Strip steel defect diagnosis expert system and apparatus

Abstract

The application provides a strip steel defect diagnosis expert system and equipment. The system comprises a data input module for inputting to-be-diagnosed data, an image defect identification module, a data defect identification module, a knowledge base for storing a plurality of preset defect causes, a plurality of preset diagnosis results and a corresponding relationship between the preset defect causes and the preset diagnosis results, a defect cause determination module for determining a defect cause according to at least one of a defect image and abnormal data, and a defect diagnosis module for querying the knowledge base according to the defect cause to obtain a diagnosis result of the to-be-diagnosed strip steel. Through the system, the to-be-diagnosed data of a strip steel continuous production line is diagnosed in a machine judgment mode, a defect cause and a defect countermeasure of the strip steel are obtained, the time consumption of defect diagnosis is reduced, possible waste or defective loss is reduced, and the experience dependence on experienced workers is reduced.

Description

Technical Field

[0001] This invention relates to the field of metallurgical technology, and in particular to an expert system and equipment for diagnosing defects in strip steel. Background Technology

[0002] Steel industry products such as rolled hard plates, cold-rolled plates, and coated plates produced by continuous production lines (i.e., strip steel mentioned in this application) have very high requirements for surface quality. They are required to be free of defects visible to the naked eye on the surface, as well as defects hidden under the surface or extremely small defects that are not visible to the naked eye.

[0003] In related technologies, the method for detecting such defects involves a partial shutdown of the continuous production line for about two minutes, during which the steel surface is manually polished with an oilstone to remove the outer layer, exposing hidden defects such as zinc slag, zinc ash, impurities, and embossing. Because these production lines operate continuously, the analysis and judgment of defects after detection rely entirely on manual labor, which may lead to misjudgments or omissions. Furthermore, the process of investigating the causes of these defects is often relatively time-consuming, generally requiring the cooperation of multiple trades, which is labor-intensive and time-consuming, potentially resulting in significant losses of scrap or defective products. Summary of the Invention

[0004] This invention provides a strip steel defect diagnosis expert system and equipment to solve the technical problem that manual grinding, identification, and analysis of the causes of defects on the surface of strip steel is time-consuming and may result in a large number of scrap or defective products.

[0005] This invention provides an expert system for diagnosing defects in strip steel. The system includes: a data input module for diagnosing data of the strip steel to be diagnosed, the data including at least one of an image to be diagnosed and production data to be diagnosed, the production data including multiple sub-data to be diagnosed; an image defect recognition module for recognizing the image to be diagnosed to obtain an image defect diagnosis result, the image defect diagnosis result including a defect presence status, and if the defect presence status is "defect exists", the image defect diagnosis result also includes a defect image; a data defect recognition module for comparing the sub-data to be diagnosed with standard sub-data to obtain the data qualification status of each sub-data to be diagnosed; a knowledge base for storing multiple preset defect causes and multiple preset diagnosis results, as well as the correspondence between the preset defect causes and preset diagnosis results; a defect cause determination module for determining the defect cause based on at least one of the defect image and abnormal data, the abnormal data including sub-data to be diagnosed whose data qualification status is "unqualified"; and a defect diagnosis module for querying the knowledge base based on the defect cause to obtain the diagnosis result of the strip steel to be diagnosed.

[0006] In one embodiment of the present invention, the system further includes a defect query module, which comprises: a query receiving module for receiving input query data, the input query data including at least one of text statements, voice statements, and input images; a statement extraction module for extracting query features from the text statements or voice statements if the input query data includes text statements or voice statements, thereby obtaining extracted query features, the extracted query features including at least one of defect shape, defect size, defect occurrence cycle, defect strip location, defect production line location, defect quantity, and abnormal production line operation data; and a data conversion module for determining the input image as an image to be diagnosed and the extracted query features as production data to be diagnosed.

[0007] In one embodiment of the present invention, the knowledge base further includes an expert knowledge update dialogue module, which is used to store newly acquired historical defects, causes of historical defects, and countermeasures for historical defects into the knowledge base.

[0008] In one embodiment of the present invention, the defect cause determination module includes multiple defect matching sub-modules and a preset matching order for the multiple defect matching sub-modules. The data to be determined of the strip to be diagnosed is matched with the preset determination data of the defect matching sub-modules according to the preset matching order. The preset defect cause of the defect matching sub-module that is finally matched is determined as the defect cause of the strip to be diagnosed. The data to be determined includes at least one of the defect image and the abnormal data. The preset determination data includes at least one of the preset image set and the preset data range set. The defect matching sub-module is preset with preset defect causes and preset determination data.

[0009] In one embodiment of the present invention, the defect diagnosis module includes a defect countermeasure determination module, which is used to compare the defect cause with multiple preset defect causes in the knowledge base, and determine the preset diagnosis result corresponding to the preset defect cause that is successfully matched as the diagnosis result of the strip steel to be diagnosed, wherein the preset diagnosis result includes a response suggestion.

[0010] In one embodiment of the present invention, if the production data to be diagnosed includes multiple production process parameters, the system further includes a process anomaly detection module, which is used to compare the production process parameters with preset standard process parameters. If the production process parameters do not meet the preset standard process parameters, an alarm message is generated based on the production process parameters to prompt the equipment related to the production process parameters to be investigated.

[0011] In one embodiment of the present invention, the system further includes a defect statistics module, used to count the defect locations of the strip steel to be diagnosed, determine the defect occurrence interval distance, compare the defect occurrence interval distance with multiple preset roller circumferences, identify the rollers corresponding to the preset roller circumferences that are successfully compared as abnormal rollers, and prompt to investigate the abnormal rollers.

[0012] In one embodiment of the present invention, the system further includes a result display module for displaying at least one of the cause of the defect and the diagnostic result.

[0013] In one embodiment of the present invention, if the image to be diagnosed is a surface image of a strip before grinding, the image defect recognition module is used to fuse multiple surface images of the strip before grinding acquired under different lighting conditions to obtain a fused image; identify each crystal flower in the fused image and determine the initial crystal flower size of each crystal flower; determine the crystal flower uniformity and crystal flower characterization size based on the differences in the initial crystal flower sizes of all crystal flowers; determine the crystal flower quality data of the strip to be diagnosed based on at least one of the crystal flower uniformity, the crystal flower characterization size, and the number of crystal flowers, wherein the number of crystal flowers is the number of crystal flowers identified in the fused image.

[0014] In one embodiment of the present invention, if the image to be diagnosed is a surface image of a steel strip after grinding, the image defect recognition module is used to perform initial defect recognition on the surface image of the steel strip after grinding. If a defect is identified, a suggested grinding message is generated based on the defect location to guide the surface of the steel strip to be diagnosed to be ground again. If a message indicating that the grinding is completed is received, a processed surface image of the steel strip to be diagnosed is acquired. The processed surface image of the steel strip is then subjected to further defect recognition, and the defect recognition of the image to be diagnosed is completed based on the results of the further defect recognition.

[0015] The present invention also provides a strip steel defect diagnosis device, wherein the strip steel defect diagnosis device is equipped with a strip steel defect diagnosis expert system as provided in any of the above claims.

[0016] The beneficial effects of this invention are as follows: This invention proposes a strip steel defect diagnosis expert system and equipment. The system includes a data input module for the data to be diagnosed of the strip steel, an image defect recognition module for recognizing the image to be diagnosed to obtain the image defect diagnosis result, a data defect recognition module for comparing the sub-data to be diagnosed with standard sub-data to obtain the data qualification status of each sub-data to be diagnosed, a knowledge base for storing multiple preset defect causes and multiple preset diagnosis results, as well as the correspondence between preset defect causes and preset diagnosis results, a defect cause determination module for determining the defect cause based on at least one of the defect image and abnormal data, and a defect diagnosis module for querying the knowledge base based on the defect cause to obtain the diagnosis result of the strip steel to be diagnosed. Through the above system, a method of machine judgment is provided for diagnosing the data to be diagnosed in a continuous strip steel production line, obtaining the defect cause and defect countermeasure of the strip steel, reducing the time consumption of defect diagnosis, reducing possible scrap or defect losses, and reducing the reliance on the experience of experienced workers. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of a strip steel defect diagnosis expert system provided in an embodiment of the present invention;

[0018] Figure 2 This is a schematic diagram illustrating the implementation of a defect query module and knowledge base according to an embodiment of the present invention.

[0019] Figure 3 This is a schematic diagram of the execution flow of a defect cause determination module provided in an embodiment of the present invention;

[0020] Figure 4 A schematic diagram of a strip steel defect diagnosis expert system provided in another embodiment of the present invention;

[0021] Figure 5 This is a schematic diagram illustrating the information input and output of a strip steel defect diagnosis expert system provided in another embodiment of the present invention. Detailed Implementation

[0022] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that, unless otherwise specified, the following embodiments and features described therein can be combined with each other.

[0023] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Therefore, the drawings only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0024] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the invention. However, it will be apparent to those skilled in the art that embodiments of the invention may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the invention.

[0025] Please see Figure 1 , Figure 1 This is a schematic diagram of a strip steel defect diagnosis expert system provided in an embodiment of the present invention, as shown below. Figure 1 As shown, this embodiment provides a strip steel defect diagnosis expert system, specifically including the following modules: a data input module 101, used for diagnosing data of the strip steel to be diagnosed, the diagnostic data including at least one of the diagnostic image and the diagnostic production data, the diagnostic production data including multiple diagnostic sub-data; an image defect recognition module 102, used for recognizing the diagnostic image to obtain an image defect diagnosis result, the image defect diagnosis result including the defect existence status, if the defect existence status is "defect exists", the image defect diagnosis result also includes a defect image; a data defect recognition module 103, used for comparing the diagnostic sub-data with standard sub-data to obtain the data qualification status of each diagnostic sub-data; a knowledge base 104, used for storing multiple preset defect causes and multiple preset diagnosis results, as well as the correspondence between preset defect causes and preset diagnosis results; a defect cause determination module 105, used for determining the defect cause based on at least one of the defect image and abnormal data, the abnormal data including the diagnostic sub-data whose data qualification status is "unqualified"; and a defect diagnosis module 106, used for querying the knowledge base based on the defect cause to obtain the diagnosis result of the strip steel to be diagnosed.

[0026] The data input module can be a keyboard, touchscreen, image acquisition device, or data acquisition device. Multiple image acquisition devices are placed at various locations on the strip steel production line, such as the furnace opening, zinc pot outlet, and strip steel output outlet, to directly acquire the images to be diagnosed. Data acquisition devices collect data such as process parameters from the production line.

[0027] The steel strips in this embodiment include, but are not limited to, various types of steel strips such as hard-rolled steel, cold-rolled steel, galvanized steel, and color-coated steel.

[0028] The strip steel to be diagnosed can be strip steel that has not yet been removed from the production line, or other strip steel as identified by those skilled in the art.

[0029] The images to be diagnosed include, but are not limited to, at least one of the following: surface image of the strip before grinding, surface image of the strip after grinding, surface image of the strip at the furnace outlet, and surface image of the zinc pot outlet. The specific acquisition device for the images to be diagnosed can be implemented using devices known to those skilled in the art.

[0030] The production data to be diagnosed includes, but is not limited to, at least one of the following: physical diagnostic data of strip steel, feedback data from downstream production lines, and production process data of strip steel production lines. Physical diagnostic data of strip steel includes, but is not limited to, at least one of the following: shape, dimensions, roughness, waviness, mechanical properties, processing performance, and performance in use. Feedback data from downstream production lines includes, but is not limited to, feedback data from downstream production lines using the strip steel as raw material. If the strip steel is substandard, defect detection can be performed on it. Production process data of the strip steel production line refers to the production line where the strip steel is located, or the operating process data of the production line that produces the strip steel. As mentioned above, the production data to be diagnosed can include multiple types of data, data from multiple sources, and data from multiple points in time. In this case, it can be stated that the production data to be diagnosed includes multiple sub-data to be diagnosed. Each sub-data to be diagnosed can be classified by data type or according to the classification method specified by those skilled in the art. When the subsequent data defect identification module compares the sub-data to be diagnosed, the standard sub-data is also set according to the above classification rules.

[0031] In one embodiment, the strip steel defect diagnosis expert system further includes a defect query module, which comprises: a query receiving module for receiving input query data, which includes at least one of text statements, voice statements, and input images; a statement extraction module for extracting query features from the text or voice statements if the input query data includes text or voice statements, obtaining extracted query features, including but not limited to at least one of defect shape, defect size, defect occurrence cycle, defect strip steel location, defect production line location, defect quantity, and abnormal production line operation data; and a data conversion module for determining the input image as the image to be diagnosed and determining the extracted query features as the production data to be diagnosed. It is evident that one input method for the production data to be diagnosed can be obtained through user-initiated query input.

[0032] In this embodiment, the query receiving module can be a voice receiving system, or an input system such as a touch screen or keyboard. Query data can be input through voice dialogue, text input, image upload, or other methods.

[0033] Before extracting query features, the speech can be converted into text, and then query features can be extracted based on keywords of interest using methods known to those skilled in the art, such as word segmentation. These keywords of interest can be pre-defined by those skilled in the art. For example, the keywords of interest could be related terms used to characterize defect shape, defect size, defect occurrence cycle, defective strip location, defective production line location, defect quantity, and abnormal production line operation data. By extracting query features, the extracted query features can be obtained. By setting up a defect query module, a way for external entities to proactively query defects can be provided, broadening the application scenarios of the strip steel defect diagnosis expert system and improving the user experience.

[0034] In one embodiment, the difference between the defect diagnosis method using the defect query module and the diagnostic data collected directly or input by the user lies only in the data source method. Subsequent data processing can be implemented in a similar way, which will not be elaborated here.

[0035] In one embodiment, if the image to be diagnosed is a surface image of the strip before grinding, the image defect recognition module is used to fuse multiple images of the strip surface before grinding acquired under different lighting conditions to obtain a fused image; identify each crystal flower in the fused image and determine the initial crystal flower size of each crystal flower; determine the crystal flower uniformity and crystal flower characterization size based on the differences in the initial crystal flower sizes of all crystal flowers; determine the crystal flower quality data of the strip to be diagnosed based on at least one of the crystal flower uniformity, the crystal flower characterization size, and the number of crystal flowers, wherein the number of crystal flowers is the number of crystal flowers identified in the fused image. The surface quality of the strip can be determined in this way.

[0036] The surface image of the strip before grinding can be acquired by an image acquisition device set up in the strip off-line quality inspection process. At this time, the strip has not yet been ground. Multiple light sources can be set in this area, and the lighting state can be changed by controlling the position and switch of the light sources.

[0037] In this embodiment, identifying each flower in the fused image and determining the initial flower size of each flower includes: inputting the fused image into a flower recognition model to obtain a recognition result, the recognition result including the flower position and initial flower size of each flower in the fused image; wherein, the training method of the flower recognition model includes acquiring multiple sample flower data, the sample flower data including sample flower images, the flower label positions in the sample flower images, and the flower label sizes; training a preset basic model with multiple sample flower data until the preset basic model converges, and using the trained preset basic model as the flower recognition model.

[0038] In this embodiment, the flower characterization size includes any one of the average size, mode size, and median size of the initial flower size of all flowers.

[0039] In this embodiment, the method for determining the uniformity of the crystal flowers includes determining the standard deviation of the initial crystal flower size of all crystal flowers, and taking the preset uniformity corresponding to the standard deviation as the uniformity of the crystal flowers.

[0040] In one embodiment, if the image to be diagnosed is a surface image of the strip after grinding, the image defect recognition module performs initial defect recognition on the surface image. If a defect is detected, a suggested grinding message is generated based on the defect location to guide the surface of the strip to be diagnosed to be ground again. If a message indicating that grinding is complete is received, a processed surface image of the strip to be diagnosed is acquired. A second defect recognition is performed on the processed surface image, and the defect recognition of the image to be diagnosed is completed based on the result of the second defect recognition. This method can determine the intrinsic quality of the strip. The surface image of the strip after grinding can be acquired using an image acquisition device installed in the strip off-line quality inspection process, at which point the strip has already been ground.

[0041] In this embodiment, the image to be diagnosed is input into a preset defect recognition model to obtain a defect recognition result. The image to be diagnosed includes a surface image of the strip after grinding or a surface image of the strip after processing. The preset defect recognition model is trained as follows: multiple sample images are acquired, and defects on the surface of the sample strip in each sample image are labeled to obtain defect labels. At least some of the sample images contain defects on the surface of the sample strip. A sample dataset is generated based on the multiple sample images and the defect labels of each sample image. The preset basic model is trained using the sample dataset until the preset basic model converges. The trained preset basic model is then used as the preset defect recognition model.

[0042] In this embodiment, after initial defect identification of the surface image of the strip after grinding, the image defect identification module is further applied to determine whether a defect exists based on the defect label in the initial identification result. If a defect is identified, the image area corresponding to the defect is marked with a bounding box. The defect location is determined based on the position of the bounding box, and a grinding suggestion message is generated based on the defect location. The grinding suggestion message also includes a suggested grinding direction, which is different from the previous grinding direction of the strip.

[0043] In this embodiment, inputting the image to be diagnosed into a preset defect recognition model includes: acquiring the initial grinding data of the strip to be diagnosed, the initial grinding data including the grinding direction and grinding amplitude; determining the grinding mark image based on the initial grinding data; filtering the grinding marks in the image to be diagnosed based on the grinding mark image; and inputting the filtered image to be diagnosed into the preset defect recognition model.

[0044] In this embodiment, if it is determined that there is a defect in the image to be diagnosed, the defect image can be obtained according to the standard defect location.

[0045] In one embodiment, the data defect identification module stores standard sub-data for different types of sub-data to be diagnosed. This standard sub-data can be a threshold range. If the sub-data to be diagnosed is not within this threshold range, it indicates that the sub-data does not meet the standard requirements, and its data qualification status is unqualified. Otherwise, the data qualification status of the sub-data to be diagnosed is qualified. Specific evaluation criteria for data qualification status and the setting method of the standard sub-data can be set by those skilled in the art.

[0046] In one embodiment, if the production data to be diagnosed includes multiple production process parameters, the system further includes a process anomaly detection module, which is used to compare the production process parameters with preset standard process parameters. If the production process parameters do not meet the preset standard process parameters, an alarm message is generated based on the production process parameters to prompt the equipment related to the production process parameters to be checked.

[0047] It should be noted that this process anomaly detection module can also be configured as a sub-module of the data defect identification module. This module detects the sub-data to be diagnosed for a specific process type and generates alarm messages based on non-compliant sub-data. The alarm messages can be displayed by highlighting, bolding, or changing the font color of the data, or they can be exported and transmitted via email or other data transfer methods for user access. Since the data sources of the sub-data to be detected are all known, targeted checks can be performed on the equipment from which these non-compliant data originate, improving the efficiency of defect detection.

[0048] In one embodiment, the knowledge base further includes an expert knowledge update dialogue module, which stores newly acquired historical defects, their causes, and countermeasures into the knowledge base. Through this expert knowledge update dialogue, previously occurring and successfully resolved defects, their characteristics, mechanisms, and solutions are solidified as a basis for future analysis. The organization and updating of the knowledge base can also be achieved using other methods known to those skilled in the art.

[0049] Please see Figure 2 , Figure 2 This is a schematic diagram illustrating the implementation of a defect query module and knowledge base according to an embodiment of the present invention, as shown below. Figure 2 As shown, users and / or experts, through the human-computer interface of the strip steel defect diagnosis expert system, and via the inference engine and compilation system, can query problems and accumulate expert knowledge. Among these, Figure 2 The inference engine and compilation system can be implemented in a manner known to those skilled in the art, and are not limited herein.

[0050] In one embodiment, the defect cause determination module includes multiple defect matching sub-modules and a preset matching order for the multiple defect matching sub-modules. The data to be determined of the strip to be diagnosed is matched with the preset determination data of the defect matching sub-modules according to the preset matching order. The preset defect cause of the defect matching sub-module that is finally matched is determined as the defect cause of the strip to be diagnosed. The data to be determined includes at least one of defect images and abnormal data. The preset determination data includes at least one of a preset image set and a preset data range set. The defect matching sub-module is preset with preset defect causes and preset determination data.

[0051] The process involves matching the data to be determined for the strip to be diagnosed with the preset data of the defect matching submodule. This means matching the defect image with a preset image set and the abnormal data with a preset data range. If both exist and match successfully, the current data to be determined satisfies this defect matching submodule A. Then, the data to be determined is matched with the preset data in the downstream modules of defect matching submodule A. This process continues until all downstream modules are matched, and the preset data of the last defect matching submodule also matches the data to be determined. The preset defect cause of the finally successfully matched defect matching submodule is then determined as the defect cause of the strip to be diagnosed. The upstream and downstream relationships between defect matching submodules can be determined by a preset matching order.

[0052] Taking the defect image in the surface image of the strip after grinding as an example, please refer to [link to relevant documentation]. Figure 3 , Figure 3 This is a schematic diagram of the execution flow of a defect cause determination module provided in an embodiment of the present invention, combined with... Figure 3As shown, defect feature identification is performed on the defect image. If the defect feature identification result is a streak-type defect, and the current density and sharpness of the streak in the defect feature identification result are greater than the preset density and sharpness, the zinc pot outlet surface image of the strip to be diagnosed is obtained. Zinc pot outlet defect identification is performed on the zinc pot outlet surface image. If a defect is identified, the furnace mouth surface image of the strip to be diagnosed is obtained. Furnace mouth defect identification is performed on the furnace mouth surface image. If no defect is identified, the submerged roller operating status of the production line where the strip to be diagnosed is located is obtained. If the submerged roller operating status is normal, the defect type on the surface of the strip to be diagnosed is determined to be a submerged roller scratch. Defects that do not meet the above conditions are all classified as other types of defects.

[0053] The defect cause determination module categorizes defects from broad categories to narrower categories, and then analyzes them layer by layer, from primary to secondary to tertiary to N levels, until the cause is finally found.

[0054] In one embodiment, the defect diagnosis module includes a defect countermeasure determination module, used to compare the defect cause with multiple preset defect causes in a knowledge base, and determine the preset diagnosis result corresponding to the successfully matched preset defect cause as the diagnosis result of the strip steel to be diagnosed. The preset diagnosis result includes a response suggestion. This response suggestion may be a handling measure successfully adopted for the same defect cause in the past. In this way, previous successful experience can be accumulated and applied, further reducing reliance on skilled workers. Of course, one or more defect causes may fail to be matched. In this case, the defect causes that failed to match can be shown to technicians, and the defect handling status can be monitored. Once the defect is successfully handled, the handling method is collected and stored in the knowledge base to enrich the knowledge base.

[0055] In one embodiment, the system further includes a defect statistics module, used to count the defect locations of the strip to be diagnosed, determine the defect occurrence interval distance, compare the defect occurrence interval distance with the circumference of multiple preset rollers, identify the rollers corresponding to the preset roller circumferences that match the comparison as abnormal rollers, and prompt for investigation of the abnormal rollers. By statistically analyzing regular scratches, the scope of investigation can be further narrowed, and the investigation efficiency can be improved.

[0056] In one embodiment, the system further includes a result display module for displaying at least one of the defect cause and diagnostic results. The result display module can also be used to display the defect handling status, etc.

[0057] Please see Figure 4 , Figure 4 A schematic diagram of a strip steel defect diagnosis expert system provided in another embodiment of the present invention is shown below. Figure 4 As shown, this strip steel defect diagnosis expert system ( Figure 4 The defect analysis expert system shown includes a defect query submodule, a defect identification submodule, a cause analysis submodule, a countermeasure suggestion submodule, a defect statistics submodule, a knowledge base maintenance submodule, and a system maintenance submodule. The defect query submodule is the same as the defect query module in the above embodiments. The defect identification submodule includes the image defect identification module and the data defect identification module in the above embodiments. The cause analysis submodule is the defect cause determination module in the above embodiments. The countermeasure suggestion submodule is the defect diagnosis module in the above embodiments. The defect statistics submodule is the defect statistics module in the above embodiments. The knowledge base maintenance submodule is used for knowledge acquisition to enrich the knowledge base. Figure 4 The system's comprehensive database (knowledge base) and maintenance submodule serve as the maintenance module for this strip steel defect diagnosis expert system, used for updates, error correction, etc. The knowledge base, via an inference engine, provides services to the defect identification submodule, cause analysis submodule, and countermeasure suggestion submodule.

[0058] Please see Figure 5 , Figure 5 A schematic diagram illustrating the information input and output of a strip steel defect diagnosis expert system provided in another embodiment of the present invention is shown below. Figure 5 As shown, a camera can be used as an image acquisition device to acquire images to be diagnosed. These images are then processed by an image defect recognition module to obtain a preset image set, which is stored in an image database. Data from production line L2 is collected as production data to be diagnosed. This data is analyzed by a data defect recognition module to obtain a preset data range set, which is stored in a process database. A defect query module is also used to analyze the feature mechanisms based on the input query data, storing the analyzed defect mechanisms in a principle database. An expert knowledge update dialogue module is used to update the knowledge base through expert dialogue. The data from the image database, process database, principle database, and knowledge base are compiled and reasoned using compilation and reasoning mechanisms, integrating process, simulation, and experience perspectives. The final countermeasure recommendations are then obtained through the defect diagnosis module.

[0059] The strip steel defect diagnosis expert system provided in the above embodiments can analyze quality defects in various aspects of strip steel, including surface quality, internal quality, shape and dimensional quality, roughness and waviness, mechanical properties, processing performance, and service performance. It features functions such as defect cause query and expert experience summary, and provides seven major modules: defect query, defect identification, cause analysis, countermeasure suggestions, defect statistics, and knowledge base maintenance. Supported input information includes, but is not limited to: production line process parameters, defect image information collected before strip steel grinding, defect image information collected after strip steel grinding, results of various strip steel analyses and tests, downstream feedback information from subsequent production processing and use, defect feature query dialogue, and expert knowledge update dialogue. Through expert knowledge update dialogue, the characteristics, mechanisms, and solutions of previously occurring and successfully resolved defects are solidified as a basis for future analysis. Supported output results include, but are not limited to, the possible causes, mechanisms, and solutions for the corresponding defects. The defect reasoning and analysis methods provided in the above embodiments include, but are not limited to: defect feature inference, defect generation mechanism inference, defect generation cycle self-learning analysis, and process parameter deviation self-learning analysis. The defect generation cycle self-learning analysis method can compare the defect occurrence cycle length with the circumference of various rollers on the production line. When the two match, the relevant rollers are displayed on the production line diagram, and operators can then investigate based on other defect characteristics. The strip steel defect expert analysis system's process parameter deviation self-learning analysis can be as follows: based on the classification of the generated defects, various related process parameters are scanned item by item, and items that do not conform to the process standards are displayed in the form of alarms, allowing operators to investigate based on other defect characteristics.

[0060] The defect query module uses human-computer dialogue to collect input query data, including but not limited to: defect shape, defect size, occurrence cycle, whether it appears on the upper or lower surface of the strip, its location on the strip, number of defects, its location on the production line, abnormal production conditions, and whether it continues to occur after a certain elimination method, etc. When classifying defects, it provides a cause-finding method that proceeds from major categories to minor categories, dividing them into primary, secondary, tertiary...N levels, analyzing layer by layer until the final cause is found.

[0061] The strip steel defect diagnosis expert system provided in the above embodiments replaces the existing manual grinding with intelligent robotic grinding and replaces manual identification of surface defects of the steel strip after grinding with intelligent image recognition. At the same time, it provides a set of expert analysis systems for all defects. Once a defect is found, the cause of the defect can be found by following the clues and clues, so that the operator can make timely process adjustments and resolve the defect smoothly.

[0062] This invention proposes an expert system for diagnosing defects in strip steel. The system includes a data input module for the data to be diagnosed in the strip steel; an image defect recognition module for identifying the image to be diagnosed to obtain image defect diagnosis results; a data defect recognition module for comparing the sub-data to be diagnosed with standard sub-data to obtain the data qualification status of each sub-data; a knowledge base for storing multiple preset defect causes and multiple preset diagnostic results, as well as the correspondence between preset defect causes and preset diagnostic results; a defect cause determination module for determining the defect cause based on at least one of the defect image and abnormal data; and a defect diagnosis module for querying the knowledge base based on the defect cause to obtain the diagnostic result of the strip steel to be diagnosed. Through this system, a method is provided for diagnosing the data to be diagnosed in a continuous strip steel production line using machine judgment, obtaining the defect causes and countermeasures of the strip steel, reducing the time consumption of defect diagnosis, reducing potential scrap or defect losses, and reducing reliance on the experience of experienced workers.

[0063] In another embodiment of the present invention, a strip steel defect diagnosis device is provided, which is configured with the strip steel defect diagnosis expert system mentioned in any of the above embodiments.

[0064] In this embodiment, the specific functions and technical effects of the strip steel defect diagnosis device are the same as those in the above embodiments, and will not be repeated here.

[0065] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.

Claims

1. A strip steel defect diagnostic expert system characterized by, The system comprises: a data input module for diagnosing data of a strip steel to be diagnosed, the diagnosing data comprising at least one of a diagnosing image and diagnosing production data, the diagnosing production data comprising a plurality of diagnosing sub-data; an image defect identification module for identifying the diagnosing image to obtain an image defect diagnosis result, the image defect diagnosis result comprising a defect existence state, and the image defect diagnosis result further comprising a defect image if the defect existence state is that defects exist; a data defect identification module for comparing the diagnosing sub-data with standard sub-data to obtain a data qualification state of each diagnosing sub-data; a knowledge base for storing a plurality of preset defect causes and a plurality of preset diagnosis results, and a corresponding relationship between the preset defect causes and the preset diagnosis results; a defect cause determination module for determining a defect cause according to at least one of the defect image and abnormal data, the abnormal data comprising diagnosing sub-data with an unqualified data qualification state, wherein the defect cause determination module comprises a plurality of defect matching sub-modules and a preset matching order of the plurality of defect matching sub-modules, the diagnosing data of the strip steel to be diagnosed is matched with preset determination data of the defect matching sub-modules according to the preset matching order, and a preset defect cause of a defect matching sub-module that finally matches successfully is determined as the defect cause of the strip steel to be diagnosed, wherein the diagnosing data comprises at least one of the defect image and the abnormal data, the preset determination data comprises at least one of a preset image set and a preset data range set, and the defect matching sub-module is previously provided with the preset defect cause and the preset determination data; a defect diagnosis module for querying the knowledge base according to the defect cause to obtain a diagnosis result of the strip steel to be diagnosed.

2. The strip steel defect diagnostic expert system of claim 1, wherein, The system further comprises a defect query module, and the defect query module comprises: a query receiving module for receiving input query data, the input query data comprising at least one of a text statement, a voice statement, and an input image; a statement extraction module for extracting query features from the text statement or the voice statement if the input query data comprises the text statement or the voice statement, the extracted query features comprising at least one of a defect shape, a defect size, a defect occurrence period, a defect strip steel position, a defect production line position, a defect quantity, and production line running abnormal data; a data conversion module for determining the input image as a diagnosing image and determining the extracted query features as diagnosing production data.

3. The strip steel defect diagnostic expert system of claim 1, wherein, The knowledge base further comprises an expert knowledge update dialogue module for storing newly acquired historical defects, historical defect causes, and historical defect countermeasures into the knowledge base.

4. A strip steel defect diagnostic expert system according to any one of claims 1 to 3, characterised in that, The defect diagnosis module comprises a defect countermeasure determination module for comparing the defect cause with a plurality of preset defect causes in the knowledge base, determining a preset diagnosis result corresponding to a preset defect cause that is successfully compared as a diagnosis result of the strip steel to be diagnosed, and the preset diagnosis result comprising a countermeasure suggestion.

5. The strip steel defect diagnostic expert system of any one of claims 1-3, wherein, If the production data to be diagnosed includes multiple production process parameters, the system further comprises a process anomaly detection module configured to compare the production process parameters with preset standard process parameters, and if the production process parameters do not meet the preset standard process parameters, generate an alarm message based on the production process parameters to prompt an investigation of the equipment related to the production process parameters.

6. The strip steel defect diagnostic expert system of any one of claims 1-3, wherein, The system further comprises a defect statistics module configured to count the defect positions of the strip steel to be diagnosed, determine a defect occurrence interval distance, compare the defect occurrence interval distance with multiple preset roll circumferences, determine a roll corresponding to a preset roll circumference that is successfully compared as an abnormal roll, and prompt an investigation of the abnormal roll.

7. The strip steel defect diagnostic expert system of any one of claims 1-3, wherein, The system further comprises a result display module configured to display at least one of the defect causes and the diagnosis results.

8. The strip steel defect diagnostic expert system of any one of claims 1-3, wherein, If the image to be diagnosed is a strip steel pre-polishing surface image, the image defect recognition module is configured to perform image fusion on multiple strip steel pre-polishing surface images collected under different illumination states to obtain a fused image, recognize each crystal flower in the fused image, and determine the initial crystal flower size of each crystal flower; determine the crystal flower uniformity and crystal flower representation size according to the differences in the initial crystal flower sizes of all the crystal flowers; and determine the crystal flower quality data of the strip steel to be diagnosed based on at least one of the crystal flower uniformity, the crystal flower representation size, and the number of crystal flowers.

9. The strip steel defect diagnostic expert system of any one of claims 1-3, wherein, If the image to be diagnosed is a strip steel post-polishing surface image, the image defect recognition module is configured to perform primary defect recognition on the strip steel post-polishing surface image, and if a defect is recognized, generate a suggested polishing message based on the defect position of the defect to guide a re-polishing of the surface of the strip steel to be diagnosed through the suggested polishing message. If a re-polishing completion message is received, a processed strip steel surface image of the surface of the strip steel to be diagnosed is collected, and re-defect recognition is performed on the processed strip steel surface image to complete the defect recognition of the image to be diagnosed according to the re-defect recognition result of the re-defect recognition.

10. A strip steel defect diagnostic apparatus characterized by comprising: The strip steel defect diagnosis device is configured with the strip steel defect diagnosis expert system according to any one of claims 1-9.

Images