CN113649422A - Hot image-based rough rolling billet quality detection system and method - Google Patents
Hot image-based rough rolling billet quality detection system and method Download PDFInfo
- Publication number
- CN113649422A CN113649422A CN202110741872.3A CN202110741872A CN113649422A CN 113649422 A CN113649422 A CN 113649422A CN 202110741872 A CN202110741872 A CN 202110741872A CN 113649422 A CN113649422 A CN 113649422A
- Authority
- CN
- China
- Prior art keywords
- billet
- steel
- steel billet
- quality
- infrared
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B21—MECHANICAL METAL-WORKING WITHOUT ESSENTIALLY REMOVING MATERIAL; PUNCHING METAL
- B21B—ROLLING OF METAL
- B21B38/00—Methods or devices for measuring, detecting or monitoring specially adapted for metal-rolling mills, e.g. position detection, inspection of the product
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
- B07C5/36—Sorting apparatus characterised by the means used for distribution
- B07C5/361—Processing or control devices therefor, e.g. escort memory
- B07C5/362—Separating or distributor mechanisms
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B21—MECHANICAL METAL-WORKING WITHOUT ESSENTIALLY REMOVING MATERIAL; PUNCHING METAL
- B21B—ROLLING OF METAL
- B21B38/00—Methods or devices for measuring, detecting or monitoring specially adapted for metal-rolling mills, e.g. position detection, inspection of the product
- B21B38/006—Methods or devices for measuring, detecting or monitoring specially adapted for metal-rolling mills, e.g. position detection, inspection of the product for measuring temperature
Landscapes
- Engineering & Computer Science (AREA)
- Mechanical Engineering (AREA)
- Radiation Pyrometers (AREA)
Abstract
The invention relates to a rough rolling billet quality detection system and method based on thermal images, and belongs to the technical field of ferrous metallurgy automatic control. The high-pressure water dephosphorization device is arranged between the heating furnace and the roughing mill group, the steel billet conveying device is arranged between the roughing mill group and the high-pressure water dephosphorization device, and the infrared thermal imager is arranged right above the steel billet conveying device; taking image point location, point location temperature infrared, steel type and temperature upper and lower limits of historical steel billets as input, and taking whether the corresponding steel billet quality is qualified as output to carry out machine learning model training; and then judging whether the quality of the steel billet is qualified or not according to the data acquired in real time, and if the quality of the steel billet is not qualified, rejecting the steel billet. The invention adopts a machine learning algorithm to analyze the relation among the image point location, the point location temperature, the upper and lower temperature limits and the quality of the steel grade and the steel billet, judges whether the steel billet meets the rough rolling requirement, is convenient and fast, has high accuracy, avoids the production loss and is easy to popularize and apply.
Description
Technical Field
The invention belongs to the technical field of ferrous metallurgy automatic control, and particularly relates to a rough rolling billet quality detection system and method based on thermal images.
Background
In the modern steel rolling production process, the temperature control of steel billets is a key technology, and is directly related to the normal operation of production and the product quality. The traditional contact type temperature measuring instrument is high in wear speed, spare parts need to be replaced frequently, production is stopped for maintenance, and meanwhile, a measured target in a steel rolling area is difficult to contact or moves at a high speed, so that the traditional contact type temperature measuring instrument cannot meet the requirement.
The infrared temperature measurement is an advanced non-contact temperature measurement means, and when a measured target is difficult to or forbidden to contact, is in some severe working conditions and moves rapidly, the infrared temperature measurement can ensure that the surface temperature of the target can be accurately measured, and the target is not damaged or influenced. Therefore, the steel rolling production process has stronger requirements on the infrared temperature measurement technology and is mainly applied to the aspects of heating furnaces, hot rolling rough rolling, finish rolling, descaling sections, hot rolling quenching temperature and the like. The infrared thermometer is selected in the hot rolling rough rolling process, so that effective, accurate and reliable temperature measurement can be provided for the steel billet, the heating temperature and the rolling temperature are controlled, the rolling parameters are optimized, and the quality of a final product is improved. The infrared thermal imager can detect the temperature of a heating point in real time, quantitatively and online within a certain distance, can also draw a temperature gradient thermal image through scanning, has high sensitivity and is not interfered by an electromagnetic field. The infrared thermal imaging detection technology can realize non-contact detection, shoot the temperature field distribution of the steel billet and measure the temperature value of any part, and has important significance for controlling the temperature in the hot rolling and rough rolling process and ensuring the product quality.
At present, the quality of rough rolling billets is mainly judged by a method of measuring point temperature by a thermodetector, and a rough rolling billet quality detection system and method based on thermal images are not found.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides a rough rolling billet quality detection system and method based on thermal images.
In order to achieve the purpose, the technical scheme adopted by the invention is as follows:
a rough rolling billet quality detection system based on thermal images is characterized in that a high-pressure water dephosphorization device is arranged between a heating furnace and a rough rolling unit, a billet conveying device is arranged between the rough rolling unit and the high-pressure water dephosphorization device, and an infrared thermal imager is arranged right above the billet conveying device; after the steel billet is discharged from the furnace, removing phosphorus by a high-pressure water phosphorus removal device, and acquiring thermal image data of the steel billet by an infrared thermal imager;
further comprising:
the data acquisition module is used for acquiring the steel grade of the steel billet;
the production database storage module is connected with the data acquisition module and the infrared thermal imager and used for storing the billet thermal infrared image data acquired by the infrared thermal imager and the corresponding steel grade;
the model training module is connected with the production database storage module and used for performing machine learning model training by taking the thermal infrared image data of the historical steel billet as input and the steel grade as output;
the billet removing device is arranged on the billet conveying device and is arranged behind the infrared thermal imager;
the billet steel removing control unit is connected with the data acquisition module, the infrared thermal imager, the model training module and the billet steel removing device and is used for inputting the billet steel type acquired by the data acquisition module in real time and the billet steel thermal infrared image data acquired by the infrared thermal imager in real time into a machine learning model obtained by training of the model training module, the output of the model is a probability value between 0 and 1 of the steel type, when the probability value is greater than a preset threshold value, the quality of the billet steel is judged to be qualified, otherwise, the quality of the billet steel is judged to be unqualified; if the output is that the quality of the steel billet is qualified, the steel billet removing control unit controls the steel billet removing device not to act, and if the output is that the quality of the steel billet is unqualified, the steel billet removing control unit controls the steel billet removing device to remove the steel billet.
Further, it is preferable that the infrared thermal imager is backlight-mounted.
Further, it is preferable that the infrared thermal imaging is an in-line infrared thermal imager.
Further, it is preferable that a closed air-cooled shield is provided outside the infrared thermal imaging.
Further, preferably, the system further comprises a display unit, which is installed in the main operating room, is respectively connected with the data acquisition module, the infrared thermal imager and the billet removal control unit, and is used for displaying the steel type of the acquired billet, the billet thermal infrared image data acquired by the infrared thermal imager, the score threshold values of the steel types and the control result of the billet removal control unit.
Further, preferably, the infrared thermal imager is connected with the PLC control system, and after the PLC control system detects that the steel billet is discharged from the furnace, the PLC control system sends a measurement start signal to the infrared thermal imager at a certain time, and the infrared thermal imager starts to measure.
The invention also provides a rough rolling billet quality detection method based on the thermal image, and the rough rolling billet quality detection system based on the thermal image comprises the following steps:
step (1), collecting steel grade and thermal image data of a steel billet, wherein the thermal image data comprises image point positions and point position temperatures of the steel billet;
step (2), taking thermal infrared image data of historical steel billets as input and steel grades as output, and performing machine learning model training;
step (3), inputting the steel grade of the steel billet collected in real time and the thermal image data of the steel billet into a machine learning model obtained by training of a model training module, wherein the output of the model is a probability value between 0 and 1 of the steel grade, and when the probability value is greater than a preset threshold value, judging that the quality of the steel billet is qualified, otherwise, judging that the quality of the steel billet is unqualified; if the output is that the quality of the steel billet is qualified, the steel billet removing control unit controls the steel billet removing device not to act, and if the output is that the quality of the steel billet is unqualified, the steel billet removing control unit controls the steel billet removing device to remove the steel billet. Preferably, the machine learning model of the present invention is a convolutional neural network.
An infrared thermal imager is arranged in front of a roughing mill set, and when a heating furnace generates a steel tapping signal and a certain delay is passed, the infrared thermal imager receives a starting signal and starts to measure thermal image data; storing thermal image data (including image point location and point location temperature) and steel grade in a production database, and inputting the thermal image data into a rough rolling billet quality online detection model (namely a trained machine learning model) for online detection; the online detection model judges whether the quality of the steel billet meets the requirements of rough rolling and initial rolling (namely whether the quality is qualified) according to the input thermal image data and the score threshold of each steel type, and if the quality does not meet the requirements, the steel billet is rejected by a steel billet rejecting device; the data of the production database is continuously updated, the online quality detection model of the rough rolling billet periodically extracts data from the production database, training, tuning and updating are carried out, and the judgment accuracy of the production process is improved, as shown in figure 2. Preferably, the billet removing control unit is arranged in the PLC control system. Wherein, the PLC control system is the control system of the existing production line.
The infrared thermal imager is arranged above the object to be measured and is arranged in a backlight mode, so that the radiation energy loss on the surface of the steel billet is minimized, and the measurement error is reduced.
An online infrared thermal imager special for metallurgy is preferably selected and used, the online infrared thermal imager is specially used in high-temperature industrial application scenes in the metallurgy industry and the like, the work is stable, the imaging is clear, and the temperature measurement range can reach dozens of to thousands of degrees centigrade.
Because the field environment of the steel rolling system is complex, the infrared thermometer is influenced by external factors such as field high temperature, dust, vibration and the like, the display fluctuation is large, and the accuracy rate can be reduced, the field infrared thermal imager preferably adopts a closed air-cooled protective cover, so that the infrared thermal imager is in a dry and closed working environment, and the influence of the field environment on the measurement result is primarily solved.
According to the invention, the high-pressure water dephosphorization device is arranged between the heating furnace and the roughing mill group, almost all iron scales and the like before the billet enters the mill are washed away by high-pressure water spray, a clean surface is provided for rolling, the infrared thermal imager measures the real temperature of the surface of the billet, and the measurement accuracy is improved.
In order to reduce the work load of the infrared thermal imager, a reasonable time interval is set for the infrared thermal imager according to the transmission speed and the distance of a rolling line, when a heating furnace generates a steel tapping signal, the infrared thermal imager receives a starting signal and starts to measure after a certain time delay, the phenomenon that the infrared thermal imager continuously collects useless data is avoided, and the storage load of a production database is increased.
In the invention, the judged preset threshold value can be self-defined according to experience and different steel grade requirements, so that whether the steel grade is qualified or not is judged by setting the given threshold value of each steel grade.
Compared with the prior art, the invention has the beneficial effects that:
(1) the invention adopts non-contact detection, the infrared thermal imager is arranged at a certain distance vertically above the detected target, the protective cover is adopted to protect equipment, the on-line infrared thermal imager special for metallurgy is preferably adopted, the resolution is high, the imaging is clear, and the reaction is sensitive in the wave band of 8-14 μm;
(2) according to the invention, a signal is sent to the infrared thermal imager to start measuring at a certain time after the steel billet is discharged from the furnace, so that the instrument is prevented from continuously collecting useless data, and the storage burden of a production database is increased;
(3) the infrared thermal image data and the steel grade are analyzed by adopting a deep learning algorithm, whether the steel billet meets the rough rolling requirement is judged, the method is convenient and rapid, the accuracy is high, the production loss is avoided, and the method is easy to popularize and apply.
(4) The existing hot rolling rough rolling temperature detection uses a point temperature measuring instrument, cannot provide a temperature image, is difficult to verify whether the instrument is aligned to a measured point or not, and is difficult to find out the distribution condition of heat intensity.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a schematic structural diagram of an embodiment of a rough rolling billet quality detection system based on thermal image, wherein the arrow direction is data or signal trend;
FIG. 2 is a schematic structural diagram of a rough rolling billet quality inspection system based on thermal image according to another embodiment of the present invention, wherein the arrow direction is data or signal trend;
FIG. 3 is a flow chart of the rough rolling billet quality detection system method based on thermal image.
Detailed Description
The present invention will be described in further detail with reference to examples.
It will be appreciated by those skilled in the art that the following examples are illustrative of the invention only and should not be taken as limiting the scope of the invention. The examples do not specify particular techniques or conditions, and are performed according to the techniques or conditions described in the literature in the art or according to the product specifications. The materials or equipment used are not indicated by manufacturers, and all are conventional products available by purchase.
As used herein, the singular forms "a", "an", "the" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. It will be understood that when an element is referred to as being "connected" to another element, it can be directly connected to the other element or intervening elements may also be present. Further, "connected" as used herein may include wirelessly connected. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items.
In the description of the present invention, "a plurality" means two or more unless otherwise specified. The terms "inner," "upper," "lower," and the like, refer to an orientation or a state relationship based on that shown in the drawings, which is for convenience in describing and simplifying the description, and do not indicate or imply that the referenced device or element must have a particular orientation, be constructed and operated in a particular orientation, and thus should not be construed as limiting the invention.
In the description of the present invention, it should be noted that, unless otherwise explicitly specified or limited, the terms "mounted," "connected," and "provided" are to be construed broadly, e.g., as being fixedly connected, detachably connected, or integrally connected; can be mechanically or electrically connected; may be directly connected or indirectly connected through an intermediate. To those of ordinary skill in the art, the specific meanings of the above terms in the present invention are understood according to specific situations.
It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
The devices used by the invention are all the prior art, can be purchased from the market in an open mode, and have different requirements on different production sites, so the model, the specific structure and the performance parameters of the infrared thermal imager are not repeated.
The image processing and deep learning algorithms related to the invention are all the prior art, and the models constructed by combining different algorithms are different, and the accuracy is also different, so detailed algorithm formulas are not repeated.
As shown in fig. 1, a rough rolling billet quality detection system based on thermal images is provided, wherein a high-pressure water dephosphorization device is arranged between a heating furnace and a rough rolling unit, a billet conveying device is arranged between the rough rolling unit and the high-pressure water dephosphorization device, and an infrared thermal imager is arranged right above the billet conveying device; after the steel billet is discharged from the furnace, removing phosphorus by a high-pressure water phosphorus removal device, and acquiring thermal image data of the steel billet by an infrared thermal imager;
further comprising:
the data acquisition module is used for acquiring the steel grade of the steel billet;
the production database storage module is connected with the data acquisition module and the infrared thermal imager and is used for storing the thermal infrared image data (image point position and point position temperature) of the steel billet acquired by the infrared thermal imager and the corresponding steel grade;
the model training module is connected with the production database storage module and used for performing machine learning model training by taking the thermal infrared image data of the historical steel billet as input and the steel grade as output;
the billet removing device is arranged on the billet conveying device and is arranged behind the infrared thermal imager;
the billet steel removing control unit is connected with the data acquisition module, the infrared thermal imager, the model training module and the billet steel removing device and is used for inputting the billet steel type acquired by the data acquisition module in real time and the billet steel thermal infrared image data acquired by the infrared thermal imager in real time into a machine learning model obtained by training of the model training module, the output of the model is a probability value between 0 and 1 of the steel type, when the probability value is greater than a preset threshold value, the quality of the billet steel is judged to be qualified, otherwise, the quality of the billet steel is judged to be unqualified; if the output is that the quality of the steel billet is qualified, the steel billet removing control unit controls the steel billet removing device not to act, and if the output is that the quality of the steel billet is unqualified, the steel billet removing control unit controls the steel billet removing device to remove the steel billet.
Preferably, the infrared thermal imager is backlight mounted.
Preferably, the infrared thermal imaging is an online infrared thermal imager.
Preferably, a closed air-cooled protective cover is arranged outside the infrared thermal imaging device.
Preferably, as shown in fig. 2, the system further includes a display unit, installed in the main operating room, and respectively connected to the data acquisition module, the infrared thermal imager, and the billet removal control unit, and configured to display the steel type of the collected billet, the billet thermal image data (image point location and point location temperature) collected by the infrared thermal imager, the score threshold of each steel type, and the control result of the billet removal control unit.
Preferably, the infrared thermal imager is connected with the PLC control system, and after the PLC control system detects that the steel billet is discharged from the furnace, the PLC control system sends a measuring starting signal to the infrared thermal imager at a certain time, and the infrared thermal imager starts to measure.
A rough rolling billet quality detection method based on thermal images adopts the rough rolling billet quality detection system based on the thermal images, and comprises the following steps:
step (1), collecting steel grade and thermal image data of a steel billet, wherein the thermal image data comprises image point positions and point position temperatures of the steel billet;
step (2), taking thermal infrared image data of historical steel billets as input and steel grades as output, and performing machine learning model training;
step (3), inputting the steel grade of the steel billet collected in real time and the thermal image data of the steel billet into a machine learning model obtained by training of a model training module, wherein the output of the model is a probability value between 0 and 1 of the steel grade, and when the probability value is greater than a preset threshold value, judging that the quality of the steel billet is qualified, otherwise, judging that the quality of the steel billet is unqualified; if the output is that the quality of the steel billet is qualified, the steel billet rejecting control unit controls the steel billet rejecting device not to act, and if the output is that the quality of the steel billet is unqualified, the steel billet rejecting control unit controls the steel billet rejecting device to reject the steel billet
In order to facilitate parameter migration and parameter adjustment, the network structure of the machine learning model (namely the online detection model) is modified from an Al exNet network. The category of the last layer of the network billet can be set to be 1, 2, 3. Input layer- (convolutional layer-pooling layer) × 2-convolutional layer × 3-pooling layer-fully connected layer × 2-output layer.
The sample data is divided into a training set and a test set according to a certain proportion (the proportion of the training set can be properly increased along with the increase of the data, and is preferably 7: 3), a certain number of pictures are randomly extracted in each training round, and multiple rounds of training are carried out.
(1) An input layer: the thermal image is randomly cropped to a uniform size RGB image as input X. The hot image in this example is a 256 × 256 RGB image, which is unified into a size of 227 × 227 by cropping.
(2) And (3) rolling layers: the convolution layer is used for extracting image features, local simple features of the image are obtained after the convolution operation for the first time, more complex features are obtained on the basis of the convolution operation for the second time, and the simple features are gradually combined into complex features as the network depth deepens. The convolution kernel (also called filter) W is a feature detector with f × f size, and the convolution operation is to multiply each element in the area with f × f size on the extreme side of the input data with the data at the corresponding position of the convolution kernel, and then add them, and the obtained value is used as the first element of the output matrix, that is:
wherein x isijzThe pixel value, w, representing the point of the imageijzRepresenting the training weights of the corresponding point locations.
(3) Sliding the convolution kernel on the image data, performing the same convolution operation on another region, collecting the result into a single output pixel value, repeating the process until the whole image is traversed, and finally obtaining a characteristic matrix WTAnd (4) X. In this example, the convolution layers have five layers, and the convolution kernels in the network have sizes of 11 × 11, 5 × 5, 3 × 3, and 3 × 3, respectively.
(4) And adding deviation to the characteristic matrix, and applying a nonlinear activation function ReLU to finally obtain the output of the convolutional layer, which is also the input of the next layer:
a＝max(WTX+b)
where X denotes the input image matrix, W denotes the corresponding weight matrix, b denotes the offset, and a denotes the output matrix.
Activation is essentially a non-linear mapping of the output of the convolutional layer. The ReLU function is formulated as: when the input x is less than or equal to 0, the output is equal to 0, and when the input x is greater than 0, the output is equal to x.
(5) After the convolution operation, the picture is reduced, meanwhile, the times of convolution operation on the pixels at the corners and the edges are few, and useful information can be lost. In this example, the 2 nd, 3 rd, 4 th and 5 th convolution layers are filled with fillers respectively, and the sizes of the fillers are 2, 1 and 1
(6) A pooling layer: the method is mainly used for feature dimension reduction, and by reducing the length and width of the matrix and compressing the number of data and parameters, overfitting is reduced, the calculation speed is increased, and the feature robustness is improved. The maximum pooling method is selected, i.e. the maximum of all pixels in a certain area of the input matrix is taken as an element of the output matrix. If the input size is n x n, the output size after pooling is:
where n denotes the input size, p denotes the fill size, s denotes the step size, and f denotes the convolution kernel size.
The convolution step size for all layers in this example is 1.
(7) Full connection layer: all neurons of the input and output of the full connection layer are connected with weights, the connection mode is the same as that of the neurons of the traditional neural network, a ReLU nonlinear activation function is selected here, namely the output is as follows:
yi＝max(wijxij+b)
wherein x isijRepresents an input, wijRepresenting the connection weight between the ith neuron and the jth neuron. The first fully-connected layer in this example contains 4096 6 x 256 convolution kernels and the second fully-connected layer contains 4096 neurons.
(8) An output layer: obtaining an output through a softmax function, wherein the output is the probability that the input belongs to the class corresponding to the output, and the formula is as follows:
wherein i represents a label and a classification, a numerical value can be set according to actual conditions, and y representsiRepresenting the full connection layer output value.
I in this example can be set to an integer of 1, 2, 3 … size depending on the actual situation.
(9) Parameter optimization: comparing the output value with the label value of the corresponding sample, calculating the error, wherein when the error is smaller, the model output value is closer to the true value of the sample, and the error function of the m samples is as follows:
wherein, y(i)In order to output the value of the model,
for the sample label value, J (w, b) represents the error function.
In order to obtain the minimum error, parameters such as the weight, the offset and the like of all the convolution layers and the output layers are reversely obtained by adopting a gradient descent method, and then forward calculation is carried out until the minimum error is calculated, and the model training is stopped.
(9) Since the number of network neurons is too large, learning ability is strong, and overfitting may occur, dropout may be added. dropout traverses each layer of the network, sets a probability of random elimination for each neuron node, deletes some nodes at last, and forms a network with fewer nodes and smaller scale for training by the retained neurons, wherein the dropout probability is set to be 0.5.
The invention was developed based on the TensorFlow framework. TensorFlow is an open-source machine learning framework based on Python, and has rich application in the scenes of graphic classification, audio processing, recommendation system, natural language processing and the like. TensorFlow also provides interfaces for C/C + +, Java, Go, R, and other programming languages, allowing deep neural network computing to be deployed on any number of CPU or GPU servers, PCs, or mobile devices, and utilizing only one TensorFlow API. Allowing the model to be deployed into industrial production and easy to use. The invention adopts the infrared thermal imager, and has high resolution and clear imaging. Through the network structure parameter setting, tens of thousands of images are trained and tested at present, the method is convenient, rapid and high in accuracy, production loss is avoided, popularization and application are easy, and continuous optimization is achieved. Meanwhile, the measuring signal is given after the billet is discharged from the furnace for a period of time, so that the continuous acquisition of useless data by an instrument is avoided, and the storage burden of a production database is increased. The distribution condition of heat intensity can also be looked over to infrared thermal imager image, is convenient for more directly perceived understanding than the point temperature measuring appearance that the market is commonly used. The foregoing shows and describes the general principles, essential features, and advantages of the invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, which are described in the specification and illustrated only to illustrate the principle of the present invention, but that various changes and modifications may be made therein without departing from the spirit and scope of the present invention, which fall within the scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (7)
1. A rough rolling billet quality detection system based on thermal images is characterized in that a high-pressure water dephosphorization device is arranged between a heating furnace and a rough rolling unit, a billet conveying device is arranged between the rough rolling unit and the high-pressure water dephosphorization device, and an infrared thermal imager is arranged right above the billet conveying device; after the steel billet is discharged from the furnace, removing phosphorus by a high-pressure water phosphorus removal device, and acquiring thermal image data of the steel billet by an infrared thermal imager;
further comprising:
the data acquisition module is used for acquiring the steel grade of the steel billet;
the production database storage module is connected with the data acquisition module and the infrared thermal imager and used for storing the billet thermal infrared image data acquired by the infrared thermal imager and the corresponding steel grade;
the model training module is connected with the production database storage module and used for performing machine learning model training by taking the thermal infrared image data of the historical steel billet as input and the steel grade as output;
the billet removing device is arranged on the billet conveying device and is arranged behind the infrared thermal imager;
the billet steel removing control unit is connected with the data acquisition module, the infrared thermal imager, the model training module and the billet steel removing device and is used for inputting the billet steel type acquired by the data acquisition module in real time and the billet steel thermal infrared image data acquired by the infrared thermal imager in real time into a machine learning model obtained by training of the model training module, the output of the model is a probability value between 0 and 1 of the steel type, when the probability value is greater than a preset threshold value, the quality of the billet steel is judged to be qualified, otherwise, the quality of the billet steel is judged to be unqualified; if the output is that the quality of the steel billet is qualified, the steel billet removing control unit controls the steel billet removing device not to act, and if the output is that the quality of the steel billet is unqualified, the steel billet removing control unit controls the steel billet removing device to remove the steel billet.
2. The system of claim 1, wherein the infrared thermal imager is backlit.
3. The system for thermal image-based rough rolling billet quality inspection according to claim 1, wherein the infrared thermal imaging is an on-line infrared thermal imager.
4. The system of claim 1, wherein the infrared thermal imaging device is provided with a closed air-cooled shield.
5. The system for inspecting rough rolling billet quality based on thermal image according to claim 1, further comprising a display unit installed in the main operating room and respectively connected to the data acquisition module, the infrared thermal imager and the billet removal control unit for displaying the steel grade of the collected billet, the thermal infrared image data of the billet collected by the infrared thermal imager, the score threshold of each steel grade and the control result of the billet removal control unit.
6. The system of claim 1, wherein the infrared thermal imager is connected to the PLC control system, and the PLC control system sends a start measurement signal to the infrared thermal imager at a predetermined time after the slab is discharged from the furnace, and the infrared thermal imager starts the measurement.
7. A rough rolling billet quality detection method based on thermal images, which adopts the rough rolling billet quality detection system based on thermal images of any one of claims 1-6, and is characterized by comprising the following steps:
step (1), collecting steel grade and thermal image data of a steel billet, wherein the thermal image data comprises image point positions and point position temperatures of the steel billet;
step (2), taking thermal infrared image data of historical steel billets as input and steel grades as output, and performing machine learning model training;
step (3), inputting the steel grade of the steel billet collected in real time and the thermal image data of the steel billet into a machine learning model obtained by training of a model training module, wherein the output of the model is a probability value between 0 and 1 of the steel grade, and when the probability value is greater than a preset threshold value, judging that the quality of the steel billet is qualified, otherwise, judging that the quality of the steel billet is unqualified; if the output is that the quality of the steel billet is qualified, the steel billet removing control unit controls the steel billet removing device not to act, and if the output is that the quality of the steel billet is unqualified, the steel billet removing control unit controls the steel billet removing device to remove the steel billet.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110741872.3A CN113649422A (en) | 2021-06-30 | 2021-06-30 | Hot image-based rough rolling billet quality detection system and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110741872.3A CN113649422A (en) | 2021-06-30 | 2021-06-30 | Hot image-based rough rolling billet quality detection system and method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113649422A true CN113649422A (en) | 2021-11-16 |
Family
ID=78489833
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110741872.3A Pending CN113649422A (en) | 2021-06-30 | 2021-06-30 | Hot image-based rough rolling billet quality detection system and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113649422A (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN203621092U (en) * | 2013-12-17 | 2014-06-04 | 安徽马钢工程技术有限公司 | Heating device for hot delivery of continuous casting square billets |
| CN106053479A (en) * | 2016-07-21 | 2016-10-26 | 湘潭大学 | System for visually detecting workpiece appearance defects based on image processing |
| CN106734252A (en) * | 2016-12-14 | 2017-05-31 | 四川德胜集团钒钛有限公司 | A kind of slab heat send technique |
| US20180239987A1 (en) * | 2017-02-22 | 2018-08-23 | Alibaba Group Holding Limited | Image recognition method and apparatus |
| CN109465295A (en) * | 2018-08-06 | 2019-03-15 | 首钢集团有限公司 | A method of preventing hot continuous-milling steel plate brisement band in side in cold rolling |
| CN109772755A (en) * | 2019-01-20 | 2019-05-21 | 敬业钢铁有限公司 | Unqualified steel billet removal equipment in hot rolled alloy steel production process |
| CN110918655A (en) * | 2019-11-30 | 2020-03-27 | 宝钢特钢韶关有限公司 | Refined heating control method |
| CN111618104A (en) * | 2020-06-03 | 2020-09-04 | 北京首钢股份有限公司 | Slab hot rolling temperature drop detection method |
-
2021
- 2021-06-30 CN CN202110741872.3A patent/CN113649422A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN203621092U (en) * | 2013-12-17 | 2014-06-04 | 安徽马钢工程技术有限公司 | Heating device for hot delivery of continuous casting square billets |
| CN106053479A (en) * | 2016-07-21 | 2016-10-26 | 湘潭大学 | System for visually detecting workpiece appearance defects based on image processing |
| CN106734252A (en) * | 2016-12-14 | 2017-05-31 | 四川德胜集团钒钛有限公司 | A kind of slab heat send technique |
| US20180239987A1 (en) * | 2017-02-22 | 2018-08-23 | Alibaba Group Holding Limited | Image recognition method and apparatus |
| CN109465295A (en) * | 2018-08-06 | 2019-03-15 | 首钢集团有限公司 | A method of preventing hot continuous-milling steel plate brisement band in side in cold rolling |
| CN109772755A (en) * | 2019-01-20 | 2019-05-21 | 敬业钢铁有限公司 | Unqualified steel billet removal equipment in hot rolled alloy steel production process |
| CN110918655A (en) * | 2019-11-30 | 2020-03-27 | 宝钢特钢韶关有限公司 | Refined heating control method |
| CN111618104A (en) * | 2020-06-03 | 2020-09-04 | 北京首钢股份有限公司 | Slab hot rolling temperature drop detection method |
Non-Patent Citations (2)
| Title |
|---|
| 孙照阳: "热成像仪在热轧带钢温度均匀性评价上的应用实践", no. 4, pages 52 - 54 * |
| 张理齐: "热轧带钢生产中的红外辐射测温技术", vol. 20, no. 3, pages 36 - 41 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103439342B (en) | Based on the Infrared Non-destructive Testing method of thermal map temporal aspect | |
| CN102159940B (en) | Method for detecting defect in material and system for same | |
| CN104914111A (en) | Strip steel surface defect on-line intelligent identification and detection system and detection method | |
| CN108917960B (en) | Device and method for measuring billet heating temperature uniformity | |
| CN108897354B (en) | Aluminum smelting process hearth temperature prediction method based on deep belief network | |
| CN104215334A (en) | Real-time online monitoring method of temperature of molten steel in RH refining furnace | |
| CN109900363A (en) | A kind of object infrared measurement of temperature method and apparatus based on contours extract | |
| Ye et al. | Real-time quality prediction of casting billet based on random forest algorithm | |
| Pan et al. | Compensation method for molten iron temperature measurement based on heterogeneous features of infrared thermal images | |
| CN115060376A (en) | Aluminum alloy temperature field measuring method based on infrared thermal imaging and iterative algorithm | |
| Wang et al. | Design of machine vision applications in detection of defects in high-speed bar copper | |
| CN105956591B (en) | A kind of online hot parts infrared image spectrum sampling Detection method | |
| CN110189321B (en) | Method and system for determining uniformity of concrete surface coating | |
| CN110732559A (en) | method for evaluating temperature uniformity of hot-rolled strip steel intermediate billet in width direction | |
| CN116842402B (en) | Blast furnace abnormal furnace condition detection method based on stable characteristic extraction of twin neural network | |
| CN113649422A (en) | Hot image-based rough rolling billet quality detection system and method | |
| CN113084193A (en) | In-situ quality comprehensive evaluation method for selective laser melting technology | |
| Jiang et al. | A new monitoring method for the blocking time of the taphole of blast furnace using molten iron flow images | |
| CN108920428B (en) | Fuzzy distance discrimination method based on joint fuzzy expansion principle | |
| CN105866168A (en) | Identification method and apparatus for lower matrix material of coating | |
| CN115446276A (en) | Continuous casting breakout early warning method for recognizing V-shaped bonding characteristics of crystallizer copper plate based on convolutional neural network | |
| CN111680448B (en) | Continuous casting billet longitudinal crack prediction method based on SVM classification | |
| CN114596296A (en) | High-sensitivity hot-rolled steel coil end surface defect identification system and method | |
| Gu et al. | A reweighting offset bin classification network for surface defect detection and location of metal components | |
| CN112765219B (en) | Stream data abnormity detection method for skipping steady region |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |