Heat treatment furnace steel wire cutting method and control device
By using transparent channels and dual CCD image sensors for real-time defect detection and automatic cutting during wire cutting, the inefficiency problem caused by offline defect processing in traditional methods is solved, and an efficient wire cutting process is achieved.
Patent Information
- Application Number
- CN202411120549.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-15
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2044-08-15
AI Technical Summary
In traditional wire cutting methods, defect treatment is usually carried out offline, resulting in low processing efficiency and affecting the continuity and overall efficiency of the production line.
Transparent wire channel and dual CCD image sensor are used for real-time image acquisition and defect detection. The pre-constructed defect detection model is used to automatically identify the surface defects of the steel wire, and a cutting instruction is generated after the defect is detected for automatic cutting, ensuring that the defective steel wire is separated in time.
Real-time defect detection and cutting processing are realized, the continuity and efficiency of the production process are improved, manual intervention and error are reduced, and product quality is improved.
Smart Images

Figure CN118635413B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of steel wire cutting, and in particular to a method and a control device for cutting steel wire emerging from a heat treatment furnace. Background Art
[0002] Surface defects such as cracks, pits, and oxidation spots are easily generated during the heat treatment process, which will reduce the quality of the steel wire. In the traditional steel wire cutting method, defect processing is usually carried out offline, and the detection and processing steps are separated, resulting in low processing efficiency, affecting the continuity and overall efficiency of the production line. In addition, the defect detection process cannot be closely matched with the transmission speed of the steel wire, which is easy to cause delays, resulting in defective steel wires that may enter the subsequent process, affecting production efficiency. Summary of the invention
[0003] The present application provides a method and a control device for cutting heat-treated steel wire, aiming to solve the technical problem that defect processing in traditional steel wire cutting methods is usually performed offline, and the detection and processing steps are separated, resulting in low processing efficiency and affecting the continuity and overall efficiency of the production line.
[0004] The first aspect disclosed in the present application provides a method for cutting heat-treated steel wire, the method comprising: transferring the heat-treated and cooled steel wire to a cutting area, wherein the cutting area comprises a transparent steel wire channel, and the steel wire channel sequentially deploys a defect detection component, a steel wire cutting component, and a steel wire stopping component along the steel wire transfer direction, the defect detection component comprises an image acquisition module and a data processing module, the image acquisition module comprises a first CCD image sensor located above the steel wire channel, and a second CCD image sensor located below the steel wire channel; when the steel wire passes through the defect detection component, the image acquisition module performs real-time image acquisition of the steel wire based on the first CCD image sensor and the second CCD image sensor to obtain a steel wire image data stream; and the data processing module performs real-time image acquisition of the steel wire based on the first CCD image sensor and the second CCD image sensor to obtain a steel wire image data stream. The embedded steel wire surface defect detection model performs surface defect detection on the steel wire image data stream to obtain a steel wire surface defect detection result, wherein the steel wire surface defect detection model is a pre-built and trained defect detection model for identifying steel wire surface defects; when the steel wire surface defect detection result is that a surface defect exists, a first cutting instruction is generated, and the steel wire cutting component is started based on the first cutting instruction to cut the steel wire to obtain scrap steel wire, and the scrap steel wire is placed in a scrap steel wire recovery area; when the steel wire surface defect detection result is that there is no surface defect, the steel wire is continuously conveyed until it reaches the steel wire stop component, and a second cutting instruction is generated, and the steel wire cutting component is started based on the second cutting instruction to cut the steel wire to obtain qualified steel wire, and the qualified steel wire is placed in a qualified steel wire collection area.
[0005] Another aspect disclosed in the present application provides a heat-treated steel wire cutting control device, the heat-treated steel wire cutting control device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the data interaction of the heat-treated steel wire cutting method when executing the computer program.
[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages:
[0007] The first and second CCD image sensors are used to capture images of the steel wire from above and below respectively, ensuring high-resolution image data covering the entire surface of the steel wire; through the real-time collaboration of the image acquisition module and the data processing module, image data can be collected and processed in real time during the steel wire transmission process, and surface defects can be detected in time; after the defect is detected, the first cutting instruction is generated, and the steel wire cutting component is automatically started to cut the defective part, ensuring that the defective steel wire is separated and put into the recycling area in time; when the test result shows no defect, the second cutting instruction is generated, and the qualified steel wire is cut and put into the collection area, ensuring the continuity and efficiency of the production process. In general, from defect detection to cutting processing, the entire process is automatically controlled by the system without manual intervention, reducing labor intensity and human errors, and further improving product quality and production efficiency by analyzing defect data.
[0008] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS
[0009] Figure 1 A schematic flow chart of a method for cutting heat-treated steel wire is provided for an embodiment of the present application;
[0010] Figure 2 A schematic flow chart of constructing a steel wire surface defect detection model in a method for cutting heat-treated steel wire is provided for an embodiment of the present application;
[0011] Figure 3 An internal structure diagram of a heat treatment furnace wire cutting control device is provided for an embodiment of the present application. DETAILED DESCRIPTION
[0012] The embodiments of the present application provide a method for cutting heat-treated steel wire, thereby solving the technical problem that defect processing in traditional steel wire cutting methods is usually performed offline, and the detection and processing steps are separated, resulting in low processing efficiency and affecting the continuity and overall efficiency of the production line.
[0013] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically described below in conjunction with the drawings in the specification.
[0014] like Figure 1 As shown, the embodiment of the present application provides a method for cutting heat-treated steel wire, the method comprising:
[0015] The steel wire that has been heat treated and cooled is conveyed to a cutting area, wherein the cutting area comprises a transparent steel wire channel, and a defect detection component, a steel wire cutting component, and a steel wire stopping component are sequentially arranged along the steel wire conveying direction. The defect detection component comprises an image acquisition module and a data processing module, and the image acquisition module comprises a first CCD image sensor located above the steel wire channel, and a second CCD image sensor located below the steel wire channel.
[0016] The heat-treated and cooled steel wire is sent to the cutting area through a conveyor. The cutting area includes a transparent steel wire channel, which is a channel for conveying steel wire. The transparent design facilitates steel wire image acquisition. Defect detection components, steel wire cutting components, and steel wire stop components are deployed in sequence along the steel wire conveying direction. Among them, the defect detection component is a component used to detect surface defects of steel wire, including an image acquisition module and a data processing module; the steel wire cutting component is a component used to cut the steel wire; the steel wire stop component is a component used to stop the steel wire conveying when it reaches a predetermined position.
[0017] The image acquisition module includes two CCD image sensors, wherein the first CCD image sensor is installed above the steel wire channel to capture the image of the steel wire from above; the second CCD image sensor is installed below the steel wire channel to capture the image of the steel wire from below. The data processing module is used to process the image data obtained from the image acquisition module to detect surface defects of the steel wire.
[0018] In the direction of wire transmission, the defect detection component, wire cutting component and wire stopping component are arranged in sequence, which ensures that the wire can be cut immediately after detection and stops transmission after reaching a specific position. The design of transparent wire channel and dual CCD image sensor improves the accuracy and efficiency of detection.
[0019] When the steel wire passes through the defect detection component, the image acquisition module performs real-time image acquisition on the steel wire based on the first CCD image sensor and the second CCD image sensor to obtain a steel wire image data stream.
[0020] When the steel wire passes through the defect detection component, the detection process starts automatically, and the first CCD image sensor and the second CCD image sensor work simultaneously to synchronously capture images of the steel wire passing through the sensor field of view, wherein the first CCD image sensor captures the surface image of the steel wire from top to bottom, and the second CCD image sensor captures the surface image of the steel wire from bottom to top. After continuous image acquisition, a continuous steel wire image data stream is formed, which includes the image data of each part of the surface of the steel wire when it passes through the detection component, ensuring seamless image data acquisition during the movement of the steel wire.
[0021] The steel wire image data stream is subjected to surface defect detection by a steel wire surface defect detection model embedded in the data processing module to obtain a steel wire surface defect detection result, wherein the steel wire surface defect detection model is a pre-built and trained defect detection model for identifying steel wire surface defects.
[0022] The data processing module is used to process and analyze the image data stream transmitted from the image acquisition module. The module has an embedded steel wire surface defect detection model, which is a trained algorithm used to identify various defects on the steel wire surface. Specifically, the steel wire surface defect detection model is obtained through supervised training based on historical image data of steel wire surface defects. These historical image data integrate the factor indicators of steel wire surface defects, such as heating temperature, heating time, etc. The training process includes learning features from a large number of images marked with defects. Combined with the influence of these factors, the model can more accurately determine the defect type and location. The specific training process of the model is detailed in subsequent steps. Based on the pre-built and trained defect detection model, the steel wire image data stream is subjected to defect recognition, and possible surface defects such as cracks, pits, oxidation spots, etc. are identified to generate steel wire surface defect detection results.
[0023] When the surface defect detection result of the steel wire shows that there are surface defects, a first cutting instruction is generated, and the steel wire cutting component is started to cut the steel wire based on the first cutting instruction to obtain scrap steel wire, and the scrap steel wire is put into a scrap steel wire recycling area.
[0024] When the test results show that the steel wire has surface defects, a first cutting instruction is generated. The first cutting instruction is used to start the cutting operation when the steel wire has defects and to cut off the defective part in time. Based on the first cutting instruction, the steel wire cutting component is activated to perform the cutting operation to cut the defective steel wire part from the whole steel wire. After the cutting operation, scrap steel wire is generated. The scrap steel wire refers to the part with surface defects that has been cut off. The scrap steel wire is transported to a designated scrap steel wire recycling area, such as a waste bin, for centralized processing and recycling of the scrap steel wire.
[0025] When the result of the steel wire surface defect detection is that there are no surface defects, the steel wire is continuously conveyed until it reaches the steel wire stopping component, a second cutting instruction is generated, and the steel wire cutting component is started based on the second cutting instruction to cut the steel wire to obtain qualified steel wire, and the qualified steel wire is put into a qualified steel wire collection area.
[0026] When the test results show that there are no defects on the surface of the steel wire, the steel wire continues to be transported forward until it reaches the preset position of the steel wire stop assembly. The steel wire stop assembly is used to control the end point of the steel wire transmission to ensure the accuracy of the cutting position. When the steel wire reaches the steel wire stop assembly, a second cutting instruction is generated. The second cutting instruction is used to start the cutting operation when the steel wire length reaches the preset requirement, and instruct the cutting assembly to cut the steel wire at the appropriate position.
[0027] Based on the second cutting instruction, the wire cutting component is started to perform the cutting operation, and the qualified wire is cut from the whole wire. After the cutting operation is completed, the obtained wire portion is a qualified wire, and no defects are detected on its surface. The cut qualified wire is transferred to the qualified wire collection area for centralized collection and storage of qualified wires.
[0028] Furthermore, if Figure 2 As shown, the steel wire surface defect detection model is constructed, including:
[0029] Obtain steel wire surface defect factor indicators, wherein the steel wire surface defect factor indicators include heating temperature factors, heating time factors, oxygen content factors, mechanical stress factors, and furnace impurity factors; establish a steel wire surface defect image database based on big data, classify the steel wire surface defect image database based on the steel wire surface defect factor indicators, and obtain a steel wire surface defect factor classification sample set; perform supervised training on the steel wire surface defect factor classification sample set, and construct the steel wire surface defect detection model.
[0030] The historical steel wire surface defect data are collected, cleaned and standardized to form a set of standard steel wire surface defect indicators, the covariance matrix is calculated, and the eigenvalue decomposition is performed to obtain the eigenvalues and eigenvectors, the main eigenvectors are selected as the principal components, the principal component matrix is established, and the data are projected into the principal component matrix to obtain the steel wire surface defect factor indicators, which include heating temperature factor, heating time factor, oxygen content factor, mechanical stress factor, and impurity factor in the furnace.
[0031] Collect a large number of steel wire surface defect images, which can come from the real-time monitoring system on the production line, historical defect data, experimental data, etc. The images contain different types of defects, such as cracks, pits, oxidation spots, etc. The images are annotated according to the causes of the defects in the images, and the annotated images are stored in a pre-built steel wire surface defect image database. According to the defect factor indicators such as heating temperature, heating time, oxygen content, mechanical stress, impurities in the furnace, etc., the image data is classified to obtain a collection of classified samples of steel wire surface defect factors.
[0032] The classified sample set is divided into training set and test set according to a certain ratio. Appropriate supervised learning algorithms, such as random forest, support vector machine, convolutional neural network (CNN), etc., are selected to build defect detection models. The selected model is trained with the training set. The model parameters are optimized through repeated iterations so that the model can accurately identify and classify surface defects of steel wires. The trained model is tested and evaluated with the test set. According to the evaluation results, the model is optimized and the training and evaluation process is repeated until the model performance reaches the preset standard. The obtained model is saved and deployed in the data processing module for real-time defect detection and improving the quality control of the production process.
[0033] Furthermore, the steel wire surface defect factor indicators are obtained, including:
[0034] Acquire a set of historical steel wire surface defect indicators; preprocess the set of historical steel wire surface defect indicators to acquire a set of standard steel wire surface defect indicators, wherein the preprocessing includes data cleaning and data standardization; based on the standard steel wire surface defect indicator set, calculate the covariance matrix of the data, perform eigenvalue decomposition on the covariance matrix, and obtain eigenvalues and eigenvectors; select principal components according to the eigenvalues, and establish a principal component matrix in combination with the eigenvectors; project the standard steel wire surface defect indicator set onto the principal component matrix to obtain the steel wire surface defect factor indicators.
[0035] The historical data of steel wire surface defects are collected from past production records, quality control systems, experimental data and maintenance reports. These data include defect type, location, severity, and process parameters related to defects, such as heating temperature, heating time, oxygen content, mechanical stress, impurities in the furnace, etc. The historical defect data from different sources are integrated to obtain a set of historical steel wire surface defect indicators.
[0036] Perform data cleaning, including detecting and processing missing values in the data, such as using interpolation, mean filling or deleting data entries containing missing values; identifying and processing outliers, such as determining outliers through statistical analysis methods and processing them, such as deleting or correcting outliers; identifying and deleting duplicate data to ensure the uniqueness of each record.
[0037] Data standardization is performed, for example, using the minimum-maximum scaling method to scale the data to the same range, such as between 0 and 1, so that different indicators are comparable. After preprocessing, a set of standard steel wire surface defect indicators is obtained.
[0038] The preprocessed standard steel wire surface defect index set is used as input data. These data sets are in matrix form, each row represents a sample, each column represents a defect factor index, and the mean of each defect factor index (each column) is calculated. These means are used for data centering, that is, for each value in the data matrix, the corresponding column mean is subtracted, and the centered data matrix is used to calculate the covariance matrix. The covariance matrix is a square matrix whose elements represent the covariance between different defect factor indicators. The specific calculation method is that for the centered data matrix, each pair of defect factor indicators calculates the covariance between them. The covariance matrix reflects the correlation between different indicators in the data.
[0039] The calculated covariance matrix is subjected to eigenvalue decomposition. The result of the eigenvalue decomposition includes a set of eigenvalues and corresponding eigenvectors. The eigenvalues represent the variance of the data in the direction of the corresponding eigenvectors, and the eigenvectors represent the principal components of the data in these directions. The eigenvalues are arranged in descending order. The eigenvector corresponding to the largest eigenvalue represents the direction in which the data changes the most. Each eigenvector is a new benchmark after a dimension reduction. These vectors together constitute a new coordinate system to represent the main features of the original data.
[0040] Sort by the size of the eigenvalues and select the first k principal components with larger eigenvalues. The cumulative variance contribution rate can be used to determine the value of k. For example, the number of principal components that can explain 95% of the variance is selected. These principal components explain the most variance in the data. The principal component matrix is composed according to the eigenvectors corresponding to the selected eigenvalues.
[0041] The standardized set of steel wire surface defect indicators is projected onto the principal component matrix. The projected data is represented in the principal component space, which retains most of the information of the original data but has a lower dimension, which is convenient for subsequent analysis.
[0042] Furthermore, supervised training is performed on the steel wire surface defect factor classification sample set to construct the steel wire surface defect detection model, including:
[0043] Based on the steel wire surface defect factor index, the steel wire surface defect factor classification sample set is labeled; according to a preset division ratio, the labeled steel wire surface defect factor classification sample set is divided into data to obtain a training set and a test set; the initial hyperparameters of the random forest structure are preset to build a random forest model; the random forest model is trained using the training set, and multiple decision trees are built by maximizing information gain, wherein each decision tree corresponds to one of the steel wire surface defect factor indexes; the model is tested using the test set to obtain a model evaluation result, the initial hyperparameters are optimized according to the model evaluation result, and the optimization process is repeated until the preset model evaluation standard is reached, and then the steel wire surface defect detection model is output.
[0044] Using the obtained steel wire surface defect factor indicators, different labels are defined according to factors such as defect type and severity. The labels can be binary, such as defect / no defect, or multi-classification, such as different types of defects. The labeling tools are used to automatically label the samples to form a labeled data set.
[0045] Determine the ratio of the training set and the test set according to specific needs and actual conditions, for example, the ratio of the training set to the test set is 8:2. According to the preset ratio, the randomized sample set is divided into a training set and a test set. The training set is used for model training and accounts for the majority of the total samples; the test set is used for model evaluation and accounts for a small portion of the total samples. If the distribution of various labels in the sample is uneven, a hierarchical division method can be used to ensure that each type of label maintains the same ratio in the training set and the test set.
[0046] Preset the initial hyperparameters of the random forest structure, including determining the number of decision trees to be built in the random forest model, such as 10 trees; determining the maximum depth of each tree, which can be infinite depth or a specific value, such as 10 layers; setting the minimum number of samples required for each node split, such as 2 samples; determining the minimum number of samples contained in each leaf node, such as 1 sample; determining the maximum number of features that can be used for each split, and you can choose the square root of the number of all features. Based on the above initial hyperparameters, build the basic structure of the random forest model.
[0047] The random forest model is trained using the prepared training set data. The random forest generates multiple decision trees. Each decision tree is built based on different features and learns the training data. During the training process, the model determines how to split each node by calculating the information gain, so as to maximize the classification information obtained by each split. By continuously splitting nodes, each decision tree is gradually built up until the preset stop condition is met, such as reaching the maximum depth or the minimum number of leaf node samples.
[0048] Use an independent test set to evaluate the trained model and test its performance on new data. The evaluation indicators can include accuracy, etc., to evaluate the performance of the model on the test set. According to the evaluation results, adjust the initial hyperparameters of the model, such as changing the number, depth or minimum number of decision trees, retrain the model and evaluate it again, repeating this process until the performance of the model reaches the preset standard. When the performance of the model meets the requirements, it is output as the final steel wire surface defect detection model.
[0049] Furthermore, it also includes:
[0050] Acquire first surface defect attribute information of a first steel wire surface defect factor index; perform criticality analysis on the first surface defect attribute information to obtain first surface defect criticality; and so on, obtain multiple surface defect criticalities of multiple steel wire surface defect factor indexes in the steel wire surface defect factor index; normalize the surface defect criticality, and generate an index weight according to the normalization result; use the index weight as a model fusion parameter to perform weighted fusion of the multiple decision trees.
[0051] A specific steel wire surface defect factor index, such as heating temperature, is randomly selected, and attribute information related to the factor index is extracted from the steel wire surface defect index set. The attribute information includes specific values of the factor index in different samples, such as data of different temperatures during each heating process.
[0052] Analyze the criticality of the first surface defect attribute information, for example, by calculating the number of split nodes and information gain of the factor in the classification decision tree to evaluate its importance to defect prediction. Through analysis and calculation, the criticality of the first steel wire surface defect factor index is obtained. The criticality reflects the importance of the factor in detecting steel wire surface defects.
[0053] Similarly, other steel wire surface defect factor indicators are selected one by one, such as heating time, oxygen content, mechanical stress, impurities in the furnace, etc. For each selected defect factor indicator, the corresponding surface defect attribute information is extracted, and the same method is used to perform criticality analysis on each defect factor indicator to obtain multiple surface defect criticalities.
[0054] All calculated surface defect criticalities are normalized, for example, using the Min-Max normalization method to normalize them to between 0 and 1 to ensure that all criticalities are on the same scale. The weight of each defect factor indicator is generated based on the normalized criticality value. The weight represents the importance of each defect factor in the model, and the sum of all weights is 1.
[0055] In the random forest, each decision tree corresponds to a defect factor indicator, and the normalized indicator weight is used as the fusion parameter of the decision tree, so that each decision tree has a different influence in the final prediction. During the prediction process, the prediction results of each decision tree are weighted according to the model fusion parameters to train the fused random forest model.
[0056] Furthermore, starting the steel wire cutting assembly to cut the steel wire based on the first cutting instruction to obtain the scrap steel wire includes:
[0057] The first cutting instruction includes a delay time interval; when a surface defect is detected, the delay time timer is started, the wire is continuously transmitted within the delay time interval, and defect detection is continuously performed; when it is detected that no surface defect exists in any defect detection, the first cutting instruction is started to perform wire cutting; if surface defects are still detected at the end of the delay time interval, an abnormal alarm is generated.
[0058] The cutting instruction includes a preset delay time interval, which is used to continue to feed the wire for a period of time after the defect is detected in order to determine the best cutting point. The delay time interval is determined according to specific needs and historical data.
[0059] When defects are detected on the surface of the steel wire, the delay timer is started to start timing, and the steel wire continues to be transmitted. During the delay time interval, the steel wire is continuously inspected for defects until it is determined that there are no defects on the current surface of the steel wire during any inspection, indicating that the defective area has ended. The first cutting instruction is started to instruct the cutting component to cut at the current position to separate the defective portion from the non-defective portion.
[0060] If defects are still detected on the wire surface at the end of the delay time interval, it means that the defective area is too long, and an abnormal alarm is generated to prompt the operator that there is a persistent defect problem so that necessary measures can be taken, such as stopping the production line, checking equipment, adjusting process parameters, etc.
[0061] Furthermore, it also includes:
[0062] Obtain a continuous cutting times threshold and a frequency monitoring time interval; after wire cutting, monitor the continuous cutting times within the frequency monitoring time interval; when the continuous cutting times reaches the continuous cutting times threshold, generate an abnormal alarm.
[0063] The continuous cutting times threshold refers to the maximum continuous cutting times allowed within the frequency monitoring time interval, which can be determined based on historical data and production experience. For example, if the average number of cuts per hour in normal production should not exceed 5, the threshold can be set to 5. The frequency monitoring time interval refers to the period of time used to monitor the continuous cutting times, which can be determined based on the operating speed of the production line, for example, set to 1 hour.
[0064] After each wire cutting operation, the timestamp and number of the cutting event are recorded, and the number of cutting times is continuously monitored during the frequency monitoring time interval. During the frequency monitoring time interval, the current number of cutting times is continuously compared with the set threshold. For example, if the set frequency monitoring time interval is 1 hour, the number of cutting times within this hour is continuously checked. When it is detected that the number of cutting times reaches or exceeds the set continuous cutting number threshold, an abnormal alarm is immediately generated to notify the operator to conduct an abnormal inspection.
[0065] In summary, the heat-treated steel wire cutting method provided in the embodiment of the present application has the following technical effects:
[0066] The first and second CCD image sensors are used to capture images of the steel wire from above and below respectively, ensuring high-resolution image data covering the entire surface of the steel wire; through the real-time collaboration of the image acquisition module and the data processing module, image data can be collected and processed in real time during the steel wire transmission process, and surface defects can be detected in time; after the defect is detected, the first cutting instruction is generated, and the steel wire cutting component is automatically started to cut the defective part, ensuring that the defective steel wire is separated and put into the recycling area in time; when the test result shows no defect, the second cutting instruction is generated, and the qualified steel wire is cut and put into the collection area, ensuring the continuity and efficiency of the production process. In general, from defect detection to cutting processing, the entire process is automatically controlled by the system without manual intervention, reducing labor intensity and human errors, and further improving product quality and production efficiency by analyzing defect data.
[0067] Based on the same inventive concept as the heat-treated steel wire cutting method in the aforementioned embodiment, the present application provides a heat-treated steel wire cutting control device, the internal structure of which is shown in FIG. Figure 3 As shown, the heat-treated steel wire cutting control device includes a memory and a processor, the memory stores a computer program, and the processor implements data interaction of the heat-treated steel wire cutting method when executing the computer program.
[0068] Through the above-mentioned detailed description of the method for cutting heat-treated steel wire, those skilled in the art can clearly understand the heat-treated steel wire cutting control device in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.
[0069] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A method for cutting heat-treated steel wire, characterized in that: The method comprises: The steel wire after heat treatment and cooling is conveyed to a cutting area, wherein the cutting area comprises a transparent steel wire channel, and the steel wire channel is sequentially provided with a defect detection component, a steel wire cutting component, and a steel wire stopping component along the steel wire conveying direction, and the defect detection component comprises an image acquisition module and a data processing module, and the image acquisition module comprises a first CCD image sensor located above the steel wire channel, and a second CCD image sensor located below the steel wire channel; When the steel wire passes through the defect detection component, the image acquisition module performs real-time image acquisition on the steel wire based on the first CCD image sensor and the second CCD image sensor to obtain a steel wire image data stream; Performing surface defect detection on the steel wire image data stream through a steel wire surface defect detection model embedded in the data processing module to obtain a steel wire surface defect detection result, wherein the steel wire surface defect detection model is a pre-built and trained defect detection model for identifying steel wire surface defects; When the surface defect detection result of the steel wire shows that there is a surface defect, a first cutting instruction is generated, and based on the first cutting instruction, the steel wire cutting component is started to cut the steel wire to obtain scrap steel wire, and the scrap steel wire is put into a scrap steel wire recycling area; When the steel wire surface defect detection result is that there is no surface defect, the steel wire is continuously conveyed until it reaches the steel wire stopping component, a second cutting instruction is generated, and the steel wire cutting component is started to cut the steel wire based on the second cutting instruction to obtain qualified steel wire, and the qualified steel wire is put into a qualified steel wire collection area; Wherein, constructing the steel wire surface defect detection model comprises: Obtaining steel wire surface defect factor indicators, wherein the steel wire surface defect factor indicators include heating temperature factors, heating time factors, oxygen content factors, mechanical stress factors, and furnace impurity factors; Establishing a steel wire surface defect image database based on big data, classifying the steel wire surface defect image database based on the steel wire surface defect factor index, and obtaining a steel wire surface defect factor classification sample set; Conducting supervised training on the steel wire surface defect factor classification sample set to construct the steel wire surface defect detection model; Among them, the indicators of steel wire surface defect factors are obtained, including: Obtain a historical set of steel wire surface defect indicators; Preprocessing the historical steel wire surface defect index set to obtain a standard steel wire surface defect index set, wherein the preprocessing includes data cleaning and data standardization; Based on the standard steel wire surface defect index set, the covariance matrix of the data is calculated, and the eigenvalue decomposition of the covariance matrix is performed to obtain eigenvalues and eigenvectors; Selecting principal components according to the eigenvalues, and establishing a principal component matrix in combination with the eigenvectors; Projecting the standard steel wire surface defect index set onto the principal component matrix to obtain the steel wire surface defect factor index; The supervised training is performed on the steel wire surface defect factor classification sample set to construct the steel wire surface defect detection model, including: Labeling the steel wire surface defect factor classification sample set based on the steel wire surface defect factor index; According to a preset division ratio, data is divided on the marked steel wire surface defect factor classification sample set to obtain a training set and a test set; Preset the initial hyperparameters of the random forest structure and build a random forest model; The random forest model is trained using the training set, and a plurality of decision trees are constructed by maximizing information gain, wherein each decision tree corresponds to one of the steel wire surface defect factor indicators; Using the test set to perform model testing, obtaining model evaluation results, optimizing the initial hyperparameters according to the model evaluation results, repeating the optimization process until a preset model evaluation standard is reached, and outputting the steel wire surface defect detection model; Among them, multiple decision trees are constructed by maximizing information gain, and also include: Acquire first surface defect attribute information of a first steel wire surface defect factor index; Performing a criticality analysis on the first surface defect attribute information to obtain a first surface defect criticality; By analogy, a plurality of surface defect criticalities of a plurality of steel wire surface defect factor indicators in the steel wire surface defect factor indicators are obtained; Normalizing the criticality of the surface defects, and generating an indicator weight according to the normalization result; The indicator weight is used as a model fusion parameter to perform weighted fusion of the multiple decision trees.
2. The method for cutting heat-treated steel wire according to claim 1, characterized in that: Starting the steel wire cutting assembly to cut the steel wire based on the first cutting instruction to obtain the scrap steel wire includes: The first cutting instruction includes a delay time interval; When a surface defect is detected, a delay timer is started, and the steel wire is continuously conveyed and defect detection is continuously performed within the delay time interval; When no surface defects are detected in any defect detection, a first cutting instruction is initiated to perform wire cutting; If surface defects are still detected at the end of the delay time interval, an abnormality alarm is generated.
3. The method for cutting heat-treated steel wire according to claim 1, characterized in that: Also includes: Get the continuous cutting times threshold and frequency monitoring time interval; After the wire cutting is performed, the number of continuous cutting times is monitored within the frequency monitoring time interval; When the continuous cutting times reaches the continuous cutting times threshold, an abnormality alarm is generated.
4. A heat treatment furnace steel wire cutting control device, comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, data interaction of the method for cutting heat-treated steel wire according to any one of claims 1 to 3 is implemented.
