether there is any abnormality in the current flame combustion status based on their own experience. This method has great limitations:

[0028] 1) There is no guarantee that the supervisor can continuously observe the flame status through the fire viewing hole. Therefore, when a sudden abnormality occurs in the furnace, there is no guarantee that the supervisor can discover the abnormality in the first time.

[0029] 2) The abnormal judgment of flame status needs to rely on the personal experience of the supervisor, and the accuracy of the monitoring results cannot be guaranteed.

[0030] In response to this problem, some solutions exist that monitor the flame status by setting up image monitoring equipment and then collecting flame image information. However, because the furnace environment is very complex and there are multiple furnace nozzles, the accuracy of existing image monitoring solutions cannot be guaranteed due to too many interference factors. Existing image monitoring solutions all use image monitoring equipment to collect images of a wide area in the furnace, and then identify abnormal flames in the images. However, this method is not very targeted and is prone to identifying abnormal flames of other furnace nozzles as the flame of the current furnace nozzle, and also prone to identifying flame areas of other noise.

[0031] In view of the problem that the accuracy of existing furnace flame status monitoring solutions cannot be guaranteed, the present invention proposes an image monitoring equipment control method, which is applied to the image monitoring equipment control in the refining heating furnace. The present invention sets up an image monitoring device in the furnace to form an image monitoring array, and continuously collects the flame image information of each furnace nozzle. Then, through the corresponding image processing method, the flame status is automatically monitored through the flame image information, and the abnormality probability is output. For the furnace nozzles whose abnormal probability is greater than the preset threshold, the posture information of the image monitoring device is adjusted in a targeted manner, and the abnormal furnace nozzles are continuously monitored, thereby realizing differentiated supervision of abnormal furnace nozzles and ensuring monitoring accuracy.

[0032] Figure 1 FIG. 1 is a flowchart of a method for controlling an image monitoring device provided by an embodiment of the present invention. Figure 1 As shown, an embodiment of the present invention provides an image monitoring device control method, the method comprising:

[0033] Step S10: collecting real-time image information of each furnace nozzle flame, and evaluating the probability of abnormal state of each furnace nozzle flame based on the image information.

[0034] Specifically, the scheme of the present invention aims to achieve differentiated monitoring of each furnace nozzle, that is, for the furnace nozzle initially identified as possibly abnormal, the scheme of the present invention will automatically perform image monitoring equipment allocation and posture adjustment to ensure that the image monitoring field is suitable for the suspected abnormal furnace nozzle, and conduct differentiated key monitoring on the furnace nozzle to determine whether it is really abnormal, so as to avoid false alarms. In this way, the image of the abnormal furnace nozzle can also be focused on, further improving the quality of the state judgment image, which is also conducive to improving the accuracy of image monitoring.

[0035] Embodiment 1:

[0036] An image acquisition array is arranged in the furnace, which is arranged at different positions to perform all-round image acquisition in the furnace. The abnormal probability assessment of the flame state of each furnace nozzle based on the image information includes: preprocessing the acquired image to obtain preprocessed image information; performing flame area recognition in the preprocessed image; performing three-dimensional rubbing based on the pre-constructed furnace three-dimensional information and the identified flame area to identify the furnace nozzle serial number to which each flame area belongs; based on the pre-constructed abnormal recognition model, performing abnormal probability assessment on the flame of each furnace nozzle one by one to obtain the abnormal probability prediction result corresponding to each furnace nozzle.

[0037] In the embodiment of the present invention, because there are multiple image monitoring devices installed in the furnace, and the installation positions of each image monitoring device are different, the corresponding fields of view of each image monitoring device when performing image acquisition are extremely different. When acquiring images from different angles, there may be multiple furnace nozzles in the field of view, but because of the different installation positions, the positions of each furnace nozzle in different images are different. Based on this, when performing subsequent abnormal judgment, how to correspond the judgment result with the furnace nozzle is a technical problem that needs to be solved.

[0038] Based on this, the scheme of the present invention first preprocesses the image, and then locates the flame in the preprocessed image. Specifically, the image information is denoised to obtain denoised image information; the denoised image information is binarized to obtain a binarized image, and the binarized image is used as the preprocessed image. In the preprocessed image, flame area recognition is performed, including: grayscale recognition of each pixel in the binarized image, and the areas connected to the pixels with grayscale lower than the preset grayscale threshold are all used as background areas; the background area is deducted, and the area is connected based on grayscale similarity in the remaining image information to obtain multiple independent calibration areas, and these calibration areas are used as flame areas.

[0039] In an embodiment of the present invention, in order to reduce the impact of noise on image quality and avoid misidentifying image noise as flame information, it is necessary to perform denoising on the image before flame recognition. Then the denoised image is binarized. Image binarization is to set the grayscale value of the pixel on the image to 0 or 255, that is, to present the entire image with an obvious black and white effect. There are differences in the color, brightness, etc. of the original image, which will eventually be distinguished by the grayscale value on the binary image. Because the scheme of the present invention is to collect flame images in the furnace, the background is pure black. In this case, the background color is relatively simple and the grayscale difference will be very large. This method can easily distinguish the flame area and realize the rapid positioning of the flame area. Because different furnace nozzles have separate flames, there will also be multiple identifiable flame areas, and each flame area corresponds to at least one furnace nozzle.

[0040] Embodiment 2:

[0041] Based on the pre-constructed three-dimensional furnace information and the identified flame area, a three-dimensional rubbing is performed to identify the furnace nozzle serial number belonging to each flame area, including: determining the image monitoring device from which the image information corresponding to the flame area identification result comes, and reading the posture information of the image monitoring device; determining the field of view information of the corresponding image monitoring device based on the posture information; based on the field of view information, a three-dimensional rubbing is performed in the pre-constructed three-dimensional furnace information, and image calibration information under the corresponding field of view is projected; based on the image calibration information, the furnace nozzle serial number belonging to each flame area is calibrated for the image that determines the flame area identification result.

[0042] In an embodiment of the present invention, in order to realize the association between the flame area and the furnace nozzle in the field of view images of different image monitoring devices, the scheme of the present invention performs three-dimensional rubbing based on the three-dimensional information of the furnace, and reversely rubs the furnace nozzle information based on the collected image as an implicit two-dimensional image. The orientation of the monitoring device determines its viewing angle, and the focal length of the image monitoring device determines the size of its field of view. To read the posture information of the image monitoring device, it is necessary to read the orientation information and focal length information at the same time, and then simulate the image information of the corresponding field of view size under the corresponding field of view. Perform three-dimensional rubbing based on the three-dimensional information of the furnace, find the furnace nozzle information under the field of view, and determine the furnace nozzle serial number.

[0043] Preferably, the method further comprises: constructing three-dimensional information of the furnace, including: collecting design information of the refining heating furnace, and reading structural information of the furnace in the design information; importing the structural information into MATLAB to obtain three-dimensional information of the furnace.

[0044] Embodiment 3:

[0045] Based on the pre-built abnormality recognition model, the abnormal probability of the flame of each furnace nozzle is evaluated one by one to obtain the abnormal probability prediction result corresponding to each furnace nozzle. The method also includes: constructing the abnormality recognition model, including: collecting historical image information of the furnace nozzle flame, the historical image information includes: normal flame image information and annotated abnormal flame image information; using the historical image information as a training sample, performing model training in the pre-neural network, and obtaining the abnormality recognition model.

[0046] Step S20: Locate the furnace nozzles whose abnormal probability is greater than a preset threshold value to form an abnormal furnace nozzle set.

[0047] Specifically, after obtaining the abnormal probability assessment result of each furnace nozzle, the abnormal assessment result of each furnace nozzle is compared with the preset threshold. When it is judged that the abnormal probability assessment result of the furnace nozzle is greater than the preset threshold, it means that the furnace nozzle has abnormal combustion, and it is defined as an abnormal furnace nozzle.

[0048] Preferably, because there are multiple image monitoring devices in the furnace, multiple image monitoring devices may have flame image information of the same furnace nozzle at the same time. Because the image acquisition is performed from different postures, the flame state of the same furnace nozzle may be different in the image information of different image monitoring devices. This makes it possible that the same furnace nozzle is judged to be abnormal and normal at the same time. In this case, as long as a furnace nozzle is judged to be abnormal in the image of any image monitoring device, in order to avoid missing abnormal conditions, even if it is judged to be normal in the images of other image monitoring devices, it still needs to be identified as an abnormal furnace nozzle.

[0049] Step S30: generating a posture adjustment plan for the image monitoring device array based on the position of each abnormal furnace nozzle in the abnormal furnace nozzle set.

[0050] Embodiment 4:

[0051] Identify the serial numbers of the furnace nozzles whose abnormal probability is greater than a preset threshold, and locate the position information of each abnormal furnace nozzle in the refining heating furnace design information based on the determined serial numbers of the furnace nozzles; based on the position information of each abnormal furnace nozzle, identify the priority information of each image monitoring device relative to each abnormal furnace nozzle in the preferred image monitoring device comparison table; wherein, the higher the priority, the better the field of view of the image information of the corresponding abnormal furnace nozzle collected by the corresponding image monitoring device; based on the priority information of each image monitoring device relative to each abnormal furnace nozzle, allocate image monitoring devices in the image monitoring device array for each abnormal furnace nozzle, and each abnormal furnace nozzle is allocated with at least two image monitoring devices; each image monitoring device generates a posture adjustment plan for the corresponding abnormal furnace nozzle based on the allocation result.

[0052] In the embodiment of the present invention, because the image monitoring devices are set at different positions, the optimal viewing angles of the image monitoring devices are different. Even if some furnace nozzles appear in the field of view of a certain image monitoring device, because they may greatly deviate from the optimal viewing angle of the device, their image information is distorted to a certain extent. In this case, the accuracy of the flame status monitoring of these furnace nozzles based on the image monitoring device cannot be guaranteed. Based on this, when the image monitoring device is set in the scheme of the present invention, it stores the corresponding preferred image monitoring device comparison table. When it is necessary to focus on the abnormal furnace nozzles in the future, it is necessary to select the preferred image monitoring device for image acquisition to ensure that the abnormal furnace nozzle is within the optimal viewing angle of the assigned image monitoring device, avoid image distortion, and improve monitoring accuracy.

[0053] Preferably, each image monitoring device generates a posture adjustment plan for the corresponding abnormal furnace nozzle based on the allocation result, including: if an image monitoring device is assigned to multiple abnormal furnace nozzles at the same time, then the supervision operation order of the image monitoring device is determined based on the priority between the image monitoring device and each abnormal furnace nozzle, and the higher the priority, the higher the supervision operation order of the image monitoring device relative to the abnormal furnace nozzle; if an image monitoring device is assigned to multiple abnormal furnace nozzles at the same time, and the image monitoring device has the same priority as at least two of the abnormal furnace nozzles, then the abnormal probabilities of the abnormal furnace nozzles with the same priority are compared, and the greater the abnormal probability, the higher the priority, and the higher the supervision operation order of the image monitoring device relative to the abnormal furnace nozzle; based on the determined supervision operation order, a field of view alignment order and a focusing plan are generated for each image monitoring device, and the field of view alignment order and the focusing plan are used as the posture adjustment plan.

[0054] Preferably, the method of generating a field of view alignment sequence and a focusing scheme for each image monitoring device based on the determined supervision operation sequence includes: generating a sequential alignment scheme for each image monitoring device relative to each abnormal furnace nozzle based on the determined supervision operation sequence, and generating an alignment timing table based on the sequential alignment scheme; determining the distance information between the abnormal furnace nozzle corresponding to each alignment moment and the assigned image monitoring device, and determining the focusing information at each alignment moment based on the distance information, and sequentially adding the focusing information to the alignment timing table to obtain the corresponding field of view alignment sequence and focusing scheme.

[0055] In the embodiment of the present invention, the scheme of the present invention aims to realize the key supervision of abnormal furnace nozzles. In addition to selecting the image monitoring device with the best viewing angle, in order to avoid the interference of the flames of other furnace nozzles, it is ensured that only the flame information of the target furnace nozzle exists in the field of view, which is the expected state. Based on this, when the scheme of the present invention performs key supervision, the image monitoring device will be automatically aligned and focused to realize the unique monitoring of abnormal furnace nozzles and ensure the monitoring accuracy. However, because different image monitoring devices cannot serve only one furnace nozzle, it is necessary to generate a corresponding timing table to perform sequential key supervision based on priority or abnormality probability, so as to ensure accuracy while reducing the delay of the overall solution and avoid missing key abnormalities.

[0056] Step S40: executing the posture adjustment scheme to perform posture adjustment of each image monitoring device in the image monitoring array.

[0057] Specifically, based on the field of view alignment sequence and the focusing scheme, the posture adjustment of each image monitoring device is synchronously executed to realize the monitoring of all abnormal furnace nozzles, thereby achieving secondary key supervision after identifying suspected abnormalities and avoiding false alarms caused by misidentification.

[0058] Preferably, the method also includes: after completing the posture adjustment of each image monitoring device, increasing the sampling frequency on the basis of the original sampling frequency, and completing a preset period of flame image acquisition corresponding to the abnormal furnace nozzle as a supervision image; performing abnormal probability evaluation on the supervision image again, if the abnormal probability evaluation result is still greater than the preset threshold, outputting an alarm message; otherwise, correcting the abnormal furnace nozzle to a normal furnace nozzle.

[0059] Embodiment 5:

[0060] Figure 2 1 is a system structure diagram of an image monitoring device control system provided by an embodiment of the present invention. Figure 2 As shown, an embodiment of the present invention provides an image monitoring device control system, the system comprising:

[0061] The acquisition unit is used to acquire real-time image information of the flames of each furnace nozzle, and to evaluate the probability of abnormal state of the flame of each furnace nozzle based on the image information.

[0062] Specifically, the scheme of the present invention aims to achieve differentiated monitoring of each furnace nozzle, that is, for the furnace nozzle initially identified as possibly abnormal, the scheme of the present invention will automatically perform image monitoring equipment allocation and posture adjustment to ensure that the image monitoring field is suitable for the suspected abnormal furnace nozzle, and conduct differentiated key monitoring on the furnace nozzle to determine whether it is really abnormal, so as to avoid false alarms. In this way, the image of the abnormal furnace nozzle can also be focused on, further improving the quality of the state judgment image, which is also conducive to improving the accuracy of image monitoring.

[0063] Specifically, an image acquisition array is provided in the furnace, which is respectively arranged at different positions to perform all-round image acquisition in the furnace. The abnormal probability assessment of the flame state of each furnace nozzle based on the image information includes: preprocessing the acquired image to obtain preprocessed image information; performing flame area recognition in the preprocessed image; performing three-dimensional rubbing based on pre-constructed furnace three-dimensional information and the identified flame area to identify the furnace nozzle serial number to which each flame area belongs; based on the pre-constructed abnormal recognition model, performing abnormal probability assessment on the flame of each furnace nozzle one by one to obtain the abnormal probability prediction result corresponding to each furnace nozzle.

[0064] In the embodiment of the present invention, because multiple image monitoring devices are installed in the furnace, and the installation positions of each image monitoring device are different, the corresponding fields of view of each image monitoring device when performing image acquisition are extremely different. When acquiring images from different angles, there may be multiple furnace nozzles in the field of view, but because of the different installation positions, the positions of each furnace nozzle in different images are different. Based on this, when performing subsequent abnormal judgment, how to correspond the judgment result with the furnace nozzle is a technical problem that needs to be solved.

[0065] Based on this, the scheme of the present invention first preprocesses the image, and then locates the flame in the preprocessed image. Specifically, the image information is denoised to obtain denoised image information; the denoised image information is binarized to obtain a binarized image, and the binarized image is used as the preprocessed image. In the preprocessed image, flame area recognition is performed, including: grayscale recognition of each pixel in the binarized image, and the areas connected to the pixels with grayscale lower than the preset grayscale threshold are all used as background areas; the background area is deducted, and the area is connected based on grayscale similarity in the remaining image information to obtain multiple independent calibration areas, and these calibration areas are used as flame areas.

[0066] In an embodiment of the present invention, in order to reduce the impact of noise on image quality and avoid misidentifying image noise as flame information, it is necessary to perform denoising on the image before flame recognition. Then the denoised image is binarized. Image binarization is to set the grayscale value of the pixel on the image to 0 or 255, that is, to present the entire image with an obvious black and white effect. There are differences in the color, brightness, etc. of the original image, which will eventually be distinguished by the grayscale value on the binary image. Because the scheme of the present invention is to collect flame images in the furnace, the background is pure black. In this case, the background color is relatively simple and the grayscale difference will be very large. This method can easily distinguish the flame area and realize the rapid positioning of the flame area. Because different furnace nozzles have separate flames, there will also be multiple identifiable flame areas, and each flame area corresponds to at least one furnace nozzle.

[0067] Specifically, three-dimensional rubbing is performed based on pre-constructed furnace three-dimensional information and identified flame areas, and the furnace nozzle serial number belonging to each flame area is identified, including: determining the image monitoring device from which the image information corresponding to the flame area identification result comes, and reading the posture information of the image monitoring device; determining the field of view information of the corresponding image monitoring device based on the posture information; based on the field of view information, three-dimensional rubbing is performed in the pre-constructed furnace three-dimensional information, and image calibration information under the corresponding field of view is projected; based on the image calibration information, the furnace nozzle serial number belonging to each flame area is calibrated for the image that determines the flame area identification result.

[0068] In an embodiment of the present invention, in order to realize the association between the flame area and the furnace nozzle in the field of view images of different image monitoring devices, the scheme of the present invention performs three-dimensional rubbing based on the three-dimensional information of the furnace, and reversely rubs the furnace nozzle information based on the collected image as an implicit two-dimensional image. The orientation of the monitoring device determines the viewing angle, and the focal length of the image monitoring device determines the size of its field of view. To read the posture information of the image monitoring device, it is necessary to read the orientation information and focal length information at the same time, and then simulate the image information of the corresponding field of view size under the corresponding field of view. Perform three-dimensional rubbing based on the three-dimensional information of the furnace, find the furnace nozzle information under the field of view, and determine the furnace nozzle serial number.

[0069] Preferably, the method further comprises: constructing three-dimensional information of the furnace, including: collecting design information of the refining heating furnace, and reading structural information of the furnace in the design information; importing the structural information into MATLAB to obtain three-dimensional information of the furnace.

[0070] Specifically, based on the pre-built abnormality recognition model, the abnormal probability of the flame of each furnace nozzle is evaluated one by one to obtain the abnormal probability prediction result corresponding to each furnace nozzle. The method also includes: constructing an abnormality recognition model, including: collecting historical image information of the furnace nozzle flame, the historical image information includes: normal flame image information and annotated abnormal flame image information; using the historical image information as a training sample, performing model training in a pre-neural network, and obtaining an abnormality recognition model.

[0071] The positioning unit is used to locate the furnace nozzles whose abnormal probability is greater than a preset threshold value to form an abnormal furnace nozzle set.

[0072] Specifically, after obtaining the abnormal probability assessment result of each furnace nozzle, the abnormal assessment result of each furnace nozzle is compared with the preset threshold. When it is judged that the abnormal probability assessment result of the furnace nozzle is greater than the preset threshold, it means that the furnace nozzle has abnormal combustion, and it is defined as an abnormal furnace nozzle.

[0073] Preferably, because there are multiple image monitoring devices in the furnace, multiple image monitoring devices may have flame image information of the same furnace nozzle at the same time. Because the image acquisition is performed from different postures, the flame state of the same furnace nozzle may be different in the image information of different image monitoring devices. This makes it possible that the same furnace nozzle is judged to be abnormal and normal at the same time. In this case, as long as a furnace nozzle is judged to be abnormal in the image of any image monitoring device, in order to avoid missing abnormal conditions, even if it is judged to be normal in the images of other image monitoring devices, it still needs to be identified as an abnormal furnace nozzle.

[0074] The scheme generating unit is used to generate a posture adjustment scheme for the array of image monitoring devices based on the position of each abnormal furnace nozzle in the abnormal furnace nozzle set.

[0075] Specifically, the serial numbers of the furnace nozzles with abnormal probabilities greater than a preset threshold are identified, and the position information of each abnormal furnace nozzle is located in the refining heating furnace design information based on the determined serial numbers of the furnace nozzles; based on the position information of each abnormal furnace nozzle, the priority information of each image monitoring device relative to each abnormal furnace nozzle is identified in the preferred image monitoring device comparison table; wherein, the higher the priority, the better the field of view of the image information of the corresponding abnormal furnace nozzle collected by the corresponding image monitoring device; based on the priority information of each image monitoring device relative to each abnormal furnace nozzle, image monitoring devices are allocated in the image monitoring device array for each abnormal furnace nozzle, and each abnormal furnace nozzle is allocated with at least two image monitoring devices; each image monitoring device generates a posture adjustment plan for the corresponding abnormal furnace nozzle based on the allocation result.

[0076] In the embodiment of the present invention, because the image monitoring devices are set at different positions, the optimal viewing angles of the image monitoring devices are different. Even if some furnace nozzles appear in the field of view of a certain image monitoring device, because they may greatly deviate from the optimal viewing angle of the device, their image information is distorted to a certain extent. In this case, the accuracy of the flame status monitoring of these furnace nozzles based on the image monitoring device cannot be guaranteed. Based on this, when the image monitoring device is set in the scheme of the present invention, it stores the corresponding preferred image monitoring device comparison table. When it is necessary to focus on the abnormal furnace nozzles in the future, it is necessary to select the preferred image monitoring device for image acquisition to ensure that the abnormal furnace nozzle is within the optimal viewing angle of the assigned image monitoring device, avoid image distortion, and improve monitoring accuracy.

[0077] Preferably, each image monitoring device generates a posture adjustment plan for the corresponding abnormal furnace nozzle based on the allocation result, including: if an image monitoring device is assigned to multiple abnormal furnace nozzles at the same time, then the supervision operation order of the image monitoring device is determined based on the priority between the image monitoring device and each abnormal furnace nozzle, and the higher the priority, the higher the supervision operation order of the image monitoring device relative to the abnormal furnace nozzle; if an image monitoring device is assigned to multiple abnormal furnace nozzles at the same time, and the image monitoring device has the same priority as at least two of the abnormal furnace nozzles, then the abnormal probabilities of the abnormal furnace nozzles with the same priority are compared, and the greater the abnormal probability, the higher the priority, and the higher the supervision operation order of the image monitoring device relative to the abnormal furnace nozzle; based on the determined supervision operation order, a field of view alignment order and a focusing plan are generated for each image monitoring device, and the field of view alignment order and the focusing plan are used as the posture adjustment plan.

[0078] Preferably, the method of generating a field of view alignment sequence and a focusing scheme for each image monitoring device based on the determined supervision operation sequence includes: generating a sequential alignment scheme for each image monitoring device relative to each abnormal furnace nozzle based on the determined supervision operation sequence, and generating an alignment timing table based on the sequential alignment scheme; determining the distance information between the abnormal furnace nozzle corresponding to each alignment moment and the assigned image monitoring device, and determining the focusing information at each alignment moment based on the distance information, and sequentially adding the focusing information to the alignment timing table to obtain the corresponding field of view alignment sequence and focusing scheme.

[0079] In the embodiment of the present invention, the scheme of the present invention aims to realize the key supervision of abnormal furnace nozzles. In addition to processing and selecting the image monitoring device with the best viewing angle, in order to avoid interference from the flames of other furnace nozzles, it is ensured that only the flame information of the target furnace nozzle exists in the field of view, which is the expected state. Based on this, when the scheme of the present invention performs key supervision, the image monitoring device will be automatically aligned and focused to achieve the only monitoring of abnormal furnace nozzles and ensure the monitoring accuracy. However, because different image monitoring devices cannot serve only one furnace nozzle, it is necessary to generate a corresponding timing table to perform sequential key supervision based on priority or abnormality probability, while ensuring accuracy, reducing the delay of the overall solution and avoiding missing key abnormalities.

[0080] The execution unit is used to execute the posture adjustment scheme and perform posture adjustment of each image monitoring device in the image monitoring array.

[0081] Specifically, based on the field of view alignment sequence and the focusing scheme, the posture adjustment of each image monitoring device is synchronously executed to realize the monitoring of all abnormal furnace nozzles, thereby achieving secondary key supervision after identifying suspected abnormalities and avoiding false alarms caused by misidentification.

[0082] Preferably, the system also includes a processing unit, which is used to increase the sampling frequency on the basis of the original sampling frequency after completing the posture adjustment of each image monitoring device, and complete a preset period of flame image acquisition corresponding to the abnormal furnace nozzle as a supervision image; perform abnormality probability evaluation on the supervision image again, if the abnormality probability evaluation result is still greater than the preset threshold, output an alarm message; otherwise, the abnormal furnace nozzle is corrected to a normal furnace nozzle.

[0083] The embodiment of the present invention further provides a computer-readable storage medium, on which instructions are stored, which, when executed on a computer, enable the computer to execute the above-mentioned image monitoring device control method.

[0084] Those skilled in the art will understand that all or part of the steps in the method for implementing the above-mentioned embodiments can be completed by instructing the relevant hardware through a program, and the program is stored in a storage medium, including several instructions for making a single-chip microcomputer, a chip or a processor (processor) perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0085] The optional embodiments of the present invention are described in detail above in conjunction with the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details in the above embodiments. Within the technical concept of the embodiments of the present invention, the technical scheme of the embodiments of the present invention can be subjected to a variety of simple modifications, and these simple modifications all belong to the protection scope of the embodiments of the present invention. It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. In order to avoid unnecessary repetition, the embodiments of the present invention will not further describe various possible combinations.

[0086] In addition, various embodiments of the present invention may be arbitrarily combined, and as long as they do not violate the concept of the embodiments of the present invention, they should also be regarded as the contents disclosed in the embodiments of the present invention.

Claims

1. An image monitoring device control method, applied to the image monitoring device control in a refining heating furnace, characterized in that: The method comprises: Collecting real-time image information of each furnace nozzle flame, and evaluating the probability of abnormal state of each furnace nozzle flame based on the image information; Locate and evaluate the furnace nozzles whose abnormal probability is greater than a preset threshold to form an abnormal furnace nozzle set; Based on the position of each abnormal furnace nozzle in the abnormal furnace nozzle set, a posture adjustment plan for the image monitoring device array is generated; The posture adjustment scheme is executed to perform posture adjustment of each image monitoring device in the image monitoring array.

2. The method according to claim 1, characterized in that The abnormal probability assessment of each furnace nozzle flame state based on the image information includes: Preprocessing the image information to obtain a preprocessed image; Perform flame area recognition on the preprocessed image; Based on the pre-built 3D information of the furnace and the identified flame area, 3D printing is performed to identify the nozzle serial number to which each flame area belongs; Based on the pre-built anomaly recognition model, the abnormal probability of the flame of each furnace nozzle is evaluated one by one to obtain the abnormal probability prediction result of each furnace nozzle.

3. The method according to claim 2, characterized in that The preprocessing of the image information to obtain the preprocessed image information includes: Perform denoising on the image information to obtain a denoised image; The denoised image is binarized to obtain a binarized image, and the binarized image is used as a preprocessed image.

4. The method according to claim 3, characterized in that The flame region recognition of the pre-processed image includes: The grayscale of each pixel is identified in the binary image, and the area connected by the pixels whose grayscale is lower than the preset grayscale threshold is regarded as the background area; The background area is removed, and the regions in the remaining image are connected based on grayscale similarity to obtain multiple independent calibration regions, which are used as flame regions.

5. The method according to claim 2, characterized in that: Based on the pre-built 3D information of the furnace and the identified flame area, 3D printing is performed to identify the nozzle number of each flame area, including: Determine the image monitoring device from which the image information corresponding to the flame area recognition result comes, and read the posture information of the image monitoring device; Determine the field of view information of the corresponding image monitoring device based on the posture information; Based on the field of view information, three-dimensional rubbing is performed in the pre-constructed three-dimensional information of the furnace, and image calibration information under the corresponding field of view is projected; Based on the image calibration information, the nozzle serial number of each flame area of ​​the image that determines the flame area recognition result is calibrated.

6. The method according to claim 2, characterized in that The method further comprises: Construct furnace 3D information, including: Collecting design information of a refining heating furnace and reading structural information of the furnace from the design information; The structural information is imported into MATLAB to obtain the three-dimensional information of the furnace.

7. The method according to claim 2, characterized in that: The method further comprises: Build an anomaly recognition model, including: Collecting historical image information of the flame of the furnace nozzle, wherein the historical image information includes: normal flame image information and marked abnormal flame image information; The historical image information is used as a training sample, and model training is performed in a preset neural network to obtain an anomaly recognition model.

8. The method according to claim 1, characterized in that: Based on the position of each abnormal furnace nozzle in the abnormal furnace nozzle set, a posture adjustment plan for the image monitoring device array is generated, including: Identify the nozzle serial numbers of the furnaces with abnormal probability greater than a preset threshold, and locate the positions of the abnormal nozzles in the design information of the refining heating furnace based on the determined nozzle serial numbers; Based on the position of each abnormal furnace nozzle, in the preferred image monitoring device comparison table, the priority of each image monitoring device relative to each abnormal furnace nozzle is identified; wherein the higher the priority, the better the field of view of the image information of the abnormal furnace nozzle collected by the corresponding image monitoring device; Based on the priority of each image monitoring device relative to each abnormal furnace nozzle, image monitoring devices are allocated in the image monitoring device array for each abnormal furnace nozzle, and each abnormal furnace nozzle is allocated with at least two image monitoring devices; Each image monitoring device generates a posture adjustment plan for the corresponding abnormal furnace nozzle based on the allocation result.

9. The method according to claim 8, characterized in that Each image monitoring device generates a posture adjustment plan for the corresponding abnormal furnace nozzle based on the allocation results, including: If an image monitoring device is assigned to multiple abnormal furnace nozzles at the same time, the supervision operation order of the image monitoring device is determined based on the priority between the image monitoring device and each abnormal furnace nozzle. The higher the priority, the higher the supervision operation order of the image monitoring device relative to the abnormal furnace nozzle. If an image monitoring device is assigned to multiple abnormal furnace nozzles at the same time, and the image monitoring device has the same priority as at least two of the abnormal furnace nozzles, then the abnormal probabilities of the abnormal furnace nozzles with the same priority are compared. The greater the abnormal probability, the higher the supervision operation order of the image monitoring device relative to the abnormal furnace nozzle; Based on the determined supervision operation sequence, a field of view alignment sequence and a focus scheme are generated for each image monitoring device, and the field of view alignment sequence and the focus scheme are used as a posture adjustment scheme.

10. The method according to claim 9, characterized in that The method of generating a field of view alignment sequence and a focusing scheme for each image monitoring device based on the determined supervision operation sequence includes: Based on the determined supervision operation sequence, a sequential alignment scheme of each image monitoring device relative to each abnormal furnace nozzle is generated, and an alignment timing table is generated based on the sequential alignment scheme; Determine the distance information between the abnormal furnace nozzle and the assigned image monitoring device corresponding to each alignment moment, and determine the focusing information of each alignment moment based on the distance information, and add the focusing information to the alignment timing table in sequence to obtain the corresponding field of view alignment sequence and focusing scheme.

11. The method according to claim 1, characterized in that: The method further comprises: After completing the posture adjustment of each image monitoring device, the sampling frequency is increased on the basis of the original sampling frequency, and the flame image collection of the corresponding abnormal furnace nozzle is completed for a preset period as the monitoring image; The abnormal probability evaluation is performed again on the supervision image. If the abnormal probability evaluation result is still greater than the preset threshold, an alarm message is output; otherwise, the abnormal furnace nozzle is corrected to a normal furnace nozzle.

12. An image monitoring equipment control system, applied to the image monitoring equipment control in a refining heating furnace, characterized in that: The system comprises: A collection unit, used to collect real-time image information of each furnace nozzle flame, and to evaluate the probability of abnormal state of each furnace nozzle flame based on the image information; A positioning unit, used for locating furnace nozzles whose abnormal probability is greater than a preset threshold, to form an abnormal furnace nozzle set; A scheme generating unit, used for generating a posture adjustment scheme of the image monitoring device array based on the position of each abnormal furnace nozzle in the abnormal furnace nozzle set; The execution unit is used to execute the posture adjustment scheme and perform posture adjustment of each image monitoring device in the image monitoring array.

13. A computer-readable storage medium having instructions stored thereon, which, when executed on a computer, enables the computer to execute the image monitoring device control method according to any one of claims 1 to 11.

Images