A zinc rotary kiln working condition intelligent monitoring method and device, terminal and medium

Abstract

The application discloses a kind of zinc rotary kiln working condition intelligent monitoring method, device, terminal and medium, comprising: obtaining the flame video data of kiln head crater collected by industrial camera;According to the preset time length, the key combustion characteristics of material in flame video data are extracted;According to the preset time length, the corresponding key monitoring data are read from equipment monitoring system;The key combustion characteristics and the key monitoring data are combined to form observation parameter matrix according to different arrangement order;Based on convolutional neural network, the fusion features of the observation parameter matrix are calculated, and the kiln condition information is output according to the fusion features.The application obtains the running state of volatile kiln by computer vision and artificial intelligence technology, solves the problem of large time lag and randomness in traditional manual observation of flame and monitoring data based on experience to judge working condition.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of metallurgical equipment modeling and application, and particularly relates to a zinc rotary kiln working condition intelligent monitoring method and device, a terminal and a medium. BACKGROUND

[0002] Zinc is an important raw material for the development of national economy and national defense industry in China. Wet metallurgical zinc is the main method of modern zinc smelting, and a large amount of leaching slag containing zinc and other valuable metals is generated in the process. At present, the high temperature environment generated by zinc oxide rotary volatilization kiln is used to realize the treatment and recovery of the mixture of leaching slag and coke. At present, the volatilization kiln is adjusted by manually observing the flame at the kiln head and related monitoring data. This way of judging the kiln condition by relying on manual experience has great randomness and hysteresis, which makes it difficult to run low-carbon and efficiently, and the quality of equipment operation fluctuates greatly.

[0003] Although the development of flame image acquisition and visual analysis technology provides data support for the kiln condition perception of the zinc rotary kiln, it still faces problems such as incomplete extraction of kiln head material combustion characteristics and difficulty in effective fusion of heterogeneous data, which leads to difficulty in accurately identifying the working condition. Comprehensive perception of the combustion state in the kiln, and fusion of the key operating parameters of the volatilization kiln, so as to accurately identify the kiln condition, is the key basis for realizing stable and efficient operation of the zinc rotary kiln.

[0004] Therefore, the prior art still needs to be improved. SUMMARY

[0005] The technical problem to be solved by the present application is that, in view of the defects of the prior art, the present application provides a zinc rotary kiln working condition intelligent monitoring method, device, terminal and medium to solve the problems of large time lag and randomness in judging the working condition by relying on experience of traditional manual observation of flame and monitoring data.

[0006] The technical scheme adopted by the present application to solve the technical problem is as follows:

[0007] In a first aspect, the present application provides a zinc rotary kiln working condition intelligent monitoring method, comprising:

[0008] acquiring flame video data of a kiln head fire hole collected by an industrial camera;

[0009] extracting key combustion characteristics of the material in the flame video data according to a preset time length;

[0010] reading corresponding key monitoring data from an equipment monitoring system according to the preset time length;

[0011] combining the key combustion characteristics and the key monitoring data according to different arrangement orders to form an observation parameter matrix;

[0012] Calculate a fusion feature of the observation parameter matrix based on a convolutional neural network, and output kiln condition information according to the fusion feature.

[0013] In an implementation manner, the flame video data of the kiln head fire hole collected by the industrial camera includes the following steps:

[0014] Adjust the position, orientation and focal length of the industrial camera, and control the industrial camera to face the kiln head fire hole.

[0015] In an implementation manner, the key combustion features of the material in the flame video data are extracted according to a preset time length, including:

[0016] Select a preset time length of video, and obtain an image set from the preset time length of video according to a preset rule;

[0017] According to a pre-defined dynamic feature calculation method, the key combustion features of the material are extracted from the image set.

[0018] In an implementation manner, the key combustion features of the material are extracted from the image set according to a pre-defined dynamic feature calculation method, including:

[0019] Calculate the dynamic texture features, optical flow features and flicker frequencies corresponding to each image in the image set to obtain the combustion features of each image;

[0020] According to the correlation between the combustion features of each image and the operation data at the corresponding time point, the key combustion features of the material are screened.

[0021] In an implementation manner, the key monitoring data includes historical monitoring data and control data, and current monitoring data and control data.

[0022] In an implementation manner, the corresponding key monitoring data is read from the equipment monitoring system according to the preset time length, including:

[0023] According to the preset time length, historical monitoring data and control data are read from the equipment monitoring system, and current monitoring data and control data are read.

[0024] In an implementation manner, the key combustion features and the key monitoring data are combined to form an observation parameter matrix according to different arrangement orders, including:

[0025] The key combustion features and the key monitoring data are respectively converted into corresponding splicing matrices;

[0026] All obtained splicing matrices are arranged according to a preset arrangement order along a non-time dimension to form different splicing matrices;

[0027] combining all the sorted splicing matrices into the observation parameter matrix.

[0028] In an implementation, the separately transforming the key combustion features and the key monitoring data into corresponding splicing matrices comprises:

[0029] transforming the key combustion features and the key monitoring data into corresponding time-dimension matrices respectively, and splicing the transformed time-dimension matrices into a corresponding splicing matrix along a corresponding non-time dimension.

[0030] In an implementation, the calculating the fusion feature of the observation parameter matrix based on the convolutional neural network comprises:

[0031] setting a training sample set of the convolutional neural network according to historical monitoring data and control data, and training the convolutional neural network by taking each training sample in the training sample set as an input of an observation parameter matrix and taking a corresponding time condition label as an output.

[0032] In an implementation, the calculating the fusion feature of the observation parameter matrix based on the convolutional neural network, and outputting kiln condition information according to the fusion feature comprises:

[0033] inputting the observation parameter matrix into the convolutional neural network to calculate a fusion feature of the observation parameter matrix;

[0034] outputting a corresponding time condition label of the material according to the fusion feature; wherein the time condition label comprises: underburning second level, underburning first level, normal, overburning first level, and overburning second level.

[0035] In a second aspect, the present application provides a zinc rotary kiln working condition intelligent monitoring device, comprising:

[0036] a data acquisition module configured to acquire flame video data of a kiln head fire hole collected by an industrial camera;

[0037] a key combustion feature module configured to extract key combustion features of a material in the flame video data according to a preset time length;

[0038] a key monitoring data module configured to read corresponding key monitoring data from an equipment monitoring system according to the preset time length;

[0039] an observation parameter matrix module configured to combine the key combustion features and the key monitoring data into an observation parameter matrix according to different arrangement orders;

[0040] a kiln condition information module configured to calculate a fusion feature of the observation parameter matrix based on a convolutional neural network, and output kiln condition information according to the fusion feature.

[0041] In a third aspect, the present application provides a terminal, comprising a processor and a memory, the memory storing a zinc rotary kiln working condition intelligent monitoring program, the zinc rotary kiln working condition intelligent monitoring program being used to implement the operations of the zinc rotary kiln working condition intelligent monitoring method according to the first aspect when executed by the processor.

[0042] In a fourth aspect, the present application further provides a medium, which is a computer readable storage medium, the medium storing a zinc rotary kiln working condition intelligent monitoring program, the zinc rotary kiln working condition intelligent monitoring program being used to implement the operations of the zinc rotary kiln working condition intelligent monitoring method according to the first aspect when executed by a processor.

[0043] The technical scheme of the present application has the following effects:

[0044] The present application converts the key combustion features and the key monitoring data into the form of matrices respectively, and combines them according to different arrangement orders, so as to not only fully integrate the information from two different sources, but also make full use of the advantage of the convolutional neural network that can well analyze the spatial relationship of data, thereby more effectively extracting and fusing features to improve the working condition recognition accuracy. Meanwhile, the present application only needs to collect video images from the fire hole at the kiln head and obtain the state monitoring data of the rotary kiln from the monitoring system, without the need to touch the main structure of the kiln and without affecting the normal operation of the on-site workers, so as to facilitate on-site deployment and application, and to provide guidance for on-site workers to operate and realize low-carbon and efficient operation of the zinc rotary kiln, thereby solving the problems of large time lag and randomness in the traditional manual observation of flame and monitoring data and the experience-based judgment of working conditions. BRIEF DESCRIPTION OF DRAWINGS

[0045] In order to more clearly illustrate the technical schemes in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained from the structures shown in the drawings without creative labor.

[0046] Figure 1 is a flowchart of a zinc rotary kiln working condition intelligent monitoring method in an implementation manner of the present application.

[0047] Figure 2 is a structural diagram of a working condition monitoring device in an implementation manner of the present application.

[0048] Figure 3 is an example diagram of a combustion feature calculation method in an implementation manner of the present application.

[0049] Figure 4It is a combination mode schematic diagram of key combustion features and key monitoring data in an implementation manner of the present application.

[0050] Figure 5 It is a functional schematic diagram of a terminal in an implementation manner of the present application.

[0051] The implementation, functional features and advantages of the present application will be further described with reference to the accompanying drawings and embodiments. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solutions and advantages of the present application more clear and explicit, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0053] Exemplary method

[0054] The development of existing flame image acquisition and visual analysis technology provides data support for the kiln condition perception of the zinc rotary kiln, but still faces problems such as incomplete extraction of kiln head material combustion features, difficulty in effective fusion of heterogeneous data, and the like, thereby leading to difficulty in accurately identifying the working condition. Comprehensive perception of the combustion state in the kiln, and fusion of the key operating parameters of the volatile kiln, thereby accurately identifying the kiln condition, is the key basis for realizing stable and efficient operation of the zinc rotary kiln.

[0055] In view of the above technical problems, the present embodiment provides a zinc rotary kiln working condition intelligent monitoring method, which is a zinc rotary kiln working condition intelligent monitoring method based on multi-source information fusion. This method can effectively fuse multi-source heterogeneous flame video and state monitoring data, thereby accurately identifying the current operating state of the rotary kiln and providing guidance for on-site worker operation. Moreover, computer vision and artificial intelligence technology are used to obtain the operating state of the volatile kiln, without the need to contact the main structure of the rotary kiln and without affecting the normal operation of the workers, thereby facilitating on-site deployment and application and long-term monitoring.

[0056] As shown in Figure 1 The present embodiment provides a zinc rotary kiln working condition intelligent monitoring method, which includes the following steps:

[0057] Step S100, acquiring flame video data of a kiln head fire hole collected by an industrial camera.

[0058] In the present embodiment, the zinc rotary kiln working condition intelligent monitoring method is applied to a terminal, which includes but is not limited to a computer and the like.

[0059] In the zinc rotary kiln working condition intelligent monitoring method, first, the visible light industrial camera is used to collect the flame video from the fire hole of the kiln head and transmit it to the edge processor; then, the key combustion characteristics of the material in the flame video are extracted every preset time interval, and the key monitoring data is read from the equipment monitoring system every preset time interval; finally, the key combustion characteristics and the key monitoring data are combined to form an observation parameter matrix according to different arrangement orders, and the fusion features of the observation parameter matrix are calculated based on the convolutional neural network, and the kiln condition information is transmitted from the edge processor to the host computer.

[0060] Before implementing the zinc rotary kiln working condition intelligent monitoring method, the installation position, orientation and focal length of the industrial camera need to be adjusted so that it can shoot clear material combustion flame video from the rotary kiln head.

[0061] Specifically, in one implementation manner of the embodiment, the step S100 includes the following steps:

[0062] Step S001, adjust the position, orientation and focal length of the industrial camera, control the industrial camera to face the fire hole of the kiln head.

[0063] In the embodiment, as shown in the figure, Figure 2 The zinc rotary kiln working condition intelligent monitoring system in the embodiment includes a zinc rotary kiln 1, a kiln head cover 2, a fire hole 3, an industrial camera 4, a bracket 5, an edge processor 6 and a host computer 7; wherein the industrial camera 4 is a visible light industrial camera.

[0064] In the process of adjusting the position, orientation and focal length of the industrial camera, first, the industrial camera 4 is installed on the kiln head cover 2 of the zinc rotary kiln 1 to look at one side of the fire hole 3, wherein the industrial camera 4 is fixed on the bracket 5; then, the installation position, orientation and focal length of the industrial camera 4 are adjusted so that it can shoot clear flame combustion images from the fire hole 3.

[0065] In the embodiment, after adjusting the industrial camera, the frame rate parameters and transmission path of the industrial camera 4 are configured, for example, the industrial camera 4 is set to collect material combustion flame video at a rate of 25 frames per second in real time, and transmit it to the edge processor 6 for processing, so as to extract the key combustion characteristics of the material in the flame video data through the edge processor 6.

[0066] As shown in the figure, Figure 1 In one implementation manner of the embodiment of the application, the zinc rotary kiln working condition intelligent monitoring method further includes the following steps:

[0067] Step S200, extract the key combustion characteristics of the material in the flame video data according to the preset time interval.

[0068] In the embodiment, after collecting the flame video data of the kiln head fire hole, the edge processor extracts the key combustion features of the material in the flame video every preset time length, wherein the extraction process is as follows: first, select a video of a preset time length; then, obtain an image set from the video of the preset time length according to a preset rule; finally, extract the key combustion features of the material from the image set according to a pre-defined dynamic feature calculation method.

[0069] Specifically, in an implementation manner of the embodiment, step S200 includes the following steps:

[0070] Step S201: selecting a video of a preset time length, and obtaining an image set from the video of the preset time length according to a preset rule;

[0071] Step S202: extracting the key combustion features of the material from the image set according to a pre-defined dynamic feature calculation method.

[0072] In an implementation manner, taking 5 minutes as the preset time length, the flame video within 5 minutes from the current time is obtained first; then, a certain number (for example, 300 frames) of video images are extracted from the flame video of each minute at equal time intervals, and an image block (for example, the image block where the flame is located) is cropped from each image according to a preset region of interest, so that 300 image blocks of each minute constitute an image set; finally, n kinds of key combustion features of the material are extracted from each minute image set according to a pre-defined dynamic feature calculation method.

[0073] Specifically, in an implementation manner of the embodiment, step S202 includes the following steps:

[0074] Step S202-1: calculating the dynamic texture features, the optical flow features and the flicker frequency corresponding to each image in the image set to obtain the combustion features of each image;

[0075] Step S202-2: screening the key combustion features of the material according to the correlation between the combustion features of each image and the operation data of the corresponding time point.

[0076] In the embodiment, the n kinds of key combustion features of the material are determined by the following manner.

[0077] As Figure 3As shown: First, a relatively long historical flame video is selected, and an image set is constructed from 300 image blocks per minute; second, operational data within the same time period as the historical flame video is obtained from the rotary kiln monitoring system, namely the material feed rate and rotation speed of the rotary kiln; third, a series of dynamic feature calculation methods are defined based on the workers' experience in observing the fire to obtain various combustion features in the image set per minute; finally, the correlation between each combustion feature and the operational data is calculated (for example, using the maximum mutual information coefficient MIC to calculate the correlation), and n combustion features with high correlation are selected according to a preset threshold (for example, 0.2), which are the key combustion features of the material.

[0078] In one implementation of this embodiment, the series of dynamic feature calculation methods defined based on the worker's experience in observing fire include various combustion features such as texture features, optical flow features, and flicker frequency.

[0079] This embodiment uses dynamic texture features and optical flow features as examples to illustrate the calculation process of combustion features, such as... Figure 3 As shown:

[0080] (1) Calculate the dynamic texture features of the image set: First, convert the image set into grayscale images and stitch them together into a three-dimensional matrix along the time dimension. Then, use the gray-level co-occurrence matrix (GLCM) to calculate the texture feature values ​​such as mean, variance, contrast, and correlation of the three-dimensional matrix.

[0081] (2) Calculate the optical flow characteristics of the image set: First, the dense optical flow between each pair of adjacent images is calculated using the DIS optical flow method proposed by Till Kroeger. Then, the mean of the optical flow magnitude and direction between each pair of images is calculated. Finally, the mean and variance of the optical flow magnitude and direction of the entire image set are calculated.

[0082] like Figure 1 As shown, in one implementation of this invention, the intelligent monitoring method for zinc rotary kiln operating conditions further includes the following steps:

[0083] Step S300: Read the corresponding key monitoring data from the equipment monitoring system according to the preset duration.

[0084] In this embodiment, key monitoring data is read from the intelligent monitoring system for the zinc rotary kiln at preset intervals. The preset interval is the same as the preset interval for flame video extraction in the previous step S200. The intelligent monitoring system for the zinc rotary kiln in this embodiment can collect equipment operating status data and perform real-time control in real time. Therefore, key monitoring data of the equipment can be read from the intelligent monitoring system for the zinc rotary kiln. The key monitoring data includes: historical monitoring data and control data, and current monitoring data and control data.

[0085] Specifically, in one implementation of this embodiment, step S300 includes the following steps:

[0086] Step S301: Read historical monitoring data and control data from the equipment monitoring system according to the preset duration, and read the monitoring data and control data at the current moment.

[0087] In this embodiment, the intelligent monitoring system for the zinc rotary kiln can be a DCS control system. This system can read m types of key monitoring data of the rotary kiln within the same preset time period (e.g., 5 minutes) at the same moment as the flame video in step S200. The read monitoring and control data include monitoring variables such as kiln head temperature, kiln tail temperature, and carbon monoxide concentration. One data point is read per minute for each key monitoring variable. The intelligent monitoring system for the zinc rotary kiln includes many monitoring variables; key monitoring variables can be determined through worker experience or by calculating the correlation between all monitoring variables and operational data.

[0088] like Figure 1 As shown, in one implementation of this invention, the intelligent monitoring method for zinc rotary kiln operating conditions further includes the following steps:

[0089] Step S400: Combine the key combustion features and the key monitoring data in different arrangements to form an observation parameter matrix.

[0090] In this embodiment, after obtaining the key combustion features and key monitoring data, the key combustion features and key monitoring data are combined in different arrangements to form an observation parameter matrix. Specifically, the key combustion features and key monitoring data are each converted into two matrices of the same length in the time dimension, and then spliced ​​together along another dimension, called the non-time dimension, to form a single matrix. The spliced ​​matrix is ​​then arranged in different preset orders along the non-time dimension to form different spliced ​​matrices. All the spliced ​​matrices are combined into a new matrix called the observation parameter matrix to more comprehensively integrate the combustion features and monitoring data.

[0091] Specifically, in one implementation of this embodiment, step S400 includes the following steps:

[0092] Step S401: Convert the key combustion features and the key monitoring data into corresponding splicing matrices.

[0093] In one implementation of this embodiment, step S401 includes the following steps:

[0094] Step S401-1, respectively, the key combustion characteristics and the key monitoring data into the corresponding time dimension matrix, and the converted time dimension matrix along the corresponding non-time dimension splicing into a corresponding splicing matrix.

[0095] As shown in the embodiment, in the process of converting into a splicing matrix, first, the n kinds of key combustion characteristics and the m kinds of key monitoring data within the preset time length are respectively converted into n*5 and m*5 matrices, where 5 represents a 5-minute time length; then, the two obtained matrices are spliced into an (n+m)*5 matrix along the dimension where the number of rows is located, and the splicing matrix is formed into different splicing matrices along the dimension where the number of rows is located according to a preset different arrangement order. Figure 4

[0096] Specifically, in one implementation manner of the embodiment, step S400 further includes the following steps:

[0097] Step S402, forming different splicing matrices along the non-time dimension according to the preset arrangement order for all obtained splicing matrices;

[0098] Step S403, combining all splicing matrices after sorting into the observation parameter matrix.

[0099] In the embodiment, the arrangement order in which the normal order and the reverse order are the same is only one kind, and then there are ((n+m)! / 2) different arrangement orders; the (n+m)! / 2 (n+m)*5 splicing matrices in different arrangement orders are combined into a two-dimensional observation parameter matrix with an integer multiple of n+m as the number of rows, an integer multiple of 5 as the number of columns, and the number of rows and the number of columns being closest, so as to more comprehensively fuse the combustion characteristics and the monitoring data.

[0100] As shown in the embodiment, in one implementation manner of the embodiment, the zinc rotary kiln working condition intelligent monitoring method further includes the following steps: Figure 1 Step S500, calculating the fusion features of the observation parameter matrix based on the convolutional neural network, and outputting the kiln condition information according to the fusion features.

[0101] Before the feature fusion of the observation parameter matrix, the convolutional neural network also needs to be trained, so as to obtain a convolutional neural network that can be used in an actual calculation scenario, so as to calculate the fusion features of the observation parameter matrix based on the trained convolutional neural network.

[0102] Specifically, in one implementation manner of the embodiment, step S500 includes the following steps before step S500:

[0103]

[0104] ​​Step S501a, set the training sample set of the convolutional neural network according to the historical monitoring data and control data, and train the convolutional neural network by taking the observation parameter matrix of each training sample in the training sample set as input and the working condition label of the corresponding time as output.

[0105] In the embodiment, the training sample set of the convolutional neural network is obtained from historical data, that is, the historical monitoring data and control data are taken as the training sample set, and in the training sample set, the observation parameter matrix corresponding to each training sample is taken as input and the working condition label of the corresponding time is taken as output; wherein the working condition label of each training sample is labeled by a worker with rich field experience, and the working condition label includes five categories, namely, under-burning secondary, under-burning primary, normal, over-burning primary, and over-burning secondary.

[0106] Specifically, in an implementation manner of the embodiment, step S500 includes the following steps:

[0107] Step S501, input the observation parameter matrix into the convolutional neural network to calculate the fusion feature of the observation parameter matrix.

[0108] Step S502, output the working condition label corresponding to the material according to the fusion feature.

[0109] In the process of real-time monitoring of the working condition of the zinc rotary kiln, the observation parameter matrix from the current time to the previous 5 minutes is input into the trained convolutional neural network, and the fusion feature corresponding to the observation parameter matrix is calculated. The working condition label corresponding to the material is output in the output layer of the trained convolutional neural network, that is, the working condition label corresponding to the material burning in the zinc rotary kiln is output.

[0110] In another implementation manner of the embodiment, the GLCM is used to calculate the dynamic texture feature and the DIS is used to calculate the optical flow feature. Other methods such as local binary pattern (LBP) can be used to calculate the texture feature, and Farneback algorithm and other methods can be used to calculate the optical flow feature. In addition, in the embodiment of the application, the texture feature and the optical flow feature are taken as examples, and other combustion features such as flicker frequency and static features can be used to replace them.

[0111] The above technical scheme of the embodiment achieves the following technical effects:

[0112] The embodiment converts the key combustion features and the key monitoring data into the form of matrices respectively, and combines them according to different arrangement orders, so as to not only fully integrate the information of two different sources, but also make full use of the advantage that the convolutional neural network can well analyze the spatial relationship of data, thereby more effectively extracting and fusing features to improve the working condition recognition accuracy. Meanwhile, the embodiment only needs to collect video images from the fire hole at the kiln head and acquire the state monitoring data of the rotary kiln from the monitoring system, without needing to contact the volatile kiln main structure and affecting the normal operation of the on-site workers, so as to facilitate on-site deployment and application, can long-term provide guidance for on-site workers to operate and realize low-carbon and efficient operation of the zinc rotary kiln, and solve the problems of large time lag and randomness in the traditional manual observation of the flame and the monitoring data to judge the working condition according to experience.

[0113] Exemplary device

[0114] Based on the above embodiment, the application further provides a zinc rotary kiln working condition intelligent monitoring device, comprising:

[0115] A data acquisition module is configured to acquire flame video data of a fire hole at a kiln head collected by an industrial camera.

[0116] A key combustion feature module is configured to extract key combustion features of materials in the flame video data according to a preset time length.

[0117] A key monitoring data module is configured to read corresponding key monitoring data from an equipment monitoring system according to the preset time length.

[0118] An observation parameter matrix module is configured to combine the key combustion features and the key monitoring data according to different arrangement orders to form an observation parameter matrix.

[0119] A kiln condition information module is configured to calculate fusion features of the observation parameter matrix based on a convolutional neural network, and output kiln condition information according to the fusion features.

[0120] Based on the above embodiment, the application further provides a terminal, and a principle block diagram thereof can be as shown in Figure 5 .

[0121] The terminal comprises a processor, a memory, an interface, a display screen and a communication module connected through a system bus; wherein the processor of the terminal is configured to provide computing and control capabilities; the memory of the terminal comprises a storage medium and an internal memory; the storage medium stores an operating system and a computer program; the internal memory provides an environment for the operating system and the computer program in the storage medium to run; the interface is configured to connect external devices, such as mobile terminals and computer devices; the display screen is configured to display corresponding information; and the communication module is configured to communicate with a cloud server or a mobile terminal.

[0122] The computer program is used to realize the operation of the zinc rotary kiln working condition intelligent monitoring method when executed by the processor.

[0123] Those skilled in the art can understand that, Figure 5 The principle block diagram shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the terminal to which the scheme of the present application is applied. The specific terminal can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement.

[0124] In one embodiment, a terminal is provided, comprising: a processor and a memory, the memory storing a zinc rotary kiln working condition intelligent monitoring program, the zinc rotary kiln working condition intelligent monitoring program being used to realize the operation of the zinc rotary kiln working condition intelligent monitoring method as above when executed by the processor.

[0125] In one embodiment, a storage medium is provided, wherein the storage medium stores a zinc rotary kiln working condition intelligent monitoring program, the zinc rotary kiln working condition intelligent monitoring program being used to realize the operation of the zinc rotary kiln working condition intelligent monitoring method as above when executed by the processor.

[0126] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by a computer program instructing related hardware. The computer program can be stored in a non-volatile storage medium. When executed, the computer program can include the processes of the above-mentioned embodiments. Any reference to memory, database or other medium used in the embodiments provided by the present application can include non-volatile and / or volatile memory.

[0127] To sum up, the present application provides a zinc rotary kiln working condition intelligent monitoring method, device, terminal and medium. The method comprises: acquiring flame video data of a kiln head fire hole collected by an industrial camera; extracting key combustion features of materials in the flame video data according to a preset time length; reading corresponding key monitoring data from an equipment monitoring system according to the preset time length; combining the key combustion features and the key monitoring data according to different arrangement orders to form an observation parameter matrix; calculating a fusion feature of the observation parameter matrix based on a convolutional neural network, and outputting kiln condition information according to the fusion feature. The present application uses computer vision and artificial intelligence technology to acquire the running state of a volatile kiln, and solves the problems of large time lag and randomness in the traditional manual observation of flame and monitoring data and the experience-based judgment of working conditions.

[0128] It should be understood that the application of the present application is not limited to the above examples. Those skilled in the art can make improvements or changes according to the above description, and all these improvements and changes shall fall within the protection scope of the claims of the present application.

Claims

1. A method for intelligent monitoring of zinc rotary kiln operation, characterized in that, include: Acquire video data of flames at the kiln head vent captured by an industrial camera; Extract key combustion features of materials from flame video data based on a preset duration; Read the corresponding key monitoring data from the equipment monitoring system according to the preset duration; The key monitoring data include one or more combinations of kiln head temperature, kiln tail temperature, and carbon monoxide concentration. The key combustion characteristics and the key monitoring data are combined in different arrangements to form an observation parameter matrix; The fusion features of the observation parameter matrix are calculated based on the convolutional neural network, and kiln condition information is output based on the fusion features; The step of combining the key combustion features and the key monitoring data in different arrangements to form an observation parameter matrix includes: The key combustion features and the key monitoring data are respectively converted into corresponding splicing matrices; All the obtained splicing matrices are arranged in a preset order along the non-time dimension to form different splicing matrices; All sorted concatenated matrices are combined to form the observation parameter matrix; The process of converting the key combustion features and key monitoring data into corresponding spliced ​​matrices includes: The key combustion features and key monitoring data are respectively converted into corresponding time dimension matrices, and the converted time dimension matrices are spliced ​​together along the corresponding non-time dimensions to form a corresponding spliced ​​matrix.

2. The intelligent monitoring method for zinc rotary kiln operating conditions according to claim 1, characterized in that, The acquisition of flame video data from the kiln head fire mouth captured by the industrial camera includes, prior to: Adjust the position, orientation, and focal length of the industrial camera to control the industrial camera to face the kiln head nozzle.

3. The intelligent monitoring method for zinc rotary kiln operating conditions according to claim 1, characterized in that, The step of extracting key combustion features of materials from flame video data according to a preset duration includes: Select a video of a preset duration, and obtain an image set from the video of the preset duration according to preset rules; Key combustion features of the material are extracted from the image set according to a predefined dynamic feature calculation method.

4. The intelligent monitoring method for zinc rotary kiln operating conditions according to claim 3, characterized in that, The step of extracting key combustion features of the material from the image set according to a predefined dynamic feature calculation method includes: The dynamic texture features, optical flow features, and flicker frequency corresponding to each image in the image set are calculated to obtain the combustion features of each image; Based on the correlation between the combustion characteristics of each image and the operation data at the corresponding time point, the key combustion characteristics of the material are selected.

5. The intelligent monitoring method for zinc rotary kiln operating conditions according to claim 1, characterized in that, The key monitoring data includes: historical monitoring and control data, and current monitoring and control data.

6. The intelligent monitoring method for zinc rotary kiln operating conditions according to claim 5, characterized in that, The step of reading the corresponding key monitoring data from the equipment monitoring system according to the preset duration includes: Based on the preset time interval, historical monitoring and control data are read from the equipment monitoring system, and the current monitoring and control data are also read.

7. The intelligent monitoring method for zinc rotary kiln operating conditions according to claim 1, characterized in that, The process of calculating the fusion features of the observation parameter matrix based on a convolutional neural network includes, prior to: The training sample set of the convolutional neural network is set based on historical monitoring and control data. Each training sample in the training sample set is used as input with the observation parameter matrix and as output with the corresponding time condition label to train the convolutional neural network.

8. The intelligent monitoring method for zinc rotary kiln operating conditions according to claim 1, characterized in that, The process of calculating the fusion features of the observation parameter matrix based on a convolutional neural network and outputting kiln condition information according to the fusion features includes: The observation parameter matrix is ​​input into the convolutional neural network to calculate the fusion features of the observation parameter matrix; The operating condition label corresponding to the material is output according to the fusion feature; wherein, the operating condition label includes: under-burning level 2, under-burning level 1, normal, over-burning level 1, and over-burning level 2.

9. A zinc rotary kiln operating condition intelligent monitoring device, used to implement the zinc rotary kiln operating condition intelligent monitoring method as described in any one of claims 1-8, characterized in that, include: The data acquisition module is used to acquire flame video data from the kiln head vent captured by an industrial camera; The key combustion feature module is used to extract key combustion features of materials from flame video data according to a preset duration. The key monitoring data module is used to read corresponding key monitoring data from the equipment monitoring system according to the preset duration. The observation parameter matrix module is used to combine the key combustion characteristics and the key monitoring data in different arrangements to form an observation parameter matrix. The kiln condition information module is used to calculate the fusion features of the observation parameter matrix based on a convolutional neural network, and output kiln condition information according to the fusion features.

10. A terminal, characterized in that, include: The processor and memory, wherein the memory stores a zinc rotary kiln intelligent monitoring program, and the zinc rotary kiln intelligent monitoring program, when executed by the processor, is used to implement the operation of the zinc rotary kiln intelligent monitoring method as described in any one of claims 1-8.

11. A medium, characterized in that, The medium is a computer-readable storage medium that stores a zinc rotary kiln intelligent monitoring program. When the zinc rotary kiln intelligent monitoring program is executed by a processor, it is used to implement the operation of the zinc rotary kiln intelligent monitoring method as described in any one of claims 1-8.

Images